E&EG Journal Volume XXVIII, Number 2 by Association of Environmental & Engineering Geologists (AEG)

EDITORIAL OFFICE: Environmental & Engineering Geoscience journal, Kent State University, Kent, OH 44242, U.S.A., 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, 3053 Nationwide Parkway, Brunswick, OH 44212. 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, 3053 Nationwide Parkway, Brunswick, OH 44212. 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.

Oommen, Thomas Board Chair, Michigan Technological University Sasowsky, Ira D. University of Akron

ASSOCIATE EDITORS Ackerman, Frances Ramboll Americas Engineering Solutions, Inc. Bastola, Hridaya Lehigh University Beglund, James Montana Bureau of Mines and Geology Bruckno, Brian Virginia Department of Transportation Clague, John Simon Fraser University, Canada Dee, Seth University of Nevada, Reno Fryar, Alan University of Kentucky Gardner, George Massachusetts Department of Environmental Protection

Hauser, Ernest Wright State University Keaton, Jeff AMEC Americas May, David USACE-ERDC-CHL Pope, Isaac Book Review Editor Santi, Paul Colorado School of Mines Schuster, Bob Shlemon, Roy R.J. Shlemon & Associates, Inc. Stock, Greg National Park Service Ulusay, Resat Hacettepe University, Turkey West, Terry Purdue University

Environmental & Engineering Geoscience May 2022

VOLUME XXVIII, NUMBER 2

Submitting a Manuscript

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 $75 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 $310 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, 3053 Nationwide Parkway, Brunswick, OH 44212. © 2022 by the Association of Environmental and Engineering Geologists

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER EDITORS

Eric Peterson Department of Geography, Geology, and the Environment Illinois State University Normal, IL 61790 309-438-5669 ewpeter@ilstu.edu

Karen E. Smith, Editorial Assistant, kesmith6@kent.edu

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 https://www.editorialmanager.com/EEG/ default.aspx. 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. Upon submission, manuscripts must meet, exactly, the criteria specified in the revised Style Guide found at https://www.aegweb.org/e-eg-supplements. Manuscripts of fewer than 10 pages may be published as Technical Notes. The new optional feature of Open Access is available upon request for $750 per article. For further information, you may contact Dr. Abdul Shakoor at the editorial office.

Cover photo Rotational slumps mantling the retreating escarpment of the Vermilion Cliffs in Arizona. Many of these relatively intact slides are also mantled by chaotic rockslide debris-avalanches generated by the collapse and disintegration of the overlying cliffs. Photo courtesy of Conor M. Watkins.

Volume XXVIII, Number 2, May 2022

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.

Abdul Shakoor Kent State University Kent, OH 44242 ashakoor@kent.edu

ADVISORY BOARD Watts, Chester “Skip” F. Radford University Hasan, Syed University of Missouri, Kansas City Nandi, Arpita East Tennessee State University

ENVIRONMENTAL & ENGINEERING GEOSCIENCE

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, 3053 Nationwide Parkway, Brunswick, OH 44212 and additional mailing offices.

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 Volume 28, Number 2, May 2022 Table of Contents 147

Geomorphology, Three-Dimensional Geology, and Seismologic Hazards of the New Madrid Seismic Zone in Dyer County, Tennessee Renee Reichenbacher, Roy Van Arsdale, Randel Cox, Chris Cramer

173

A New Look at Landslides of the Vermilion and Echo Cliffs, Northern Arizona Conor M.Watkins, J. David Rogers

193

Characterization and Analysis of the Cedar Pass Landslide Complex, Badlands National Park Kyle C. Radach, Paul M. Santi

215

Stratigraphic and Geochemical Evidence for the Alteration of Calcareous Glauconitic Marine Sediments to Calcium Bentonite James H. May, Wayne C. Isphording, David Patrick, David R.Williamson, James E. Lyles, Sr. Technical Note

233

The Yaqui Flat Long Run-Out Rock Avalanche: Anza-Borrego Desert State Park, California Michael W. Hart, David B. Evans

Geomorphology, Three-Dimensional Geology, and Seismologic Hazards of the New Madrid Seismic Zone in Dyer County, Tennessee RENEE REICHENBACHER* ROY VAN ARSDALE RANDEL COX Department of Earth Sciences, University of Memphis, 488 Patterson Street, Memphis, TN 38152

CHRIS CRAMER Center for Earthquake Research and Information, University of Memphis, 3890 Central Avenue, Memphis, TN 38152

ABSTRACT Geomorphic and three-dimensional geologic mapping reveals two major fault systems of the seismically active Reelfoot rift pass beneath Dyer County in northwestern Tennessee, the Reelfoot South fault, and the eastbounding faults of the Reelfoot rift. The Dyer County mapping also indicates that the two principal Reelfoot South fault hanging wall structures, the Lake County uplift and Tiptonville dome, pass beneath the county. Quaternary displacement was identified on southeastern Reelfoot rift margin faults in Dyer County, thus indicating that this rift margin has been active during the Quaternary from adjacent Obion County through Dyer County to Lauderdale County, Tennessee, for a distance of at least 60 km. The three-dimensional geologic mapping also provides stratigraphic thicknesses of surface sediment and underlying Paleogene and Cretaceous strata that significantly contribute to the estimation of ground motion in the event of a future large New Madrid seismic zone earthquake. The new ground motion maps using the three-dimensional geology of Dyer County are compared to the current U.S. Geological Survey earthquake hazard maps. This comparison reveals generally lower acceleration for buildings less than four stories high and greater acceleration for buildings greater than 10 stories high in the event of a large New Madrid seismic zone earthquake.

seismic zone (NMSZ) (Figure 1) and thus is particularly vulnerable to both earthquake ground shaking and liquefaction in the event of future large earthquakes (Chiu et al., 1992; Cramer, 2006; Csontos and Van Arsdale, 2008; Cramer and Boyd, 2014; and Cramer et al., 2018a, 2018b, 2020). The topography of Dyer County consists of Holocene lowland ﬂoodplains of the Mississippi, Obion, and Forked Deer rivers; Pliocene and Pleistocene loess–covered terraces of these rivers; and Pleistocene loess overlying Eocene Upland sediments. Beneath the near-surface geology are older Paleogene, late Cretaceous, and early Paleozoic strata (Figure 2) (Hardeman, 1966; Hosman, 1996; and Weathers and Van Arsdale, 2019). Structurally, Dyer County overlies Cambrian age Reelfoot rift faults (Figure 1) (Csontos et al., 2008), which have Quaternary reactivation in adjacent Lake, Obion, and Lauderdale counties (Figure 1) (Kelson et al., 1996; Cox et al., 2001, 2006; Greenwood et al., 2016; Weathers and Van Arsdale, 2019; and Gold et al., 2019). Of particular interest are the seismically active Reelfoot South fault (RSF) and Reelfoot rift eastern margin faults because they appear to pass beneath Dyer County. In this study, we present the geomorphology, surface and subsurface geology, and geologic history and provide important information for ongoing seismologic and liquefaction hazard modeling of Dyer County (e.g., Cramer et al., 2018a, 2018b, 2020; Reichenbacher, 2020; and Reichenbacher et al., 2020).

INTRODUCTION Dyer County in northwestern Tennessee is located within the easternmost portion of the New Madrid

*Corresponding author email: rmrchnbc@memphis.edu

New Madrid Earthquakes of 1811–1812 The great NMSZ earthquakes of 1811–1812 occurred on three major Reelfoot rift faults (Figure 1) (Johnston and Schweig, 1996). On December 16, 1811, an ∼M 7.5 earthquake and its aftershock occurred

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

147

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 1. New Madrid seismic zone epicenters and Reelfoot Rift faults. The three major earthquakes of 1811–1812 are shown as yellow stars. NMNF = New Madrid North fault; RF = Risco fault; RNF = Reelfoot North fault; RSF = Reelfoot South fault; AF = axial fault; NMNF = New Madrid North fault; C = Charleston; CA = Cairo; H = Hickman; PG = Porter’s Gap fault; UC = Union City fault; D = Dyersburg; N = New Madrid. Ticks on downthrown sides of the western and eastern margin faults of the Reelfoot Rift. Inset map locates Memphis (M), Reelfoot Rift bounding faults, epicenters of the New Madrid seismic zone, Mississippi embayment perimeter, and Crowley’s Ridge (stippled) (modiﬁed from Pryne et al., 2013).

on the northeast-striking right-lateral strike-slip axial fault. An ∼M 7.3 earthquake took place on January 1, 1812, along the northeast-striking rightlateral strike-slip New Madrid North fault (NMNF), and on February 7, 1812, an ∼M 7.7 earthquake occurred on the northwest-striking Reelfoot thrust (step-over) fault (Cramer and Boyd, 2014). Eyewitness accounts of the 1811–1812 earthquakes describe catastrophic damage to the landscape (Penick, 1981). These

148

accounts describe geyser-like sand blows reaching treetops, ground ﬁssures that in some places required the downing of trees to traverse, major cabin damage, and landslides along the Mississippi River bluﬀs from Memphis, Tennessee, to near Cairo, Illinois, that accompanied the severe ground shaking (Fuller, 1912; Jibson and Keefer, 1989). Today, an earthquake of this intensity would be catastrophic throughout the region (Cramer, 2006), resulting in $300 billion in

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 2. Stratigraphic column of West Tennessee. Quat. = Quaternary; Plio. = Pliocene; EGRP = Eastern Granite-Rhyolite Province (modiﬁed from Ward, 2016).

direct losses and at least twice that in indirect losses, about 2.6 million households without power, nearly 715,000 buildings and 3,500 bridges damaged, and 86,000 expected injuries and fatalities (Elnashai et al., 2009). Regional Stratigraphy and Structure Dyer County, Tennessee, lies within the central Mississippi River valley, its dominant geologic events be-

ing the formation of the Cambrian Reelfoot Rift and the Cretaceous Mississippi Embayment (Figure 1). The Reelfoot Rift is a northeast-striking, 300-kmlong, 70-km-wide aulacogen that formed during the breakup of Rodinia (Hildenbrand, 1985; Hildenbrand and Hendricks, 1995; Dart and Swolfs, 1998; Csontos et al., 2008; and Van Arsdale, 2014). The failed rift produced marginal high-angle normal faults, which allowed for marine transgression and deposition of the early to middle Cambrian Potsdam Megagroup

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

149

Reichenbacher, Van Arsdale, Cox, and Cramer

(Thomas, 1989, 1991; Parish and Van Arsdale, 2004) (Figure 2). The Potsdama Megagroup lies unconformably above the Precambrian crystalline basement rock of the 1.47 Ga Granite-Rhyolite Province and consists of Cambrian Lamotte arkosic sandstone, Bonneterre dolomite, and Elvins shale (Howe, 1985; McKeown et al., 1990). The Potsdam Megagroup was followed by deposition of the Cambrian and Ordovician Knox Megagroup (Schwalb, 1982; Lumsden and Caudle, 2001). The lithology of the Knox Megagroup consists of shallow-marine carbonates and nearshore clastic sediments. Subsequent deposition of sediments occurred from the middle Ordovician through the Pennsylvanian. However, most of that section was removed by regional uplift and erosion during the Cretaceous passing of the North American Plate over the Bermuda hotspot, resulting in an unconformity at the top of the Paleozoic section (Cox and Van Arsdale, 1997, 2002). The Middle Cretaceous hotspot uplift also caused reactivation of the Reelfoot Rift basement faults and the intrusion of plutons (Cox and Van Arsdale, 2002). When the central United States drifted westward off the Bermuda hotspot during the late Cretaceous, cooling and subsidence occurred, thus forming the southwest-plunging Mississippi Embayment trough (Figure 1 inset). Formation of the Mississippi Embayment trough caused rerouting of the ancestral Mississippi River from a northwest to a south flow direction and allowed inundation by the Gulf of Mexico (Cox and Van Arsdale, 2002). Regional cooling subsidence and marine transgressions resulted in deposition of Late Cretaceous and Paleogene sediments. The Late Cretaceous units include the Coffee, Demopolis, McNairy Sand, and Owl Creek formations (Figure 2), which consist of shallow-marine to fluvial-deltaic deposits made up of marls, chalk, clays, sands, and lignite (Cushing et al., 1964; Crone, 1981; Van Arsdale and Tenbrink, 2000; Parrish and Van Arsdale, 2004; Van Arsdale, 2009; and Martin and Van Arsdale, 2017). The overlying Paleogene units are differentiated into marine and fluvial-lacustrine deposits of the Paleocene Midway Group, Eocene Wilcox Group, Eocene Claiborne Group, and Eocene-Oligocene Jackson Formation. The Jackson Formation is locally exposed along the Mississippi River bluffs and underlies the fluvial Pliocene Upland Complex in western Tennessee (Hardeman, 1966; Van Arsdale et al., 2007). The Upland Complex disconformably overlies Eocene formations and is disconformably overlain by Pleistocene loess (Markewich et al., 1998; Lumsden et al., 2016). The Upland Complex is an ∼3.6-Ma (Odum et al., 2020) high-level fluvial deposit composed of ferruginous sand and gravel of the Pliocene Mississippi

150

River (Potter, 1955a, 1955b; Autin et al., 1991; Saucier, 1994; Van Arsdale et al., 2007; Van Arsdale, 2009; Van Arsdale and Cupples, 2013; Cox et al., 2014; Cupples and Van Arsdale, 2014; and Lumsden et al., 2016). The Mississippi River and its tributary river ﬂoodplains are made up of Pleistocene and Holocene braided and meandering river sediments (Cushing et al., 1964; Crone, 1981; Saucier, 1987, 1994; Guccione et al., 2002; Rittenour et al., 2007; Csontos and Van Arsdale, 2008; and Csontos et al., 2008). Reelfoot Rift The Reelfoot rift faults (Figure 1) originated as Cambrian normal faults, but the northeast-striking faults within the rift have undergone Neogene rightlateral strike-slip oﬀset (Cox et al., 2001, 2006; Pratt, 2012; Pratt et al., 2012; and Van Arsdale and Cupples, 2013). Continued right-lateral movement is revealed in fault plane solutions (Chiu et al., 1992; Mueller and Pujol, 2001; and Martin and Van Arsdale, 2017). There are also seismically active northwest-striking reverse faults, like the Reelfoot fault, within step-over zones (Csontos and Van Arsdale, 2008; Van Arsdale and Cupples, 2013; and Cramer and Boyd, 2014). Additionally, east-to-west trending Neogene grabens indicate north- to-south extension during right-lateral reactivation of the northeast-striking Reelfoot rift basement faults (Van Arsdale and Cupples, 2013; Martin and Van Arsdale, 2017). The NMSZ earthquakes occur along rift faults at depths between 4 and 14 km, primarily within the Precambrian section (Chiu et al., 1992; Mueller and Pujol, 2001). The most seismogenic faults of the NMSZ are the vertical right-lateral axial fault, which trends down the center of the Reelfoot rift, and the southwestdipping Reelfoot reverse fault, which extends from 15 km northwest of New Madrid, Missouri, to Dyersburg, Tennessee (Figure 1) (Johnston and Schweig, 1996; Csontos and Van Arsdale 2008; Csontos et al., 2008; Magnani and McIntosh, 2009; Van Arsdale et al., 2013; Guo et al., 2014; and Magnani et al., 2017). The axial fault bisects the northwest-striking Reelfoot fault into two segments: the Reelfoot North fault (RNF) and the Reelfoot South fault (RSF). From the shallow subsurface to ∼4-km depth, both the RNF and the RSF have a steep southwest dip of ∼80°. However, below ∼4 km, the RNF dips ∼30° southwest, and the RSF dips ∼44° southwest (Figures 3 and 4) (Csontos and Van Arsdale, 2008). Studies in adjacent Lake, Obion, and Lauderdale counties suggest Reelfoot rift faults extend into Dyer County (Figures 1 and 4) (Van Arsdale et al., 1999; Csontos and Van Arsdale, 2008; Greenwood et al., 2016; and Weathers and Van Arsdale, 2019). These

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 3. (A) Reelfoot North fault and its hanging wall Lake County uplift (outlined in thick black line) and Tiptonville dome culmination on the east side of the Lake County uplift. The hanging wall faults bounding the Lake County uplift and Tiptonville dome are the Lake County uplift backthrust, Tiptonville dome backthrust, and Reelfoot North fault. RS = Reelfoot scarp; RL = Reelfoot Lake; KY = Kentucky; MO = Missouri. Dyer County bounds Lake County on the south. (B) Cross section A–A illustrating Lake County uplift (LCU) and Tiptonville dome (TD). Earthquake hypocenters deﬁne the Reelfoot North fault. K = top of Cretaceous; Pz = top of Paleozoic; Pc = top of Precambrian (from Purser and Van Arsdale, 1998).

structures include the southeast-striking RSF and its hanging wall structures of the Lake County uplift (LCU) and Tiptonville dome (TD). The faults bounding these structural uplifts are the Lake County uplift backthrust (LCUB), Tiptonville dome backthrust

(TDB), and RSF (Figure 4). At the Mississippi River bluﬀs in southwestern Obion County, the RSF displaces the top of the Paleozoic 65 m, Cretaceous 40 m, Midway Group 31 m, Wilcox Group 20 m, and Memphis Sand 16 m (Greenwood et al., 2016).

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

151

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 4. (A) Red dots are earthquake epicenters in Dyer and adjacent Lake (northwest) and Obion (northeast) counties. RSF = Reelfoot South fault; TDB = Tiptonville dome backthrust; LCUB = Lake County uplift backthrust; L = lane. UTM coordinates. (B) Reelfoot South fault cross section showing 1,200 m of displacement on top of Precambrian (from Csontos and Van Arsdale, 2008).

152

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 5. One-meter-resolution LiDAR digital elevation model of central Dyer County drainage with faults as red lines and drainage divides. Measured drainage basins are yellow with blue lines demarcating streams draining the three drainage basins: DB1, DB2, and DB3. LCUB = Lake County uplift backthrust; TDB = Tiptonville dome backthrust; RSF = Reelfoot South fault; L = lane.

Displacement of Quaternary stream terraces and Pliocene Upland Complex has been documented at this same location (Greenwood et al., 2016). Farther south, the RSF is interpreted to cross into Dyer County, where historical records document a temporary lake formed in 1812 on the Obion River at Lane, Tennessee (Figures 4A and 5) (Van Arsdale et al. 1999). Strong evidence for a southern continuation of the RSF into Dyer County includes the distribution of earthquake epicenters (Figure 4A) and the interpretation that the hypocenters of the faults at >4-km depth deﬁne the RSF geometry shown in Figure 4B (Csontos and Van Arsdale, 2008). The southeastern Reelfoot rift margin consists of two down-to-the west Cambrian basement faults and is projected from Lauderdale County through Dyer County to Obion County (Cox et al., 2001; Parish and Van Arsdale, 2004) (Figures 1 and 4A). While NMSZ earthquakes are concentrated along the Axial and Reelfoot faults of the NMSZ (Figures 1 and 4A), seismicity and Quaternary faulting along the southeastern Reelfoot rift margin support the possibility of

strain buildup and the potential to produce moderate to large earthquakes (Cox et al., 2001, 2006). Earthquake Liquefaction Earthquake liquefaction studies within the NMSZ reveal extensive 1811–1812 and pre-1811–1812 liquefaction (Saucier, 1987, 1994; Obermeier, 1988, 1989; Rodbell and Bradley, 1993; Rodbell, 1996; Tuttle et al., 1999, 2002, 2005; and Wolf et al., 2006). A large portion of Dyer County consists of Holocene lowland ﬂoodplain alluvium of the Mississippi, Obion, and Fork Deer rivers (Saucier, 1994; Rittenour et al., 2007). Four northeast-oriented sand blow trends, 10 km southwest of Dyersburg, reveal three liquefaction deposits with estimated ages of 350–650 years BP and one that formed in 1811–1812 (Tuttle et al., 1999, 2002, 2005). A survey of the Obion River ﬂoodplain ∼10 km north of Dyersburg revealed sand blows that may have formed after 1811–1812 (Tuttle et al., 1999, 2002, 2005). Therefore, Dyer County is of major concern for liquefaction in the event of a large fu-

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

153

Reichenbacher, Van Arsdale, Cox, and Cramer

ture earthquake (Reichenbacher, 2020; Reichenbacher et al., 2020). METHODS Principal objectives of this research were to map the surface and subsurface geology of Dyer County and determine if and where the LCUB, TDB, RSF, and the southeastern Reelfoot Rift margin faults pass through Dyer County. This was undertaken through geomorphic analyses, interpretation of geologic boring logs, and previously interpreted seismic reflection profiles within Dyer County and portions of adjacent counties. Drainage Asymmetry Factor Analysis Central Dyer County, east of the Mississippi River floodplain, was geomorphically studied to ascertain whether there is evidence for Quaternary activity on the LCUB, TDB, and RSF. Streams in uplands of central Dyer County were extracted from a 1-m-resolution LiDAR digital elevation model (DEM) using the ArcMap 10.7 Spatial Analyst Hydrology tool set. The drainage basins were delineated from the selected stream networks between drainage divides in the uplands of central Dyer County for analysis (Figure 5). The drainage basins were manually digitized into ESRI shapefile polygons, and each polygon was bisected by the trunk stream. The drainage basin asymmetry factor (AF) of three basins was calculated as follows: AF = 100 (Ar /At ) , where Ar is drainage area on the right side of the main drainage line looking downstream and At is the total drainage area (Hare and Gardner, 1985; Baioni, 2007). The premise of this technique is that if a stream basin has tilted about an axis parallel to the major stream, the stream will lie on the down-tilted side of the basin axis and thus reveal a drainage basin asymmetry (Hare and Gardner, 1985). Where the AF is greater than 50, the main channel has shifted toward the left side of the drainage basin looking downstream. If AF is less than 50, the channel has shifted toward the right side of the drainage basin looking downstream (Hare and Gardner, 1985; Baioni, 2007). Stream Sinuosity Analyses Five river reaches in Dyer County were analyzed with respect to stream sinuosity in ArcMap 10.7 (Figure 6) to ascertain whether there is any indication of tectonic influence on the river reaches. A map of the Obion River, prior to channelization by the U.S. Army Corps of Engineers in the 1960s, was georefer-

154

enced and digitized, from which the Obion River reach was selected. Four reaches of the Forked Deer River were also analyzed in this study. Two reaches of the modern Forked Deer River that have not been channelized and pre-channelization former channels of the Forked Deer River were digitized in the other two reaches using a 1-m-resolution LiDAR DEM. The ArcMap 10.7 Proximity tool set was used to place numbered points along each river segment at 1-km spacing. The straight-line distance between each 1-km point was measured, and sinuosity was calculated using the following formula: Sinuosity = length of stream channel (1 km)/ length of straight line distance (km).

Geologic Mapping Subsurface lithologic boring logs were acquired from the North American Coal Company, the U.S. Army Corps of Engineers, the Tennessee Department of Transportation, petroleum exploration wells, the Tennessee Department of Environmental Control, and the Center of Applied Earth Science and Engineering Research at the University of Memphis. The lithologic boring logs were geologically interpreted and recorded in Excel spreadsheets. A LiDAR based 1-m DEM of Dyer County was the base map on which surface and near-surface geology was mapped from 1,115 bore holes (Figure 7). Geologic mapping was conducted from the ground surface down to the top of the Paleozoic strata, which is ∼700 m below Dyer County. Some of the river terraces along the Obion and Forked Deer rivers were derived from previously published maps (Saucier, 1994; Rodbell, 1996) with additional terrace locations mapped in this study. The top of the Eocene was mapped using 968 borings. Contact between Eocene sediment and the overlying alluvium is identified by the basal conglomeratic facies of the alluvium. The contact between surface Pleistocene loess and Eocene sediment was picked on the basis of silt overlying fine sand or clay. Where Pleistocene loess overlies Pleistocene and Pliocene river terraces, the loess could not be differentiated from the underlying overbank silt/clay of the terrace alluvium, and thus the loess and overbank alluvium were combined into one silt/clay unit. The top of the sand and gravel alluvium of the terraces was picked at the base of the silt/clay overbank alluvium. The underlying tops of the Cretaceous and Paleozoic were mapped from Dow Chemical seismic reflection lines and oil exploration wells (Parrish and Van Arsdale, 2004; Csontos et al., 2008).

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 6. One-meter-resolution LiDAR digital elevation model with the modern rivers in blue, river reaches used for sinuosity analyses in yellow, 1-km point locations in black, and faults in red lines. Northern box enlarged in Figure 8A and southern box enlarged in Figure 8B. LCUB = Lake County uplift backthrust; TDB = Tiptonville Dome backthrust; RSF = Reelfoot South fault.

Structure contour maps were made of the tops of the Eocene, Cretaceous, and Paleozoic sections beneath Dyer County. All three surfaces were mapped using the Natural Neighbor contouring algorithm. Bedrock faults have been mapped beneath Dyer County (Parrish and Van Arsdale, 2004; Csontos et al., 2008; and Martin and Van Arsdale, 2017) and immediately adjacent Lake and Obion counties (Figures 3 and 4) (Weathers and Van Arsdale, 2019). The previously mapped faults in Lake and Obion counties in Figure 3 were located along the northern margin of Dyer County, and the faults were continued south through Dyer County (Figure 4A) using geomor-

phic data and by interpreting the irregular top of the Eocene as a faulted surface. The bounding faults (LCUB, TDB, RSF) of the LCU and TD were picked where major north-to-south drainage divides pass through central Dyer County (Figures 4A and 5–7) and where changes in slope occurred on the top of the Eocene. Similarly, two previously mapped faults (Parrish and Van Arsdale, 2004; Csontos et al., 2008) in southeastern Dyer County appear to continue northeast based on a structural low on the top of the Eocene. These faults were then used to make faulted structure contour maps of the tops of the Eocene, Cretaceous, and Paleozoic using the Spline with Barriers contour-

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

155

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 7. Geologic borings (1,115) used to map the near-surface geology of Dyer County in Figure 10. Black arrows demarcate drainage divides that correspond to underlying faults in Figure 4A.

ing algorithm. The Spline with Barriers contouring algorithm permits breaking a surface into blocks by inserting faults and contouring within the fault-bounded blocks (Briggs, 1974). A west-to-east cross section was made across Dyer County to the top of the Paleozoic from the faulted structure contour maps using the profile tool in Surfer 18.1.186 Golden Software, which allows the user to manually adjust parameters, such as the location, distance, and vertical exaggeration, while re-creating the cross section in real time. Therefore, this tool was the preferred method for generating cross sections in this research. In addition, shallow cross sections were made across the Forked Deer River terraces where drill data are dense to determine if the top of the Eocene strata and overlying terrace alluvium were faulted. Isopach maps were made of the Lowland river floodplain alluvium and the silt/clay, surface overbank portion, of the floodplain alluvium. These isopach maps will provide information for ongoing liquefaction susceptibility mapping in Dyer County.

156

RESULTS Drainage Basin Asymmetry Drainage basin asymmetry has been used to determine where tectonics have controlled river migration (Cox, 1994; Burbank and Anderson, 2012). For stream networks that formed and continue to ﬂow in stable settings, the mean value of AF should be approximately 50 (Keller & Pinter, 2002; Baioni, 2007). In this study, Drainage Basin 1 (DB1) has an AF of 33, suggesting an easterly tilt; Drainage Basin 2 (DB2) has an AF of 74, suggesting a westerly tilt; and Drainage Basin 3 (DB3) has an AF of 28, suggesting an easterly tilt (Figure 5 and Table 1).

Stream Sinuosity Streams react to tectonic inﬂuences by changing their longitudinal proﬁle, channel pattern, and/or sediment discharge. Studies have documented the

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee Table 1. Drainage basin asymmetry.

Drainage Basin DB1 DB2 DB3

Ar (km2 )

At (km2 )

Asymmetry Factor

Tilt Direction

4.4 21.9 4.6

13.4 29.6 16.4

33 74 28

East West East

Ar = drainage basin area on the downstream right of the main drainage line looking downstream; At = total drainage basin area; DB1 = drainage basin 1; DB2 = drainage basin 2; DB3 = drainage basin 3.

influence of vertical crustal movements on channel pattern (Schumm, 1986; Burbank and Anderson, 2012; and Sahu and Roy, 2015). Thus, the sinuosity was analyzed in this research for possible fault/fold influence. The sinuosity values on the Obion and Forked Deer river reaches analyzed in this research (Figure 6) provide evidence of Quaternary displacement on the LCUB, TDB, and RSF and the northeast-trending Reelfoot Rift margin faults in Dyer County. Figure 8A shows the Obion river reach where sinuosities were calculated. High sinuosity values occur at river points 62, 67, and 72 (Figure 9A). Figure 8B shows the reaches of the Forked Deer River where sinuosities were calculated. Figure 9B illustrates the sinuosity of the reach containing points 1–10 with an anomalous low at point 5. The sinuosity of the Forked Deer River reaches containing points 11–37 reveal anomalously high sinuosity values at points 27–29 (Figures 8B and 9C). Sinuosity of the Forked Deer River reach containing points 38–49 (Figure 8B) reveal high sinuosity values at points 41–42 (Figure 9D). Finally, the sinuosity of a Forked Deer River tributary reach that contains points 50–52 (Figure 8B) has high sinuosity values at points 51–52 (Figure 9E). Near-Surface Geologic Map The near-surface geology of Dyer County was geologically differentiated into three mapping units (Lowlands, Intermediate, and Uplands) based on their elevation and near-surface geology (Figure 10) (Reichenbacher, 2020; Reichenbacher et al., 2020). The cross section in Figure 10B does not represent a particular place on Figure 10A but is a general depiction of the near-surface geology. Unit surface elevations and unit thicknesses in Figure 10B are averages from measured values throughout the county. The Lowlands are at an elevation of <82 m above mean sea level and consist primarily of Holocene (<12 ka) river floodplain alluvium (Saucier, 1994; Rittenour et al., 2007). In general, the alluvium consists of surface silt and clay overbank sediment that over-

lies sand and gravel sediment. Between elevations of 82 and 107 m are four Intermediate units. Intermediate units are diﬀerentiated into Pleistocene river terraces of the Obion and Forked Deer rivers and areas where loess overlies Eocene strata. The terraces have been mapped from topographically highest to lowest as the Humboldt, Hatchie, and Finley terraces (Saucier, 1987; Rodbell, 1996). All Intermediate units are covered with Peoria loess that is <20 ka. The Hatchie and Humboldt terraces have 65–26-ka Roxana and possibly older loess units beneath the Peoria loess (Rodbell, 1996; Markewich et al., 1998); however, these older loess deposits have not been mapped throughout Dyer County. Beneath the loess, the terrace alluvium consists of silt and clay overbank sediment and underlying sand and gravel that is ∼22 ka (Finley) and >65 ka (Hatchie and Humboldt). The Upland unit is at an elevation >107 m and is a loess-covered highlevel terrace of the Pliocene (3.6 Ma) Mississippi River (Van Arsdale et al., 2007; Odum et al., 2020). This high-level terrace alluvium, called the Upland Complex, consists of sand and gravel that is regionally a major source of aggregate (Van Arsdale et al., 2012). The area mapped as uplifted ﬂoodplain in Figure 10A is 1–2 m higher than the adjacent Lowlands alluvium. It may be a low-level Pleistocene terrace of the Mississippi River, but we interpret the 1–2-mhigher alluvium as having been tectonically uplifted by Holocene displacement on the LCU discussed below (Figure 4A). Structure Contour Maps There is a lot of subsurface relief on the top of the Eocene (Figure 11) because the Eocene has been eroded by the Mississippi, Obion, and Forked Deer rivers and, we infer, because it has also undergone structural deformation. A north-trending high on the top of the Eocene passes beneath Dyersburg, and a northeast-trending low on the top of the Eocene is east of Dyersburg (Figure 11A). Modeling using faults bounding the north-trending high in Figure 11A best interprets the high as a southern continuation of the RSF and its hanging wall TD and LCU (Figure 11B). Fault locations are also supported by the north-to-south drainage divides (Figures 5 –7). Similarly, on insertion of Reelfoot rift eastern margin faults bounding the northeast-trending low, we interpret the low to be a graben. The graben is on strike with well-constrained northeast-striking faults in southeastern Dyer County (Figure 4A) (Parrish and Van Arsdale, 2004; Martin and Van Arsdale, 2017). The less well constrained tops of the Cretaceous strata and Paleozoic strata were contoured and reveal west-sloping surfaces (Figures 12A and 13A). When

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

157

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 8. One-meter-resolution LiDAR digital elevation models with the modern Obion and Forked Deer rivers in blue and the reaches analyzed in yellow. Areas A and B located in Figure 6. The numbered boxes indicate the 1-km spaced points where the sinuosity values were calculated in Figure 9. L = lane; TDB = Tipton Dome backthrust; RSF = Reelfoot South fault.

inserting the faults, identiﬁed as displacing the top of the Eocene strata in Figure 11B, the tops of Cretaceous and Paleozoic were contoured with faults (Figures 12B and 13B).

