May 2018, Volume 24, Number 2 by Association of Environmental & Engineering Geologists (AEG)

Environmental & Engineering Geoscience MAY 2018

VOLUME XXIV, NUMBER 2

THE JOINT PUBLICATION OF THE ASSOCIATION OF ENVIRONMENTAL AND ENGINEERING GEOLOGISTS AND THE GEOLOGICAL SOCIETY OF AMERICA SERVING PROFESSIONALS IN ENGINEERING GEOLOGY, ENVIRONMENTAL GEOLOGY, AND HYDROGEOLOGY

Environmental & Engineering Geoscience (ISSN 1078-7275) is published quarterly by the Association of Environmental & Engineering Geologists (AEG) and the Geological Society of America (GSA). Periodicals postage paid at AEG, 201 East Main St., Suite 1405, Lexington, KY 40507 and additional mailing offices. EDITORIAL OFFICE: Environmental & Engineering Geoscience journal, Department of Geology, Kent State University, Kent, OH 44242, U.S.A. phone: 330-672-2968, fax: 330-672-7949, ashakoor@kent.edu. CLAIMS: Claims for damaged or not received issues will be honored for 6 months from date of publication. AEG members should contact AEG, 201 East Main St., Suite 1405, Lexington, KY 40507. Phone: 844-331-7867. GSA members who are not members of AEG should contact the GSA Member Service center. All claims must be submitted in writing. POSTMASTER: Send address changes to AEG, 201 East Main St., Suite 1405, Lexington, KY 40507. Phone: 844-331-7867. Include both old and new addresses, with ZIP code. Canada agreement number PM40063731. Return undeliverable Canadian addresses to Station A P.O. Box 54, Windsor, ON N9A 6J5 Email: returnsil@imexpb.com. DISCLAIMER NOTICE: Authors alone are responsible for views expressed inarticles. Advertisers and their agencies are solely responsible for the content of all advertisements printed and also assume responsibility for any claims arising therefrom against the publisher. AEG and Environmental & Engineering Geoscience reserve the right to reject any advertising copy. SUBSCRIPTIONS: Member subscriptions: AEG members automatically receive digital access to the journal as part of their AEG membership dues. Members may order print subscriptions for $60 per year. GSA members who are not members of AEG may order for $60 per year on their annual GSA dues statement or by contacting GSA. Nonmember subscriptions are $295 and may be ordered from the subscription department of either organization. A postage differential of $10 may apply to nonmember subscribers outside the United States, Canada, and Pan America. Contact AEG at 844-331-7867; contact GSA Subscription Services, Geological Society of America, P.O. Box 9140, Boulder, CO 80301. Single copies are $75.00 each. Requests for single copies should be sent to AEG, 201 East Main St., Suite 1405, Lexington, KY 40507. © 2018 by the Association of Environmental and Engineering Geologists All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage and retrieval system, without permission in writing from AEG. THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER Abdul Shakoor Department of Geology Kent State University Kent, OH 44242 330-672-2968 ashakoor@kent.edu

EDITORS

Brian G. Katz Florida Department of Environmental Protection 2600 Blair Stone Rd. Tallahassee, FL 32399 850-245-8233 eegeditorbkatz@gmail.com

EDITORIAL BOARD Jerome V. DeGraff CSU Fresno Chester (Skip) F. Watts Radford University Thomas Oommen Michigan Technological Univ. Syed E. Hasan University of Missouri

Thomas J. Burbey Virginia Polytechnic Institute Abdul Shakoor Kent State University Brian G. Katz Florida Department of Environmental Protection

ASSOCIATE EDITORS John W. Bell Nevada Bureau of Mines and Geology Richard E. Jackson Geofirma Engineering, Ltd. Jeffrey R. Keaton AMEC Americas Paul G. Marinos National Technical University of Athens, Greece June E. Mirecki U.S. Army Corps of Engineers Peter Pehme Waterloo Geophysics, Inc Nicholas Pinter Southern Illinois University

Paul M. Santi Colorado School of Mines Robert L. Schuster U.S. Geological Survey Roy J. Shlemon R. J. Shlemon & Associates, Inc. Greg M. Stock National Park Service Resat Ulusay Hacettepe University, Turkey Chester F. “Skip” Watts Radford University Terry R. West Purdue University

SUBMISSION OF MANUSCRIPTS Environmental & Engineering Geoscience (E&EG), is a quarterly journal devoted to the publication of original papers that are of potential interest to hydrogeologists, environmental and engineering geologists, and geological engineers working in site selection, feasibility studies, investigations, design or construction of civil engineering projects or in waste management, groundwater, and related environmental fields. All papers are peer reviewed. The editors invite contributions concerning all aspects of environmental and engineering geology and related disciplines. Recent abstracts can be viewed under “Archive” at the web site, “http://eeg. geoscienceworld.org”. Articles that report on research, case histories and new methods, and book reviews are welcome. Discussion papers, which are critiques of printed articles and are technical in nature, may be published with replies from the original author(s). Discussion papers and replies should be concise. To submit a manuscript go to http://eeg.allentrack.net. If you have not used the system before, follow the link at the bottom of the page that says New users should register for an account. Choose your own login and password. Further instructions will be available upon logging into the system. Please carefully read the “Instructions for Authors”. Authors do not pay any charge for color figures that are essential to the manuscript. Manuscripts of fewer than 10 pages may be published as Technical Notes. For further information, you may contact Dr. Abdul Shakoor at the editorial office. Cover photo View to the north from the Waipio Valley overlook of cliffs that reach a height of 400 m along the Kohala volcano northeast coast on the island of Hawaii. Evidence indicates the cliffs formed as the headwall of a catastrophic debris avalanche several thousand years ago, while the volcanic material exposed in the cliffs is part of a giant translational landslide that extends from the coast to the volcano’s summit. A landslide deposit derived from the cliffs is present near the center of the photo. See article on page 187.

Environmental & Engineering Geoscience

Volume 24, Number 2, May 2018

Table of Contents

137

Technical Note: Crossing the East River of New York in the Era of Mixed Face Tunneling David M. Cregger

137

Technical Note: Crossing the East River of New York in the Era of Mixed Face Tunneling David M. Cregger

143

Evaluation of Rockfall-Hazard Potential for Rockville, Utah, Following a 2013 Fatal Rockfall Carl Jacklitch, Abdul Shakoor and William R. Lund

143

Evaluation of Rockfall-Hazard Potential for Rockville, Utah, Following a 2013 Fatal Rockfall Carl Jacklitch, Abdul Shakoor and William R. Lund

165

Monitoring Thermal Springs to Improve Land Management Decision-Making, Sierra Nevada, California Jerome V. De Graff, Christopher J. Pluhar, Alan J. Gallegos, Kellen Takenaka and Bryant Platt

165

Monitoring Thermal Springs to Improve Land Management Decision-Making, Sierra Nevada, California Jerome V. De Graff, Christopher J. Pluhar, Alan J. Gallegos, Kellen Takenaka and Bryant Platt

187

Landslide Interpretation of the Northeast Flank of Kohala Volcano, Hawaii Kim M. Bishop

187

Landslide Interpretation of the Northeast Flank of Kohala Volcano, Hawaii Kim M. Bishop

207

Correlations between Fluvial Knickpoints and Recurrent Landslide Dams along the Upper Indus River M. Farooq Ahmed, J. David Rogers and Elamin H. Ismail

207

Correlations between Fluvial Knickpoints and Recurrent Landslide Dams along the Upper Indus River M. Farooq Ahmed, J. David Rogers and Elamin H. Ismail

221

Identification and Analysis of Large Paleo-Landslides at Mount Burnaby, British Columbia Mirko Francioni, Doug Stead, John J. Clague and Allison Westin

221

Identification and Analysis of Large Paleo-Landslides at Mount Burnaby, British Columbia Mirko Francioni, Doug Stead, John J. Clague and Allison Westin

237

Using Google Earth and Google Street View to Rate Rock Slope Hazards William Swanger and Yonathan Admassu

237

Using Google Earth and Google Street View to Rate Rock Slope Hazards William Swanger and Yonathan Admassu

Technical Note

Crossing the East River of New York in the Era of Mixed Face Tunneling

DAVID M. CREGGER

AECOM Technology, 250 Apollo Drive, Chelmsford, MA 01824

Key Terms: Engineering, Construction, Tunnels, Geology Engineering geology and case history experience have been relied upon as predictors for several historical East River tunnel crossings, as shown in Figure 1. Differing rock types were crossed, consisting of gneisses, schist, and limestone or dolomite nearly vertical in orientation and part of a complex tightly folded sequence, but deeply weathered along faults parallel to the river. Each crossing had vertical grade requirements, forcing the alignment through known geologic hazards of poor-quality rock and mixed face. The presence of decomposed and highly crushed zones at depth and glacial or man-made deposits further complicated the mixed face conditions for these shallow river crossings. This review illustrates the case history method of precedence, with incremental changes adapted to site conditions as practiced by geo-engineers. EARLY EXPERIENCE—THE STEINWAY TUNNELS The earliest tunnel crossing of the East River was the Steinway tunnel, which used the bolted iron segment tunnel-lining technology of the time. Planning started in the 1880s by the famous piano maker, who was interested in running a horse-drawn trolley line to his properties east of the river. After a decade of politics and economic competition with a proposed bridge crossing, construction was started on the 42nd Street shaft in 1892 at a location due west of the island in the East River known as the Man-O-War Reef. Groundwater inflow problems attributed to an underground spring, broken windows due to blasting and lodging of complaints by residents, and accidental deaths due to the handling of frozen dynamite darkened the outlook for the project. Shaft 1 was sunk to 85 ft (26 m) depth into solid Fordham Gneiss, and the project was abandoned for another decade. With joint ownership by the newly formed Interborough Rapid Transit (known as the IRT) and the Brooklyn-Manhattan Transit (BMT), grades of the tunnel were adjusted, 1 Corresponding

DAVID M. CREGGER

author email: david.cregger@aecom.com

and a new survey was made so that the new electric railcars could make passage from Brooklyn to Manhattan. The Steinway crossings were designed as two round tunnels 15.5 ft (4.7 m) in diameter and supported by cast iron bolted rings at elevations required for the subway. Construction of the twin tunnel 8,000 linear foot (2,438 linear m) segment was started in 1905 and completed in 26 months, with as many as 1,500 men working under 40 psi (276 kPa) compressed air pressures at different times. The tunnel shields were exposed to the water of the East River on descents from the Brooklyn shaft toward the west, but the tunnel eventually went under cover of 25 ft (8 m) of river silt on the bottom below 88 ft (27 m) of water, as illustrated in Figure 2. Shields were also staged from the bedrock “reef” that became known as Belmont Island after its enlargement. Tunneling on double headings from this shaft started from horseshoe sections blasted out at the base of shafts in rock, from which the shields crossed the west channel towards Manhattan and east channel towards Brooklyn. A survey irregularity resulted in a kink in the alignment where the tunnels met the Brooklyn heading. One can only imagine the air bubbling in the river from the shields crossing beneath the west channel where the top of rock drops off and where the overburden was thin, but no further deaths were experienced. The SteinwayBelmont tunnels were the first East River crossings, ultimately becoming the No. 7 Flushing Line of the IRT, which continues to pose modern challenges due to tight clearances and lack of niches for safety of maintenance workers. BROOKLYN BATTERY TUNNEL INNOVATIONS Great success was achieved on the Brooklyn Battery tunnel, with reliance on geological consultants Charles Berkey and Thomas Fluhr, who were brought in from the New York Board of Water Supply. A geologic map of New York City was developed based on field inspection of outcrops and existing excavations, and it was checked by available borings. Their engineering geology guidelines (see following, from Sunborn, 1950, p 61–62) were considered by tunnel designers and

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

137

author email: david.cregger@aecom.com

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

137

Cregger

Figure 1. Locus map of East River crossings (base from Rossum, 1983).

provided to the contractors as information on which to base their bid: 1. Avoid location in or near a contact between two formations likely to be a zone of weakness underground. 2. Nearly all the fault lines and crush zones are also lines of weakness. 3. Behavior of the Inwood limestone is believed to be more uncertain and to be avoided. 4. Igneous units such as the Yonkers gneiss-granite and Ravenswood granodiorite are more uniform and consistent due to their simple crystalline character; Fordham Gneiss and Manhattan Schist may also be satisfactory but less so. 5. Along lines of contact and crush zones, superficial decay and disintegration extend much deeper than expected, where inflow of water and falling rock may occur during construction. 6. The deepest erosion, marking pre-glacial stream channels, follows these lines of weakness to elevations that cannot be determined without exploration. 138

Cregger

7. The overlying glacial drift is of variable quality but is porous and water bearing even where sufficient cover is expected.

The Brooklyn Battery tunnel pushed the East River precedent, increasing the diameter to 30 ft (9 m) and requiring excavation in stages. The rock invert was taken out as a bottom drift, followed by a top heading protected by a shield supported on rock haunches. Sanborn (1950) provided geologic maps of Brooklyn Battery tunnel. Twisting of the support steel due to sloping rock cover and excessive settlement above the shields were the only reported geotechnical problems; the vehicular tunnel did not encounter the limestone or dolomite and was holed through in 1917.

NORTH RIVER CROSSINGS

Plans for the new North River crossings were accompanied by two novel approaches based on that experience. To mitigate the danger of blow-outs into the river, clay blankets were to be deposited on the river bottom where gravels were present. Rock-fill dikes were placed underwater paralleling each side of the twin tunnel crossings, and clay obtained from dredging projects in New Jersey was puddled between the dikes 125 ft (38 m) apart, creating a cover of clay as much as 5 to 15 ft (1.5 to 4.5 m) thick. The second innovation was the use of tail grouting, in which roofing gravel was blown into the tail of the shield as it progressed to back-fill the voids and prevent settlement of the strata and distortion of the ground ring supports, which had been overcut 6 to 8 in. (15 to 20 cm) by the shield. The back end of the shield is known as “the tail,” where a smaller-diameter tunnel liner was installed, typically in wet conditions. Grouting provides savings in water treatment/disposal costs and prevents dewatering and consolidation of varved clays. In 1933, the 53rd Street tunnel opened as an underwater crossing, listed as 5,589 ft (1,703 m) in length. It was not until 1955, however, that the 60th Street tunnel to Queens opened. These small-diameter shield-driven tunnels were driven below the East River without incident, as far as is known. Crossing of the East River at Welfare (Roosevelt) Island was more predictable, geologically, because the width of the solid rock was greater, and the stretches of open water were shorter. Figure 3 illustrates the engineering geology as shown by Fluhr’s interpretation of the core borings. Overall, the geology is remarkably similar to the Steinway-Belmont tunnel, with carbonate rocks of Inwood Formation predicted in the west channel. The 63rd Street crossing was ultimately constructed as a sunken tube tunnel at a slightly higher elevation.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

Figure 1. Locus map of East River crossings (base from Rossum, 1983).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

Crossing the East River

Figure 2. Steinway tunnel profile and construction sections (Rogoff, 1960).

Figure 3. Fluhr’s geological profile of the East River (Fluhr, 1965).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

139

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

139

Cregger

QUEENS MIDTOWN MIXED FACE The Queens-Midtown vehicular tunnel was driven in 1936 through the worst ground in which any shield had ever been attempted, providing a classic case of mixed face tunneling (Sanborn, 1950; Fluhr, 1964). The 31ft-diameter (9-m-diameter) tunnel was advanced at a rate of 18 ft (5 m) per week on each of two headings; the southbound Queens tube is 6,272 ft (1,912 m) long, while the Manhattan-bound tube is 6,414 ft (1,955 m) long from portal to portal. The shield air-compressor system was designed for 50 psi (345 kPa) pressure. Too high of a pressure would push the moisture out of the Bull’s Liver silts, which then would dry out and unravel onto the workers, but too low of a pressure would not compensate for the elevation difference between the crown and the invert, resulting in seepage up through the bottom of the workings and running ground below the spring line. The shield was large enough for an upper work crew to work on a flying gangway, mudding up the earth face, breasting, and placing pea gravel and grout into the tail, while a lower crew was drilling and blasting rock, mucking, and placing the invert on which the shield would travel. Once, the air loss was so great in one of the headings that the other had to be shut down, and all pressure was diverted to save the face. Extensive diamond-drill core borings were taken to prepare rock-floor contour maps, profiles, and geologic sections of expected conditions for the construction. The bedrock formations were highly folded and faulted and metamorphosed, generally striking parallel to the course of the East River. Driving bottom drifts from the Queens shaft passed through Brooklyn Injection Gneiss for 1,000 ft (305 m) before hitting soft ground. Silts and sands and gravels of a lacustrine glacial deposit were penetrated and soon followed by stiff blue clay, possibly a pre-ice age Cretaceous deposit, then directly into a white clay residual soil on top of dolomitic limestone, back into open water, and then finally into the Fordham Gneiss, a solid rock projecting up into the middle of the East River. The white clay zone was highly variable vertically due to weathering and contributed to mixed face conditions, but it was laterally limited to the stratigraphic extent of the Inwood limestone. The blue clay, however, being a much younger deposit, was less predictable laterally. Driving from the west shaft presented its own set of challenges. Beginning from a high ridge along the Manhattan shore, the alignment had to be curved to maintain grade for vehicles, resulting in a longer tunnel. Passing from hard rock schist, foliated and striking northeast, the shields exited through ancient shoreline into artificial fill. Sometimes, fragments of coal the size of eggs were excavated, all in a jumbled-up matrix through which the air pressure could be lost. Passing 140

Cregger

through 150 ft (45 m) of rip-rap stone placed for erosion protection and construction of the bulkhead forming the Manhattan shoreline, each piece of stone had to be removed individually from the face and plugged with straw and mud to maintain the air pressure. Under special permit from the War Department, in control of navigation in the East River, extensive clay blankets were deposited over the rip-rap and bottom of the river, which had to be dredged out later. Special grout mixes of bentonite and sawdust were pumped into the tail of the shield to prevent water intrusion through the porous glacial gravels and rip-rap. As the shields progressed eastward, pinnacles of limestone and dolomite of the Inwood Formation came up through the invert, some mixed with decomposed rock, and never reaching the crown, which remained in soft ground. Reportedly, rock fill from the enlargement of Belmont Island was encountered. Finally, the middle “reef” of Fordham Gneiss was reached, and a full face of hard rock was driven without air pressure to meet the shield face from Queens. Hole-through in 1939 was well documented in Engineering News magazine (Tam, 1998) at the time, and the geology was recorded in a Geological Society of America case history volume (Fluhr, 1964) for those with further interest.

MAPPING MANHATTAN ISLAND Engineering geology maps (Baskerville, 1994) of Manhattan Island summarize the bedrock types and are overprinted with top of rock contours, geologic structure data, and historic shoreline locations. Viele’s (1865) map, a portion of which is shown in Figure 4, is an appropriate supplementary tool used by engineering geologists to evaluate landforms prior to urbanization. Note that Blackwell’s (Roosevelt) Island forms an anticlinal band of land down the center of the East River broken into small islands towards the south. Three southeast-trending diagonal river drainages, parallel to each other and colored dark green in Viele’s map, suggest major linear fracture zones in the rock, which are known to cut across Manhattan. A cove in the west bank of the river mapped at 46th Street at the largest of these drainage features is further evidence of poorquality rock conditions. The drainage trends bend to a north-south orientation along a valley parallel to the East River and then return to an east-west orientation intersecting the East River. This is likely an intersecting shear zone of bad rock, as further evidenced by the break in the line of islands south of Blackwell on the map. Springs encountered in the Steinway shafts, which are continuing to cause seepage in the Queens Midtown portal on the Manhattan side, can be related to these types of bedrock fracture zones. Projection of

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

Crossing the East River

Figure 4. Viele’s (1865) pre-settlement geologic map.

such linear trends into the East River can be helpful for planning future test borings for bridge structures, riverfront parkways, and future river crossings. LESSONS LEARNED In tunneling, keen understanding of the engineering geology of the alignment is fundamental to risk mitigation during construction. Complexities of the pressurized mined caverns used to create early East River tunnels are intriguing examples of how conditions impact the execution of a project. The issues are best understood in the context of an eroded section of several different geologic formations, with varying rock types, nearly vertical in orientation, as part of a com-

Crossing the East River

plex tightly folded sequence. Factors of great interest to geological engineers are the decomposed and highly crushed zones each crossing had to penetrate to achieve grade. Geomorphology of the bedrock surface is revealed by pre-urbanization topographic maps. The crushed and decomposed zones were investigated by extensive diamond core borings and confirmed by geologic mapping during construction of the East River tunnels. This case history information is extremely valuable for future generations and illustrates the legacy and vision of these engineering geologists.

ACKNOWLEDGMENTS

The author wishes to acknowledge Karen Armfield, AECOM associate vice president and project manager of the East Midtown Waterfront Esplanade, built over the 7 Line, for her review of the paper and help in collecting the figures.

REFERENCES

BASKERVILLE, C. A., 1994, Bedrock and Engineering Geologic Maps of New York County and Parts of Kings and Queens Counties, New York and Parts of Bergen and Hudson Counties, New Jersey: U.S. Geological Survey Miscellaneous Investigations Series Map I-2306 (Sheet 2 of 2) Engineering Sheet. FLUHR, T. W., 1964, Geology of the Queens Midtown Tunnel, New York. In Trask, P. D. (Editor), Engineering Geology Case Histories Number 1: Geological Society of America, New York, pp. 1–9. FLUHR, T. W, 1965, New York Transit Authority, Proposed East River Subway Tunnel 63rd Street Manhattan–41st Street Queens Profile and Geologic Section as Interpreted from Test Borings, May 10, 1965. ROGOFF, D., 1960, The Steinway Tunnels: Electric Railroads, Vol. 29, April 1960, http://www.nycsubway.org. ROSSUM, L., 1983, New York City Transportation Tunnels Case Histories and the Designer’s Approach: Tunneling and Underground Construction: Foundations and Soil Mechanics Group of the Metropolitan Section American Society of Civil Engineers, New York, NY, 84 p. SANBORN, J. F., 1950, Engineering Geology in the Design and Construction of Tunnels. In Application of Geology to Engineering Practice: Engineering Geology (Berkey) Volume, Sidney Paige, Ed. Geological Society of America, New York, pp. 45–81. TAM, J. G., 1998, The Influence of Geology on the Construction of the Queens Midtown Tunnel, New York: University of California, Berkeley, CA, http://www.ocf.berkeley.edu. VIELE, E. L., 1865, Sanitary & Topographical Map of the City and Island of New York: Prepared for the Council of Hygiene and Public Health of the Citizens Association, New York.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

141

Figure 4. Viele’s (1865) pre-settlement geologic map.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 137–141

141

Evaluation of Rockfall-Hazard Potential for Rockville, Utah, Following a 2013 Fatal Rockfall

CARL JACKLITCH

Gannett Fleming, Inc., Foster Plaza 8, Suite 400, 730 Holiday Drive, Pittsburgh, PA 15220

ABDUL SHAKOOR1

Department of Geology, Kent State University, Kent, OH 44242

WILLIAM R. LUND

Utah Geological Survey, 436 North Main Street, Cedar City, UT 84721

Key Terms: Rockfall, Shinarump Conglomerate, Chinle Formation, Upper Red Member, Moenkopi Formation, Hazard Map

close to the hazardous slopes. Other possible remedial measures include removing loose rock blocks, installing rock anchors, and using drapery mesh.

Key Terms: Rockfall, Shinarump Conglomerate, Chinle Formation, Upper Red Member, Moenkopi Formation, Hazard Map

ABSTRACT In December 2013, a rockfall in the town of Rockville, Utah, released an estimated 2,700 tons (2,450 tonnes) of rock from a 400-ft (122-m) high slope; the rock struck a house at the base of the slope, resulting in two fatalities. We performed detailed field and laboratory investigations to (1) identify the modes of failure and factors contributing to rockfalls along the east-west trending, south-facing slope where it passes through the town; (2) identify sections of the slope that pose the highest hazard for future property damage or injury; and (3) suggest potential remedial measures. Field investigations included mapping discontinuities, establishing stratigraphy, measuring slope geometry, and evaluating potential failure modes at four selected sites. Laboratory investigations included determining dry density, friction angle, and slake durability index of rock samples. Using the Dips software, we determined the principal joint sets and performed a kinematic analysis. The maximum rollout distances for rock blocks of various sizes were determined for each of the study sites using the RocFall software. Results of the kinematic analysis and field observations indicate that wedge, plane, and toppling failures are possible within the Shinarump Conglomerate Member of the Chinle Formation and the Upper Red Member of the Moenkopi Formation along the entire slope. Based on the results of the study, we developed a rockfall-hazard map that indicates that the western portion of the town faces the highest hazard from potential rockfalls. The most feasible future remedial measure is not to build

1 Corresponding

author email: ashakoor@kent.edu

close to the hazardous slopes. Other possible remedial measures include removing loose rock blocks, installing rock anchors, and using drapery mesh.

ABSTRACT INTRODUCTION About 4:54 p.m. on December 12, 2013, a rockfall occurred in the town of Rockville, Washington County, Utah (Figure 1). The rock mass, weighing approximately 2,700 tons (2,450 tonnes), fell from a 400-ft (122-m) high slope, destroying a house and resulting in the death of its two inhabitants (Dalrymple and Piper, 2013; Mabbutt, 2013; Sharp, 2013; and Lund et al., 2014). Figure 2 shows the house before and after the rockfall, the size of the rockfall, and the scattered debris from the rockfall. Rockville has a record of large, extremely hazardous rockfalls. At least six large rockfalls have been reported within the past 38 years (1976, October 2001, October 2002, spring of 2007, February 2010, and November 2010), with five of the six occurring in the past 16 years (Knudsen, 2011). Figure 3 shows a generalized rockfallhazard map (1:24,000 scale) of the Rockville slope prepared by the Utah Geological Survey (UGS) (Knudsen and Lund, 2013). The map also shows the locations and pathways of the six rockfall events mentioned above.

Geologic Setting of the Study Area Regionally, the study area lies within the western margin of the Colorado Plateau. The stratigraphic units present in the area include the well-indurated Upper Triassic Shinarump Conglomerate Member (cliffforming unit) of the Chinle Formation, which caps the less resistant Lower Triassic Moenkopi Formation

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

143

In December 2013, a rockfall in the town of Rockville, Utah, released an estimated 2,700 tons (2,450 tonnes) of rock from a 400-ft (122-m) high slope; the rock struck a house at the base of the slope, resulting in two fatalities. We performed detailed field and laboratory investigations to (1) identify the modes of failure and factors contributing to rockfalls along the east-west trending, south-facing slope where it passes through the town; (2) identify sections of the slope that pose the highest hazard for future property damage or injury; and (3) suggest potential remedial measures. Field investigations included mapping discontinuities, establishing stratigraphy, measuring slope geometry, and evaluating potential failure modes at four selected sites. Laboratory investigations included determining dry density, friction angle, and slake durability index of rock samples. Using the Dips software, we determined the principal joint sets and performed a kinematic analysis. The maximum rollout distances for rock blocks of various sizes were determined for each of the study sites using the RocFall software. Results of the kinematic analysis and field observations indicate that wedge, plane, and toppling failures are possible within the Shinarump Conglomerate Member of the Chinle Formation and the Upper Red Member of the Moenkopi Formation along the entire slope. Based on the results of the study, we developed a rockfall-hazard map that indicates that the western portion of the town faces the highest hazard from potential rockfalls. The most feasible future remedial measure is not to build

1 Corresponding

author email: ashakoor@kent.edu

INTRODUCTION About 4:54 p.m. on December 12, 2013, a rockfall occurred in the town of Rockville, Washington County, Utah (Figure 1). The rock mass, weighing approximately 2,700 tons (2,450 tonnes), fell from a 400-ft (122-m) high slope, destroying a house and resulting in the death of its two inhabitants (Dalrymple and Piper, 2013; Mabbutt, 2013; Sharp, 2013; and Lund et al., 2014). Figure 2 shows the house before and after the rockfall, the size of the rockfall, and the scattered debris from the rockfall. Rockville has a record of large, extremely hazardous rockfalls. At least six large rockfalls have been reported within the past 38 years (1976, October 2001, October 2002, spring of 2007, February 2010, and November 2010), with five of the six occurring in the past 16 years (Knudsen, 2011). Figure 3 shows a generalized rockfallhazard map (1:24,000 scale) of the Rockville slope prepared by the Utah Geological Survey (UGS) (Knudsen and Lund, 2013). The map also shows the locations and pathways of the six rockfall events mentioned above.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

143

Jacklitch, Shakoor, and Lund

Figure 1. (a) Location of Rockville in the State of Utah (Google Earth). (b) Location of Rockville in Washington County, Utah (Google Earth).

144

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

Evaluation of Rockfall Hazard

Figure 2. (a) The house before the December 12, 2013, fatal rockfall (Lund et al., 2014). (b) The house after the rockfall (Lund et al., 2014). (c) Debris from the rockfall (Mabbutt, 2013).

Evaluation of Rockfall Hazard

(slope-forming unit) (Figure 4). The Shinarump Conglomerate Member is 60–135 ft (18–41 m) thick, dips 1◦ to 2◦ to the east, and contains widely spaced vertical joints (Knudsen, 2011). It is a medium- to coarsegrained pebbly sandstone and lesser pebbly conglomerate. The Shinarump Conglomerate Member represents deposits from braided streams that filled paleo valleys (Graham, 2006). The Moenkopi Formation is a reddish-orange to reddish-brown siltstone, mudstone, and fine-grained sandstone. It generally thins to the northeast from about 400–600 ft (122–183 m) thick, west of St. George, to 200–280 ft (61–85 m) thick in the Zion National Park area (Utah Geological Survey, 2013). The Moenkopi Formation slopes are covered with talus deposits, derived from the Shinarump Conglomerate, which are more than 45 ft (14 m) thick (Knudsen, 2011) (Figure 4). The Upper Red Member of the Moenkopi Formation consists of shallow marine and mudflat deposits, whereas the Shnabkaib Member of the Moenkopi Formation represents a sabkha environment (Graham, 2006). An unconformity is present between the Chinle Formation and the Moenkopi Formation.

Current Methods for Identifying Rockfall Hazard in Utah

Between 2001 and 2006, the Utah Department of Transportation (UDOT) developed a Rockfall Hazard Rating System (RHRS) involving two phases. The first phase consisted of creating a rockfall-hazards inventory and a subjective evaluation of the rockfall hazard to roadways. This was based on a general assessment of all potentially hazardous slopes, using the Oregon Department of Transportation (ODOT) RHRS developed by Pierson (1991). In order to categorize the slopes with respect to hazard, the slopes were assigned ratings of A (moderate to high), B (low to moderate), or C (low to none), based on their current hazard potential and their history of rockfalls. The hazard potential was based on the estimated sizes of the rock blocks reaching the road, the quantity of material involved in a rockfall event, and the ditch effectiveness if present. The evaluation of historical rockfall activity included the frequency of rocks found on the highway, the quantity of material, the size of the material, and the frequency of ditch clean-out required (Pierson, 1991; Pack et al., 2006). Due to the difficulty of rating some of the slopes with a moderate hazard potential, the ratings “A−” and “B + ” were used (Pack et al., 2006). The second phase of the Utah RHRS was a detailed investigation of the slopes ranked as A, A−, or B + in Phase 1 of the study. The UDOT detailed hazard

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

145

Figure 2. (a) The house before the December 12, 2013, fatal rockfall (Lund et al., 2014). (b) The house after the rockfall (Lund et al., 2014). (c) Debris from the rockfall (Mabbutt, 2013).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

145

Jacklitch, Shakoor, and Lund

Figure 3. Excerpt from a 1:24,000-scale, regional rockfall-hazard map showing rockfall hazard in the town of Rockville (Knudsen and Lund, 2013). Notice the locations and pathways of previous major rockfalls.

study was developed by applying three previously established rockfall-hazard rating systems: the original ODOT RHRS (Pierson, 1991), the ODOT Unstable Slope Management System (ODOT, 2001; Pack et al., 2006), and the New York Rock Slope Rating Procedure (New York State Department of Transportation, 1996). The UGS used an empirical modeling approach, referred to as the rockfall shadow zone, to define the maximum rollout distance of a rock block to develop the

146

rockfall-hazard zones shown on its regional rockfallhazard map (Figure 3) (Knudsen and Lund, 2013). A shadow zone is the area extending away from the base of a cliff (rockfall source) to the most distant rock blocks that are identified as products of rockfall events originating from the cliff (Turner and Duffy, 2012). Figure 5 illustrates the concept of shadow angle or shadow zone for the Rockville slope. Any properties located within the shadow zone face the risk of being damaged by a rockfall event.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

146

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

Evaluation of Rockfall Hazard

Figure 4. An overview of the stratigraphy near the fatal rockfall site showing the Shinarump Conglomerate Member of the Chinle Formation, the Upper Red Member of the Moenkopi Formation, and the Shnabkaib Member of the Moenkopi Formation.

Objectives The objectives of this study were to (1) identify the various modes of failure affecting the Rockville slope, (2) identify the factors contributing to rockfall activity in Rockville, (3) develop a rockfall-hazard-potential map for the Rockville slope based on rockfall simulations, and (4) suggest possible remedial measures.

Evaluation of Rockfall Hazard

RESEARCH METHODS

Site Selection

We selected four primary sites along the Rockville slope for detailed investigations and several supplementary sites, in the vicinity of primary sites, to collect additional data. Two primary criteria were used for site selection, the first being accessibility of the steep slopes (whether the slopes were safe to climb) and the second being obtaining permission from the landowners for data collection. While applying these criteria, we ensured that there was no bias and that the selected sites were representative of the overall stratigraphy and the associated rockfall problems. Figure 6 shows the locations of all sites; Figure 7 provides an overview of the four primary sites.

Field Investigations We conducted field investigations to evaluate slope geometry, establish slope stratigraphy, collect discontinuity data, evaluate modes of failure, and obtain samples for laboratory testing. We used the discontinuity data to determine the principal joint sets, which were required to evaluate the kinematically possible modes of failure affecting the slopes at the four study sites. Slope geometry involves slope orientation (strike and dip), slope height, and slope length. We used a Brunton compass to measure slope orientation, a topographic map to measure slope height, and pace to measure

Figure 5. Illustration of the shadow angle for the Rockville slope (Lund et al., 2014).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

Figure 5. Illustration of the shadow angle for the Rockville slope (Lund et al., 2014).

147

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

147

Jacklitch, Shakoor, and Lund

Figure 6. Location of the four primary sites and the supplemental data sites on a USGS topographic map overlay (Google Earth). Note: 1 mi = 1.6 km.

slope length for the four sites. Because of the irregular nature of the slopes, a minimum of three locations at each site were chosen to measure slope orientation. For slope height, we obtained KMZ files of U.S. Geological Survey (USGS) topographic maps for each site and set them as overlays in Google Earth. This made it easier to reference the topographic maps to the site locations. We established slope stratigraphy in two steps: the first consisted of a general description of what we could see from the toe of the slope, and the second involved a more detailed analysis of the rock types as we traversed the slopes. The detailed analysis included identifying the composition of the resistant units and locating the contact between the Upper Red Member and the Shnabkaib Member of the Moenkopi Formation, as it is covered by colluvium at most places. We used available literature (Lund et al., 2014) for information about thicknesses of various stratigraphic units in the Rockville area. For discontinuity mapping, we used the window mapping and random mapping methods (Wyllie and

148

Mah, 2004) to record the following aspects of discontinuities: geometry, continuity (extent or persistence), spacing, aperture, in-filling material (if present), surface irregularities, and presence or absence of water. The geometry of a discontinuity involves its orientation in space and its position relative to the slope face (West, 1995). We used a Brunton compass to measure discontinuity orientations. The extent and spacing of the discontinuities define the sizes of the rock blocks that may fall. We estimated continuity visually. Spacing was measured using a line mapping method (Wyllie and Mah, 2004). We stretched a 100-ft (30.5-m) tape parallel to the slope face, and every discontinuity that intersected the tape was counted to determine average spacing. The aperture of a discontinuity is the distance between two adjoining rock blocks. In-filling material is any material (soil, rock fragments, plants, etc.) that fills the joint space (aperture) between two rock blocks. Surface irregularities on discontinuity surfaces contribute to frictional resistance along the discontinuities. We categorized surface roughness as smooth or moderately rough. No water was observed on the slopes, and no

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

148

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

Evaluation of Rockfall Hazard

Figure 7. (a) Site 1, looking northwest across State Route 9. (b) Site 2, looking east, near the contact of the Shnabkaib Member and Upper Red Member of the Moenkopi Formation. (c) Site 3 looking north. (d) Site 4, looking west.

precipitation occurred during the two-and-a-half-week field investigation. The modes of failure affecting the slopes in Rockville (rockfalls, plane failures, wedge failures, toppling failures) were evaluated using both field observations and the results of kinematic analysis. We used two sampling methods: random sampling and block sampling. Small hand samples of various units from the Moenkopi Formation and lenses of material between the Moenkopi Formation and the Shinarump Conglomerate Member were collected for

slake durability testing. We collected block samples of the Shinarump Conglomerate Member and Moenkopi Formation to obtain a minimum of four NX-size (2.125 in. [5.4 cm]) cores of each rock type by coring in the laboratory. Laboratory Investigations We performed laboratory investigations to determine density (mass/volume), friction angle, and second-cycle slake durability index values for various

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

149

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

149

Jacklitch, Shakoor, and Lund

rock units making up the slopes in the study area. We used density and friction angle for rockfall simulations and kinematic analysis, respectively, and slake durability index to evaluate the potential for undercutting of the resistant rock units (Shinarump Conglomerate Member) by the less resistant rock units (Moenkopi Formation). We determined density on six core samples of the Shinarump Conglomerate Member (from two different sites) and three core samples of the Upper Red Member. The core samples were oven dried for 24 hours at 105◦ C, cooled to room temperature, and weighed. The volume of each core was determined using the average diameter and length, based on five measurements of each. We calculated density by dividing the mass of the core by its volume. We used the Stimpson (1981) method to determine the basic angle of friction. Three cores of the Upper Red Member and six cores of the Shinarump Conglomerate Member, with each core having a diameter of 2 in. (56 mm) and a length-to-diameter ratio of 2.5, were used for the test. We stacked three cores from each rock unit on a tilting table such that two cores were fixed in place side by side and the third core was placed on top, free to slide. The table was tilted until the top core slid over the lower two cores. We measured the tilt angle of the table using a Brunton compass. The three cores were rotated to obtain a fresh surface and the test was repeated three times to obtain the average tilt angle. We used the tilt angle in the following equation to calculate the basic friction angle: = tan−1 (1.155 tan ), where is the internal angle of friction and is the tilt angle (Stimpson, 1981). We used method D-4644 of the American Society for Testing and Materials (2010) to determine the secondcycle slake durability index of the Moenkopi Formation. Six samples collected from two of the sites (four samples from Site 1 and two samples from Site 2) were used for the test. Data Analysis The data analyses for this study consisted of (1) stereonet analysis, (2) kinematic analysis, and (3) rockfall analysis. The purpose of the stereonet analysis was to determine the principal joint sets. The kinematic analysis was used to evaluate the potential for planar failures, wedge failures, and toppling failures. We conducted the rockfall analysis to determine the trajectories and the rollout distances of rock blocks of specific sizes. We used Dips software (Rocscience, 2014) for the stereonet analysis. The software plots discontinuity orientation data as poles on an equal area stereonet. A pole density of 6 percent or higher was used to define the principal joint sets for each site. 150

We used the principal joints obtained from the stereonet analysis, the slope angles measured in the field, and the basic friction angle determined by the Stimpson (1981) method for kinematic analysis. The Dips software (Rocscience, 2014) was also used to perform the kinematic analysis. Because of the relatively smooth to moderately rough nature of the discontinuities, we decided to be conservative and used the basic friction angle for a worst-case-scenario analysis. We conducted the rockfall analysis using the RocFall software (Rocscience, 2015). Slope profiles were developed for each of the four sites using USGS topographic maps and aerial imagery from Google Earth and Bing Maps. We entered the slope profiles into RocFall and selected the parameters for the slope material from the suggested values in the RocFall software manual (Rocscience, 2015). The parameters included the coefficient of normal restitution, the coefficient of tangential restitution, the dynamic friction coefficient, and the rolling resistance. Coefficient of restitution refers to the elasticity of an object. A perfectly elastic material will have a coefficient of restitution of 1, and an object striking this material would rebound without losing any velocity. A perfectly inelastic material will have a coefficient of restitution of 0, and an object striking this material will not rebound. The coefficient of restitution is broken into normal and tangential components where the tangential coefficient is the ratio of the outgoing velocity to the incoming velocity tangent to the striking surface and the normal coefficient is the same ratio but normal to the striking surface (Rocscience, 2015). The dynamic friction coefficient depends on the friction angle of the material and is calculated by taking the tangent of the friction angle. The rolling resistance coefficient represents the energy lost to factors other than dynamic friction, such as plastic deformation, hysteresis, and slippage of contact surfaces (Rocscience, 2015). In order to calibrate the parameters to ensure accuracy of the model, the parameters were tested with the known block shape, density, mass, and rollout distance of the largest block at the fatal rockfall site. The block shape, density, and mass are the three parameters that can be altered for the rollout distance analysis. As the density remains constant for a given lithology, we altered the block shape and mass at each site to find the combination that yielded the greatest rollout distance. In RocFall, 22 shape options are available for the general shape of the rock block. Our field observations indicated that rockfall block sizes and shapes were highly variable. Therefore, in order to accommodate maximum variability, we let the computer program select all 22 shape options. With all 22 shapes selected, 11,000 rock blocks were selected for each run. This provided 500 blocks of each of the 22 block shapes. With

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

Jacklitch, Shakoor, and Lund

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

Site Average slope angle Slope height Slope length

Evaluation of Rockfall Hazard

Table 1. Geometric data for the four primary data sites.

40◦ 260 ft (79 m) 1,500 ft (457 m)

40◦ 430 ft (131 m) 450 ft (137 m)

35◦ 290 ft (88 m) 900 ft (275 m)

35◦ 276 ft (84 m) 600 ft (138 m)

the block shape and density fixed, the mass of the block was manually altered for each run until the maximum rollout distance was determined. This process was repeated for all four sites. RESULTS Results of Field Investigations Table 1 summarizes the geometrical data for the four sites. The average slope angle ranges from 35◦ (Sites 3 and 4) to 40◦ (Sites 1 and 2), the slope height ranges from 260 ft (79 m) at Site 1 to 430 ft (131 m) at Site 2, and the slope length ranges from 450 ft (137 m) at Site 2 to 1,500 ft (457 m) at Site 1. The three stratigraphic units present in the Rockville slope are the Shinarump Conglomerate Member of the Chinle Formation and the Upper Red and Shnabkaib members of the Moenkopi Formation. Figure 8 is a stratigraphic column showing the maximum exposure of these three units in the Rockville area. The Shinarump Conglomerate Member is a yellow to tan, medium- to coarse-grained, massive, cross-bedded sandstone with lenses of quartz-pebble conglomerate and a basal conglomerate. The sandstone can be classified as a sublitharenite, using Folk’s (1974) classification or a lithic wacke using Blatt et al.’s (1996) classification. The rock contains calcareous cement and, while fairly well cemented in most locations, is friable locally. The Upper Red Member is a red mudstone with interbedded siltstone and fine-grained sandstone. The mudstone contains small lenses and seams of gypsum. The siltstone and sandstone units contain muscovite mica and ripple marks. The sandstone classifies as a lithic arkose according to Folk’s (1974) classification or a feldspathic wacke according to Blatt et al.’s (1996) classification. The Shnabkaib Member is a red to pink mudstone with interbedded gypsum layers and lenses. Between the Shinarump Conglomerate Member and the Upper Red Member are lenses of a cream to maroon colored silty mudstone. These lenses were observed at all four sites; however, the distribution of the lenses was variable. Table 2 summarizes the range and average values for the various aspects of discontinuities. The most prevalent discontinuity in the Rockville area consists of valley stress relief joints. While there is some variation among the discontinuities at the four sites,

Site

generally the orientation, continuity, spacing, aperture, and roughness are fairly consistent. Results of Laboratory Investigations Laboratory tests indicated that the Shinarump Conglomerate Member has an average density of 141.6 lb/ft3 (2.27 g/cm3 ) and that the Upper Red Member has an average density of 164.7 lb/ft3 (2.64 g/cm3 ). The lower density of the Shinarump Conglomerate Member reflects its relatively porous nature. The average friction angle values for the Shinarump Conglomerate Member and the Upper Red Member are 45.8◦ and 39◦ , respectively. Six samples were tested for slake durability index and classified according to the International Society for Rock Mechanics (1981) classification system based on the second-cycle slake durability index (Id2 ). Two samples, one from the Shnabkaib Member (Site 2) and the other from the lower portion of the Upper Red Member (Site 1), have very low durability (Id2 = 25 and 9 percent, respectively). The three samples from the resistant units within the Upper Red Member all have a high to very high durability (Id2 = 94, 96, and 97 percent). One sample from a silty mudstone lens, present between the Upper Red Member and the Shinarump Conglomerate Member, has medium durability (Id2 = 69 percent). The large difference in slake durability index values between the upper resistant units (94 to 97 percent) and the lower weaker units (9 to 25 percent) suggests the potential for significant undercutting. While undercutting was observed at all four sites, it did not appear to be as significant as the difference in Id2 values would suggest. This could be a result of low precipitation in the region. Results of Stereonet Analysis We used stereonet analysis to determine the principal joint sets for use in the kinematic analysis. The discontinuity orientation data, collected from the four study sites and the supplemental sites were imported into the Dips software (RocScience, 2014) in the form of poles. We selected pole concentrations >6 percent, as indicated by contouring, to represent the principal joint sets. Because of the inaccessible nature of certain areas of the study sites, sufficient discontinuity orientation data could not be collected from a single site. Therefore, we combined orientation data from all sites to generate contour plots for the Shinarump Conglomerate

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

151

Average slope angle Slope height Slope length

40◦ 260 ft (79 m) 1,500 ft (457 m)

40◦ 430 ft (131 m) 450 ft (137 m)

35◦ 290 ft (88 m) 900 ft (275 m)

35◦ 276 ft (84 m) 600 ft (138 m)

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

151

Jacklitch, Shakoor, and Lund

Figure 8. Stratigraphic section showing the rock units observed at the study sites. The only unit in the study area that is fully exposed is the Upper Red Member of the Moenkopi Formation. The Shnabkaib Member extends into the subsurface, and an unknown portion of the Shinarump Conglomerate Member of the Chinle Formation has eroded away.

152

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

Evaluation of Rockfall Hazard

Figure 9. Contour plots of discontinuity-pole data, representing discontinuity orientations within (a) the Shinarump Conglomerate Member of the Chinle Formation and (b) the Moenkopi Formation.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

153

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

153

Parameter Continuity Range Average Spacing Range Average Roughness Range Average Aperture Range Average Average Infilling Range Average Water

Jacklitch, Shakoor, and Lund

Table 2. Summary of discontinuity data.

Shinarump Conglomerate Member

Upper Red Member

Parameter

4 in.–150 ft (10 cm–46 m) 17 ft (5 m)

2.5 in.–50 ft (6 cm–15 m) 11 ft (3 m)

Not available 6 ft (1.8 m)

Not available 5 ft (1.7 m)

Smooth to very rough with undulations Moderately smooth to moderately rough with undulations

Very smooth to very rough with undulations Smooth to moderately rough with undulations

0–7 in. (0–18 cm) 0.1 in. (0.25 cm) 17 ft (5 m)

0–9 in. (0–23 cm) 0.2 in. (0.5 cm) 11 ft (3 m)

No infilling to colluvial soil and small vegetation No infilling None

No infilling to clayey soil and small vegetation No infilling None

Member and Moenkopi Formation (Figure 9). Table 3 lists the orientations of bedding and the principal joints as dips and dip directions.

examples of various modes of failure from the study area.

Results of Kinematic Analysis

We used RocFall software from Rocscience (2015) to determine rockfall trajectories and the maximum rollout distances at each of the four sites. It should be noted that Shinarump Conglomerate Member is not the only source for rockfalls; they occur wherever weaker layers on Rockville slope undercut the stronger layers. Table 4 summarizes the rockfall simulation results. Several block sizes had significant rollout distances at each site. The farthest rollout distance at each site was as follows: a 90-lb (41-kg) block traveled 522 ft (159 m) at Site 1, a 50-ton (45-tonne) block traveled 830 ft (253 m) at Site 2, a 35-ton (32-tonne) block traveled 548 ft (167 m) at Site 3, and a 600-lb (272-kg) block traveled 505 ft (154 m) at Site 4. Figures 11 through 14 show the rockfall simulations for the blocks that traveled the farthest at each site.

We performed the kinematic analysis using Dips software (Rocscience, 2014) to evaluate the potential for different modes of failure at each of the four sites. The principal joint sets, determined from the stereonet analysis, and the friction angle, determined from the Stimpson (1981) method, were input parameters for the analysis. The kinematic analysis showed the potential for planar failures, wedge failures, and toppling failures at Sites 1, 3, and 4. Site 2 has the potential for only toppling and wedge failures. While kinematic analysis did not indicate a potential for planar failures at Site 2, the possibility for a planar failure at Site 2 still exists. The kinematic analysis does not consider differential weathering that can cause steeply dipping discontinuities to daylight in the undercut portion of the slope face (Figure 4), potentially causing a failure (Admassu et al., 2012). Rockfalls, the most dominant mode of failure in the Rockville area, are not subject to kinematic analysis. However, field observations indicated that a vast majority of the rockfalls initiate as plane, wedge, or toppling failures higher up in the slopes and then descend as rockfalls. Figure 10 shows

Results of Rockfall Simulations Analysis

ROCKFALL-HAZARD EVALUATION Factors Contributing to Rockfalls and the Associated Hazard Important factors contributing to rockfall hazard in Rockville include unfavorable orientation of disconti-

Upper Red Member

4 in.–150 ft (10 cm–46 m) 17 ft (5 m)

2.5 in.–50 ft (6 cm–15 m) 11 ft (3 m)

Not available 6 ft (1.8 m)

Not available 5 ft (1.7 m)

Smooth to very rough with undulations Moderately smooth to moderately rough with undulations

Very smooth to very rough with undulations Smooth to moderately rough with undulations

0–7 in. (0–18 cm) 0.1 in. (0.25 cm) 17 ft (5 m)

0–9 in. (0–23 cm) 0.2 in. (0.5 cm) 11 ft (3 m)

No infilling to colluvial soil and small vegetation No infilling None

No infilling to clayey soil and small vegetation No infilling None

Member and Moenkopi Formation (Figure 9). Table 3 lists the orientations of bedding and the principal joints as dips and dip directions.

examples of various modes of failure from the study area.

Results of Kinematic Analysis

Results of Rockfall Simulations Analysis

ROCKFALL-HAZARD EVALUATION Factors Contributing to Rockfalls and the Associated Hazard Important factors contributing to rockfall hazard in Rockville include unfavorable orientation of disconti-

Table 3. Orientations of bedding and the principal joint sets.

Principal Discontinuity Sets

Unit

Bedding

Joint Set 1

Joint Set 2

Joint Set 3

Shinarump Conglomerate Member Moenkopi Formation

04, 086 02, 074

84, 197 84, 228

87, 158 88, 314

87, 229 —

154

Continuity Range Average Spacing Range Average Roughness Range Average Aperture Range Average Average Infilling Range Average Water

Shinarump Conglomerate Member

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

Unit

Bedding

Joint Set 1

Joint Set 2

Joint Set 3

Shinarump Conglomerate Member Moenkopi Formation

04, 086 02, 074

84, 197 84, 228

87, 158 88, 314

87, 229 —

154

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

Evaluation of Rockfall Hazard

Figure 10. Examples of (a) plane failures and (b) wedge failures and rockfalls from the Rockville area. Part (b) of the figure demonstrates how the wedge failures occurring in the uppermost layer can cover the slope below as rockfalls.

nuities, differential weathering, regional climate, presence of gullies, and proximity of homes to slopes. Among the aspects of discontinuities recorded in the field, orientation is the most important with regard

to slope failure. If the orientation of discontinuities present at a given site kinematically precludes a failure, other aspects of the discontinuities are irrelevant. Continuity and spacing are also important aspects of

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

155

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

155

156

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

156

505 (154) 600 lbs (272 kg) 500 (152) 900 lbs (408 kg) 752 (229) 900 lbs (408 kg) 800 lbs (363 kg) Block size 4

486 (148)

502 (153) 100 tons (91 tonnes) 548 (167) 35 tons (32 tonnes) 50 tons (45 tonnes) 45 tons (41 tonnes) Block size 3

517 (158)

830 (253)

475 (145) 100 lbs (45 kg) 501 (152) 100 lbs (45 kg) 750 (229) 150 lbs (68 kg) 90 lbs (41 kg) Block size 2

522 (159)

503 (153) 610 tons (553 tonnes) 500 (152) 590 tons (535 tonnes) 610 tons (553 tonnes) 610 tons (553 tonnes) Block size 1

460 (140)

808 (246)

425 (130) 490 tons (445 tonnes) 130 (40) 490 tons (445 tonnes) 740 (226) 490 tons (445 tonnes) 490 tons (445 tonnes) Second-largest block at fatal site

418 (127)

422 (129) 520 tons (472 tonnes) 448 (137) 520 tons (472 tonnes) 661 (201) 520 tons (472 tonnes) 520 tons (472 tonnes) Largest block at fatal site

408 (124)

424 (129) 2,700 tons (2,450 tonnes) 520 (159) 2,700 tons (2,450 tonnes) 646 (197) 2,700 tons (2,450 tonnes) 407 (124) 2,700 tons (2,450 tonnes) Total material at fatal site

Site 4 Distance, ft (m) Mass Site 3 Distance, ft (m) Mass Site 2 Distance, ft (m) Mass Site 1 Distance, ft (m) Mass Parameter

discontinuities relevant to rockfall hazard because they define the size of the rock block that may detach and fall. Most discontinuities extended through the full thickness of the Shinarump Conglomerate Member or the Upper Red Member. Although the average spacing of the discontinuities is approximately 6 ft (1.8 m), for some blocks spacing between discontinuities is as much as 150 ft (46 m). At many locations, we observed spacing of 15 ft (5 m) to 30 ft (9 m). Based on visual observations of discontinuity exposures, the majority of the discontinuities at the study sites have little or no aperture; however, some have nearly a foot (0.3 m) of separation between adjacent rock blocks. The presence of aperture contributes to the potential for failure. Most discontinuities at the study sites are moderately rough to smooth. The relatively low roughness of the discontinuities provides less resistance to failure. Slake durability tests show that there is a significant potential for differential weathering within the Moenkopi Formation, resulting in undercutting of the Upper Red Member by the underlying slope-forming unit (Figure 15). The rock units within the Moenkopi Formation that exhibit the lowest Id2 values and result in undercutting are those that contain lenses and seams of gypsum and are also more clayey in nature. The amount of undercutting of the Shinarump Conglomerate Member by the Moenkopi Formation (Figure 15) is variable across the study area. The primary factors that contribute to the variability is whether the portion of the Moenkopi Formation in contact with the Shinarump Conglomerate Member consists of more resistant or less resistant rock, that is, whether the cream to maroon colored silty mudstone lenses are present along the contact (Figure 16) or whether the contact between the Shinarump Conglomerate Member and the Moenkopi Formation is covered with colluvium. The mudstone lenses, because of their low durability, promote undercutting. Colluvium covering the contact between the Shinarump Conglomerate Member and the Moenkopi Formation prevents undercutting of the Shinarump Conglomerate Member by the Moenkopi Formation. The climate of the Rockville area (sporadic but intense rainfall and freeze–thaw cycles) is a contributory factor to rockfall activity. Climate may have had a significant effect on initiating the fatal rockfall on December 12, 2013, and some of the other major recorded rockfalls. During the end of November and the beginning of December 2013, a total of 2.6 in. (6.6 cm) of precipitation was recorded at the station GHCND: USC00429717 near the Springdale entrance to Zion National Park, Utah (Menne et al., 2012). During the same time period, temperatures fell below freezing at night and rose above freezing during the day. The lowest temperature (−4◦ F/−20◦ C) for the year (and

Jacklitch, Shakoor, and Lund Table 4. Results of rockfall simulation analysis. The block sizes and the corresponding rollout distances are listed for each site. The top three rows show results for various block sizes from the fatal rockfall site. The next four rows show other block sizes that resulted in significant rollout distances. Note: The size is expressed in terms of the mass of the rock block; in addition to the mass of the block, the rollout distance depends on the block shape and slope characteristics.

505 (154) 600 lbs (272 kg) 500 (152) 900 lbs (408 kg) 752 (229) 900 lbs (408 kg) 800 lbs (363 kg) Block size 4

486 (148)

475 (145)

502 (153) 100 tons (91 tonnes)

100 lbs (45 kg) 501 (152)

548 (167) 35 tons (32 tonnes)

100 lbs (45 kg) 750 (229)

830 (253)

150 lbs (68 kg)

50 tons (45 tonnes) 45 tons (41 tonnes) Block size 3

522 (159) 90 lbs (41 kg) Block size 2

517 (158)

503 (153) 610 tons (553 tonnes) 500 (152) 590 tons (535 tonnes) 610 tons (553 tonnes) 610 tons (553 tonnes) Block size 1

460 (140)

808 (246)

425 (130) 490 tons (445 tonnes) 130 (40) 490 tons (445 tonnes) 740 (226) 490 tons (445 tonnes) 490 tons (445 tonnes) Second-largest block at fatal site

418 (127)

422 (129) 520 tons (472 tonnes) 448 (137) 520 tons (472 tonnes) 661 (201) 520 tons (472 tonnes) 520 tons (472 tonnes) Largest block at fatal site

408 (124)

424 (129) 2,700 tons (2,450 tonnes) 520 (159) 2,700 tons (2,450 tonnes) 646 (197) 2,700 tons (2,450 tonnes) 407 (124) 2,700 tons (2,450 tonnes) Total material at fatal site

Site 4 Distance, ft (m) Mass Site 3 Distance, ft (m) Mass Site 2 Distance, ft (m) Mass Site 1 Distance, ft (m) Mass Parameter

Table 4. Results of rockfall simulation analysis. The block sizes and the corresponding rollout distances are listed for each site. The top three rows show results for various block sizes from the fatal rockfall site. The next four rows show other block sizes that resulted in significant rollout distances. Note: The size is expressed in terms of the mass of the rock block; in addition to the mass of the block, the rollout distance depends on the block shape and slope characteristics.

Jacklitch, Shakoor, and Lund

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

Evaluation of Rockfall Hazard

Table 5. Rockfall-hazard rating results for the four study sites using the modified ODOT II RHRS.

Factor Failure type/hazard House impact Annual frequency Potential for property damage/injury or fatality Accident history Sum Rating (%)

Site 1

Site 2

Site 3

Site 4

9 3 25 30 3 70 14

100 81 25 100 100 406 81

100 27 25 100 9 261 52

27 27 25 100 9 188 38

Factor Failure type/hazard House impact Annual frequency Potential for property damage/injury or fatality Accident history Sum Rating (%)

Site 1

Site 2

Site 3

Site 4

9 3 25 30 3 70 14

100 81 25 100 100 406 81

100 27 25 100 9 261 52

27 27 25 100 9 188 38

Figure 11. Rockfall simulation plot for Site 1 using RocFall software (Rocscience, 2015). The rock block weighed 90 lbs (41 kg). After 11,000 simulated rockfalls with variable random block sizes and initiation points along the face of the Shinarump Conglomerate Member, the farthest rollout distance from the point of release was 522 ft (159 m). This distance represents the farthest distance a rock block can travel at Site 1 for the various block sizes tested.

the lowest temperature in a 10-year span) occurred 3 days prior to the incident (Menne et al., 2012), yet the daily highs were still in the 50◦ F (10◦ C) to 60◦ F (15.5◦ C) range. These freeze-thaw cycles likely helped detach the rock that fell. The temperature data were obtained from the same station as the precipitation data. The presence of gullies on the slope in the study area tends to enhance rockfall hazard by channelizing rockfall debris (Figure 17). A structure built near the mouth of a gully has a higher risk of damage from rockfalls. Some gullies are quite deep (>6 ft [1.8 m]) and are more prevalent within the western slope, becoming less prominent toward the east. This is probably because the Shnabkaib Member (weaker unit prone to gully-

ing) is below the ground surface in the eastern portion of the study area. Proximity of homes to slope toes or the mouths of the gullies is the most important factor contributing to rockfall hazard (Figure 4). Modified RHRS for Rockville We used a modified ODOT RHRS (ODOT II) to quantify rockfall hazard at the four study sites. The modifications included the following: “Roadway” was changed to “House” because houses occupy the space between the slope toe and the roadway through the town, “Annual Maintenance Frequency” was changed to “Annual Frequency,” and “Average Daily

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

157

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

157

Jacklitch, Shakoor, and Lund

Figure 12. Rockfall simulation plot for Site 2 using RocFall software (Rocscience, 2015). The rock block weighed 50 tons (45 tonnes). After 11,000 simulated rockfalls with variable random block sizes and initiation points along the face of the Shinarump Conglomerate Member, the farthest rollout distance from the point of release was 830 ft (253 m). This distance represents the farthest distance a rock block can travel at Site 2 for the various block sizes tested.

Figure 13. Rockfall simulation plot for Site 3 using RocFall software (Rocscience, 2015). The rock block weighed 35 tons (32 tonnes). After 11,000 simulated rockfalls with variable random block sizes and initiation points along the face of the Shinarump Conglomerate Member, the farthest rollout distance from the point of release was 548 ft (167 m). This distance represents the farthest distance a rock block can travel at Site 3 for the various block sizes tested.

158

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

Evaluation of Rockfall Hazard

Figure 14. Rockfall simulation plot for Site 4 using RocFall software (Rocscience, 2015). The rock block weighed 600 lbs (272 kg). After 11,000 simulated rockfalls with variable random block sizes and initiation points along the face of the Shinarump Conglomerate Member, the farthest rollout distance from the point of release was 505 ft (154 m). This distance represents the farthest distance a rock block can travel at Site 4 for the various block sizes tested.

Traffic” was changed to “Potential for Property Damage/Injury or Fatality.” The “Potential for Property Damage/Injury or Fatality” evaluates the possible damage to facilities that are within the rollout distance as well as the possibility that an injury or fatality may occur. The same scoring system was used as for the ODOT II RHRS, with low values representing a low hazard potential and high numbers representing a high hazard potential. High hazard ranges from 100 to 66 percent, moderate hazard ranges from 65 to 33 percent, and low hazard ranges from 32 to 0 percent. The percentages represent the score for a slope divided by a total possible score of 500. Based on the qualitative and quantitative evaluations of rockfall hazard discussed above and the maximum rollout distance determined using the RocFall software, we created a rockfall-hazard map (Figure 18). The map uses three categories of hazard—high, medium, and low—indicated by red, yellow, and green colors, respectively. The western portion of the study area has the highest hazard due to the higher slopes, larger exposure of the Shnabkaib Member (the weaker rock unit), higher number of deep gullies, and the higher density of residences within the rollout zone. The middle to mideast-

ern portion of the study area has a moderate hazard, as the effect of all previously mentioned factors decreases eastward. Finally, the easternmost portion of the study area exhibits a low rockfall hazard. This is due primarily to the lack of residences in the area. It is important to state that these hazard ratings do not indicate the potential for rockfall. Rather, they indicate the degree to which rockfall activity poses a hazard (risk) and can change if additional structures are built in rockfall runout zones. Figure 18 compares the rockfall-hazard map developed in this study, using the RocFall software, with the UGS rockfall-hazard map based on the shadow zone method (Knudsen and Lund, 2013). The difference in the rollout distances by the two methods is shown in blue. In the western part of the study area, the shadow zone method results in a rollout distance that is larger than the rollout distance indicated by the RocFall software; that is, the shadow zone method is more conservative. However, the two methods appear to provide fairly similar results within the eastern portion of the study area. Also, it is important to note here that the UGS rockfall-hazard map is part of a regionallevel study as opposed to our more detailed site-specific study.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

159

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

159

Jacklitch, Shakoor, and Lund

Figure 15. (a) Undercutting of the cliff-forming resistant rock unit within the Upper Red Member of the Moenkopi Formation by the less resistant slope-forming rock unit of the same formation at Site 1. (b) Undercutting of the Shinarump Conglomerate Member by the Moenkopi Formation near the central park area between Site 3 and Site 4, resulting in potentially unstable blocks of the Shinarump Member. Central Park is approximately 0.5 mi (0.8 km) east of the fatal rockfall site.

REMEDIAL MEASURES Possible remedial measures for the Rockville slope are limited primarily because of the limited economic 160

resources of the town of Rockville. For any remediation plan to be practical, we must consider whether it is logistically and economically feasible. Removing rock overhangs is not practical because of the height of the

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

REMEDIAL MEASURES Possible remedial measures for the Rockville slope are limited primarily because of the limited economic 160

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

Evaluation of Rockfall Hazard

Figure 16. One of the lenses of red to gray mudstone that separate the Shinarump Conglomerate Member from the Moenkopi Formation at several locations.

Evaluation of Rockfall Hazard

slope and the proximity of homes at the toe of the slope. Dislodged rocks could easily roll downslope and strike existing homes. Barriers and catchment ditches are not practical because of the large sizes of potential rockfall blocks and the limited space between the slope toe and residential structures. Drapery mesh, shotcrete, riprap, and horizontal drains are cost prohibitive for the limited economic resources of the town. Based on the proximity of structures to the slopes, inaccessibility of slopes needing remediation, and the limited economic resources of the town, the best mitigation option is to avoid hazardous slopes. Avoiding hazardous slopes entails moving those homes that are currently within the hazard zone (i.e. rockfall rollout distance) and restricting future development within the hazard zone. If avoidance is not possible, the portions of the slope that pose the greatest hazard to structures will require some level of remediation.

Figure 17. (a) Gullies at Site 2 with varying sizes of rock blocks. (b) A large gully at Site 4 with large rock blocks. (c) Gullies near the base of the slope at Site 3 with large rock blocks.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

161

Figure 16. One of the lenses of red to gray mudstone that separate the Shinarump Conglomerate Member from the Moenkopi Formation at several locations.

Figure 17. (a) Gullies at Site 2 with varying sizes of rock blocks. (b) A large gully at Site 4 with large rock blocks. (c) Gullies near the base of the slope at Site 3 with large rock blocks.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

161

Jacklitch, Shakoor, and Lund

Figure 18. Rockfall-hazard map of Rockville showing the hazard categories. The width of the color bands indicates the extent a rockfall may travel, based on rockfall trajectories. The blue band represents the extent of the current hazard map, based on the shadow zone concept. The topographic map used in the figure is outdated with respect to the locations of houses. Note: 1 mi = 1.6 km.

There are two remediation methods that may be cautiously considered if government assistance (external source of money) becomes available. The first method involves removing loose blocks within the Shinarump Conglomerate. Extreme caution would be necessary to avoid dislodging any unintended rock blocks or losing control of the rock blocks that are removed. The second remediation method would involve using rock anchors to stabilize the larger blocks and draped wiremesh nets to restrain the smaller rockfall debris closer to the slope. A contractor with significant experience in mitigating difficult and hazardous slopes would be required to perform this work. Any remediation action should focus on the western end of Rockville, where the highest degree of rockfall hazard exists (Figure 19).

LIMITATIONS OF RESEARCH The hazard map developed in this research indicates current hazard conditions. The ratings provided on the map will change based on factors such as an increase of population within the hazard zone or implementation 162

of remedial measures. The area designated as the lowhazard area could become a moderate- or even a highhazard (risk) area if new homes are built within the hazard zone. The coefficient of restitution and the coefficient of friction values we used for the rollout distance were obtained from charts in the Rocscience (2015) manual. While the charts provide a reasonable estimate for these values, they may or may not represent actual field conditions. Furthermore, the RocFall software cannot model factors such as a rock block breaking up while moving down the slope or the transfer of energy from the falling rock to a stationary rock at the base of the slope. Another limiting factor of the RocFall software is the limited number of rock block shapes used. All of these factors can either increase or decrease the rollout distance. CONCLUSIONS The conclusions of this study are the following: 1. Rockfalls, plane failures, wedge failures, and toppling failures are the potential modes of failure

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143â&#x20AC;&#x201C;163

Evaluation of Rockfall Hazard

3. 4.

affecting the east-west trending, south-facing slope through the town of Rockville, Utah. Because of the height at which these failures occur, they all descend as rockfalls. The rockfalls in the Rockville area result from the combined effects of multiple freeze–thaw cycles, increase in water pressure in discontinuities from rainfall events, and differential weathering. Gullies in the weak material in the lower part of the slope add to rockfall hazard by channelizing the rockfall debris. Rockfall hazard is higher in the western portion of Rockville due to greater slope height, greater exposure of the weak Shnabkaib Member leading to undercutting, and the higher concentration of homes within the rockfall rollout zone. The hazard to homes and inhabitants decreases eastward. However, even the eastern slopes have potential for large rockfalls and resulting property damage, injury, or death in the vicinity of the slopes. Large rockfalls will continue to occur if slope stabilization measures are not implemented, and the associated hazard will continue to exist as long as Rockville residents live within the rockfall rollout zones. REFERENCES

AMERICAN SOCIETY FOR TESTING AND MATERIALS, 2010, Standard Test Method for Slake Durability of Shales and Similar Weak Rock D4644-08, Vol. 1, No. 4: ASTM International, West Conshohocken, PA. doi:10.1520/D4644-08. ADMASSU, Y.; SHAKOOR, A.; AND WELLS, N. A., 2012, Evaluating selected factors affecting the depth of undercutting in rocks subject to differential weathing: Engineering Geology, Vol. 124, pp. 1–11. BLATT, H.; TRACY, R. J.; AND OWENS, B., 1996, Petrology: Igneous, Sedimentary, and Metamorphic: WH Freeman and Co., New York, 530 p. DALRYMPLE, J. AND PIPER, M., 2013, Rock Slide Destroys Home, Kills Two in Rockville: Electronic document, available at http://archive.sltrib.com/story.php?ref=/sltrib/ news/57256787-78/amp-rockville-slide-wright.html.csp FOLK, R. L., 1974, Petrology of Sedimentary Rocks: Himphill Publishing Company, Austin TX, 182 p. GRAHAM, J., 2006, Zion National Park Geologic Resource Evaluation Report: NPS/NRPC/GRD/NRR-2006/014: National Park Service, Denver, CO. INTERNATIONAL SOCIETY FOR ROCK MECHANICS, 1981, Suggested Method for Determination of Slake Durability Index: Rock Characterization, Testing and Monitoring: Pergamon Press, London, U.K., pp. 92–94.

Evaluation of Rockfall Hazard

KNUDSEN, T. R., 2011, Investigation of the February 10, 2010 Rock Fall at 274 West Main Street, and Preliminary Assessment of Rock Fall Hazard, Rockville, Washington County, Utah: Report of Investigation 270, Utah Geological Survey, pp. 1–17. KNUDSEN, T. R. AND LUND, W. R., 2013, Geologic Hazards of the State Route 9 Corridor, La Verkin City to Town of Springdale, Washington County, Utah: Utah Geological Survey Special Study 148, 13 p., 9 plates, scale 1:24,000, DVD. LUND, W. R.; KNUDSEN, T. R.; AND BOWMAN, S. D., 2014, Investigation of the December 12, 2013, Fatal Rock Fall at 368 West Main Street, Rockville, Utah: Report of Investigation 273, Utah Geological Survey, 20 p. MABBUTT, D., 2013, Rockslide in Rockville, One Home Destroyed, 2 Fatalities: Electronic document, available at http:// www.stgeorgeutah.com/news/archive/2013/12/12/mabrockslide-in-rockville-one-home-destroyed-photo-gallery MENNE, M. J.; DURRE, I.; KORZENIEWSKI, B.; MCNEAL, S.; THOMAS, K.; YIN, X.; AND HOUSTON, T. G., 2012, Global Historical Climatology Network—Daily, Version 3.22: NOAA National Climatic Data Center. doi:10.7289/V5D21VHZ. NEW YORK STATE DEPARTMENT OF TRANSPORTATION, 1996, Rock Slope Hazard Rating Procedure: Geotechnical Engineering Manual 15, Geotechnical Engineering Bureau, Albany, NY. OREGON DEPARTMENT OF TRANSPORTATION, 2001, Landslide and Rockfall Pilot Study Final Report: Geo-Hydro Section, Oregon Department of Transportation, Salem, OR. PACK, R. T.; BOIE, K.; MATHER, S.; AND FARRELL, J., 2006, UDOT Rockfall Hazard Rating System: Final Report and User’s Manual: Utah Department of Transportation Research and Development Division Report Number UT-06.07, 81 p. PIERSON, L. A., 1991, The Rockfall Hazard Rating System: Oregon Department of Transportation Technical Report FHWA-ORGT-92-05, pp. 1–11. ROCSCIENCE INC., 2014, Dips Version 6.016: University of Toronto, Toronto, ON, Canada. ROCSCIENCE INC., 2015, RocFall Version 5.014: University of Toronto, Toronto, ON, Canada. SHARP, A., 2013, Boulder the Size of an Elephant Crushes Entire House and Instantly Kills Two Inhabitants in Utah Landslide: Electronic document, available at http://www. dailymail.co.uk/news/article-2523067/Boulder-sizeelephant-crushes-entire-house-instantly-kills-inhabitantsUtah-landslide.html STIMPSON, B., 1981, A suggested technique for determining the basic friction angle of rock surfaces using core: International Journal of Rock Mechanics, Mining Sciences, and Geomechanics Abstracts, Vol. 18, No. 1, pp. 63–65. TURNER, K. A. AND DUFFY, J. D., 2012,. Modeling and prediction of rockfall: In Turner, K. A. and Schuster, R. L. (Editors), Rockfall Characterization and Control: Transportation Research Board, Washington, DC, pp. 334–406. UTAH GEOLOGICAL SURVEY, 2013, Interactive Utah Geologic Map: Electronic document, available at http:// geology.utah.gov/maps/geomap/interactive/viewer/index.html WEST, T. R., 1995, Geology Applied to Egineering: Waveland Press, Long Grove, IL, 560 p. WYLLIE, D. C. AND MAH, C. W., 2004, Rock Slope Engineering, 4th ed.: Spon Press, New York, NY, 431 p.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

163

3. 4.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 143–163

163

Monitoring Thermal Springs to Improve Land Management Decision-Making, Sierra Nevada, California

JEROME V. DE GRAFF1 CHRISTOPHER J. PLUHAR

Department of Earth & Environmental Sciences, California State University Fresno, 2576 East San Ramon Avenue, Mail Stop ST-24, Fresno, CA 93740

ALAN J. GALLEGOS KELLEN TAKENAKA

U.S. Department of Agriculture Forest Service, 1600 Tollhouse Road, Clovis, CA 93611

BRYANT PLATT

Department of Earth & Environmental Sciences, California State University Fresno, 2576 East San Ramon Avenue, Mail Stop ST-24, Fresno, CA 93740

Key Terms: Thermal Springs, Land-Use Planning, Site Investigation, Hydrogeology, Sierra Nevada, California ABSTRACT The Sierra National Forest administers Mono Hot Springs and other nearby geothermal features, a concentration of more than a dozen springs, pools, and seeps in the high Sierra Nevada, California. The Native American Mono Tribe traditionally uses Mono Hot Springs for spiritual purposes, while simultaneously the Mono Hot Springs Resort holds a special-use permit for some of the geothermal waters. To support environmental assessments for area management, the Sierra National Forest studied thermal spring chemistry and temperature, evaluating potential use conflicts. An initial multi-year monitoring of 11 representative thermal springs was followed a decade later by another multi-year sampling of the same springs, providing insight into the geothermal character of Mono Hot Springs. Measured water temperatures ranged from 44.5◦ C to 24.3◦ C and pH from 8.0 to 7.03, depending upon the thermal spring, higher pH values correlating with lower temperatures. Thermal spring temperatures varied seasonally with higher temperatures in springtime and lower ones in autumn. pH did not exhibit a coherent seasonal variation. Mono Hot Spring temperature decreased and pH increased during the decade-long study period, with even greater longer-term temperature change evidenced at nearby Mono Crossing. Silica and cation geothermometry at Mono Hot Springs suggests

1 Corresponding

author email: jdegraff@csufresno.edu

that the geothermal waters reached equilibrium with 74– 79◦ C rock at depth at estimated pH of 5 to 6. The spatial distribution of neighboring thermal springs, regional seismicity, and mapped faults suggests that Mono Hot Springs rises along faults running nearly north-south, connecting to Mammoth Mountain and Long Valley, California, 30 km to the north. INTRODUCTION Mono Hot Springs, California, represents a hydrological resource on National Forest System land that needs wise management and also provides insight into the hydrogeologic and tectono-magmatic character of the region. Water resources are a significant component of land management on national forests in the United States, with early objectives including “. . . securing favorable conditions of waterflows . . .” as a primary reason for establishing a national forest (Forest History Society, 2009). This direction reflected the fact that many rivers flow from a source within national forests, especially in the western United States, and foreshadowed the many different water resource uses now common to today’s national forests. Presentday water-resource management encompasses streams, reservoirs, natural lakes, and springs and attempts to balance the needs of downstream agriculture and communities, electrical power generation, fisheries and wildlife habitat, and recreational activities. To address how best to manage these sometimes-competing water uses, decision-makers need information developed through studies of specific water resources or management activities that might alter them (De Graff, 1979,

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

165

1 Corresponding

author email: jdegraff@csufresno.edu

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

165

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Figure 1. Map showing the region surrounding Mono Hot Springs, including the San Joaquin River South Fork. Figures 2 and 3 are within the outlined area. Among the features shown are Long Valley Caldera (LVC), San Joaquin River (SJR), and Casa Diablo Geothermal Area (CD). Identified thermal springs include Keough Hot Spring (KHS), Red’s Meadow Hot Spring (RMHS), Fish Creek/Iva Bell Hot Springs (FCIBHS), Crater Lake Meadow Hot Spring (CLMHS), Mono Crossing Hot Spring (MCHS), Mono Hot Springs (MHS), and Blayney Meadow Hot Spring (BMHS). Data sources: faults USGS and CGS, 2006; seismicity data (August 4, 1980, to August 11, 2016; USGS, 2017); Walker Lane from Dong et al., 2014; thermal spring locations from Berry et al., 1980 and NCEI; hillshade based on USGS 30-m digital elevation (DEM) map.

1982; Jager and Rose, 2003; Berg et al., 2005; and De Graff et al., 2007). Springs may not contribute greatly to fluvial discharge from a national forest but may constitute significant water resources for wildlife habitat, local potable water, and support of groundwater-dependent ecosystems. Thermal springs are those with a water temperature considerably greater than the local mean annual atmospheric temperature (Neuendorf et al., 2011). A thermal spring is designated as either a “warm” or “hot” spring based on whether its temperature is less than or greater than that of the human body. Thermal springs in national forests are often used for recreation, and a number of these locations are scattered throughout the southern Sierra Nevada. Mono Hot Springs within the Sierra National Forest is a wellknown and easily accessible example. Mono Hot Springs is a concentration of springs, seeps, pools, and concrete structures, clustered adjacent to the South Fork of the San Joaquin River on

166

the boundary of the Ansel Adams Wilderness (Figure 1). The hot springs saw use from pre-Columbian through historic time by Native Americans for spiritual purposes and later development by the federal government and commercial interests in the 1930s and thereafter (Rose, 1994; The Mono Nation, pers. comm., 1999). Today, the site includes a riverside campground and a rustic resort. Despite their importance to management issues, the character and possible source of Mono Hot Springs’ geothermal water have received minimal study (Lockwood et al., 1972; Mariner et al., 1977). The site lies within what is normally considered the “stable” Sierra Nevada batholith but is near Long Valley Caldera and faults of the Walker Lane, as well as a substantial concentration of earthquake epicenters within the Sierra itself (Figure 1; ANSS Comprehensive Earthquake Catalog). In this article, we describe characteristics of Mono Hot Springs, including 1) temperature and pH values and variability identified at individual

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

166

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs

springs and among the thermal springs in general, 2) decade-long trends in these parameters, 3) the relationship between temperature and pH observed at the springs, and 4) chemical geothermometers applied to several of the thermal springs, providing insight into the geothermal heat source. We also review existing geological and geophysical data in order to speculate on the geothermal system giving rise to this resource. BACKGROUND Geologic and Tectonic Setting The study area lies in the upper San Joaquin River watershed at the eastern-central edge of the Sierra Nevada Microplate, a portion of relatively unfaulted lithosphere between the California Coast Ranges and the Walker Lane (Figure 1; Argus and Gordon, 1991, 2001; Unruh et al., 2003). Despite being in a region of seemingly low deformation rate, one or more factors have or could have contributed to the existence of the Mono Hot Springs geothermal area: 1) tectonic activity of the Walker Lane and Sierra Nevada frontal fault zone located 30 km to the north, northeast, and east along the margin of the Sierra Nevada microplate; 2) volcanic activity of the Long Valley, 30 km to the northeast, and Mammoth Mountain, 30 km to the north (Figure 1); 3) local incipient tectonic activity evident in mapped faults and small earthquakes (Figures 1 and 2A); 4) the Miocene and Pliocene San Joaquin Volcanic Field (Figure 2B); and 5) Sierra Nevada delamination. A review of these follows. The Sierra Nevada Microplate consists of the Sierra Nevada and the Great Valley physiographic provinces as a single coherent piece of lithosphere (Argus and Gordon, 1991, 2001; Kreemer et al., 2009). The microplate is caught up between motion of the Pacific and North American plates, with most displacement occurring on faults of the San Andreas system and Coast Range faults (Argus and Gordon, 2001) and the remaining 20–25 percent occurring in the Walker Lane and Basin and Range (Dixon et al., 1995, 2000). The Sierra Nevada Microplate moves northwestward relative to stable North America (Argus and Gordon, 1991, 2001; Unruh et al., 2003; and Unruh and Hauksson, 2009), separated from it by the extensional Basin and Range and dextral Walker Lane tectonic provinces (Unruh et al., 2003; Unruh and Hauksson, 2009; Dixon et al., 1995, 2000; and Kreemer et al., 2009). The study area lies within 30 km of the Microplate’s eastern edge, where faults mark the Walker Lane, Sierra Nevada frontal fault, and the edge of the Long Valley Caldera (Figure 1). The Walker Lane hosts numerous distributed faults, hot springs, and geothermal features. North-

Monitoring Thermal Springs

northwest–oriented dextral faults and north to northnortheast–striking normal faults dominate this belt of dextral to trans-tensional shear (Unruh et al., 2003). The Walker Lane has been active since the Miocene (Surpless et al., 2002; Oldow et al., 2008), with the westernmost Walker Lane, the current edge of the Sierra Nevada microplate, becoming tectonically active between 8 and 3 Ma (e.g., Bacon et al., 1982; Jones et al., 2004; Oldow et al., 2008; and Unruh et al., 2014). Many geothermal features in the Walker Lane derive their heat from tectonically induced magmatic systems, such as the Coso Geothermal Area, which generates ∼200 MW of electricity (Monastero, 2002). Others do not exhibit an obvious spatial association with recent volcanic activity (e.g., Keough, Dirty Socks, and Buckeye Hot Springs). Long Valley constitutes one of the most significant concentrations of geothermal features in the Walker Lane (Berry et al., 1980; National Centers for Environmental Information [NCEI]) and includes hot springs such as that at Red’s Meadow and Hot Creek, as well as Casa Diablo Geothermal Area, hosting a 29-MW geothermal plant (Mammoth Pacific, 2017). Today’s Long Valley (Figure 1) has been volcanically active since the early Pleistocene (Bailey, 1989) and was the site of one of the most significant Quaternary caldera-forming eruptions in the conterminous United States, producing the Bishop Tuff/Ash (0.772 Ma; Sarna-Wojcicki et al., 2000). Long Valley Caldera experiences earthquake swarms that include probable diking events (Templeton and Dreger, 2006) and surface elevation uplift interpreted to result from magma inflation (Langbein et al., 1993). Long Valley hosts a very active geothermal system (Berry et al., 1980; Hilton, 1996). Mammoth Mountain, at the boundary between the Sierra Nevada Microplate and Long Valley and 33 km north of the study area (Figure 1), became active during the Pleistocene (0.23 Ma) and continued into the Holocene, most recently erupting ∼8 ka (Hildreth et al., 2014; Hildreth and Fierstein, 2016). Earthquake swarms and other seismicity, ground deformation, and magmatic gas release all demonstrate unrest beneath Mammoth Mountain (Sorey et al., 2000; Hill and Prejean, 2005). The magmatic gas release includes CO2 and mantle helium, emanating from parts of the north, west, and south flanks of Mammoth Mountain, especially to the south near Horseshoe Lake, resulting in tree kills (Sorey et al., 2000; Hill and Prejean, 2005). In contrast to the tectonically and volcanically active Walker Lane, the Sierra Nevada is relatively quiescent (e.g., Figures 1 and 2A). Earthquakes do occur within the Sierra Nevada Microplate (Figures 1 and 2A), but their magnitudes and abundance (ANSS

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

167

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

167

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Figure 2. Study area maps, showing Crater Lake Meadow Hot Spring (CLMHS), Mono Crossing Hot Spring (MCHS), Mono Hot Springs (MHS), and Blayney Meadow Hot Spring (BMHS). (A) USGS 30-m DEM hillshade showing earthquake epicenters spanning August 4, 1980., to August 11, 2016 (USGS, 2017), and seismic focal mechanisms for events for which USGS provided them. (B) Topographic map showing remnant volcanic rock of the Miocene and Pliocene San Joaquin Volcanic Field found within the study area (from Bateman, 1965; Bateman et al., 1971; and Lockwood and Lydon, 1975).