158

Geologic Cross Sections A west-to-east geologic cross section across Dyer County illustrates the general subsurface stratigraphy

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 9. The numbers on the X axis indicate the 1-km points where river sinuosity values were calculated along the reaches located in Figure 8. TDB = Tipton dome backthrust; RSF = Reelfoot South fault; LCUB = Lake County Uplift backthrust in C crossed at two locations. Faults are located at or very near changes in stream sinuosity.

and structure (Figures 14A and B). Cross section C– C reveals the bounding faults of the Lake County uplift (LCUB and TDB) and Tiptonville dome (TDB and RSF) as well as the graben within the southeastern Reelfoot rift margin. However, the west half of the LCU mapped on the top of the Eocene is a graben on the top of the Cretaceous and Paleozoic, thus suggesting local Neogene tectonic inversion.

The southeastern corner of Dyer County has hundreds of drill holes that penetrate the top of the Eocene section (Figure 7) and northeast-striking faults that have been mapped from Dow Chemical seismic lines (Figure 14A, inset) (Parrish and Van Arsdale, 2004; Martin and Van Arsdale, 2017). These northeaststriking faults underlie northwest-trending Pleistocene Finley, Hatchie, and Humboldt terraces of the Forked

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

159

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 10. (A) Surface geology of Dyer County. (B) Generalized cross section illustrating the units mapped in Dyer County.

160

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 11. (A) Structure contour map of top of the Eocene in Dyer County. Dots are boring data used in mapping. (B) Structure contour map of the top of the Eocene in Dyer County made by inserting faults into the model.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

161

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 12. (A) Structure contour map of top of Cretaceous strata in Dyer County. Red dotted lines are Dow Chemical seismic reﬂection lines, and blue squares are petroleum exploration wells. (B) Structure contour map of top of Cretaceous strata in Dyer County made by inserting faults into the model.

162

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 13. (A) Structure contour map of top of Paleozoic strata in Dyer County. Purple dotted lines are Dow Chemical seismic reﬂection lines, and green squares are petroleum exploration wells. (B) Structure contour map of top of Paleozoic strata in Dyer County made by inserting faults into the model.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

163

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 14. (A) Dyer County faults and cross sections C–C , D–D , and E–E locations. Small faults are shown as short black lines on seismic line 5EB. (B) Cross sections C–C showing shallow (−100 to +100 m) and deep (−900 to +100 m) sections. (C) Shallow cross sections (+50 to +100 m) D–D and E–E .

Deer River, thus allowing determination of Quaternary faulting (Figure 14A). Cross section D–D passes through the Humboldt (>65 ka) terrace (Figure 14C). A 13-m up-to-the-northwest step on the top of the Eocene and the top of the overlying Humboldt ter-

164

race sand and gravel horizon is evident at cross-section location ∼1,300 m. At D–D , the Pleistocene Forked Deer River ﬂowed from southeast to northwest; thus, the 13-m step ascends in the downstream direction, contrary to a normal river proﬁle. Cross section E–E

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

Figure 14. (continued)

reveals an apparent uplift on the top of the Eocene and an ∼3-m down-to-the-northwest step on the top of the ∼20-ka Finley terrace sand and gravel horizon (Figure 14C) (Reichenbacher, 2020; Reichenbacher et al., 2020). Holocene Floodplain Isopach Maps The Holocene floodplain alluvium, in general, consists of two layers: a surface silt and clay layer that averages 8 m thick and an underlying sand and gravel layer that averages 30 m thick. The isopach map of the entire Quaternary floodplain alluvium reveals variation in thickness due to rivers of different sizes and scour depths (Figure 15A). Thickness of the uppermost silt and clay layer is also variable (Figure 15B). CONCLUSIONS Pliocene–Holocene Geologic History of Dyer County The near-surface geology of Dyer County reveals the incision of the ancestral Mississippi/Ohio river

system through time (Figure 10). Approximately 3.6 Ma (Odum et al., 2020), the ancestral Mississippi River flowed across a vast floodplain depositing the Upland Complex that extended beyond Dyer County (Van Arsdale et al., 2007; Cox et al., 2014; and Cupples and Van Arsdale, 2014). Incision through the Upland Complex throughout the lower Mississippi/Ohio river system began in the early Pleistocene with growth of continental ice sheets and resultant lower sea levels (Saucier, 1994; Van Arsdale et al., 2007). During the Pleistocene, up to four loess layers were deposited in western Tennessee (Markewich et al., 1998). In Dyer County, Rodbell (1996) identified Peoria loess (<20 ka) covering the Upland and Intermediate surfaces with underlying Roxanna silt (loess) that is 65–26 ka on top of the alluvium of the Humboldt and Hatchie terraces. Based on these observations, we know the ancestral Mississippi/Ohio river system and its tributaries entrenched >65 ka because the Humboldt terrace alluvium is overlain by the Roxana and perhaps older loess (Figure 10B). Subsequently, the Mississippi/Ohio river system and its tributaries further entrenched to form the Hatchie (also >65 ka)

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

165

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 15. (A) Floodplain alluvium thickness for the Obion, Mississippi, Forked Deer, and South Forked Deer rivers in Dyer County. (B) Floodplain overbank silt/clay thickness for the Obion, Mississippi, Forked Deer, and South Forked Deer rivers in Dyer County.

166

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

and then Finley (∼22 ka) terrace levels. The final entrenchment of the Mississippi River and its Obion and Forked Deer river tributaries in Dyer County occurred within the last 22 ka, resulting in most of the Lowlands being underlain by floodplain alluvium that is <12 ka. Geomorphic Analyses of Dyer County The AFs of the three drainage basins in central Dyer County appear to indicate tilting on the LCU and TD (Table 1 and Figure 5). This drainage basin asymmetry could be due to differences in the underlying geology; however, we think this unlikely because the underlying lithology is Eocene sediments in all three fault bounded basins. If indeed the three drainage basins illustrated in Figure 5 are controlled by underlying bedrock tilting, then it appears that the tilted blocks are bounded by the LCBT, TDB, and RSF. However, a drainage divide between DB1 and DB2 suggests that a fourth fault may underlie and create this divide. Stream reach sinuosity along the Obion River supports the interpretation that the TDB and RSF pass beneath the Obion River near locations 59 and 68 in Figure 8A as indicated in Figure 9A. Furthermore, the locations of the Obion River sinuosity anomalies are immediately north of the drainage divides interpreted to be faults in Figures 5 –7. Sinuosity measurements of the Forked Deer River reaches in Figure 9B–D suggest Quaternary displacement on the southeast Reelfoot rift margin faults. However, the variability of sinuosity values of the Forked Deer River reach in Figure 9E is likely due to deposition of fluvial sediments from upstream tributaries. Geologic Mapping of Dyer County Geologic mapping of Dyer County has provided insight into the county’s Quaternary history, threedimensional geology, continuation of the seismically active RSF across most of the county, and Quaternary displacement on eastern margin faults of the Reelfoot rift. The maps and three-dimensional model also provide stratigraphic thicknesses and sediment types from the ground surface to the top of the Paleozoic bedrock strata. This geologic information, in combination with geophysical and geotechnical data, is important for the determination of earthquake and liquefaction hazards for Dyer County in the event of a future large earthquake (Cramer et al., 2020; Reichenbacher, 2020; Reichenbacher et al., 2020). Faults underlying Dyer County include the LCUB, TDB, RSF, and eastern margin faults of the Reelfoot

rift. These faults are revealed in north-to-south– trending drainage divides that we interpret to be surface manifestations of these faults, relief on the top of the Eocene, and earthquake epicenter distribution (Figures 4A, 5, and 7). We interpret the northeasttrending depression on the top of the Eocene in eastern Dyer County (Figure 11) as a graben because the bounding faults overlie Reelfoot rift margin faults and the depression does not coincide with a river valley and thus is not erosional. The RSF in adjacent Lake and Obion counties (Kelson et al., 1996; Greenwood et al., 2016; and Gold et al., 2019) and a fault within the eastern margin of the Reelfoot rift in adjacent Lauderdale (Porter’s Gap fault) and Obion (Union City fault) counties (Cox et al., 2001, 2006) have Quaternary displacement (Figure 1). Dyer County faults in cross sections D–D and E–E (Figure 14C) appear to displace Pleistocene river terrace alluvium and thus have also been active in the Quaternary. The Quaternary fault scarp in Obion County (Saucier, 1987; Cox et al., 2006), Quaternary rift margin faulting in Dyer County identiﬁed in this study and the Quaternary faulting at Porter’s Gap (Cox et al., 2001, 2006) all appear to lie within the southeastern Reelfoot rift margin fault zone (Figure 1), thus supporting the proposition by Cox et al. (2006) that the southeastern Reelfoot rift margin has been active over a distance of at least 60 km during the Quaternary and thus poses a seismic threat to the region. We conclude by illustrating the importance of using detailed local geology in seismic hazard analyses. Cramer et al. (2020) and Reichenbacher et al. (2020) compare the ground accelerations for speciﬁc earthquake scenarios by overlaying their Dyer County hazard map, determined using the geologic data provided in this study, with U.S. Geological Survey (USGS) earthquake hazard maps (Figure 16). Figure 16A illustrates that the expected ground accelerations for ground motion of 1.0-second frequency is higher in Dyer County than currently estimated by the USGS. However, Figure 16B illustrates that the expected ground accelerations for 0.2-second frequency is locally higher but over most of Dyer County lower than estimated by the USGS. These local geology-based ground acceleration estimates indicate generally lower acceleration for buildings less than four stories high and greater acceleration for buildings greater than 10 stories high in the event of a large earthquake than currently indicated in the USGS earthquake hazard maps, thus demonstrating the importance of integrating local geology into earthquake hazard analyses. A comprehensive presentation of Dyer County seismic hazard determinations and scenarios is beyond the scope of this paper and is detailed in Cramer et al. (2020).

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

167

Reichenbacher, Van Arsdale, Cox, and Cramer

Figure 16. Overlay of the estimated earthquake ground accelerations determined using Dyer County local geology with faults on the regional ground accelerations currently in use by the U.S. Geological Survey. (A) Dyer County 1.0-second frequency and (B) Dyer County 0.2-second frequency (from Cramer et al., 2020).

168

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee

ACKNOWLEDGMENTS This research was funded by a Housing and Urban Development research grant. This project combined the resources of the University of Memphis Department of Earth Sciences, the Department of Civil Engineering, and the Center for Earthquake Research and Information with other outside entities working in tandem as part of a five-year multidisciplinary project, the University of Memphis Hazard Mapping, Assessment and Education (HazMAE) project, which will provide multi-hazard data to communities for hazard mitigation and resilience efforts. We wish to thank those who provided the data necessary for this project, including David Arellano, Hamed Tohidi, Steve Horton, Luke Ewing, and William Mann of the Tennessee Department of Environmental Control and the U.S. Army Corps of Engineers. REFERENCES Autin, W. J.; Burns, S. F.; Miller, B. J.; Saucier, R. T.; and Snead, J. I., 1991, Quaternary geology of the Lower Mississippi Valley, in Morrison, R. B. (Editor), Quaternary Nonglacial Geology: Conterminous U.S.: Geological Society of America, Boulder, CO, pp. 547–582. Baioni, D., 2007, Drainage basin asymmetry and erosion processes relationship through a new representation of two geomorphic indices in the Conca river (northern Apennines): BollettinoSocieta Geologica Italiana, Vol. 126, pp. 573–579. Briggs, I.C., 1974, Machine contouring using minimum curvature: Geophysics, Vol. 39, pp. 39–48. Burbank, D. W. and Anderson, R. S., 2012, Tectonic Geomorphology, 2nd ed.: Wiley-Blackwell, New York. Chiu, J. M.; Johnston, A. C.; and Yang, Y. T., 1992, Imaging the active faults of the central New Madrid seismic zone using PANDA array data: Seismological Research Letters, Vol. 63, pp. 375–393. Cox, R. T., 1994, Analysis of drainage-basin symmetry as a rapid technique to identify areas of possible Quaternary tilt-block tectonics: An example from Mississippi Embayment: Geological Society America Bulletin, Vol. 106, pp. 571–581. Cox, R. T.; Cherryhomes, J.; Harris, J. B.; Larsen, D.; Van Arsdale, R. B.; and Forman, S. L., 2006, Paleoseismology of the southeastern Reelfoot rift in western Tennessee and implications for intraplate fault zone evolution: Tectonics, Vol. 25, pp. 1–17. Cox, R. T.; Lumsden, D. N.; and Van Arsdale, R. B., 2014, Possible relict meanders of the Pliocene Mississippi River and their implications: Journal Geology, Vol. 122, pp. 609–622. Cox, R. T. and Van Arsdale, R. B., 1997, Hotspot origin of the Mississippi embayment and its possible impact on contemporary seismicity: Engineering Geology, Vol. 46, pp. 5–12. Cox, R. T. and Van Arsdale, R. B., 2002, The Mississippi Embayment, North America: A first order continental structure generated by the Cretaceous superplume mantle event: Journal Geodynamics, Vol. 34, pp. 163–176. Cox, R. T.; Van Arsdale, R. B.; Harris, J. B.; and Larsen, D., 2001, Neotectonics of the southeastern Reelfoot Rift zone margin, central United States, and implications for regional strain accommodation: Geology, Vol. 29, pp. 419–422.

Cramer, C. H., 2006, Quantifying the uncertainty in site amplification modeling and its effects on site-specific seismic-hazard estimation in the Mississippi embayment and adjacent area: Bulletin Seismological Society America, Vol. 96, pp. 2008–2020. Cramer, C. H. and Boyd, O. S., 2014, Why the New Madrid earthquakes are M7-8 and the Charleston earthquake is ∼M 7: Bulletin Seismological Society America, Vol. 104, pp. 2884–2903. doi:10.1785/0120120257. Cramer, C. H.; Dhar, M. S.; and Arellano, D., 2018a, Update of the urban seismic and liquefaction hazard maps for Memphis and Shelby County, Tennessee: Liquefaction probability curves and 2015 hazard maps: Seismological Research Letters, Vol. 89, pp. 688–701. Cramer, C.; Van Arsdale, R.; Arellano, D.; Pezeshk, S.; Horton, S.; Weathers, T.; Nazemi, N.; Imenez, J. A.; Tohidi, H.; and Ogweno, L. P., 2018b, Seismic and liquefaction hazard maps for Lake County, northwestern Tennessee (2018 SE-GSA abstract): Geological Society America Abstracts Programs, Vol. 50. doi:10.1130/abs/2018SE-312570. Cramer, C.; Van Arsdale, R. B.; Reichenbacher, R.; Arellano, D.; Tahidi, H.; Pezeshk, S.; Horton, S. P.; Bhattarai, R.; Nazemi, N.; and Farhadi, A., 2020. Dyer County Seismic and Liquefaction Hazard Maps: CERI Report. 90 p. Crone, A. J., 1981. Sample Description and Stratigraphic Correlation of the New Madrid Test Well-1-X, New Madrid County, Missouri: U.S. Geological Survey Open-File Report 81-426. 26 p. Csontos, R. and Van Arsdale, R., 2008, New Madrid seismic zone fault geometry: Geosphere, Vol. 4, No. 5, pp. 802–813. Csontos, R.; Van Arsdale, R.; Cox, R.; and Waldron, B., 2008, The Reelfoot Rift and its impact on Quaternary deformation in the central Mississippi River Valley: Geosphere, Vol. 4, No. 1, pp. 145–158. doi:10.1130/GES00107.1. Cupples, W., and Van Arsdale, R., 2014, The preglacial “Pliocene” Mississippi River: Journal Geology, Vol. 122, pp. 1–15. Cushing, E. M.; Boswell, E. H.; and Hosman, R. L., 1964, General Geology of the Mississippi Embayment: U.S. Geological Survey Professional Paper 448-B. 28 p. Dart, R. and Swolfs, H. S., 1998, Contour mapping of relic structures in the Precambrian basement of the Reelfoot Rift, North American Midcontinent: Tectonics, Vol. 17, pp. 235–249. Elnashai, A. S.; Cleveland, L.; Jefferson, J. T.; and Harrald, J., 2009, Impact of New Madrid Seismic Zone Earthquakes on the Central USA, Vol. I: Mid-America Earthquake Center Release 09-03. 153 p. Fuller, M. L., 1912, The New Madrid Earthquake: U.S. Geological Survey Bulletin 494. 119 p. Gold, R. D.; DuRoss, C. B.; Delano, J. E.; Jibson, R. W.; Briggs, R. W.; Mahan, S. A.; Williams, R. A.; and Reide Corbetts, D., 2019, Four major Holocene earthquakes on the Reelfoot fault recorded by Sackungen in the New Madrid seismic zone, USA: Journal Geophysical Research: Solid Earth, Vol. 124, pp. 3105–3126. Greenwood, M. L.; Woolery, E. W.; Van Arsdale, R. B.; Stephenson, W. J.; and Patterson, G. L., 2016, Continuity of the Reelfoot fault across the Cottonwood Grove and Ridgely faults of the New Madrid seismic zone: Seismological Society America Bulletin, Vol. 106, pp. 2674–2685. doi:10.1785/0120150290. Guccione, M. J.; Mueller, K.; Champion, J.; Sheperd, S.; Carlson, S. D.; Odhiambo, B.; and Tate A., 2002, Stream response to repeated co-seismic faulting, Tiptonville Dome, New Madrid seismic zone: Geomorphology, Vol. 43, pp. 313–349.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

169

Reichenbacher, Van Arsdale, Cox, and Cramer Guo, L.; Magnani, M. B.; McIntosh, K.; and Waldron, B., 2014, Quaternary deformation and fault structure in the northern Mississippi Embayment as imaged by near-surface seismic reﬂection data: Tectonics, Vol. 33, pp. 807–823. Hardeman, W. D., 1966, Geologic Map of Tennessee: West Sheet. Hare, P. H. and Gardner, T. W., 1985, Geomorphic indicators of vertical neotectonism along converging plate margins, Nycoya Peninsula, Costa Rica. In Morisawa, M. and Hack, J. T. (Editors), Tectonic Geomorphology: Allen and Unwin, Boston, pp. 75–104. Hildenbrand, T. G., 1985, Rift structure of the northern Mississippi Embayment from the analysis of gravity and magnetic data: Journal Geophysical Research, Vol. 90, pp. 12607– 12622. Hildenbrand, T. G. and Hendricks, J. D., 1995, Geophysical setting of the Reelfoot Rift and relation between rift structures and the New Madrid seismic zone. In Shedlock, K. M. and Johnson, A. C. (Editors), Investigations of the New Madrid Seismic Zone: U.S. Geological Survey Professional Paper 1538-E, pp. E1–E30. Hosman, R. L, 1996, Regional Stratigraphy and Subsurface Geology of Cenozoic Deposits, Gulf Coastal Plain, South-Central United States: U.S. Geological Survey Professional Paper 1416-G, 35 p. Howe, J. R., 1985, Tectonics, Sedimentation, and Hydrocarbon Potential of the Reelfoot Aulocogen: Unpublished M.S. Thesis, University of Oklahoma, Norman. Jibson, R. W. and Keefer, D. K., 1989, Statistical analysis of factors aﬀecting landslide distribution in the New Madrid seismic zone, Tennessee and Kentucky: Engineering Geology, Vol. 27, pp. 509–542. Johnston, A. C. and Schweig, E. S., 1996, The enigma of the New Madrid earthquakes of 1811–1812: Annual Review Earth Planetary Sciences, Vol. 24, pp. 339–384. Keller, E. A. and Pinter, N., 2002, Active Tectonics: Earthquakes, Uplift and Landscape: Prentice Hall, Upper Saddle River, NJ. Kelson, K. I.; Simpson, G. D.; Van Arsdale, R. B.; Harris, J. B.; Haraden, C. C.; and Lettis, W. R., 1996, Multiple Late Holocene earthquakes along the Reelfoot fault, central New Madrid seismic zone: Journal Geophysical Research, Vol. 101,No. B-3, pp. 6151–6170. Lumsden, D. N. and Caudle, G. C., 2001, Origin of massive dolostone: The Upper Knox model: Journal Sedimentary Research, Vol. 71, No. 3, pp. 400–409. doi:10.1306/2DC4094E-0E4711D7-8643000102C1865D. Lumsden, D. N.; Cox, R. T.; Van Arsdale, R. B.; and Cupples, W. B., 2016, Petrology of Pliocene Mississippi River alluvium: Provenance implications: Journal Geology, Vol. 124, pp. 501–517. Magnani, M. B.; Blanpied, M. L.; Deshon, H. R.; and Hornbach, M. J., 2017, Discriminating between natural versus induced seismicity from long-term deformation history of intraplate faults: Science Advances, Vol. 3, No. 11, e1701593. doi:10.1126/sciadv.1701593. Magnani, M. B. and McIntosh, K., 2009, Toward an Understanding of the Long-Term Deformation in the Mississippi Embayment: National Science Foundation NEHRP Report 8HQGR0089/08HQGR0090.. Markewich, H. W.; Wysocki, D. A.; Pavich, M. J.; Rutledge, E. M.; Millard, H. T.; Rich, F. J.; Maal, P. B.; MeyerRubin; and McGeehin, J. P., 1998, Paleopedology plus TL, 10Be, and 14C dating as tools in stratigraphic and paleoclimatic investigations, Mississippi River valley, U.S.A.: Quaternary International, Vol. 51, pp. 143–167. doi:10.1016/s10406182(97)00041-4.

170

Martin, R. V. and Van Arsdale R. B., 2017, Stratigraphy and structure of the Eocene Memphis Sand above the eastern margin of the Reelfoot rift in Tennessee, Mississippi, and Arkansas, USA: Geological Society America Bulletin, Vol. 129, pp. 970–996. McKeown, F. A.; Hamilton, R. A.; Diehl, S. F.; and Glick, E. E., 1990, Diapiric origin of the Blytheville and Pascola arches in the Reelfoot rift, east-central United States: Relation to New Madrid seismicity: Geology, Vol. 18, pp. 1158–1162. Mueller, K. and Pujol, J., 2001, Three-dimensional geometry of the Reelfoot blind thrust: Implications for moment release and earthquake magnitude in the New Madrid seismic zone: Bulletin Seismological Society America, Vol. 91, pp. 1563–1573, doi:10.1785/0120000276. Obermeier, S. F., 1988, Liquefaction Potential in the Central Mississippi Valley: U.S. Geological Survey Bulletin 1832. 21 p. Obermeier, S. F., 1989, The New Madrid Earthquakes: An Engineering-Geologic Interpretation of Relict Liquefaction Features: U.S. Geological Survey Professional Paper 1336-B. 114 p. Odum, W.; Hofmann, F.; Van Arsdale, R.; and Granger, D., 2020, New 26 Al/10 Be and (U-Th)/He constraints on the age of the Upland Complex, central Mississippi River Valley: Geomorphology, Vol. 371, pp. 1–9. Parrish, S. and Van Arsdale, R., 2004, Faulting along the southeastern margin of the Reelfoot Rift in northwestern Tennessee revealed in deep seismic reﬂection proﬁles: Seismological Research Letters, Vol. 75, pp. 782–791. Penick, J. L., 1981, The New Madrid Earthquakes: University of Missouri Press, Columbia. Potter, P. E., 1955a, The petrology and origin of the Lafayette Gravel: Part 1. Mineralogy and petrology: Journal Geology, Vol. 63, No. 1, pp. 1–38. Potter, P. E., 1955b, The Petrology and origin of the Lafayette Gravel: Part 2. Geomorphic history: Journal Geology, Vol. 63, No. 2, pp. 115–132. Pratt, T. L., 2012, Kinematics of the New Madrid seismic zone, central U.S., based on stepover models: Geology, Vol. 40, pp. 371–374. Pratt, T. L.; Williams, R. A.; Odum, J. K.; and Stephenson, W. J., 2012, Origin of the Blytheville arch, and long-term displacement on the New Madrid seismic zone, central U.S. In Cox, R. T.; Tuttle, M. P.; Boyd, O. S.; and Locat, J. (Editors), Recent Advances in North American Paleoseismology and Neotectonics East of the Rockies: Geological Society of America Special Paper 493, pp. 1–16. doi:10.1130/2012.2493(01). Pryne, D.; Van Arsdale, R.; Csontos, R.; and Woolery, E., 2013, Northeastern extension of the New Madrid North fault– New Madrid seismic zone, central United States: Bulletin Seismological Society, Vol. 103, pp. 2277–2294. Purser, J. L. and Van Arsdale, R. B., 1998, Structure of the Lake County uplift: New Madrid seismic zone: Bulletin Seismological Society America, Vol. 88, pp. 1204–1211. Reichenbacher, R., 2020, Geologic Mapping of Dyer County, Tennessee. Unpublished M.S. Thesis, University of Memphis. 91 p. Reichenbacher, R.; Harrison, V.; Weathers, T.; Van Arsdale, R.; Cramer, C.; Arellano, D.; Tohidi, H.; and Horton, S., 2020, 3D Geologic Mapping in and Adjacent to the New Madrid Seismic Zone, Central United States: Geological Society of America abstract. https://gsa.confex.com/gsa/2020AM/ meetingapp.cgi/Paper/357982. Rittenour, T. M.; Blum, M. D.; and Goble, R. J., 2007, Fluvial evolution of the lower Mississippi River valley during the last 100 k.y. glacial cycle: Response to glaciation and sea-

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

Geology and Earthquake Hazard in Dyer County, Tennessee level change: Geological Society America Bulletin, Vol. 119, pp. 586–608. Rodbell, D. T., 1996, Subdivision, Subsurface Stratigraphy, and Estimated Age of Fluvial Terrace Deposits in Northwestern Tennessee: U.S. Geologic Survey Bulletin 2128. 24 p. Rodbell, D. T. and Bradley, L. A., 1993, Logs of Exploratory Trenches through Liquefaction Features on Late Quaternary Terraces in the Obion River Valley, Northwestern Tennessee: U.S. Geological Survey Miscellaneous Field Studies Map MF2218. Sahu, A. S. and Roy, S., 2015, Quaternary tectonic control on channel morphology over sedimentary low land: A case study in the Ajay-Damodar interﬂuve of Eastern India: Geoscience Frontiers, Vol. 6, pp. 927–946. Saucier, R. T., 1987, Geomorphological Interpretations of Late Quaternary Terraces in Western Tennessee and Their Regional Tectonic Implication: U.S. Geologic Survey Professional Paper 1336-A. 19 p. Saucier, R. T., 1994, Geomorphology and Quaternary Geologic History of the Lower Mississippi Valley, 2 vols.: U.S. Army Engineer Waterways Experiment Station, Vicksburg, MS. Schwalb, H. R., 1982, Paleozoic Geology of the New Madrid Area, U.S.: Nuclear Regulatory Commission NURG/CR-2909. 6 p. Schumm, S. A., 1986, Alluvial river response to active tectonics. In Wallace, R. E. (Panel Chair), Studies in Geophysics— Active Tectonics: National Academies Press, Washington, DC, pp. 80–94. Thomas, W. A., 1991, The Appalachian-Ouachita rifted margin of southeastern North America: Geological Society America Bulletin, Vol. 103, pp. 415–431. Thomas, W. A., 1989, The Appalachian-Ouachita orogen beneath the Gulf Coastal Plain between the outcrops in the Appalachian and Ouachita Mountains. In Hatcher, R. D.; Thomas, W. A.; and Viele, G.W. (Editors), The AppalachianOuachita Orogen in the United States: Geological Society of America, Boulder, CO, pp. 537–553. Tuttle, M. P.; Collier, J.; Wolf, L. W.; and Lafferty, R.H., III, 1999, New evidence for a large earthquake in the New Madrid seismic zone between A.D. 1400 and 1670: Geology, Vol. 27, No. 9, pp. 771–774. Tuttle, M. P.; Schweig, E. S., III; Campbell, J.; Thomas, P. M.; Sims, J. D.; and Lafferty, R. H., III, 2005, Evidence for New Madrid earthquakes in A.D. 300 and 2350 B.C.: Seismological Research Letters, Vol. 76, pp. 489–501. Tuttle, M. P.; Schweig, E. S.; Sims, J. D.; Lafferty, R. H.; Wolf, L. W.; and Haynes, M. L., 2002, The earthquake potential of the New Madrid seismic zone: Seismological Society America Bulletin, Vol. 92, pp. 2080–2089. doi:10.1785/01200 10227.