168

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165â&#x20AC;&#x201C;185

Monitoring Thermal Springs

Comprehensive Earthquake Catalog) are low compared to the Walker Lane, and Sierran lithosphere experiences little deformation overall (Argus and Gordon, 2001; Kreemer et al., 2009). Very low reduced heat flow measurements (<25 mW/m2 ) within most of the Microplate support this idea (Saltus and Lachenbruch, 1991). For the most part, volcanic activity within the Sierra Nevada occurred millions of years ago (e.g., Manley et al., 2000). The study area lies within the Pliocene San Joaquin volcanic field (see Figure 2B), which may have been produced by delamination of an ecologite root and replacement by warm mantle during the Pliocene or late Miocene (Farmer et al., 2002). The volcanic activity near the study area occurred sufficiently long ago that volcanism-associated heat would have already dissipated. Indeed, there is no obvious spatial relationship between Neogene volcanic deposits and thermal springs (Figure 2B). The delamination likely continues to influence regional heat flow in the study area. However, there are notable exceptions to the current tectonic and volcanic quiescence of the Sierra Nevada. For example, the Cascade Arc’s southern terminus, Lassen, lies at the northern end of the Sierra Nevada. In addition, the southern Sierra Nevada currently exhibits significant tectonic and volcanic activity at the Kern Canyon Fault (e.g., Brossy et al., 2012) and nearby hot springs as well as middle and late Quaternary activity in the Golden Trout Volcanic Field (Moore and Sisson, 1985; Manley et al., 2000). Saltus and Lachenbruch (1991) interpreted higher heat flow and seismicity in the southeastern Sierra to represent incipient “calving off” of a segment of the eastern batholith and intrusions at depth, and Unruh and Hauksson (2009) demonstrate seismogenic deformation in the southern Sierra Nevada. In addition, small-magnitude seismicity (ANSS Comprehensive Earthquake Catalog) and small-offset faults (Wakabayashi and Sawyer, 2000, 2001) occur within the Sierra Nevada Microplate at a very low rate (Figures 1 and 2A). Rocks in the study area are composed of Paleozoic and Mesozoic metamorphic rocks, Jurassic and Cretaceous rocks of the Sierra Nevada batholith, Neogene volcanic rocks, and unconsolidated Quaternary glacial, fluvial, and colluvial sediments and soils (Bateman, 1965; Bateman et al., 1971). Bedrock underlying Mono Hot Springs is the Cretaceous-aged (Tobisch et al., 1995; Frazer et al., 2014) Mount Givens Granodiorite, an extensive pluton within this part of the Sierra Nevada (Bateman et al., 1971). Similarly, the Mount Givens Granodiorite underlies the thermal springs at Mono Crossing and Crater Lake Meadow, both northwest of Mono Hot Springs (Bateman et al., 1971). It may also underlie the Blayney Meadow thermal springs, though the extensive alluvium present and

Monitoring Thermal Springs

the proximity to other bedrock units makes this determination uncertain (Bateman, 1965). According to Mariner and others (1977), the Blayney Meadow thermal spring present at Muir Trail Ranch issues from granodiorite. Historical Context A trans-Sierra trail that passes by Mono Hot Springs facilitated trade between the Mono tribe west of the Sierran crest and tribal groups on the eastern side. The western Mono traditions include both this seasonal trading activity and spiritual use of the springs (The Mono Nation, pers. comm., 1999). The area remained accessible only by foot or horseback until major hydroelectric development stimulated both road and recreational development in the area (Rose, 1994; Sierra Nevada Geotourism, 2017). A road from Huntington Lake across Kaiser Pass was built to facilitate excavation of Ward Tunnel and construction of Florence Lake dam. The tunnel began carrying water diverted from the South Fork of the San Joaquin River in 1925 and then from Florence Lake in 1926. A siphon completed in 1927 still conveys water from the Mono and Bear Creek diversions to the Ward Tunnel, crossing the South Fork San Joaquin just upstream from Mono Hot Springs. In the 1930s, the Civilian Conservation Corps built the High Sierra Guard Station and a bathhouse at Mono Hot Springs. The popularity of the bathhouse resulted in approval of a permit for the Mono Hot Springs Resort. Opening in 1937, the resort consisted of a general store and a dozen small cabins. Lake Thomas A. Edison was completed in 1954, extending the Kaiser Pass Road beyond the Mono Hot Springs area and facilitating increased recreational access. In the early 1960s, the Forest Service removed bathhouse buildings at the Mono Hot Springs, leaving their concrete foundations in place. At the same time, a spring box and storage tank were built for supplying hot water via a pipeline suspended over the river to a bathhouse at the resort across the river. This bathhouse was rebuilt in the early 1980s. In the late 1990s, the reissuance of the Mono Hot Springs Resort special-use permit revealed differences among the various stakeholders interested in the management decisions at Mono Hot Springs. The resulting discussions revealed a lack of basic information about the character of the thermal springs. Stakeholders and managers lacked fundamental data, such as the typical temperature and pH of individual springs, as well as spatial and seasonal variability in these factors among the springs. Consequently, the Forest Service initiated a multi-year monitoring program to gather these data. The goal of this monitoring can be described as

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

169

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

169

De Graff, Pluhar, Gallegos, Takenaka, and Platt

“phenomena monitoring,” the intent of which is to obtain sufficient data to better describe or characterize an event or process (De Graff, 2011). Because national forests cover large areas, land management often requires site-specific studies or monitoring efforts upon which to base decisions (De Graff and Romesburg, 1981; De Graff and Gallegos, 1987; and De Graff et al., 2015). Monitoring was undertaken to improve the understanding of Mono Hot Springs and to provide data with which to underpin future land management decisions affecting these water resources.

Physical Setting The Mono Hot Springs lie in the southern Sierra Nevada, California (Figure 1). The South Fork of the San Joaquin River flows northwest from its headwaters to its junction with the main stem near Balloon Dome (Figure 2A). In the vicinity of Mono Hot Springs, the glaciated valley of the South Fork is bounded by Mount Givens and the other prominent peaks of Kaiser Ridge on the south and west and the main crest of the Sierra Nevada lying to the east of Mt. Hooper and Graveyard Peak (Figure 2A). Mono Hot Springs consists of pools, springs, and seeps within an approximately 9-ha area bounded by the South Fork of the San Joaquin River on the north and the Kaiser Pass Road on the east (Figure 3). The main concentration of springs is on the south side of the South Fork of the San Joaquin River across from the Mono Hot Springs Resort (Figure 3). The thermal springs occur within a landscape of granitic bedrock outcrops interspersed with low marshy areas or accumulations of tufa. In a few places, springs flow from either granitic bedrock or tufa deposits. Inflow to pooled water can be detected where gas bubbles rise from the subsurface or by higher temperature zones at pool bottoms. Mariner and others (1977) identified the bubbles in two of the Mono Hot Springs as being predominantly nitrogen and indicative of a lower-temperature thermal system. The majority of springs rise within pools that are formed in native soil. Some are enclosed within the remnant concrete foundations of historic bath-house structures, where water rises from the open bottom of the foundations. One spring is confined within a concrete box enclosing the spring but leaving the bottom open and permitting a concrete cover to be placed over the top. Water from this spring is piped into a tank supplying the bath house at the Mono Hot Springs Resort via a pipeline across the river. Water outflow from the pools issues through low points, across the top of enclosing concrete structures, or from pipes installed in spring boxes. Thermal spring users have sometimes 170

partially blocked pool water egress in order to raise the water level within a pool. This study included 11 spring-fed pools scattered across the Mono Hot Springs area (Table 1 and Figure 3). The highest-elevation pools sampled occur at approximately 2,085 m elevation, well above the level of the San Joaquin River (pools 2 and 3; Table 1). Both of these springs form pools within native soil at the base of granodiorite outcrops present on the upslope and downslope sides of Kaiser Pass Road, next to where the road is bridged across the penstock from Lake Thomas A. Edison to the Ward Tunnel (Figure 3). A large accumulation of tufa with extensive warm water seepage occurs on the downslope side of the road adjacent to the west side of the outcrop where pool 3 is located. Roughly downslope from springs 2 and 3, springs 4 and 6 (Table 1) are associated with a large tufa mound along the San Joaquin River’s south bank. A concrete bath-house foundation provides a pool for spring 4 at the base of the tufa mound, about 1 m from the river channel, and 4 m above river level (elevation 2,000 m) (Figures 3 and 4A). At 2013 m in elevation, spring 6 issues from the tufa mound apex into a concrete spring box supplying water to a tank for the Mono Hot Springs Resort bath house (Figures 3 and 4B). This spring box is approximately 185 m due south of spring 4 and 17 m higher than the river level. Three spring-fed pools (pools 7, 8, and 9; Table 1) cluster in a flat, open area northeast of the large tufa mound along the San Joaquin River’s south bank and 23 m south of the river channel. A concrete bath-house foundation with two deep sections forms the spring 7 pool at about 2,001 m in elevation. Springs 8 (Figures 3 and 4C) and 9 consist of two pools within soil within 3 to 4 m of the concrete structure containing the pool for spring 7. Their elevation and distance from the river channel are essentially the same as those of spring 7. On the bank of the South Fork of the San Joaquin River at an elevation of 1,997 m (within 1 to 2 m of the river, depending on the time of the year), spring 10 supplies a single pool in soil lying 18 m away and almost due north from the cluster formed by springs 7, 8, and 9. Springs 11 and 12 and their respective pools are about 73 m from Kaiser Pass Road. Spring 11 forms a pool adjacent to a large granodiorite outcrop at an elevation of 2,006 m. Spring 12 can be found upslope and to the south of spring 11. Spring 12 supplies one of the larger thermal pools within the top of a small tufa mound at an elevation of about 2,012 m (Figures 3 and 4D). Spring 13 supplies a pool in a small concrete tub similar in appearance to a watering trough and is found at an elevation of 2,023 m at the base of a vertical granodiorite face, just west of the Kaiser Pass Road. The Mono Hot Springs are not the only thermal springs present along the South Fork of the San

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs

Figure 3. A Google Earth image showing (A) Location of the 11 springs sampled at Mono Hot Springs. The yellow circles indicate the springs formed as a pool in soil, and the red ones are those springs enclosed in concrete structures. The spring numbers correspond to those in Table 1. (B) Average temperature (◦ C) for the initial monitoring period (Table 2) separated by a slash from the average temperature for the later monitoring period (Table 5).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

171

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

171

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Table 1. The 11 pool-forming springs monitored as part of this study at Mono Hot Springs. The spring number, Global Positioning Satellite (GPS) coordinates (WGS-84), description of physical location, and spring description are listed.

Spring No.

GPS Coordinates

Physical Location

Spring Occurrence

Upslope from Kaiser Pass Forest Road just west of bridge over Edison penstock on small tufa deposit Downslope from Kaiser Pass Forest Road and spring 2 on top of large tufa mound At the base of large tufa mound just upslope from the South Fork of the San Joaquin River On the top of the large tufa mound just upslope from the South Fork of the San Joaquin River

Natural pool with sandy bottom; gas bubbles observed Natural pool with sandy bottom; gas bubbles observed Double-concrete tub with open bottom, gas bubbles observed Concrete spring box (open bottom) with lockable concrete cover; gas bubbles observed Double-concrete tub with open bottom; gas bubbles observed

N37 19.299 , W119 1.115

N37◦ 19.348� , W119◦ 1.108�

N37◦ 19.501� , W119◦ 1.011�

N37◦ 19.464� , W119◦ 1.022�

N37◦ 19.556� , W119◦ 0.937�

Small natural pool with sandy bottom; gas bubbles observed

N37◦ 19.552� , W119◦ 0.932�

Small natural pool with sandy bottom; gas bubbles observed

N37◦ 19.556� , W119◦ 0.940�

Small natural pool with sandy bottom augmented by rocks placed on side nearest the river; gas bubbles observed Small natural pool with sandy bottom adjacent to granitic outcrop augmented by rocks placed at the discharge point Natural pool with sandy bottom; gas bubbles observed Concrete trough abutting a granitic outcrop; little outflow

N37◦ 19.568� , W119◦ 0.934�

N37◦ 19.561� , W119◦ 0.854�

By a granitic bedrock knob midway between the open flat and the bridge crossing the South Fork of the San Joaquin River

N37◦ 19.545� , W119◦ 0.867�

At the top of a small tufa mound upslope from spring 11

N37◦ 19.473� , W119◦ 0.923�

Downslope from Kaiser Pass Road and upslope from the bridge crossing the South Fork of the San Joaquin River on small tufa mound

N37 19.299 , W119 1.115

N37◦ 19.348� , W119◦ 1.108�

N37◦ 19.501� , W119◦ 1.011�

N37◦ 19.464� , W119◦ 1.022�

N37◦ 19.556� , W119◦ 0.937�

N37◦ 19.552� , W119◦ 0.932�

N37◦ 19.556� , W119◦ 0.940�

N37◦ 19.568� , W119◦ 0.934�

N37◦ 19.561� , W119◦ 0.854�

By a granitic bedrock knob midway between the open flat and the bridge crossing the South Fork of the San Joaquin River

N37◦ 19.545� , W119◦ 0.867�

At the top of a small tufa mound upslope from spring 11

N37◦ 19.473� , W119◦ 0.923�

Downslope from Kaiser Pass Road and upslope from the bridge crossing the South Fork of the San Joaquin River on small tufa mound

◦

�

◦

�

In the open flat near the South Fork of the San Joaquin River between the large tufa mound and the Bailey bridge crossing In the open flat near the South Fork of the San Joaquin River between the large tufa mound and the Bailey bridge crossing In the open flat near the South Fork of the San Joaquin River between the large tufa mound and the Bailey bridge crossing On the river’s edge in the open flat near the South Fork of the San Joaquin River midway between the large tufa mound and the bridge crossing

GPS Coordinates ◦

�

◦

�

Physical Location

Spring Occurrence

From comparison with information in Mariner et al. (1977) and phone conversation with Dr. Mariner, this spring appears to be the same spring as the one sampled in their report.

Joaquin River (Figure 1). There are other thermal springs in close proximity to the current channel, both upstream and downstream from the Mono Hot Springs. During a geochemical survey, Lockwood and others (1972) found several thermal springs along the South Fork of the San Joaquin River extending 21 km northwestward (downstream) from Mono Hot Springs. However, these springs were described as being cooler than the Mono Hot Springs. One of these downstream spring locations is about 3 km northwest of Mono Hot Springs near Mono Crossing (Figures 1 and 2). Lockwood and others (1972) described the location as having an extensive area of travertine terraces, similar to features in Yellowstone National Park. A single measurement taken on June 6, 2013, from the only spring visible within that terraced landscape near Mono Crossing yielded a temperature of 18.2◦ C. The nearly 1.2 ha of travertine terraces and 0.5 ha of tufa deposit imply that several more hot springs previously arose within this area, either in series or simultaneously.

172

Approximately 21 km northwest (downstream) of Mono Hot Springs, a 35◦ C thermal spring (http://maps.ngdc.noaa.gov/viewers/hot_springs/) emanates from the north side of the South Fork San Joaquin River Canyon, north of Crater Lake Meadow (Figures 1 and 2). Another concentration of thermal springs is found along the South Fork San Joaquin River about 15 km southeast (upstream) from Mono Hot Springs at Blayney Meadow within the John Muir Wilderness (Figures 1 and 2). The Blayney Meadow thermal springs include a spring issuing from a bedrock fracture at Muir Trail Ranch and a number of spring-fed pools on the opposite (south) side of the river channel. The spring at Muir Trail Ranch is the same one identified by Mariner et al. (1977) as their “Blayney Meadow Hot Spring.” The composition of gas in the bubbles rising at Mono Hot Springs and Blayney Hot Spring was found to be 95 and 97 percent nitrogen by volume, respectively (Mariner et al., 1977). In addition to the thermal springs already mentioned, geothermal features lying to the north of Mono

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Small natural pool with sandy bottom; gas bubbles observed Small natural pool with sandy bottom; gas bubbles observed Small natural pool with sandy bottom augmented by rocks placed on side nearest the river; gas bubbles observed Small natural pool with sandy bottom adjacent to granitic outcrop augmented by rocks placed at the discharge point Natural pool with sandy bottom; gas bubbles observed Concrete trough abutting a granitic outcrop; little outflow

From comparison with information in Mariner et al. (1977) and phone conversation with Dr. Mariner, this spring appears to be the same spring as the one sampled in their report.

172

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs

Figure 4. Representative thermal spring pools from among the 11 monitored between 1999 and 2013 at Mono Hot Springs. (A) Spring 4 is the double-concrete tub at the base of tufa hill near the South Fork of the San Joaquin River. It is representative of several springs rising within the foundations of former bath houses. The snow banks around the spring are typical for mid-winter conditions at Mono Hot Springs. (B) Spring 6 is the spring box at the top of tufa hill. Temperatures measured at this spring were the hottest and varied the least. (C) Spring 8 is a spring pool in soil by the board bridge leading to the deep double-concrete tubs. The temperatures at this spring were found to vary the most during the monitoring periods. (D) Spring 12 is a pool in soil at the top of a small tufa mound upslope from spring 11. Temperatures measured at this spring were the lowest among the 11 springs being monitored.

Hot Springs may also be related. These lie along the surface trace of Quaternary active normal faults striking generally southward from Mammoth Mountain (Figure 1; USGS and CGS, 2006) and include Fish Creek/Iva Bell Hot Spring and the geothermal and magmatic gas release area at Horseshoe Lake just south of Mammoth Mountain. Red’s Meadow Hot Spring lies in this general vicinity as well, but off of this linear trend (Figure 1). Mono Hot Springs and Mono Crossings Hot Springs lie along the strike of the Quaternary active faults that run south of Mammoth Mountain, but the mapped faults end long before reaching the study area.

METHODS Field Temperature and pH Measurements Field pH and temperature monitoring of the Mono Hot Springs began in October 1999. Subsequently, spring season field pH and temperature monitoring took place once the Kaiser Pass Road was clear of snow, and fall season monitoring occurred after the main tourist season and prior to early snowfall on Kaiser Pass. Between 1999 and 2001, three fall (late September/early October) monitoring events and two spring (late May/early June) monitoring events

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

173

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

173

De Graff, Pluhar, Gallegos, Takenaka, and Platt

occurred. Hereafter, this field pH and temperature monitoring will be referred to as the “initial monitoring period.” A decade later another series of monitoring events took place to address ongoing management concerns. This round of monitoring between June 2011 and September 2013 consisted of three spring (early June) monitoring events and three fall (late September/early October) monitoring events. Hereafter, this field pH and temperature monitoring will be referred to as the “later monitoring period.” When initial monitoring began in October 1999, we tested 13 thermal springs distributed across the Mono Hot Springs area. The thermal spring numbers reflect their order during the first field pH and temperature monitoring event and for future reference were named in reference to some identifying site feature. Photos of each thermal spring during this period provided visual reference for future monitoring events. As previously described, these 13 thermal springs consisted of flowing springs and springs rising in pools within soil, springs within concrete bath-house foundations, and one enclosed within a concrete spring box. The two flowing thermal springs proved unreliable for repeated measurement, leaving 11 natural and artificial containment pools fed by thermal springs as the measuring points for the initial 1999–2001 monitoring effort (Table 1). These same 11 pools were evaluated during the later 2011–2013 monitoring period. Protocols involving both the instrumentation used and field data collection methods were established to ensure the reliability of the monitoring data and to avoid introducing systematic error (Romesburg, 2009). A WTW (Wissenschaftlich-Technische Werkstätten GmbH) Multiline P3 portable meter was used for collecting all temperature and pH measurements during both monitoring periods at Mono Hot Springs, Mono Crossing, and Blayney Meadow (Muir Trail Ranch). This battery-operated meter employs a combined temperature and pH probe. The device was calibrated with standards a day or two prior to each sampling event. We transported the Multiline P3 to and from Mono Hot Springs in the padded plastic carrying case provided by the manufacturer. The senior author or someone under his direct supervision conducted all spring measurements taken with the Multiline P3. The same protocol was followed during each temperature and pH monitoring event. We collected water in a Teflon plastic cup supplied as part of the equipment for the Multiline P3 device. At a pool, water was dipped by hand with the cup. First, the cup and instrument probe were rinsed with water from the pool. The cup would then be filled by reaching as far below the surface as practicable. In practice, this was usually 40 to 50 cm below the water surface. We strived to col174

lect water closest to where gas bubbles were seen rising to the surface during the initial monitoring event on the assumption the bubbles coincided with the greatest rate of water inflow. During subsequent monitoring visits we attempted to repeat efforts to obtain water at the same point. The temperature and pH probe was placed in the cup of collected water and measurement was initiated per meter instructions. The temperature and pH values were written in a dedicated field notebook along with the spring name, collection date, and time. The order in which the springs were measured was established during the first monitoring event and was generally duplicated in later sampling events. Temperature and pH monitoring typically occurred between late morning and early afternoon. Geothermometry Thermal springs gain their heat from deep groundwater circulation to heat sources typically within the upper few kilometers of the Earth’s surface (Williams et al., 2008). The impetus of both scientific curiosity and geothermal exploration produced a number of chemical geothermometers for determining the temperature of these deep heat sources (D’Amore and Arnórsson, 1990; Williams et al., 2008). These geothermometers use either silica or various cations found in the emerging spring water to estimate water temperature at depth. These methodologies assume that chemical fluid-rock equilibrium exists in the source aquifer and that the fluid composition remains essentially unmodified as it rises to flow from the thermal spring (D’Amore and Arnórsson, 1990; Williams et al., 2008). Consequently, some thermal spring properties, such as low pH, low thermal spring flow rate, shallow nonthermal groundwater mixing, and localized hydrogeologic conditions (such as rising through salt-rich sediments), can potentially lead to spurious/inaccurate temperature estimates (Mariner et al., 1977; Kolesar and De Graff, 1978; Williams et al., 2008; and Oerter, 2011). To minimize the likelihood of interference with geothermometer accuracy due to the factors mentioned, three chemical geothermometers were used. Oerter (2011) provides a compilation of silica, cation, and isotope geothermometer formulations with any associated temperature range and source information. The silica (quartz) geothermometer used in this study applies to thermal springs with temperatures ranging from 25 to 250◦ C (Fournier, 1977). The choice of the Na-K-Ca geothermometer (Fournier and Truesdall, 1973) was partly due to the fact that it is one of the most commonly used cation geothermometers and provides a specific formulation for thermal waters at temperatures of less than 100◦ C (Williams et al., 2008).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs

Additionally, both of the foregoing geothermometers were used by Mariner and others (1977), permitting direct temperature comparison between their 1974 Mono Hot Springs #6 sample and our 2013 sampling for the same spring. Williams and others (2008) suggested that a K-Mg geothermometer such as the one developed by Giggenbach (1988) might be more appropriate for non– chloride-rich water, so, we used this geothermometer as well. In June 2013, we collected water from four of the 11 Mono Hot Springs for geothermometry analysis. The four chosen springs reflect the range of spring temperature and temperature variability and will be fully discussed in the results section. Geothermometry water samples derived from the same location within each spring from which temperature and pH measurements were made. The Teflon cup was rinsed several times in the thermal spring water and emptied on the ground nearby. The cup was then used to fill 1,000-mL

Monitoring Thermal Springs

amber glass bottles containing nitric acid preservative and 1,000-mL HTPE bottles supplied by the analytical laboratory. All sample bottles were placed on ice in a cooler. Following collection, the samples were transported by vehicle to BSK Laboratories (Fresno, CA) to ensure laboratory analysis within the required hold times. RESULTS Temperature and pH of Mono Hot Springs Temperature During the initial monitoring period (1999–2001), the 11 thermal springs at Mono Hot Springs yielded instantaneous temperatures ranging between 44.5 and 30.7◦ C, with variability possibly attributable to spatial factors, physical conditions, or seasonality (Table 2). The average annual temperature for individual

Table 2. Summary of temperature data. These data include instantaneous temperature measurements, seasonal average temperature within each monitoring period, average temperature for each monitoring period, and range of temperature within each monitoring period. The mean values for the seasonal averages are in bold at the bottom of their respective columns.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

175

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

175

De Graff, Pluhar, Gallegos, Takenaka, and Platt

thermal springs ranged from a high of 44◦ C at spring 6 to a low of 32◦ C at spring 12. Results also demonstrate that individual thermal springs exhibited unique temperature variability during the initial monitoring period. At one end of the variability spectrum, temperatures at the spring box on top of the large tufa mound (spring 6) fluctuated only 0.6◦ C, whereas at the other end of the spectrum the thermal spring by Board Bridge (spring 8) seasonally oscillated across a range of 8.0◦ C (Table 2). The monitored thermal springs were found to vary during the year, showing temperature maxima in spring and minima in fall. During the later (2011–2013) monitoring period, the 11 thermal springs at Mono Hot Springs produced instantaneous temperatures ranging from 43.7 to 24.3◦ C (Table 2). Again, the highest average annual temperature was at spring 6 (43◦ C) and the lowest was at spring 12 (27◦ C) (Table 2). In addition, the difference between the greatest and smallest temperatures recorded for individual thermal springs revealed that some springs varied more than others, similar to findings from the initial monitoring period. As before, the least variable thermal spring was spring 6 at the summit of the large tufa mound, with a difference of only 3.0◦ C. The thermal spring by Board Bridge (spring 8) again exhibited the greatest difference at 6.3◦ C (Table 2). The seasonal temperature variability again displayed a maxima in spring and minima in fall. The average annual thermal spring temperatures exhibit no obvious spatial or surface characteristics– related pattern (Figure 3). We considered all of the following factors, and none provided a clear influence on water temperature: location/elevation within the study area, type of spring containment, distance from the river channel, or proximity to other springs with similar temperature characteristics. For example, spring 6, which has the highest average temperature and little seasonal temperature variability, is only 1 m higher in elevation than spring 12, which has the lowest average temperature with the greatest variability in seasonally measured values. The nature of the spring pool and proximity to the San Joaquin River appear to play no role in thermal spring temperature. For example, springs 4 (contained within a bath-house foundation) and 10 (a within-soil pool) are found at nearly the same elevation only tens of meters apart and are the two thermal springs located closest to the river (within 1 to 2 m), yet both thermal springs produced some of the higher average temperatures of the 11 springs monitored. Temperature variability appears to be unrelated to water temperature between thermal springs in close proximity. Both springs 11 and 12 display a similar moderate variability in their temperature measurements over time, but the average annual temperature of spring 11 is among the higher ones 176

measured, while spring 12, located only 25 m away and 6 m higher on the slope, has the lowest average temperature. Seasonality could influence thermal spring temperatures through mixing with shallow groundwater from snowmelt. The mean thermal spring seasonal (fall and spring) temperatures were computed for the initial monitoring period (Table 2). The mean annual temperature for early fall monitoring (late September to early October) was 37.3◦ C, while the value for late spring monitoring (late May to early June) was 39.1◦ C (Table 2). The autumn temperatures at the 11 thermal springs were consistently lower than or equal to those taken during springtime. A similar comparison for the mean thermal spring seasonal (fall and spring) temperatures was computed for the later sampling period (Table 2). The mean annual temperature for early fall monitoring (late September to early October) was 34.6◦ C, while the value for monitoring done during late spring (late May to early June) was 36.7◦ C (Table 2). Comparison of these mean seasonal values for the initial and later monitoring periods demonstrates a consistent 2◦ C difference between mean fall and mean spring temperatures. In order to test the statistical significance between the mean seasonal thermal spring temperature differences, we applied the Kolmogorov-Smirnov (K-S) test, a non-parametric and distribution-free method applicable to our uncertainty about the nature of the population distribution and suitable for the relatively small number of data points (Cheeney, 1983). An Internetaccessible program for computing the K-S was used to make these calculations using all of the initial monitoring period measurements taken in the fall compared to those taken in the spring (CSBSJU, 2016). Table 3 displays the results of the K-S test comparing fall and spring temperatures for the initial (1999– 2001) and later (2011–2013) monitoring period and the descriptive statistics for their respective fall and spring temperature populations. For the initial (1999– 2001) monitoring period, the maximum difference in cumulative fraction or sample K-S statistic is D = 0.3485, with a P value of 0.062. For the later (2011– 2013) monitoring period, maximum difference in cumulative fraction is D = 0.3118, with a P value of 0.082. In both cases, the sample statistic is greater than the critical D statistic (Zaiontz, 2016) for the same P value. So we reject the null hypothesis that the fall and spring temperature values are all drawn from the same population during both initial (1999–2001) and later (2011–2013) monitoring periods. This indicates that a statistical difference exists between the mean fall and spring temperature populations during both the initial (1999–2001) and later (2011–2013) monitoring periods.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs

Table 3. The descriptive statistics for seasonal temperature for each monitoring period with their respective Kolmogorov-Smirnov (K-S) statistic values. Also provided are the descriptive statistics comparing temperature and pH data between the two monitoring periods.

Groups

No.

Mean

Standard Deviation

High

Low

K-S Test Results

Groups

No.

Mean

Standard Deviation

High

Low

K-S Test Results

Initial monitoring period fall temperatures Initial monitoring period spring temperatures Later monitoring period fall temperatures Later monitoring period spring temperatures Initial monitoring period temperatures Later monitoring period temperatures Initial monitoring period pH values Later monitoring period pH values

37.31

4.14

44.5

30.7

D = 0.3485, with P = 0.062

37.31

4.14

44.5

30.7

D = 0.3485, with P = 0.062

39.10

3.20

44.2

33.0

39.10

3.20

44.2

33.0

34.32

4.30

42.5

24.3

34.32

4.30

42.5

24.3

35.86

5.02

43.7

20.2

35.86

5.02

43.7

20.2

55 62 55 62

38.03 35.09 7.35 7.52

3.86 4.70 0.245 0.197

44.5 43.7 7.8 8.0

30.7 20.2 7.0 7.15

55 62 55 62

38.03 35.09 7.35 7.52

3.86 4.70 0.245 0.197

44.5 43.7 7.8 8.0

30.7 20.2 7.0 7.15

Visually, it would seem that the later monitoring period average annual temperatures are nearly all consistently cooler by a few degrees than the average annual temperatures during the initial monitoring period (Figures 3 and 5). To test the statistical significance of this apparent difference, the K-S test was applied to the distribution of all the temperature measurements for the initial (1999–2001) monitoring compared to those of the later (2011–2013) monitoring period (Table 3). The maximum difference in cumulative fraction is D

Figure 5. Plot of springs and average annual temperatures for both the initial and later monitoring periods. The actual data show a trend line (dashed) subparallel to the 1:1 line with a temperature offset of about 3◦ C, indicating that the initial monitoring period temperatures were generally warmer than the later monitoring period temperatures. The formula on the graph is the regression defining the data points relative to the trend line.

D = 0.3118, with P = 0.082

D = 0.2903, with P = 0.120 D = 0.3429, with P = 0.001

= 0.2903, with a P value of 0.120. We reject the null hypothesis that the initial and later monitoring period temperatures are drawn from the same population because the sample statistic is greater than the critical D statistic (Zaiontz, 2016). From this, we conclude that a statistically significant temperature difference exists between the earlier and later monitoring periods, in addition to the seasonal variation found between spring and fall temperature within each monitoring period.

D = 0.3118, with P = 0.082

D = 0.2903, with P = 0.120 D = 0.3429, with P = 0.001

The initial monitoring period revealed pH values ranging from 7.0 to 7.8, whereas the later monitoring period obtained similar pH values of between 7.15 and 8.0 (Table 4). The mean pH for the 11 thermal springs during the initial monitoring period was 7.35 and for the later monitoring period was 7.52. Figure 6 shows an observable relationship of higher pH values associated with lower temperatures in both monitoring periods. While this linear relationship is not identical for both monitoring periods, they are similar. The coefficient of determination (R2 ) for the line fitted to the data accounting for 78 percent of the variability present during the initial monitoring period and 62 percent of the variability for the later monitoring period. Given the statistically significant temperature differences between initial and later monitoring periods, a similar assessment of pH was warranted (Table 3). The K-S test yielded D = 0.3429 at P = 0.001. The sample D value is greater than the critical D statistic (Zaiontz, 2016) for the same P value, meaning we reject the null hypothesis that the initial and later monitoring period pH values are drawn from the same population. Therefore, a statistically significant increase in pH between the initial and later monitoring periods exists concomitant with a decrease in temperature. In contrast to the seasonal temperature variation and long-

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

177

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

177

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Table 4. Summary of pH data. These data include instantaneous pH measurements, seasonal average pH within each monitoring period, average pH for each monitoring period, and range of pH within each monitoring period.

term pH trend described earlier, there appears to be no coherent seasonal fluctuation in thermal spring pH values (Table 4). A review of studies providing pH information for thermal springs in the vicinity of Mono Hot Springs provides context for understanding the findings described above. Feth et al. (1964) conducted a broad study of chemical content in natural water in the Sierra Nevada. Water samples were collected within the batholithic rocks of the Sierra Nevada from south of Visalia, California, north to near Downieville, California (Feth et al., 1964, plate 1). Their sampling points included eight thermal springs in granitic rock exclusive of the ones in this study. Their field pH values for these thermal springs ranged from 7.0 to 9.4, with a mean of 8.6. The pH values measured at Mono Hot Springs, which rises through Mount Givens Granodiorite, falls within Feth’s range of thermal spring pH values. The pH of the Blayney Meadows (Muir Trail Ranch) thermal spring in 2015 was 8.68, compared to the 8.0 during monitoring in 1974 (Mariner et al., 1977), and is also within the pH range found during the study of natural water in the Sierra Nevada (Feth et al., 1964). In contrast, the only spring found flowing within the travertine deposits near Mono Crossing in

178

2013 had a pH of 6.4, which is just outside the range of the thermal spring pH values determined by Feth and others (1964) and significantly less than their mean value. The long-term pH increase at Blayney Meadow Hot Spring mirrors that seen at Mono Hot Springs during the current study. ANALYSIS OF THERMAL SPRING SOURCE We chose a subset of Mono Hot Springs for geothermometer analysis. The thermal springs selected for sampling included spring 6, with the highest temperatures and most consistent temperature; spring 7, which is also consistent, but at a notably lower temperature; spring 8, which displays the most variable temperatures annually; and spring 12, which also shows significant annual variability in temperature and consistently has the lowest temperature (Table 1 and Figure 2A). The calculated rock-water equilibration temperatures for Mono Hot Springs water is between 70 and 79◦ C using the two cation geothermometers (Table 5). Estimating temperature at depth by this means is a standard approach in geothermal exploration (e.g., Williams et al., 2008). The values computed by any chemical or isotopic geothermometers rest on the key

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

178

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs

Figure 6. Graphs showing the relationship between pH and average annual temperature for both the initial and later monitoring periods. The formula on each graph is the regression expressing this relationship.

assumption that fluid composition is not changed during its ascent from the hot source aquifer to the spring. Additionally, it is assumed that there are chemical equilibria for the relevant cations within that source aquifer (D’Amore and Arnórsson, 1990). If either of these assumptions is violated, the temperatures of 70– 79◦ C for these four springs may not be the hottest temperature for the heat source but would instead represent a minimum temperature. For springs 6, 7, 8, and 12, the heat source temperature calculated from both cation geothermometers is about twice the temperature of water emerging at the surface. Mariner and others (1977)

Monitoring Thermal Springs

computed a source temperature in their 1974 sample of 80◦ C for spring 6 using the Na-K-Ca geothermometer, which compares well to our 79◦ C (Table 5). Table 5 also shows the source temperature varying from 105 to 109◦ C, using the silica geothermometer for springs 6, 7, 8 and 12. These temperature values are about 25 to 30◦ C greater than those computed using the cation geothermometers, suggesting an overestimation of the heat source temperature. With silica geothermometers, reactions can change the silica content to produce this effect. D’Amore and Arnórsson (1990) note that boiling is one way this increase in aqueous silica within the hot water can happen. Another means for increasing silica concentration is through dilution and mixing with cooler water from shallow aquifers (Kolesar and De Graff, 1978; Oerter, 2011). Given that our emerging waters at Mono Hot Springs are less than the 100◦ C necessary for boiling, and given that shallow groundwater in the region is known to increase in silica content during snowmelt recharge (Feth et al., 1964), dilution by shallow groundwater seems a likely mechanism for this proposed overestimation of source temperature. Mariner and others (1977) obtained a silica geothermometer–based source temperature of 110◦ C for spring 6. They attributed this higher value to dilution with high-silica surface water from a creek upslope of spring 6. Our geothermometry results for spring 6 were consistent with those obtained in 1974 (Mariner et al., 1977). Our cation and silica geothermometer differences at spring 6 are mirrored at springs 7, 8, and#12, despite their physical location being some distance from spring 6 and despite the differing physical characteristics (distance from river, elevation, pool within native soil vs. concrete walls). Based on these findings, it seems reasonable to conclude that mixing of shallow cooler groundwater with rising thermal water is more likely the rule rather than the exception. Our data establish that despite having different average annual temperatures and seasonal temperature variability, thermal springs existing in the Mono Hot Springs area have a source with a similar temperature at depth.