Van Arsdale, R. B., 2009, Adventures through Deep Time: The Central Mississippi River Valley and Its Earthquakes: Geological Society of America Special Paper 455. 107 p. Van Arsdale, R., 2014, The New Madrid seismic zone of the Central United States: In Talwani, P. (Editor), Intraplate Earthquakes: Cambridge University Press, Cambridge, U.K., pp. 162–197. Van Arsdale, R. B.; Arellano, D.; Stevens, K. C.; Hill, A. A.; Lester, J. D.; Parks, A. G.; Csontos, R. M.; Rapino, M. A.; Deen, T. S.; Woolery, E. W.; Harris, J. B., 2012, Geology, geotechnical engineering, and natural hazards of Memphis, Tennessee, USA: Environmental Engineering Geoscience, Vol. 18, pp. 113–158. doi:10.2113/gseegeosci.18.2.113. Van Arsdale, R. B.; Bresnahan, R. P.; McCallister, N. S.; and Waldron, B., 2007, The Upland Complex of the central Mississippi River Valley: Its origin, denudation, and possible role in reactivation of the New Madrid seismic zone. In Stein, S. and Mazzotti, S. (Editors), Continental Intraplate Earthquakes: Science, Hazard, and Policy Issues: Geological Society of America Special Paper 425, pp. 177–192. Van Arsdale, R. B.; Cox, R. T.; Johnston, A. C.; Stephenson, W. J.; and Odum, J. K., 1999, Southeastern extension of the Reelfoot fault: Seismological Research Letters, Vol. 70, pp. 348–359. Van Arsdale, R. and Cupples, W., 2013, Late Pliocene and Quaternary deformation of the Reelfoot rift: Geosphere, Vol. 9, No. 6, pp. 1819–1831. doi:10.1130/GES00906.1. Van Arsdale, R.; Pryne, D.; and Woolery, E., 2013, Northwestern extension of the Reelfoot North fault near New Madrid, Missouri: Seismological Research Letters, Vol. 84, No. 6, pp. 1114–1123. Van Arsdale, R. B. and Tenbrink, R., 2000, Late Cretaceous and Cenozoic geology of the New Madrid seismic zone: Bulletin Seismological Society America, Vol. 90, No. 2, pp. 345–356. Ward, A., 2016, Progress toward Quaternary Displacement Rates on the Meeman-Shelby Fault and Joiner Ridge Horst, Eastern Arkansas: Unpublished M.S. Thesis, University of Memphis. 446 p. https://umwa.memphis.edu/etd. Weathers, T. and Van Arsdale, R., 2019, Lake County Tennessee: In the Heart of the New Madrid seismic zone: Frontiers Earth Science, Vol. 7, pp. 1–20. Wolf, L. W.; Tuttle, P. M.; Browning, S.; and Park, S., 2006, Case history: Geophysical surveys of earthquake-induced liquefaction deposits in the New Madrid seismic zone: Geophysics, Vol. 71, No. 6, pp. 223–230. doi:10.1190/1.2353801.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 147–171

171

A New Look at Landslides of the Vermilion and Echo Cliﬀs, Northern Arizona CONOR M. WATKINS* J. DAVID ROGERS Department of Geosciences and Geological Engineering, Missouri University of Science & Technology, Rolla, MO 65409

Key Terms: Landslides, Toreva Blocks, Geomorphology, Block Kinematics, Rock Mechanics, Geotechnical Strength Parameters, Slope Stability, Factor of Safety, Erosion, Colorado River, Arizona ABSTRACT The Vermilion and Echo Cliffs form a nearly continuous escarpment more than 160 km long within the Colorado Plateau physiographic province of North America. The cliffs overlie the Marble Platform in northern Arizona and are located along the Colorado River, just upstream of the Grand Canyon. Large rotational block landslides mantle the erosional escarpment along most of its extent. Although these landslides have been noted for over 100 years, their likely origin has never been explained. Landslide failure surfaces appear to be influenced by the Petrified Forest Member of the Triassic Chinle Formation, a shale layer containing smectite clay weathered from volcanic ash. Although landslides are common along the majority of escarpments comprising the Colorado Plateau where the Petrified Forest Member and other shales outcrop, most appear to have been inactive since the early Holocene. Multiple generations of landslides and remnants of previous slides exist up to 3 km from the present cliff face. Multiple working hypotheses explaining these landslides are explored in this article, including past landslides and/or lava dams along the Colorado River within the Grand Canyon, periods of wetter climate with higher groundwater levels, and earthquakes related to nearby faulting and volcanism. Various sliding modes along these cliffs are described along with potential triggering mechanisms. Back-analysis of these landslides has been conducted using mechanical properties of the formations involved as well as varying groundwater levels. Calculated factors of safety for existing slides under present conditions are greater than unity, consistent with their apparent stability.

*Corresponding author email: cwatkin@mst.edu

INTRODUCTION AND BACKGROUND The Vermilion and Echo Cliffs form a nearly continuous escarpment overlying the Marble Platform in northern Arizona. This escarpment is mantled with landslides throughout the greater part of its more than 160-km length (see Figure 1). The sequence begins at the southern end of the Echo Cliffs near Cameron, AZ, and continues northward toward Lee’s Ferry, AZ, to where it is interrupted by the mouths of the Glen and Paria Canyons. It then continues to the west as the Vermilion Cliffs. The Echo Cliffs reach their maximum height of ∼550 m near Lee’s Ferry, but diminish in height to the south. Although intermittent landslides are present along the southern portion of the Echo Cliffs, slides are continuous along the northernmost 24 km, from Bitter Springs to Lee’s Ferry. These cliffs are developed along the eroded Echo Cliffs Monocline, which turns to the northwest and crosses the Colorado River at Lee’s Ferry, where it gradually diminishes along Paria Canyon. This structural feature allowed for the historic river crossing at this location by placing the resistant Shinarump Bench at river level. The bench is overlain by the erodible Petrified Forest Member of the Chinle Formation, which creates a natural topographic bench, suitable for wagons, horses, and cattle or sheep approaching the Colorado River. Although the crossing is no longer used, it was of significant historical importance since the Colorado River is confined within incised bedrock gorges both up- and downstream of this location for hundreds of kilometers. The Echo Cliffs were named in 1871 during John Wesley Powell’s second mapping expedition, while it was being led by his brother-in-law, Almon Thompson. One of the expedition members fired a pistol near Lee’s Ferry and allegedly counted 22 echoes. The Vermilion Cliffs were also named by this same party for their brilliant reddish coloration, which the explorers admired on their journey to Kanab, where they spent the winter and spring of 1871–72. The Echo Cliffs Monocline structurally controls the location of Paria Canyon where the Paria River runs along the eroded axis of the fold. The monocline becomes less pronounced upstream, where Paria Canyon

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

173

Watkins and Rogers

Figure 1. This overview map shows the distribution of landslides throughout the Vermilion and Echo Cliﬀs escarpments. The location of Figure 8 is shown with line A-A’, while Figure 11 is shown with line B-B’. The location of Figure 6a is shown by C. Line D-D’ shows the location of Figure 6b.

is deeply incised. The Echo Cliffs transition to the Vermilion Cliffs along the lower portion of Paria Canyon. The lower 12 km of Paria Canyon has incised into the Chinle Formation, allowing the canyon walls to undergo significant retreat due to mass wasting. This regression has allowed this portion of the canyon to form an open amphitheater, in contrast to the narrow slot canyon upstream. The Vermilion Cliffs separate the Marble Platform from the Paria Plateau (see Figure 1). The cliffs be-

174

gin at the mouth of Paria Canyon near Lee’s Ferry and achieve a maximum height of ∼610 m. From Lee’s Ferry, they trend southwesterly for 24 km toward a small settlement known as “Cliff Dwellers.” Here, the cliffs turn northwest, continuing another 32 km until they turn due north, along the House Rock Valley. The cliffs gradually decrease in height along the East Kaibab Monocline, or “Cockscomb,” as it is locally known, decreasing in height to about ∼305 m along the southern portion of the House Rock Valley. The

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides

Figure 2. Hummocky landslides mantle much of the Vermilion and Echo Cliﬀs escarpments. This shows the rotational slumps (Toreva blocks) that typify the western Vermilion Cliﬀs.

cliffs continue northward and west through southern Utah, although that segment of the cliffs was not examined as a part of this study. The Vermilion Cliffs are continuously mantled with landslides for a distance of approximately 56 km, between the mouth of Paria Canyon and the northern end of House Rock Valley, where they cease to occur. Figure 2 shows an example of rotational slumps along the Vermilion Cliffs near House Rock. STRATIGRAPHY AND STRUCTURE The stratigraphic units exposed along the Vermilion and Echo Cliff escarpments include the Navajo, Kayenta, Moenave, Chinle, Moenkopi, and Kaibab Formations, of Jurassic to Permian age. Landsliding in the region generally floors in the Petrified Forest Member of the upper Triassic Chinle Formation. This member is well known for its inclusion of Petrified Wood and development of badlands topography. The Petrified Forest Member comprises shales, siltstones, sandstones, and channel lag gravels deposited in an extensive fluvial system (Dubiel, 1989), making its composition quite variable. The shales/clays in the Petrified Forest Member are derived from ∼60 percent volcanic materials, including ash (Riggs et al., 1993), and are rich in smectite clays (Chasteen et al., 2001). Smectites are recognized for their contribution to slope instability because of their low shear strength (Tiwari and Marui, 2005). This member overlies the more resistant Shinarump Member, which forms the basal unit of the Chinle. The Chinle has a thickness of ∼130 m throughout the Vermilion and Echo Cliffs area. Figure 3 presents a stratigraphic section of the formations exposed along the Vermilion and Echo Cliffs.

The relatively competent Jurassic age Navajo, Kayenta, and Moenave Formations overlie the Chinle and cap the cliffs. The Jurassic section has a cumulative thickness of ∼600 m, although the entire section isn’t exposed along the faces of the Vermilion and Echo Cliffs. The resistant Shinarump Member of the Chinle Formation, which comprises cross-bedded channel sands and lag gravels, forms a prominent bench overlying the Triassic Moenkopi Formation. The Moenkopi in the vicinity of Lee’s Ferry consists of siltstones, gypsum lenses, sandstone ledges, and minor amounts of limestone. It is easily eroded where the overlying Shinarump Bench has been removed. An undifferentiated member is also present in the vicinity of Lee’s Ferry that is usually a slope former (Stewart et al., 1972). The Triassic Moenkopi is extensively composed of red beds. The Shinarump Bench and Moenkopi Formation are pronounced elements forming the base of the escarpment along the eastern Vermilion Cliffs, lower Paria Canyon, and portions of the Echo Cliffs. The Moenkopi overlies the Permian Kaibab Formation, which forms the Marble Platform and the rims of Marble and Grand Canyons downstream. ECHO CLIFFS MONOCLINE The Echo Cliffs are structurally controlled by the Echo Cliffs Monocline, a prominent Laramide fold in the western Colorado Plateau. This monocline trends north-south and is upthrown on its western side. The upthrown side of the monocline has been eroded, creating the Echo Cliffs escarpment, which dips to the east at an inclination of 10–30° (Strahler, 1940), creating obsequent bedding (dipping into the slope). The brittle formations comprising the Echo Cliffs appear more intensely jointed than the relatively undeformed strata exposed in the Vermilion Cliffs. This may be due to the extension caused by the monoclinal flexure. The monocline turns to the northwest near Lee’s Ferry, where it crosses the Colorado River and trends just north of the axis of Paria Canyon. VERMILION CLIFFS/PARIA PLATEAU The Vermilion Cliffs form the southern and western limits of the Paria Plateau, which dips to the northwest (eastern cliffs) and north (western cliffs) at inclinations of 2° to 2.5°. The cliffs gradually decrease in height to the west. The cliffs have a maximum height of around 610 m near Lee’s Ferry and diminish to a height of 305 m in House Rock Valley. This is due to a combination of their dip and the increasing elevation of the Marble Platform to the west, approaching the East Kaibab Monocline, a Laramide fold separating the Paria Plateau from the Kaibab Plateau. The

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

175

Watkins and Rogers

Figure 3. Principal formations comprising the Vermilion Cliffs escarpment near Lee’s Ferry. The Shinarump Member of the Chinle Formation forms a prominent bench and armors the less resistant Moenkopi Formation from erosion. The Kaibab, Moenkopi, and Shinarump are not exposed along the western Vermilion Cliffs or southern Echo Cliffs.

Shinarump Bench and Moenkopi Formation are not prominently exposed along the western portion of the cliffs, where headward incision by streams has yet to occur. DESCRIPTION OF LANDSLIDING Prior Work in the Area Davis (1901) appears to have been the first geologist to describe landsliding along the Vermilion and Echo Cliffs. Prior reports by Powell and others make no mention of landslides, although they discuss the Triassic strata exposed in the Vermilion and Echo Cliffs. Davis thought these slides were related to a hiatus and then renewed incision by the Colorado River in the vicinity of Lee’s Ferry. He noted two forms of landsliding: coherent rotational slides and disaggregated rock slide-debris avalanches (Hungr et al., 2014), or “sturztroms,” which he believed formed when rotational slides broke apart as they cascaded over the Shinarump Bench. His 1901 map (reproduced in Figure 4) shows the occurrence of landslides along both sets of cliffs. Reich (1937) also identified landsliding along the Vermilion Cliffs, although it doesn’t appear that he studied this area in much detail. His 1937 article defines the Toreva block, or rotational bedrock slide, which he named after the type locality near Toreva, AZ (approximately 150 km to the east-southeast). He makes brief mention of their occurrence at several other locations across the southern Colorado Plateau, including the Vermilion Cliffs. He curiously states that

176

Toreva blocks are not present along the Echo Cliffs and are somewhat rare along the Vermilion Cliffs. Prior and subsequent researchers identified numerous rotational bedrock slides along both sets of cliffs. Strahler (1940) studied the landslides of the Vermilion and Echo Cliffs in greater detail. He made note of both Toreva blocks and disaggregated “rockslides” along both the Vermilion and Echo Cliffs where the cliffs are tallest and the Shinarump Bench is prominent, as noted by Davis (1901). Like Davis, he believed

Figure 4. This diagram from Davis (1901) shows the areal extent of the landslides mantling the lower elevations of Vermilion and Echo Cliﬀs near Lee’s Ferry, AZ.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides

Figure 5. Series of cross sections modiﬁed from Strahler (1940) illustrate the retrogressive nature of rotational landslides observed along the western Vermilion Cliﬀs in the vicinity of House Rock.

that these features formed when Toreva blocks cascaded over the Shinarump Bench and disaggregated. The upper cliff-forming formations are brittle and regionally jointed, aiding in their disintegration as they cascade over the lower Shinarump Bench. Our research has found similar disaggregated slides in the western Vermilion Cliffs near House Rock. Strahler also identified eroded remnants of Toreva blocks nearly a mile (1.6 km) in front of the escarpment. He also observed that the most recent rotational slides adjacent to the Vermilion Cliffs are only rotated 10–15° and exhibit a rougher and more ragged appearance than those farther from the face of the escarpment, indicative of relative youthfulness (see Figure 5). The younger slides exhibit increasing basal elevations of the failure surfaces, a phenomenon likely caused by the buttressing effect of the pre-existing landslides at the toes of the slopes. Echo Cliffs Strata comprising the Echo Cliffs dips to the east, or back into the slope, forming adverse bedding. Such inclinations normally increase stability and reduce the propensity to landsliding. The southern Echo Cliffs exhibit intermittent occurrences of landsliding, likely because the Chinle Formation is not fully exposed along the base of the cliffs. Most of the slope failures are relatively shallow (tens of meters) rotational slumps that appear to have initiated along regional systematic joints. The northernmost portion of the cliffs, running ∼24 km from Bitter Springs to Lee’s Ferry, is continuously mantled by landslides (see Figure 1). The entire stratigraphic section of the Chinle Formation is ex-

posed in this area. Landslides along this portion of the Echo Cliffs are present as rotational slump blocks and highly disaggregated rockfalls, which appear to have flowed down the slopes, whether or not they cascaded over the Shinarump Bench. All of the landslides observed along the Echo Cliffs appear to have failed against bedding, with the exception of one major landslide complex on the east side of the Colorado River near Lee’s Ferry. Figure 6a presents an overview of this feature from the Spencer Trail, which overlooks Lee’s Ferry. This composite landslide translated within the Chinle Formation directly above the Shinarump Member, which is inclined in a ramp-like form toward the Colorado River, where the Echo Cliffs Monocline crosses the channel. This landslide consists of a series of relatively intact blocks covered by a veneer of disaggregated rubble. The complex also includes enormous rock slidedebris avalanches, which may be of partially subaqueous origin. These lobes appear to have flowed down the inclined Shinarump Bench. At least one intermittent drainage has since dissected the parent mass. This slide appears to have deflected the Colorado River slightly to the north, just upstream of Lee’s Ferry. A typical cross section of the Echo Cliffs is shown in Figure 6b. Vermilion Cliffs Landslides mantling the Vermilion Cliffs exhibit two distinctive styles, as suggested by Davis (1901) and Strahler (1940). The classic rotational slide, or Toreva block–style landslide, is common throughout the western portion of the cliffs (see Figures 2 and 7), while disaggregated, rock slide-debris avalanche de-

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

177

Watkins and Rogers

Figure 6. a. Composite landslide mass exposed at the northern end of the Echo Cliffs at Lee’s Ferry. These slides appear to floor in the Chinle Formation just above the Shinarump Member, which forms a sloping, resistant platform. b. This view of the Echo Cliffs looking south from Marble Canyon, AZ, illustrates the dipping strata within the Echo Cliffs Monocline. Obsequent dip back into the slope limits the scale of landslides along the Echo Cliffs. The Shinarump Member caps the underlying Moenkopi Formation, creating a sloping, resistant platform.

posits are more numerous in the eastern Vermilion Cliﬀs, where the escarpment reaches its greatest height. This disaggregation appears to have occurred where rotational slides cascaded over the Shinarump Bench, transitioning to rock slide-debris avalanches capable of

178

ﬂowing long distances (Legros, 2002; Iverson, 2003). These form the distinctive debris lobes shown in Figures 8 through 13 below. All of these debris fans are highly eroded and dissected, suggesting that these features formed some time ago, likely during the late

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides

Rock Slide-Debris Avalanches

Figure 7. The rotational slumps of the western Vermilion Cliffs were first described by Davis (1901) and Strahler (1940). Multiple sequences of landslides are present along the escarpment, suggesting a sustained history of cliff retreat via retrogressive mass wasting.

Pleistocene (>14.5 ka). Landsliding ceases where the cliffs turn north along the East Kaibab Monocline, slightly north of House Rock (Figure 1). The cliffs decrease in height along their western limits north of House Rock. There is a lesser degree of incision/headward erosion from Colorado River tributaries, mainly House Rock Wash, and the floor of the House Rock Valley increases in elevation proceeding northward along its axis. The apparent dip of the Paria Plateau also shifts northward. The combination of these factors has led to a condition in which the underlying Chinle Formation has yet to be exposed along this stretch of the cliffs. The combination of a lower overburden stress and a buttressing of the weak Petrified Forest Member of the Chinle Formation has allowed this portion of the cliffs to remain inherently stable. Disaggregated masses of material are present atop many of the slump blocks in the western Vermilion Cliffs, indicative of rock slide-debris avalanches, described as “sturzstroms” in the literature (Skerner, 1989). These long run-out landslides are influenced by the conservation of momentum (Legros, 2002; Iverson, 2003) and consist of disaggregated debris emanating from the Jurassic section capping the cliffs. These deposits have been dissected by intermittent streams, exposing their contacts with underlying materials, including pre-existing rotational slump blocks. A cross section through one of these features near House Rock is presented in Figure 8, an aerial view in Figure 9, and a ground view in Figure 10. Figure 5 presents an additional cross section through the House Rock area, modified from Strahler (1940).

Rock slide-debris avalanches are most common where they litter the base of the eastern Vermilion and northern Echo Cliffs, above and below the prominent Shinarump Bench. These physical relationships suggest that different triggers and/or mechanisms of mass wasting might have been operative in this area, closer to the Colorado River. Figure 11 presents a cross section of a slide that disaggregated and flowed over the Shinarump Bench. Figure 12 shows an aerial photo of a disaggregated slide, while Figure 13 shows how these features appear on the ground. Many of the landslide sequences along the Vermilion and Echo Cliffs appear to be of approximately the same age and morphology, increasing in age with offset from the current escarpment. In the western Vermilion Cliffs near House Rock it appears that at least three major episodes of cliff retreat occurred in distinct groupings. Each sequence of slides appears to be laterally contiguous and of similar age, suggesting that common and/or recurring mechanisms likely played a role in triggering each landslide sequence.

Potential Triggers of Landsliding along the Vermilion and Echo Cliffs A number of variables appear to have triggered repeated sequences of landsliding along the Vermilion and Echo Cliffs. The likely triggers are described and discussed below. Although landslides are extensive along this escarpment, few, if any, appear to have occurred during the Holocene.Strain Incompatibility and Strain Softening Within the Chinle Formation The Shinarump conglomerate, which is the basal member of the Chinle Formation, is a stiff sandstone/conglomerate deposited by outwash channels during the Triassic. The member is ∼30 m thick and typically forms a resistant ledge, commonly referred to as the “Shinarump Bench.” This bench, along with the underlying Moenkopi Formation, accounts for up to 190 m of the escarpment height and is most prominent in the vicinity of Lee’s Ferry. The Shinarump is an anomalously stiff material, as compared to the shaly units lying immediately beneath and above it, and it exhibits strength parameters on par with any of the massive sandstones deposited during the Mesozoic Era in this area (Rogers, 1982). The Shinarump caps several nearby mesas and buttes, with one of the largest being Shinumo Altar. The Petrified Forest and other shalerich members of the Chinle overlie the Shinarump. These members are considerably less stiff than the Shinarump and are susceptible to strain softening, especially when wetted (Mesri and Shahien, 2003).

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

179

Watkins and Rogers

Figure 8. Cross section through the western Vermilion Cliﬀs reveals multiple sequences of rotational slump blocks with relatively intact bedding. An older rotational block is present beyond the rightmost extent of this cross section, over 3 km from the face of the parent escarpment.

This stiffness contrast between the Shinarump and Petrified Forest Members promotes strain incompatibility at the contact, promoting a disproportionate amount of strain softening at this boundary within the overlying Petrified Forest Member when lateral support is removed by mass movements or erosion. Most of the landslides in this area appear to initiate just above the Shinarump, retrogressing upward through

the Petriﬁed Forest Member. Existing slides buttress the lower slopes, forcing succeeding slope failures to occur at higher elevations. Both Strahler (1940) and Ahnert (1960) observed dilated joints in the Navajo Sandstone capping the Vermilion Cliﬀs. Joint apertures increase markedly toward the escarpment (see Figure. 14), likely because of creep of the underlying Chinle Formation. The Chinle

Figure 9. Aerial oblique image of the western Vermilion Cliﬀs showing three sequences of landsliding, which appear to have occurred in separate episodes. Deposits of highly disaggregated materials, likely emanating from sturzstroms, lie between two sequences of back-rotated slump blocks.

180

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides

Figure 10. This image, taken in the western Vermilion Cliﬀs, shows the most recent rotational slump, adjacent to the escarpment, along with sturzstrom debris overlying the variegated shales of the intact Petriﬁed Forest Member of the Chinle Formation. An older slump block is just out of view to the right. The cross section shown in Figure 8 was taken through the eroded saddle seen in the upper third of this image.

shales exhibit high plasticity and strain softening with increasing loss of conﬁnement, often driven by repeated cycles of wetting and drying. Given the static strength parameters exhibited by these units, it would appear that strain softening must necessarily precede any gross landsliding along the Vermilion Cliﬀs escarpment. Nygard et al. (2006) summarized behaviors of other overconsolidated mudstone formations

and found similar losses of shear strength as confining stress is lessened.Seismicity Related to Volcanism and Faulting The extensive volcanic activity and faulting that have helped shape northern Arizona and New Mexico (Duffield, 1993) may have played an important role in the occurrence of landslides along the Vermilion and Echo Cliffs as well as other locations in the study

Figure 11. This cross section illustrates the disaggregated nature of the landslides along the eastern Vermilion Cliﬀs, where landslides have cascaded over the Shinarump Bench.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

181

Watkins and Rogers

Figure 12. Aerial oblique view of a disaggregated rockslide-debris avalanche (sturzstrom) along the eastern Vermilion Cliffs, where a slide flooring in the Petrified Forest Member of the Chinle Formation appears to have cascaded over the Shinarump Bench. The Shinarump Conglomerate stands out in this image as a resistant buff-colored band intermittently exposed near the base of the escarpment.

Figure 14. Aerial oblique view of the crest of the Vermilion Cliﬀs showing increasing dilation of regional joints (up to 1.5 m) in the Navajo Sandstone at the precipice of the escarpment. This is likely caused by creep of the softer Chinle beds, lying below the Jurassic section. Strain softening likely precedes landsliding along the escarpment.

area. Movement of magma within the earth’s crust often produces long-duration harmonic tremors. Hundreds of eruptions have taken place in the San Francisco Peaks Volcanic Field, with more than 600 cinder cones. This complex lies about 140 km to the south, along with the Uinkaret and Shivwits Volcanic Fields, located ∼120 km to the west. Evidence also points to sizable earthquakes triggered by faulting throughout the western margins of the Colorado Plateau.

It is possible that earthquakes coincided with wetter periods during the Pleistocene, which, acting together, could be expected to elevate pore water pressures and decrease slope stability. Several factors may have combined to trigger the observed landslides along the base of the cliﬀs today. The “episodic” nature of the diﬀerent landslide complexes supports this theory.

Lava Dams in the Western Grand Canyon or a Rockfall/Landslide Downstream

Figure 13. Disaggregated landslide debris mantling the Vermilion and Echo Cliffs, above and below the prominent Shinarump Bench near Lee’s Ferry. These landslides are strikingly different from those observed elsewhere along the Vermilion Cliffs, which are much further from the Colorado River. The clasts are angular and chaotic, without organized structure.

182

During the Pleistocene (last 1.2 million years), at least 13 lava dams infilled the Colorado River channel in the western Grand Canyon (Hamblin, 1994). These natural dams were short-lived, yet they must have backed water up to Lee’s Ferry and beyond. Prospect Lake is the name given by Hamblin (1994) to the highest lava dammed lake, formed by a succession of lava flows that occurred between 1.8 and 0.5 Ma. The highest lava dam appears to have reached a maximum elevation of 1,260 m, about 757 m above the present river level (∼35 to ∼45 m above the likely channel surface 0.5 Ma). The waters of Hamblin’s Prospect Lake would have extended to present-day Moab, UT, and covered Lee’s Ferry in ∼313 m of water. This would have easily saturated the toes of the eastern Vermilion Cliffs, northern Echo Cliffs, and the mouth of Paria Canyon, shown in Figure 15. The ponding of so much water against these cliffs would have saturated the Chinle Formation within just a few years, hastening loss of cohesion and triggering a successive series of slides

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides

Figure 15. Map showing the extent of Prospect Lake, with a surface elevation of 1,260 m above sea level, overlain on modern-day topography. Lee’s Ferry would have been under more than 300 m of water, while the slopes of Paria Canyon, the eastern Vermilion Cliﬀs, and northern Echo Cliﬀs would have also been inundated.

through simple strain softening (Trollope, 1969, 1973; Rogers and Pyles, 1980). The sudden failure of any of the lava dams would have allowed rapid draining of the sizable reservoirs, creating a “rapid drawdown condition” wherein negative pore pressures would develop within the saturated slopes as the reservoir waters evacuated. If any of the lava dams that failed suddenly retained reservoir pools above an elevation of ∼1,250 m, their sudden drainage could easily have triggered the massive landslide observed around Lee’s Ferry (elevation 947 m). This elevation coincides with the base of the Vermilion and Echo Cliﬀs near Lee’s Ferry, where the Shinarump Bench crosses the Colorado River (Figure 6). The Shinarump Bench creates a resistant platform, allowing landslides from above to cascade over a precipice, promoting their disintegration (Figures 11 and 12). The disaggregated rock slide-debris avalanches at the bases of the eastern Vermilion and northern Echo Cliﬀs, above and below the prominent Shinarump

Bench (Figures 12 and 13), suggest that different mechanisms of mass wasting were once operative in this area, closer to the Colorado River (Figure 14). The rapid draining of the temporary reservoirs impounded behind the lava dams could have triggered landslides floored in the overconsolidated Chinle shale, which is exposed along the eastern Vermilion and northern Echo Cliff escarpments. The concept of slope failures triggered by rapid drawdown was introduced by Casagrande (1950). Recent data suggest that incision rates of the Colorado River between Lee’s Ferry and the eastern Grand Canyon have varied between 100 and 150 m per million years (Pederson et al., 2002; Karlstrom, 2004; and Karlstrom et al., 2007). Research by Willis and Biek (2001) suggests that incision rates accelerated in the middle to late Quaternary. Polyak et al. (2004) discovered evidence that rates of incision in Marble Canyon near Redwall Cavern (32 mi downstream of Lee’s Ferry) have ranged as high as ∼550 m/million years during the last 82 ka. If we accept that the age of the Prospect Canyon Lava Dam at its maximum height was ∼0.50 Ma and use the above-cited incision rates, the Colorado River was likely between 50 and 75 m higher than it is at present. Direct shear tests by Rogers and Pyles (1980) demonstrated that overconsolidated shales lose about two-thirds of their intrinsic cohesion upon complete saturation (Figure 4). Such significant strength loss could easily explain the observed failures. This strength loss could also account for repeated episodes of cliff retreat near Lee’s Ferry, the formation of the Lee’s Ferry re-entrant, and widening of the mouth of Paria Canyon each time a lava dam reservoir breached and drained. The rounding of Moenkopi slopes in the vicinity of Lee’s Ferry and Marble Canyon (Figure 16) resembles shorelines inundated and subsequently exposed in modern reservoirs. The texture of the Moenkopi throughout this area contrasts markedly with other Moenkopi outcrops in the western Colorado Plateau, suggesting that the area was once submerged by a natural lake. The two large rockfall sites within Marble Canyon at Nankoweap Creek and President Harding Rapid may have also played a role in backing up water to Lee’s Ferry and beyond. Driftwood logs and fine-grained sediment discovered within Stanton’s Cave support the theory that a large lake was once present in this portion of the canyon (Hereford, 1984). Since landslides along the Vermilion and Echo Cliffs also occur in large numbers, well above the level of any potential prehistoric lakes and throughout the surrounding region, additional components of de-stabilization (or, triggering mechanisms) may have been operative at the time of sliding.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

183

Watkins and Rogers

Chinle Formation form a relatively impermeable barrier, which would serve to perch groundwater within the cliﬀs, elevating pore pressures and reducing unit cohesion, both of which hasten slope creep and strain softening. Both Strahler (1940) and Ahnert (1960) mention the presence of joint planes in the Navajo Formation and have asserted that these structures likely played a role in the channeling of surface water into the subsurface. Ahnert (1960) notes that almost all retreating escarpments mantled by landslide complexes in the Colorado Plateau consist of permeable sandstones underlain by weaker rocks, mainly shales.

Figure 16. Rounded topography near Lee’s Ferry resembles that of modern-day dry lake beds. This shows the smoothed bedding of the Moenkopi outcrops unique to this area.