Table 5. Geothermometer results. Data for each of the springs sampled includes instantaneous temperature of each thermal spring sampled for geothermometry, computed heat source temperature for the two cation geothermometers and the silica geothermometer, and average thermal spring temperature for the later monitoring period, when the sampling took place. Each spring is represented on Figure 3A and shown in Figure 4. Spring No. 6: Spring box on tufa hill 7: Double deep concrete tubs at base of tufa hill 8: Spring by board bridge 12: Pool upslope from spring 11 *

Sampled Temperature (◦ C)

Na-K-Ca (◦ C)

K-Mg (◦ C)

Silica (◦ C)

Average Thermal Spring Temperature (◦ C)*

43.7 42.3 38.8 29.8

79 79 74 74

74 74 74 70

106 107 109 105

43 35 36 27

Calculated from the samples taken during the later monitoring period (2011–2013).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

179

Sampled Temperature (◦ C)

Na-K-Ca (◦ C)

K-Mg (◦ C)

Silica (◦ C)

Average Thermal Spring Temperature (◦ C)*

43.7 42.3 38.8 29.8

79 79 74 74

74 74 74 70

106 107 109 105

43 35 36 27

Calculated from the samples taken during the later monitoring period (2011–2013).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

179

De Graff, Pluhar, Gallegos, Takenaka, and Platt

DISCUSSION As noted earlier, Mono Hot Springs is one of several thermal spring localities along the valley bottom of the South Fork of the San Joaquin River. The chemistry of thermal springs of the Sierra Nevada in general (Feth and others, 1964) and of Mono Hot Springs and Blayney Meadow Hot Spring, specifically (Mariner et al., 1977), indicates outflow reflective of deep circulation by meteoric water. Although Lockwood and others (1972) indicate that thermal springs are present along the South Fork of the San Joaquin for a distance of about 19 km downstream from Mono Hot Springs, their map only shows one spring in the vicinity of Crater Lake Meadow and the one at Mono Crossing. Blayney Meadow Hot Spring, including the one at Muir Trail Ranch, lies 18 km upstream from Mono Hot Springs. We note that these four localities—Crater Lake Meadow, Mono Crossing, Mono Hot Springs, and Blayney Meadow—form a linear array (Figures 1 and 2), though this corresponds to the orientation of the San Joaquin River South Fork. Only two known locations, Mono Crossing Hot Spring and Mono Hot Springs, have extensive deposits. It is reasonable to assume that thermal water was rising at these locations at least since the retreat of glacial ice from this area at about 15 ka (Phillips, 2016) to create the extensive travertine terraces near Mono Crossing and the tufa mounds at Mono Hot Springs. The nature of these deposits makes them unlikely to have survived being overridden by glacial ice. The thermal springs downstream from Mono Crossing Hot Springs are described by Lockwood and others (1972) as being cooler than Mono Hot Springs. However, the National Oceanic and Atmospheric Administration hot springs database provides a temperature of 35◦ C for the Crater Lake Meadow thermal spring, while Mono Crossing thermal spring was 18.2◦ C when we measured it on June 6, 2013. The extensive travertine deposits at Mono Crossing indicate a higher temperature in the past, given their comparison to the Mammoth Hot Springs in Yellowstone Park (Bargar, 1978). This suggests some significant change to the thermal water paths to Mono Crossing. Mono Hot Springs and the Blayney Meadow (Muir Trail Ranch) Hot Spring have comparable higher temperatures of about 44◦ C. We currently have insufficient data to explain the wide differences in South Fork San Joaquin thermal spring temperatures. Variable dilution by shallow aquifers and/or independent geothermal systems could separately or together account for the variability. What is the nature of the geothermal heat source at Mono Hot Springs? Certainly, thermal springs are often related to magma bodies at depth, as exemplified by those present in Yellowstone National Park (Bar180

gar, 1978). Remnants of Pliocene flows of trachybasalt are found along much of the South Fork of the San Joaquin River valley (Figure 2B) (Bateman, 1965; Bateman et al., 1971; and Lockwood and Lydon, 1975). Devil’s Table, an example of one of these remnants, is less than 2 km north-northwest of Mono Hot Springs. Mariner and others (1977) suggested that the proximity of these flows to Mono Hot Springs might indicate a magmatic heat source, but their age (Manley et al., 2000) makes it unlikely that this or related intrusions continue to warm the area. Any magmatic heating of the Mono Hot Springs water would have to be from more recent or ongoing intrusions. Conductive heating through thinned lithosphere, while insufficient without fracture porosity, may contribute to thermal springs in the vicinity of the study area. Ducea and Saleeby (1996, 1998) demonstrated that lithosphere in the study area thinned during the Neogene by delamination and has been replaced by (warmer) asthenosphere. Today, lithosphere in the study area is 35–45 km thick (Frassetto et al., 2011), similar to the distance from Mono Hot Springs to the east edge of the Sierra Nevada microplate (Figure 1). However, reduced heat flow in the area is low (Saltus and Lachenbruch, 1991), and warming from hypothesized replacement of the Sierran root with warm asthenosphere will not have reached the surface yet, if purely by conduction (Brady et al., 2006). The proximity of Mono Hot Springs to the active faults within Walker Lane, Long Valley Caldera, and Mammoth Mountain suggests a possible relationship to these large-scale features (Faulds et al., 2004). Mammoth Mountain and the Long Valley geothermal systems are the largest most-proximal possible influences. These systems include geothermal waters in excess of 200◦ C that are clearly being heated by magma at depth (Hilton, 1996; Sorey et al., 2000). The likely presence of a magma body (Stroujkova and Malin, 2000), continuing probable diking (Hill and Prejean, 2005; Templeton and Dreger, 2006), and the large number of active and extinct thermal spring features (Berry et al., 1980; Sorey et al., 2000) in Long Valley highlight the substantial heat input there. However, the ∼30-km distance between Long Valley or Mammoth Mountain and Mono Hot Springs reduces the likelihood of direct flow of water between them and the study area. What seems more likely is that related tectonic effects exist within the study area. Significant small-magnitude seismicity occurs within the Sierra Nevada around the Long Valley Caldera (Figure 2A and ANSS Comprehensive Earthquake Catalog). A concentration of epicenters forms a cluster between Long Valley and Mono Hot Springs, as well as a large concentration north of Blayney Meadow (Figure 2A). The seismicity trend between Long

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs

Valley and Mono Hot Springs passes through Mammoth Mountain, the Fish Creek/Iva Bel Hot spring, and unnamed Quaternary active normal faults near Mammoth (USGS, 2017). This low-magnitude (M 2– 3) seismicity is mostly shallow (<10 km in depth) and exhibits dominantly normal fault focal mechanisms mostly oriented subparallel to this hypothesized lineament (Figures 1 and 2A). The proposed feature, composed of mapped faults, seismicity, and thermal springs, is subparallel to known left-stepping northsouth–oriented normal faults, such as the Hilton Creek Fault (Figure 1), that accommodate NNW-oriented dextral shear in the area (Unruh et al., 2003). Such faults are consistent with formation under dextral trans-tension in this part of the eastern Sierra Nevada (Unruh et al., 2003). In a trans-tensional setting, this faulting and seismicity would likely produce porosity and geothermal fluid-flow pathways and might also include dike emplacement. We speculate that at a minimum, this seismicity indicates one or more faults that permit deep circulation of thermal water to Mono Hot Springs. Seismicity also has the potential to affect the flow paths of rising thermal water in ways that might alter their flow rate or temperature (e.g., Wang et al., 2004; City of El Paso de Robles, 2007). The Blayney Meadow (Muir Trail Ranch) spring located within the John Muir Wilderness provides insight on this potential impact due to an observed earthquake event. The thermal spring at Muir Trail Ranch, on a private land holding at Blayney Meadow within the John Muir Wilderness (Sierra National Forest), was affected by an earthquake in the 1990s. Unfortunately, the date and time were not recorded for this event. Because of the operation of the Ranch, it would have taken place between June and September. Witnesses recalled the earthquake to have probably occurred in late morning or midday sometime in July (H. Painter, Muir Trail Ranch, written comm., 2015). After the earthquake shaking ceased, reports indicate that the spring had stopped flowing. Shortly afterwards, it restarted in spurts for nearly an hour before resuming a continuous flow. However, the flow for the remainder of the day was a cloudy gray color (H. Painter, Muir Trail Ranch, written comm., 2015). The thermal spring flowed normally the following day. No other change, such as temperature change, was noted. However, precise temperature measurement of this spring is only limited to the 1974 sample reported in Mariner and others (1977) and our 2015 sample. The lack of impact to the spring’s temperature from the 1990s earthquake is even more important because of the severe shaking sustained by this area on May 25, 1980. Three earthquakes on the Hilton Creek Fault near Mammoth Lakes, California (of magnitude 6.2, 5.0, and 5.9), took place between 8:33 AM and 11:44

Monitoring Thermal Springs

AM (Stover and Coffman, 1993). The shaking would have affected both the Blayney Meadow (Muir Trail Ranch) and Mono Hot Springs areas. Yet the temperatures taken by Mariner and others (1977) during their 1974 field work and those we took in 2015 at Muir Trail Ranch and those taken at Mono Hot Spring 6 during our monitoring do not reflect any persistent temperature change. There are no records of any short-term effects from the May 1980 earthquakes, such as those reported for the thermal spring at Muir Trail Ranch, because they occurred prior to the seasonal occupancy of the area. This was likely the case also for observable short-term effects at any of the thermal springs in the Mono Hot Springs area, because the snow blocking Kaiser Pass Road prevented site access at that time. In order to robustly investigate effects of seismicity on thermal water temperature in the area, regular monitoring would be necessary. Nonetheless, the observations outlined here appear to indicate that earthquakes during the study period did not alter the temperature of thermal water issuing from either Mono Hot Springs or the hot springs at Blayney Meadow. Current monitoring data for Mono Hot Springs demonstrate a slight cooling trend of ∼2 to 3◦ C over the decade between the initial and later monitoring periods (Figure 6), though some springs seem more stable than that. Data from 1974 indicate a temperature of 43◦ C for spring 6 (Mariner et al, 1977), slightly lower than that of the 1999–2001 monitoring period and identical to its temperature during the 2011–2013 monitoring period. Similarly, temperatures at one of the thermal springs at Blayney Meadow (Muir Trail Ranch) upstream from Mono Hot Springs was also stable between 1974 (43◦ C; Mariner et al., 1977) and our monitoring on July 16, 2015 (42.4◦ C). The stability of temperatures at the hottest springs (Mono Hot Spring 6 and Blayney Meadow/Muir Trail Ranch) over three decades (1974– 2015) compared to the decade-long trend of about 2 to 3◦ C cooling for the group of Mono Hot Springs suggests other possibilities. Differences in near-surface mixing with meteoric waters may be a cause, as could changes in the nearsurface flow paths of the lower-temperature springs. The dormancy or extinction of activity at some Mono Crossing thermal spring features clearly signals temporal variability at that locality as well. Robust conclusions about the temporal evolution of the thermal springs in the study area await continued monitoring through seismicity, changes in surface water availability (e.g., drought), and other uncontrolled variability. Shallow cooler groundwater is more available in the springtime after snowmelt, based on surface observations. Yet thermal spring temperatures are higher during springtime than during fall after the long, dry summer. This apparent contradiction suggests

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

181

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

181

De Graff, Pluhar, Gallegos, Takenaka, and Platt

that increased deep groundwater flow occurs due to seasonal-runoff–triggered aquifer pressure changes. Testing this hypothesis would likely require flow data for the thermal springs during spring and fall. At this time, flow data on spring 6 taken at the time of sampling in 1974 are the only data available (Mariner et al., 1977). While understanding how the Mono Hot Springs are influenced by groundwater circulation in the surrounding aquifer is desirable, it is outside the scope of this initial study. CONCLUSIONS Our findings show that the individual springs at Mono Hot Springs display consistent temperature and pH relationships seasonally and over multiple years. Primary control is exerted by heating at depth and groundwater flow through underground conduits. As guidance for management of this natural resource, passive catchment of the thermal spring outflow for recreational uses is unlikely to alter these relationships. Consequently, the attributes of the springs that have sustained Native American spiritual values should persist under such current uses. Management actions that might introduce additional surface water into the thermal spring area, such as modifying existing road drainage, should be avoided, as additional surface water mixing could decrease some thermal spring temperatures. Similarly, drilling wells and pumping water from within the spring area or affecting the conduits along which hot water rises obviously could alter the thermal spring conditions. An important aspect of managing a natural resource is understanding how, or if, it varies over time and what factors influence this normal variability. Such basic data enable a better analysis of how management action may or may not alter the natural operation of the resource. In the case of Mono Hot Springs, we needed to understand the thermal character of the springs. The springs in the Mono Hot Springs area are now known to range in temperature from a high of 44.5◦ C to a low of 24.3◦ C and from a pH of 8.0 to 7.05. Some springs display a very consistent annual temperature and others show significant variability. The decade between periods of data gathering showed a 2 to 3◦ C decrease in average thermal spring temperatures and an increase in pH. However, this trend may only be a minor variation because this decade falls within the three-decade trend, represented by Mono Hot Spring 6 and Blayney Meadow (Muir Trail Ranch) hot spring, showing a constant temperature trend. Both data-gathering periods detected a statistically verified seasonal (late spring vs. early fall) difference in the temperatures, but not in pH. Because the temperature variability is in the water outflow from the springs, the mixing of shallow cooler 182

groundwater with rising thermal water seems a likely mechanism. Combined data sets suggest that the Mono Hot Springs may be caused by tectonic activity along a minor fault system within the Sierra Nevada Microplate. Mapped Quaternary active normal faults and earthquake epicenters with dominantly normal-fault focal mechanisms coincide along a lineament that includes geothermal activity at Mammoth Mountain, Fish Creek/Iva Bell Hot Spring, and Mono Hot Springs (Figures 1 and 2A). Finally, assuming water-rock equilibrium, the temperature of the heat source generating these thermal springs is at least 74–79◦ C. The temperatures computed using cation geothermometers produced the same source temperature for one of the sampled springs as an investigation by the U.S. Geological Survey conducted 39 years earlier. The difference in calculated heat source temperature between cation and silica geothermometers is indicative of shallow groundwater mixing with rising hot water before flowing from the thermal springs. The pathways permitting meteoric water to circulate at depth to this heat source can be temporarily impacted by earthquake activity and shaking. However, the Blayney Meadow (Muir Trail Ranch) spring observation described above suggests that changes only last a few days at most. Future monitoring in support of land management decision-making should not be limited to only one or two springs. The 1974 sampling of Mono Hot Spring 6 provided a valuable snapshot at that time. However, the monitoring of 11 springs in this study demonstrates the more comprehensive understanding of spring conditions made possible with a larger data set drawn over a longer monitoring period. Any future monitoring should take into account the seasonal differences detected and provide a means for continuing assessment of long-term temperature trends. Given the lack of data on flow rates from the springs, collecting flow data might provide valuable information to add to the temperature and pH data. ACKNOWLEDGMENTS The manuscript substantially benefited from reviews by Allen Glazner, Elizabeth Haddon, and an anonymous reviewer. REFERENCES ARGUS, D. F. AND GORDON, R. G., 1991, Current Sierra Nevada– North America motion from very long baseline interferometry: Implications for the kinematics of the western United States: Geology, Vol. 19, pp. 1085–1088. ARGUS, D. F. AND GORDON, R. G., 2001, Present tectonic motion across the Coast Ranges and San Andreas fault system in

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

De Graff, Pluhar, Gallegos, Takenaka, and Platt

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs Central California: Geological Society America Bulletin, Vol. 113, No. 12, pp. 1580–1592. BACON, C. R.; GIOVANNETTI, D. M.; DUFFIELD, W. A.; DALRYMPLE, G. B.; AND DRAKE, R. E., 1982, Age of the Coso Formation, Inyo County, California: U.S. Geological Survey Bulletin 1527, 18 p. BAILEY, R. A., 1989, Geologic Map of Long Valley Caldera, MonoInyo Craters Volcanic Chain, and Vicinity, Eastern California: U.S. Geological Survey Miscellaneous Investigations Map I1933, scale 1:62,500, p. 11. BARGAR, K. E., 1978, Geology and Thermal History of Mammoth Hot Springs, Yellowstone National Park, Wyoming: U.S. Geological Survey Bulletin 1444, 55 p. BATEMAN, P. C., 1965, Geologic Map of the Blackcap Mountain Quadrangle, Fresno County, California: U.S. Geological Survey Geologic Quadrangle Map GQ-428. BATEMAN, P. C.; LOCKWOOD, J. P.; AND LYDON, P. A., 1971, Geologic Map of the Kaiser Peak Quadrangle, Central Sierra Nevada, California: U.S. Geological Survey Geologic Quadrangle Map GQ-894. BERG, N. H.; GALLEGOS, A.; DELL, T., FRAZIER, J.; PROCTER, T.; SICKMAN, J.; AND ARBAUGH, M., 2005, A screening procedure for identifying acid-sensitive lakes from catchment characteristics: Environmental Monitoring Assessment, Vol. 105, No. 1–3, pp. 285–307. BERRY, G. W.; GRIM, P. J.; AND IKELMAN, J. A., 1980, Thermal Spring List for the United States: National Oceanic and Atmospheric Administration Key to Geophysical Records Documentation No. 12, 59 p. BRADY, R. J.; DUCEA, M. N.; KIDDER, S. B.. and SALEEBY, J. B., 2006, The distribution of radiogenic heat production as a function of depth in the Sierra Nevada Batholith, California: Lithos, Vol. 86, No. 3, pp. 229–244. BROSSY, C. C.; KELSON, K. I.; AMOS, C. B.; BALDWIN, J. N.; KOZLOWICZ, B.; SIMPSON, D.; TICCI, M. G.; LUTZ, A. T.; KOZACI, O.; STREIG, A.; TURNER, R.; AND ROSE, R., 2012, Map of the late Quaternary active Kern Canyon and Breckenridge faults, southern Sierra Nevada, California: Geosphere, Vol. 8, No. 3, pp. 581–591, 511. CHEENEY, R. F., 1983, Statistical Methods in Geology for Field and Lab Decisions: George Allen & Unwin, London, U.K. 169 p. CITY OF EL PASO DE ROBLES, 2007, Supplemental Environmental Assessment to the Programmatic Environmental Assessment (PEA) for Typical Recurring Actions Resulting from Flood, Earthquake, Fire, Rain, and Wind Disasters in California as Proposed by the Federal Emergency Management Agency, FEMA-1505-DR-CA: Electronic document, available at https://www.fema.gov/media-library-data/201307261609-20490-1143/pasoroblescityhall_sea.txt CSBSJU, 2016, Kolmogorov-Smirnov Test, Department of Physics, College of Saint Benedict Saint John’s University, Minnesota: Electronic document, available at http://www.physics.csbsju. edu/stats/KS-test.html D’AMORE, F. AND ARNÓRSSON, S., 2000, Geothermometry. In Arnórsson, S. (Editor), Isotopic and Chemical Techniques in Geothermal Exploration, Development, and Use: International Atomic Energy Agency, Vienna, Austria, pp. 152–199. DE GRAFF, J. V., 1979, Initiation of shallow mass movement by vegetative-type conversion: Geology, Vol. 7, pp. 426–429. DE GRAFF, J. V., 1982, An approach for avoiding damage to springs from shock waves generated during seismic exploration. In Nielson, D. L.(Editor), Overthrust Belt of Utah: Utah Geological Association Publication 10, pp. 163–166. DE GRAFF, J. V., 2011, Perspectives for systematic landslide monitoring: Environmental Engineering Geoscience, Vol. 17, No. 1, pp. 67–76.

Monitoring Thermal Springs

DE GRAFF, J. V. AND GALLEGOS, A. J., 1987, Measuring soil loss to address erosion control needs in natural resources management. In Proceedings of the International Erosion Control Association, Annual Meeting, Sparks, NV, pp. 267–276. DE GRAFF, J. V.; GALLEGOS, A. J.; REID, M. E.; LAHUSEN, R. G.; AND DENLINGER, R. P., 2015, Using monitoring and modeling to define the hazard posed by the reactivated Ferguson rock slide, Merced Canyon, California: Natural Hazards, Vol. 76, No. 2, pp. 769–789. DE GRAFF, J. V.; ROATH, B.; AND FRANKS, E., 2007, Monitoring to improve the understanding of herbicide fate and transport in the southern Sierra Nevada, California. In Furniss, M.; Clifton, C.; and Ronnenberg, K. (Editors), Advancing the Fundamental Sciences: Proceedings of the Forest Service National Earth Sciences Conference: San Diego, CA, October 18– 22, 2004, PNW-GTR-689, Pacific Northwest Research Station, USDA Forest Service, Portland, OR, Vol. 2, pp. 352–360. DE GRAFF, J. V. AND ROMESBURG, H. C., 1981, Subsidence crack closures: Rate, magnitude, and sequence: Bulletin International Association Engineering Geology, No. 23, pp. 123–127. DIXON, T. H.; MILLER, M.; FARINA, F.; WANG, H.; AND JOHNSON, D., 2000, Present-day motion of the Sierra Nevada block and some tectonic implications for the Basin and Range province, North American Cordillera: Tectonics, Vol. 19, No. 1, pp. 1–24. DIXON, T. H.; ROBAUDO, S.; LEE, J.; AND REHEIS, M. C., 1995, Constraints on present-day Basin and Range deformation from space geodesy: Tectonics, Vol. 14, No. 4, pp. 755–772. DONG, S.; UCARKUS, G.; WESNOUSKY, S.; MALONEY, J.; KENT, G.; DRISCOLL, N.; AND BASKIN, R.; 2014, Strike-slip faulting along the Wassuk Range of the northern Walker Lane, Nevada: Geosphere, Vol. 10, No. 1, pp. 40–48. DUCEA, M. AND SALEEBY, J. B., 1998, A case for delamination of the deep batholithic crust beneath the Sierra Nevada, California: International Geology Review, Vol. 40, No. 1, pp. 78–93. DUCEA, M. N. AND SALEEBY, J. B., 1996, Buoyancy sources for a large, unrooted mountain range, the Sierra Nevada, California; Evidence from xenolith thermobarometry: Journal Geophysical Research, Vol. 101, No. B4, pp. 8229–8244. FARMER, G. L.; GLAZNER, A. F.; AND MANLEY, C. R., 2002, Did lithospheric delamination trigger late Cenozoic potassic volcanism in the southern Sierra Nevada, California?: Geological Society America Bulletin, Vol. 114, No. 6, pp. 754–768. FAULDS, J. E.; COOLBAUGH, M.; BLEWITT, G.; AND HENRY, C. D., 2004, Why is Nevada in Hot Water? Structural controls and tectonic model of geothermal systems in the northwestern Great Basin: Geothermal Resources Council Transactions, Vol. 28, pp. 649–654. FETH, J. H.; ROGERS, S. M.; AND ROBERSON, C. E., 1964, Chemical Composition of Snows in the Northern Sierra Nevada and Other Areas: U.S. Geological Survey Water-Supply Paper 1535. Forest History SOCIETY, 2009, Introduction: Managing Multiple Uses on National Forests, 1905–1995: Electronic document, available at http://www.foresthistory.org/ASPNET/ Publications/multiple_use/chap1.htm FOURNIER, R. O., 1977, Chemical geothermometers and mixing models for geothermal systems: Geothermics, Vol. 5, pp. 41– 50. FOURNIER, R. O. AND TRUESDALL, A. H., 1973, An empirical NaK-Ca geothermometer for natural waters: Geochimica Cosmochimica Acta, Vol. 37, pp. 1255–1275. FRASSETTO, A. M.; ZANDT, G.; GILBERT, H.; OWENS, T. J.; AND JONES, C. H., 2011, Structure of the Sierra Nevada from receiver functions and implications for lithospheric foundering: Geosphere, Vol. 7, No. 4, pp. 898–921. FRAZER, R. E.; COLEMAN, D. S.; AND MILLS, R. D., 2014, Zircon U-Pb geochronology of the Mount Givens Granodiorite:

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

183

Central California: Geological Society America Bulletin, Vol. 113, No. 12, pp. 1580–1592. BACON, C. R.; GIOVANNETTI, D. M.; DUFFIELD, W. A.; DALRYMPLE, G. B.; AND DRAKE, R. E., 1982, Age of the Coso Formation, Inyo County, California: U.S. Geological Survey Bulletin 1527, 18 p. BAILEY, R. A., 1989, Geologic Map of Long Valley Caldera, MonoInyo Craters Volcanic Chain, and Vicinity, Eastern California: U.S. Geological Survey Miscellaneous Investigations Map I1933, scale 1:62,500, p. 11. BARGAR, K. E., 1978, Geology and Thermal History of Mammoth Hot Springs, Yellowstone National Park, Wyoming: U.S. Geological Survey Bulletin 1444, 55 p. BATEMAN, P. C., 1965, Geologic Map of the Blackcap Mountain Quadrangle, Fresno County, California: U.S. Geological Survey Geologic Quadrangle Map GQ-428. BATEMAN, P. C.; LOCKWOOD, J. P.; AND LYDON, P. A., 1971, Geologic Map of the Kaiser Peak Quadrangle, Central Sierra Nevada, California: U.S. Geological Survey Geologic Quadrangle Map GQ-894. BERG, N. H.; GALLEGOS, A.; DELL, T., FRAZIER, J.; PROCTER, T.; SICKMAN, J.; AND ARBAUGH, M., 2005, A screening procedure for identifying acid-sensitive lakes from catchment characteristics: Environmental Monitoring Assessment, Vol. 105, No. 1–3, pp. 285–307. BERRY, G. W.; GRIM, P. J.; AND IKELMAN, J. A., 1980, Thermal Spring List for the United States: National Oceanic and Atmospheric Administration Key to Geophysical Records Documentation No. 12, 59 p. BRADY, R. J.; DUCEA, M. N.; KIDDER, S. B.. and SALEEBY, J. B., 2006, The distribution of radiogenic heat production as a function of depth in the Sierra Nevada Batholith, California: Lithos, Vol. 86, No. 3, pp. 229–244. BROSSY, C. C.; KELSON, K. I.; AMOS, C. B.; BALDWIN, J. N.; KOZLOWICZ, B.; SIMPSON, D.; TICCI, M. G.; LUTZ, A. T.; KOZACI, O.; STREIG, A.; TURNER, R.; AND ROSE, R., 2012, Map of the late Quaternary active Kern Canyon and Breckenridge faults, southern Sierra Nevada, California: Geosphere, Vol. 8, No. 3, pp. 581–591, 511. CHEENEY, R. F., 1983, Statistical Methods in Geology for Field and Lab Decisions: George Allen & Unwin, London, U.K. 169 p. CITY OF EL PASO DE ROBLES, 2007, Supplemental Environmental Assessment to the Programmatic Environmental Assessment (PEA) for Typical Recurring Actions Resulting from Flood, Earthquake, Fire, Rain, and Wind Disasters in California as Proposed by the Federal Emergency Management Agency, FEMA-1505-DR-CA: Electronic document, available at https://www.fema.gov/media-library-data/201307261609-20490-1143/pasoroblescityhall_sea.txt CSBSJU, 2016, Kolmogorov-Smirnov Test, Department of Physics, College of Saint Benedict Saint John’s University, Minnesota: Electronic document, available at http://www.physics.csbsju. edu/stats/KS-test.html D’AMORE, F. AND ARNÓRSSON, S., 2000, Geothermometry. In Arnórsson, S. (Editor), Isotopic and Chemical Techniques in Geothermal Exploration, Development, and Use: International Atomic Energy Agency, Vienna, Austria, pp. 152–199. DE GRAFF, J. V., 1979, Initiation of shallow mass movement by vegetative-type conversion: Geology, Vol. 7, pp. 426–429. DE GRAFF, J. V., 1982, An approach for avoiding damage to springs from shock waves generated during seismic exploration. In Nielson, D. L.(Editor), Overthrust Belt of Utah: Utah Geological Association Publication 10, pp. 163–166. DE GRAFF, J. V., 2011, Perspectives for systematic landslide monitoring: Environmental Engineering Geoscience, Vol. 17, No. 1, pp. 67–76.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

183

De Graff, Pluhar, Gallegos, Takenaka, and Platt Implications for the genesis of large volumes of eruptible magma: Journal Geophysical Research: Solid Earth, Vol. 119, No. 4, pp. 2907–2924. GIGGENBACH, G. F., 1988, Geothermal solute equilibria: Geochimica Cosmochimica Acta, Vol. 52, pp. 2749–2765. HILDRETH, W. AND FIERSTEIN, J., 2016, Eruptive History of Mammoth Mountain and its Mafic Periphery, California: U.S. Geological Survey Professional Paper 1812, 128 p. HILDRETH, W.; FIERSTEIN, J.; CHAMPION, D.; AND CALVERT, A., 2014, Mammoth Mountain and its mafic periphery—A late Quaternary volcanic field in eastern California: Geosphere, Vol. 10, No. 6, pp. 1315–1365. HILL, D. P. AND PREJEAN, S., 2005, Magmatic unrest beneath mammoth mountain, California: Journal Volcanology Geothermal Research, Vol. 146, No. 4, pp. 257–283. HILTON, D. R., 1996, The helium and carbon isotope systematics of a continental geothermal system: Results from monitoring studies at Long Valley caldera (California, USA): Chemical Geology, Vol. 127, No. 4, pp. 269–295. JAGER, H. I. AND ROSE, K. A., 2003, Designing optimum flow patterns for fall Chinook Salmon in a Central Valley, California river: North American Journal Fisheries Management, Vol. 23, pp. 1–21. JONES, C. H.; LANG FARMER, G.; AND UNRUH, J., 2004, Tectonics of Pliocene removal of lithosphere of the Sierra Nevada, California: Geological Society America Bulletin, Vol. 116, No. 11–12, pp. 1408–1422. KOLESAR, P. T. AND DE GRAFF, J. V., 1978, A comparison of the silica and Na-K-Ca geothermometers for thermal springs in Utah: Geothermics, Vol. 6, pp. 221–226. KREEMER, C.; BLEWITT, G.; AND HAMMOND, W. C., 2009, Geodetic constraints on contemporary deformation in the northern Walker Lane: 2. Velocity and strain rate tensor analysis. In Oldow, J. S. and Cashman, P. H. (Editors), Late Cenozoic Structure and Evolution of the Great Basin-Sierra Nevada Transition, Volume 447, Geological Society of America, Boulder, CO. pp. 17–31. LANGBEIN, J. O.; HILL, D. P.; PARKER, T. N.; AND WILKINSON, S. K., 1993, An episode of reinflation of the Long Valley Caldera, eastern California; 1989–1991: Journal Geophysical Research, Vol. 98, No. B9, pp. 15851–15870. LOCKWOOD, J. P.; BATEMAN, P. C.; AND SULLIVAN, J. S., 1972, Mineral Resource Evaluation of the U.S. Forest Service Sierra Demonstration Project Area, Sierra National Forest, California: U.S. Geological Survey Professional Paper 714, 30 p. LOCKWOOD, J. P. AND LYDON, P. A., 1975, Geologic Map of the Mount Abbot Quadrangle, Central Sierra Nevada, California: U.S. Geological Survey Geologic Quadrangle Map GQ-1155. Mammoth Pacific, 2017, Welcome to the Mammoth Geothermal Complex Website: Electronic document, available at http://www.mammothpacific.com/ MANLEY, C. R.; GLAZNER, A. F.; AND FARMER, G. L., 2000, Timing of volcanism in the Sierra Nevada of California; Evidence for Pliocene delamination of the batholithic root?: Geology, Vol. 28, No. 9, pp. 811–814. MARINER, R. H.; PRESSER, E. S.; AND EVANS W. C., 1977, Hot Springs of the Central Sierra Nevada, California: U.S. Geological Survey Open-File Report 77-559, 27 p. MONASTERO, F. C., 2002, Model for success: An overview of industry-military cooperation in the development of power operations at the Coso Geothermal Field in Southern California: Geothermal Resources Council Bulletin, Vol. 31, No. 5, pp. 188–195. MOORE, J. G. AND SISSON, T. W., 1985, Geologic Map of the Kern Peak Quadrangle, Tulare County, California: U.S. Geological Survey, Reston, VA, p. 1 sheet.

184

NATIONAL CENTERS FOR ENVIRONMENTAL INFORMATION (NCEI), Thermal Springs in the US: Electronic document, available at https://maps.ngdc.noaa.gov/viewers/hot_springs/, National Ocean and Atmospheric Administration. Neuendorf, K. K. E.; Mehl, J. P. Jr.; and Jackson, J. A. (Editors), 2011, Glossary of Geology, Fifth Edition, Revised: American Geosciences Institute, Alexandria, VA. OERTER, E. J., 2011, Geothermometry of Thermal Springs in the Rico, Dunton, and West Fork Dolores River Areas, Dolores County, Colorado: Colorado Geological Survey, Denver, CO, 56 p. OLDOW, J. S.; GEISSMAN, J. W.; AND STOCKLI, D. F., 2008, Evolution and strain reorganization within late Neogene structural stepovers linking the central Walker Lane and northern Eastern California shear zone, western Great Basin: International Geology Review, Vol. 50, No. 3, pp. 270–290. PHILLIPS, F. M., 2016, Cosmogenic nuclide data sets from the Sierra Nevada, California, for assessment of nuclide production models: I. Late Pleistocene glacial chronology: Quaternary Geomorphology, Vol. 35, pp. 119–129. ROSE, G., 1994, Sierra Centennial: Three Forest Interpretative Association, Auberry, CA. 183 p. SALTUS, R. W. AND LACHENBRUCH, A. H., 1991, Thermal evolution of the Sierra Nevada; Tectonic implications of new heat flow data: Tectonics, Vol. 10, No. 2, pp. 325–344. SARNA-WOJCICKI, A. M.; PRINGLE, M. S.; AND WIJBRANS, J., 2000, New (super 40) Ar/(super 39) Ar age of the Bishop Tuff from multiple sites and sediment rate calibration for the MatuyamaBrunhes boundary: Journal Geophysical Research, Vol. 105, No. B9, p. 21. SIERRA NEVADA GEOTOURISM, 2017, Southern California Edison Big Creek Hydroelectric Project: Electronic document, available at https://www.sierranevadageotourism.org/content/southerncalifornia-edison-big-creek-hydroelectric-project/ sie5ee9ccaf96c262c3c SOREY, M. L.; FARRAR, C. D.; GERLACH, T. M.; MCGEE, K. A.; EVANS, W. C.; COLVARD, E. M.; HILL, D. P.; BAILEY, R. A.; ROGIE, J. D.; HENDLEY, J. W., II; AND STAUFFER, P. H., 2000, Invisible CO2 Gas Killing Trees at Mammoth Mountain, California: Fact Sheet 172-96 (Version 2.0): U.S. Geological Survey, 2 p. SURPLESS, B. E.; STOCKLI, D. F.; DUMITRU, T. A.; AND MILLER, E. L., 2002, Two-phase westward encroachment of Basin and Range extension into the northern Sierra Nevada: Tectonics, Vol. 21, No. 1, pp. 2-1-2-10. TEMPLETON, D. C. AND DREGER, D. S., 2006, Non-double-couple earthquakes in the Long Valley Volcanic Region: Bulletin Seismological Society America, Vol. 96, No. 1, pp. 69–79. TOBISCH, O. T.; SALEEBY, J. B.; RENNE, P. R.; MCNULTY, B.; AND TONG, W., 1995, Variations in deformation fields during development of a large-volume magmatic arc, central Sierra Nevada, California: Geological Society America Bulletin, Vol. 107, No. 2, pp. 148–166. UNRUH, J. AND HAUKSSON, E., 2009, Seismotectonics of an evolving intracontinental plate boundary, southeastern California. In Oldow, J. S. and Cashman, P. H. (Editors), Late Cenozoic Structure and Evolution of the Great Basin–Sierra Nevada Transition: Geological Society of America Special Paper 447, pp. 351–372. UNRUH, J.; HAUKSSON, E.; AND JONES, C. H., 2014, Internal deformation of the Southern Sierra Nevada Microplate associated with foundering lower lithosphere, California: Geosphere, Vol. 10, No. 1, pp. 107–128. UNRUH, J.; HUMPHREY, J.; AND BARRON, A., 2003, Transtensional model for the Sierra Nevada frontal fault system, eastern California: Geology, Vol. 31, No. 4, pp. 327–330.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

184

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

Monitoring Thermal Springs U.S. GEOLOGICAL SURVEY (USGS) AND CALIFORNIA GEOLOGICAL SURVEY (GSC), 2006, Quaternary Fault and Fold Database for the United States: U.S. Geological Survey Web Site: Electronic document, available at http//earthquake.usgs.gov/ hazards/qfaults/ USGS, 2017, Advanced National Seismic System (ANSS) Comprehensive Earthquake Catalog (ComCat): U.S. Geological Survey: Electronic document, available at https://earthquake. usgs.gov/earthquakes/ WAKABAYASHI, J. AND SAWYER, T. L., 2000, Neotectonics of the Sierra Nevada and the Sierra Nevada-Basin and Range transition, California, with field trip stop descriptions for the northeastern Sierra Nevada. In Brooks, E. R. and Dida, L. T. (Editors), Field Guide to the Geology and Tectonics of the Northern

Monitoring Thermal Springs

Sierra Nevada: California Division of Mines and Geology Special Publication 122, pp. 173–212. WAKABAYASHI, J. AND SAWYER, T. L., 2001, Stream incision, tectonics, uplift, and evolution of topography of the Sierra Nevada, California: Journal Geology, Vol. 109, No. 5, pp. 539–562. WANG, C. Y.; MANGA, M.; DREGER, D.; AND WONG, A., 2004, Streamflow increase due to rupturing of hydrothermal reservoirs: Evidence from the 2003 San Simeon, California, Earthquake: Geophysical Research Letters, Vol. 31, No. 10. WILLIAMS, C. F.; REED, M. J.; AND MARINER, R. H., 2008, A Review of Methods Applied by the U.S. Geological Survey in the Assessment of Identified Geothermal Resources: U.S. Geological Survey Open-File Report 2008-1296, 27 p.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

185

U.S. GEOLOGICAL SURVEY (USGS) AND CALIFORNIA GEOLOGICAL SURVEY (GSC), 2006, Quaternary Fault and Fold Database for the United States: U.S. Geological Survey Web Site: Electronic document, available at http//earthquake.usgs.gov/ hazards/qfaults/ USGS, 2017, Advanced National Seismic System (ANSS) Comprehensive Earthquake Catalog (ComCat): U.S. Geological Survey: Electronic document, available at https://earthquake. usgs.gov/earthquakes/ WAKABAYASHI, J. AND SAWYER, T. L., 2000, Neotectonics of the Sierra Nevada and the Sierra Nevada-Basin and Range transition, California, with field trip stop descriptions for the northeastern Sierra Nevada. In Brooks, E. R. and Dida, L. T. (Editors), Field Guide to the Geology and Tectonics of the Northern

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 165–185

185

Landslide Interpretation of the Northeast Flank of Kohala Volcano, Hawaii

KIM M. BISHOP1

Department of Geosciences and Environment, California State University–Los Angeles, 5151 State University Drive, Los Angeles, CA 90032

Key Terms: Volcano, Landslide, Engineering Geology, Kohala, Hawaii, Pololu Landslide ABSTRACT The Hawaiian Island volcanic edifices have shed at least 15 giant submarine landslides, each classified as either a slump or debris avalanche. Controversy exists regarding the number, size, and type of landslides on the northeast flank of Kohala Volcano. This study provides a new interpretation for the Kohala flank based on contour and balanced cross-section analysis. Specifically, contours indicate that there is a landslide extending from the summit to the coast between Pololu and Waipio Valleys. The contour evidence also shows that the slide plane is planar and dips less steeply than the topographic slope. Balanced cross sections show the slide plane to be approximately 950 m deep immediately downhill from the zone of depletion, and the slide plane presumably reaches the surface at the base of the coastal cliffs on the northeast coast of Kohala mountain. The lower part of the landslide once extended from the coast to approximately 10 km offshore, but this portion now has been completely removed, apparently as a debris avalanche. Removal of this distal landslide mass created a 200 to 450 m headwall that is now topographically represented by sea cliffs. This newly identified slide/debris avalanche is informally named the “Kohala landslide.” Based on cross-cutting relations of landslide faults with Hawi series lava flows, the upper slide part of the landslide moved sometime between 270 and 60 ka. The age of the lower, debris avalanche part is even less certain and depends on whether canyons cut in the seafloor after the avalanche movement were eroded in the subaerial or submarine environment. INTRODUCTION Giant landslide deposits derived from the volcanic flanks of the Hawaiian Islands mantle the seafloor surrounding the island chain (Figure 1; Moore et al., 1989). The landslides, classified as slumps and debris

1 Corresponding

author email: kbishop@ calstatela.edu

avalanches, individually cover hundreds to thousands of square kilometers (Moore et al., 1989). Slumps, the less common of the two types, consist of thick (a kilometer or more), coherent blocks that apparently move intermittently, a few meters at a time, over many years (Moore et al., 1994, 1989). In contrast, debris avalanches are catastrophically emplaced landslides that result in relatively thin deposits averaging a few tens of meters thick and consist of strongly brecciated rock. Rapid emplacement is indicated by runout lengths measuring tens to hundreds of kilometers from the source area (Moore et al., 1994, 1989). An area of controversy regarding the number, size, and types of landslides is on the northeast flank of Kohala Volcano at the north end of Hawaii Island (Figures 1 and 2). Initially, Moore et al. (1989) and Moore and Clague (1992) hypothesized the presence of a single debris avalanche with its head at Kohala Volcano’s summit and its toe 130 km offshore (Figures 1 and 3A). Moore et al. (1994) re-interpreted this feature such that the head is at Kohala’s coast, and the proximal one third of the landslide consists of slump rather than debris avalanche material. From improved bathymetric data, Smith et al. (2002) proposed that the northeast Kohala area contains four separate landslides, including two slumps, which were named by them as the Pololu and Laupahoehoe slumps, and the deposits of two unnamed debris avalanches further offshore (Figures 1 and 3B), sourced from the southern Maui and eastern Hawaii Island edifices, respectively. Smith et al. (2002) suggested the possibility of another debris avalanche with source area immediately offshore of Kohala Volcano (Figure 3C). All of these various landslide interpretations were based solely on morphologic observations. In contrast, this paper characterizes and, where possible, quantifies the Kohala flank landslides using contour line observations and balanced cross sections. Accordingly, the “Kohala slide/debris avalanche” extends from the volcano summit to about 10 km offshore. Contour maps indicate the existence of a translational landslide on the subaerial slope northeast of Kohala, and contour lines and balanced cross sections document the geometry, depth, and extent of the slide plane. Evidence is presented suggesting that the lower part of the

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

187

1 Corresponding

author email: kbishop@ calstatela.edu

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

187

Bishop

Figure 1. Map of the Hawaiian Islands showing the study area and limits of numerous giant submarine landslides. Note the differing landslide interpretations within and near the study area for the Pololu debris avalanche (Moore et al, 1989; dashed lines) and the Pololu and Laupahoehoe slumps (Smith et al., 2007; dotted lines). The ornamented area with large dots at the distal end of the Pololu debris avalanche designates an area of hummocky topography. The limits of all other giant landslides are re-drawn from Moore et al. (1989).

landslide was completely removed when mobilized as a debris avalanche.

Pertinent Kohala Volcano Geology and Geomorphology Several geologic and geomorphic features of Kohala Volcano, including faults, deeply eroded valleys, coastal cliffs, and a coastal embayment, were identified in previous studies as important to landslide interpretation of the area (Moore et al., 1989, 1994; Smith et al., 2002; Lamb et al., 2007; and Sherrod et al., 2007). As will be 188

shown, all of these features are also germane to this study’s conclusions. Several normal faults that define graben and halfgraben structures are present at the summit of Kohala Volcano (Figure 4). In most places, the faults cut the Hawi series lava flows and vents, which are the youngest magmatic extrusions on the volcano (Wolfe and Morris, 1996; Sherrod et al., 2007), indicating relatively recent extensional deformation of the mountain. Moore et al. (1989) and Moore and Clague (1992) proposed that the faults formed in the pull-away zone at the head of their hypothesized Pololu debris avalanche

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 2. Image of Kohala Volcano with view toward the southwest (from Google Earth, 2017).

(Figure 1). Other suggestions for the origin of the faults include that they are the remnants of ancient caldera walls (MacDonald et al., 1983), the result of volcanic rifting associated with volcanism (Smith et al., 2002), or the product of far-field gravitational instability created by movement of a giant slump with a headwall at Kohala Volcano’s coastline (Sherrod et al., 2007). Another pertinent geomorphic feature is an embayment that is 20 km long by 2 km wide bordered by cliffs 200 to 450 m high along the northeast Kohala coast (Figure 4). The presence of such high cliffs strongly contrasts to the rest of the island’s coastline, where coastal cliffs rarely exceed 30 m. The embayment has been interpreted as the head of a giant landslide with its headwall forming the cliffs (Moore et al., 1989, 1994; Smith et al., 2002; Eakins and Robinson, 2006; Lamb et al., 2007; and Sherrod et al., 2007). Other features that have been proposed as landsliderelated are deep valleys eroded in the volcanic slopes above the embayment. From north to south, the most prominent are Pololu, Honokane, Waimanu, and Waipio Valleys (Figure 4). Pololu and Waipio Valleys discharge to the ocean at the north and south ends of the coastal embayment, respectively. Near Kohala’s summit, Honokane Valley, which is adjacent to Pololu Valley at the north end of the embayment, curves southward, oblique to the mountain’s downhill direction.

Waipio Valley, at the south end, mirrors Honokane Valley by curving northward. In the interpretation of Moore et al. (1989), Pololu and Waipio Valleys formed at the lateral boundaries of their proposed Pololu debris avalanche, and Honokane and Waipio Valleys curve laterally across the slopes at the landslide head. Lamb et al. (2007) alternatively interpreted all of the large valleys solely as the result of stream incision resulting from base-level change occurring when the coastal cliffs were formed as the headwall of the giant Pololu slump. CONTOUR AND BALANCED CROSS-SECTION ANALYSIS Landslide Evidence from Contour Map Analysis The contour maps presented in Figures 4 and 5 provide evidence for the existence and geometry of a giant landslide on Kohala Volcano’s northeast subaerial flank. Comparisons of contour lines north and south of Pololu/Honokane Valleys show that the positions of contours south of Pololu/Honokane Valleys are shifted seaward relative to those north of the valley (Figure 5). Similarly, contours north of Waipio Valley are shifted seaward relative to those south of the valley. The simplest explanation for the slope mismatches

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

189

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

189

Bishop

Figure 3. Maps that depict previous landslide proposals for the northeast flank of Kohala Volcano. (A) Pololu debris avalanche. (Moore et al., 1989). (B) Pololu slump (Smith et al., 2002) with cross section. (C) Un-named debris avalanche (Smith et al., 2002) with cross section. Dashed line indicates proposed travel path.

across Pololu/Honokane and Waipio Valleys demonstrated by the contour shifts is past landslide movement of the volcanic slope between the valleys. Support for this explanation is the presence of normal faults at the summit, which are explained as the result of pull-apart movement at the landslide’s head. The contour shifts across Waipio Valley are less obvious than for those of Pololu/Honokane Valleys partly because most of the south rim of Waipio Valley has received lava flows from Mauna Kea that overlap older Kohala flows (Figure 4). The Mauna Kea flows affect the topography that

190

would otherwise be formed exclusively from Kohala’s lava flows, thereby partially masking the contour shifts. Figure 4 provides evidence as to whether the landslide is translational (i.e., planar slide plane) or rotational (i.e., curved slide plane). The slope gradients along line B on the landslide surface and along line C on the intact slope on the opposite side of the volcano are both 8 degrees. With the equal slope gradients, there is no evidence for rotation of the landslide’s topographic surface, thereby leading to the interpretation of a translational landslide and planar slide plane.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

190

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 4. Geologic map of Kohala volcano re-drawn from Sherrod et al. (2007) and Wolfe and Morris (1996).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

191

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

191

Bishop

Figure 5. Contour maps at 200 ft (61 m) contour intervals, illustrating the seaward shift of contour lines between Pololu and Waipio Valleys on Kohala Volcano’s northeast subaerial flank. The contour maps are re-drawn from Sherrod et al. (2007). (A) Contour lines between sea level and 5,200 ft (1,585 m). (B) Contour lines between 1,000 and 1,800 ft (305 and 549 m). (C) Contour lines between 2,000 and 2,800 ft (610 and 854 m). (D) Contour lines between 3,000 and 4,000 ft (915 and 1,220 m).

The contour line shift direction for the landslide surface provides information as to the slide plane gradient. As shown in Figure 6, a contour line shift in the direction of landslide movement, as is the case for the landslide on Kohala’s slope, results when the slide plane is less steep than the topographic slope. Because the volcanic slope gradient across most of the landslide’s surface is about 8 degrees, the slide plane’s dip must be less than 8 degrees seaward. Slide Plane Depth Depth Determination by Cross-Section Area Balance An estimate of the Kohala mountain landslide depth and geometry can be determined using 192

balanced cross-section concepts. A balanced cross section depicts a deformed earth mass that can be retro-deformed in a kinematically plausible manner to reconstruct the original pre-deformation configuration (Dahlstrom, 1969; Woodward et al., 1989). For a balanced deformed/retro-deformed cross-section pair, an important constraint to be met is conservation of mass (Suppe, 1985). Assuming constant density and no movement of material in or out of the line of cross section during deformation, conservation of mass equates to conservation of area on a balanced cross-section pair. Deformed/retrodeformed cross-section pairs meeting the criteria for balance represent permissible, though not necessarily accurate, geologic interpretations (Dahlstrom, 1969; Rowan and Kligfield, 1989; and Fossen, 2010).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 6. Diagram illustrating contour line shift caused by movement of a translational landslide where the slide plane is less steep than the topographic surface. Point b moves to point b� in the identical direction and distance as the movement of point a to a� caused by displacement, D. Similar movement of all points on the surface of the landslide causes the displacement of the slide’s surface as shown. A contour line at point c will move in the direction of landslide movement to point c� . In contrast to the situation diagramed, if the slide plane and topographic surface are parallel, there will be no contour line shift, and if the slide plane is steeper than the topographic surface, contour lines will shift in the opposite direction of the landslide displacement direction.

Cross sections not meeting balance criteria cannot be valid. Construction of a balanced cross section is an iterative process (Davis and Reynolds, 1996). The first step is to interpret the subsurface geometry of a deformed earth mass based on surface and, if available, subsurface data. The cross-sectional mass is then conceptually retro-deformed to an un-deformed state assuming realistic kinematic processes. If the pre- and post-deformation cross-section interpretations violate the balance criteria, then the deformed state cross section must be adjusted, retro-deformed once again, and re-checked for balance. These steps are continued until cross-section balance is achieved. For retro-deformation of extensionally deformed earth masses where internal deformation occurs in the hanging walls of high-angle normal faults that merge with a basal low-angle normal fault (which is the case in the head area of many translational landslides, including the landslide of this study, where the low-angle normal fault is represented by the planar slide plane and the high-angle normal faults are represented by slip surfaces in the pull-away zone), one of three deformation styles is commonly modeled (Rowan and Kligfield, 1989; Fossen, 2010). Internal deformation is

Landslide Interpretation, Kohala Volcano, Hawaii

assumed to result from gravitational collapse of the hanging wall along high-angle faults into the void that would be created as the footwall block pulls away laterally (Hamblin, 1965; de Matos, 1993), an example of which is shown in Figure 7. One of the three deformation styles treat the hanging-wall blocks as single rigid blocks that translate and/or rotate during hangingwall collapse (Fossen, 2010). For the other two styles, one involves vertical shear and the other involves 60 degree antithetic-dipping shear (Rowan and Kligfield, 1989; Fossen, 2010), with the shear deformation in both cases accommodated by strain and/or internal faulting (Figure 7). Experience has shown that either vertical or 60 degree antithetic-dipping shear is generally adequate for modelling the shear deformation (Fossen, 2010). Using a mass balance, the depth of the slide plane for a translational landslide can be calculated for a given cross section drawn parallel to the direction of landslide movement using the concepts depicted in Figure 8 (e.g., Hansen, 1965; Bishop, 1999). As shown, two parameters needed for the calculation of slide plane depth, H, are the following: (1) the area of the zone of depletion (more accurately referred to as the area of depletion for a cross section), A, and (2) the landslide displacement, D. The principle of mass conservation results in the equation: A = D × H,

(1)

where A is cross-sectional value for the area of depletion; D is landslide displacement magnitude above the planar part of the slide plane at the point just downslope from the area of depletion; and H is the depth of the slide plane just downslope from the area of depletion. Re-arrangement of Eq. 1 allows the determination of the slide plane depth, H, from the equation: H = A/D.

(2)

The size of the area of depletion is determined by scaled measurement of the cross section. Figure 9 presents a cross section of Kohala Volcano showing the area of depletion defined from the existing topographic profile and estimation of the pre-landslide topographic surface location. This area of depletion is measured to be 232,900 m2 . Determination of displacement, D, is achieved using the cross-section balancing process. For this study, a graphical approach was utilized for retro-deformation of the landslide mass. Retro-deformation was performed by dividing the landslide mass into blocks and then sequentially moving the blocks uphill such that the area of depletion is re-filled with landslide material. From the procedure, the magnitude of retrodisplacement needed to re-fill the area of depletion is

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

193

(1)

(2)

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

193

Bishop

Figure 7. Concept of creation of the area of depletion by gravitational hanging-wall collapse as the footwall moves away from the hanging wall of a fault during extensional deformation. In the example shown, collapse occurs by antithetic shear on planes dipping 60 degrees.

determined. This value defines the slide displacement, D. The line of cross section for the retro-deformation analysis is presented in Figure 4, and the topographic profile is shown in Figure 9. Along the profile line, the dominant deformation structures at the landslide

Figure 8. Concept of a mass balance for translational landslides with a pull-away head area and consequent area of depletion. Based on the mass balance and assumption of constant density, the area of depletion, A, equals rectangular area B, which is the product of landslide displacement, D and the slide plane depth, H.

194

head are three half grabens developed in the hanging walls of faults labeled F1, F2, and F3 (Figure 9). The area of depletion was mainly created by subsidence of these three half grabens. Although other mapped landslide faults are present in the line of cross section (Figure 4), their displacements are sufficiently small that surficial offsets are not discernible at the scale of the cross section, and the faults were ignored. Because the mass configuration of a balanced crosssection pair is permissible but not necessarily accurate, the slide plane depth determined from a single balanced cross-section analysis cannot be assumed to represent the actual value of the depth. Instead, the depth is only one possibility in a range of values. In order for balanced cross-section analysis to provide meaningful results, the range of depths from multiple analyses using different deformation styles must be determined. For this study four different deformation styles were used to determine the slide plane depth. Retro-deformation analyses were run using each of the three deformation styles previously mentioned (single rigid blocks, vertical shear, and 60 degree antithetic shear), plus one additional style with 45 degree antithetic shear as an added slide plane depth determination. In the first analysis, the three subsided half-graben blocks were modeled as large blocks that underwent translation and rotation (Figure 10). The hanging-wall

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

194

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 9. Cross section showing the area of depletion existing at the summit of Kohala Volcano.

blocks of faults F2 and F3 were modeled as single blocks, whereas the hanging-wall block of fault F1 was modeled as two blocks, because the retro-deformed geometry resulted in a poor fit with the topographic surface if only a single block was used. For the other three analyses, shear deformation was modeled in the hanging-wall blocks of faults F1 and F2 and, in one case, F1, F2, and F3 (Figures 11–13). To model the shear, the hanging-wall blocks were divided into a number of mostly long, slender blocks that dipped vertically in one case (Figure 11) and 60 degrees and 45 degrees antithetically to the underlying faults in the other two cases (Figures 12 and 13). The first step in the retro-deformation procedure for each model was to assume a slide plane depth and dip. As described earlier, the main, translational slide plane surface dips less steeply than the 8 degree coastward dip of the topographic surface. In order to meet this shallow dip requirement for the slide plane and yet have a positive seaward dip that provides at least a modicum of gravitational driving force for the landslide, a slide plane dip of 3 degrees was used in each model. After the slide plane was drawn on a cross section to be retrodeformed, the internal blocks and faults F1, F2, and F3 were added. Also, for each cross section to be retrodeformed, a reference line was marked on the main

translational landslide block, which is the block that extends from the area of depletion to the coast. Starting with the most proximal block of the half graben in the hanging wall of fault F1, the landslide was then retro-deformed by sequentially “un-sliding” each block so that the block’s topographic surface was aligned as best as possible with the pre-landslide topographic surface. After this half graben was retrodeformed, the F1 footwall block was un-slipped to close the gap between the F1 footwall block and the already retro-deformed half-graben mass. Using the same procedure, the half grabens and footwall blocks for faults F2 and F3 were retro-deformed. Once the footwall block of fault F3 was retro-deformed, the process was complete. The distance that the reference line moved upslope from its position before retro-deformation was scaled in the direction parallel to the slide plane. This retro-deformation distance was noted as the landslide displacement. Using the displacement value, the depth of the landslide slide plane was calculated by Eq. 2. The calculated depth was then compared to the depth that was initially assumed in the model trial. If the calculated depth and the model depth did not reasonably match, then the model was not geometrically valid. A new slide plane depth was chosen, and the retro-deformation process

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

195

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

195

Bishop

Figure 10. Balanced cross sections showing the present and retro-deformed landslide configuration for the large block-style deformation model. (A) Present configuration. (B) Retro-deformed configuration.

was repeated. This iterative process was continued until the depth calculated from the retro-deformation procedure and the assumed depth for the model reasonably matched. In this study, a reasonable match in depth was considered to be agreement within 25 m. This distance was found to be approximately the shortest distance within the limits of precision for the procedure. Generally, about eight to ten trial-and-error iterations were needed to reach a reasonable match. Figures 10B through 13B present the final balanced retro-deformed cross-section diagrams. Numerical results are summarized in Table 1. The displacement results range between 225 and 310 m, and the calculated slide plane depths between 750 and 1035 m. From the complete results, the slide plane depth is concluded to be 950 ± ∼150 m.

196

In Figures 10B through 13B, the model blocks, which were assumed to be rigid, do not perfectly fit with one another, thereby resulting in gap and overlap areas. The presence of these areas indicates that none of the models can be exactly correct. Because the objective of the retro-deformation procedure is to determine an estimate for the depth of the slide plane and not to determine a precisely correct deformation model, some imperfection is acceptable. Still, it is useful to consider the issue of the presence of gap and overlap areas with regard to the error they represent. In order to achieve an exact mass balance for the area with the presence of the gaps and overlaps in the retro-deformed cross sections, the total area of gaps would need to equal the total area of overlap. Although it was found impractical to achieve exact equality of

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

196

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 11. Balanced cross sections showing the present and retro-deformed landslide configuration for the vertical shear deformation model. (A) Present configuration. (B) Retro-deformed configuration.

areas, the difference between the size of the total gap and overlap areas was minimized to the extent possible. In the final iteration(s) of retro-deformation for each model, the total area of gap and area of overlap in the retro-deformed cross section were determined, and the two values were compared. If the discrepancy was greater than 2,000 m2 , then the placement of the retro-deformed blocks was adjusted to better equalize the difference between the gap and overlap areas. After adjustment, the displacement value was measured, the slide plane depth was recalculated, and the depth was compared to the model depth. If these values then differed by 25 m or more, another iteration of retrodeformation was performed. The difference between total gap and overlap area in the final cross section for each model is presented in Table 2. An estimate of the

error in the slide plane depth calculation caused by the difference in the areas was made by dividing the difference in gap and overlap areas by the slide plane depth determined from retro-deformation. This ratio represents an approximation of the displacement error caused by the imbalance between gap area and overlap area. The largest calculated displacement error was 2 m, which represents a 1 percent error (10 m) for the calculated depth. This percentage error is sufficiently small that it does not change the conclusions of the analysis. On the cross sections, faults F1, F2, and F3 are modeled as planar, although it is known that high-angle normal faults are often listric. No field data were available to indicate if the faults are indeed planar. In order to determine if different results would be obtained if the

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

197

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

197

Bishop

Figure 12. Balanced cross sections showing the present and retro-deformed landslide configuration for the 60 degree antithetic shear deformation model. (A) Present configuration. (B) Retro-deformed configuration.

faults were instead listric, retro-deformation analyses were performed using listric high-angle faults for the 60 degree antithetic shear and the single block deformation styles. The differences in results from the planar fault models were insignificant.

Depth Determination from Contour Line Shift Data Use of contour line misalignment across the landslide’s lateral boundary is an alternate method with which to determine the landslide displacement. The relationship between contour shift and landslide displacement is diagramed in Figure 6. An equation that determines displacement, D, based on contour shift, X, is: D = X × sin / sin ( − ) , 198

(3)

where is slide plane dip, and is topographic surface gradient. Because the landslide boundary along the lower Pololu/Honokane Valley area provides the most unambiguous interpretation area for contour line shifts, this area was used to determine the contour shift value, X. The extrapolated positions of six contour lines across Pololu/Honokane Valleys and their scaled shift distances are shown in Figure 14. Using the six values, the average shift is 192 m. With a slope gradient, , of 8 degrees, and slide plane gradient, , of 3 degrees, a displacement value of 307 m was calculated from Eq. 3. Using this value in Eq. 2 yielded a slide plane depth of 759 m. This result is essentially the same as that obtained from the single large block retro-deformation procedure and, thus,

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

(3)

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 13. Balanced cross sections showing the present and retro-deformed landslide configuration for the 45 degree antithetic shear deformation model. (A) Present configuration. (B) Retro-deformed configuration.

supports the validity of the balanced cross-section analysis results. Landslide Interpretation Based on Depth and Contour Map Analysis Based on the observations and analysis of this study, a contour map and cross-section line showing the

interpreted landslide location and geometry are presented in Figure 15. The critical parameters determined from the study for constructing the cross-sectional slide plane geometry (Figure 15B) are the following: (1) The slide plane is approximately 950 m deep just downslope from the area of depletion, (2) the slide plane is planar downslope of the pull-apart head area, and (3) the slide plane gradient below

Table 1. Retro-deformation results. Model Large blocks Vertical blocks 60-degree dipping blocks 45-degree dipping blocks

Table 1. Retro-deformation results.

Displacement (m)

Model Depth (m)

Calculated Depth (m)

Average of Model and Calculated Depth (m)

310 245 230 225

765 970 985 1,060

750 950 1,010 1,035

760 960 1,000 1,050

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

Model Large blocks Vertical blocks 60-degree dipping blocks 45-degree dipping blocks

199

Displacement (m)

Model Depth (m)

Calculated Depth (m)

Average of Model and Calculated Depth (m)

310 245 230 225

765 970 985 1,060

750 950 1,010 1,035

760 960 1,000 1,050

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

199

Bishop

Figure 14. Contour map of the Pololu and Honokane Valley areas between elevations of 800 and 2,800 ft (244 and 854 m) at a contour interval of 100 ft (30 m). Contour lines at an interval of 400 ft (122 m), starting with the 800 ft (244 m) contour, are shown and have been straightened by projecting them across valleys without “V”-ing. Note that the contour lines across Pololu and Honokane Valleys projected from the northwest and southeast are misaligned. Also note that no misalignment occurs across Pololu Valley at and above the 2,400 ft (732 m) contour. The boxed numbers indicate the total scaled horizontal misalignment distance across Pololu and Honokane Valleys for each contour line. Base map is from Google Maps (2017).

the head area is less than the topographic slope gradient. The interpretation that the slide plane reaches the surface at the base of the coastal cliffs, as depicted in Figure 15B, is logical from two perspectives. First, there is no outcrop evidence that the slide plane reaches

Table 2. Gap and overlap area data.

Model Large blocks Vertical blocks 60-degree dipping blocks 45-degree dipping blocks

200

Difference in Gap Approximate and Overlap Areas Error in in Final Model (m2 ) Displacement (m) 0 650 1,850 101

0.0 0.7 2.0 0.1

the surface above the base of the cliffs. Although the slide plane is parallel or sub-parallel to lava flows in the volcanic stratigraphic section and would possibly not be conspicuous in outcrop on the cliffs, it seems likely that evidence such as brecciation or discordance of basalt flows along the slide plane, at least locally, would be detected by geologists who have studied the area (e.g., Stearns and MacDonald, 1946). No studies have reported such features. The second logical aspect of the slide plane reaching the surface at the base of the cliffs is that the slide plane aligns well with the seafloor surface offshore from the cliffs. This alignment strongly suggests that that the seafloor downslope of the coastal cliffs is an offshore continuation of the slide plane and that a lower part of the original landslide mass has been entirely removed.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Table 2. Gap and overlap area data.

Model Large blocks Vertical blocks 60-degree dipping blocks 45-degree dipping blocks

200

Difference in Gap Approximate and Overlap Areas Error in in Final Model (m2 ) Displacement (m) 0 650 1,850 101

0.0 0.7 2.0 0.1

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 15. (A) Interpreted map limits of the Kohala landslide. (B) Cross section showing the slide plane above the coastal cliffs. (C) Cross section showing estimated location of the topographic surface downslope of the coastal cliffs before removal by debris avalanche movement.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

201

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187â&#x20AC;&#x201C;205

201

Bishop

Because, as discussed in the next section, there is a reasonable mechanism by which the landslide mass above the lower part of the landslide could have been entirely removed down to the slide plane, it is rational to place the slide plane surfacing location at that base of the cliffs. The topographic surface of the missing part of the landslide before it was removed was estimated by projecting the subaerial volcanic slope seaward of the headwall coastal cliffs (Figure 15C). From the projection, the estimated toe location was found to be approximately 10 km northeast of the base of the cliffs. The submarine map limits of the landslide in Figure 15A were drawn based on the location of the toe. No previous studies have proposed this landslide, and it is herein named the Kohala landslide. DISCUSSION Removal of the Lower Part of the Landslide and Formation of the Coastal Cliffs An important question is: What was the mechanism by which the lower part of the landslide was removed? Erosion is one possibility. If erosion was, indeed, the cause, the effect was to remove the slide mass all the way to the slide plane, below which erosion of the bedrock was relatively nil. One scenario by which this could occur is through the formation of widely distributed, small-scale faults and fractures during displacement, which would cause the landslide mass to be significantly more erodible than underlying intact rock. Visual inspection of the landslide material exposed in the sea cliffs at Waipio and Pololu Valleys, however, revealed no unusual small-scale fracturing and/or faulting. Furthermore, significant fracturing and/or faulting are/is not consistent with the translational nature of the landslide. An alternative possibility to explain the missing mass is that it was removed as a catastrophic debris avalanche. In that scenario, the present-day coastal sea cliffs on Kohala Volcano are the debris avalanche headwall. This hypothesis for the origin of the cliffs was advanced by Smith et al. (2002) and Eakins and Robinson (2006, Plate I), although neither study recognized the portion of the landslide between the summit and the cliffs. A “debris chute” that trends laterally across the head of the Laupahoehoe slump (Smith et al., 2002) is the likely path of the debris avalanche. Hummocky debris avalanche material is present downslope of the debris chute. Circumstantial evidence supporting the debris avalanche hypothesis includes the relatively common occurrence of large-scale debris avalanche deposits surrounding the Hawaiian Islands (Figure 1). 202

Bishop

Because of the problems with the erosion hypothesis and the evidence supporting the debris avalanche hypothesis, the debris avalanche model is accepted here. In the terminology of Cruden and Varnes (1996), the Kohala landslide (Figure 15A), consisting of an upper slide area and lower debris avalanche area, is classified as a complex landslide. More specifically, the landslide is a slide/debris avalanche. Relation of Deep Valleys to the Landslide The deeply incised Pololu/Honokane and Waipio Valleys define the northwest and southeast lateral boundaries of the landslide, respectively. Rapid erosion of shear-weakened rock is likely a major factor for the large size of the valleys. Another important factor is the diversion of streams laterally across the slope at the landslide’s head along uphill-facing landslide fault scarps. These diverted streams feed into the large valleys, causing enhanced erosion. Conversely, streams downslope of the scarps are relatively starved. Pololu and Honokane Valleys, on the northwest landslide boundary, are separated by a narrow ridge. Mostly, the ridgeline is a knife edge, but in a few places, the ridge top is a planar remnant of Kohala’s volcanic slope (Figure 14). Where the ridge top is planar, the locations of map contours strongly suggest the presence of contour shifts across both valleys, with the larger shift occurring across Honokane Valley. The contour shifts are evidence for the presence of a fault in each valley, which helps to explain the large size of both Pololu and Honokane Valleys, both located at the northeast landslide margin. For Pololu Valley, the amount of contour shift appears to decrease upstream, suggesting that the fault displacement dies out upstream. Based on no apparent shift along the 2,400 ft (732 m) and 2,800 ft (854 m) contour lines, the fault under Pololu Valley terminates roughly 5 km from the coast. Waimanu Valley, with breadth and depth similar to Waipio Valley, is not near a landslide margin, where longitudinal landslide faults are primarily expected. Nevertheless, evidence indicates that Waimanu Valley is the location of a landslide fault. When projected across the valley, contour lines on the volcanic slopes bordering the valley do not align (Figure 16). Measured parallel to contour lines, the misalignment shows the slope southeast of the valley is 50 to 60 m lower than the volcanic slope to the northwest. In addition to the slope misalignment, the valley contains an anomalous bend toward the south 2 km inland of the coast. Above this bend, the valley trends oblique to the downhill slope direction for 5 km. The anomalous oblique valley trend is readily explained by the presence of an easily eroded fault zone underlying the valley (Figure 16).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii

Figure 16. Contour maps showing the Waimano and Waipio Valley areas with contour interval of 200 ft (61 m). Contour lines are re-drawn from the U.S. Geological Survey Kawaihae (U.S. Geological Survey, 1998a), Kukuihaele (U.S. Geological Survey, 1998b), and Honokane (U.S. Geological Survey, 1982) 7.5 minute quadrangle maps. (A) Contour lines on the volcano slopes adjacent to Waimano and Waipio Valleys without “V”s across relatively minor stream valleys. The dotted lines show the upper edges of steep slopes developed along Waimanu and Waipio Valleys and the coast. Contours for the coastal cliff are omitted for clarity. (B) Contour line projections across Waimanu Valley showing slope misalignment and the possible location of the fault responsible for the misalignment.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

203

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

203

Bishop

The presence of a fault below Waimanu Valley was first interpreted by Powers (1917) based on his observations of the slope misalignment. Landslide Age With the interpretation that the lower part of the landslide was removed as a debris avalanche, there are several possibilities for the chronology of events. One possibility is that the entire landslide began as a slowmoving slide, and at some point in time, the lower part mobilized to become a debris avalanche. Another possibility is that the debris avalanche occurred first, which destabilized the upper slope, causing the upper slide to form. As a third possibility, both the slide and debris avalanche parts of the landslide may have happened catastrophically, but the upper part slowed and did not mobilize as a debris avalanche. Age constraints to help decipher the chronology are few. For the slide part of the overall landslide, crosscutting relationships between the summit normal faults and Hawi basalts provide the only evidence for the slide age. The summit faults mostly cut Hawi flows but are locally buried by them, indicating only that the slide was active during the time of Hawi eruptions, which places the landslide movement at between 250 and 60 ka (Wolfe and Morris, 1996; Sherrod et al., 2007). For the debris avalanche part of the landslide, the best-known age constraint is based on the presence of valleys incised into the slide plane that forms the seafloor. The valleys extend from near the shoreline to about 20 km offshore, where water depth reaches 1,000 m and could only have been eroded after the debris avalanche occurred. An important unknown is whether the valleys were carved by subaerial streams or by submarine processes such as turbidity currents. If they were created by subaerial erosion, then the debris avalanche pre-dates subsidence of the area below sea level. Based on a subsidence rate of 2.6 mm/yr (Ludwig et al., 1991) and the toe area presently 700 m below sea level, the toe was at sea level approximately 270 ka. Hence, the debris avalanche would have occurred at 270 ka, or more, an age that would suggest a significant gap between the age of the debris avalanche and that of the slide. On the other hand, if the valleys were created largely or entirely by submarine processes, then the only constraint provided by the valleys is that the debris avalanche is older than the amount of time required for the valleys to be eroded. This would allow, but not require, the debris avalanche to have been contemporaneous with the slide. Until the erosional environment of the submarine valleys is known, even a rough of understanding of the chronology of the Kohala landslide movement is lacking. 204

Bishop

SUMMARY Contour map analysis of the subaerial northeast flank of Kohala Volcano indicates that the volcanic slope between Pololu and Waipio Valleys is underlain by a landslide mass that extends from the volcano’s summit to the coast. Contour map evidence shows that the slide plane is planar and dips less steeply than the topographic surface. Balanced cross-section analysis indicates that the slide plane is approximately 950 m below the surface just downslope from the zone of depletion. Using the depth of the slide plane and the constraint that the slide plane is less steep than the topographic surface, the interpretation presented here is that the slide plane reaches the surface at the base of the coastal cliffs between Pololu and Waipio Valleys. The slide plane thusly defined is in alignment with the seafloor offshore from the cliffs. The alignment suggests that the seafloor is a continuation of the slide plane and that a section of the landslide that had been present downslope from the sea cliffs has been entirely removed. A reasonable inference is that the missing material was removed far offshore as a catastrophic debris avalanche. The name proposed for the previously unrecognized slide/debris avalanche is the Kohala landslide. The sequence of movement for the slide and debris avalanche is uncertain. The upper slide part of the overall landslide moved sometime during the time of eruption of the Hawi series basalts that occurred between 250 and 60 ka. The debris avalanche involving the lower part of the landslide may have moved before or after 270 ka, depending on whether seafloor valleys eroded in the area vacated by the debris avalanche mass formed in a subaerial or submarine environment. Based on these constraints, the slide and debris avalanche could have been contemporaneous, or they could have occurred sequentially. ACKNOWLEDGMENTS Thoughtful reviews by John Clague, Eldon Gath, and Roy Shlemon greatly improved this manuscript. Their efforts are greatly appreciated. REFERENCES BISHOP, K. M., 1999, Determination of translational landslide slip surface depth using balanced cross-sections: Environmental and Engineering Geoscience, Vol. 5, No. 2, pp. 147–156. CRUDEN, D. M. AND VARNES, D. J., 1996, Landslide types and processes. In Turner A. K. and Schuster R. L. (Editors), Landslides Investigation and Mitigation, Special Report 247: Transportation Research Board, National Research Council, National Academy Press, Washington, D.C., pp. 36– 75.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

Landslide Interpretation, Kohala Volcano, Hawaii DAHLSTROM, C. D. A., 1969, Balanced cross-sections: Canadian Journal Earth Sciences. Vol. 6, No. 4, p. 743–757. DAVIS, G. H. AND REYNOLDS, S. J., 1996, Structural Geology of Rocks and Regions, 2nd ed.: John Wiley and Sons, Inc., New York, NY. 776 p. DE MATOS, R. M. D., 1993, Geometry of the hanging wall above a system of listric normal faults—A numerical solution: American Association Petroleum Geologists Bulletin, Vol. 77, pp. 1839–1859. EAKINS, B. W. AND ROBINSON, J. E., 2006, Submarine geology of Hana Ridge and Haleakala Volcano’s northeast flank, Maui: Journal Volcanology Geothermal Research, Vol. 151, pp. 229– 215. FOSSEN, H., 2010, Structural Geology: Cambridge University Press, Cambridge, U.K. 463 p. Google Earth, 2017, Google Earth: Electronic document, available at https://www.google.com/earth Google Maps, 2017, Google Maps: Electronic document, available at https://www.google.com/maps HAMBLIN, W. K., 1965, Origin of “reverse drag” on the downthrown side of normal faults: Geological Society America Bulletin, Vol. 76, pp. 1145–1164. HANSEN, W. R., 1965, Effects of the Earthquake of March 27, 1964, at Anchorage, Alaska: U.S. Geological Survey Profession Paper 542-A, 68 p. LAMB, J. R.; HOWARD, A. D.; DIETRICH, W. F.; AND PERRON, J. T., 2007, Formation of amphitheater-headed valleys by waterfall erosion after large-scale slumping on Hawai’i: Geological Society America Bulletin, Vol. 119, No. 7/8, pp. 805–822. LUDWIG, K. R.; SZABO, B. J.; MOORE, J. G.; AND SIMMONS, K. R., 1991, Crustal subsidence rate off Hawaii determined from 234 U/238 U ages of drowned coral reefs: Geology, Vol. 19, pp. 171–174. MACDONALD, G. A.; ABBOTT, A. T.; AND PETERSON, F. L., 1983, Volcanoes in the Sea, The Geology of Hawaii, 2nd ed.: University of Hawaii Press, Honolulu, HI. 517 p. MOORE, J. G. AND CLAGUE, D. A., 1992, Volcano growth and evolution of the island of Hawaii: Geological Society America Bulletin, Vol. 104, pp. 1471–1484. MOORE, J. G.; CLAGUE, D. A.; HOLCOMB, P. W.; LIPMAN, P. W.; NORMARK, W. R.; AND TORRENSAN, M. E., 1989, Prodigious submarine landslides on the Hawaiian Ridge: Journal Geophysical Research, Vol. 94, No. B12, pp. 17465–17484.

Landslide Interpretation, Kohala Volcano, Hawaii

MOORE, J. G.; NORMARK, W. R.; AND HOLCOMB, R. T., 1994, Giant Hawaiian landslides: Annual Review Earth and Planetary Science, Vol. 22, pp. 119–144. POWERS, S., 1917, Tectonic lines in the Hawaiian Islands: Geological Society America Bulletin, Vol. 28, pp. 501–514. ROWAN, M. G. AND KLIGFIELD, R., 1989, Cross section restoration and balancing as an aid to seismic interpretation in extension terranes: American Association Petroleum Geologists Bulletin, Vol. 75, No. 8, pp. 955–966. SHERROD, D. R.; SINTON, J. M.; WATKINS, S. E.; AND BRUNT, K. M., 2007, Geologic Map of the State of Hawai’i: U.S. Geological Survey Open-File Report 2007-1089, 83 p., 8 plates, scales 1:100,000 and 1:250,000, with GIS database. SMITH, J. R.; SATAKE, K.; MORAN, J. E.; AND LIPMAN, P. W., 2002, Submarine landslides and volcanic features on Kohala and Mauna Kea Volcanoes and the Hana Ridge, Hawaii. In Takahashi, E.; Lipman, P. W.; Garcia, M. O.; Naka, J.; and Aramaki, S. (Editors), Hawaiian Volcanoes: Deep Underwater Perspectives: Geophysical Monograph 128, American Geophysical Union, Washington, D.C., pp. 11–28. STEARNS, H. T. AND MACDONALD, G. A., 1946, Geology and Ground-Water Resources of the Island of Hawaii: Hawaii (Terr.) Division of Hydrography Bulletin 9, 363 p. SUPPE, J., 1985, Principles of Structural Geology: Prentice-Hall, Inc., Englewood Cliffs, NJ. 537 p. U.S. GEOLOGICAL SURVEY, 1982, Honokane Hawaii (map): 7.5 Minute Series, U.S. Department Interior, U.S. Geological Survey, Reston, VA, scale 1:24,000. U.S. GEOLOGICAL SURVEY, 1998a, Kawaihae, Hawaii (map): 7.5 Minute Series, U.S. Department Interior, U.S. Geological Survey, Reston, VA, scale 1:24,000. U.S. GEOLOGICAL SURVEY, 1998b, Kukuihaele Hawaii (map): 7.5 Minute Series, U.S. Department Interior, U.S. Geological Survey, Reston, VA, scale 1:24,000. WOLFE, E. W. AND MORRIS, J., 1996, Geologic Map of the Island of Hawaii: U.S. Geological Survey Miscellaneous Investigations Series I-2524-A, scale 1:100,000. WOODWARD, N. B.; BOYER, S. E.; AND SUPPE, J., 1989, Balanced geological cross-sections. In Balanced Geological CrossSections: An Essential Technique in Geological Research and Exploration: American Geophysical Union, Washington, D.C. 132 p.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

205

DAHLSTROM, C. D. A., 1969, Balanced cross-sections: Canadian Journal Earth Sciences. Vol. 6, No. 4, p. 743–757. DAVIS, G. H. AND REYNOLDS, S. J., 1996, Structural Geology of Rocks and Regions, 2nd ed.: John Wiley and Sons, Inc., New York, NY. 776 p. DE MATOS, R. M. D., 1993, Geometry of the hanging wall above a system of listric normal faults—A numerical solution: American Association Petroleum Geologists Bulletin, Vol. 77, pp. 1839–1859. EAKINS, B. W. AND ROBINSON, J. E., 2006, Submarine geology of Hana Ridge and Haleakala Volcano’s northeast flank, Maui: Journal Volcanology Geothermal Research, Vol. 151, pp. 229– 215. FOSSEN, H., 2010, Structural Geology: Cambridge University Press, Cambridge, U.K. 463 p. Google Earth, 2017, Google Earth: Electronic document, available at https://www.google.com/earth Google Maps, 2017, Google Maps: Electronic document, available at https://www.google.com/maps HAMBLIN, W. K., 1965, Origin of “reverse drag” on the downthrown side of normal faults: Geological Society America Bulletin, Vol. 76, pp. 1145–1164. HANSEN, W. R., 1965, Effects of the Earthquake of March 27, 1964, at Anchorage, Alaska: U.S. Geological Survey Profession Paper 542-A, 68 p. LAMB, J. R.; HOWARD, A. D.; DIETRICH, W. F.; AND PERRON, J. T., 2007, Formation of amphitheater-headed valleys by waterfall erosion after large-scale slumping on Hawai’i: Geological Society America Bulletin, Vol. 119, No. 7/8, pp. 805–822. LUDWIG, K. R.; SZABO, B. J.; MOORE, J. G.; AND SIMMONS, K. R., 1991, Crustal subsidence rate off Hawaii determined from 234 U/238 U ages of drowned coral reefs: Geology, Vol. 19, pp. 171–174. MACDONALD, G. A.; ABBOTT, A. T.; AND PETERSON, F. L., 1983, Volcanoes in the Sea, The Geology of Hawaii, 2nd ed.: University of Hawaii Press, Honolulu, HI. 517 p. MOORE, J. G. AND CLAGUE, D. A., 1992, Volcano growth and evolution of the island of Hawaii: Geological Society America Bulletin, Vol. 104, pp. 1471–1484. MOORE, J. G.; CLAGUE, D. A.; HOLCOMB, P. W.; LIPMAN, P. W.; NORMARK, W. R.; AND TORRENSAN, M. E., 1989, Prodigious submarine landslides on the Hawaiian Ridge: Journal Geophysical Research, Vol. 94, No. B12, pp. 17465–17484.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 187–205

205

Correlations between Fluvial Knickpoints and Recurrent Landslide Dams along the Upper Indus River

M. FAROOQ AHMED1

Department of Geological Engineering, University of Engineering and Technology, Lahore, Pakistan

J. DAVID ROGERS

Department of Geosciences and Geological and Petroleum Engineering, Missouri University of Science & Technology, Rolla, MO 65409, USA

ELAMIN H. ISMAIL

Department of Geosciences and Geological and Petroleum Engineering, Missouri University of Science & Technology, Rolla, MO 65409, USA

Key Terms: Indus River, Landslide Dam, Lithology, Knickpoint, DEM, Longitudinal Profile ABSTRACT This article summarizes an investigation into the likely role of landsliding in the formation of knickpoints along the Indus River in northern Pakistan. The knickpoints and their related geomorphic parameters (channel profile, concavity, drainage area, steepness index, etc.) were extracted from ASTER digital elevation models (DEMs) with 30 m resolution using ArcGIS and Matlab software. In total, 251 knickpoints were extracted from the longitudinal profile of the Indus River along an ∼750km-long reach upstream of Tarbela Dam. The identified knickpoint locations, along with their respective normalized steepness index (ksn values), were compared with the lithologic contacts, mapped faults, a regional-level landslide inventory, and the locations of prehistoric rockslides. The knickpoints identified adjacent to the prehistoric landslide dams (e.g., Katzarah, Gol-Ghone, and Lichar Gah, etc.) exhibited normalized steepness index (ksn ) in the range of 500–1800 m0.9 at various locations along the river channel. The highest normalized ksn values (>1800 m0.9 ) were observed in the tectonically active Nanga Parbat Haramosh Massif region, where the river flows through narrow gorges, and/or where active thrust faults cross the river channel. This study reveals that the landslide dams appear to be one of the significant trigger factors in the formation of knickpoints along the Indus River.

1 Corresponding

author email: mfanr5@mst.edu

INTRODUCTION Fluvial bedrock incision plays an active role in initiating rockslides that impact channels in mountainous areas (Leopold et al., 1964; Whipple, 2004). The migration of knickpoints through fluvial processes has been studied by a number of researchers, for example, Schumm et al. (1987); Seidl and Dietrich (1994); Whipple (2001); Harbor et al. (2005); Bishop et al. (2005), Crosby et al. (2007); and Ahmed and Rogers (2013), among others. These studies have shown that the upstream migration of knickpoints plays a significant role in initiating bedrock incision along stream channels, even triggering changes in the local base level. The most common mechanisms of concentrating detrital influx (choking the channel) are tributary accretion at the mouths of major tributaries and landslides and/or rockslides. Landslide dams often trigger bedrock incision by shifting the point of down-cutting onto the opposing, un-failed river bank. Most landslide dams tend to constrict channels and divert flow to the distal margins of debris dams, where overtopping often ensues, cutting downward (Ahmed and Rogers, 2014). These accumulations of coarse debris hinder the river’s transporting power and eventually retard the erosion process, if the blockages cannot be dislodged by normal river flow. This situation often occurs at the confluence of tributaries with main channels or at the foot of large landslides (Crosby et al., 2007; Ahmed et al., 2014). The oversize blocks left in the channel usually are found in clusters sufficient to form rapids that retard subsequent channel flow, locally increasing the channel gradient. These features are often observed as knickpoints in the longitudinal profile (Morisawa, 1960; Leopold et al., 1964). These knickpoints typically exhibit steep downstream and gentle upstream slopes,

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

207

1 Corresponding

author email: mfanr5@mst.edu

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

207

Ahmed, Rogers, and Ismail

and they resist bedrock erosion and upstream incision (Wang et al., 2012). In their study of the eastern margin of the Tibetan Plateau in Sichuan, Ouimet et al. (2007) noted the importance of large landslides in forming stable knickpoints. These knickpoints exert local control on river morphology and the channel profile by inhibiting incision and preventing the adjustment of channel beds to regional tectonic, climatic, and lithological factors. In most instances, this resistance to bed erosion appears to be a result of the coarse debris serving to “armor” the channel in these locations, while leaving a locally disturbed longitudinal profile (Hewitt, 1998; Korup, 2004a). In other words, a “step pool system” (Wang et al., 2012) is often formed by the rearrangement of the oversize bedrock blocks and boulders that have slid into the channel, obstructing and absorbing much of the flow energy, and thereby reducing bedrock incision (Whittaker and Jaeggi, 1982). There are many short-lived rockslide avalanches and glacial/moraine dams concentrated in the high-relief zones of the western Himalayan syntaxes, along the Indus, Gilgit, Hunza, and Shyok Rivers (Hewitt, 1982, 1998, 2002, 2009; Shroder and Bishop, 1998; Korup et al., 2010; Hewitt et al., 2011; and Ahmed and Rogers, 2014). The slope morphologies expressed at most of these mapped landslide dam sites (Ahmed and Rogers, 2012, 2014) suggest that the hillslopes were perturbed by previous episodes of landsliding and are continuously eroding. Each event truncates evidence of previous landslide sequences/events, making the older events increasingly difficult to discern, morphologically. The terraces observed along the Indus River tributaries in the Nanga Parbat Haramosh Massif (NPHM), as well as most of the Karakoram and Himalayas, appear to be composed of remnants of prehistoric landslides. In this study, the knickpoints identified along the Indus River’s longitudinal profile were analyzed, with a focus on measured normalized steepness indices (ksn ), to understand whether these knickpoints might be utilized as indirect criteria to identify the sites of unmapped landslide dams and/or rock avalanches. STUDY AREA This study was carried out along the Upper Indus River, which plays a significant role in the overall drainage systems within Pakistan. The river originates from the Tibetan Plateau, flowing through northern Pakistan from the Kashmir region through the entire country, eventually terminating at the Arabian Sea to the south (Figure 1). The irregular longitudinal profile of the Indus River is caused by tectonically active 208

Ahmed, Rogers, and Ismail

thrust faulting and associated uplift, especially across the NPHM region (Leland et al., 1998; Shehzad et al., 2009; and Korup et al., 2010). In this area, the rates of bedrock incision can be as much as 12 mm/yr. However, in many stretches, the river is obstructed by a significant number of landslide dams with volumes exceeding 10 × 106 m3 , which are expressed in the river’s longitudinal profile as prominent knickpoints. The most likely trigger factors for the high density of mapped slides (other than lithology) could be the high seismicity (Keefer, 1984) associated with a number of active thrust faults and periodic increases in the river discharge volume associated with climate variation actively triggering incision, which results in increased landslide activation. At the locations of historic landslide dams that have breached (e.g., Gol-Ghone, Katzarah, and Lichar Gah, etc.), the river is still actively down-cutting through the slide debris (Hewitt et al., 2011; Ahmed and Rogers, 2014). The enormous volumes of slide debris choking the main channel of the Indus appear to protect the Tibetan Plateau from more rapid dissection (Korup et al., 2010). This situation is also observed in the Skardu Valley (Figure 1), where the river flows upon a thick sequence of alluvial detritus. Most of the valley’s channel terraces have been stabilized by vegetation and support argillic B soil horizons, indicating considerable periods of subaerial exposure, especially, in the center of the river valley.