Wetter Pleistocene Climate Based on evidence provided by oxygen isotope ratios, pollen abundances, and sediment accumulation, it appears that the climate was cooler and wetter throughout the southern Colorado Plateau during the Wisconsin glaciation (Betancourt, 1984; Cole, 1990; Anderson, 1993; Dryer, 1994; Anderson et al., 2000; Rogers et al., 2004; and Watkins et al., 2007). Based on the references cited above, analysis of sediments throughout the region suggests a higher annual rate of precipitation, possibly by as much as 35–60 percent, and a cooler climate. This precipitation likely arrived in gentle, long-duration events, not via the infrequent torrential downpours common throughout the arid region today. This would have allowed more precipitation to infiltrate the subsurface instead of running off. A cooler climate would also reduce losses to evaporation. Morgenstern and Eigenbrod (1974) described how shales lose shear strength with complete saturation (which can take considerable time). As recently as April 1995, a large paleolandslide partially reactivated, damming the North Fork of the Virgin River within Zion National Park. The previous winter season had been wetter than average, saturating basal shales (Wieczorek and Schuster, 1995). Permeable Sandstone Overlying Impermeable Shales The Navajo Sandstone is permeable and systematically jointed, with at least two major sets being nearly vertical (Rogers, 1982; see Figure 14, above). These characteristics allow precipitation falling on the Paria Plateau to easily infiltrate the sandstone and the underlying formations. However, the underlying shales of the

184

Triggers Appear Widespread but Are Not Presently Active Similar landslide complexes are present throughout the Colorado Plateau, and most appear inactive at the present time. The presence of landslides throughout the entire region suggests that the various triggering mechanisms were once widespread across the entire region. All of the landslides mantling the Vermilion and Echo Cliffs appear to predate the Holocene (>14.5 ka). Drainage patterns have been re-established, and there are presently no enclosed basins in the headscarp grabens. It appears that such features existed at one time, prior to the renewed incision that likely accompanied the increasingly arid climatic cycle that enveloped in the region during the Holocene. Similar retrogressive landslide complexes mantle other escarpments in the southern Colorado Plateau (Reiche, 1937; Strahler, 1940; Ahnert, 1960; and Radbruch-Hall et al., 1982). The landslides blanketing the base of the Vermilion and Echo Cliffs exhibit evidence of recurring episodically, suggesting a recurring trigger mechanism in the area. Ahnert (1960) noted what appeared to be groupings of landslides from distinct episodes, based on similar morphologies (weathering, local erosion, and offset from the present escarpment). He assumed the retreat sequences likely occurred during humid glacial cycles of the late Pleistocene. A wetter Pleistocene climate (Anderson and Betancourt, 2000; Huth et al., 2020) and/or seismic activity associated with regional tectonism (Keefer, 1993) and volcanism may have played a role, or may actually been responsible for, the widespread distribution of landslide features throughout this area and the surrounding region. Analysis and Likely Failure Mechanisms The authors’ detailed reconnaissance mapping of the landslides blanketing the Vermilion Cliffs suggests that the failures floor in thin seams of volcanic ash

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides

Figure 17. Stratigraphy, mechanical properties, assumed groundwater conditions, and incipient failure surface of a typical landslide along the western Vermilion Cliﬀs.

within the Petrified Forest Member of the Chinle Formation. The ash appears to have weathered into smectite clays, rich in montmorillonite. X-ray diffraction tests were conducted in the Materials Research Center at the Missouri University of Science and Technology, confirming this assumption. These beds are highly anisotropic and behave like overconsolidated shales, which tend to exhibit significantly lower shear strength parallel to bedding. For this reason, the basal sliding surfaces of most landslides along the Vermilion Cliffs sole along these weaker beds or horizons. The Petrified Forest Member is markedly heterogeneous, comprising old channel sands and volcanic ash, which are laterally variable, even at a localized scale. Back-analyses of slope stability were conducted using Geo-Slope’s Geostudio Slope/W Version 2007, a limit equilibrium package that allows for several different methods of analysis. We employed the Morgenstern and Price (1965) Method because it has been shown to exhibit greater accuracy when applied to any slope, regardless of soil strength parameters or slope geometry (Duncan and Wright, 2005), because of its calculation of interslice forces. The back-analyses were predicated upon assumed groundwater levels, shown in Figures 17 and 18. The factor of safety for the slope was then held equal to 1.0 for a spectrum of likely groundwater levels, beginning with that assumed to be the highest possible groundwater condition for the paleo-escarpment. Groundwater levels were altered to high, low, and intermediate levels to back-calculate likely combinations of strength parameters. This analysis does not include the influence of Hamblin’s (1994) Prospect Lake on

groundwater levels. The location analyzed is to the west, near House Rock, above the maximum level of the lake. Table 1 summarizes the various properties, both measured and derived, used in the stability analyses. Ubiquitous vertical joints (without any tensile strength) were employed to model the massive Jurassic age sandstone units, negating the influence of cohesion on the Jurassic section, which comprises the cliffforming units. The mobilized cohesion of the Petrified Forest Member was assumed to be a fairly low value, just 4.8 kPa (100 psf). Duncan and Stark (1992) suggest that the use of an assumed cohesion in the backanalysis of the angle of internal friction (phi angle) for slope stability calculations will be self-compensating. This is based upon the fact that other conditions, such as changes in pore water conditions, will change the overall factor of safety in a similar manner when assumed cohesion and friction strength parameters are used. On the other hand, an incorrect assumption for pore water levels can lead to significant over- or underestimation of strength parameters, depending on whether the groundwater level is specified too high or too low within the section being analyzed, respectively. Zero cohesion values are often used for effective stress analyses (Cornforth, 2005), but the authors assumed 4.8 kPa cohesion for this analysis. Case of Retrogressive Toreva Blocks An effective friction angle of 37° was used for the initial rupture of the basal slip surface where it cuts across intact beds of the Petrified Forest Member, as

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

185

Watkins and Rogers

Figure 18. These four plots were generated by GeoTesting Express during direct shear tests (ASTM D3080) under variable conﬁnement of 7.5, 20, and 50 tsf. A phi angle of 11.7° was calculated, and a signiﬁcant strength loss was noted when sheared to residual strengths, as illustrated in the lower left quadrant.

shown in Figure 17. The weathered volcanic ash beds were assumed to have an eﬀective friction angle of just 11°. A composite value of these materials would be around 22°, so this was assumed to represent the residual shear strength along the slip surface after extensive smearing and mixing, caused by continued translation (movement). This is also shown in Figure 18. The assumed groundwater levels were then lowered, based upon reasonable assumptions with respect to fracture density, aperture, and geometry observed in

186

outcrops, as well as present-day seepage zones/springs observed along the cliffs. This model assumed a saturated zone slightly above the Chinle beds, with groundwater filling ubiquitous joint elements to a depth not exceeding 20 percent the total height of the joints (shown in Figure 18). This lowering of the likely groundwater profile increased the factor of safety from ∼1.0 to ∼1.27. The factor of safety for the pre-existing landslide (Qls) buttresses, independent of the Toreva block,

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides Table 1. Factor of safety (FS) for varying phi angles and water levels within low-strength horizon. This table compares conditions and friction (phi) angles derived from parametric back-analyses and laboratory testing. “Transient very high water level” refers to extreme high groundwater conditions that are not expected to persist for long periods of time, possibly ranging from days to months. Water Level FS Phi angle (°)

Transient Very High

High

Medium

Low

1.000 0.812 0.866 0.909 15.450

1.110 0.878 0.948 1.000 11.700

1.148 0.910 0.983 1.036 10.550

1.236 1.000 1.071 1.120 7.700

Extreme high (15.45) Extreme low (7.7) ADOT (10) Geotesting (11.7) Back-calculated phi angle (°) at FS = 1.0 for speciﬁed water levels ADOT = Arizona Department of Transportation.

shown in Figure 17 was found to be around 1.05, close to instability. Such a low factor of safety likely contributed to the numerous reactivations and regression of the escarpment, both of which are apparent by direct observation. For these analyses, an incipient slide surface was selected from several sliding surfaces observed in the field. Preliminary slope stability analyses showed that this slip surface produced the lowest factor of safety when compared to other alternatives, making it the logical choice for back-analysis. Case of Sturzstrom Toe Buttress In order to evaluate the impact of buttressing by previous landslides, we modeled the remnants of a sturzstrom-type landslide, similar to those observed along the western margins of the escarpment (see Figures 9 and 10). The strength parameters of the sturzstrom debris were assumed, based on literature review and experience with analyzing modern sturzstroms in the San Juan Mountains (Rogers and Beckmann, 2003). These analyses suggest the factor of safety for the sturzstrom buttress is about 1.97, while the factor of safety of the incipient landslide mass is ∼1.34. We can conclude, therefore, that the slopes buttressed by sturzstrom debris appear to be inherently more stable than those buttressed by rotational slumps because a low-strength basal slip surface is preserved in the former, but is absent in the latter. A sketch of the analysis setup is shown in Figure 17, in which the incipient failure surface is shown by a dashed line. These analyses revealed the stabilizing effects of confinement on successive sequences of mass movement. The older slides serve to buttress the Petrified Forest Member, which is likely the weakest horizon of the escarpment. The back-analyses suggest that the main escarpment slopes should remain stable (factor of safety

> 1.3) until the sturzstrom debris blanketing the Petrified Forest Member is substantially removed, through erosion or re-activation. Critical Slope Stability Analysis of the Vermilion Cliffs The Bitter Springs Landslide, named for the nearby community of Bitter Springs, AZ, closed Highway 89 in 2013. This site is where Highway 89 climbs over the Echo Cliffs from its junction with Highway 89A north toward Page, AZ. The landslide destroyed Highway 89, requiring extensive reconstruction, which took 2 years to complete. Analyses conducted by Kleinfelder West, Inc. (McCormick and Richmond, 2014) as part of a geotechnical report prepared for the Arizona Department of Transportation produced a wealth of geotechnical testing information about the Echo Cliffs. These cliffs are the stratagraphic counterpart to the Vermilion Cliffs to the west. The results of the analysis conducted by Kleinfelder closely correlate with the authors’ testing and analysis of similar samples. The authors’ samples were recovered at a site along the Vermilion Cliffs near House Rock, AZ, approximately 20 mi (straight line distance) to the west. As with the landslides near House Rock, the Petrified Forest Member of the Chinle Formation appeared to be the controlling weak material in which the Bitter Springs Landslide initiated. Kleinfelder contracted with Cooper Testing Labs of Palo Alto, CA, who conducted torsional ring shear tests (ASTM D7608) on a re-molded sample of the slide plane material. The authors of this article also conducted a parametric back-analysis using extreme and actual conditions at the site to establish the likely range of strengths. The analysis was conducted using GeoStudio’s GeoSlope Slope/W (version 2007) slope stability application. The Morgenstern and Price Method was employed to conduct a limit equilibrium analysis.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

187

Watkins and Rogers

Figure 19. A phi angle of 15.45° was required within the Chinle shales to preclude mass movement of the Toreva block during maximum transient water levels, shown as a dashed blue line.

The main variable in the analysis was the groundwater level, with four conditions set. An extreme high level is illustrated in Figure 19, intended to replicate conditions likely present during the cooler and wetter climate of the Pleistocene. This value provides a maximum shear strength at which the cliﬀs remain stable. An extreme low level was then set (Figure 20), simulating extended drought periods. Such an analysis provides the lowest likely values by which the cliﬀs would

remain stable under these conditions. A likely groundwater surface for current conditions was also estimated and evaluated (Figure 21). A toe buttress comprising dissected landslide and disaggregated talus materials was included in each analysis. These analyses used actual slope proﬁles based on topography from the western Vermilion Cliﬀs near House Rock, where the samples were collected. In this area the Shinarump conglomerate has not been incised to form a bench, as it

Figure 20. In this back-analysis a phi angle of 7.7° within the Chinle shales was required to maintain stability during periods of extreme low water levels (dashed blue line).

188

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliﬀs Landslides

Figure 21. In this analysis a phi angle of 11.7° within the Chinle shales would be required to maintain stability during present-day groundwater conditions.

has to the east, near Bitter Springs. The factor of safety was held at 1.0 for each analysis. Both the Vermilion and Echo Cliffs exhibit adverse bedding where the strata dip back into the cliffs, opposite of the slope movement. This typically tends to resist failure. The axis of the Echo Cliffs Monocline closely parallels the face of the cliffs, resulting in extensive jointing of the bedrock. The jointing likely negates much of the stabilizing impact caused by the adverse bedding because it facilitates moisture infiltration along the Echo Cliffs. The authors’ parametric analyses provided a range of friction (phi) angles from 7.70° to 15.45°, with a likely value of ∼10.55°. Multiple samples of Chinle shale were collected by the authors along the base of the Vermilion Cliffs for analysis. Because of the cost of the testing involved, the samples were initially evaluated using conventional diagnostic tests, such as determining Atterberg limits and X-Ray diffraction. Both tests were conducted at Missouri Science and Technology. Although the samples were physically similar in appearance, the two analyses pinpointed one sample in particular that was the weakest of the collection. This sample was sent off for further testing, as described below. The authors also contracted with GeoTesting Express of Acton, MA, to conduct three direct shear tests (ASTM D 3080) with a maximum normal force of 4.79 MPa (50 tsf = 100,000 psf) to approximate pre-failure overburden conditions. From these tests, the friction (phi) angle of the material was determined to be 11.7°. The intrinsic cohesion was determined to be approximately 11 MPa (1.1 tsf). A significant loss of strength

was also noted during repeated shearing to residual strengths. Mobilized shear strength plots generated by these tests are summarized in Figure 18. All values from the various tests coincide well, with no more than a 17 percent variance between the key material properties. Table 1 compares the values determined from the parametric back-analyses, as well as the laboratory tests. CONCLUSIONS The Vermilion and Echo Cliffs of northern Arizona are mantled by extensive landslide complexes, which have contributed to 3 km or more of scarp regression. These failures are almost exclusively confined to basal translation within the Petrified Forest Member of the Triassic Chinle Formation and have occurred regardless of whether the stratigraphy dips into or away from the direction of movement. Both rotational and disaggregated sturzstrom-type landslides occur along the cliffs, although the sturzstroms are most pronounced where the cliffs reach their maximum height (550–610 m), near Lee’s Ferry. Landslides are almost continuous along the escarpment where the Petrified Forest Member of the Chinle Formation is fully exposed. The scale and intensity of landsliding gradually diminish in proportion to exposure of the Petrified Forest Member of the Chinle Formation. The landslides appear to be relicts of a past climatic regime, during the Pleistocene. They also appear to have occurred in separate, distinct episodes, indicative of recurrent or cyclical environmental triggers.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

189

Watkins and Rogers

Paleoecological evidence suggests a significantly wetter climate existed during much of the Pleistocene. Studies of paleovegetation also suggest that Pleistocene rainfall was probably less intense than that afforded by present-day storms and likely extended over a much longer period of time each year, similar to the kind of patterns experienced today in the lower latitudes of Canada (Rogers et al., 2004; Watkins et al., 2007). These sorts of environmental conditions would have allowed a much greater percentage of the precipitation to infiltrate the subsurface. The systematically jointed Jurassic sandstones likely channeled this water into the underlying shales of the Petrified Forest Member, gradually reducing effective shear strength and cohesion through simple saturation and gravityinduced creep under the enormous load of the steep escarpment. Seismicity associated with regional tectonism and volcanism might have played a role as well by temporarily elevating pore pressures in the shale and thereby reducing the effective stress and overall strength of the saturated strata. Other escarpments mantled by landslides within the Colorado Plateau share similar characteristics, both in terms of the landslides themselves and the formations involved. Occurrences of similar landslides throughout the entire region suggest that the environmental triggers were widespread.

We would also like to thank Gary T. Torosian, Joseph D. Tomei, and Ethan Maro of GeoTesting Express in Acton, MA, who supervised the shear tests on the Petrified Forest Member of the Chinle Formation. These were performed with normal confining stresses of 8, 20, and 50 tons/ft2 (100,000 lb/ft2 ), These were the highest normal force loads available from any geotechnical lab in the United States. They also performed a five-cycle residual shear test at a normal force of 20 tons/ft2 . We would also like to thank James J. Lemmon, RG of the Arizona Department of Transportation, for allowing us access to and copies of the geotechnical test data Arizona DOT and their consultants Kleinfelder West developed during their studies of the Bitter Springs Landslide along US Highway 89 in February 2013. This slide was in the Echo Cliffs across the Colorado River from the Vermilion Cliffs in the same stratigraphic units. The slide damage closed the highway for more than two years and cost $25 million to repair. There was an extensive program of geotechnical testing that included ring shear and soil plasticity tests by Cooper Testing Labs in Palo Alto, CA. These data corroborated similar testing we performed as part of our slope stability evaluations of the Vermilion Cliffs.

ACKNOWLEDGMENTS

Ahnert, F., 1960, The inﬂuence of Pleistocene climates upon the morphology of Cuesta scarps on the Colorado Plateau: Annals Association American Geographers, Vol. 50, No. 2, pp. 139– 156. Anderson, S. A., 1993, A 35,000 year vegetation and climate history from Potato Lake, Mogollon Rim, Arizona: Quaternary Research, Vol. 40, pp. 351–359. Anderson, S. A.; Betancourt, J. L.; Mead, J. I.; Hevly, R. H.; and Adam, D. P., 2000, Middle- and late-Wisconsin paleobotanic and paleoclimatic records from the southern Colorado Plateau, USA: Paleogeography, Paleoclimatology, Paleoecology, Vol. 155, pp. 31–57. Betancourt, J. L., 1984, Late Quaternary plant zonation and climate in Southeastern Utah: Great Basin Naturalist, Vol. 44, pp. 1–35. Casagrande, A., 1950, Notes on the design of earth dams: Journal Boston Society Civil Engineers, Vol. 37, No. 4, pp. 405–429. Chasteen, K. R., 2001. Spectroradiometer analysis of clays from the Petriﬁed Forest Member of Triassic Chinle Formation, Southwestern Utah. 13th Annual SSM Poster Session, College of Charleston, Charleston, SC. Cole, K. L., 1990, Late Quaternary vegetation gradients through the Grand Canyon. Packrat middens: the last 40 ka, 40:240– 258. Cornforth, D. H., 2005, Landslides in practice: Investigation, Analysis and Remedial/Preventive Options in Soil, John Wiley & Sons, Hoboken, NJ, 596 p. Davis, W. M., 1901, An excursion to the Grand Canyon of the Colorado: Bulletin: Museum of Comparative Zoology, 201 p. Dryer, J. D., 1994, Late Pleistocene Vegetation Change at Stanton’s Cave, Colorado River, Grand Canyon National Park, Arizona: M.S. Thesis, Northern Arizona University.

This study was made possible through funds provided by the Hasselmann Endowment of the Department of Geosciences, Geological and Petroleum Engineering, and the Natural Hazards Mitigation Institute at the Missouri University of Science and Technology. We are also thankful for support from the Norman R. Tilford Field Studies Scholarship, 2005, which also assisted in funding this research. The authors would like to express their appreciation to individuals who helped collect soil samples at the site, including Dr. J. David Rogers, Ronald Watkins, John Cable, and Garrett Euler, Ryan Seabaugh, Craig Kaibel, and Kevin James. Shamrock Consulting performed an array of laboratory tests on soil and rock samples recovered from the Vermilion Cliffs. These included evaluations of liquid limit, plastic limit, and plasticity indices of various horizons of the Petrified Forest Member of the Chinle Formation, X-Ray diffraction tests of clay minerals recovered from the same samples, onedimensional consolidation using incremental loading, one-dimensional soil swell or collapse potential of the clay shale beds, measurement of swell pressure with increasing confinement, and specific gravity tests. These were performed under the supervision of Dr. Kerry R. Magner, PE of Shamrock Consulting.

190

REFERENCES

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Vermilion and Echo Cliffs Landslides Dubiel, R. F., 1989, Depositional Environments of the Upper Triassic Chinle Formation in the Eastern San Juan Basin and Vicinity, New Mexico: U.S. Geological Survey Bulletin. Bulletin No. 1808:1–22B. Duffield, W. H., 1993, Volcanoes of Northern Arizona: Grand Canyon Association, 68 p. Duncan, J. D. and Wright, S. G., 2005, Soil Strength and Slope Stability: John Wiley and Sons. 297 p. Duncan, J. M. and Stark, T. D., 1992, Soil Strengths from Back Analysis of Slope Failures, Stability and Performance of Slopes and Embankments-II: American Society of Civil Engineers Geotechnical Special Publication No. 31, pp. 890–904. Hamblin, K., 1994, Late Cenozoic Lava Dams in the Western Grand Canyon: Geological Society of America Memoir 183, 139 p. Hereford, R., 1984, Driftwood in Stanton’s Cave: The Case for Temporary Damming of the Colorado River at Nankoweap Creek in Marble Canyon, Grand Canyon National Park, Arizona, The Archaeology, Geology, and Paleobiology of Stanton’s Cave, Grand Canyon National Park, Arizona: Grand Canyon Natural History Association Monograph No. 6, pp. 99–106. Hungr, O.; Leroueil, S.; and Picarelli, L., 2014, The Varnes classification of landslide types, an update: Landslides, Vol. 11, No. 2, pp. 167–194. Huth, T. E.; Cerling, T. E.; Marchetti, D. W.; Bowling, D. R.; Ellwein, A. L.; Passey, B. H.; Fernandez, D. P.; Valley, J. W.; and Orland, I. J., 2020, Laminated soil carbonate rinds as a paleoclimate archive of the Colorado Plateau: Geochimica Cosmochimica Acta, Vol. 282, pp. 227–244. Iverson, R. M., 2003, How should mathematical models of geomorphic processes be judged? In Wilcock, P. R. and Iverson, R. M. (Editors), Prediction in Geomorphology: American Geophysical Union Geophysical Monograph Series, Vol. 135. 256 p. Karlstrom, K. E., 2004, Colorado River System of the Southwestern U.S.: Longitudinal profiles, differential incision, and a hypothesis for Quaternary tectonism at both ends: Geological Society America Abstracts Programs, Vol. 36, No. 5, p. 550. Karlstrom, K. E.; Crow, R. S.; Peters, L.; McIntosh, W.; Raucci, J.; Crossey, L. J.; Umhoefer, P.; and Dunbar, N., 2007, 40Ar/39Ar and field studies of Quaternary basalts in Grand Canyon and model for carving Grand Canyon: Quantifying the interaction of river incision and normal faulting across the western edge of the Colorado Plateau: Geological Society America Bulletin, Vol. 119, No. 11/12, pp. 1283–1312. Keefer, D. K., 1993, The susceptibility of rock slopes to earthquake-induced failure: Bulletin Association Engineering Geologists, Vol. 30, No. 3, pp. 353–361. Legros, F., 2002, The mobility of long-runout landslides: Engineering Geology, Vol. 63, pp. 301–331. McCormick, W. V.; Richmond, J. C.; and Dahlen, K., 2014, Revised Geotechnical Assessment Bitter Springs Landslide US Highway 89 MP 526 ER-089E(205)T Tracs H 8641 01D, Coconino County, AZ. 58 p. Mesri, G. and Shahien, M., 2003, Residual shear strength mobilized in first-time slope failures: Journal Geotechnical Geoenvironmental Engineering, Vol. 129, No. 1, pp. 12–31. Morgenstern, N. R. and Eigenbrod, K. D, 1974, Classification of Argillaceous soils and rocks: Journal Geotechnical Engineering Division, American Society Civil Engineers, Vol. 100, No. GT10, pp. 1137–1158.

Morgenstern, N. R. and Price, V. E., 1965, An analysis of the stability of general slip surfaces: Geotechnique, Vol. 15, pp. 79–93. Nygard, R.; Gutierrez, M.; Bratli, R. K.; and Hoeg, K., 2006, Brittle-ductile transition, shear failure, and leakage in shales and mudrocks: Marine Petroleum Geology, Vol. 23, pp. 201–212. Pederson, J.; Karlstrom, K.; Sharp, W.; and McIntosh, W., 2002, Differential incision of the Grand Canyon related to Quaternary faulting–Constraints from U-series and Ar/Ar dating: Geology, Vol. 30, No. 8, pp. 739–742. Polyak, V. J.; Hill, C. A.; and Asmerom, Y., 2004, History of headward erosion of the Grand Canyon from a study of speleothems: Geological Society America Abstracts Programs, Vol. 36, No. 5, p. 549. Radbruch-Hall, D. H.; Colton, R. B.; Davies, W. E.; Luchitta, I.; Skipp, B. A.; and Varnes, D. J., 1982, Landslide Overview Map of the Conterminous United States: U.S. Geological Survey, Geological Survey Professional Paper 1183. Reiche, P., 1937, The Toreva-Block: A distinctive landslide type: Journal Geology, Vol. 45, No. 5, pp. 538–548. Riggs, N. R.; Stewart, J. H.; Haxel, G. B.; and Conway, C.M., 1993, Sources of Chinle Formation volcanic detritus: Where and how big were they: Geological Society America Abstracts Programs, Vol. 25, No. 5, p. 89. Rogers, J. D., 1982, The Genesis, Properties and Significance of Fracturing in Colorado Plateau Sandstones: Ph.D. Dissertation, University of California, Berkeley, 706 p. Rogers, J. D. and Beckmann, N., 2003, West Lost Creek Trail Sturzstrom: A composite landslide. In Boyer, P.; Rogers, P.; and Santi, P. (Editors), Engineering Geology in Colorado: Contributions, Trends and Case Histories: Association of Engineering Geologists Special Publication No. 15/Colorado Geological Survey Special Publication No. 55, 20 p. Rogers, J. D. and Pyles, M. R., 1980, Evidence of cataclysmic erosional events in the Grand Canyon of the Colorado River, Arizona: Proceedings of the Second Conference on Research in the National Parks, San Francisco: National Park Service, Washington, D.C.: Physical Sciences, Vol. 5, pp. 392–454. Rogers, J. D.; Watkins, C. M.; and Oboh-Ikuenobe, F. E., 2004, Analysis and dating of composite bedrock landslides in the Grand Canyon: Programs with Abstracts, 47th Annual Meeting, Association of Engineering Geologists: Dearborn, MI, September, pp. 72. Skerner, N. A., 1989, Landslides and Human Lives, Translation of Heim, A. (1932), pp. 126–136. Stewart, J. H.; Poole, F. G.; and Wilson, R. F., 1972, Stratigraphy and Origin of the Triassic Moenkopi Formation and Related Strata in the Colorado Plateau Region: U.S. Geological Survey Professional Paper 691, U.S. Government Printing Office, Washington, D.C., 195 p. Strahler, A. N., 1940, Landslides of the Echo and Vermillion Cliffs, Northern Arizona: Journal Geomorphology, Vol. 3, pp. 285–301. Tiwari, B. and Marui, H., 2005, A new method for the correlation of residual shear strength of the soil with mineralogical composition: Journal Geotechnical Geoenvironmental Engineering, Vol. 131, No. 9, pp. 1139–1150. Trollope, D. H., 1969, The Mechanics of Discontinua or Clastic Mechanics in Rock problems. In Stagg, K. G. and Zienkiewicz, O. C. (Editors), Rock Mechanics in Engineering Practice: John Wiley & Sons, London, U.K., pp. 275–320. Trollope, D. H., 1973, Sequential failure in strain-softening soils. In Proceedings of the 8th International Conference on Soil

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

191

Watkins and Rogers Mechanics and Foundation Engineering, Moscow: Vol. 2, Pt. 2, pp. 227–232. Watkins, C. M.; Rogers, J. D.; and Oboh-Ikuenobe, F., 2007, Use of Palynology to Complement the Analysis of Megalandslides in the Colorado Plateau: First North American Landslide Conference, Vail, CO: Association of Environmental and Engineering Geologists Special Publication No. 23, pp. 810–819.

192

Wieczorek, G. F. and Schuster, R. L., 1995, The 1995 Landslide in Zion Canyon, Zion National Park, Utah, U.S.A.: Landslide News, No. 9, pp. 28–31. Willis, G. C., Biek, R. F., Young, R. A. and Spamer, E. E., 2001, Quaternary incision rates of the Colorado River and major tributaries in the Colorado Plateau, Utah. Colorado River: Origin and Evolution: Grand Canyon Association Monograph, 12, pp. 125–127.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 173–192

Characterization and Analysis of the Cedar Pass Landslide Complex, Badlands National Park KYLE C. RADACH* Colorado Geological Survey, 1801 Moly Road, Golden, CO 80401

PAUL M. SANTI Colorado School of Mines, 1500 Illinois Street, Golden, CO 80401

Key Terms: slope stability, paleo-landslide, landslide complex, residual, mapping ABSTRACT The Cedar Pass Landslide Complex is located in Badlands National Park, South Dakota, and has created the need for regular maintenance and repair of Badlands Loop Road (South Dakota State Highway 240), the main highway traversing the park. Although there have been previous studies done in small portions of the complex, there has not been a comprehensive study to evaluate the interactions between smaller landslides and the sensitivity of the landslides within the complex to changing conditions. This study used extensive field mapping and slope stability modeling to delineate the boundaries of landslides within the complex, assess the stability, and investigate the sensitivity to fluctuations in groundwater, reduction in material strength, and erosion within the landslide mass. Results show that highway surface damage in the complex is related to the interaction of movement in both smaller and larger landslides. For instance, damage to the Cliff Shelf parking lot is related to small, destabilized blocks above the head scarp of the larger Prairie Island Landslide located to the southeast. In the Upper and Lower Wedge areas, previously mapped landslides were not confirmed, but highway damage may relate to settlement and erosion of an embankment fill and continued deformation of the massive Cliff Shelf paleo-landslide, which was thought to be dormant until the late 1990s. The overall slow and episodic movement of the landslides observed over the past 30 years may be attributed to dilatant strengthening.

INTRODUCTION AND SCOPE The Cedar Pass Landslide Complex (CPLC) located in Badlands National Park, South Dakota, has *Corresponding author email: kcradach@gmail.com

been the focus of many geotechnical investigations over the previous three decades due to the persistent infrastructure damage caused by slope instability in this region (Figure 1). A majority of the damage involves South Dakota State Route 240 (SR-240). SR-240, also known as Badlands Loop Road, is the main artery for local, commercial, and tourist traffic through the northern unit of the park. Slope movements at multiple locations along the highway within the complex have created a financial burden for the National Park Service (NPS), which is responsible for maintaining this portion of the highway through the park. Past geotechnical studies have provided general comments about failure mechanisms, landslide sensitivity, and approximate landslide boundaries, but have not provided a global analysis of stability or an integrated analysis of these issues. Because of the high rates of erosion that characterize Badlands National Park, the extents of the landsides are difficult to distinguish and many features traditionally attributed to landslides, such as tension cracks and localized slumps, may actually be piping and erosion features. Furthermore, as with many landslide complexes, the complicated interactions between individual landslides must be understood before an effective mitigation program can be implemented. The main objective of this study was to provide a comprehensive characterization of the landslides in the complex in order to better understand their interrelationships and behavior. Specifically, the research sought to answer the following questions, which reflect the unique geology of the Badlands: 1. Landslide Characteristics: What are the boundaries of the various landslides within the complex, how do they interact, and, in particular, how can the eroded and buried toe area be reasonably delineated and modeled? Are any of the features that are currently mapped as landslides actually the result of other processes? What are the effects of continued and rapid erosion on landslide stability?

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

193

Radach and Santi

Figure 1. The boundary of the Cedar Pass Landslide Complex and the extent of other identiﬁed landslides within the complex as well as estimated directions of movement adapted from NPS (2016) based on observations made in the ﬁeld.

2. Landslide Triggering: What geologic or climatic factors control instability in this area? It is hypothesized that landslide movement is driven by periodic ﬂuctuations of groundwater and the associated dilatant strengthening that is induced during movement. Additionally, movement is sensitive to topographic changes in the landslide mass from continued and rapid erosion, especially near the toe. 3. Landslide Mitigation: What is the potential effectiveness of diﬀerent mitigation techniques such as improved drainage or the construction of earthen berms or retaining walls in stabilizing untreated landslides within the complex? This question is not addressed in this paper, but is mentioned here as it was one of the tasks completed for this study. Details can be found in Radach (2020).