DATA AND METHODOLOGY Data The regional geomorphic analysis of the Indus River longitudinal profile was performed using data extracted from the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) global digital elevation model (GDEM) v2 data tiles, with 30 m resolution, using ArcGIS 10.2 (http://www.esri. com/software/arcgis/arcgis-for-desktop) and Matlab 12 (https://www.mathworks.com/products/matlabproduction-server/) software (Ahmed et al., 2017). The ASTER GDEM v2 is pre-processed, freely downloadable global elevation data with minimum artifacts, spikes, and errors. This data set provides reasonably accurate elevation measurements over Shuttle Radar Topography Mission (SRTM) 90-m-resolution data in steeply inclined terrain (Tachikawa et al., 2011). The GDEM data tiles were mosaicked and geo-referenced with Pakistan Grid Map World Geodetic System (WGS) 1984 Universal Transverse Mercator (UTM) Zone N430 using Envi 4.8 software (http://www. exelisvis.co.uk/ProductsServices/ENVIProducts.aspx)

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Knickpoints and Recurrent Landslide Dams

Figure 1. The upper Indus River basin above Tarbela Dam. The study area is outlined in black along the main stem of the upper Indus River in northern Pakistan.

to perform the regional geomorphic analysis in the ArcGIS and Matlab environment.

The ksn values were obtained by modifying the relationship given in Eq. 2 (from Whipple, 2004; Wobus et al., 2006),

to perform the regional geomorphic analysis in the ArcGIS and Matlab environment.

The ksn values were obtained by modifying the relationship given in Eq. 2 (from Whipple, 2004; Wobus et al., 2006),

Methodology

ksn = ks A( ref − ) ,

Methodology

ksn = ks A( ref − ) ,

The morphology of the river’s longitudinal profile can be explained by a variety of stream power models (Montgomery, 1994; Whipple, 2004; and Wobus et al., 2006). The basic stream power relationship between different geomorphic parameters is given in the following equation: S = ks A − ,

(1)

where S is the local slope of the channel; A is the mid-point area (Acent ) of the segment analyzed in the regression analyses in the form of Acent = 10(log Amax + log Amin )/2 (Wobus et al., 2006); ks is the local steepness index, which is the ratio of the channel gradient at specific locations (knickpoints) in the drainage area; and is the concavity of the stream channel profile (Kirby and Whipple, 2001; Wobus et al., 2006). The values of A and S could be assessed from regression analyses (Montgomery, 1994; Wobus et al., 2006). The value of ksn , expressing channel steepness, is normalized with respect to the upstream drainage area.

(2)

where ref is the reference concavity, and is the observed concavity of the stream channel at different channel segments. The ref value can be computed by averaging the local values observed along the respective channels for multiple reaches. In steady-state river systems that are in equilibrium with tectonic, climatic, or other environmental conditions, the longitudinal stream profile can be modeled by a single combination of steepness index and concavity. However, in transient river systems or river reaches where the stream profile presents abrupt changes in channel gradient, convex reaches may be found. These phenomena are called knickpoints (Wobus et al., 2006). In this analysis, multiple ksn values were generated for short segments of all the rivers in a catchment above a minimum drainage area to identify and characterize individual knickpoints. To obtain representative values of steepness indices, appropriate values of the reference concavity ( ref ),

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

209

(1)

(2)

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

209

Ahmed, Rogers, and Ismail

Figure 2. (A) Longitudinal profile of the Indus River upstream of Tarbela Dam, generated by Matlab 2012, using ASTER DEM 30-mresolution data. (B) An enlarged portion of the plot, illustrating the actual and smoothed channel profiles with user-specified knickpoints. Notice the two consecutive major knickpoints having ksn values of 3,225 and 2,139, respectively, observed a few kilometers upstream of Nanga Parbat Haramosh Massif (NPHM) region.

smoothing window size (250 m), and sampling contour interval (10 m) were selected to extract longitudinal profiles from the ASTER DEM data. The ref value of 0.45 was taken (which is a typical value for rivers in active mountain belts) for the Indus River channel in the Himalayan Region (Whipple, 2004). The relative steepness indices (ksn ) for different river reaches were obtained by applying a linear regression relationship between log slope and log drainage area for a fixed concavity index ( ref = 0.45) in the Matlab environment. The term ( ref − ) was calculated from the channel profile, considering the channel concavity at each point, and then it was used to estimate the normalized steepness index (ksn ) values. The user-identified knickpoints were marked within Matlab on the extracted longitudinal profiles. The user-identified and marked knickpoints with their respective ksn values (Figure 2) were then exported to ArcGIS to compare them with the mapped landslides and related features (Ahmed and Rogers, 2014). Korup et al. (2006) also followed similar criteria to verify whether the ksn values of channel segments were influenced by landslides, by noting if these values were significantly dissimilar from the rest of the channel profile. Higher ksn values, >175–2,000 m0.9 , were observed for those river segments previously impacted by sizable rock slope failures in the Indian Himalayas, Tien Chen (China), and the Southern Alps in New Zealand (Korup, 2004b; Korup et al., 2006). In a recent study (Ahmed et al., 2017), 251 useridentified knickpoints were marked at different places along the Indus River’s longitudinal profile (see Figure 2). The green line shows the actual longitudinal profile, while the cyan color line delineates the smoothed longi-

210

tudinal profile using the reference concavity value ( ref = 0.45) over an average 250 m reach of channel with 10 m contour interval in Matlab environment. This specification is appropriate to minimize the spikes, noise, and artifacts from the DEM data to highlight notable knickpoints along the channel sections (Ismail and Abdelsalam, 2012). The user-identified knickpoints were divided into “major” and “minor” groups, depending upon the magnitude of the elevation drop (difference in elevation). Using these criteria, there were 134 major and 117 minor knickpoints identified along the Indus River profile. “Major knickpoints” were arbitrarily characterized as those expressing a significant drop in elevation over a relatively short horizontal distance (i.e., >10 m vertical drop over an averaged 250 m reach on the profile). “Minor knickpoints” were those that exhibited topographic features similar to major knickpoints, but with less elevation change (i.e., <10 m vertical drop over an averaged 250 m reach on the profile). The knickpoints with sharp vertical drops over small distances (>15 m vertical drop over an averaged 250 m reach on the profile) were identified where mapped faults crossed the river channel. The enlarged view of the longitudinal profile (Figure 2B) clearly shows marked major and minor knickpoints. This section is taken along the Indus River where it enters the NPHM region. In this area, the river appears to be structurally influenced by linear features, most likely faults. At a horizontal distance of 540 km from Tarbela Dam, the river takes a sudden 90 degree turn, which likely follows another linear feature. The very gentle gradient (nearly horizontal) between these two consecutive knickpoints might be because the river is flowing parallel to these tectonic features, which are lifting it at near-constant rates.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

210

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Knickpoints and Recurrent Landslide Dams

Figure 3. Plot illustrating the comparison of landslide dams and other geomorphic features with the knickpoint locations along with the observed ksn values on the longitudinal profile of the Indus River upstream of Tarbela Dam. Note the location of MMT fault zone, Stak (2), Baroluma (6), Raikot (8), and other faults (modified from Ahmed et al., 2017).

Figure 3 exhibits the variation of normalized ksn values (hollow circles), adjacent to the user-identified knickpoints along the profile. Lithologic contacts are provided as different color bars, and active faults are indicated as dotted lines. The thin blue lines denote prominent landslide/rock avalanche dam features with their likely ksn values.

Landslide Inventory Maps and Prehistoric Landslides Data gleaned from the regional landslide inventory maps of the Indus River corridor (Ahmed and Rogers, 2012, 2014) and documented rockslides within this corridor (Hewitt, 1982, 2002; Shroder and Bishop, 1998; Korup et al., 2010; and Hewitt et al., 2011) were utilized to compare the spatial distribution of knickpoints (along with their associated ksn values) with mapped landslide features. Anomalous topographic protocols, which focused on bedrock slides more than half a kilometer long, were utilized to identify landslide features along the Indus River (Ahmed and Rogers, 2014). The topographic keys included: descending slopes with evidence of isolated topographic benches, divergent contours, crenulated contours, arcuate headscarps, extended topographic ridges, isolated topographic benches, and sudden up/downturns in hillslope contours (Rogers, 1980, 1994; Cruden and Varnes, 1996; Crosta, 2001; Glade, 2001; Doyle and Rogers, 2005; Van Den Eeckhaut et al., 2005; and Ahmed and Rogers, 2014). This regional inventory tentatively identified more than 2,200 deep-seated bedrock slides and secondary flow slides.

The landslide inventory map included 451 landslide features along both sides of the Indus River channel. These identified landslides were used to validate this study. Most of the bedrock landslide features identified were more than 1 km in length along the direction of landslide motion, and they appeared to be structurally controlled by bedding or foliation, pervasive jointing, lithologic contacts, and/or active faults. Subsequent movements, usually not related to the rock discontinuities, can be due to toe erosion, surcharging by adjacent slide debris, and earthquakes. The flanks of the parent bedrock could be fractured and subject to erosion, resulting in secondary slope failures as rotational slumps and earth flows. These secondary slope failures can move into the river channels, displacing the channel or forming significant knickpoints (Ahmed and Rogers, 2014). These secondary movements often trigger smaller slump blocks, flows, and rock avalanches. The mapped landslide features were spatially correlated to many of the documented historic landslide dams and rockslide features along the Indus River (Hewitt, 1982, 2002; Shroder and Bishop, 1998; Korup et al., 2010; and Hewitt et al., 2011). This provided an encouraging indication of validity for a regional-level study (covering a land area of >18,000 km2 ) and also made the mapped landslide features suitable for further utilization in the current study. Skeletal Blocks as Persistent Knickpoints The apparent preservation of landslide dams as knickpoints suggests that coarse debris remaining in the channel likely retards the rate of bedrock incision

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207â&#x20AC;&#x201C;220

211

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207â&#x20AC;&#x201C;220

211

Ahmed, Rogers, and Ismail

Figure 4. (A) Schematic section view of initial conditions through the inner gorge of an asymmetric bedrock canyon, sculpted by rock joints and foliation. (B) Schematic juxtaposition of blocks in a multiple block-glide rockslide emanating from a bank, along pre-existing suites of discontinuities. The blockage can then serve to elevate the river surface and water table in the adjacent banks, as shown on the left side at the profile. (C) Sketch illustrating the impacts of sudden drawdown of the local water level triggered by a landslide dam outbreak flood. This sudden drawdown often leads to new failures occurring along formerly submerged banks, on the opposite side of the channel.

at these locations. This inadvertent armoring could also impact local base levels as much as several kilometers upstream of old landslide dam sites (as observed at the Gol-Ghone and Katzarah sites). Thousands of years may pass before the river erodes the landslide debris and recovers its equilibrium grade. In most instances, the landslides appear to recur at the same locations, resulting in an accumulation of “skeletal blocks” at that location. This may explain why “major knickpoints” are formed at sites that do not exhibit overt tectonic offset along active fault zones. Figure 4 illustrates the typical mechanisms by which landslide dams tend to recur at the same general locations in the study area, likely because each blockage may influence and/or disturb the local equilibrium. After the initial slope failure occurs along pre-existing sets of discontinuities, it leaves an unsupported slope in the headscarp evacuation scar. The accumulated debris at the toe of the slide is then eroded by the river and subsequent slope failures and partial reactivations may 212

ensue, depending on how much debris is dispersed or removed by the river. Subsequent movements may be caused by rapid drawdown of the groundwater in the banks during catastrophic “outbreak floods” that breach the landslide debris dams. Figure 4A shows the initial site conditions along a hypothetical channel reach exhibiting the asymmetry typical of opposing canyon walls, common in layered or foliated rocks (typical of the Himalaya). The morphology of most canyon slopes is influenced by pervasive discontinuity suites, shown in this schematic view. Earthquake shaking or a rapidly falling channel (after debris dam is breached) can trigger mobilization of the joint-bordered blocks, which then obstruct the main channel. If a sufficient volume of debris effectively blocks the channel, the river’s flow can be temporarily suspended until it begins to overflow the debris dam. Upon breach, the flow is usually concentrated to the topographic low side of the dam, which is typically the opposing bank of the channel, as shown by the yellow alluvial gravels in Figure 4B. Such masses of blocky debris can obstruct the channel for considerable periods of time, between a few decades to as much as a few thousand years (Schuster and Costa, 1986). In this case, a temporary reservoir would be impounded upstream of the blockage (Grater, 1945). If the reservoir persists for more than a few weeks, the groundwater table will rise within the submerged banks, as sketched in the left side of Figure 4B. Figure 4C presents a conceptual view of a landslide dam shortly after overtopping. The water stored behind the debris dam will rapidly excavate a new channel, cutting downward from the point of initial overtopping, usually on the opposite channel bank (Rogers, 1994). Additional slide debris can be expected to slide and erode into the channel as the dam is rapidly excavated and debris is moved downstream. The rapid drawdown of the temporarily elevated groundwater within the debris and valley sides associated with the rapid erosion of the debris dam often triggers new block movements along inclined discontinuities at the lower extremities of the opposing slope, as sketched in Figure 4C. Note the possible recurrence of a multiplicity of slides, not only at old landslide dam sites, but along either bank of the temporary reservoirs formed behind debris dams. Also note that the displaced blocks on one bank are often semi-stabilized and in-filled with finer debris. The sequence of events presented in Figure 4A, B, and C suggests that the impact of sizable rock slope failures on a river’s longitudinal profile can be significant and is likely worthy of more attention in fluvial geomorphology, similar to other geomorphic parameters, such as climatic and tectonic factors.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Ahmed, Rogers, and Ismail

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Knickpoints and Recurrent Landslide Dams

Figure 5. Landslide inventory map (modified after Ahmed and Rogers, 2014) with red and yellow dots denoting the exported knickpoints (from ArcGIS software). The blue ovals show highlighted areas where the landslides and active faults apparently have significant correlation with identified knickpoints; a few of these areas were selected for further discussion.

Knickpoints and Recurrent Landslide Dams

Figure 6. Excerpt of landslide hazard map showing the area surrounding the prehistoric 450-m-high Gol-Ghone landslide dam. Outlined red circle shows the probable extent of the landslide dam, whereas the arrows show movement of the rock avalanche from the valley sides along the river: A cross-valley profile upstream of the Gol-Ghone landslide dam (C–D) is shown at lower left, while a cross-valley profile through the middle portion of the landslide dam (A–B) is presented at upper right.

RESULTS AND DISCUSSION Description of a Few Landslide Dams and Persistent Knickpoints in the Himalayas The identified knickpoints with their respective normalized steepness index (ksn ) values were exported to ArcGIS software and co-registered on the hillshade topographic map (Ahmed and Rogers, 2014) with the topographic features interpreted to be river-damming landslides. The highlighted areas, shown on Figure 5 as circles and ovals, were chosen to evaluate the likely association between the observed knickpoints and the landslide dams. The locations of a few mapped larger prehistoric landslide dams (volumes >10 × 106 m3 ) are shown on Figure 5. This figure also reveals that significant knickpoints, or “steps,” occur where active faults cross the channel, or where the channel has been obstructed by the landslide dams. For these larger prehistoric events, such as Gol-Ghone, Katzarah, and historic Lichar Gah, 1841 (see Figure 5), the normalized steepness index (ksn ) increases markedly. Figure 6 presents the prehistoric (Holocene age) GolGhone landslide dam, which impounded a reservoir at least 450 m deep (Hewitt, 2002; Hewitt et al., 2011). This mega-event (>10 km3 ) occurred just downstream of the confluence of the Shyok and Indus Rivers. At this location, the Indus River has incised 400 m through landslide dam debris, with another ∼50 m of debris still lying beneath the current channel. This debris deposit serves as the local base level (Korup et al., 2010; Hewitt et al., 2011). In this area, the Indus River flows through

RESULTS AND DISCUSSION a narrow gorge, and a number of knickpoints were observed within the footprint of the old landslide debris dam. One of those knickpoints exhibited a ksn value of 877 m0.9 , suggesting a major channel blockage from large resistant blocks in the breached debris dam. Cross-valley profiles (Figure 6) were developed through the center of the old landslide dam (section A–B) and approximately ∼2 km upstream of the dam (section C–D). The section through the landslide barrier is rather narrow and linear, indicative of a channel that is rapidly down-cutting its bed and flowing on a fairly high gradient, with numerous rapids. A few kilometers upstream, the channel is noticeably wider and shallower, due to aggradation along a reach exhibiting a much lower hydraulic gradient. Note the significant topographic benches 300 to 800 m above the present channel formed by the massive slope failures (section A–B). One of the largest documented, more likely prehistoric, rock avalanche features (Hewitt et al., 2011) in the study area is at Katzarah, on the west side of the Skardu-Shigar Basin. Figure 7 presents an example of the geomorphic impacts of catastrophic rockslide avalanches, including shallow braided channels and backwater aggradation upstream of the blockage, resulting in rapid deposition of braid bars, fine sand, and lacustrine silts downstream of the slide. Large remnant blocks from the old debris dam are scattered around the footprint of the breached landslide dam. In this area, the river has yet to retrench itself to the pre-slide riverbed profile. Figure 7 also shows a major

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

213

Description of a Few Landslide Dams and Persistent Knickpoints in the Himalayas The identified knickpoints with their respective normalized steepness index (ksn ) values were exported to ArcGIS software and co-registered on the hillshade topographic map (Ahmed and Rogers, 2014) with the topographic features interpreted to be river-damming landslides. The highlighted areas, shown on Figure 5 as circles and ovals, were chosen to evaluate the likely association between the observed knickpoints and the landslide dams. The locations of a few mapped larger prehistoric landslide dams (volumes >10 × 106 m3 ) are shown on Figure 5. This figure also reveals that significant knickpoints, or “steps,” occur where active faults cross the channel, or where the channel has been obstructed by the landslide dams. For these larger prehistoric events, such as Gol-Ghone, Katzarah, and historic Lichar Gah, 1841 (see Figure 5), the normalized steepness index (ksn ) increases markedly. Figure 6 presents the prehistoric (Holocene age) GolGhone landslide dam, which impounded a reservoir at least 450 m deep (Hewitt, 2002; Hewitt et al., 2011). This mega-event (>10 km3 ) occurred just downstream of the confluence of the Shyok and Indus Rivers. At this location, the Indus River has incised 400 m through landslide dam debris, with another ∼50 m of debris still lying beneath the current channel. This debris deposit serves as the local base level (Korup et al., 2010; Hewitt et al., 2011). In this area, the Indus River flows through

a narrow gorge, and a number of knickpoints were observed within the footprint of the old landslide debris dam. One of those knickpoints exhibited a ksn value of 877 m0.9 , suggesting a major channel blockage from large resistant blocks in the breached debris dam. Cross-valley profiles (Figure 6) were developed through the center of the old landslide dam (section A–B) and approximately ∼2 km upstream of the dam (section C–D). The section through the landslide barrier is rather narrow and linear, indicative of a channel that is rapidly down-cutting its bed and flowing on a fairly high gradient, with numerous rapids. A few kilometers upstream, the channel is noticeably wider and shallower, due to aggradation along a reach exhibiting a much lower hydraulic gradient. Note the significant topographic benches 300 to 800 m above the present channel formed by the massive slope failures (section A–B). One of the largest documented, more likely prehistoric, rock avalanche features (Hewitt et al., 2011) in the study area is at Katzarah, on the west side of the Skardu-Shigar Basin. Figure 7 presents an example of the geomorphic impacts of catastrophic rockslide avalanches, including shallow braided channels and backwater aggradation upstream of the blockage, resulting in rapid deposition of braid bars, fine sand, and lacustrine silts downstream of the slide. Large remnant blocks from the old debris dam are scattered around the footprint of the breached landslide dam. In this area, the river has yet to retrench itself to the pre-slide riverbed profile. Figure 7 also shows a major

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

213

Ahmed, Rogers, and Ismail

Figure 7. Landslide inventory map surrounding the Katzarah landslide dam location. Outlined red circle shows the probable extent of the landslide dam, whereas the arrows show movement of rock avalanche from the source along the river. A major knickpoint exists just downstream of the dam site, exhibiting a ksn value of 1,610 m0.9 . An aerial view of the Katzarah rock avalanche debris dam is shown at lower left, which shows the areal extent of the source area and debris fan developed in the Indus River channel, as well as the remnant lake.

knickpoint downstream of this dam site, which exhibits a ksn value of 1,610 m0.9 . This knickpoint might have been developed by some other landslide event, since the area is densely mapped with slides, or from the possible migration of debris downstream of the main event in the past several thousand years. A few kilometers downstream of Katzarah, the Indus River enters a very narrow chasm, with numerous knickpoints (Figure 7). This steep-sided gorge is armored with large boulders and skeletal blocks eroded from other landslide debris that once filled the channel. Further downstream, the Indus River turns sharply, entering the NPHM region (see Figure 8). This area comprises the northwestern part of the Greater Himalayas, south of Kohistan. The whole region is blanketed by mapped landslide features (Ahmed and Rogers, 2014), and significant numbers of knickpoints were identified in this reach. On the eastern flank of the NPHM region, a cluster of major and minor knickpoints was observed, with exceptionally high normalized steepness index (ksn ) values (>1,000 and up to 4,196 m0.9 ). Figure 9 shows what appears to be a breached landslide dam with identified knickpoints a few kilometers upstream of the Main Mantle Thrust (MMT) zone. The knickpoint with the highest ksn value (1,876 m0.9 ) occurs at the site of a former landslide dam (see photo at top right of Figure 9). Along the western flanks of the NPHM, downstream of the Indus-Gilgit River junction, the river becomes a braided channel where it flows through a Quaternary214

Figure 8. Excerpt of landslide inventory map where the Indus River flows through a steep-sided gorge in the eastern side of the NHPM, near the confluence of the Gilgit River, which flows through a broad, flat valley.

age alluvial valley (Figure 8). Widening of the channel in this reach likely resulted from the significant influx of sediment from both the Gilgit and Hunza Rivers, which have an aggregate watershed of ∼26,000 km2 . This aggradation sequence is locally restricted because of the increased stream power and active uplift associated with the MMT, as well as a complex system of active faults, such as the Raikot, Baroluma, and Stak Faults (Dipietro et al., 2000; Hewitt et al., 2011). Further downstream, in the Lichar Gah area (shown in Figure 10A and B), a seismically triggered rockslide in 1841 blocked the main Indus River channel (Code and Sirhindi, 1986). This dam was breached

Figure 9. Portion of the landslide inventory map a few kilometers upstream of MMT in the NHPM region. The image at upper right shows the breached landslide dam section (photo by Arve Tvedt, 2011).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Ahmed, Rogers, and Ismail

Figure 9. Portion of the landslide inventory map a few kilometers upstream of MMT in the NHPM region. The image at upper right shows the breached landslide dam section (photo by Arve Tvedt, 2011).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Knickpoints and Recurrent Landslide Dams

Figure 10. (A) (a) Ground view of the current situation, where the channel cuts through the dissected landslide dam, and (b) image showing the severe incision of the channel through sediment deposited a few kilometers upstream of the landslide dam (photo by from A. Mughal, 2010). (B) A portion of the landslide inventory map showing the Lichar Gah landslide dam location (outlined in red), triggered by an earthquake in 1841.

catastrophically after 6 months by significant inflows produced by excessive rainfall and snowmelt the following summer. The knickpoint associated with the breached Lichar Gah landslide dam exhibited one of the highest steepness index values (ksn = 2,951 m0.9 ) of any of the landslides identified along the main channel (see Figure 10B). Figure 11A (a–d) shows the physical characteristics of the Indus River channel at various locations upstream of the mapped Kes Gah landslide dam. These images show remnants of lacustrine silt and clay deposits that were trapped in the reservoir formed by the landslide dam. Figure 11B presents an excerpt of the landslide inventory map of this same area, where

the anomalous topographic features suggest that a series of debris dams likely blocked the channel in this reach. Note the mapped landslide features along with their respective knickpoints in the Kes Gah area. These knickpoints were probably generated from these masswasting events, as there is no evidence of fault activity or lithologic contrasts in this area. Figure 12 shows the current course of river flow near Hodar Gah, a few kilometers downstream of Kes Gah, along the Indus River (photo on lower left). The landslide inventory interpreted this area as a series of breached debris dams, with a buried channel lying beneath the largest run-out fan, causing the present channel to make a hairpin turn around this seemingly recent (but prehistoric) blockage. The hillshade topographic

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

215

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

215

Ahmed, Rogers, and Ismail

Figure 11. (A) Indus River channel near the Kes Gah landslide dam: (a) current incised channel conditions about 6 km upstream, (b) evidence of river deposits on the adjacent slopes before breaching, almost 3 km upstream from the site of the mapped landslide dam, (c) breached section showing the landslide debris close to the Kes Gah event, and (d) a little downstream of the Kes Gah landslide dam. (B) Map showing locations of probable landslide dams near the Kes Gah area of Chillas, along the Indus River, which are likely associated with the knickpoints, shown as red and yellow dots (with their respective ksn values).

map also shows the locations of probable landslide dams and associated blockages of the channel. There are many other examples of mapped landslides and faults exhibiting spatial correlations with knickpoints across the study area. A few of them are provided on the Figure 13. These comparisons reveal that moderate to high ksn values were observed at historically documented debris dam sites, as well as interpreted landslide/rockslide debris dam sites of unknown ages, especially the GolGhone (ksn = 877 m0.9 ), Katzarah (ksn = 1,610 m0.9 ), 216

Ahmed, Rogers, and Ismail

Figure 12. Excerpt of the landslide inventory map showing the locations of probable landslide dams, a few kilometers downstream of Kes Gah area along the river. Ground view shows the breached section of the river near Hodar Gah, looking upstream (photo by Der Fuchs, 2007).

Lichar Gah (ksn = 2,951 m0.9 ), and Kes Gah (ksn = 1,278 m0.9 ) historic events. The knickpoints observed adjacent to these historic slides suggest the likely correlation between mapped landslide dams and knickpoints, in the absence of any noticeable lithologic change or known faults. At a few locations, the knickpoints exhibited very high ksn values (between 1,800 and 4,200 m0.9 ). These tend to occur where the breached debris dams overlap with the active faults crossing the channel (especially in the NPHM region), which, in turn, highlights the composite role of landslide dams, uplift, and active faulting in knickpoint formation. The landslide inventory map (Ahmed and Rogers, 2014) included 451 mapped landslide features along both sides of the Indus River, in a corridor approximately 10 km wide (i.e., around 5 km along either side of the river). Close screening of the inventory map revealed that 337 landslides out of the 471 mapped slides extended to the banks of the Indus River. The other slides are either small, or their toes terminate more than 1 km from the river. This suggests that 337 landslides likely impacted the Indus River channel. In this way, the gross comparison of landslides to knickpoints (Table 1) suggests that 193 of the 337 (or 57 percent) mapped landslides appear to be spatially associated with knickpoints, while 28 of the 39 (72 percent) documented prehistoric and historic landslide dams, reported by other researchers in the study area (Hewitt, 1982, 1998, 2002, 2009; Shroder and Bishop, 1998; Korup et al., 2010; and Hewitt et al., 2011), also appear to exhibit marked knickpoints.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207â&#x20AC;&#x201C;220

Knickpoints and Recurrent Landslide Dams

Figure 13. Rockslide avalanches likely blocked the Indus River at different locations. (A) Excerpts of landslide inventory a few kilometers upstream of Dasu in the Pani Ba area. (B) Upstream of Char Nala (Dasu), showing the locations of knickpoints likely linked with active faulting (running along the river) and mapped landslides. (C) Probable landslide blockage along the Indus River near Thakot, where the knickpoints are probably influenced by active faults (running along the river) and mapped landslides. (D) Landslide inventory map in vicinity of a probable landslide dam (red stipple) near Bulder Gah, along the Indus River. The linear cyan line is a mapped fault cutting through the area.

Figure 14. Normalized relationship between mapped landslides and number of knickpoints with geologic map units. These histograms summarize knickpoint data normalized over 50 km of Indus River profile length.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207â&#x20AC;&#x201C;220

217

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207â&#x20AC;&#x201C;220

217

Ahmed, Rogers, and Ismail Table 1. Summary of observed knickpoints and frequency of related landslides. Description

No.

Mapped landslides Mapped landslides exhibiting knickpoints

451 193

Historic landslide dams Historic landslides exhibiting knickpoints

39 28

significant factors in triggering knickpoints along the channel.

Table 1. Summary of observed knickpoints and frequency of related landslides.

Calculation (193/451) × 100 = 43% of the observed landslides appear to be associated with identified knickpoints. (28/39) × 100 = 72% of documented landslide dams exhibit spatial agreement with identified knickpoints.

Normalized Relationships among Knickpoints, Lithology, and Landslides A gross comparison of knickpoints and landslide dams with underlying lithologic units was also performed to investigate any relationship between these geomorphic features. Figure 14 shows the normalized relationship between mapped landslides and the major geologic units. The massive bedrock units, including Tkm, Tkb, Pct, and Eg, are associated with the least number of slides. The bedrock correlations suggest that the areas associated with massive landslides are those of the more highly weathered gneisses, amphibolites, and graphitic blueschist and greenschist of the Indian Basement Complex of Precambrian age, the HazaraKashmir Basement Complex (PCb-PCs), undifferentiated metasedimentary rocks (MPzm), and Mesozoic to Paleozoic rocks of the Northern Suture Mélange (MPzs). Mica schist terrains are particularly susceptible to triggering large bedrock landslides, likely because of the low friction developed along micaceous planes of foliation and their tendency to retard pore pressure dissipation, especially during translatory movement (Morton and Sadler, 1989). Figure 14 shows the majority of knickpoints were observed in the rock units MPzm, MPzs, PCb, and PCs. Overall, these units are composed of materials with medium to low strength parameters. The likely correlation between knickpoints and prehistoric landslides in the NPHM region and the prehistoric Gol-Ghone, Katzarah, and historic Lichar Gah (1841) landslide dams suggests that the majority of observed knickpoints appear to correlate well with landslide dams, which highlights the importance of landslide dams as a significant factor in knickpoint creation. This does not diminish the significance of regional tectonics, active faults, and changes in bedrock lithology and bedrock incision as additional 218

Ahmed, Rogers, and Ismail

Description

CONCLUSIONS This study was conducted to examine whether knickpoint analysis using publicly available data and analytical tools could be used to identify the presence of significant landslide dams on the Indus River. The knickpoints and their related geomorphic parameters (channel profile, concavity, drainage area, and normalized steepness index, etc.) were extracted from the analysis of ASTER GDEMs with 30 m resolution, using ArcGIS and Matlab software. In total, 251 major and minor knickpoints were extracted from the river’s longitudinal profile, extending approximately 750 km upstream of Tarbela Dam. The knickpoint locations and their respective normalized steepness index (ksn ) values were compared with the regional-level landslide inventory maps and the locations of historic rockslides. These analyses suggest that correlations can be drawn between spatial locations of knickpoints and mapped and documented landslides dams. This correlation is reflected in moderate to high normalized steepness index (ksn ) values at several landslide debris dam sites, particularly in the Gol-Ghone (ksn = 877), Katzarah (ksn = 1,610), Lichar Gah (ksn = 2,951), and Kes Gah (ksn = 1,278) areas. Gross comparison between knickpoints and landslides shows that 57 percent of the mapped landslides and 72 percent of the historic landslide dams impacting the channel are sufficiently pervasive to have formed recognizable knickpoints. These geomorphic observations and comparisons appear to validate the concept that rock slope failures have impacted the Indus River’s longitudinal profile on multiple occasions in the geologic past. This study further concluded that the highly complex and irregular longitudinal profile of the Indus River has likely resulted from the complex interaction of various geomorphic processes, and that landslide dams are one of the significant mechanisms by which knickpoints form in the channel. Using only publicly available data analysis, it is difficult to attribute each knickpoint to a definite cause, such as landslides or other possible causes of knickpoints, but this analysis demonstrates that there are numerous locations along the channel where a better correlation exists between knickpoints and landslide damming processes. More detailed site-specific analyses aided by higherresolution data would need to be undertaken to verify the assumptions used for this investigation, which focused mainly on identification of anomalous topography (identifying landslide features) and perturbations of the channel profile (identifying knickpoints).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

No.

Calculation

Mapped landslides Mapped landslides exhibiting knickpoints

451 193

Historic landslide dams Historic landslides exhibiting knickpoints

39 28

(193/451) × 100 = 43% of the observed landslides appear to be associated with identified knickpoints. (28/39) × 100 = 72% of documented landslide dams exhibit spatial agreement with identified knickpoints.

significant factors in triggering knickpoints along the channel. CONCLUSIONS This study was conducted to examine whether knickpoint analysis using publicly available data and analytical tools could be used to identify the presence of significant landslide dams on the Indus River. The knickpoints and their related geomorphic parameters (channel profile, concavity, drainage area, and normalized steepness index, etc.) were extracted from the analysis of ASTER GDEMs with 30 m resolution, using ArcGIS and Matlab software. In total, 251 major and minor knickpoints were extracted from the river’s longitudinal profile, extending approximately 750 km upstream of Tarbela Dam. The knickpoint locations and their respective normalized steepness index (ksn ) values were compared with the regional-level landslide inventory maps and the locations of historic rockslides. These analyses suggest that correlations can be drawn between spatial locations of knickpoints and mapped and documented landslides dams. This correlation is reflected in moderate to high normalized steepness index (ksn ) values at several landslide debris dam sites, particularly in the Gol-Ghone (ksn = 877), Katzarah (ksn = 1,610), Lichar Gah (ksn = 2,951), and Kes Gah (ksn = 1,278) areas. Gross comparison between knickpoints and landslides shows that 57 percent of the mapped landslides and 72 percent of the historic landslide dams impacting the channel are sufficiently pervasive to have formed recognizable knickpoints. These geomorphic observations and comparisons appear to validate the concept that rock slope failures have impacted the Indus River’s longitudinal profile on multiple occasions in the geologic past. This study further concluded that the highly complex and irregular longitudinal profile of the Indus River has likely resulted from the complex interaction of various geomorphic processes, and that landslide dams are one of the significant mechanisms by which knickpoints form in the channel. Using only publicly available data analysis, it is difficult to attribute each knickpoint to a definite cause, such as landslides or other possible causes of knickpoints, but this analysis demonstrates that there are numerous locations along the channel where a better correlation exists between knickpoints and landslide damming processes. More detailed site-specific analyses aided by higherresolution data would need to be undertaken to verify the assumptions used for this investigation, which focused mainly on identification of anomalous topography (identifying landslide features) and perturbations of the channel profile (identifying knickpoints).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Knickpoints and Recurrent Landslide Dams

Further studies would be required to differentiate knickpoints triggered by regional tectonic uplift, local fault offset, bedrock erodibility, or landslide/rockslide dams. ACKNOWLEDGMENTS The authors wish to thank the Natural Hazards Mitigation Institute at the Missouri University of Science and Technology, Rolla, MO, USA, and the Center of Excellence for Geospatial Information Science of the U.S. Geological Survey in Rolla, MO, for providing the opportunity to accomplish this work. The authors would also like to thank University of Engineering and Technology, Lahore, Pakistan, for financial support to one of the authors to conduct this research. REFERENCES AHMED, M. F. AND ROGERS, J. D., 2012, Landslide mapping and identification of old landslide dams along the Indus River in Pakistan, using GIS techniques. In Proceedings of the AEG 55th Annual Meeting, Vol. 55: Salt Lake City, UT, Association of Environmental & Engineering Geologists, 44 p. AHMED, M. F. AND ROGERS, J. D., 2013, Thalweg profiles and knickpoints as useful discriminators of prehistoric landslide dams in northern Pakistan: Geological Society of America Abstracts with Programs, Vol. 45, No. 7. AHMED, M. F. AND ROGERS J. D., 2014, Creating reliable, firstapproximation landslide inventory maps using ASTER DEM data and geomorphic indicators, an example from the upper Indus River in northern Pakistan: Environmental & Engineering Geoscience, Vol. 20, No. 1, pp. 67–83. AHMED, M. F.; ROGERS, J. D.; AND ISMAIL, E. H., 2014, A regional level preliminary landslide hazard study of upper Indus River basin: European Journal of Remote Sensing, Vol. 47, pp. 343– 373. AHMED, M. F.; ROGERS, J. D.; AND ISMAIL, E. H., 2017, Knickpoints and various geomorphic processes along the upper Indus River, Pakistan: Swiss Journal of Geosciences (under review or accepted). BISHOP, P.; HOEY, T. B.; JANSEN, J. D.; AND ARTZA, I. L., 2005, Knickpoint recession rate and catchment area: The case of uplifted rivers in eastern Scotland: Earth Surface Processes and Landforms, Vol. 30, pp. 767–778. CODE, J. A. AND SIRHINDI, S., 1986, Engineering implications of the impoundment of the Indus River by an earthquake induced landslide. In Schuster, R. L. (Editor), Landslide Dams: Processes, Risk and Mitigation: Geotechnical Special Publication No. 3, American Society of Civil Engineers, Reston, VA, pp. 97–110 CROSBY, B. T.; WHIPPLE, K.; GASPARINI, N. M.; AND WOBUS, C. W., 2007, Formation of fluvial hanging valleys; theory and simulation: Journal of Geophysical Research, Vol. 112, pp. F03S10. CROSTA, G. B., 2001, Failure and flow development of a complex slide: The 1993 Sesa landslide: Engineering Geology, Vol. 59, pp. 173–199. CRUDEN, D. M. AND VARNES, D. J., 1996, Landslide types and processes. In Turner, A. K. and Schuster, R. L. (Editors), Landslides, Investigation and Mitigation: Special Report 247, Transportation Research Board, Washington, D.C., pp. 36–75.

Knickpoints and Recurrent Landslide Dams

DIPIETRO, J. A.; HUSSAIN, A.; AHMAD, I.; AND KHAN, M. A., 2000, The Main Mantle Thrust in Pakistan: Its character and extent. In Khan, M. A.; Treloar, P. J.; Searle, M. P.; and Jan, M. Q. (Editors), Tectonics of the Nanga Parbat Syntaxis and the Western Himalaya: Special Publication No. 170, Geological Society, London, U.K., pp. 375–393. DOYLE, B. C. AND ROGERS, J. D., 2005, Seismically-induced lateral spread features in the western New Madrid Seismic Zone: Environmental & Engineering Geoscience, Vol. XI, No. 3, pp. 251–258. GLADE, T., 2001, Landslide hazard assessment and historical landslide data: An inseparable couple? In Glade, T.; Albini, P.; and Frances, F. (Editors), The Use of Historical Data in Natural Hazard Assessments: Springer, Berlin, pp. 153–167. GRATER, R., 1945, Landslide in Zion Canyon National Park, Utah: Journal of Geology, Vol. 53, pp. 116–124. HARBOR, D.; BACASTOW, A.; HEATH, A.; AND ROGERS, J., 2005, Capturing variable knickpoint retreat in the Central Appalachians, USA: Geografia Fisica e Dinamica Quaternaria, Vol. 28, pp. 23–36. HEWITT, K., 1982, Natural dams and outburst floods of the Karakoram Himalaya. In Glen, J. (Editors), Hydrological Aspects of Alpine and High Mountain Areas: Publication 138, International Hydrological Association, Exeter, UK, pp. 259–269. HEWITT, K., 1998, Catastrophic landslides and their effects on the Upper Indus streams, Karakoram Himalaya, northern Pakistan: Geomorphology, Vol. 26, pp. 47–80. HEWITT, K., 2002, Styles of rock-avalanche depositional complexes conditioned by very rugged terrain, Karakoram Himalaya Pakistan. In Evans, S. G. and Degraff, J. V. (Editors), Catastrophic Landslides: Effects, Occurrence, and Mechanisms: Reviews in Engineering Geology XV, Geological Society of America, Boulder, CO, pp. 345–377. HEWITT, K., 2009, Catastrophic rock slope failures and late Quaternary developments in the Nanga Parbat–Haramosh Massif, Upper Indus basin, northern Pakistan: Quaternary Science Review, Vol. 28, pp. 1055–1069. HEWITT, K.; GOSSE, J.; AND CLAGUE, J. J., 2011, Rock avalanches and the pace of late Quaternary development of river valleys in the Karakoram Himalaya: Geological Society of America Bulletin, Vol. 123, No. 9/10, pp. 1836–1850 ISMAIL, E. H. AND ABDELSALAM, M. G., 2012, Morpho-tectonic analysis of the Tekeze River and the Blue Nile drainage systems on the Northwestern Plateau, Ethiopia. Journal of African Earth Sciences, Vol. 69, pp. 34–47. KEEFER, D. K., 1984, Landslides caused by earthquakes: Geological Society of America Bulletin, Vol. 95, pp. 406–421. KIRBY, E. AND WHIPPLE, K. X., 2001, Quantifying differential rockuplift rates via stream profile analysis: Geology, Vol. 29, No. 5, pp. 415–418. KORUP, O., 2004a, Landslide-induced river channel avulsions in mountain catchments of southwest New Zealand: Geomorphology, Vol. 63, pp. 57–80. KORUP, O., 2004b, Large landslides and their effect on sediment flux in South Westland, New Zealand: Earth Surface Processes and Landforms, Vol. 30, pp. 305–323. KORUP, O.; MONTGOMERY, D. R.; AND HEWITT, K., 2010, Glacier and landslide feedbacks to topographic relief in the Himalayan syntaxes: Proceedings of National Academy of Sciences of the United States of America, Vol. 107, No. 12, pp. 5317–5322. KORUP, O.; STROM, L. S.; AND WEIDINGER, J. T., 2006, Fluvial response to large rock-slope failures: Examples from the Himalayas, the Tien Shan, and the Southern Alps in New Zealand: Geomorphology, Vol. 78, pp. 3–21. LELAND, J.; REID, M. R.; BURBANK, D. W.; FINKEL, R.; AND CAFFEE, M., 1998, Incision and differential bedrock uplift

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

219

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

219

Ahmed, Rogers, and Ismail along the Indus River near Nanga Parbat, Pakistan Himalaya, from 10Be and 26Al exposure age dating of bedrock straths: Earth and Planetary Science Letters, Vol. 154, No. 1–4, pp. 93–107. LEOPOLD, L. B.; WOLMAN, M. G.; AND MILLER, J. P., 1964, Fluvial Processes in Geomorphology: W. H. Freeman, San Francisco, CA. MONTGOMERY, D. R., 1994, Valley incision and the uplift of mountain peaks: Journal of Geophysical Research, Vol. 99, pp. 13913– 13921. MORISAWA, M., 1960, Erosion rates on Hebgen earthquake scarps, Montana: Geological Society of America Bulletin, Vol. 71, No. 12, pt. 2, 1932 p. MORTON, D. M. AND SADLER, P. M., 1989, The failings of the Pelona Schist: landslides and sackungen in the Lone Pine Canyon and Wrightwood areas, eastern San Gabriel Mountains, southern California. In Sadler, P. M. and Morton, D. M. (Editors), Landslides in a semi-arid environment: Redlands, California, Publications of the Inland Geological Society, Vol. 2, pp. 301–322. OUIMET, W. B.; WHIPPLE, K. X.; ROYDEN, L. R.; SUN, Z.; AND CHEN, Z., 2007, The influence of large landslides on river incision in a transient landscape: Eastern margin of the Tibetan Plateau (Sichuan China): Geological Society of America Bulletin, Vol. 119, pp. 1462–1476. ROGERS, J. D., 1980, Factors Affecting Hillslope Profile: Current Topics in Geomorphology: unpublished manuscript, University of California at Berkeley, University of California Water Resources Center Archives, Berkeley, CA, 67 p. ROGERS, J. D., 1994, Report Accompanying Map of Landslides and Other Surficial Deposits of the City of Orinda, CA: Rogers/Pacific, Inc., for the City of Orinda Public Works Department, Orinda, CA, 141 p. SCHUMM, S. A., 1987, The Fluvial System: John Wiley & Sons, New York, NY, 338 p. SCHUSTER, R. L. AND COSTA, J. E., 1986, A perspective on landslide dams. In Schuster, R. L. (Editor), Landslide Dams—Processes, Risk, and Mitigation: Geotechnical Special Publication No. 3, American Society of Civil Engineers, Reston, VA, pp. 1–20.