194

BACKGROUND Badlands National Park and Badlands Loop Road (SR-240) Badlands National Park is renowned for its colorful cliﬀs of horizontal rock strata, dramatic spires and rugged topography, and the largest assemblage of known late Eocene and Oligocene mammal fossils (NPS, 2020). This National Park, located in southwestern South Dakota on the Great Plains and covering approximately 10,350 km2 , is the largest area of badlands topography in the world (Darton, 1921; Smith, 1958) (Figure 2). In 1939, President Franklin D. Roosevelt oﬃcially designated the area as a National Monument. The monument was enlarged and given National Park status in 1978. Badlands Loop Road, which traverses the CPLC, crosses four named passes that intersect the most

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

Figure 2. The location of Badlands National Park in southwest South Dakota. Major river names are in blue.

dominant geomorphic feature in the northern part of the Park known as the Badlands Wall. This east–westtrending escarpment extends for over 100 km and is the divide between the lower prairie along the White River to the south and undissected upland (upper prairie) to the north. The first pass encountered when driving west along Badlands Loop Road from the Northeast Entrance is Cedar Pass. The road utilizes the gentler slopes created by two large paleo-landslides that define the boundaries of the CPLC (Figure 3). The section of highway that travels across the landslide complex is approximately 1.3 km long. This section of road is vital to the park as it is the most heavily used segment, with an estimated 75 percent of all visitors traveling along this section of highway as they enter the park (Anderson et al., 2004; FHWA, 2013). In addition, this portion of highway also provides a crucial route for local and commercial traffic traveling from the north side of the park and Interstate 90 to the town of Interior and the Pine Ridge Reservation located to the south. The road through Cedar Pass was first constructed in 1935, although it was not until 1957 that the highway was graded and paved. Since that time, landslide movements have been consistently observed along the highway in the CPLC (Kumar & Associates, 1999). In 1958, a scenic overlook was constructed immediately south of Cedar Pass on an embankment fill on the downslope side of the road. The scenic overlook was eventually closed in 1993 because of damage from slope movement (Kumar & Associates, 1999). A geotechnical report prepared by Parsons Brinckerhoff Quade & Douglas, Inc. in 2004 states that the highway had to be resurfaced in the summer of 1967 after the road surface across the slump block of the

CPLC dropped 15 cm. By the late 1990s, the road surface across the landslide at the Cedar Pass summit was kept as gravel. The decision to leave the highway with a soft surface was made after damage from landslide movement to at least two previous road surface overlays (Kumar & Associates, 1999). In 2000, an 80,000 m3 buttress was constructed on the slope below the old overlook and along a short portion of the highway to arrest slope movement in that area. The buttress appeared to have sufficiently slowed or stopped movement as an interferometric synthetic aperture radar (InSAR) survey conducted in the late 1990s to the early 2000s showed no appreciable movement after the construction of the buttress (Anderson et al., 2004). Maintenance and construction work shifted to the Cliff Shelf area during the next 15 years, specifically to the portion of the highway directly west of the Cliff Shelf parking lot that was experiencing slope movement and settlement. Multiple patching and resurfacing projects culminated in the construction of another earthen buttress and new stormwater collection and conveyance system in 2015. The stormwater system was designed to collect all surface water on slopes draining toward the road above the Cliff Shelf parking lot as well as much of the highway and parking lot runoff and transport it to an outlet at the base of the slope below the buttress. The earthen buttress was built to stabilize the failing slope below the highway and to protect the road from future movement. In addition to the buttress, a portion of the highway was reconstructed with a deep patch to provide a stable base for the highway and to assist in the drainage of groundwater beneath the road. A deep patch is a design method for repairing a subsiding section of roadway. It is constructed by excavating the upper meter or two of material under the subsiding section of roadway and then replacing the material with layers of compacted backfill reinforced with geosynthetic material. Ongoing road surface distress in the vicinity of the buttress has become the focus of investigations in the past few years. Most recently, crack sealing and asphalt grinding to smooth bumps in the asphalt have been the primary maintenance operations. Physiography and Climate Badlands National Park sits within the Great Plains geographic province, which is dominated by low-relief topography and extends from the eastern slope of the Rocky Mountains to the eastern sides of North Dakota on down into northwest Texas (Kiver and Harris, 1999). Temperatures in Badlands National Park can range from 40°C in the summer to −40°C

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

195

Radach and Santi

Figure 3. The approximate boundary of the Cedar Pass Landslide Complex, the Cedar Pass paleo-landslide, and the much larger Cliﬀ Shelf paleo-landslide. Basemap imagery is georeferenced Google Earth imagery from 2016. Inset shows the approximate location of the complex within the park boundaries.

in the winter (NPS, 2020). Fifty years (1970–2019) of precipitation data recorded at the Ben Reifel Visitor Center and compiled by the National Oceanic and Atmospheric Administration, National Centers for Environmental Information show the park receives an average of 450 mm of precipitation annually. The wettest time of year is typically late spring through the summer months. On average, about 70 percent of the annual precipitation can fall between April and August. Rainfall events can be long or short duration, and convective type events are common throughout the late spring and summer months. These storms are capable of producing tens of millimeters of rain in the matter of a few hours. Geologic Setting Badlands National Park is named for the badland topography that dominates the landscape in this por-

196

tion of southwestern South Dakota. “Badlands” refers to a heavily dissected and channelized landscape created as the result of erosion in poorly consolidated sediments (Stoffer, 2003). The erosion can occur due to rain, surface water flow, or groundwater flow, and it can create a large system of intricate channels, ravines, and piping features such as underground conduits and sinkholes. The oldest geologic unit exposed in Badlands National Park is the Cretaceous age Pierre Shale, which was deposited approximately 75 million years ago when the Western Interior Seaway covered a large portion of what is now the Great Plains. The Pierre Shale is not exposed in the vicinity of the CPLC, and outcrops of it are generally only found in gullies many kilometers to the west (Kiver and Harris, 1999; Benton et al., 2015). Overlying the Pierre Shale are the thin, silty shales and fine-grained sandstones of the Fox Hills For-

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

mation, deposited within the delta of a river system flowing into the Western Interior Seaway. There was continuous deposition between the Pierre Shale and Fox Hills Formation, with deposition of the Fox Hills Formation persisting in the Badlands until approximately 67 million years ago (Benton et al., 2015). The majority of the exposed rocks in Badlands National Park are part of the White River Group, which unconformably overlies the Fox Hills Formation. The White River Group consists of the Eocene Chamberlain Pass Formation, the Eocene Chadron Formation, the Oligocene Brule Formation, and the Oligocene Sharps Formation. Additionally, a thick layer of ash known as the Rockyford Ash lies between the Brule and the Sharps Formation. White River Group rocks are made up of sediments eroded from the ancient core of the Black Hills (Stoffer, 2003) and volcanic ash and dust (Evanoff et al., 2010) transported to the area by rivers and wind. Due to the volcanic origin of many of the sediments in the White River Group, the rocks and soils contain abundant amounts of smectite clays (Van Houten, 1953; Benton et al., 2015). The landslide material within the CPLC is stratigraphically (from bottom to top) derived from the upper 20 m of the Upper Scenic Member of the Brule Formation and the lower 52 m of the Lower Poleslide Member of the Brule Formation (Starck, 2017). Benton et al. (2015) describe the Upper Scenic as dominated by grey to brown mudstone beds and the Lower Poleslide as dominated by massive, thick siltstone beds with thinner sandstone beds and occasional thin intervals of red to red-brown mudstone. The stratigraphy within the park is generally horizontal with a regional dip of about 1–2 degrees to the southeast (Smith, 1958). Stoffer (2003) states that there are a few faults present but they generally only show offsets on the order of a few meters. A geotechnical report produced by the Central Federal Lands Highway Division of the Federal Highway Administration (FHWA) in 2013 regarding a landslide in the Cedar Pass area states that, “there are no Quaternary faults mapped within the general vicinity of the project site.” Continental glaciation during the Pleistocene did not reach Badlands National Park, and during the last glacial maximum between 14,000 and 25,000 years ago, the Laurentide ice sheet was 240 km to the east near the present position of the Missouri River (Benton et al., 2015). During the late Pleistocene, the climate in the Badlands was dry and very windy as winds out of the northwest traveled along the margins of the ice sheet (Benton et al., 2015). These winds contributed to a significant amount of erosion as evidenced by

the extensive loess deposits located to the southeast in Nebraska. Landslides Within Badlands National Park Slope instability has been a persistent problem in many locations around Badlands National Park. Movement within the CPLC was noted 100 years ago by Wanless (1920), who observed at least two small lakes located in the Cedar Pass area and noted that they were likely formed by landslides that had disrupted drainage networks. Another area of noted instability is along Norbeck Ridge, located approximately 10 km west of Cedar Pass along the highway. A study conducted from 2010 to 2011 identified one of the landslides along Norbeck Ridge as a rotational slump with only periodic movement (Baldauf and Burkhart, 2011). In general, the typical mode of failure within the badlands of western South and North Dakota is slump or earthflow (Trimble, 1979; Gonzalez, 2010); however, a translational failure mechanism has also been proposed for the CPLC (Kumar & Associates, 1998, 1999; Anderson et al., 2004). Overview of Previous Geotechnical Studies The CPLC consists of two paleo-landslides identified as the Cedar Pass Landslide, also called the Bowl Landslide in post-2016 maps, and the larger Cliff Shelf Landslide (Kumar & Associates, 1998) (see Figure 3). The exact date of these larger paleo-landslides is unknown, however, Parsons Brinckerhoff Quade & Douglas, Inc. (2004) states that they occurred, “several hundred years ago,” whereas interpretative signs along the Cliff Shelf Trail located in the landslide complex suggest the slumps occurred on the order of thousands of years ago. It is likely these landslides happened as a result of instability along the Badlands Wall caused by rapid northward erosion by the White River and smaller creeks. Kumar & Associates (1998) describe the Cedar Pass paleo-landslide as consisting mainly of bedrock blocks and debris. This interpretation is supported by the presence of bedrock blocks still containing defined bedding planes observed within the landslide mass as well as a large (5–6 m wide) block with bedding rotated out of the horizontal sitting directly below the head scarp just south of the Cedar Pass summit. The uppermost failure point of the Cedar Pass paleo-landslide could be the N 65 W to N 80 W striking, near-vertical joints along Millard Ridge that provide a discontinuity on which bedrock blocks are able to separate from intact portions of the Millard Ridge (Kumar & Associates, 1998). Focusing on the reactivation within the Cedar Pass paleo-landslide near the summit, and based on

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

197

Radach and Santi

exploration of the subsurface of the slide, Kumar & Associates (1999) concluded that the shear plane is likely 15–20 m below the surface with a dip of approximately 2 degrees to the south–southwest, possibly following the low angle dip of bedding within the underlying Brule Formation. As a result, it is thought that movement of the Cedar Pass paleo-landslide is mainly translational. Observations of groundwater within the Cedar Pass paleo-landslide range from approximately 4 m below the surface to greater than 25 m, with the shallower depths located in uphill portions of the slide near the highway and Millard Ridge (Kumar & Associates, 1999). These levels mean groundwater is located anywhere from 0–15 m above the failure surface of the landslide. Variations in groundwater levels have been attributed to seasonal fluctuations in precipitation, including rainfall, snowmelt, and local drainage patterns. Perched zones may also exist (Yeh and Associates, 2016). The much larger Cliff Shelf paleo-landslide was thought to be dormant through the late 1990s; however, field observations by engineers and sub-surface exploration and monument surveys in 1998 and 1999 noted reactivation of the slide mass and localized areas of settlement in the highway, possibly related to largescale landslide movement. The sliding mechanism of this landslide is thought to be complex due to its large size, and as of 2000, geotechnical investigations were unable to establish the depth of the failure surface or the sliding mechanism (Kumar & Associates, 2000). Subsequent investigations of the Cedar Pass and Cliff Shelf paleo-landslides in the late 1990s and early 2000s confirmed that translational movement was likely the predominant movement type across the entire complex (Anderson et al., 2004). In the late 2000s and early 2010s, attention shifted to the Cliff Shelf paleo-landslide because of the onset of highway surface distress in the area and other observations of movement based on ground and air surveys. Several investigations between 2010 and 2016 attempted to identify the failure surfaces of several suspected smaller landslides within the Cliff Shelf paleolandslide. Investigations included borings (FHWA, 2013; Yeh and Associates, 2016) and geophysical methods (Zonge International, 2013). At a location west of the Cliff Shelf parking lot, two failure surfaces were identified at approximately 10 and 15 m below the ground surface. These depths are slightly shallower than the estimated slide plane of the Cedar Pass paleo-landslide. A significant slope failure in 2013 at the same location was mitigated with another earthen buttress. The consensus among investigators has been that periods of above normal precipitation increase

198

groundwater levels and the saturation of highly plastic clay and claystone layers and are the major cause of landslide movement in the CPLC (Kumar & Associates, 1999, 2000; Anderson et al., 2004; and FHWA, 2012, 2013). Field observations, as well as monitoring data, show increased movement during the months following wetter than normal conditions. Specifically, this trend is noted in 1998 and 2011, when more significant movement of landslides in the area was directly preceded by several years of either normal or above normal precipitation (Figure 4). Soil Strength Conditions at CPLC In cases where soils experience relatively large amounts of shear strain, such as in reactivated landslides like the CPLC, the residual shear condition may be reached. Several factors can influence the residual strength behavior as noted in Abramson et al. (2002), including clay mineralogy, clay content of cohesive soils, and the proportion of platy particles versus spherical particles (i.e., the proportion of clay minerals versus that of non-clay minerals with diameters less than 2 microns). The use of residual strength for modeling soils in the landslides at the CPLC is considered appropriate because the landslides have been moving episodically for at least the last 5 years and because some of the landslides may be failing along reactivated surfaces within larger paleo-landslides. Although measuring shear strength directly is preferred, it is not always possible to collect samples from the exact shear surface, and samples may not be representative of the behavior of the entire unit. Therefore, methods, including those proposed by Stark and Eid (1994), Wesley (2003), Stark et al. (2005), and Stark and Hussain (2013), have been developed to correlate residual friction angle with other, more easily measured material index properties such as Atterberg limits. The first of these methods was used by Kumar & Associates (1999) during their investigation of the Cedar Pass paleo-landslide to estimate the residual friction angle of the claystone unit expected to contain the failure surface. Stark and Eid (1994) tested 32 different clays and clayshales and found that the drained residual failure envelope is non-linear and is controlled by the liquid limit, as an indicator of clay mineralogy; the clay fraction, representing the abundance of fine particles; and the effective normal stress, which influences the interaction between clay particles. Stark and Hussain (2013) modified this work to conduct more laboratory testing to increase the number of data points and to produce trend line equations to enable direct calculation of the drained residual friction angle. It should be noted that correlations of shear strength with

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

Figure 4. Annual precipitation from 1980 and 2012 recorded at the Ben Reifel Visitor Center based on data from the National Oceanic and Atmospheric Administration, National Centers for Environmental Information. Years when more signiﬁcant landslide movement was observed are highlighted by black boxes.

index properties may contain uncertainties, as noted in Hamel (2004), and, therefore, the values calculated using the correlation described above are only used as estimates.

Upper Scenic Member of the Brule Formation and the siltstone of the Lower Poleslide Member of the Brule Formation, as shown on Table 1. Mapping included identifying landslide-induced geomorphic features as well as other features that may

METHODS Field Reconnaissance Fieldwork consisted of landslide mapping, topographic profile construction, and soil sampling. Collected samples were used for laboratory tests to estimate shear strength, grain-size distribution, and plastic and liquid limits. Seventeen soil samples were collected from within the landslide complex at the surface using a 63.5-mm-diameter, 152- mm-long brass sample tube. Fourteen samples were extracted from four different geologic units, two samples were collected from an earthen buttress and one was collected from a highway embankment fill. Samples were collected from units expected to contain the sliding surfaces based on the stratigraphy observed in intact buttes adjacent to the landslides. Care was taken to sample a variety of different geologic materials to account for the fact that failure surfaces may pass through multiple units. These units include the mudstones/claystones of the

Table 1. Collected samples and the geologic units or materials from which they were extracted. Sample Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Geologic Unit/Material Nodular mudstone 2000 buttress Nodular mudstone Upper Scenic mudstone/claystone

Lower Poleslide siltstone

Poleslide siltstone Highway embankment

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

199

Radach and Santi

contribute to slope movement. Those features included scarps, tension cracks, areas of hummocky topography, areas of standing water or where water could accumulate, and areas of erosion and drainage channels. Additionally, damage to park infrastructure, including asphalt cracking, pavement offset, broken curbs, and other signs of deformation, was recorded. The locations of springs and seeps were also noted to help identify areas where groundwater was reaching the surface. Mapping began with a review of historical imagery available from Google Earth. Additionally, the Park Service provided a set of 1:2,300 high-resolution aerial photographs taken in 2003 that were reviewed in stereo to help focus mapping efforts and to identify prominent landslide features. An important slope stability model input is the topography along a cross section of the landslide that is oriented parallel to the predicted direction of movement. A total of 14 different topographic profiles were collected in the field in different orientations so that various directions of slope movement could be analyzed if necessary: The locations of the six profiles that were eventually analyzed are shown in Figure 5. Topographic profiles were surveyed using a slope profiler to collect slope angles over 0.9 m intervals. The slope profiler used in this study was a 0.9-m-long wood board connected to two legs of equal length. Profiles were collected in as straight of a line as possible. If an obstruction such as a tree was encountered, the slope directly adjacent and parallel to the profile was measured until the obstruction was bypassed. Some profiles extended into highly eroded areas (badlands topography) that were too difficult to access, so a 1/3 arc-second (approximately 10 meter) digital elevation model (DEM) was employed to create those sections of the profile. Because badlands are dominated by deep, narrow channels and overall rugged terrain, an attempt was made to record the average elevation profile in areas of badlands topography. Laboratory Testing A series of tests was conducted on soil specimens collected in the field in order to characterize the physical and strength properties of the slide materials (extensive testing details are included in Radach, 2020). Density/unit weight (ASTM D7263-09; ASTM, 2018) tests were performed to provide input data for numerical modeling. Atterberg limits (ASTM D431805; ASTM, 2005) and grain-size distribution (ASTM D422-63; ASTM, 2007) were measured to estimate the friction angle using an empirical correlation (described previously) and to compare these values to those measured by direct shear testing. The natural moisture content of each sample was not recorded because, in

200

most cases, the sampling procedure included the addition of water. For direct shear testing, ASTM D3080-90 (ASTM, 1990) guidelines were followed as much as was reasonable, but with some specific differences, mainly the shearing rate, which was set at 0.5 mm/min. This is approximately an order of magnitude faster than the rate suggested by ASTM. The faster rate was chosen for this study because it was shown to produce reasonable results in Walker (1999), who also tested clayey soil and rock units for slope stability analysis. Furthermore, this movement rate is expected to be similar to the irregular movements of the landslides at the site observed over time. Cylindrical samples (height of 25 mm and diameter of 63 mm) were remolded to their plastic limit inside the shear device in three to five lifts. The test was run to high strain levels to evaluate residual strengths, by running the sample forward and backward for five cycles. Total displacement averaged 190 mm but varied depending on the sample. Samples were tested at three different normal effective stress levels, approximately 90, 600, and 1200 kPa, to simulate an average overburden thickness of 5, 40, and 75 m, respectively. The testing procedure described previously includes several inherent assumptions, including that remolded samples can be reasonably recompacted to their in situ density, and any material lost during shearing (i.e., the volume of soil that squeezes out of the box during testing) does not have a significant impact on the measured strength. A specific weakness of this approach is that the sample is sheared in two directions instead of a single direction: Shearing in a single direction is normally preferred to reach residual conditions. Numerical Modeling of Landslides in the Complex In order to assess the current stability and sensitivity to various input parameters, computer modeling was conducted using two-dimensional limit equilibrium methods with RocScience Slide (RocScience, 2018). Based on the assumed planar shape of the failure surface for the different landslides, the Spencer’s, general limit equilibrium (GLE)/Morgenstern-Price, and corrected Janbu methods were used. For the sensitivity analysis, the GLE/Morgenstern-Price method was used because it consistently provided the lowest and most conservative factor of safety of the three methods. For smaller landslides, where the failure surface was assumed to be circular, the Bishop method was used. Circular surfaces were assumed for the smaller landslides because observations in the field showed some rotational component to slope movement. Topography for each analyzed profile was created as described previously, and initial water table elevation

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

Figure 5. Location of topographic proﬁles (blue lines) used for slope stability modeling. Solid polygon outlines show landslides mapped in this study. Dashed polygon outlines are approximate boundaries of landslides mapped by the Park Service that could not be identiﬁed in this research. Contour lines are in 10-m intervals. The basemap is a slope function draped over a 10-m DEM.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

201

Radach and Santi

was estimated based on a combination of field observations and borehole data (Kumar & Associates, 1999; FHWA, 2013; and Yeh and Associates, 2016). Initial Mohr-Coulomb strength parameters were estimated based on the results of laboratory tests carried out on samples collected for this study and values found in the literature, as well as the empirical correlations described previously. Calibration was completed on the large landslide located south of the Cliff Shelf Trail, identified in this study as the Prairie Island Landslide. Three profiles for the Prairie Island Landslide (Profile D, E, and F on Figure 5) were used to iteratively adjust slope stability model input parameters until they converged on specific material strengths (unit weight, cohesion, friction angle), failure surface geometry, and water table depth, a methodology demonstrated by Santi (2014) and Scheevel (2017) by using the following steps: 1. Develop a range of piezometric surfaces using available data. 2. Bracket strength values (cohesion and friction angle) and unit weight from laboratory and literature sources. 3. Calculate the factor of safety from the mid-range values, iterate with the largest bracket, and then narrow next largest bracket and so on. Iteration was completed when the factor of safety across all three profiles was within the range of 0.9 to 1.1 and all the input parameters across the three profiles were equal. A range of factor of safety values was used because differences in topography kept some sections from completely converging. Back-calculated strength values from the Prairie Island Landslide were then used in the models for the three other landslides because of the similarities between soils present at each location. Additional modeling details are included in Radach (2020). A sensitivity analysis was carried out to evaluate the impact to the overall stability of changes in material properties, fluctuation of groundwater, and erosion within and near the toe by varying these properties individually across a wide range. The sensitivity of the landslides to erosion at the toe was simulated by incrementally lowering the topography in the areas with little to no vegetation, which usually coincided with heavily channelized areas near the toe of the landslide. The topography was lowered by increments up to a total of 60 cm. As estimated average erosion rates within the park are about 2 cm/yr (Stoffer, 2003; Benton et al., 2015; and NPS, 2020), the simulated erosion equated to between 0.5 years and 24 years of erosion. For the landslides investigated in this study, high rates of erosion meant that landslide boundaries were difficult to distinguish, even in the field. Therefore,

202

great care was taken in selecting the head, toe, and location of failure surface to provide the most accurate and consistent models. Modeling software limits for the head and toe exit points of the failure surface were set at the narrowest width defined by field mapping, then further narrowed by iterative modeling. Available inclinometer and borehole data from the complex (Kumar & Associates, 1999; FHWA, 2012, 2013; and Monarco, 2018) helped constrain the minimum depth at which a failure surface was located and decreased the total number of possible model solutions. The failure surface geometry and head and toe for other landslides were adjusted incrementally in order to define a reasonable failure surface. Multiple circular and non-circular surfaces were modeled but an emphasis was placed on failure surfaces with a similar geometry to the Cedar Pass paleo-landslide, which has a basal shear surface of only a few degrees and a rear sliding surface of about 60 degrees because of the suspected translational behavior of the landslides in the complex. RESULTS Field Reconnaissance Observations and Interpretations Fieldwork was conducted in December 2017, May– August 2018, and April 2019. Details of field observations are included in Radach (2020), and summaries for each landslide, as well as maps of various features, are presented below. Cliff Shelf Trail Area The Cliff Shelf area defined by this research encompasses the area to the east of the upper switchback of Badlands Loop Road and the Cliff Shelf parking lot (see Figure 3). Landslide features include settlement and cracking of the southeast side of the parking lot and adjacent sidewalk, vertical scarps at the head of the landslide, internal scarps, tension cracks, and generally hummocky topography, which are shown in Figure 6. Field observations indicate that movement of the Prairie Island Landslide is destabilizing smaller blocks above the main landslide mass (the Trail Block and Parking Lot Block in Figure 6), and movement of these smaller blocks is causing the observed damage to the parking lot and trail. The toe of the Prairie Island Landslide could not be identified in aerial photographs or in the field, so it was assumed to be located at the base of the slope, a common assumption made during landslide investigations (Schulz, 2004). Movement of the Prairie Island Landslide appears to be mainly translational as indicated by many

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

Figure 6. The boundaries of landslides in the vicinity of the Cliﬀ Shelf parking lot identiﬁed by mapping landslide-induced geomorphic features. Arrows indicate the direction of movement. Basemap credit: 2016 Google Earth image.

linear ridges and grabens near the top of the landslide. These observations are consistent with an InSAR survey carried out in 1999–2002 by Anderson et al. (2004) that showed very few extensive areas of uplift and subsidence that could have indicated rotational landslide movement. This area also contains the smaller Lateral Shear Landslide located southwest of the Cliff Shelf parking lot entrance. This landslide is characterized by several areas of pavement cracking and offset as well as multiple scarps and tension cracks located within close proximity to the highway. Upper and Lower Wedge Landslides The major concern in the Upper Wedge area is recurring cracking and offset in the highway surface

(Figure 7). An overview of aerial photographs and Google Earth historical imagery shows the cracks developed in the adjacent slope sometime between 2003 and 2011 with highway damage occurring between 2011 and 2012. However, the Upper Wedge area has been subjected to slope stability issues for the last 100 years, as Wanless (1920) observed a landslidedammed pond directly west of the present-day highway embankment. Presently, water accumulates in a catchment basin behind a concrete buttress on the east side of the road, and this may increase groundwater levels in the embankment. This structure was built in 2015 to collect surface water as part of the new stormwater system and was lined between August 2018 and April 2019 to prevent inﬁltration. Our interpretation of the Upper Wedge area is that the recurring oﬀset in the highway is caused by

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

203

Radach and Santi

Shelf paleo-landslide, as the slope is located near the boundary of that landslide. The head scarp of the Lower Wedge Landslide was not observed during mapping. Landslide-induced geomorphic features, including minor scarps and tension cracks, were generally restricted to the north side of the slide area. Additionally, there is no obvious head scarp near the highway. The highway is experiencing significant surface distress at that location with a short section of multiple bumps, dips, and cracks in the pavement. However, deformation in the highway is not always laterally continuous and all features do not extend onto either shoulder. Damage to the roadway in this area, like the Upper Wedge, could be due to issues in the highway fill or ongoing deformation near the boundary of the Cliff Shelf paleo-landslide. Laboratory Testing

Figure 7. The north slope of the Upper Wedge area. The slope in the foreground is moving to the right (south) and was mapped as the Upper Wedge Landslide A in this study. The truck is passing across the highway embankment that is experiencing some deformation. Millard Ridge makes up the cliﬀs in the background. The blue line indicates the approximate location of the head scarp of the Upper Wedge Landslide A.

settlement in the embankment fill on which the highway is constructed, possibly related to movement in the natural slope adjacent to the highway (“Upper Wedge Landslide A” in Figure 8). The highway surface drops as it transitions onto fill from the north and then has some slight undulations before it transitions off fill to the south. Settlement of fill has been an issue in other parts of the park, including the inclines to Norbeck Pass and Dillon Pass to the west. Settlement has been attributed to saturation and swelling of clay embankment soils from changes in moisture content after construction (FHWA, 1999, 2011), piping of fines, and differential frost heave (Parsons Brinckerhoff Quade & Douglas, 2004). Therefore, the boundaries of the Upper Wedge Landslide as mapped by the Park Service have been revised to only include the slope in the foreground of Figure 7. This smaller landslide is likely the result of the steepness of the slope and soils, but may also be related to large-scale movement of the Cliff

204

All of the samples tested, excluding the embankment fill, were grouped together for characterization and modeling purposes. This approach was used to provide a range of initial values with the expectation that a single geologic unit would be modeled during the slope stability analysis due to the uncertainty of the stratigraphy in the landslides and the uncertainty in the identities of the geologic units sampled. Consistency testing showed that almost all the samples were classified as high plasticity clay (CH) using the United Soil Classification System. The soils contained 97–99 percent fines (silt and clay sizes) with 36– 47 percent clay sizes, with the exception of the embankment fill sample, which contained only 80 percent fines and 27 percent clay sizes. The liquid limit ranged from 64 to 87 and plasticity index ranged from 35 to 57. These ranges generally agree with values from other studies involving soils in this portion of the park (Kumar & Associates, 1999; FHWA, 2013; Zhang, 2013; and Yeh and Associates, 2016). Direct shear testing to estimate the residual strength properties of the soils was done on samples 7, 9, 11, 14, and 17, and the results show a wide range of values, reflecting the wide range of materials tested. Cohesion ranged from 27 to 37 (average of 33) kPa and the friction angle ranged from 18 to 24 (average of 21) degrees. Whereas Stark and Eid (2005) suggest that the residual strength envelope of clays should have a cohesion value of 0, the rate of landslide movement observed and measured by inclinometers, InSAR, and survey monuments at the CPLC likely at times exceeds the rate required for drained conditions (Kumar & Associates, 2000; FHWA, 2002; and Anderson et al., 2004). Therefore, it is expected that a small value of cohesion accurately models real conditions in the complex. The use of a small value of residual cohesion has been

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

Figure 8. The boundaries of landslides in the Wedge area identiﬁed by mapping landslide-induced geomorphic features. Arrows indicate the direction of movement. Basemap credit: Google Earth, 2016.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

205

Radach and Santi Table 2. Ranges of eﬀective residual cohesion and eﬀective residual friction angle from various sources and this study.

Source This study (direct shear) This study (taking into account multipliers from Vithana et al., 2012) This study (using correlation from Stark and Hussain, 2013) This study (from back analysis) Literature sources* Other Badlands studies**

Eﬀective Residual Cohesion (kPa)

Eﬀective Residual Friction Angle (degrees)

27.0–36.9 8.4–17.6

18.5–23.6 9.7–14.8

—

10.2–13.4

5.0

10.5

0.0–36.2 0.0

5.0–20.9 8.5–13.0

*Stark and Eid, 1994; Baum et al., 1998; Wan and Kwong, 2002; Dewoolkar and Huzjak, 2005; Stark et al., 2005; Tiwari et al., 2005. **Kumar & Associates, 1999.

used for other analyses with similar conditions (Lupini et al., 1981; Skempton, 1985; Tiwari et al., 2005; and Vithana et al., 2012). Furthermore, the contribution of cohesion to the overall stability is very small compared to friction angle values under normal stresses as deep as 45 m for most of the landslides. Overall, friction angle and cohesion values measured in this study are higher than expected based on ranges for these parameters provided by other studies and back-calculated values used for investigation in the park. This may be because a direct shear device, which shears a sample forward and backward, may not accurately model residual conditions. Specifically, Vithana et al. (2012) found that residual values of cohesion and friction angle for clay and mudstone samples obtained using a direct shear device compared to using a ring shear device were 2.1–3.2 and 1.6–1.9 times higher, respectively. If the multipliers calculated by Vithana et al. (2012) are applied to the strength values of the Badlands samples, the resulting “adjusted” values are well within ranges reported by other investigators when testing samples of similar materials (Table 2). We conﬁrmed the range of adjusted friction angles in Table 2 using the empirical relationship developed and revised by Stark and Eid (1994), Stark et al. (2005), and Stark and Hussain (2013). Eq. 1, from Stark and Hussain (2013), is valid for a clay fraction between 25 and 45 percent, a liquid limit between 30 and 130, and an applied normal stress of 700 kPa. ϕ r σ n =700 kPa = 28.05 − 0.2083 (LL) − 8.183 × 10−4 (LL)2 + 9.372 × 10−6 (LL)2 (1)

206

Table 3. Calculated drained residual friction angle values for diﬀerent geologic units within the Cedar Pass Landslide Complex using Eq. 1 from an empirical correlation between drained residual friction angle and liquid limit developed by Stark and Hussain (2013).

Unit 1 2 3 4

Samples

Average Liquid Limit

Average Clay Fraction

Drained Residual Friction Angle (degrees)

1, 2, 5 6–10 11–14 15–16

76.1 66.2 74.7 74.0

40.9% 44.4% 44.7% 44.3%

11.6 13.4 11.8 12.0

Samples 3, 4, and 17 are excluded because they were extracted from compacted ﬁll.

where, ϕr = the drained effective residual friction angle, σn = the effective normal stress for which the equation is applicable, and LL = the liquid limit of the soil. Using Eq. 1 from Stark and Hussain (2013) provided a drained residual friction angle range of 12– 13 degrees (Table 3). This range falls within the range of 8.5–13 degrees estimated by Kumar & Associates (1999) using the graphical form of the same correlation for their samples. Numerical Modeling Prairie Island Landslide The Prairie Island Landslide was modeled as a homogeneous geologic unit because consistent stratigraphy within the landslide could not be identified. Modeling was initiated with a range of material properties and groundwater levels, and the range was narrowed down using the back analysis described previously. The final selected values were an effective cohesion value of 5 kPa, an effective friction angle of 10.5 degrees, and a unit weight of 18.0 kN/m3 . These values are within the range of (1) adjusted residual strength parameters for laboratory tests from Table 2, (2) residual friction angles calculated for similar materials by other researchers (Baum et al., 1998, Kumar & Associates, 1999; Dewoolkar and Huzjak, 2005; and Tiwari et al., 2005), and (3) within the range of values predicted by empirical correlation in Table 3 using Stark and Hussain (2013). The water table depth range of 5–10 m was chosen based on piezometric data provided in past geotechnical studies (Kumar & Associates, 1999; FHWA, 2012). The final calculated critical depth of 5 m matches the highest groundwater level observed in the complex (Kumar & Associates, 1999), but varies slightly within the profiles due to variations in the topography. Figure 9 shows the final adjusted slope model for one of the three analyzed cross sections (E–E’) shown

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

Figure 9. Slope model for the Prairie Island Landslide Profile E. A block search was used to help constrain the shape of the slide plane. The horizontal and vertical scales are in meters. A single material was used to model the slope because the stratigraphy could not be confidently identified in the field and no geotechnical drilling data exist for this landslide. The orange and yellow colors in the failed landslide mass are different failure surfaces produced by various model outcomes. The failure surface with the lowest FS is highlighted by the dark line.

on Figure 5. The upper landslide scarp was tightly constrained by a set of limits based on the location of the head scarp compared to the toe of the landslide, which was allowed more flexibility because it was unclear in the field where the toe was located. Models were constrained to produce basal shear surfaces with a dip of only a few degrees and a rear sliding surface of about 60 degrees to match the profile of the Cedar Pass paleo-landslide by using a block search. The basal shear surface was generally up to a couple of degrees steeper than the regional bedding dip of 1–2 degrees. The factor of safety for this cross section, in the middle of the landslide, is slightly higher. The sensitivity of the Prairie Island Landslide to changes in material properties (unit weight, cohesion, and friction angle), water table fluctuation, and toe erosion were investigated. Among material properties, the highest sensitivity was to variation in friction angle, as expected for the deep failure surface for this landslide. For example, a change in friction angle of 25 percent resulted in a change in factor of safety on the order of 30 percent. To test the sensitivity to fluctuations in groundwater, the water table was lowered in 1 m increments from the ground surface to a depth of 25 m. With the water table at the surface, the factor of safety decreased by 11–20 percent (Figure 10). When the water table was lowered to a depth of 25 m, the factor of safety increased by 38–47 percent. The maximum water table depth is expected to be approximately 25 m based on data from drilling during and after an extended period (several years) of drier than average conditions. Assuming a critical range of factor of safety from 0.9 to 1.1 to account for variability in modeling parameters across the landslide, Figure 10 shows that groundwater levels shallower than 5–15 m could initiate movement in parts of the landslide.

The sensitivity of the Prairie Island Landslide to toe erosion was modeled by lowering topography in the bottom third of the landslide by increments as described previously. All three profiles showed very little change in stability, with a maximum change in factor of safety of slightly more than 1 percent (Figure 11). The unstable blocks located above the head scarp of the Prairie Island Landslide, called the Parking Lot Block and Trail Block (shown in Figure 6), are responsible for the observed damage in the Cliff Shelf parking lot and along the trail. The Trail Block, assumed to be representative of both blocks, was modeled as a shallow rotational failure to simulate retrogression of the head scarp of the Prairie Island Landslide, using the uppermost portion of Profile D, which is closest to the parking lot.

Figure 10. Sensitivity of the Prairie Island Landslide to changes in the groundwater level. Each profile has relatively the same sensitivity to the water table, indicated by the similar shape of each line. Differences in the positions of the lines are a result of each profile having a slightly different base factor of safety. Red shading indicates factor of safety values ࣘ1.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

207

Radach and Santi

Figure 11. Sensitivity of the Prairie Island Landslide to toe erosion. Each profile shows minimal sensitivity to erosion. Differences in the positions of the lines are a result of each profile having a slightly different base factor of safety. Red shading indicates factor of safety values ࣘ1.

The base case produced a factor of safety of 1.02. This value is relatively low because the stability of this block is tied to the movement and stability of the Prairie Island Landslide located on the slope below. Slope movement below the Cliff Shelf Trail and in the parking lot is expected to increase if movement of the Prairie Island Landslide increases or if the water table rises to or above the failure surface of the block. Lateral Shear Landslide The Lateral Shear Landslide is located west and downhill of the Cliff Shelf parking lot (see Figure 6). In the field it was difficult to identify the distinct boundaries of this landslide. However, an inclinometer installed by Yeh and Associates and FHWA in 2016 showed movement at approximately 11 m below the shoulder of the highway (Monarco, 2018). The stratigraphy of the landslide was based on a borehole log from Yeh and Associates (2016), and the model used the same soil properties back-calculated for the Prairie Island Landslide, with the water table approximately 5 m below the highway. Figure 12 shows the slope stability model for the Lateral Shear Landslide, with a factor of safety of 0.997. Sensitivity analysis showed similar trends as the Prairie Island Landslide, with strong controls exerted by strength parameters and groundwater levels, but with little change resulting from toe erosion. Upper Wedge Landslide Field mapping for this study did not locate the boundaries for the Upper Wedge Landslide indicated in earlier mapping by the National Park Service (2016).

208

Figure 12. Slope stability model for the Lateral Shear Landslide. The shape of the water table was based off of piezometer data and observations made in the field. The vertical line passing through the landslide is the approximate location and depth of the inclinometer. The traffic load is placed where the highway intersects the cross section, and it is the same magnitude as the load used by FHWA (2013) along the highway for slope analysis. The horizontal and vertical scales are in meters.

Furthermore, there was no evidence of uplift or a toe bulge that would be expected at the base of a landslide of this size and height. Nevertheless, the feature was still modeled to gauge the current and potential future stability, using profile A1 on Figure 5. Stratigraphy was based on borehole data from Yeh and Associates (2016); natural material properties were the same as those back-calculated from the Prairie Island Landslide; and cohesion, friction angle, and unit weight values for the embankment fill were the same as those used by Kumar & Associates (1999). The water table was modeled at a depth of 6 m below the right shoulder of the highway, which matches the shallowest water level encountered during drilling (Yeh and Associates, 2016) and is considered the highest likely level at this location. Both shallow and deeper circular and non-circular failure surfaces were investigated, but the lowest calculated factors of safety ranged from 1.2 to 1.4, even with a relatively high water table and low material strength properties. The failure surfaces that produced the lowest factor of safety were rotational surfaces extending more than 70 m below the ground surface and with high angle scarp and toe exit points. This failure surface shape was considered unrealistic as it would include a significant rotation component that did not match observations made in the field. Therefore, based on modeling and field mapping, we could not confirm the presence of a landslide as originally mapped by the National Park Service (2016). We also modeled a transect parallel to the highway (A2 on Figure 5) since field mapping indicated scarps and steep slopes in this area (Figure 8). Assuming a circular failure surface, a factor of safety for this smaller

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis Table 4. Conﬁdence level of landslides mapped in the ﬁeld and point values for primary landslide characteristics. Landslide Prairie Island Landslide

Total Lateral Shear Landslide

Landslide Feature

Points

Head scarp Flanks Toe Internal scarps, sag ponds or closed depressions, compression ridges, etc.

10 8 3 9 30 4 6 6 5 21 0 4 1 2 7 1 4 2 3 10 9 7 9 10 35

Head scarp Flanks Toe Internal scarps, sag ponds or closed depressions, compression ridges, etc.

Total Lower Wedge Landslide

Total Upper Wedge (NPS)

Head scarp Flanks Toe Internal scarps, sag ponds or closed depressions, compression ridges, etc. Head scarp Flanks Toe Internal scarps, sag ponds or closed depressions, compression ridges, etc.

Total Upper Wedge (A)

Head scarp Flanks Toe Internal scarps, sag ponds or closed depressions, compression ridges, etc.

Total

landslide was ∼1, indicating marginal stability. Therefore, we concluded that the geomorphic features identified on the slope adjacent to the highway are more likely related to this small, recent landslide oriented parallel to the highway and not caused by the previously mapped larger scale movement perpendicular to the highway. Lower Wedge Landslide As with the Upper Wedge area, the boundaries of this landslide could not be definitively identified in the field, and slope stability analysis for a variety of model geometries produced factors of safety no lower than 1.2 to 1.3. Therefore, we concluded that this feature is not a landslide, but observed road damage and ground cracking may be the result of other causes, such as settlement, swelling of clays in the subgrade, frost heave, or deformation from large-scale slope movement in the Cliff Shelf paleo-landslide. DISCUSSION Confidence of Landslide Identification in the Field Because of the difficulty in identifying specific landslide boundaries in the highly erosive environment of the Badlands, we used a confidence rating system as another method to gauge the likelihood that observed features indicate a landslide. The system developed by

Conﬁdence

High

Moderate

Low

High

the Oregon Department of Geology and Mineral Industries (Burns and Madin, 2009) was used. While this system was designed for LiDAR-based mapping projects, it was adopted for this study based on evidence from field mapping. Each landslide is classified into a confidence category by scoring four main topographic features associated with landslides, with a maximum of 10 points assigned to each feature (Table 4). A score of 10 for a feature means it was easily identifiable, and a score of 0 means the feature cannot be identified. The sum of the scores is used to assign a level of confidence: 0–10 points = low, 11–29 points = moderate, and 30–40 points = high. The Prairie Landslide is one of the most well defined landslides studied, and the confidence rating of this landslide is high (see Table 4). This landslide was previously unmapped, but movement of this landslide is likely responsible for much of the deformation occurring just above the head scarp. The lowest confidence was in the location of the toe because of the highly eroded nature of the lower part of the slope. The Lateral Shear Landslide was mapped with a moderate confidence rating due to the overall lack of readily identifiable surface features, but a shear surface identified by an inclinometer helped constrain some of the boundaries. The cracks and bumps crossing the highway were interpreted as the landslide flanks as some of these features were continuous and extended beyond the edge of the highway.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

209

Radach and Santi

The Lower Wedge Landslide was mapped with an overall low confidence rating. There was no observable head scarp and only a change in slope angle to indicate a toe. There were few tension cracks, scarps, and depressions mapped in this area, and any that were mapped were generally oriented in a direction parallel to any potential downslope movement. The Upper Wedge area was divided into two different landslides. Upper Wedge (NPS) is the landslide mapped by the National Park Service (defined by profile A1 on Figure 5). Movement of this landslide was predicted to be to the southwest, perpendicular to the highway. Upper Wedge (A) is a landslide identified in this study and encompasses only the northern slope of the Upper Wedge area (Figure 8, and profile A2 in Figure 5). The National Park Service (2016) mapped this area as the northern flank of the Upper Wedge Landslide. Movement of the Upper Wedge (A) Landslide is southward and into the valley. A low confidence rating was assigned to Upper Wedge (NPS) because a head scarp and toe could not be identified, nor were there widely distributed internal features. The separation of blocks along nearly vertical joints along Millard Ridge can be attributed to the typical mechanism responsible for producing rockfall in the area and not necessarily the head of a landslide. This process of separation in the cliff was observed in many locations along Millard Ridge above the Cliff Shelf area. The distribution of scarps and tension cracks is limited to the northern slope valley, associated with Upper Wedge (A), which consequently has a high confidence rating. It is interesting to note that the location of most landslide-induced geomorphic features is on the northern side of both the Upper and Lower Wedge areas, closest to Millard Ridge, which in this area is thought to be the head scarp of the Cliff Shelf paleo-landslide. These features may suggest a more southerly direction of movement. It is possible that mapped features and highway damage in these locations are associated with continued or reactivated deformation at and near the head of the large paleo-landslide. Survey data from 1999–2001, collected from monuments installed by park and FHWA personnel showed a relatively consistent southwest direction of movement across much of the western side of the Cliff Shelf paleo-landslide ranging from 1 to 13 mm/week (Kumar & Associates, 2000; FHWA, 2002). Cedar Pass Landslide Complex Landslide Characteristics and Failure Mode Based on a review of previous geotechnical investigations, fieldwork, and computer modeling conducted for this study, the primary mechanism of fail-

210

ure within the CPLC is translational sliding along lowangle shear surfaces. This is supported by observations made in the field for this study as well as borehole data (Kumar & Associates, 1999), monument surveying (Kumar & Associates, 1999, 2000), and an InSAR survey (Anderson et al., 2004). Sliding is possibly occurring on the low dipping surfaces between different geologic units; however, the continuity of these dipping surfaces is unknown. Sliding may also be occurring along stepped surfaces, although there is no evidence to support this, and computer modeling of approximations of a stepped surface produced higher factor of safety values than those of a single, low-angle slide plane. The velocity of the landslides in the complex can be classified as moderate to extremely slow (Cruden and Varnes, 1996) based on measurements that showed a maximum rate of 17.8 mm/d and a minimum rate of 0.14 mm/d (Kumar & Associates, 1999). A review of photographs provided by the Park Service of highway damage and observations made during this study show that movement rates are likely closer to the lower end of this range and possibly zero at times. The semi-continuous movement of landslides in the complex may be attributed to groundwater drainage through cracks or other piping features in the slide mass or capillary tensions in the soil as the water table drops after a precipitation event. Another possible explanation for this type of movement is a mechanical feedback called dilatant strengthening, which has been explained theoretically, tested in a laboratory setting, and observed in the field at other locations (e.g., Iverson, 2005; Schulz et al., 2009). Iverson (2005) and Schulz et al. (2009) describe dilatant strengthening as the process where an increase in groundwater levels from rain or snowmelt increases pore pressures and initiates movement along a shear plane. Shearing can cause soil along the shear plane to dilate, which lowers the pore pressures, therefore increasing the effective stress and shear resistance and ultimately slowing landslide movement. With time, reconsolidation of the sheared soils may then occur until increased pore pressures reactivate landslide movement. This cyclic process depends on the ability of the shear zone to dilate and reconsolidate in cycles and the amount of time it takes for pore pressures to develop, dissipate, and redevelop (Iverson, 2005). Dilatant strengthening has been used to explain movement patterns of landslides in fine-grained materials (Baum and Johnson, 1993; Schulz et al., 2009). We suspect this process controls movement in the CPLC: The area shows semi-continuous slow movement based on field observations and other survey methods, and soil samples from this study showed dilative response during shear testing. However, in order to confirm whether dilatant

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis

Figure 13. Annual precipitation averages divided into 5-year intervals from 1970–2019. A dashed linear trendline through the data is used to show that annual precipitation has increased over the last 50 years.

strengthening is contributing to the control of landslide movement in the CPLC, more laboratory testing (on samples collected from the failure surface) and monitoring of pore pressures in the landslide mass and landslide movement is required. Slope deformation and movement of landslides in the complex is sensitive to groundwater levels, as shown by sensitivity analysis, and appears to be tied to periods of above normal precipitation. Periods of accelerated movement, most notably in the late 1990s to early 2000s and again in the early to mid-2010s were directly preceded by years with above normal precipitation. Most recently, 2018 and 2019 saw higher than average precipitation, with 2019 receiving more precipitation than any year in the last 50 years. Consequently, as of February 2020, highway surface distress, especially in the Cliff Shelf Trail parking lot and above the Lateral Shear Landslide, was more severe than at any point in the previous 5 years. Short-term climate trends based on 50 years of precipitation data show that annual precipitation amounts have generally increased slightly over the last half century (Figure 13). This may explain why instability throughout the complex has become more widespread in that time. It also provides evidence that continued movement of already existing landslides, and the development of new landslides, is possible in the future. CONCLUSIONS The stability and sensitivity to changes in material strength, groundwater fluctuations, and landslide toe erosions in the CPLC was estimated using twodimensional limit equilibrium slope stability models. These models were also used to investigate the effectiveness of different mitigation techniques in im-

proving slope stability. Individual landslide boundaries within the complex were refined through detailed field mapping. Three landslides within the Complex identified by the Park Service (the Lateral Shear Landslide and the Upper Wedge and Lower Wedge Landslides) and the Prairie Island Landslide identified during this study were investigated. Neither field observations nor computer modeling could confirm the existence of the Upper and Lower Wedge Landslides as mapped by the NPS. Highway surface damage in these areas could be explained by deformation of embankment fill from piping, frost heave, or swelling of subgrade clays, or by settlement along the upper boundary of the Cliff Shelf paleolandslide. In the case of the Upper Wedge area, highway surface damage may also be the result of a smaller, previously unidentified landslide adjacent to an embankment fill. The Lateral Shear Landslide caused highway surface damage west of the Cliff Shelf parking lot, and monitoring shows a shear surface located approximately 11 m below the south shoulder of the highway. The Prairie Island Landslide identified in this investigation is the largest landslide that was investigated and encompasses much of the slope below the Cliff Shelf Trail. Movement of this slide is destabilizing blocks above the head scarp, and movement of these blocks is causing the damage to the boardwalk and settlement of the southeastern side of the parking lot. The landslides showed sensitivity to changes in groundwater levels with up to a 47 percent change in the factor of safety when groundwater is raised to seasonally high levels. Undrained conditions in which the rate of pore pressure increase exceeds the rate of pore pressure dissipation in the landslide mass at the failure surface may cause periods of faster movement. The landslides did not show sensitivity to toe erosion over a projected 24-year period, based on erosion rates measured elsewhere in the park. Some samples showed dilative behavior during shearing, so this behavior, coupled with related changes in pore-water pressure, could explain the cyclic movement of these landslides through the process of dilatant strengthening. However, more research is needed in order to eliminate other possible explanations for the cyclical landslide movement such as intermittent and localized drainage through cracks and piping conduits in the slide mass. ACKNOWLEDGMENTS Support for author K. Radach was provided by the Geologist-in-the-Parks program from the U.S. National Park Service and the Geological Society of America. Dr. Rachel Benton and Ellen Starck, of the

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

211

Radach and Santi

National Park Service, provided access to background information, aerial photographs, and a historical perspective of the landslide complex. The views expressed in this paper do not necessarily reﬂect those of the National Park Service. REFERENCES Abramson, L. W.; Lee, T. S.; Sharma, S.; and Boyce, G. M., 2002, Slope Stability and Stabilization Methods: John Wiley & Sons, New York, 712 p. Anderson, S. A.; Surdahl, R.; and Young, B., 2004, InSAR evaluation of landslides and alternative transportation routes. In Yegian, M. K. and Kavazanjian, E. (Editors), Proceedings of Geo-Trans 2004, ASCE, Reston, VA. ASTM D3080-90, 1990, Standard Test Method for Direct Shear Test of Soils Under Consolidated Drained Condition: ASTM International, West Conshohocken, PA, 9 p. ASTM D4318-05, 2005, Standard Test Methods for Liquid Limit, Plastic Limit, and Plasticity Index of Soils: ASTM International, West Conshohocken, PA, 14 p. ASTM D422-63, 2007, Standard Test Method for Particle-size Analysis of Soils: ASTM International, West Conshohocken, PA, 8 p. ASTM D7263-09, 2018, Standard Test Methods for Laboratory Determination of Density (Unit Weight) of Soil Specimens: ASTM International, West Conshohocken, PA, 7 p. Baldauf, P. E. and Burkhart, P. A., 2011, Mass wasting and quaternary landscape development, Badlands National Park, South Dakota. In Abstracts with Programs, Vol. 43, No. 5: Geological Society of America, Boulder, CO. Baum, R. L. and Johnson, A. M., 1993, Steady Movement of Landslides in Fine-Grained Soils – A Model for Sliding Over an Irregular Slip Surface: Landslide Processes in Utah—Observation and Theory, U.S. Geological Survey Bulletin 1842, 28 p. Baum, R. L.; Messerich, J.; and Fleming, R. W., 1998, Surface deformation as a guide to kinematics and three-dimensional shape of slow-moving, clay-rich landslides, Honolulu, Hawaii: Environmental Engineering Geoscience, Vol. 4, No. 3, pp. 283–306. Benton, R. C.; Terry, D. O., Jr.; Evanoff, E.; and McDonald, H. G., 2015, The White River Badlands: Geology and Paleontology: Indiana University Press, Bloomington, IN, 222 p. Burns, W. J. and Madin, I. P., 2009, Protocol for Inventory Mapping of Landslide Deposits from Light Detection and Ranging (LIDAR) Imagery: Oregon Department of Geology and Mineral Industries, Special Report 42, 30 p. Cruden, D. M. and Varnes, D. J., 1996, Landslides 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, Washington, DC, pp. 36–75. Darton, N. H., 1921, Badlands of South Dakota and Nebraska. In Andree, K. (Editor), Geologische Charakterbilder, Heft 25, Berlin, 7 p. Dewoolkar, M. M. and Huzjak, R. J., 2005, Drained residual shear strength of some claystones from Front Range, Colorado: Journal Geotechnical Geoenvironmental Engineering, Vol. 131, No. 12, pp. 1543–1551. Evanoff, E.; Terry, D. O., Jr.; Benton, R. C.; and Minkler, H., 2010, Field guide to geology of the White River Group in the North Unit of Badlands National Park. In Terry, M. P., Duke, E. F. and Tielke, J. A. (Editors), Geologic Field Trips in

212

the Black Hills Region, South Dakota: South Dakota School of Mines and Technology, Bulletin 21, pp. 96–127. Federal Highway Administration (FHWA), 1999, Pavements: Federal Highway Administration, Central Federal Lands Highway Division, 9 p. Federal Highway Administration (FHWA), 2002, South Dakota Badlands National Park Alternative Alignment Study: Federal Highway Administration, Central Federal Lands Highway Division, Lakewood, CO, 34 p. Federal Highway Administration (FHWA), 2011, Badlands National Park Pavement Report, SD: Report No. 11-006, Federal Highway Administration, Central Federal Lands Highway Division, 9 p. Federal Highway Administration (FHWA), 2012, Technical Memorandum: Cliff Shelf Landslide Investigation, Badlands Loop Road, Badlands National Park, SD, SD-PRA-BADL. 10(7): Federal Highway Administration, Central Federal Lands Highway Division, Lakewood, CO, 31 p. Federal Highway Administration (FHWA), 2013, Cliff Shelf Landslide, Badlands National Park: Geotechnical Design Report: Report No. SD-PX-BADL-13-01, Federal Highway Administration, Central Federal Lands Highway Division, 18 p. Gonzalez, M. A., 2010, Badlands of the northern Great Plains: Hell with the fires out. In Migon, P. (Editor), Geomorphological Landscapes of the World: Springer Science+Business Media B. V., Heidelberg, Germany, pp. 29–38. Hamel, J. V., 2004, Discussion of “Residual Shear Strength Mobilized in First-Time Slope Failures” by G. Mesri and M. Shahien: Journal Geotechnical Geoenvironmental Engineering, Vol. 130, No. 5, pp. 544–546. Iverson, R. M., 2005, Regulation of landslide motion by dilatancy and pore pressure feedback: Journal Geophysical Research, Vol. 110, F02015, 16 p. Kiver, E. P. and Harris, D. V., 1999, Geology of U.S. Parklands: John Wiley & Sons, New York, 912 p. Kumar & Associates, Inc., 1998, Summary Report of Reconnaissance of Landslide State Highway 240 at Cedar Pass, Badlands National Park, Jackson County, South Dakota: Project No. 981-198, Kumar & Associates, Inc., Denver, CO. Kumar & Associates, Inc., 1999, Geotechnical Engineering Study Cedar Pass Landslide, Badlands National Park, South Dakota: Project No. 98-1-198C, Kumar & Associates, Inc., Denver, CO. Kumar & Associates, Inc., 2000, Geotechnical Evaluation of Material Losses and Landslide Movements, Cedar Pass and Cliff Shelf Landslides, Badlands National Park, South Dakota: Project No. 00-1-139, Kumar & Associates, Inc., Denver, CO. Lupini, J. F.; Skinner, A. E.; and Vaughan, P. R., 1981, The drained residual strength of cohesive soils: Geotechnique, Vol. 31, No. 2, pp. 181–213. Monarco, D., 2018, personal communication, Federal Highway Administration, Central Federal Lands Highway Division, Lakewood, CO. National Park Service (NPS), 2016, Buttresses and Landslides within the Cedar Pass Landslide Complex, internal report, National Park Service, Interior, SD. National Park Service (NPS), 2020, Badlands: available at https://www.nps.gov/badl/index.htm Parsons Brinckerhoff Quade & Douglas, Inc., 2004, Badlands Loop Road Geotechnical/Pavement Report: Parsons Brinckerhoff Quade & Douglas, Inc., Murray, UT, 21 p. Radach, K. C., 2020, Characterization, Analysis, and Remediation of the Cedar Pass Landslide Complex, Badlands National Park, South Dakota: Unpublished M.S. Thesis, Department of Geology and Geological Engineering, Colorado School of Mines, 120 p.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

Cedar Pass Landslide Analysis RocScience, 2018, Slide, Version 8.0: RocScience Inc., Toronto, Canada. Santi, P. M., 2014, Calibration of modeling parameters for a large landslide, West Salt Creek, Mesa County, Colorado. In Abstracts with Programs, Vol. 46, No. 6: Geological Society of America, Boulder, CO, p. 610. Scheevel, C. R., 2017, Predicting Landslide Stability, Runout, and Failure Velocity at Cook Lake Landslide, Wyoming: Unpublished M.S. Thesis, Department of Geology and Geological Engineering, Colorado School of Mines, 75 p. Schulz, W. H., 2004, Landslides Mapped Using LIDAR Imagery, Seattle, Washington: U.S. Geological Survey Open-File Report 2004-1396, 11 p. Schulz, W. H.; McKenna, J. P.; Kibler, J. D.; and Blavati, G., 2009, Relations between hydrology and velocity of a continuously moving landslide—evidence of pore-pressure feedback regulating landslide motion?: Landslides, Vol. 6, pp. 181–190. Skempton, A. W., 1985, Residual strength of clays in landslides, folded strata and the laboratory: Geotechnique, Vol. 35, pp. 3–18. Smith, K. G., 1958, Erosional processes and landforms in Badlands National Monument, South Dakota: Geological Society America Bulletin, No. 69, pp. 975–1008. Starck, E., 2017, personal communication, National Park Service, Interior, SD. Stark, T. D.; Choi, H.; and McCone, S., 2005, Drained shear strength parameters for analysis of landslides: Journal Geotechnical Geoenvironmental Engineering, Vol. 131, No. 5, pp. 575–588. Stark, T. D. and Eid, H. T., 1994, Drained residual strength of cohesive soils: Journal Geotechnical Engineering, Vol. 120, No. 5, pp. 856–871. Stark, T. D. and Hussain, M., 2013, Empirical correlations: Drained shear strength for slope stability analyses: Journal Geotechnical Geoenvironmental Engineering, Vol. 139, No. 6, pp. 853–862. Stoffer, P., 2003, Geology of Badlands National Park: A Preliminary Report: U.S. Geological Survey Open-File Report 03-35, 63 p.

Tiwari, B.; Brandon, T. L.; Marui, H.; and Tuladhar, G. R., 2005, Comparison of residual shear strengths from back analysis and ring shear tests on undisturbed and remolded specimens: Journal Geotechnical Geoenvironmental Engineering, Vol. 131, No. 9, pp. 1071–1079. Trimble, D. E., 1979, Unstable Ground in Western North Dakota: U.S. Geological Survey Circular 798, 19 p. Van Houten, F. B., 1953, Clay minerals in sedimentary rocks and derived soils: American Journal Science, Vol. 251, pp. 61–82. Vithana, S. B.; Nakamura, S.; Gibo, S.; Yoshinaga, A. and Kimura, S., 2012, Correlation of large displacement drained shear strength of landslide soils measured by direct shear and ring shear devices: Landslides, Vol. 9, pp. 305–314. Walker, S. R., 1999, Analysis of the Red Creek Slide, U.S. Highway 50 Near Gunnison, Colorado: Unpublished M.S. Thesis, Department of Geological and Petroleum Engineering, University of Missouri–Rolla, 105 p. Wan, Y. and Kwong, K., 2002, Shear strength of soils containing amorphous clay-size materials in a slow-moving landslide: Engineering Geology, Vol. 65, pp 293–303. Wanless, H. R., 1920, Expedition of Princeton University to South Dakota: Princeton University, Princeton, NJ, 196 p. Wesley, L. D., 2003, Residual strength of clays and correlations using Atterberg limits: Geotechnique, Vol. 53¸ No. 7, pp. 669–672. Yeh and Associates, Inc., 2016, Geotechnical Services Summary Report, Cliff Shelf Landslide Repair Instrumentation, Badlands Loop Road, Badlands National Park, Jackson County, South Dakota: Project No. 216-249, Yeh and Associates, Inc., Denver, CO, 16 p. Zhang, M., 2013, Soil Erosion and Slope Stability at Significant Fossil-Bearing Sites within Badlands National Park: Unpublished M.S. Thesis, Department of Geology and Geological Engineering, South Dakota School of Mines and Technology, 85 p. Zonge International, Inc., 2013, Geophysical Investigation Report – Cliff Shelf Landslide Seismic and Resistivity Survey, Badlands National Park, South Dakota: Zonge International, Inc., Lakewood, CO, 15 p.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 193–213

213

Stratigraphic and Geochemical Evidence for the Alteration of Calcareous Glauconitic Marine Sediments to Calcium Bentonite JAMES H. MAY Alpha Geological Assessments, LLC, 301 Silver Creek Drive, Vicksburg, MS 39180

WAYNE C. ISPHORDING University of South Alabama (Emeritus Professor), 5506 Richmond Road, Mobile, AL 36608

DAVID PATRICK University of Southern Mississippi (Emeritus Professor), 118 College Drive #0001, Hattiesburg, MS 39406

DAVID R. WILLIAMSON Williamson and Associates, LLC, 10003, Stonehaven Drive Shreveport, LA 71118

JAMES E. LYLES, SR. Lyles Land and Mineral Services, 430 West Bontemps, Marksville, LA 71351

Key Terms: Ca Bentonite, Stratigraphy, Alteration, Geochemical, Origin, Marine Sediment, Volcanic Ash, Glauconitic, Calcareous ABSTRACT Calcium bentonite mined in Smith County, Mississippi, has been reported in numerous publications to be derived from the weathering of volcanic ash. These interpretations were based on the bentonite having similar properties to bentonites actually formed from volcanic precursors. No recent detailed stratigraphic mapping in combination with modern laboratory analyses had been conducted for this area. Calcium bentonite, found at Olmstead, Illinois, in contrast, formed in situ from the weathering of biotite, glauconite, quartz, and other materials found in shallow calcareous marine deposits. Detailed stratigraphic mapping in Smith County in 2011 was carried out by Mississippi State University as part of a proposed surface reservoir study. The mapping, utilizing GPS and continuous sonic core samples, indicated that calcium bentonite was present and similarly formed by the in situ weathering of calcareous, glauconitic, marine marl in several formations in the Oligocene Vicksburg Group. The bentonite was not restricted to one stratigraphic interval as would be the case for a true ashfall deposit. Additional research conducted at Mississippi State University and more recently by others confirmed that the bentonite was formed by weathering

and contemporaneous microbial action of the calcareous glauconitic marls. No volcanic ash was detected in any of the samples tested. The preponderance of other material present in the Smith County bentonite renders the presence of minute amounts of volcanic detritus volumetrically insignificant in the formation of the clay mass. A conceptual model is presented showing how the bentonite was formed and why it is restricted to this small area of the Oligocene outcrop.