220

SEIDL, M. A. AND DIETRICH, W. E., 1994 Longitudinal profile development into bedrock: An analysis of Hawaiian channels: Journal of Geology, Vol. 102, pp. 457–474. SHEHZAD, F.; MAHMOOD, S. A.; AND GLOAGUEN, R., 2009, Drainage network and seismological analysis of active tectonics in the Nanga Parbat Haramosh Massif, Pakistan. In Proceedings of the IEEE International Geoscience and Remote Sensing Symposium, Vol. I: IEEE, Cape Town, South Africa, pp. 9–12. SHRODER, J. F., JR., AND BISHOP, M. P., 1998, Mass movement in the Himalaya: New insights and research directions: Geomorphology, Vol. 26, pp. 13–35. TACHIKAWA, T.; KAKU, M.; IWASAKI, A.; GESCH, D.; OIMOEN, M.; ZHANG, Z.; DANIELSON, J.; KRIEGER, T.; CURTIS, B.; HAASE, J.; ABRAMS, M.; AND CARABAJAL, C., 2011, ASTER Global Digital Elevation Model Version 2—Summary of Validation Results: Report to the ASTER GDEM Validation Team, NASA, Washington, D.C., pp. 15–24. VAN DEN EECKHAUT, M.; POESEN, J.; VERSTRAETEN, G.; VANACKER, V.; MOEYERSONS, J.; NYSSEN, J.; AND VAN BEEK, L. P. H., 2005, The effectiveness of hillshade maps and expert knowledge in mapping old deep-seated landslides: Geomorphology, Vol. 67, pp. 351–363. WANG, Z.; CUI, P.; YU, G.; AND ZHANG, K., 2012, Stability of landslide dams and development of knickpoints: Environmental Earth Sciences, Vol. 65, pp. 1067–1080. WHIPPLE, K. X., 2001, Fluvial landscape response time; how plausible is steady-state denudation?: American Journal of Science, Vol. 301, pp. 313–325. WHIPPLE, K. X., 2004, Bedrock rivers and the geomorphology of active orogens: Annual Review of Earth and Planetary Sciences, Vol. 32, pp. 151–185. WHITTAKER, J. G. AND JAEGGI, N. R., 1982, Origin of step-pool system in mountain streams: Journal of the Hydraulics Division, ASCE, Vol. 108, pp. 758–773. WOBUS, C.; WHIPPLE, K. X.; KIRBY, E.; SNYDER, N.; JOHNSON, J.; SPYROPOLOU, K.; CROSBY, B.; AND SHEEHAN, D., 2006, Tectonics from topography: Procedures, promise, and pitfalls, Special Paper of the Geological Society of America, Vol. 398, pp. 55– 74. DOI: 10.1130/2006.2398(04)

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

220

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 207–220

Identification and Analysis of Large Paleo-Landslides at Mount Burnaby, British Columbia

MIRKO FRANCIONI1

Camborne School of Mines, University of Exeter, Penryn Campus, Penryn, Cornwall, TR10 9FE, UK

DOUG STEAD JOHN J. CLAGUE ALLISON WESTIN

Department of Earth Sciences, Simon Fraser University, 8888 University Dr, Burnaby, BC V5A 1S6, Canada

Key Terms: Paleo-Landslides, LiDAR, GIS, Engineering Geomorphology ABSTRACT This article presents a multi-scale and multidisciplinary study of large, late Pleistocene or early Holocene slumps in Eocene sedimentary rocks at Mount Burnaby, just east of Vancouver, British Columbia (BC). Airborne Light Detection and Ranging (LiDAR) and field data were integrated into a Geographic Information System to understand the origin, kinematics, and subsequent history of the landslides. Products derived from the bare-earth LiDAR data include an engineering geomorphology map, shaded relief maps, and several LiDAR slope profiles. To understand the landslides better, we analyzed discontinuities and structural lineaments. The structure of the Eocene rocks underlying Mount Burnaby was compared with trends of local lineaments and the shape of the coastline of Burrard Inlet and Indian Arm as well as trends of regional faults and lineaments identified by previous researchers working in southwest BC. Two main joint systems likely played a key role in conditioning the north slope of Mount Burnaby for failure. The landslides likely happened during or soon after deglaciation of the area at the end of the Pleistocene on the steep north face of Mount Burnaby after a 200-m fall in relative sea level caused by glacio-isostatic uplift of the crust. INTRODUCTION In this article, we document a previously unknown complex of large slumps in the Vancouver, British Columbia (BC), metropolitan area. The slumps, which occurred on the steep north face of Mount Bur1 Corresponding

author email: M.Francioni@exeter.ac.uk

naby at the edge of Simon Fraser University, are the largest known mass movement in south-coast British Columbia. We were able to document and infer the cause of the landslides through an integrated study involving field mapping, new Light Detection and Ranging (LiDAR) imagery, and data analysis within a multilayered Geographic Information System (GIS). The slumps are likely of late Pleistocene or early Holocene age, but given their size, questions arise about whether or not similar landslides might occur in the future. If this were to happen, parts of Simon Fraser University and an allied community (UniverCity), as well as a railway and highway at the base of Mount Burnaby might be damaged. Acquisition of a LiDAR point cloud of Mount Burnaby provided an opportunity to create thematic maps, slope profiles, and an engineering geomorphological map that enabled an interpretation of the geometry and original positions of slump blocks and sites subject to reactivation that could damage public works. In this article, we describe the multi-disciplinary approach we used to identify and characterize the slumps. We also characterize the local and regional structural environment in which the slumps occurred and consider how likely it is that a similar slope failure might happen in the future. Study Area Mount Burnaby is a small forested mountain (peak elevation 350 m above sea level (m.a.s.l.)) located approximately 10 km east of Vancouver, BC, and just south of Burrard Inlet (Figure 1A). Simon Fraser University (SFU), which opened in 1965, is situated at the top of the mountain. The mountain is a remnant of formerly more extensive Eocene terrestrial sedimentary rocks that are part of a Cretaceous-Cenozoic fill in the Georgia Basin, which lies between the Coast Mountains to the north, the spine of Vancouver Island to the west, and the

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

221

author email: M.Francioni@exeter.ac.uk

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

221

Francioni, Stead, Clague, and Westin

Figure 1. Mount Burnaby (Google Earth, 2015). (A) Location of Mount Burnaby in the Vancouver metropolitan area (the inset shows the location of the study area in western Canada). (B) 3D view of Mount Burnaby and Simon Fraser University, situated at the top of the mountain. View to the southeast.

Cascade Range in Washington State (United States), to the southeast (Armstrong, 1990; Mustard and Rouse, 1994; and Turner et al., 1998). The rocks consist of conglomerate, sandstone, and mudstone of alluvial and deltaic origin deposited in a subsiding structural basin. There are few natural exposures of the rocks on Mount Burnaby, but the same sequence of rocks is commonly exposed in building excavations in downtown Vancouver.

222

Mount Burnaby has a pronounced asymmetric north-south profile, reflecting the ca. 11◦ southerly dip of the entire Eocene sequence (Figure 1B). A south-thickening wedge of Fraser Glaciation till and glaciomarine sediments overlies the inclined bedrock surface on the southerly dip slope (Roddick, 1965; Armstrong, 1990). The glacial sediments partially, but not completely, mask the underlying bedrock structure. A series of steps on the south-facing slope, which are

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

222

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

Paleo-Landslides at Mount Burnaby

Figure 2. Flow chart showing steps followed in the Mount Burnaby study.

only evident in LiDAR images, reflect daylighting beds of sandstone, conglomerate, and mudstone that have different resistances to erosion. In contrast, the forestcovered north slope of the mountain is very steep and, for this reason, has not been studied in the past. METHODS Several tasks were completed to gain a better understanding of the geology of Mount Burnaby. Specifically, we created thematic maps and slope profiles, produced an engineering geomorphology map, identified major structural lineaments visible in the LiDAR data, and analyzed coastline trends to determine if they might be structurally controlled. The results were validated and integrated with stratigraphic and structural data acquired through fieldwork. We pinpointed areas in which relict slump blocks are prone to re-activation due to human activity and, finally, compiled all data in a GIS database to interpret, visualize, and share the data. Figure 2 is a flow chart summarizing the methodology used in this study. LiDAR and GIS Analyses Airborne LiDAR data for the area shown in Figure 3A were provided in LASer (LAS) file format and processed using the GIS software ESRI’s ArcMap version 10.2. LAS was created and is maintained by the American Society for Photogrammetry and Remote Sensing; it is a standard file format for the exchange of

Paleo-Landslides at Mount Burnaby

LiDAR data. Each LAS file contains records of all laser pulses recorded. The first returns are typically associated with the highest features in the landscape, for example a treetop or the top of a building, whereas the last returns are from the ground surface. In this research, we filtered out all but the last returns to produce a bareearth layer, from which we built a 1-m-resolution digital terrain model of the mountain. Numerous authors have discussed the use of GIS in landslide and structural geology investigations (e.g., Van Westen, 1998; Xie et al., 2006; and Francioni et al. 2014, 2015). We produced and used three main thematic maps: hillshade, slope, and aspect (respectively, Figure 3B– D). A hillshade map (Figure 3B) was generated using a lighting effect based on differences in elevation within the landscape. It provides synthetic three-dimensional (3D) views of the landscape. The light angle used in Figure 3B has an azimuth of 315◦ (light cast from the northwest) and an inclination of 25◦ . The slope map (Figure 3C) shows the steepness of the slopes. The aspect map (Figure 3D) shows the dip direction of slopes and was used in this study to highlight abrupt changes in slope orientation. Engineering Geomorphology Map and Slope Profiles We produced an engineering geomorphology map (presented and described in Section 3 of this article) of the north side of Mount Burnaby using guidelines provided in Cooke and Doornkamp (1990). We mapped (i) linear terrain features, including concave and convex slopes, cliffs (terrain steeper than 45◦ ), lineaments, and gullies, and (ii) landforms, including debris flow fans and slump blocks. Faults, bedding, and joints were represented as point values on the map. We also created topographic profiles across the north, west, and south sides of Mount Burnaby to better characterize its topography and geometry. Structural Analysis We measured trends of lineaments on Mount Burnaby (red lines in Figure 4). In addition, we measured conspicuous linear sections of the coastlines of Burrard Inlet and Indian Arm in the vicinity of the mountain (blue lines in Figure 4) to determine if their geometries might be structurally controlled and in agreement with the lineaments on the mountain. The same approach was also used to measure the average orientations of mapped faults in southwest BC (British Columbia Ministry of Energy and Mines, 2014) (Figure 5). Field Survey All natural and man-made exposures on Mount Burnaby were documented to obtain as much

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

223

Figure 2. Flow chart showing steps followed in the Mount Burnaby study.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

223

Francioni, Stead, Clague, and Westin

Figure 3. Mount Burnaby thematic maps (reference system: NAD 83 Zone 10N, Simon Fraser University as reference). (A) Satellite image (Google Earth, 2015) showing the area where LiDAR and GIS analyses were carried out. (B) Hillshade map. (C) Slope map. (D) Aspect map.

information as possible about the stratigraphy and structure of the mountain. As a result of the presence of dense vegetation and the inaccessibility of most of the steep north slope of the mountain, we could only doc-

Figure 4. Mount Burnaby lineaments (red lines) interpreted from thematic maps and linear shoreline segments (blue lines; Dominion of Canada, 1859).

224

ument 11 outcrops in the area (Figure 6A). Figure 6B shows an example of geological characterization of one of the outcrops (SFU2). The orientations of all discontinuities and faults exposed in outcrops were measured. Finally, we also examined borehole logs and reports provided by BGC Engineering, which include geotechnical information on foundation conditions at SFU.

Geodatabase and Data Sharing

We developed a GIS database to manage the thematic maps, shape files, and field data (outcrop locations and observations and structural elements). LiDAR maps and Google Earth were used as base maps; through hyperlinks it was possible to electronically link information to specific sites. The GIS was also an important resource for sharing data with others, using the freeware platform ArcReader (ESRI, 2014) and Google Earth (2015).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

Figure 4. Mount Burnaby lineaments (red lines) interpreted from thematic maps and linear shoreline segments (blue lines; Dominion of Canada, 1859).

224

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

Paleo-Landslides at Mount Burnaby

Figure 5. Faults in southwest BC (from British Columbia Ministry of Energy and Mines, 2014).

RESULTS We rendered 3D visualizations of the twodimensional thematic maps, such as those shown in Figure 3B–D, to identify and interpret the paleolandslides on the north side of Mount Burnaby. Figure 7 is an example of a 3D representation of the aspect map that clearly shows back-tilted slump blocks on the north side of the mountain. The faces of the back-tilted slump blocks are approximately parallel to the headscarp, indicating that they detached from that area. Figure 8 illustrates the difference between the steep north face of Mount Burnaby, with slopes typically ranging from 40◦ to 80◦ , and the much gentler south face inclined 5–15◦ , reflecting the ca. 11◦ southerly dip of the Eocene sequence. The main gully that indents the north face of the mountain marks the border between two different slope environments. East of the gully, a gentle, north-dipping slope separates the top of the mountain from the steep landslide headscarp, whereas to the west, the mountain top abruptly borders the steep north slope. Barnet Highway extends across the toes of some of the slumps blocks (Figures 7 and 8). Since construction of the highway in the early 1940s, there have been small slope failures at its margins, which we attribute to reactivation of the slump blocks. Figures 9 and 10 show the engineering geomorphology map of Mount Burnaby. We separated the north side of the mountain into two areas to highlight geomorphic and geological features (Figure 9). Area 1 (Figure 10A) contains 14 slump blocks that are partially covered by fan deposits and might be connected

to one another below the surface. All fan deposits slope away from the cliff face. All except one of the convex breaks of the slump blocks align with the cliff face, indicating their source and supporting the interpretation drawn from the aspect map. Area 2 (Figure 10B) contains four slump blocks, one on the north side of Barnet Highway and the other three south of the highway. The convex breaks in these blocks align and are approximately parallel to the cliff face. Numerous steep gullies extend down the north slope of the mountain. The largest gully, located at northwest corner of SFU (Figure 3B), has a northeast trend and a gradient ranging from 40◦ to 85◦ ; it has retrogressed toward the south much more that other gullies. Representative slope profiles on the north, west, and south sides of the mountain were used to highlight geomorphic features and slope geometries (Figure 11A). Profiles of the north slope show its steep upper face and the presence and geometry of the slump blocks (Figure 11B). The west slope has a series of topographic steps reflecting the alternation of sandstone, conglomerate, and mudstone units with different resistances to erosion (Figure 11C). The steps are up to about 10 m high, have average gradients of 14–24◦ , and are separated by treads with gradients of 3◦ to 6◦ . The slope on the south side of the mountain is much more gentle, reflecting the ca. 11◦ southerly dip of the entire Eocene sequence (Figure 11D). Several lines of evidence reveal two main sets of structures, both throughout the region and on and near Mount Burnaby (Figure 12). First, regional faults extracted from the databases of the British Columbia Ministry of Energy and Mines strike

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

225

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

225

Francioni, Stead, Clague, and Westin

Figure 6. (A) Hillshade map showing locations of investigated outcrops and boreholes. (B) Exposure at SFU2 (shovel for scale). (C) conglomerate; SS, sandstone; M, mudstone.

northeast-southwest (ca. N 046◦ ) and northwestsoutheast (ca. N 125◦ ) (Figure 12A). Second, the shorelines of Burrard Inlet and Indian Arm are rectilinear, with ca. N 045◦ and N 125◦ trends, respectively (Figures 4 and 12B). Third, lineaments evident on thematic maps trend ca. N 55◦ and N 145◦ (Figure 12C). Together, the two sets delineate the bimodal geometry of crest of the escarpment on the north side of the mountain (Figure 4). Fourth, field measurements of fractures and faults con-

226

firm these two main sets on an outcrop scale—one striking about 055◦ and the other 135◦ (Figure 12D). Although there is strong structural control on the geomorphology of the mountain, we found no evidence that any of the structures are currently active. Field observations indicate that the exposed Eocene rocks are highly weathered rock and dense soils, mainly of sand and gravel size. Using field techniques (Hoek and Brown, 1997), we estimated the strength of these rocks to be between 1 and 5 MPa. Rock cores recovered

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

226

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

Paleo-Landslides at Mount Burnaby

Figure 7. 3D representation of aspect map with back-tilted slump blocks on the north face of Mount Burnaby, road cuts, and inferred slump headscarp.

in four boreholes recently drilled by BGC Engineering yielded uniaxial compressive strength values of 19 MPa for mudstone, 10 MPa for sandstone, and 18 MPa for conglomerate (Schmid and Baumgard, 2014). The difference between unconfined compressive strengths of the deep rocks and those at the surface is due to weathering, which has leached cement and matrix from rock in the near-surface environment (Clague et al., 2015).

Figure 13 shows the result of a slope profile analysis that helped us to define the geometry of the paleolandslides and interpret the failure mechanism better. Slope profile D-D� (Figure 13A and B) shows three slump blocks sourced on the steep headwall of the mountain to the south. The headscarp reaches an elevation of 200 m.a.s.l. and has an average gradient of 51◦ toward the northwest. The slump block closest to the headwall, which is the youngest of the three, has a

Figure 8. 3D representation of the slope map.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

Figure 8. 3D representation of the slope map.

227

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

227

Francioni, Stead, Clague, and Westin

Figure 9. Engineering geomorphology map of Mount Burnaby.

back-tilted face dipping 38◦ parallel to the headscarp. The base of the back-tilted face is about 100 m below the top of the headscarp and about 215 m away from it. The other two, older slump blocks are about 110 m and 160 m below the top of the headscarp and 330 m and 585 m away from it, respectively. Again, the slump blocks are parallel to the headscarp and dip backward at 12◦ and 18◦ . Slope profile E-E� (Figure 13C) shows two slump blocks 190 m and 220 m below the top of the headscarp (270 m.a.s.l.) and 475 m and 645 m from it, respectively. The headscarp has an average gradient of 60◦ . The slump blocks dip 32◦ and 15◦ toward the headscarp. Slope profile F-F� (Figure 13D) shows a single slump block 190 m below the top of the headscarp (230 m.a.s.l.) and 460 m from it. The headscarp has an average gradient of 65◦ . This slump block dips 20◦ toward the headscarp. Figure 14 illustrates a possible failure mechanism for the landslides based on the maps produced in this study. Measurements made using LiDAR slope profiles and the thematic maps allowed us to calculate the length, width, and angle of the back-tilted face of the uppermost slump block (respectively, 250 m, 50 m, and 38◦ ). Assuming that the surface of Mount Burnaby extended to the north beyond the headscarp with a gradient similar to that of the present surface, we conclude that a block failed along a curved surface and rotated 38◦ before coming to rest. We note, however, that post-landslide fan deposits partially en228

velop the slump block, introducing uncertainties in our estimates of block dimensions and volume. Considering the similarities in the geometry and orientation of all slump blocks on the north side of the mountain (Figure 9), we assume that the blocks were generated by the same failure mechanism. The many gullies incising the north slope are evidence of the relative ease with which the sedimentary rocks underlying the mountain are eroded. The main gully is likely located on a northeast-southwest structure, possibly a fault, along which there has been enhanced erosion. Erosion along this and other northeast-trending lineaments may have favored the formation of lateral relief surfaces for the landslide. DISCUSSION Airborne LiDAR is now widely used in landslide studies (e.g., Chen et al., 2006; Ventura et al., 2011; Brideau et al., 2012; and Haas et al., 2012) because it provides detailed georeferenced 3D models of the land surface free of vegetation. In this article, we feature the symbiotic integration of airborne LiDAR with GIS techniques in a study of the Mount Burnaby paleolandslide, as has been done successfully by other landslide researchers in the past (Schulz. 2007; Van Den Eeckhaut et al., 2012; and Jebur et al., 2014). The information obtained from the thematic maps and LiDAR interpretation were used to produce an

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

Paleo-Landslides at Mount Burnaby

Figure 10. Engineering geomorphology maps of (A) area 1 and (B) area 2 (see Figure 9).

engineering geomorphology map of Mount Burnaby that could not otherwise have been made. This map and the aspect map allowed us to define the position and geometry of the slump blocks and to suggest their source and movement direction. Analysis of the engineering geomorphology map showed

that fan deposits overlie the slump blocks and are derived from the same source, but clearly postdate them. LiDAR data were also critical for analyzing Mount Burnaby slope profiles, which aided in understanding the geometry of the mountain and the paleo-landslides.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

229

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

229

Francioni, Stead, Clague, and Westin

Figure 11. Slope profiles. (A) Locations of slope profiles shown in B, C, and D. (B) Slope profile A-A . (C) Slope profile B-B . (D) Slope profile C-C .

The steps reflecting the alternation of layers of sandstone, conglomerate, and mudstone could not be seen in conventional aerial photographs and only became obvious when we examined the LiDAR DEM.

230

The slope profiles also supported the interpretation of the geometry and source of the slump blocks. Our analysis highlighted two main discontinuity systems that are likely a product of Paleogene or Neogene

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

230

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

Paleo-Landslides at Mount Burnaby

Figure 12. Rose diagrams showing the results of all structural analyses (rose diagrams created using DIPS [Rocscience, 2014]). (A) Regional faults. (B) Shoreline segments. (C) GIS lineaments. (D) Faults and fractures measured in the field.

tectonic deformation. Although we found no evidence of active (Holocene) faults in the area, the consistent structure at all scales suggests that deformation related to these two main trends played an important role in the evolution of both Mount Burnaby and Burrard Inlet. Lineaments with these trends also define the main scarp and the main gully on the mountain. The presence on the north side of Mount Burnaby of multiple slump blocks with similar geometries and orientations (Figures 11 and 13) suggests a series of retrogressive events that were closely spaced in time. Although we were unable to determine the exact ages of the landslides, the large volume of fan material overlying the slump blocks suggests that the failures happened during late Pleistocene or early Holocene time, either during local deglaciation or shortly thereafter. The Vancouver area, including

Mount Burnaby, was deglaciated about 13,500 years ago (Clague, 1981); at that time relative sea level was about 200 m higher than it is today, and only the upper slopes of Mount Burnaby were above the sea surface (Mathews et al., 1970; Clague et al., 1982). Shortly after deglaciation, and by no later than 11,000 years ago, glacio-isostatic rebound rapidly lowered local sea level to below its present datum. Glacier retreat and glacioisostatic rebound may have contributed to the massive slope failures on the north side of Mount Burnaby. The importance of glacier de-buttressing as a cause of slope failure has been highlighted by Evans and Clague (1994). Removal of ice confinement allows for kinematic failure, while the fall in sea level due to rebound may have been accompanied by increased erosion of the toe of the slope. Earthquakes caused by glacio-isostatic rebound may also be implicated in the landslides.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

231

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

231

Francioni, Stead, Clague, and Westin

Figure 13. Slope profiles and paleo-landslide geometry. (A) Slope map showing the locations of three profiles across the north slope of Mount Burnaby. (B) Slope profile D-D . (C) Slope profile E-E . (D) Slope profile F-F .

232

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221â&#x20AC;&#x201C;235

Paleo-Landslides at Mount Burnaby

Figure 14. A possible failure interpretation for paleo-landslides on the north slope of Mount Burnaby. The slump block used in this example has a length of 250 m and a 50-m back-tilted face.

Recognition of the landslides on Mount Burnaby has raised concerns over the possibility that similar failures in the densely populated Vancouver metropolitan area could happen in the future. In our view, however, the Mount Burnaby slumps occurred in response to processes that no longer affect this area, namely deglaciation and large and rapid sea-level change. Therefore, we consider it unlikely that a slump like the ones we have documented will happen again in the Vancouver area.

CONCLUSION Mount Burnaby consists of several hundred meters of Eocene sandstone, conglomerate, and mudstone that dip about 11◦ to the south. The gently dipping south slope of the mountain is mantled by a southwardthickening wedge of Pleistocene glacial and glaciomarine sediments. The surface Eocene rocks are highly weathered and typically have the strength of soils (as

the term is used in an engineering sense of the word). Non-weathered rocks at depth have unconfined compressive strength properties characteristic of indurated sedimentary rocks. Our integrated study of LiDAR-derived thematic maps, slope profiles, a derived engineering geomorphological map, and field observations show that the north slope of Mount Burnaby failed as a series of large slump blocks that are now partially buried by coalescent debris flow fans sourced on the steep slope to the south. Although the slump blocks at the base of the mountain are relicts of latest Pleistocene or early Holocene landslides, they have spawned small reactivation failures along Barnet Highway over the past 60 years. The identification of these previously unrecognized landslides is significant in showing the power of a modern, integrated approach for analyzing the land surface, as well as the implications that they can carry for managing landslide hazards in urban areas. We advocate the use of thematic and engineering geomorphology maps

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

233

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

233

Francioni, Stead, Clague, and Westin

to identify landslides prior to construction or improvement of private and public infrastructure. Our LiDAR-assisted interpretation of structural trends helped document two main fracture and fault sets, one oriented NW-SE and the other NE-SW. Paleogene or Neogene deformation responsible for these joint sets played a key role in creating the present landscape of the Vancouver metropolitan area. Field study and an examination of LiDAR imagery revealed no evidence of Holocene faulting in the area, although the limited exposure of rock on Mount Burnaby does not allow us to definitively preclude this possibility. ACKNOWLEDGMENTS We are grateful to BGC Engineering for its support of our research, and in particular we acknowledge Alex Baumgard, who helped us secure LiDAR imagery and funding that allowed us to undertake the project. The research was supported with grants provided by Kinder Morgan Canada and the Natural Sciences and Engineering Research Council of Canada (NSERC Discovery Grants to D.S. and J.J.C.). Three journal reviewers (Kim Bishop, Keith Loague, and one anonymous referee) provided critiques that allowed us to improve the article. REFERENCES ARMSTRONG, J. E., 1990, Vancouver Geology: Geological Association of Canada, Cordilleran Section, Vancouver, BC. 128 pp. BRIDEAU, M. A.; STURZENEGGER, M.; STEAD, M.; JABOYEDOFF, M.; LAWRENCE, M.; ROBERTS, N. J.; WARD, B. C.; MILLARD, T. H.; AND CLAGUE, J. J., 2012, Stability analysis of the 2007 Chehalis Lake landslide based on long-range terrestrial photogrammetry and airborne LiDAR data: Landslides, Vol. 9, pp. 75–91. BRITISH COLUMBIA MINISTRY OF ENERGY AND MINES, 2014, BC Digital Geology Data: Electronic document, available at http://www.empr.gov.bc.ca/Mining/Geoscience/Bedrock Mapping/Pages/BCGeoMap.aspx CHEN, R. F.; CHANG, K. J.; ANGELIER, J.; CHAN, Y. C.; DEFFONTAINES, B.; LEE, C. T.; AND LIN, M. L., 2006, Topographical changes revealed by high-resolution airborne LiDAR data: The 1999 Tsaoling landslide induced by the Chi-Chi earthquake: Engineering Geology, Vol. 88, pp. 160–172. CLAGUE, J. J., 1981, Late Quaternary Geology and Geochronology of British Columbia, Part 2: Summary and Discussion of Radiocarbon-Dated Quaternary History: Geological Survey of Canada Paper 80–35, 41 pp. CLAGUE, J. J.; HARPER, J. R.; HEBDA, R. J.; AND HOWES, D. E., 1982, Late Quaternary sea levels and crustal movements, coastal British Columbia: Canadian Journal Earth Sciences, Vol. 19, pp. 597–618. CLAGUE, J. J.; STEAD, D.; FRANCIONI, M.; AND WESTIN, A., 2015, Geology of Mount Burnaby: Kinder Morgan unpublished report, Calgary, AB.

234

COOKE, R. U. AND DOORNKAMP, J. C., 1990, Geomorphology in Environmental Management: A New Introduction: Oxford University Press, Oxford, 434 pp. DOMINION OF CANADA, 1859, Burrard Inlet [Map]. Dominion of Canada; surveyed by Mr. W.J. Stewart, under the direction of Staff Commander J.G. Boulton, R.N.: Electronic document, available at http://searcharchives.vancouver.ca/burrardinlet-6 ESRI, 2014, ArcMap, ArcReader and ArcScene Version 10.2: Electronic document, available at http://www.arcgis.com EVANS, S. G. AND CLAGUE, J. J., 1994, Recent climatic change and catastrophic geomorphic processes in mountain environments: Geomorphology, Vol. 10, pp. 107–128. FRANCIONI, M.; SALVINI, R.; STEAD, D.; GIOVANNINI, R.; RICCUCCI, S.; VANNESCHI, C.; AND GULLÌ, D., 2015, An integrated remote sensing-GIS approach for the analysis of an open pit in the Carrara marble district, Italy: Slope stability assessment through kinematic and numerical methods: Computers Geotechnics, Vol. 67, pp. 46–63. FRANCIONI, M.; SALVINI, R.; STEAD, D.; AND LITRICO, S., 2014, A case study integrating remote sensing and distinct element analysis to quarry slope stability assessment in the Monte Altissimo area, Italy: Engineering Geology, Vol. 183, pp. 290–302. GOOGLE EARTH, 2015, DigitalGlobe 2012: Electronic document, available at http://www.earth.google.com HAAS, F.; HECKMANN, T.; WICHMANN, V.; AND BECHT, M., 2012, Runout analysis of a large rockfall in the Dolomites/Italian Alps using LIDAR derived particle sizes and shapes: Earth Surface Processes Landforms, Vol. 37, pp. 1444–1455. HOEK, E. AND BROWN, E. T., 1997, Practical estimates of rock mass strength: International Journal Rock Mechanics Mining Sciences, Vol. 34, pp. 1165–1186. JEBUR, M. N.; PRADHAN, B.; AND TEHRANY, M. S., 2014, Optimization of landslide conditioning factors using very highresolution airborne laser scanning (LiDAR) data at catchment scale: Remote Sensing Environment, Vol. 152, pp. 150–165. MATHEWS, W. H.; FYLES, G.; AND NASMITH, H., 1970, Postglacial crustal movements in southwestern British Columbia and adjacent Washington State: Canadian Journal Earth Sciences, Vol. 7, pp. 690–702. MUSTARD, P. S. AND ROUSE, G. E., 1994, Stratigraphy and evolution of Tertiary Georgia basin and subjacent Upper Cretaceous sedimentary rocks, southwestern British Columbia and northwestern Washington State. In Monger, J. W. H. (Editor), Geology and Geological Hazards of the Vancouver Region, Southwestern British Columbia: Geological Survey of Canada Bulletin, Vol. 481, pp. 97–169. ROCSCIENCE, 2014, Dips Version 6.016: Electronic document, available at https://www.rocscience.com RODDICK, J. A., 1965, Vancouver North, Coquitlam and Pitt Lake Map Areas, British Columbia, with Special Emphasis on the Evolution of the Plutonic Rocks: Geological Survey of Canada Memoir 335: Electronic document, available at http://geogratis.gc.ca/api/en/nrcan-rncan/ess-sst/ 0062e577-e050-593c-a248-2ddc421744f8.html#distribution SCHMID, C. AND BAUMGARD, A., 2014, Trans Mountain Pipeline ULC, TMEP Westridge Tunnel Investigation. 2014. Site Investigation Data Report: BGC Technical Report, Project No. 0095-150-15, 40 pp. SCHULZ, W. H., 2007, Landslide susceptibility revealed by LIDAR imagery and historical records, Seattle, Washington: Engineering Geology, Vol. 89, pp. 67–87. TURNER, R. J. W.; CLAGUE, J. J.; GROULX, B. J.; AND JOURNEAY, J. M., 1998, GeoMap Vancouver, Geological Map of the Vancouver Metropolitan Area: Geological Survey of Canada Open File 3511.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

Francioni, Stead, Clague, and Westin

234

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

Paleo-Landslides at Mount Burnaby VAN DEN EECKHAUT, M.; KERLE, N.; POESEN, J.; AND HERVÁS, J., 2012, Object-oriented identification of forested landslides with derivatives of single pulse LiDAR data: Geomorphology, Vol. 173–174, pp. 30–42. VAN WESTEN, C. J., 1998, GIS in landslide hazard zonation: A view, with cases from the Andes of Colombia. In Martin, F. P. and Heywood, D. I. (Editors), Mountain Environment and Geographic Information Systems: Taylor & Francis, London, U.K., pp. 35–165.

Paleo-Landslides at Mount Burnaby

VENTURA, G.; VILARDO, G.; TERRANOVA, C.; AND BELLUCCI SESSA, E., 2011, Tracking and evolution of complex active landslides by multi-temporal airborne LiDAR data: The Montaguto landslide (southern Italy): Remote Sensing Environment, Vol. 115, pp. 3237–3248. XIE, M.; ESAKI, T.; QIU, C.; AND WANG C. X., 2006, Geographical information system-based computational implementation and application of spatial three-dimensional slope stability analysis: Computers Geotechnics, Vol. 33, pp. 260–274.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

235

VAN DEN EECKHAUT, M.; KERLE, N.; POESEN, J.; AND HERVÁS, J., 2012, Object-oriented identification of forested landslides with derivatives of single pulse LiDAR data: Geomorphology, Vol. 173–174, pp. 30–42. VAN WESTEN, C. J., 1998, GIS in landslide hazard zonation: A view, with cases from the Andes of Colombia. In Martin, F. P. and Heywood, D. I. (Editors), Mountain Environment and Geographic Information Systems: Taylor & Francis, London, U.K., pp. 35–165.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 221–235

235

Using Google Earth and Google Street View to Rate Rock Slope Hazards

WILLIAM SWANGER YONATHAN ADMASSU1

Department of Geology and Environmental Science, James Madison University, Harrisonburg, VA 22807

Key Terms: Rockfall, Rockfall Hazard Rating, Google Earth, Google Street View ABSTRACT Rockfall hazard from cut slopes along highways are caused primarily by unfavorable orientations of discontinuities, presence of unconsolidated cobble/boulder deposits, undercutting of strong rocks by weaker rocks, or degradation of weak rock masses. The rockfall hazard rating system (RHRS) was introduced in Oregon to evaluate the hazard and associated risk to an adjacent transportation facility for a cut slope’s potential for releasing rockfalls. RHRS is a numerical score–based rating of parameters that characterize rockfalls. The parameters include slope geometry (height, angle, roughness, orientation), geologic information (discontinuity characterization, undercutting susceptibility), driver’s line of sight, and climate. Geologic information, such as discontinuity orientation data, is traditionally collected using a transit compass and measuring tape at the site. The method is time consuming and expensive and can be dangerous. This study tests the use of Google Earth and Google Street View tools to remotely collect data for selected parameters that characterize rockfall hazard. The selected parameters are categorized under slope profile, geologic characteristics, and impact factor parameters, which are quantitatively and qualitatively measurable using Google Street View and Google Earth. A section of U.S. 33 with a high density of road cuts and two more sites along Interstate 64, all located in Virginia, were selected for the study. Sites were evaluated by using a combination of measurement tools available in Google Earth and a visual inspection of the rock units in Google Street View. The results of seven of the sites were re-evaluated using field-derived data.

1 Corresponding

author email: yonathan.admassu@gmail.com

INTRODUCTION Rockfalls typically occur from unstable cut slopes proximal to highways constructed in mountainous regions. Rockfalls can cause traffic delays, and remediation work can cost millions (Baillifard et al., 2003). Rockfalls are caused by internal and external influences (Higgins and Andrew, 2012). Internal influences include rock mass properties, such as discontinuity properties and rock type, whereas external influences refer to rainfall, snowmelt, groundwater seepage, water runoff, weathering, erosion, freeze–thaw cycles, tree roots, disturbance by animals, and earthquakes. Pierson and Van Vickle (1993) and Pierson (2012) outline steps to proactively assess rock slopes along highways with respect to the potential of rockfall hazard. The proposed steps begin with slope inventory, preliminary rating to be followed by detailed rating with respect to the potential of rockfall release, preliminary design and cost estimate for most serious sections, rockfall mitigation project identification and development, and routine review of rockfall sites and a regular update of database information. According to Pierson and Van Vickle (1993), Oregon’s detailed ranking of slopes with respect to rockfall hazard was based on a rockfall hazard rating method known as the rockfall hazard rating system (RHRS) adopted by the Oregon Department of Transportation (ODOT) in 1989. The basis of the RHRS were earlier works by Brawner and Wyllie (1976) and Wyllie (1987). Brawner and Wyllie (1976) grouped slopes based on their rockfall event potential, whereas Wyllie (1987) suggested scoring certain parameters that control rockfall events. Wyllie’s (1987) approach was later adopted and modified by ODOT to become the widely used RHRS (Pierson and Van Vickle, 1993). RHRS parameters include slope height, catchment ditch effectiveness, average vehicle risk, percent of decision sight distance, roadway width, geologic character, rock block size, climate conditions/presence of water, and rockfall history. Each of these parameters is scored based on an exponential scoring of 3, 9, 27, and 81. The overall rating is the sum of individual scores. Higher scores indicate higher rockfall hazard.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

237

1 Corresponding

author email: yonathan.admassu@gmail.com

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

237

Swanger and Admassu

The benefits of the RHRS according to Pierson and Van Vickle (1993) and Pierson (2012) are (1) creation of a rockfall inventory, (2) education of the public, (3) reduction of liability for transportation agency, (4) provision of tools for strategic planning, and (5) usefulness to other applications, such as housing projects and dams. RHRS evaluations are also useful in the decision-making process by state DOTs for designing a rockfall warning system and identifying cut slopes that have high potentials for generating rockfalls and therefore are in heightened need of remediation. Many state DOT offices adopted the RHRS, and some modified it to better fit the geology and associated hazards in their respective states (Bateman, 2003; Pierson, 2012). According to Russell et al. (2008) and references therein, California, Kentucky, Nevada, Pennsylvania, Virginia, Wyoming, North Carolina, Utah, and West Virginia use the RHRS unmodified, whereas Colorado, Arizona, New Jersey, Vermont, New York, Tennessee, Idaho, Ohio, New Hampshire, and Washington have either modified the RHRS or formulated their own. This information may have changed since the publication of Russell et al. (2008). For example, in 2012 Virginia adopted a hazard rating system based on rock mass rating, rockfall frequency, and rockfall energy (http://www.virginiadot.org/business/resources/bumat-moi-3.pdf). One example of modification of the RHRS is that of Colorado’s, which was originally a slight modification of the RHRS (Andrew, 1994) and in 1997 and 2003 included more parameters pertaining to geologic attributes that are unique to the Colorado Front Range. Other rockfall rating methods, such as New York’s, is organized based on three factors—geological, section, and human exposure factors—to calculate the rockfall risk (Maerz et al., 2005). The geological factor refers to rock properties that may lead to failure, section factor considers the risk of a rockfall reaching the roadway, and human exposure considers the risk of a motorist being hit by a rockfall or driving over a fallen rock (Maerz et al., 2005). The Missouri DOT modified the rating method to separate consequences and risks of failure (Maerz et al., 2005; Youssef and Maerz, 2010). Some parameters, such as geologic characteristics, are rated based on risk of failure, whereas catchment ditch effectiveness are rated in terms of consequence. The overall Missouri RHRS provides a rating reflecting both risk and consequence-of-failure scores (Maerz et al., 2005; Youssef and Maerz, 2010). Although rockfall hazard rating methods are demonstrably beneficial, some problems have been cited by Russell et al. (2008). One major problem is that some of the hazard parameters, such as slope roughness, geologic characteristics, climate, and pres238

ence of water in the original RHRS and the modified systems, are rated based on subjective terminology that may lead to inconsistent rating, depending on the evaluator. For example, the difference between “possible launching features” and “some minor launching features” can be vague. RHRS surveyors need to be well trained in order to maintain consistent results (Pierson, 1991). Even with sufficient training, inconsistencies are still likely to occur. Modifications by several states have removed subjective language (Vandewater et al., 2005; Russell et al., 2008; Santi et al., 2009; and Metzger et al., 2014). The latest Colorado rockfall hazard evaluation chart, which is a modified RHRS, has attempted to address the problem by attaching numerical explanations to qualitative subjective terminologies to minimize inconsistent scoring. For example, “< 2ft surface variation” is added in parentheses to the “Minor Launching Features” rating of surface roughness to further qualify the rating. In addition to the problem of inconsistency, Russell et al. (2009) point out the problem of equally weighing all parameters, which may lead to higher-rated slopes with favorable geologic conditions due to higher scores from non-geological parameters. The mixing of hazard-related parameters, such as geologic characteristics, with risk-related ones, such as catchment ditch effectiveness, also makes interpretation of RHRS ratings more complicated (Maerz et al., 2005). The data to be collected for the RHRS or other rockfall hazard evaluation require physical investigations through field visits (Piteau and Martin, 1977). The process is potentially dangerous and time intensive. However, the use of remote sensing methods to evaluate rockfall hazard can shorten the rating process and make it much less expensive. Largearea GIS-based, rockfall susceptibility, and rockfallrunout-distance studies have become advantageous with the growth of geographic information techniques and availability of high-resolution aerial digital elevation models (Ayala-Carcedo et al., 2003; Baillifard et al., 2003; Shirzadi et al., 2012; and Dorren and Seijmonsbergen, 2003). Road-level remote sensing methods, such as videography, photogrammetry, and terrestrial LiDAR, that are suitable for rockfall hazard rating purposes are not widely available and are prohibitively expensive. The Missouri DOT, as part of the preliminary and detailed rockfall survey, use a scaled video that is captured from a vehicle (Maerz et al., 2005; Maerz and Youssef, 2012). Slope length, slope height, ditch width, ditch depth, rock height, and rock length can be measured from the scaled video logs (Maerz et al., 2005; Maerz and Youssef, 2012). The Tennessee DOT also uses video and photograph logs for preliminary screening of the potentially most hazardous sites (Pierson and Turner, 2012). Lato et al. (2009) show the

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Swanger and Admassu

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Rock Fall Hazard Rating

use of terrestrial LiDAR acquired from a vehicle to collect discontinuity parameters for kinematic analysis of rock slopes along transport corridors. Since its inception in 2005, Google Earth’s highresolution, color aerial photographs have been used as an aid for geological mapping and interpreting existing geologic maps. Google Earth is also capable of storing geologic features in the form of points, lines, and polygons. Google Street View was introduced in 2008, providing seamless photographs of natural street-level views and built-up features adjoining roadways. Google Street View presented an opportunity to use photos of outcrops along roadways for geologic research and education. Similar to video logs used by the Missouri and Tennessee DOTs, Google Street View allows one to visually inspect slopes adjacent to roadways, make linear measurements, and collect qualitative information. This unique tool also allows virtual driving along roadways to identify rockfall problem areas. The purpose of this study is to demonstrate the use of Google Earth and its Google Street View to perform rockfall hazard ratings of rock slopes along highways with respect to parameters that are measureable with Google Earth and Google Street View and to examine the accuracy of the method. RESEARCH METHODS This research evaluates the use of Google Earth and Google Street view as an aid for rockfall hazard rating. The parameters to be evaluated using the proposed method were selected from Colorado’s original (Andrew, 1994) and revised rockfall hazard rating charts (Russell et al., 2008). The selected parameters shown in Table 1 were found to be measurable by Google Earth and Google Street View. Below is a list of the rockfall hazard parameters that are used for this study: 1. Slope profile a. Slope height b. Slope length c. Slope aspect d. Slope inclination e. Slope roughness 2. Geologic characteristics a. Structural condition b. Degree of interbedding c. Degree of undercutting d. Average block size 3. Impact factors a. Sight distance b. Catchment ditch width The level of precision needed to score some of the parameters, as outlined in Colorado’s improved rockfall hazard rating chart (Russell et al., 2008; Santi et al.,

Rock Fall Hazard Rating

2009), is not attainable using Google Street View. Because of this, some subjective language from the original RHRS is kept to evaluate parameters, such as slope roughness and structural conditions. All other parameters are quantifiable. Data for this study were taken primarily from cut slopes located along U.S. 33 on the Virginia side of Shenandoah Mountain (Figure 1). The main geologic factor leading to rockfall hazard along U.S. 33 is the presence of undercut sandstone layers belonging to the Hampshire Formation, a Devonian-age rock unit comprised of interbedded red to brown sandstone and red shale. Forty-one sites were evaluated for rockfall hazards along a 3.5-mi section of U.S. 33. Due to the winding nature of the road, cut slopes that were otherwise continuous around turns were treated as separate sites due to changes in slope aspect. Sites were also separated based on significant differences in slope heights and the degree of interbedding. Two additional cut slope sites were included, one located on Interstate 64 near the city of Waynesboro at mile marker 90 (I-64-1) consisting of limestone of the Beekmantown Formation and another where I-64 crosses Afton Mountain at mile marker 101 (I-64-2) consisting of a Precambrian metabasalt (Catoctin Formation). The I-64-1 site has a structural condition of low-hazard rating score, whereas the Afton cut site, I-64-2, was chosen because of its history of significant rockfalls and was expected to have a high-hazard rating score. Therefore, the sites used for this study represent both undercutting-induced and discontinuity orientation control. Using the rockfall hazard parameters shown in Table 1, a rockfall hazard rating data collection sheet was created in Microsoft Excel. Table 2 shows an example of a completed data collection sheet containing data that were collected from Google Earth for one of the sites along U.S. 33. The data collection form is organized based on site information, slope profile, geologic properties, and impact factors (Table 1). The site profile includes slope height, slope length, slope aspect, slope roughness, and slope inclination. Geologic properties include structural condition, degree of interbedding, degree of undercutting, and average rock block size. Given similar geologic conditions, the impact of rockfalls from two cut slopes might vary depending on the presence of adequate catchment ditch and longer decision sight distances. Therefore, decision sight distance and catchment ditch width are grouped under the “Impact Factors” category. Google Street View screenshots of at least two views were included to accompany the data collection sheet. Screenshots of potential problem areas were also provided (Figure 2). Cut slope sites in the form of lines were saved as KML files. Every site’s rockfall hazard

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

239

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

239

Swanger and Admassu

Table 1. RHRS parameters that are measureable with Google Earth.

Score Factor

3 Points

Slope profile Slope height (ft) Slope length (ft) Slope aspect Slope inclination (◦ ) Slope roughness Geologic characteristics Structural condition Degree of interbedding Degree of undercutting (ft) Average block size Impact factors Sight distance (%) Catchment ditch effectiveness (%)

9 Points

Score 27 Points

81 Points

25–50 0–250 N 15–25 Possible launching features

50–75 250–500 E, W, NE, NW 25–35 Some minor launching features

75–100 500–750 SE, SW 35–50 Many launching features

>100 >750 S >50 Major rock launching features

Discontinuous or continuous fractures, favorable orientation 1–2 weak interbeds, <6 in. 0–1 6–12 in.

Discontinuous or continuous fractures, random orientation 1–2 weak interbeds, >6 in. 1–2 1–2 ft

Discontinuous fractures, adverse orientation >2 weak interbeds, <6 in. 2–4 2–5 ft

Continuous fractures, adverse orientation

>80 100–95

60–80 95–65

40–60 65–30

<40 30–0

>2 weak interbeds, >6 in. >4 >5 ft

Factor Slope profile Slope height (ft) Slope length (ft) Slope aspect Slope inclination (◦ ) Slope roughness Geologic characteristics Structural condition Degree of interbedding Degree of undercutting (ft) Average block size Impact factors Sight distance (%) Catchment ditch effectiveness (%)

Figure 1. Location map of study sites on U.S. 33 and I-64 (Google Earth Imagery).

240

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

3 Points

9 Points

27 Points

25–50 0–250 N 15–25 Possible launching features

50–75 250–500 E, W, NE, NW 25–35 Some minor launching features

75–100 500–750 SE, SW 35–50 Many launching features

>100 >750 S >50 Major rock launching features

Discontinuous or continuous fractures, favorable orientation 1–2 weak interbeds, <6 in. 0–1 6–12 in.

Discontinuous or continuous fractures, random orientation 1–2 weak interbeds, >6 in. 1–2 1–2 ft

Discontinuous fractures, adverse orientation >2 weak interbeds, <6 in. 2–4 2–5 ft

Continuous fractures, adverse orientation

>80 100–95

60–80 95–65

40–60 65–30

<40 30–0

Figure 1. Location map of study sites on U.S. 33 and I-64 (Google Earth Imagery).

240

81 Points

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

>2 weak interbeds, >6 in. >4 >5 ft

Rock Fall Hazard Rating

Table 2. An example of a completed RHRS data sheet.

Site Information Date Location Site Designation Lithology/Age Comments

8/2/2015 West Rockingham County, VA/U.S. 33/Shenandoah Mt. Rt. 33-41 Interbedded sandstone and shale/Miss-Devonian

Slope Profile Slope Height Slope Length Slope Aspect Slope Inclination Slope Roughness

Rank 3 3 27 81 9 Slope Crest 1 38.578895◦ −79.168191◦ 3,466 ft Slope Toe 38.578834◦ −79.168183◦ 3,455 ft

Values 13 ft 84 ft 158◦ >50◦ Some minor launching features Slope Crest 2 38.578918◦ −79.168059◦ 3,469 ft

Slope Crest 3 38.578973◦ −79.167962◦ 3,472 ft

Latitude Longitude Elevation

38.578863◦ −79.168024◦ 3,455 ft

38.578915◦ −79.167915◦ 3,457 ft

Latitude Longitude Elevation

Geologic Characteristics Structural Condition Plane Wedge Toppling Degree of Interbedding Degree of Undercutting Block Size

Rank

Values

3 3 3 3 3 9

Fractures, favorable orientation Fractures, favorable orientation Fractures, favorable orientation 0 weak interbeds, <6 in. <1 ft 1.5 ft

Other Impact Factors Sight Distance (Westbound) Sight Distance (Eastbound) Catchment Ditch Width Total Hazard Score Relative Risk

Rank 81 27 9 231 Minimum Score Maximum Score

Values 30.7 43.5 6 ft

Latitude Longitude Elevation Latitude Longitude Elevation

rating table was prepared in HTML that a user can access by clicking on an individual site in Google Earth (Figure 3). DATA COLLECTION Site Information Site information contains basic descriptors of the survey site that include location, date, site designation, and lithology (Tables 1 and 2). The comments section is to note any site-specific information about the site that does not have a place on the data sheet. Examples of such information can be the presence or absence of a rockfall barrier, extreme overhang, large visible fractures, or large dislodged blocks that could make the site hazardous. Slope Profile Slope Height Slope height is the vertical height from the road to the highest point on the crest of the slope (Tables 1

33 891

and 2). Rockfalls from high slopes possess more energy and therefore are more hazardous. This value is obtained by calculating the average height difference between the crest and the toe of the slope. Multiple elevation and latitude and longitude readings were taken in Google Earth along the crest and toe of the slope and recorded on the RHRS worksheet (Tables 1 and 2). An average slope height is determined using the difference in elevation from the crest to the toe. Another method to calculate slope height is from the elevation profile of a path drawn perpendicular to the slope (Figure 4).

Slope Length Longer slopes have a higher probability of releasing more rockfalls. Slope length is the length of road that runs parallel to the cut slope (Tables 1 and 2). It is measured in Google Earth using the Ruler tool (Figure 4).

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

241

8/2/2015 West Rockingham County, VA/U.S. 33/Shenandoah Mt. Rt. 33-41 Interbedded sandstone and shale/Miss-Devonian

Slope Profile Slope Height Slope Length Slope Aspect Slope Inclination Slope Roughness

Rank 3 3 27 81 9 Slope Crest 1 38.578895◦ −79.168191◦ 3,466 ft Slope Toe 38.578834◦ −79.168183◦ 3,455 ft

Values 13 ft 84 ft 158◦ >50◦ Some minor launching features Slope Crest 2 38.578918◦ −79.168059◦ 3,469 ft

Slope Crest 3 38.578973◦ −79.167962◦ 3,472 ft

38.578863◦ −79.168024◦ 3,455 ft

38.578915◦ −79.167915◦ 3,457 ft

Geologic Characteristics Structural Condition Plane Wedge Toppling Degree of Interbedding Degree of Undercutting Block Size

Rank

Values

3 3 3 3 3 9

Fractures, favorable orientation Fractures, favorable orientation Fractures, favorable orientation 0 weak interbeds, <6 in. <1 ft 1.5 ft

Other Impact Factors Sight Distance (Westbound) Sight Distance (Eastbound) Catchment Ditch Width Total Hazard Score Relative Risk

Rank 81 27 9 231 Minimum Score Maximum Score

Values 30.7 43.5 6 ft

33 891

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

241

Swanger and Admassu

Figure 2. Examples of Google Street View images of one U.S. 33 site showing undercutting-induced failures.

Swanger and Admassu

Slope Aspect

Slope aspect indicates the direction the cut slope is facing. South-facing slopes will carry more risk because they get the most direct sunlight and thus are more affected by freeze–thaw cycles. This parameter is measured in Google Earth by using the Ruler tool. Using the Ruler tool, a line can be digitized perpendicularly from the crest to the toe of the slope, and the tool will calculate the “heading” of the line, which is the aspect of the slope (Figure 5). So, a slope aspect of 0◦ faces north, a slope aspect of 180◦ faces south, and slope aspects of 90◦ and 270◦ face east and west, respectively (Tables 1 and 2).

Slope Inclination

Slope inclination correlates with velocity—and thus energy and trajectory of rockfalls—and therefore needs to be rated. Slope inclination can be calculated using the Ruler tool, which provides the map distance, which is the horizontal distance between the beginning and end of the ruler, and the ground length, which is the profile length along a slope face (Figure 5). The slope inclination is therefore cos−1 (map length/ground length). Slope inclination can also be estimated from the elevation profile (Figure 4).

Slope Roughness

Slope roughness gives an indication of potential launching features present on a slope surface. More launching features means that there is a higher risk that falling rocks will be directed farther toward the roadway. One must use photographic examples of slopes with various ratings to infer what score is appropriate for each slope (Russell et al., 2008; Santi et al., 2009). By changing the viewing angle in Google Street View, the presence of launching features can be qualitatively evaluated. By comparing dimensions of launching features visible in Google Street View with objects of known dimensions (road width and slope height), semi-quantitative slope roughness can be assigned (Tables 1 and 2).

Figure 2. Examples of Google Street View images of one U.S. 33 site showing undercutting-induced failures.

Geologic Characteristics

Figure 3. Cut slopes on U.S. 33 are subdivided into separate sites based on slope aspect and slope height. A table accessible by clicking the sites shows the RHRS data collected and the total rating.

242

Geologic Characteristics

Structural Condition

There are three types of structurally controlled slope failures that may lead to rockfall generation: plane, wedge, and toppling failures. The potential for each type of failure is characterized by the dip angle of the discontinuities and the plunge of their intersections. The risk for plane and wedge failures depends on whether discontinuities such as bedding planes,

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Figure 3. Cut slopes on U.S. 33 are subdivided into separate sites based on slope aspect and slope height. A table accessible by clicking the sites shows the RHRS data collected and the total rating.

242

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Rock Fall Hazard Rating

Figure 4. (a) The slope length is calculated by drawing a path along the toe of the slope and obtaining its length. (b) Slope height is calculated as the difference in elevation from the crest to the toe. Slope height and inclination can also be measured from the elevation profile of any path drawn in Google Earth.

foliation, joints, or their lines of intersections are oriented in such a way that they daylight onto the slope face and are steeper than the friction angle (Hoek and Bray, 1981). Toppling failure result from discontinuities that are parallel with the slope and dip into the slope.

Evaluation of potentials of plane, wedge, and toppling failures is traditionally performed by plotting orientation data on a stereonet along with slope face orientation and the friction angle (Hoek and Bray, 1981). The assessment for structural failures was performed in

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237â&#x20AC;&#x201C;250

243

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237â&#x20AC;&#x201C;250

243

Swanger and Admassu

Figure 5. Slope inclination and slope aspect can be determined using the Ruler tool. The ground length is the length along the slope, and map length is the horizontal distance. The slope angle can be calculated as cos−1 (map length/ground length). The Ruler tool also provides the slope aspect as the “Heading.”

Google Street View by visually examining orientations of discontinuities. Discontinuities or their intersections daylighting on the slope face can be visible in Google Street View. The potentials for each failure type— planar, wedge, and toppling—is rated and scored. To be conservative, only the highest score is used for the total hazard score. The bedding along this section of U.S. 33 is subhorizontal, so rockfalls are typically generated by undercutting (Figure 2). I-64-2 consists of discontinuities (foliation) daylighting on the slope face, promoting plane failure (Figure 6), whereas I-64-1 consists of bedding nearly perpendicular to the slope face, causing no daylighting conditions (Figure 7). These interpretations are subjective and dependent on visual inspection and not a result of factor of safety calculations. Degree of Interbedding The degree of interbedding is dependent on alternating lithology among rocks on the cut slope. This

Figure 6. Site I-64-2 consists of metabasalt (Catoctin Formation) with foliation daylighting on the slope face promoting plane failure.

244

Swanger and Admassu

Figure 7. Site I-64-1 consists of sub-vertical limestone (Beekmantown Formation) oriented nearly perpendicular to the slope face.

is common primarily with interbedded sandstone and shale layers. Weak rock units, such as shale and claystone/mudstone, can cause undercutting-induced rockfalls as they erode faster than interbedded stronger rocks (sandstone, limestone, dolomite). It is not only the number of weak layers present but also their thickness that increase rockfall hazard (Tables 1 and 2). Thicker layers of weak interbedded material increase the risk of undercutting, which will allow large blocks to be fall onto the roadway. This parameter is analyzed visually in Google Street View. Thickness of layers was estimated using slope height and road as a scale of reference. Degree of Undercutting Undercutting on cut slopes shows where weak layers have been eroded and more resistant layers are still present. The deeper that a rock layer is undercut, the higher the potential for a large mass to fall. Undercutting-induced rockfalls occur not only where interbedded rocks exist but also where differential joint spacing exists. Highly jointed rocks can undercut rocks with wider discontinuity. The depth of undercutting is estimated in Google Street View using road width and slope height as scale references (Tables 1 and 2).

Figure 7. Site I-64-1 consists of sub-vertical limestone (Beekmantown Formation) oriented nearly perpendicular to the slope face.

Table 3. Required decision sight distance based on posted speed limits (Pierson et al., 1993).

Posted Speed Limit (mph)

Decision Sight Distance (ft)

Posted Speed Limit (mph)

Decision Sight Distance (ft)

25 30 35 40 45 50 55 60 65

375 450 525 600 675 750 875 1,000 1,050

25 30 35 40 45 50 55 60 65

375 450 525 600 675 750 875 1,000 1,050

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Figure 6. Site I-64-2 consists of metabasalt (Catoctin Formation) with foliation daylighting on the slope face promoting plane failure.

244

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Rock Fall Hazard Rating

Average Block Size

Block size is a function of discontinuity spacing and controls the energy of rockfalls. The average block size of rocks that are likely to fall onto the road can be rated using Google Street View (Tables 1 and 2). By looking at rocks in the ditch and on the slopes that have already fallen, along with cavities (Figure 6) on the slope from where rocks have fallen, an estimate for average block sizes can be made. The slope height and roadway width were used as scale for estimating rock block dimensions.

Impact Factors

Figure 8. Rockfall modeling was performed using RocFall. Various situations were modeled to determine the minimum catchment ditch width that would contain 100 percent of rockfalls.

Sight Distance

Sight distance is a measure of how much time a driver would have to react after noticing an object that has fallen onto the road. The more time a driver has to maneuver a vehicle, the smaller the chance of driving into a fallen rock. This usually depends on the curviness of the roadway. Sight distance is a percentage ratio of the actual decision sight distance with the required decision sight distance. The actual decision sight distance is the farthest straight line distance from a point on the roadway that a fallen rock is visible. The required decision sight based on the posted speed limit is provided in Pierson and Van Vickle (1993), who used the AASHTO decision sight distances to develop a scale for the relative risk of a section based on the speed limit of the road (Table 3). The sight distance is calculated as follows: Sight distance =

Impact Factors

Figure 8. Rockfall modeling was performed using RocFall. Various situations were modeled to determine the minimum catchment ditch width that would contain 100 percent of rockfalls.

Actual decision sight distance ×100. Required decision sight distance

The actual decision site is measured in Google Earth by using a combination of Google Earth and Google Street View. While virtually driving in Google Street View toward the slope cut, the farthest point where the cut slope is visible is noted and then marked with a place marker in Google Earth. The distance from the place marker to the cut slope is measured using the Ruler tool in Google Earth as the actual decision sight distance. The sight distance parameter was measured for eastbound and westbound lanes. The highest rating of the two is used for the overall RHRS (Tables 1 and 2). Catchment Ditch Effectiveness The catchment ditch is located at the base of a cut slope to contain rockfalls so that they do not reach the roadway. Ritchie (1963) and Pierson and Van Vickle (1993) provide recommendations for catchment ditch width and depth. The recommended catchment ditch

and width are based on slope heights and angles. The catchment ditch effectiveness for the revised Colorado RHRS is calculated as follows (Russell et al., 2008; Santi et al., 2009): Ditch effectiveness = ((Da + Wa) / (Dr + Wr)) × 100, where Da and Wa are actual ditch depth and width, respectively, and Dr and Wr are Ritchie-recommended catchment depth and width, respectively. Using Google Street View, catchment ditch width is estimated based on roadway width as scale reference. Estimating the depth was difficult from Google Street View, but most catchment ditches at the study sites appeared to be flat. The slope heights along U.S. 33 were shorter than the slopes in Ritchie’s (1963) or Pierson’s (1991) catchment ditch recommendations. Therefore, rockfall modeling was performed using the RocFall program from Rocscience (Rocscience Inc., 2001) to determine catchment ditch dimensions that can contain 100 percent of rockfalls (Figure 8). Two types of slopes, slopes with talus and near vertical slopes, were modeled for multiple drop heights, and the minimum ditch width that would contain 100 percent of rockfalls was determined. The coefficients of rockfall tangential and normal restitution values used in RocFall were determined from back calculations based on actual rockfall landing sites in catchment ditches. The calculated ditch effectiveness for the sites was a ratio percentage of catchment ditch width determined from Google Street View to the minimum catchment ditch width that contains 100 percent of rocks as determined from RocFall simulation. The effectiveness of catchment ditch width is scored based on the calculated catchment effectiveness percentage. RESULTS A total hazard rating is calculated by adding the scores of the 11 individual parameters. The minimum

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

245

Actual decision sight distance ×100. Required decision sight distance

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

245

Swanger and Admassu

Figure 9. A bar graph showing the rockfall hazard rating of all U.S. 33 sites and the additional sites on I-64.

and maximum possible score is 33 and 891, respectively. The highest rating along the U.S. 33 section was 411, and the lowest was 147 (Figure 9). The results show that the highest risk parameters for the U.S. 33 section are sight distance, degree of interbedding, and slope aspect. The presence of undercutting in cut slopes is the cause of rockfalls in the U.S. 33 section, but the highest risk areas are found proximal to sharp turns where the sight distance score would be high. I-64-1 had a total hazard rating score of 168, and site I-64-2 had a total hazard rating score of 471. This agrees well with the expected results, as I-64-1 is comprised of Beekmantown limestone and dolostone with sub-vertical bedding that is oriented perpendicular to the roadway, resulting in a low rating of 168 (Figure 7). With a hazard score of 168, site I-64-1 has a hazard score close to the lowest risk slopes from the sites on U.S. 33. I-64-2, on the other hand, is comprised of Catoctin metabasalt with discontinuities (foliation) dipping toward the roadway and daylighting in the slope face (Figure 6). This makes the site highly prone to planar failures with a total hazard rating score of 471. A proactive measure has been taken at the site by placing a fence along the cut to catch rockfalls. Field Verification The results of rockfall hazard rating based on Google Earth and Google Street View were compared with field-measured values. Similar parameters were evaluated in the field at five of the sites on U.S. 33 and the two additional sites (I-64-1 and I-64-2). The methods used for field verification of the parameters are summarized below. Slope Height A laser range finder was used to determine the distance to the toe and crest of the slope from a common point facing the slope. Vertical angles were determined 246

Swanger and Admassu

using a clinometer. The height was then determined by simple trigonometry. Multiple measurements were made, and the average value was recorded.

Slope Length

A measuring tape was stretched along the toe of the slope to determine slope length.

Slope Aspect

A Brunton compass was used to determine the slope facing direction (slope aspect). Multiple measurements were taken, and the average was used.

Slope Inclination A Brunton compass was used to sight the slope profile and determine the slope angle. The average of multiple measurements was used. Most of the sites on U.S. 33 are vertical and were assigned 90◦ . Slope Roughness and Structural Condition These parameters were visually determined qualitatively. Slope roughness was evaluated in terms of the presence of rockfall launching features. Structural condition is based on visual evaluation of orientation of discontinuities and their likelihood to cause plane, wedge, and/or toppling failures. Degree of Interbedding The number and thicknesses of weak undercutting layers were counted and recorded. Degree of Undercutting Multiple measurements of depth of undercutting were taken using a ruler. The average value was used. Sight Distance Due to safety concerns, sight distance was not measured in the field. Therefore, sight distance values determined from Google Earth were used. Average Block Size Rockfalls in catchment ditches were measured with a ruler, and average values were used. Appendix 1 provides field-derived versus Google Earth/Google Street View–derived measurements. Table 4 compares the resulting scores. Out of the five sites on U.S. 33, only one, Route 33-3, resulted in similar rating. The resulting differences were mainly in degree of undercutting and block size. Sites I-64-1 and I-64-2 did not show differences.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Figure 9. A bar graph showing the rockfall hazard rating of all U.S. 33 sites and the additional sites on I-64.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

247 Slope profile Slope height (ft) Slope length (ft) Slope aspect (◦ ) Slope roughness Slope inclination (◦ ) Geologic properties Structural condition Plane Wedge Toppling Degree of interbedding (no. weak interbeds, thickness in inches) Degree of undercutting (ft) Block size (ft) Impact factors Sight distance (%) Catchment ditch width (ft) Total % difference

Parameter

Slope profile Slope height (ft) Slope length (ft) Slope aspect (◦ ) Slope roughness Slope inclination (◦ ) Geologic properties Structural condition Plane Wedge Toppling Degree of interbedding (no. weak interbeds, thickness in inches) Degree of undercutting (ft) Block size (ft) Impact factors Sight distance (%) Catchment ditch width (ft) Total % difference

Parameter

0 3

0 0

0 3

0 0 2.7

0 0

0 0 0 0 0

Score

4 9 16 0 0

Value

0 0 0 0 0

33-3

0 0 0 0 6

0 0 0 0 0

0 0

33-12

0 0

5 2 10 0 0

Value

6.7

0 0

0 0 0 0 0

Score

33-19

0 0

1.5

2.5

9 11 4 0 0

Value

0 0

0 0 0 0 0

Score

33-20

0 0

11 20 10 0 0

Value

0 18

0 0 0 0 6

0 0 0 0 0

Score

33-40

0 1

1.5

8 53 4 0 0

Value

0 0

0 0 0 0 0

Score

33-12

0 0

5 2 10 0 0

Value

6.7

0 0

0 0 0 0 0

Score

33-19

0 0

1.5

2.5

9 11 4 0 0

Value

0 0

0 0 0 0 0

Score

33-20

Site

0 0

11 20 10 0 0

Value

0 18

0 0 0 0 6

0 0 0 0 0

Score

33-40

0 1

1.5

8 53 4 0 0

Value

0 0

0 0 0 0 0

Score

Table 4. Comparison of field-derived and Google Earth/Google Street View–derived values and corresponding scores.

Score

2.7

0 0 0 0 0

Score

0 0 0 0 0

4 9 16 0 0

Value

0 0 0 0 6

0 0 0 0 0

Score

33-3

Site

Table 4. Comparison of field-derived and Google Earth/Google Street View–derived values and corresponding scores.

64-1

0 6

8 25 6 0 13

Value

0 6

8 25 6 0 13

Value

0 0

0 0 0 0 0

Score

0 0

0 0 0 0 0

Score

64-2

0 5

14 36 5 0 2

Value

0 5

14 36 5 0 2

Value

Rock Fall Hazard Rating Rock Fall Hazard Rating

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

247

Swanger and Admassu

DISCUSSION Although rockfall hazard rating parameters can be evaluated semi-quantitatively with Google Earth/Google Street View, there are observed limitations: (1) Dates of photos taken for Google Street View vary by location. Some changes may have occurred since the date of these photos were taken. (2) The time of the year that Google Street View photos were taken also affects the visibility of cut slopes, especially if taken during summer months when the vegetation cover is thick. (3) High cut slopes with gentle angles may not totally be visible in street view, as higher parts fall outside the field of view the Street View camera. (4) Direct linear and orientation measurements are not possible in Google Street View, making it impossible to quantify parameters such as the degree of undercutting. The accuracy of measurements obtained from Google Earth/Google Street View and field visits is evaluated. Below is a detailed discussion on the differences in measurements and scores of parameters evaluated using both methods. Slope Profile Parameters Four of the slope profile parameters—slope height, slope length, slope aspect, and slope inclination—are quantifiable. The slope height values show an average approximate 9-ft difference between remotely and fieldderived values for the seven sites. The differences could be due to the resolution of the digital elevation model embedded in Google Earth. It can also be due to the difficulty in identifying the crest of the cut slope on the slope profile (Figure 4). Slope length values show an average of an approximate 22-ft difference. This may be due to errors in misidentifying exact ground positions of start and end points of rock slopes that were determined in Google Earth. Slope aspect and slope inclination measurements did not show much variation (2◦ to 5◦ ). Slope aspect can be accurately measured in Google Earth. Slope inclination, however, depends on the accuracy of Google Earth’s digital elevation model. Slope roughness is a qualitative evaluation of the presence of rockfall launching features. Google Street View photos, if not affected by vegetation cover, can easily show rockfall launching features. No difference in scoring between the field and remote methods was observed. Geologic Characteristics Parameters Measuring orientation and spacing of discontinuities to evaluate the structural condition is impossible in Google Street View. However, the structural condition is evaluated visually and scored qualitatively on 248

Swanger and Admassu

Google Street View images. The scoring of the structural condition using field and Google Street View did not show any difference. In regard to determining the degree of undercutting, the number of weak interbeds can visually be determined in Google Street View. Exact measurements are required for thickness of weak interbeds, depth of undercutting, and block sizes. These measurements are indirectly determined in Google Street View by using slope height and road width as scale reference. This introduced errors in measurements that explain differences in scores of up to 54 points (Table 4). Impact Factors Parameters Sight distance was determined in Google Earth. Due to safety issues, field measurements were not accomplished. Catchment ditch width and depth were measured in Google Street View using road width and slope height as scale. This introduced errors in catchment ditch dimensions and ultimately in the assigned scores. The difference in the overall rockfall hazard rating between field-obtained and Google Earth/Google Street View was less than 19 percent (Table 4). Three of the seven field-evaluated sites resulted in the same overall score. The main source of difference in the overall rating is due to differences in scores assigned for degree of undercutting, average block size and catchment ditch dimensions. This is due mainly to the narrow range (1 to 2 ft) of quantifiable values of each score range for these parameters (degree of undercutting, and block size). Therefore, small differences in measurements can result in larger differences in scores. On the other hand, there was no difference in slope profile parameter scores despite differences in measured values, mainly because of the wide range of values for each score. For example, each slope height score is within a 25-ft range, and slope length is 250 ft. The difference in overall rockfall hazard rating between field-measured versus remotely measured values was within 19 percent based on the seven sites tested. More sites with varying geology should be used for further validation. The U.S. 33 sites showed more variation in overall rating between field-acquired and remotely acquired data sets than for I-64-1 and I-64-2 sites. I-641 and I-64-2 required fewer quantitative measurements of geologic parameters due to the absence of undercutting prone interlayered rock units, resulting in similar overall rating between field-derived and remotely derived rockfall rating. CONCLUSION The Google Earth/Google Street View–based RHRS method enables the evaluation of several rock

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

Rock Fall Hazard Rating

slopes in 1 to 2 hours. Google Street View basically allows one to “drive” virtually, examining long stretches of road within a fraction of the time required to actually drive from site to site. Once an individual is familiar with the system, the process can take less time. Road-level remote sensing methods, such as the Missouri and Tennessee DOT video surveys, does reduce surveying time but still will require significant amount of driving to acquire images, whereas Google Street View images are available at no charge. The Google Earth–based RHRS survey is also a safe practice, as on-site surveys of cut slopes can have inherent personal risk from traffic or by traversing below steep slopes. The accuracy of the rating can be improved if Google Street View adds linear and orientation measurement tools. The geological experience of the evaluator is also a determining factor in the reliability of the proposed method. The rating of slopes that do not have undercutting related problems require fewer quantitative measurements and therefore can be better evaluated remotely. Despite its limitations, the Google Earth/Google Street View method should be seriously considered for rock slope inventory and a more advanced preliminary rating, helping DOTs become more selective in identifying cut slopes requiring detailed RHRS or equivalent rockfall hazard evaluation methods. It should be stressed that final detailed RHRS evaluation should be field performed in the field by a trained individual. REFERENCES ANDREW, R. D., 1994, The Colorado Rockfall Hazard Rating System: Colorado Department of Transportation, Report CTICDOT-2-94. AYALA-CARCEDO, F. J., CUBILLO-NIELSEN, S., ALVAREZ, A., DOMÍNGUEZ, M. J., LAÍN, L., LAÍN, R., AND ORTIZ, G., 2003, Large scale rockfall reach susceptibility maps in La Cabrera Sierra (Madrid) performed with GIS and dynamic analysis at 1: 5,000: Natural Hazards, Vol. 30, No. 3, pp. 325–340. BAILLIFARD, F., JABOYEDO, M., AND SARTORI, M., 2003, Rockfall hazard mapping along a mountainous road in Switzerland using a GIS-based parameter rating approach: Natural Hazards and Earth System Science, Vol. 3, pp. 435–442. BATEMAN, V., 2003, The development of database to manage rockfall hazard: The Tennessee rockfall harzard database: 2003 Transportation Research Board Annual Meeting, 13 p. BRAWNER, C. O. AND WYLLIE D., 1976, Rock slope stability on railway projects: American Railway Engineering Assoication Bulletin 656, pp. 449–474. DORREN, L. K. AND SEIJMONSBERGEN, A. C., 2003, Comparison of three GIS-based models for predicting rockfall runout zones at a regional scale: Geomorphology, Vol. 56, No. 1, pp. 49–64. HIGGINS, J. D. AND ANDREW, R. D., 2012, Rockfall types and causes. In Turner K. A. and Schuster R. L. (Editors), Rockfall: Characterization and Control: Transportation Research Board, National Research Council, Washington, DC. 658 p.

Rock Fall Hazard Rating

HOEK, E. AND BRAY, J. W., 1981, Rock Slope Engineering: Institute of Mining and Metallurgy, London, U.K. 358 p. LATO, M., HUTCHINSON, J., DIEDERICHS, M., BALL, D., AND HARRAP, R., 2009, Engineering monitoring of rockfall hazards along transportation corridors: Using mobile terrestrial LiDAR: Natural Hazards Earth Systems Science, Vol. 9, No. 3, pp. 935–946. MAERZ, N. H., YOUSSEF, A., AND FRENNESSEY, T. W., 2005, New risk-consequence rockfall hazard rating system for Missouri highways using digital image analysis: Environmental and Engineering Geoscience, Vol. 11, No. 3, pp. 229–249. METZGER, A. T., OLSEN, M., WARTMAN, J., DUNHAM, L., AND STUEDLEIN, A., 2014, A Platform for Proactive Risk-Based Slope Asset Management: Phase I Interim Project Report: Pacific Northwest Transportation Consortium, Washington, DC. PIERSON, L. A., 1991, The Rockfall Hazard Rating System: Oregon Department of Transportation, Final Report FHWA-OR-GT92-05. PIERSON, L. A. AND VAN VICKLE, R., 1993, Rockfall Hazard Rating System Participant’s Manual: Federal Highway Administration Publication SA-93-057, Washington DC, 104 p. PIERSON, L. A., 2012, Rockfall hazard rating systems. In Turner K. A. and Schuster R. L. (Editors), Rockfall: Characterization and Control: Transportation Research Board, National Research Council, Washington, DC. 658 p. PIERSON, L. A. AND TURNER, A. K., 2012, Implementation of rock slope management systems. In Turner K. A. and Schuster R. L. (Editors), Rockfall: Characterization and Control: Transportation Research Board, National Research Council, Washington, DC. 658 p. PITEAU, D. R. AND MARTIN, D. C., 1977, Description of Detail Line Engineering Geology Mapping Method; in Rock Slope Engineering, Part G: Federal Highway Administration, Reference Manual FHWA-13-97-208, Portland, OR. 29 p. RITCHIE, A. M., 1963, Evaluation of rockfall and its control: Highway Research Record, Journal of the Transportation Research Board, Vol. 17, pp. 13–28. Rocscience Inc., 2001, RocFall Version 4.0—Statistical Analysis of Rockfalls. Available at http://www.rocscience.com RUSSELL, C. P., SANTI, P. M., AND HIGGINS, J. D., 2008, Modification and Statistical Analysis of the Colorado Rockfall Hazard Rating System: Colorado Department of Transportation, DTD Applied Research and Innovation Branch, No. CDOT-2008-7. 123 p. SANTI, P. M., RUSSELL, C. P., HIGGINS, J. D., AND SPRIET, J. I., 2009, Modification and statistical analysis of the Colorado Rockfall Hazard Rating System: Engineering Geology, Vol. 104, pp. 55– 65. SHIRZADI, A., SARO, L., JOO, O. H., AND CHAPI, K. 2012, A GIS-based logistic regression model in rock-fall susceptibility mapping along a mountainous road: Salavat Abad case study, Kurdistan, Iran: Natural Hazards, Vol. 64, No. 2, pp. 1639–1656. VANDEWATER, C. J., DUNNE, W. M., MAULDON, M., DRUMM, E. C., AND BATEMANN, V., 2005, Classifying and assessing the geologic contribution to rockfall hazard: Environmental and Engineering Geoscience, Vol. 11, No. 2, pp. 141–154. WYLLIE, D., 1987, Rock slope inventory/maintenance programs: Proceedings of FHWA Rockfall Mitigation Seminar—The 13th Northwest Geotechnical Workshop: Portland, R. YOUSSEF, A. M. AND MAERZ, N. H., 2010, Development, justification, and verification of a rock fall hazard rating system: Bulletin of Engineering Geology and the Environment, Vol. 71, pp. 171–186.

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

249

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250

249

250 Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250 Parameter Slope profile Slope height (ft) Slope length (ft) Slope aspect (◦ ) Slope roughness Slope inclination (◦ ) Geologic properties Structural condition Plane Wedge Toppling Degree of interbedding (no. weak interbeds, thickness in inches) Degree of undercutting (ft) Block size (ft) Impact factors Sight distance (%) Catchment ditch width (ft) Total

Parameter Slope profile Slope height (ft) Slope length (ft) Slope aspect (◦ ) Slope roughness Slope inclination (◦ ) Geologic properties Structural condition Plane Wedge Toppling Degree of interbedding (no. weak interbeds, thickness in inches) Degree of undercutting (ft) Block size (ft) Impact factors Sight distance (%) Catchment ditch width (ft) Total

Google

Field

Google

33-20 Field

Google

33-40 Field

Google

I-64-1 Field

Google

I-64-2 Field

3 81 27

250

34.7 6

357

81 27

3 9 27 27 81

32 6 393

81 27

3 3 3 >2, 81 >6 in. 81

25 387 102

3 3 27 3 81

32 6

273

81 9

3 3 3 >2, 9 >6 in. 9

30 385 112

3 3 27 3 81

35.2 5 255

81 9

3 3 3 1–2, 9 >6 in. 9

10 171 198

3 9 9 9 81

285

81 27

3 9 9 9 81

28.3 5

339

81 27

3 3 3 1–2, 9 >6 in. 9

21 355 75

APPENDIX 1.

35.2 5

1.5

2.5

3 3 3 1–2, 9 >6 in. 9

19 160 202

3 9 27 3 81

28.3 5

177

27 9

3 3 3 1–2, 3 >6 in. 3

32 335 65

3 9 27 3 81

40.3 6

207

27 27

3 3 3 1–2, 9 <6 in. 9

12 403 138

9 27 27 9 81

40.3 5

1.5

171

3 3

3 3 3 1–2, 3 >6 in. 3

20 350 134

9 27 27 9 81

100 28

171

3 3

3 3 3 1–2, 3 <6 in. 3

52 600 219

Field

Google

Field

Google

33-20 Field

Site and Method Google

33-40 Field

Google

I-64-1 Field

Google

100 31

I-64-2

471

3 81

Field

100 36

3 81 27

231

225

3 3 3 1–2, 9 <6 in. 9

3 3 3 3 3

34.7 3

3 3 9 3 81 3

13 174 80

34.7 6

357

81 27

3 3 3 1–2, 81 > 6 in. 81

17 165 96

3 9 27 27 81

32 6

393

81 27

3 3 3 >2, 81 >6 in. 81

25 387 102

3 3 27 3 81

32 6

273

81 9

3 3 3 >2, 9 >6 in. 9

30 385 112

3 3 27 3 81

35.2 5

255

81 9

3 3 3 1–2, 9 >6 in. 9

10 171 198

3 9 9 9 81

35.2 5

1.5

2.5

285

81 27

3 3 3 1–2, 9 >6 in. 9

19 160 202

3 9 9 9 81

28.3 5

339

81 27

3 3 3 1–2, 9 >6 in. 9

21 355 75

3 9 27 3 81

28.3 5

177

27 9

3 3 3 1–2, 3 >6 in. 3

32 335 65

3 9 27 3 81

40.3 6

207

27 27

3 3 3 1–2, 9 <6 in. 9

12 403 138

9 27 27 9 81

40.3 5

1.5

171

3 3

3 3 3 1–2, 3 >6 in. 3

20 350 134

9 27 27 9 81

100 28

171

3 3

3 3 3 1–2, 3 <6 in. 3

52 600 219

81 81 27 3 27

100 34

471

3 81

81 3 3 1–2, 3 <6 in. 3

60 575 225

81 81 27 3 27

100 31

471

3 81

81 3 3 1–2, 3 <6 in. 3

140 816 155

100 36

1–2, <6 in.

154 780 160

3 9 27 27 81

Google

33-19

471

3 81

1–2, <6 in.

154 780 160

3 3 9 3 81

Field

33-12

100 34

81 81 27 3 27

81 3 3 1–2, 3 <6 in. 3

140 816 155

Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value

33-3

81 81 27 3 27

81 3 3 1–2, 3 <6 in. 3

60 575 225

Score Value Score Value

Google

3 3 3 3 1–2, 81 > 6 in. 81

17 165 96

Table A1. Field-derived versus Google Earth/Google Street View–derived rockfall rating data and corresponding scores.

231

225

3 3 3 1–2, 9 <6 in. 9

3 3 3 3 3

34.7 3

3 3 9 3 81

13 174 80

3 9 27 27 81

Field

Site and Method

3 3 9 3 81

Google

33-19

Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value Score Value

Field

33-12

Score Value Score Value

Google

33-3

Table A1. Field-derived versus Google Earth/Google Street View–derived rockfall rating data and corresponding scores.

APPENDIX 1.

Swanger and Admassu Swanger and Admassu

Environmental & Engineering Geoscience, Vol. XXIV, No. 2, May 2018, pp. 237–250