INTRODUCTION Bentonite is commonly defined as clay that originated from the alteration of volcanic ash. The inference from the definition is that “all” bentonites are formed from volcanic ash. However, there are several categories of sedimentary bentonites that are formed from material that does not contain any volcanic precursors. Many of these sedimentary bentonites are of inferior quality and therefore are poorly sampled and documented. Historically, bentonite mined in Smith County, Mississippi, was assumed to be formed from the weathering of volcanic ash, even though no detailed stratigraphic or petrographic research had been conducted to confirm the assumption. This assumption was based primarily on the similarity in basic chemistry, as reported by Morse (1934), between the Smith County bentonites and bentonites actually

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

215

May, Isphording, Patrick, Williamson, and Lyles

derived from volcanic ash. The bentonite mines in the study area have been abandoned for more than 50 years, further limiting the gathering of new data. The Smith County bentonite has never been studied from an engineering geology perspective with regard to the volume of the clay mass. If one is arguing that a 1-m-thick layer of bentonite has formed solely from the weathering of volcanic ash, it is required that you have a thick layer of volcanic ash in the parent material. This becomes even more problematic when the parent material is being mined for agricultural lime and contains significant amounts of calcium carbonate, glauconite, silt, clay, and fossil fragments. A conceptual model showing how this bentonite formed and why it is found only in this small area of the Oligocene outcrop has never been presented. To further address the conundrum of “volcanic versus non-volcanic” origin, information is presented as part of this investigation that includes previously unavailable analyses of the Smith County bentonite in the form of detailed chemistry, scanning electron microscopy, and textural analysis. These analyses further confirm the “uniqueness” of this deposit when compared with bentonites of unquestioned igneous origin and categorically attest to the distinctive nature of this bentonite, whose origin has resulted from the transformation of former glauconitic marine sediments. Samples and Methodology The samples analyzed in this study were collected from the area of the abandoned Chisolm bentonite mine in Smith County. The samples were part of a larger geotechnical/hyrogeological data set obtained during the site investigation for a proposed surface water reservoir. The samples consist of portions of continuous core obtained using a sonic drilling rig. The field and laboratory procedures evolved over a period of several years, as the purpose of the original study involved looking for zones of high permeability, not determining the origin of the local bentonite. When it became obvious that the zones of high permeability were related to the formation of the bentonite, the focus shifted to trying to define how the bentonite actually originated. The methods for the bentonite research involved detailed stratigraphic mapping, using GPS and the core data, to determine the exact vertical and horizontal positon of the bentonite layers and identifying what geologic formation contained the bentonite. Research was also conducted to find references that could help explain what was observed in the detailed stratigraphic correlations. The literature review was hampered by the scarcity of research describing bentonites of

216

non-volcanic origin. Samples of the bentonite and parent material were initially sent to Mississippi State University for phase 1 tests that included petrography, X-ray diffraction, scanning electron microscopy, and microbial analyses. More recently, phase 2 tests consisting of additional X-ray diffraction, scanning electron microscopy analyses, and chemical analyses were performed at the Coordinated Instrument Facility at Tulane University. These results are presented in this article. Background In 2011, a geotechnical investigation was conducted in Smith County as part of a broader study to determine if an area north of Raleigh, Mississippi, east of Oakohay Creek was suitable for the construction of a reservoir (Mississippi State University and Pickering Firm, Inc., 2012) (Figure 1). A primary focus of the geotechnical investigation was to determine if the proposed reservoir site contained any subsurface cavities that would be detrimental to the integrity of the proposed reservoir. Of particular interest were areas where the Oligocene age Glendon Formation was exposed. Numerous test borings were drilled in the study area. Many of these boreholes were noted as being “blind holes.” The term “blind hole” indicates that there was a loss of circulation during the drilling of the hole. This area also contained an abandoned bentonite mine (Chisholm Mine) where clay, historically referred to in the literature as ash-derived bentonite, had been produced (Figure 2). A prior comprehensive surface reconnaissance and drilling program had been conducted to determine the stratigraphic position and engineering significance of the formations underlying the proposed reservoir site. In addition, soil boring logs conducted by the Mississippi Geological Survey (Luper, 1972) and private water well logs in the vicinity were also used for correlation and interpretation of the geologic formations for the reservoir study. As part of the geotechnical reservoir site investigation, numerous continuous core samples were taken with a sonic drill rig in and around the abandoned bentonite mine. These cores, along with GPS coordinates, allowed detailed stratigraphic correlation of the geologic units that were present at the site. It was determined that the bentonite was formed from the in situ weathering of the marine sediments and, significantly, not restricted to one stratigraphic interval. The sonic borings are critical in explaining the origin of the Smith County bentonite because that level of stratigraphic detail had not previously been conducted in the study area. Samples were sent to Mississippi State University for laboratory tests and analyses, and

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite

Figure 1. Location of the study area. Note the location of the Raleigh Salt Dome southeast of the study area.

the results confirmed that the marine sediments were weathering to bentonite. Some geologists in Mississippi were still not convinced that the bentonite formed from altered marine sediments. Another independent study was therefore initiated to address the problem utilizing a different group of scientists and a different university (Tulane). The results of this study form the basis for this article.

DISCUSSION OF RESULTS Bentonite, Bleaching Clay, and Fuller’s Earth The terms “bentonite,” “bleaching clay,” and “Fuller’s earth” historically have been used interchangeably to describe industrial clays composed primarily of the clay mineral montmorillonite. Montmo-

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

217

May, Isphording, Patrick, Williamson, and Lyles

Figure 2. The bentonite study area centered on the abandoned bentonite mine east of Little Oakohay Creek. Also shown are the core locations and the position of cross section A–A .

rillonite is a member of the smectite group of swelling clay minerals. Knight (1898) proposed the name “bentonite” for a “waxy clay” found in the Fort Benton unit of a Cretaceous Formation in Wyoming. Hewitt (1917) and Wherry (1917) determined that the Wyoming bentonite was an alteration product of volcanic ash. Ross and Shannon (1926) determined that the Wyoming bentonites were composed largely of montmorillonite. The Wyoming bentonite is dominated chieﬂy by sodium montmorillonite, while the Smith County bentonite consists largely of calcium montmorillonite (Table 1). The bentonite mined in Wyoming is widely used as a drilling mud and possesses distinctive features indicating an origin involving alteration of volcanic ash. All bentonites, however, are not necessarily of “volcanic origin.” Christidis and Huﬀ (2009) note that such clays may be formed by any of three mechanisms—(1) diagenetic alteration of volcanic glass, (2) hydrothermal alteration of volcanic glass, and (3) formation of smectite-rich sediments in salt lakes and sabkha environments, usually by dissolution and transformation of detrital smectites— and are often associated with sepiolite and/or palygorskite. The latter process does not require pyroclas-

218

tic (or volcanoclastic) precursors, and all three methods leave signature mineralogical and textural evidence that often can be used to diﬀerentiate their origin. Evidence for the volcanic nature of bentonite precursors is seen in the presence of primary igneous minerals such as β-quartz, biotite, sanidine, zircon, apatite, ilmenite, and magnetite; the presence of glass shards that may be fresh, partially altered, or pseudo-morphically replaced by smectite; and trace components, such as the rare earth elements. Köster et al. (2019) conducted research on ash-derived bentonites using boron geochemistry but, like Chistiidis et al. (1995), discusses only bentonites that have volcanic precursors. Sodium bentonites are characterized by their exceptionally high swelling capacity, typically expanding from an unglycolated d-spacing of 14 Å to 20 Å or more when glycolated. They are therefore prized by the petroleum industry for use as “drilling muds.” Calcium bentonites, in contrast, are often termed “nonswelling” bentonites, but are also in demand for other industrial applications though not to the same extent. Therefore, a major purpose of this article is to assess which origin most properly describes the genesis of the Smith County bentonites.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite Table 1. Chemical analyses of bentonite clays (after Nelson, 1922; Grim, 1933; Ross and Hendricks, 1945; Kerr et al., 1950; and Swineford et al., 1955). Bentonite Clays

Ca bentonites Smith County, MS Polkville, MS-19 Lorena, MS-20 Burns, MS-21 Porters Creek, KY, clay Polkville, MS-16 Lemon, MS-20 Lemon, MS-23 Lemon, MS-31 Booneville, MS-38 Pontotoc, MS-50 Olmstead, IL-1 Olmstead, IL-2 Olmstead, IL-3 Olmstead, IL-4 Olmstead, IL-5 Olmstead, IL-6 Olmstead, IL-7 Olmstead, IL-8 Shelbyville, TN Grand mean Na bentonites Birmingham, AL, Red Mountain (Ord) Birmingham, AL, Woodward (Ordovician) High Bridge, KY (Ordovician) Singleton, TN (Ordovician) Quilchena, BC Rosedale, BC, RH-25 Rosedale, AB Tallon, NV Carson District, NV, RH-49 Otay, San Diego, CA Otay, San Diego, CA RH-15 Otay, San Diego, CA RH-32 Amargosa Valley, NV, RH-3 Telachpi, CA, RH-4 Mariposa, CA, RH-7 Hector, CA, RH-14 San Diego, CA, RH-9 Santa Monica, CA, RH-26 Santa Monica, CA, RH-26 Santa Monica, CA, RH-26 Pala, CA, RH-36 Pala, CA, RH-42 Clairmont, CA, RH-40 Los Angeles, CA, RH-43 Conejos, CO, RH-8 Wagon Wheel, CO RH-47 Beidell, CO, RH-44 Beroil Energy, CO Belle Fourche, SD Ardmore, SD, RH-12 Wyoming SPV Wyoming SPV Rock Creek, WY WySt (Wyoming) WyR1 (Wyoming) WyR1 (Wyoming)

SiO2

Al2 O3 Fe2 O3 /FeO MgO CaO Na2 O K2 O TiO2 H2 O− H2 O+ H2 O (total)

56.15 50.95 47.64 51.18 60.68 50.2 50.37 50.39 49.48 50.53 49.95 60.78 58.06 56.2 61.7 55.42 58.82 54.46 56.2 54 54.16

19.42 15.54 17.79 16.3 15.66 16.19 17.11 17.37 17.75 19.31 27.26 13.83 15.03 15.4 15.11 15.61 15.12 16.84 13.2 24.48 17.22

3.22 1.62 3.67 2.43 6.4 4.13 2.88 2.84 3.48 7.25 2.58 4.5 4.42 4.33 4.63 3.75 3.97 3.36 5.08 3 2.90

3.28 4.65 2.63 4.41 1.79 4.12 2.49 4.56 4.42 2.6 1.39 1.85 2.18 1.96 1.67 1.46 1.98 4.84 2.92 2.75 2.90

1.42 2.26 1.23 2.12 0.29 2.18 1.23 1.29 4.42 0.72 0 1.19 0.53 0.47 0.83 0.94 0.55 3.2 1.6 2.08 1.43

0.14 0.17 0.12 0.17 0.17 0.17 0.56 0.46 0.06 0.41 0.2 0.19 0.16 0.15 0.28 * 0.17 0.06 * 0.17 * 0.17 1.74 0.31

0.79 0.47 0.26 0.38 0.16 0.16 0.09 0.04 0.06 0.34 0.36 1.16 0.9 1.13 0.98 1.41 1.34

0.26 0.32 0.12 0.28 1 0.2

0.59

0.43

58.88 55.28 54.56 54.8 67.04 54.88 59.8 61.5 45.32 59.84 50.3 49.56 54.58 53.88 50.33 53.02 53.96 50.03 70.3 65.66 50.06 50.72 49.7 50.28 54.46 48.05 47.28 52.49 60.64 51.54 57.5 59.69 60.18 65.9 68 69

22.91 24.65 19.97 22.93 13.46 19.92 16.36 21.2 24.84 11.84 15.96 15.08 16.44 11.66 16.42 18.5 15.44 16.75 12.81 12.71 21.22 22.14 22.1 20 16.84 23.01 20.27 21.58 23.26 18.25 18.3 19.93 26.58 21.5 21.3 20.4

1.64 1.59 1.97 2.36 3.02 4.32 5.75 0.1 0.8 3.2 0.86 3.44 4.6 0.92 2.42 2.46 1.12 6.29 2.22 3.54 0.22 1.49 2.12 4 3.36 6.67 8.68 10.82 3.92 2.64 8.23 3.07

2.36 3.71 5.08 3.1 1.93 2.83 2.67 1.1 0.16 2.32 6.53 7.84 4.9 8.61 4.1 4.04 6.99 2.78 0.24 0.71 4.42 4.28 2.85 4.6 4.84 2.14 0.7 3.21 2.19 3.41 2.62 1.82 1.01 2.82 2.49 2.48

0.01 0.01 1.08 1.2 1.78 2.22 1.82 0.4 2.78 2.9 1.24 1.08 0.72 1.56 1.39 0.8 0.8 1.2 1.18 1.44 1.26 1.56 1.68 1.08 3.2 1.52 2.75 0.42 0.59 2.18 0.71 0.28 0.23 1.63 1.31 1.33

2.42 2.06 1.66 4.12 0.53 1.75 2 1.8 0.1 2.13 3.19 1.19 3.02 0.15 0.12 3.8 0.94 0.26 2.1 1.1 0.83

4.3 3.71 4.06 2.04 0.22 0.26 0.27 0.2 0.12 2.34 0.45 0.45 0.81 0.39 1 0.16 0.54 0.6 2.48 1.73 0.19

1.17

0.21 0.97 0.84 4.33 2.09 2.49 2.09

1.04 0.01 0.41 0.37 0.12 0.55 0.55 1.23 0.56 0.52 0.5

3.95 4.07 4.32

15.01 19.25 15.02 5.56 15.58 12.29 13.31 18.26 10.66 11.1 5.18 5.13 6.3 5.36 6.38 4.65

8.28 6.78 8.29 6.18 7.57 10.93 10.11 5.86 7.9 10.55 11.1 13.11 13.25 10.71 14.28 12.52 16.1 20.32

0.25 0.25 0.05

4.16 4.16 7.42 5.78 6.64 8.1 5.18

4.77 3.27 4.52 3.83 4.92 4.28 4.91

0.4 0.18 0.44

11.1 8.21

5.49 9.98

11.69 14.22 13.53 0.68 4.34 14.05 11.56

5.44 6.34 7.52 7.12 8.03 7.56 8.24

6.14

10.46

13.22

6.33

0.29 0.75

0.35 0.45 0.23 0.46 0.44 0.8

0.13 0.19 0.65 0.32 0.4

0.28 0.8

2.69 2.16 1.95

1.56 0.12

0.17 10.26

15.82 23.29 26.03 23.31 11.74 23.15 23.22 23.42 24.12 18.56 21.56 16.28 18.24 19.55 16.07 20.66 17.17 16.1 20.32 9.12 19.39 8.93 7.43 11.94 9.61 11.56 12.38 10.09 8.61 22.64 10.5 23.61 22.96 16.59 18.91 23.95 17.13 20.56 21.05 1.8 12.37 21.61 19.8 21.14 19.6 16.1 16.6 19.72 8.56 2.83 19.55 7.18 15.6 10.26

0.24 0.16 0.16

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

219

May, Isphording, Patrick, Williamson, and Lyles Table 1. Continued. Bentonite Clays

WyR1 (Wyoming) WyR1 (Wyoming) WyR1 (Wyoming) Wyoming, RH-19 Wyoming, RH-29 Wyoming, RH-30 Newcastle District, WY Tatatila, Mexico, RH-6 Atzapoz, Mexico, RH-22 Mexico, location unknown, RH-24 Maniquipi, Mexico, RH-39 San Antonio, TX, RH-27 Leakey, TX, RH-34 South Bosque, TX, RH-48 Grand mean *

SiO2

Al2 O3

Fe2 O3 /FeO

MgO

CaO

Na2 O

K2 O

TiO2

66.3 67.6 67.6 55.44 49.2 53.5 66.9 52.09 50.44 54.5 49.34 52.08 51.74 51.87 56.15

21.7 21.3 21.3 20.14 17.6 21.57 15.26 18.98 16.26 19.44 22.88 18.2 22.81 26.8 19.42

4.01 4.03 4.14 3.97 1.6 3.28 2.92 0.06 5.38 0.1 2.36 2.88 0.39 1.42 3.20

2.71 2.57 2.56 2.49 5.08 1.89 2.26 3.8 3.92 5.07 1.67 4.48 3.32 4.42 3.28

1.61 1.45 1.41 0.5 1.52 1.25 0.46 3.28 0.72 0.48 2.66 2.28 0.01 1.9 1.34

2.38 2.15 2.14 2.75

0.61 0.57 0.56 0.6

0.18 0.15 0.16 0.1

1.94 2.12

1.04 0.42

0.11 0.11 0 0.42 0 0

0.26 0.09 0.89

0.01

0.2

0.17 0.03 1.71

H2 O−

H2 O+

5.8 11.75 16 14.5 14.77

3.67 7.46 6.3 8.04 6.57

11.84 0.81

9.19 12.61

0.27

H2 O (total)

14.7 25.52 15.2 9.47 19.21 22.3 22.04 21.34 20.8 21.03 13.42 15.82

Value obtained by “averaging” information supplied for all other Olmstead Group samples.

Bleaching clay and Fuller’s earth are terms used to describe a wide range of clay and clay mixtures that have the ability to remove colors from oils and other liquids. The bentonite (calcium montmorillonite) mined in Smith County was mined for use as a bleaching clay. The clay deposits at Olmstead were also utilized in a similar manner. Grim (1928) noted that no categorical evidence was found for volcanic ash in any of the samples he examined and that the clays were not the result of the in situ weathering of volcanic ash. He did not discuss the origin of the bentonite in this report, as this was prior to his Olmstead investigation, wherein he concluded that bentonite can form by differing processes from a variety of source materials. Bay (1935) discussed these clays. These are the same clays that are described by Morse (1934). Bay (1935) described the “bleaching clays” (bentonites) of Smith County as being remarkably free from grit, whereas, the Wayne County, Mississippi, bentonites contained fine sand and mica particles. Dockery and Thompson (2016) discuss a 6.5-ftthick bed of bentonite present in an outcrop on Ichusa Creek in Smith County. Samples of the bentonite were sent to the U.S. Geological Survey to determine their age by dating sanidine crystals. The sanidine crystals, however, were found to be detrital, and the age of the bentonite was not determined. Morse (1934) examined the chemical properties of the bentonites in Mississippi and, believing that they were similar to Wyoming and other western bentonites, concluded that they must have originated from volcanic ash. Morse did not have the additional chemical analyses available today, or it would have been apparent to him that significant dif-

220

ferences are seen in the chemistry of “sodium-rich” western bentonites that were derived from precursor sediments containing volcanic ash versus some of the “calcium” bentonites in the eastern United States originating from sediments with little or no volcanic ash. Table 1 is a compilation of numerous published chemical analyses for clays described as bentonites. This table clearly shows that these clays can be easily differentiated based on their distinctive calcium/sodium contents, but determination of their origin requires evaluation of other factors. A recently analyzed sample from the Smith County bentonite location is included as the first analysis in Table 1. Its chemical characteristics are clearly most similar to others included in the “calcium bentonite” category. The analysis from Burns, Mississippi (Burns, MS-21), is also from the study area described in this report and, similarly, is included with other Table 1 calcium bentonites. Several Ordovician potassium bentonite samples from the Southeast are also included in Table 1. These, as with others in eastern North America, are generally included with sodium bentonites and, likewise, are considered to have been derived from precursor volcanic ash material (see Kolata et al., 1989). Some years later, Grim and Guven (1978) further defined bentonite as a clay that was predominantly a smectite group clay mineral (montmorillonite is an example) and whose properties are controlled by this mineral regardless of how it originated. This definition is in accordance with his earlier research findings (Grim, 1933) that all bentonites are not derived from volcanic ash. However, he did describe clays from some localities as being derived from or partly derived from

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite

ash. Other investigators (e.g., Reynolds, 1962) have noted that montmorillonite continues to be a common clay mineral component in many rivers draining into the Gulf of Mexico, and its origin is thus properly described simply as a “detrital” mineral. Geological observations of weathered ash deposits in other areas further support the model for the formation of bentonite described in this article. Volcanic ash deposited in non-marine sediments does not always form bentonite. One such observation is in the Naborton Formation of the Wilcox Group in northwestern Louisiana (Williamson, 1986). A persistent layer of grayish-white clay with ﬁnely disseminated black organic matter, approximately 3 to 4 in. (7.62 to 10.16 cm) thick, was observed in the highwall of the Dolet Hills Lignite Surface Mine in DeSoto Parish, Louisiana. The clay stratum was present in the upper part of the Chemard Lake Lignite Lentil of the non-marine Naborton Formation of the lower Wilcox Group of the Paleocene Series. The clay layer has a sharply deﬁned contact with the underlying lignite seam and with the overlying approximately 10 in. (25.4 cm) of lignite. The total thickness of the entire lignite seam, including the clay layer, averaged 6.1 ft (1.86 m) over the area of the 30,000-acre (12,141-ha) surface mine permit area. Examination of the clay layer found in the lignite mine by scanning electron microscopy showed the presence of altered glass shards, indicating that the origin of the clay layer was a volcanic ashfall (i.e., tonstein). This ash layer represented a true ash deposit, which is characteristically thin, extends for miles, and can often be used as a time line. Observation of the ash layer in the deeply weathered zone along the subcrop did not reveal the presence of any bentonitic clays in the non-marine sediments. Detailed Stratigraphy in the Study Area An accurate description of the stratigraphic units in the study area was essential in determining which formations contained bentonite and the elevation of the bentonite. Cooke (1923) described and correlated the units in the Vicksburg Group. McIlwain (2007) mapped the stratigraphy in the study area during the reservoir study. The geologic formations studied in the area of the abandoned bentonite mine in Smith County are, from oldest to youngest, the Forest Hill Formation (Oligocene), the Marianna/Mint Spring Formation (Oligocene), the Glendon Formation (Oligocene), the Byram Formation (Oligocene), the Bucatunna Formation (Oligocene), the Catahoula Formation (Miocene), Terrace Deposits (Pliocene/Pleistocene), and Recent Alluvial Deposits (Figure 3). The only formations that contained

Figure 3. Stratigraphic units in the study area.

bentonite were the Bucatunna, Byram, Glendon, and Mint Springs. Marianna/Mint Spring Formation The Marianna was not diﬀerentiated from the Mint Springs during this investigation. No surface outcrops of the Mint Spring were located in the study area except for several small exposures in the banks of Oakohay Creek. The weathered Mint Spring is typically a reddish-brown, glauconitic sand. It is very fossiliferous but weathers very quickly on the outcrop. Based on the current study, the marl could form bentonite if it is exposed at the surface in an area of rugged topography and a low water table. In the subsurface, it is composed of greenish-gray, fossiliferous, glauconitic, pyritic, sandy marl and greenish-gray, medium- to coarse-grained, glauconitic, fossiliferous sand. The lower contact of the Mint Spring was picked at the appearance of dark carbonaceous clay or ﬁnegrained gray sand of the Forest Hill Formation. The thickness of the Marianna/Mint Spring ranged from 2.5 ft (.76 m) in boring FS-7 to 18.5 ft (5.64 m) in boring JTB-3. The upper contact of the Marianna/Mint Spring is placed at the contact of the marl with the overlying Glendon Limestone. Glendon Formation The un-weathered Glendon in the study area usually appears as a medium-gray to light-olive-gray,

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

221

May, Isphording, Patrick, Williamson, and Lyles

fossiliferous, glauconitic limestone. Throughout the area, hard ledges of limestone are inter-bedded with gray to greenish-gray marl. On the outcrop, the weathered Glendon has holes and cavities, and this characteristic is known locally as “horsebone weathering.” The Glendon section was deeply weathered in many of the exploratory borings. May (1974) and others have noted that the Glendon weathers to a brown montmorillonitic clay. David R. Williamson (personal observations, 1970 to present) has observed thick montmorillonitic clay deposits in Pitts Cave in Wayne County and other caves in southeastern Mississippi developed beneath the limestone ledges of the Glendon Formation. The signiﬁcance of the presence of the bentonite in the Glendon and loss of circulation in borings is important in understanding how the bentonite was actually formed. As the marls alter to bentonite, there is a volume reduction that increases local permeability. The thickness of the Glendon varied depending on the depth of geochemical weathering. The thickest unweathered section of the Glendon was 45.5 ft (13.87 m) in boring FS-7. Byram Formation The un-weathered Byram is composed of greenishgray, glauconitic, fossiliferous, clayey marl. Fragments of fossils as well as microfossils, such as foraminifera, were found in the test borings. Luper (1972) stated that the average thickness of the Byram in Smith County was 11 ft (3.35 m). The thickest section of Byram found during the reservoir investigation was 10 ft (3.05 m) in boring JTB-8. The Byram section was often truncated by erosion or geochemical weathering. The thinnest un-truncated section of Byram appeared to be darkgray clay in boring FS-13, where it was 6 ft (1.83 m) thick. The upper contact of the Byram is placed at the contact of the Bucatunna. Outcrops of Byram are rare because of the thinness of the unit and the fact that it is easily weathered and eroded. In boring BMB-4, the Byram was near the ground surface and had weathered to bentonite. Bucatunna Formation The un-weathered Bucatunna consists of dark-gray to black, micaceous, glauconitic, fossiliferous, carbonaceous, silty clay. These glauconitic zones are altered to bentonite when exposed near the ground surface (boring WHB-1). The only bentonite unit that Luper (1972) reported in a boring in Smith County was in the lower Bucatunna in boring AL-36. The thickest section of the Bucatunna in the study area was 25.5 ft (7.77 m), found

222

in boring FS-11. The thinnest section of Bucatunna, overlain by Catahoula at the site, was 7 ft (2.13 m) in boring JTB-8. The weathered Bucatunna is typically chocolate brown in color and may contain gypsum crystals and limonite staining along fractures. The contact between the Bucatunna and underlying Byram is conformable. The upper contact with the Catahoula Formation is unconformable. A stratigraphic cross section across the study area clearly demonstrates that the bentonite is not restricted to any given formation but occurs where glauconitic, calcareous sediments are exposed in the deeply weathered zone (Figure 4). The layers of bentonite that were mined in the area were much thicker than would be expected for an ashfall deposit far removed from centers of volcanic activity. The bentonite correlated with the ground surface and highly permeable zones in the weathering profile (Figure 5). The bentonite was observed to be present within the zone of weathering but not below its base. No bentonite was found below the water table (unoxidized zone) in any of the cores sampled. Field reconnaissance at other nearby abandoned bentonite mines indicated that the bleaching clay mined at those locations was also associated with near-surface, highly weathered zones positioned well above the water table. One core in particular (see Figure 6) showed a distinct contact between oxidized and un-oxidized marl. It was apparent that the contact was a result of weathering, as it cut across fossils. The unusually deep weathering could be the result of faulting caused by a nearby salt dome. The Raleigh Salt Dome is located near the study area as shown in Figure 1. Wilbur Baughman (in Luper, 1972) attributed abrupt changes in elevation of the base of fresh groundwater in the study area to faulting. The oxidized portion of the core was partly altered to bentonite, and it was obvious that the bentonite was the result of in situ geochemical alteration of the marine marl. The depth of weathering near the abandoned bentonite mine seemed to be controlled by the rugged topography and the base level of the nearby streams. A copy of the boring log for sample FS 6 is presented in Figure 7 and demonstrates both the effects of deep weathering and that bentonite is present in multiple formations. The failure to recognize the significance of severe weathering and resulting geochemically altered materials has caused significant geologic mapping and nomenclature problems in Mississippi. In particular, stratigraphic mapping has been a problem with the coarse-grained up-dip material in the Miocene Formations and mapping of the Plio-Pleistocene Citronelle /Lafayette Formation. The stratigraphic placement of silicified wood and the interpretation of Carolina Bays

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite

Figure 4. Cross section A–A . The bentonite occurs in the weathered zone and is not restricted to any one formation.

Figure 5. The occurrence of bentonite does not correlate with any speciﬁc stratigraphic horizon or formation but does correlate with the elevation of the deeply weathered ground surface.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

223

May, Isphording, Patrick, Williamson, and Lyles

ever, all reported the presence of layers of calcareous, fossiliferous, glauconitic marl and limestone. Further, many of the bulletins did report that the Glendon Formation weathered to a montmorillonitic clay The samples of bentonite from Wayne County that May (1974) tested for bleaching properties came from a severely weathered zone where the Oligocene age, Glendon Formation limestones and marls were exposed at the surface in the weathering zone. Laboratory Tests

Figure 6. Distinct weathering front showing oxidized marl over unoxidized marl in soil boring FS-7 on the east side of Oakohay Creek.

also rely on an understanding of severe geochemical weathering and silica transport (see Isphording, 1984; May and Warne, 2004). Carolina Bays are circular to elliptical Atlantic and Gulf coastal plain topographic depressions that look karst-like but are not usually underlain by carbonates. No volcanic material was found in the weathered or un-weathered core samples. A literature review was carried out to determine if similar clays and weathering conditions had been reported elsewhere. The Mississippi Geological Survey County bulletins on counties where the Vicksburg Group is exposed were examined to determine if any of the authors noted layers of volcanic ash. These counties included Warren (Mellen, 1941), Hinds (Moore, 1965), Rankin (Baughman, 1971), Copiah (Bicker, 1966), Smith (Luper, 1972), Wayne (May, 1974), and Clarke (Gilliland, 1980). None of the above authors reported seeing layers of volcanic ash in the Vicksburg section; how-

224

The phase 1 laboratory tests of the Smith County bentonite conducted at Mississippi State University in 2011 included petrography, X-ray diffraction, scanning electron microscopy, and microbial analyses. The results of the laboratory analyses confirmed the stratigraphic evidence that the calcium bentonite in the study area was formed by the in situ alteration of calcareous, glauconitic marine sediments. The details of the phase 1 testing have been presented in a master’s thesis and in several poster sessions (Calhoun, 2013; Calhoun et al., 2013). Phase 2 tests of the bentonite consisted of chemical analyses carried out by Eurofins TestAmerica Laboratories in Pensacola, Florida, and additional X-ray diffraction and scanning electron microscopy that were conducted at the Coordinated Instrument Facility at Tulane University. Scanning electron microscopy of the bentonite revealed that the clay was present in several forms. Vermiform intergrowths were common, as were surficial overgrowths on quartz and glauconite grains. Equally abundant was bentonite occurring as “replacement platelets” on precursor smectite grains (Figure 8). Grim (1933) described similar overgrowths on quartz and glauconite grains in his study of the Olmstead clay. He similarly concluded that quartz and glauconite were being altered to montmorillonite. X-ray diffraction analysis of the Smith County bentonite samples, expectedly, revealed that these clays display minimal expansion of only a few angstroms when subjected to glycol treatment (15.1 Å for unglycolated, 17.6 Å for glycolated). The analyses determined that the material at the 44-ft (13.41-m) interval in boring BMB-1 (see Figure 9) was mostly saponite, whereas, the material near the base of the weathered zone at 66 ft (18.9 m) was primarily the iron-rich smectite clay nontronite. Calcite is abundantly present with the associated bentonite, and its partial dissolution could certainly supply the calcium needed to replace the iron in the original smectite (i.e., nontronite) during the transformation of the marine sediments to calcium bentonite in the intensely weathered zone. Still yet another feature that typically distinguishes Ca bentonites from their Na-dominated counterparts

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite

Figure 7. Boring FS-6 illustrates the extreme depth of weathering in parts of the study area. The marls in various formations are altering to bentonite.

is their overall particle size distribution. Ca bentonites, which owe their origin largely to post-depositional chemical weathering processes that may take place in more restrictive environments, such as lagoons or nearshore zones of elevated salinity, are typically of coarser grain size than are bentonites whose precursor was air-fall volcanic ash. Consequently, Ca bentonites characteristically contain larger quantities of “silt-sized” and “sand-sized” components than do Na bentonites. This feature is clearly seen in size analyses of samples from Smith County shown in Table 2. Sieve and hydrometer analyses were performed according to ASTM D-422 procedure and produced the descriptive parameters and measures of central tendency (mean and median diameters) and dispersion (sorting, skewness, and kurtosis) calculated using the equations described by Inman (1952) and Folk and Ward (1957).

Texturally, the particle size percentages would classify the sample as a “slightly gravelly, sandy mud,” using Folk and Ward’s (1957) classiﬁcation, but this is misleading in view of the fact that “gravel-sized” constituents were largely absent and that those few present were composed entirely of fossil fragments. A better textural description would be that derived from the statistics using the U.S. Army Corps of Engineers classiﬁcation system, which would classify the sample as simply “silty clay.” ORIGIN OF THE SMITH COUNTY BENTONITE The discussion of the origin of the Smith County bentonite will focus on the calcium montmorillonites (bleaching clays) that were mined in the study area but has implications on the formation of other bentonites associated with shallow marine deposits.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

225

May, Isphording, Patrick, Williamson, and Lyles

Figure 8. Scanning electron micrographs of Smith County, Mississippi, bentonitic clay.

Nutting (1935) researched bleaching clay and concluded that it was the presence of chemically open bonds or free valences on the clay surface that made it selectively adsorbing. Other factors contributing to

the adsorption are platy or cleavable crystalline structure, ﬁne grain size, and characteristic bonding of the atoms in the clay mineral. Shearer (1917) stated that the bleaching clay in Georgia was formed from a

Table 2. Textural, statistical, and engineering parameters for Smith County, Mississippi, bentonite Composition *Percent gravel Percent sand Percent silt Percent clay

Size 0.09 29.71 36.68 33.52

Mean diameter Median diameter Standard deviation Skewness Kurtosis

12.49 microns (ﬁne silt) 15.87 microns (ﬁne silt) 3.350 (very poorly sorted) 0.095 (near symmetrical) 0.693 (platykurtic)

Engineering Parameters (from Cumulative Frequency Curve): D-60: 36.51 microns; D-30: 2.52 microns; D-10: 0.72 microns Coeﬃcient of uniformity 50.418; coeﬃcient of curvature 0.241 *

Descriptive statistics are from Inman, 1952, and Folk and Ward, 1957.

226

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite

Figure 9. X-ray diﬀractogram of Smith County, Mississippi, bentonite sample. Other phases present include minor kaolinite (7 Å) and prominent calcite peak (3Å).

calcareous clay that had been deposited in shallow water. He proposed that as the calcareous clay underwent leaching, the calcium carbonate was removed, leaving a large volume of openings. He also concluded that silica was deposited from solution during this leaching process. Grim (1933) noted that the bleaching clays mined at Olmstead were originally nearshore marine deposits. Bay (1935) studied the Mississippi bleaching clays and, contradicting Grim, concluded that they were the result of the devitrification and chemical alteration of volcanic ash. He concluded that the physical and chemical environments of the clay-forming materials has been very effective factors in the development of the bleaching clays. Bay stated that severe leaching has played a significant role in the development of various types of bleaching clays. In ash-derived bentonite, such as the Wyoming bentonite, some trace of the ash structure can be seen. In weathered and un-weathered samples analyzed during this study, no volcanic ash was noted, but glauconite was common. Glauconite Glauconite is a green to black iron-rich illite that is a major constituent of the marl and limestone at the old bentonite mine site (see Figure 10). There was an

apparent higher concentration of glauconite in the limestone and marls in the area of the bentonite mines than noted elsewhere along the Vicksburg outcrop. Further research revealed that glauconite can be formed from the alteration of volcanic ash under reducing conditions (such as might occur in a shallow sea). Jeans (2006) discussed Upper Cretaceous strata in England, Scotland, and Ireland, where glauconite is particularly abundant. He said that the glauconite represented the glauconitization of penecontemporanous volcanic ash. Glauconite can form as well from material other than volcanic ash. Regardless of how the glauconite formed, it is highly abundant in the un-weathered marls and limestones in the study area. The next step was to look at how glauconite weathers above the water table under oxidizing and leaching conditions similar to those at the area of the bentonite mines in Smith County. During weathering, glauconite loses iron and is altered to a smectite. Pestitschek et al. (2012) conducted a study to ﬁnd the mineralogical and chemical diﬀerences between weathered glauconite on the surface and fresh glauconite from the subsurface. They found that the color, mineralogy, and chemistry indicate that surface glauconite samples are strongly altered by weathering processes and that glauconite transforms progressively

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

227

May, Isphording, Patrick, Williamson, and Lyles

Figure 10. Glauconite-rich fossiliferous marl from soil boring JTB-1.

into Fe-rich mixed-layer illite-smectite and then into smectite-group clay. Weathering can thus completely reverse the glauconitization process. For any chemical and mineralogical characterization of glauconites at or near the surface, these weathering eﬀects must be taken into consideration. The chemistry of the in situ alteration is described in detail in the experimental transformation of glauconite to smectite by Robert (1972). He duplicated in the laboratory the natural weathering of glauconite to smectite and found that dioctahedral micaceous illites like glauconite that have a lower tetrahedral charge transform to smectite easily. Only a lowering of total charge is needed, and reduction-oxidation plays an important role. Figure 11 is a series of X-ray diﬀractograms showing the change during alteration from glauconite to smectite. The 10 Å peak in A is the unaltered glauconite, and the 17 Å peak in C is smectite. Additional evidence for in situ alteration of glauconitic marl and limestone to bentonite is given in the report by Bay (1935) in the form of molds of sand dollars and shark teeth in samples of bentonite that were being mined in Smith County. James H. May (per-

228

sonal communication, 2007), during interviews with people who had worked at the bentonite mine, was also told of the presence of shark teeth in the bentonite. Figure 12 is a conceptual model of how the bentonite deposits in Smith County were formed. The

Figure 11. Series of X-ray diﬀractograms demonstrating that glauconite alters to smectite. After Robert (1972).

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite

Figure 12. Conceptual model demonstrating the stratigraphic relationship of bentonite occurrence in Smith County, Mississippi. The glauconite-rich marine units are geochemically altered to bentonite.

glauconite-rich strata that is found in the subsurface is altered to bentonite as it is exposed to the severe weathering and leaching conditions near the surface above the water table. During the geochemical alteration of the marls and limestones to smectite, the volume of the material is reduced, causing localized increases in permeability. Zones of oxidized iron were noted under the zones where the bentonite was forming. This explains the numerous reports of loss of circulation in borings penetrating through this highly weathered material. The stratigraphic correlations from the continuous sonic cores taken at the site indicated that the bentonite was the result of in situ alteration of glauconitic-rich marine sediments. Samples of the weathered and un-weathered material from the sonic cores were sent to Mississippi State University and examined extensively using scanning electron microscopy, X-ray diﬀraction, petrographic microscopy, and microbial methods as part of a phase 1 testing effort. The results of the phase 1 and phase 2 analyses

conﬁrmed the stratigraphic observations that the bentonite was formed in situ by the alteration of marine sediments. The origin of the Smith County bentonite ﬁts the model introduced by Grim (1933), whereby a calcareous, glauconitic marine marl alters in place to bentonite. Some formations that are glauconitic, such as the Eocene Winona Formation in Mississippi and the Eocene Shark River Formation and massive Hornerstown Formation in New Jersey and Delaware, do not alter to bentonite. Apparently, weathering of glauconite alone does not form a bentonite deposit. These formations may not have the necessary percentage of calcium carbonate and other constituents needed to allow the alteration process to take place. Calciumrich groundwater apparently is a key factor in the alteration process. The glauconitic sands in New Jersey are in a climate zone (humid, warm summers and very cold winters) that is dissimilar to that of Smith County (humid, hot summers and mild winters) in severity of

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

229

May, Isphording, Patrick, Williamson, and Lyles

weathering and thereby lack any significant occurrence of bentonite. The glauconite alone has not formed the bentonite, but it is a necessary component of the marine sediments that have altered in situ to bentonite. From a geochemical standpoint, there is reasonably strong support that the Smith County bentonites owe their origin to a process involving basic metasomatic exchange of calcium-rich groundwaters acting on precursor iron-rich smectite clays in a near-surface weathering environment. By this process, the original marine sediments underwent replacement by calcium-rich groundwater, transforming them into bentonite. CONCLUSIONS The conclusions drawn from the field and laboratory investigations of the Smith County bentonites reinforce Grim’s (1933) model for the in situ alteration of marine sediments to bentonite. The opinion expressed by Morse (1934), in contrast, implied that volcanic ash was the sole material from which all Mississippi bentonites were derived. Recent and past research has clearly indicated that this is not the case. In Smith County, the marine sediments below the highly weathered zone contain iron-rich clays, such as glauconite and nontronite. Within the highly weathered zone, the calcium-rich clay saponite is prominent. The bentonite in the study area was not found at the same stratigraphic position, as would be the case for a true pure ashfall deposit, but occurred in different formations, depending on their location with regard to the ground surface and water table. The thicknesses of the bentonite layers (up to 2 m) is too great for a true ashfall deposit in an area like Smith County that is far removed from volcanic centers. No volcanic ash residues were found in any of the samples studied, and the clays appear to be the result of diagenetic alteration of shallow-shelf marine sediments by calcium-charged groundwater. The release of calcium from the limestones and marls during the weathering process apparently played a key role in the formation of the calcium bentonite. The formations in the Vicksburg Group in the area of the abandoned bentonite mine were observed to have a higher percentage of glauconite than was reported in other areas of Mississippi. The increased amount of glauconite in the marine sediments in the study area appears to be a significant contributing factor in transforming them into a commercial-grade bleaching clay. Any volcanic detritus (such as zircon crystals) that might be found in the Oligocene bentonites (and some other calcium bentonites) would be volumetrically insignificant in the formation of the resulting clay body. To have a 1-m-thick layer of calcium bentonite derived solely from volcanic ash would

230

require a thick layer of pure volcanic ash. A 2-mthick layer of pure ashfall bentonite would be unusual, but 2-m-thick layers of calcareous marl are common. The formation of the bentonite appears to have occurred after the evolution of the present-day topography, making the bentonite a relatively young material. These conclusions are significant in regard to other bentonites in the Southeast, especially those that are associated with nearshore marine deposits. This research has significance for several areas of engineering geology. With the understanding that the formation of calcium bentonite does not always require volcanic ash, regions far removed from volcanic centers can be explored for commercial clay deposits. Also, areas where marine marls are being weathered to bentonite can have a serious foundation problem because of the plasticity of the calcium bentonite. ACKNOWLEDGMENTS The authors would especially like to thank Dr. George C. Flowers and Ms. Tanja L. Goehring, Tulane University Coordinated Instrumentation Facility, for their assistance in generating the scanning electron microscopy and X-ray diffraction information included in this article. Chemical analysis of the bentonite was carried out by Eurofin TestAmerica Laboratories, Pensacola, Florida. James E. Lyles is also extended special thanks for his input as a soil scientist and for drafting the diagrams included in this article. REFERENCES Baughman, W. T., 1971, Rankin County Geology and Mineral Resources: Mississippi Geological, Economic and Topographical Survey Bulletin No. 115, pp. 41–50. Bay, H. X., 1935, A Preliminary Investigation of the Bleaching Clays of Mississippi: Mississippi Geological Survey Bulletin No. 29, pp. 15–17. Bicker, A. R., 1966, Claiborne County Geology and Mineral Resources: Mississippi Geological, Economic and Topographical Survey Bulletin No. 107, pp. 113–115. Calhoun, K., 2013, Investigation of Parent Source Material in Smith County, Mississippi: Unpublished M.S. Thesis, Mississippi State University, Starkville, 58 p. Calhoun, K.; Schmitz, D.; Kirkland, B.; and May, J., 2013, Parent Source Material of Calcium Bentonite in Smith County, Mississippi: Poster presentation given at the AAPG 2013 Annual Convention and Exhibition, Pittsburgh, PA, May 19–22, 2013. Christidis, G. E. and Huff, W. D., 2009. Geological aspects and genesis of bentonites: Elements, Vol. 5, No. 2, pp. 93–98. Christidis, G. E.; Scott, P. W.; and Marcopoulos, T., 1995. Origin of the bentonite deposits of eastern Milos, Aegean, Greece: Geological, mineralogical and geochemical evidence: Clays Clay Minerals, Vol. 43, No. 1, pp. 63–77. Cooke, C. W., 1923, Correlation of the Vicksburg Group: U.S. Geological Survey Professional Paper 133, pp. 2–3.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

Glauconitic Marine Sediments Can Alter to Calcium Bentonite Dockery, D. and Thompson, D. E., 2016, The Geology of Mississippi: Mississippi Department of Environmental Quality, Ofﬁce of Geology, Jackson, pp. 480–481. Folk, R. L. and Ward, W. C., 1957, Brazos River Bar: A study in the signiﬁcance of grain size parameters: Journal Sedimentary Petrology, Vol. 27, pp. 3–26. Gilliland, W. A. 1980, Clarke County Geology and Mineral Resources: Mississippi Geological, Economic and Topographical Survey Bulletin No. 121, pp. 75–79. Grim, R. E., 1928, Preliminary Report on Bentonite in Mississippi: Mississippi Geological Survey Bulletin No. 22, pp. 10–11. Grim, R. E., 1933, Petrography of the Fuller’s earth deposits, Olmstead, Illinois: Economic Geology, Vol. 28, No. 4, pp. 344–363. Grim, R. E. and Guven, N., 1978, Bentonites: Geology, Mineralogy, Properties and Uses, Developments in Sedimentology, Vol. 24: Elsevier, Amsterdam, pp. 14–18, 126–149. Hewitt, D. F., 1917, The origin of bentonite: Journal Washington Academy Science, Vol. 7, pp. 196–198. Inman, D. L., 1952, Measures for describing the size distributions of sediments: Journal Sedimentary Petrology, Vol. 22, pp. 125–145. Isphording, W. C., 1984, Sand craters in Gulf Coastal Plain clastic sediments: An extension of the Carolina Bays phenomenon? In Abstracts of the Southeastern/North-Central Section Meeting: Geological Society of America, Lexington, KY, pp. 148. Jeans, C. V.., 2006, Clay mineralogy of the Cretaceous strata of the British Isles: Clay Minerals, Vol. 41, No. 1, pp. 47–150. Kerr, P. F.; Hamilton, P. K.; and Pill, R. J., 1950, Analytical data on reference clay minerals: API Project No. 49, Report No. 7, 160 p. Kolata, D. R.; Huff, W.D.; and Bergstrom, S. M., 1989, Ordovician K-Bentonites of Eastern North America: Geological Society of America Special Paper No. 313, 84 p. Köster, M. H.; Williams, L. B.; Kudejova, P.; and Gilg, H. A., 2019. The boron isotope geochemistry of smectites from sodium, magnesium and calcium bentonite deposits: Chemical Geology, Vol. 510, pp. 166–187. Luper, E. E., 1972, Smith County Geology and Mineral Resources: Mississippi Geological, Economic and Topographical Survey Bulletin No. 116, pp. 29–42, 60–61. May, J. H., 1974, Wayne County Geology and Mineral Resources: Mississippi Geological, Economic and Topographical Survey Bulletin No. 117. May, J. H. and Warne, A. G., 2004, Hydrogeologic and geochemical factors required for the development of Carolina Bays along the Atlantic and Gulf of Mexico, coastal plain USA: Environmental Engineering Geoscience, Vol. 5, No. 3, pp. 261–270.

McIlwain, J. A., 2007, Hydrogeologic Assessment of a Proposed Reservoir Site, Smith County, Mississippi: Unpublished M.S. Thesis, Mississippi State University, Starkville, 140 p. Mellen, F. F., 1941, Warren County mineral resources: Mississippi Geological Survey Bulletin No. 43, 28 p. Mississippi State University and Pickering Firm, Inc., 2012, Geologic and Hydrologic Investigations for Proposed Smith County Lake, Smith County, Mississippi: Technical Report, 84 p. Moore, W. H., 1965, Hinds County Mineral Resources: Mississippi Geological, Economic and Topographical Survey Bulletin No. 105, pp. 65–67. Morse, H. M.,1934, A Supplemental Report on Bentonite in Mississippi: Mississippi State Geological Survey Bulletin No. 22-A, 26 p. Nelson, W. A., 1922, Volcanic ash bed in the Ordovician of Kentucky, Tennessee and Alabama: Geological Society America Bulletin, Vol. 33, pp. 605–616. Nutting, P. G., 1935, Technical basis of bleaching clay industry: American Association Petroleum Geologists Bulletin, Vol. 19, pp. 1043–1052. Pestitschek, B.; Kurzwell, H.; Gier, S.; and Essa, M., 2012, Effects of weathering on glauconite: Evidence from the Abu Tartur Plateau, Egypt: Clays Clay Minerals, Vol. 60, pp. 76–88. Reynolds, W. R., 1962, The Lithostratigraphy and Clay Mineralogy of the Tampa-Hawthorn Sequence of Peninsular Florida: Unpublished M.S. Thesis, Florida State University, Tallahassee, 126 p. Robert, M., 1972, The experimental transformation of mica toward smectite: Relative importance of total charge and tetrahedral substitution: Clays Clay Minerals, Vol. 21, pp. 167–174. Ross, C. S. and Hendricks, S. B., 1945, Minerals of the Montmorillonite Group: U.S. Geological Survey Professional Paper No. 205-B, pp. 25–79. Ross, C. S. and Shannon, E. V., 1926, Minerals of bentonite and related clays and their physical properties: Journal American Ceramic Science, Vol. 9, pp. 77–96. Shearer, H. K., 1917, Bauxite and Fuller’s Earth of the Coastal Plain of Georgia: Geological Survey of Georgia Bulletin No. 31, pp. 309–311 Swineford, A.; Frye, J.; and Leonard, B., 1955, Petrography of the Late Tertiary volcanic ash falls in the Central Great Plains: Journal Sedimentary Research, Vol. 25, No. 4, pp. 243–261. Wherry, E. T., 1917. Clay derived from volcanic dust in the Pierre Shale of South Dakota: Journal Washington Academy Science, Vol. 7, pp. 576–583. Williamson, D. R., 1986, Lignites of northwest Louisiana and the Dolet Hills lignite mine. In Finkelman, R. B. and Casagrande, D. J. (Editors), Geology of Gulf Coast Lignites: Geological Society of America, Environmental and Coal Associates, Houston, pp. 13–28

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 215–231

231

Technical Note The Yaqui Flat Long Run-Out Rock Avalanche: Anza-Borrego Desert State Park, California MICHAEL W. HART* P.O. Box 261227, San Diego, CA 92196 DAVID B. EVANS Geocon, Inc., 6960 Flanders Drive, San Diego, CA 92121

Key Terms: Avalanche, Long Run-Out, Landslide INTRODUCTION The Yaqui Flat landslide is a previously unrecognized Late Pleistocene long run-out (LRO) rock avalanche, or sturzstrom, located in Anza-Borrego Desert State Park, eastern San Diego County, California (Figure 1). The Yaqui Flat LRO rock avalanche (Yaqui Flat Slide) is situated in Grapevine Canyon near the distal edge of a large alluvial fan emanating from the south side of Pinyon Ridge. We surmise that the landslide was previously unrecognized because of its partial burial by active fan deposition and because of its low proﬁle and slight morphologic resemblance to younger fanglomerate deposits in the area. LRO rock avalanches, or sturzstroms, are a unique type of landslide characterized by LRO distances of as much as 6 mi (Shreve, 1968; Shaller et al., 2020). Their signature and most confounding characteristic is their ability to travel great distances over level to gently sloping terrain seemingly in deﬁance of the laws of physics. LRO rock avalanches can also be distinguished from other types of landslides in that they are detached from their source area and typically have a boulder cap overlying a layered debris lobe that often preserves the source stratigraphy. Perhaps the most famous and one of the best-preserved examples this type of landslide is the Blackhawk Landslide located in Lucerne Valley on the north side of the San Bernardino Mountains of southern California (Shreve, 1968). Similar LRO landslides in southern and eastern California have been described in Eureka Valley of Death Valley National Park (Shaller et al., 2020), the Split Mountain area of San Diego County (Robinson and Threet, 1974; Abbott et al., 2003), Chalfant Valley, California (Bishop, personal communication), and Vallecito Valley, San Diego County (Hart, 2003).

*Corresponding author email: mwHart40@gmail.com

The mechanics of LRO rock avalanches have been discussed by various authors, but the best explanation of the sliding mechanism for the type of LRO landslides that occur on alluvial fans was set forth in papers by Yarnold and Lombard (1989) and Shaller et al. (2020); that is, they slid as an semirigid layered body over a saturated, basal substrate of alluvium that was temporarily liqueﬁed by impact. The evidence for this is that these features often exhibit clastic dikes that project upward from the failure zone through the lower matrix-rich facies of the landslide. Additionally, it is the nature of the lowermost landslide layer itself, a zone of matrix-rich breccia and clay-sized fractions giving the appearance of a mud slurry replete with ﬂow banding and slip surfaces that formed as the slide came to a halt. These features are usually apparent wherever the base of LRO landslides are exposed (Yarnold and Lombard, 1989).

Figure 1. Location and general geologic map of Yaqui Flat Area with the Yaqui Flat long run-out landslide shown in dark blue (lat. 33.135564, long. −116.421949). Geologic map modiﬁed after Wagner (1997). Q = alluvium; Qf = alluvial fans; Qtls = ancient landslides; pKgn = prebatholithic rocks; pKbc = Bitter Creek Canyon metamorphic rocks. (inset; ABDSP: Anza-Borrego Desert State Park).

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 233–236

233

Hart and Evans

Figure 2. Google Earth image of the Yaqui Flat landslide. The Eureka Valley long run-out rock avalanche shown in inset (view to southeast with the Last Chance Range at top of the image). Total width of the Yaqui Flat landslide is 1,800 ft (563 m). Width of the Eureka Valley landslide is 1,500 ft (469 m).

DISCUSSION The Yaqui Flat slide detached from granitic and metamorphic terrain on the southern slopes of Pinyon Ridge, which forms the northwest side of Grapevine Canyon, and ran out approximately 1 mi (1.6 km) to the south over an alluvial fan. The geologic map of Tubb Canyon (Wagner, 1997) indicates that the southern flanks of Pinyon Ridge are populated by massive bedrock landslides and extensive alluvial fans that project southward onto Yaqui Flat (Figures 1 and 2). These large, apparently translational landslides involve Jurassic metasediments, gneiss, and plutonic rocks consisting of coarse-grained biotite-rich granodiorite. The Yaqui Flat slide first became apparent during inspection of Google Earth images on which the feature stood out from the surrounding alluvial fan by its unique shape and difference in color from the surrounding alluvium (Figure 2). This color difference is due to the unique rock types making up the two layers of the landslide: a lower layer consisting of orangebrown, pre-batholithic metasediments (mostly schist and gneiss) and an upper light gray cap consisting of granodiorite boulders, some of which occur in 40–50ft-wide outcrops. Wind erosion and weathering of the lower layer have produced a desert pavement consisting of angular schist and gneiss clasts whose orangebrown color is readily distinguished on aerial imagery from the light gray of the adjacent alluvial fan and boulder cap. The layered nature of the landslide deposit is well expressed by the locally near-horizontal contact between the two rock types shown on the geologic map (Figure 3) and in an outcrop (Figure 4). Because of the Yaqui Flat slide’s layered stratigraphy,

234

Figure 3. Geologic map of the eastern lobe of the Yaqui Flat long run-out rock avalanche. Qal = alluvium; Qfg = fanglomerate; Qmb = metamorphic breccia; Qgbc = granodiororite boulder cap. (A) Location of Figure 4. (B): Location of Figure 6. (C): Location of Figure 5. Map base is color enhanced high-angle oblique drone photograph by David Evans. Total width of eastern lobe is 1,100 ft (563 m).

we infer that the slide likely originated from one of the large translational landslides located at the base of Pinyon Ridge to the north that in some locations exhibit a similar relationship between the granodiorite and metasediments that was eventually preserved during the long run-out. The landslide has been separated by active fan deposition into two primary segments or lobes and a small outlier located near the northwest corner of the larger eastern lobe. The western lobe has been buried to a greater degree than the eastern lobe and exhibits only the uppermost portion of the granodiorite cap and limited areas of the underlying metamorphic breccia.

Figure 4. View to east of granodiorite cap rock overlying metamorphic breccia. Location A in Figure 3. Qgbc = granodiorite boulder cap; Qmb = metamorphic breccia (rock hammer, bottom left center for scale).

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 233–236

Yaqui Flat Rock Avalanche

The eastern lobe has retained the shape of a classic LRO landslide: spatulate outline and somewhat bulbous oversteepened front. In addition, the eastern lobe exhibits several good outcrops of breccia that are not present on the western lobe. The Yaqui Flat Slide measures approximately 1800 ft (563 m) in width and has an exposed length of 2,400 ft (750 m). The current width of the slide has not likely been significantly affected by being partially buried by fan deposits; however, its true length is not known with certainty since it has been shortened an unknown amount by burial. Lateral levees that likely once defined the path of the slide from its source, characteristic of this type of feature, have long since been buried or removed by erosion. The inset on Figure 2 is an image of the Eureka Valley landslide (Shaller et al., 2020) that depicts what the Yaqui Flat slide probably looked like before partial burial occurred. Stream cuts along the right flank of the Eureka Valley landslide also provide an excellent example of preservation of source area stratigraphy where much-attenuated, multicolored beds of Paleozoic limestones are beautifully preserved in their original stratigraphic order. Shaller et al. (2020) estimate that the Eureka Valley landslide has a minimum age of 85–102 ka. This estimate is based on the presence of well-developed desert pavement and dark brown varnish on the noncarbonate clasts that characterize the alluvial surfaces overlying the landslide. The lack of strong carbonate cementation in the breccia deposits limits the age to roughly 100 ka. The breccia deposits of the Yaqui Flat slide (located in a desert environment similar to the Eureka Valley landslide) are moderately well cemented with calcium carbonate, and in some locations, a thin calcrete horizon has developed and is exposed by erosion near the surface. In addition, desert pavement consisting primarily of metamorphic rock fragments is well developed on the more gently sloping to level areas on the breccia layer. Therefore, we infer that the Yaqui

Figure 6. Thin white dike in outcrop of granodiorite boulders in cap rock. Note that the dike has not been signiﬁcantly oﬀset between boulders during transport by the landslide. Location B in Figure 3. Arrows depicts west and east ends of the dike. Total distance between arrows is 40 ft (12.5 m). Boulder at bottom left is 3 ft (1 m) wide.

Flat slide is significantly older than the Eureka Valley landslide and may have occurred during the pluvial period (marine isotope stage 6) ending around 123 ka. Figure 4 shows an outcrop of granodiorite cap rock that clearly overlies the metamorphic breccia. Figure 5 shows an example of jigsaw and crackle microbreccia (pervasively fractured rock with no or little matrix separating the clasts) observed in gullies near the toe of the slide. At location B in Figure 3 is a thin white dike that extends essentially uninterrupted and only slightly misaligned across an outcrop 40 ft (12.5 m) long on the boulder cap (Figure 6). This can be explained only if the boulder cap traveled as an essentially rigid layer on the underlying breccia. CONCLUSION There are three lines of evidence that lead to our conclusion that the breccia and boulder feature in Yaqui Flat is a partially buried LRO rock avalanche: first, the near horizontal contact relationship between the granodiorite boulder cap and metamorphic breccia; second, the microcrackle and jigsaw breccia observed in the underlying breccia facies; and, finally, the presence of the thin white dike that extends essentially uninterrupted and only slightly misaligned across an outcrop 40 ft (12.5 m) long on the boulder cap that can be explained only by transport in an LRO rock avalanche. REFERENCES

Figure 5. Close-up of breccia outcrop showing microcrackle breccia at arrow. Note preserved relict bedding. Location C in Figure 3.

Abbott, P. L.; Borron, S. E.; and Washburn, J. L., 2003, Sturzstrom deposit caprocks. In Murbach, M. and Hart, M.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 233–236

235

Hart and Evans W. (Editors), Geology of the Elsinore Fault, San Diego Region: Sunbelt Publications, El Cajon, CA, pp. 77–84. Bishop, K. M., 2009, personal communication, California State University, Los Angeles. Hart, M. W., 2003, Landslides in the peninsular ranges, southern California. In Murbach, M. and Hart, M. W. (Editors), Geology of the Elsinore Fault, San Diego Region: Sunbelt Publications, El Cajon, CA, pp. 193–209. Robinson, J. W. and Threet, R. L., 1974, Geology of the Split Mountain area, Anza Borrego Desert State Park, eastern San Diego County, California. In Recent Geologic and Hydrologic Studies Eastern San Diego County and Adjacent Areas: Sunbelt Publications, El Cajon, CA, pp. 47–56.

236

Shaller, P. J.; Doroudian, M.; and Hart, M. W., 2020, The Eureka Valley landslide: Evidence for a dual failure mechanism for a long run-out landslide: Lithosphere, Vol. 2020, pp. 1–26. Shreve, R. L., 1968, The Blackhawk Landslide: Geological Society of America Special Paper 108, 47 p. Wagner, D. L., 1997, Geologic Map of the Tubb Canyon 7.5 Quadrangle, San Diego County, California: DMG Open File Report 96-06. Yarnold, J. C. and Lombard, J. P., 1989, A facies model for large rock avalanche deposits found in dry climates. In Colburn, I. P.; Abbott, P.; and Minch, J. (Editors), Conglomerates in Basin Analysis: A Symposium Dedicated to A. O. Woodford: Society of Economic Paleontologists and Mineralogists, pp. 9–31.

Environmental & Engineering Geoscience, Vol. XXVIII, No. 2, May 2022, pp. 233–236

Oommen, Thomas Board Chair, Michigan Technological University Sasowsky, Ira D. University of Akron

Environmental & Engineering Geoscience May 2022

VOLUME XXVIII, NUMBER 2

Submitting a Manuscript

THIS PUBLICATION IS PRINTED ON ACID-FREE PAPER EDITORS

Eric Peterson Department of Geography, Geology, and the Environment Illinois State University Normal, IL 61790 309-438-5669 ewpeter@ilstu.edu

Karen E. Smith, Editorial Assistant, kesmith6@kent.edu

Volume XXVIII, Number 2, May 2022

Abdul Shakoor Kent State University Kent, OH 44242 ashakoor@kent.edu

ADVISORY BOARD Watts, Chester “Skip” F. Radford University Hasan, Syed University of Missouri, Kansas City Nandi, Arpita East Tennessee State University

ENVIRONMENTAL & ENGINEERING GEOSCIENCE