Department of Earth Sciences
CENTRE OF COMPETENCE OF THE CIVIL PROTECTION DEPARTMENT PRESIDENCY OF THE COUNCIL OF MINISTERS
Monitoring deformations through Ground-Based radar interferometry N. Casagli, F. Catani, G.Luzi, L. Guerri: Univ. Florence, Earth Sciences Department
D. Tarchi: EU Commission JRC 窶的PSC D. Leva: LisaLab Ltd A JRC spin-off company
Presentation outline • Introduction to Ground Based radar interferometry • The LiSA GB SAR system at Stromboli • The 2003 data collection: • effusion phase • 5 April explosion
• The 2007 data collection: • inflation in the pre-effusive phase • bulging and vent opening • 15 March explosion
A radar transmits and receives e.m. wave in the Microwave portion of the em. spectrum
Ku band
Îťâ‹… f =c
Radar signal is slightly affected by atmospheric propagation MW sensors work when the optical ones are blind
Low attenuation
•
In Remote Sensing we want to observe natural surfaces with fine spatial resolution.
•
Conventional RADAR were born for detecting and ranging targets and suffer from a coarse spatial resolution limited by antenna dimension and radar-target distance The map
A target Radar •
Synthetic Aperture Radar SAR is a tool to collect radar images from large distances with high spatial resolution: a few meters from hundreds of kilometres.
Basic interferometry A wave travels transferring the configuration of a physical parameter (e.m field, sound, a seismic deformation..…) with a velocity c. λ It is characterized by: A amplitude A, wavelength λ, frequency f and phase θ and can generates specific phenomena as: interference and diffraction c= light velocity ~ 3.108 m/s
f=number of cycles per time unit To perform interferometry waves need a stable phase wave
wave
λ⋅ f =c Phase strongly affects waves interaction
interference
Two waves “in phase” yield high intensity Two waves “in quadrature” yield a null intensity In phase
Every half wavelength displacement ambiguity arises Fringes are an alternating of bright and dark lines due to interference: phase ranges only from –π to+ π
Quadrature To retrieve phase value exceeding half phase Cycle, measured phase must be unwrapped.
When errors are negligible a simple relationship between the displacement and the measured phase:
λ ΔR = Δϕ 4π
SAR imaging: the Ground Based approach Azimuth resolution
Pk Rn,k ΔXa (Np step)
n
Azimuth length: L
Azimuth resolution
Main benefits from GB observations:
Range resolution
• Accurate motion • Frequency of the observation very high (<hour) • Zero spatial baseline among images
c 2B λ Δϑ = 2L
ΔR =
B = bandwidth L = scan length λ = wavelength
The SAR power image of a brigth spot (an oriented metal disk) Focalization alghorythm
1 I ( Pn ) = n f np
np
2 nk
nf
∑R ∑E k =1
ik e
(
4πf i ( Rnk o )) c
= I (n) e jϕ ( n )
i =1
Focalization improves azimuth resolution Transceiver: a Continuous-wave stepped-frequency (CW-SF) radar based on a Vectorial Network Analyser (VNA)
A SAR image is represented through a matrix of complex numbers RADAR
I i ,l = ai ,l e
⎡ 4π ⎤ j ⎢ Ri ,l −Φ noise ⎥ ⎣ λ ⎦
= si ,l e jϕ
2d Amplitude image
The amplitude of a pixel sil is related to backscattering of the illuminated area The phase ϕ is related to propagation path + noise due to other factors. When a Digital Elevation Model is available we can project the radar image on it 3d Amplitude image
Master Image
M(t , R) = a1i e
−j
4π
λ
R1i
Slave Image
S(t , R) = a2i e
−j
4π
λ
R2 i
*
si = a1,i a2,i e
Topography
B≠0
⎡ 4π ⎤ j ⎢ ( R2 i − R1i ) − Φ noise ,i ) ⎥ ⎣ λ ⎦
= si e jϕ
Displacement
Δφ∝(4π/λ) hr
B~0
B=Baseline
R
Pi Interferometric Phase φ:
φ=arg[MS*] Interferometry provides sub-wavelength sensitivity
Δφ=(4π/λ) ΔR
Interferogram
Phase
*
Time= t0
= Time= t0+Δτ Zero baseline
Measured differential phase can be affected by some “decorrelation sources”. In GB case :
φ = φinstrument al + φscattering + φ geometric + φ atmosferic + φ displacement The amplitude of the conjugate product is related to coherence
Phase wrapping and decorrelation Phase Wrapping: Large displacement generates fringes
Decorrelation: Rapid motion causes “salt & pepper” texture
Coherence, Γ, ( 0 <Γ <1) gives an estimate of the error in measured differential phase
Interferometric Ground-based Imaging Deformeter Linear Synthetic Aperture Radar
InGrID-LiSA
Ingrid Bergman on the set of the movie â&#x20AC;&#x153;Stromboliâ&#x20AC;? (1949)
2002-2003 eruption
Landslides on SDF on 30 December 2002
Photo INGV Catania (2003) Courtesy of Sonia Calvari
Data collection Heliplatform centre Optical cable
Wireless connection
Heli-platform
Radar installation
16
System set-up
17
Ground-based InSAR Continuous-wave stepped-frequency (CW-SF) radar based on a Network Analyser (NWA) operating in the frequency band 17.0-17.1 GHz
target area
Tx
Rx
sled
linear rail 2.8m
source NWA
:ď&#x20AC;ş
computer
The synthetic aperture is obtained sliding the antennas along a linear rail European Commission Joint Research Centre
Measurement parameters • Frequency range: 17.0 – 17.10 GHz • Frequency points : 1601 • Polarization: VV • transmitted power: 300 mW (25 dBm)
• Synthetic Aperture: 3.0 m • Step: 5 mm • Azimuth points : 601 • Time range: 12 min • Image number: ca. 120 per day
• distance: 650 m • Spatial Resolution: 1.0 m x ca. 1.5 m • Accuracy: < 0.5 mm
Acquisition of raw data
Network Analyzer
First step:
Second step:
Interferometry Image 1 Interferogram (phase difference)
3
Image 2
1
2
1: Flank of Sciara del Fuoco (stable) 2 and 3: Sciara del Fuoco slope 4 and 5: crater
LOS displacement (mm)
4
phase wrapping
5
Observed scenery February 2003
5
4 3
2 1
Lava flow No.3 Lava flow No.2
Lava flow No.1
Power image HIGH REFLECTIVITY
5 4 3 1 2
LOW REFLECTIVITY SHADOW
SdF velocity history since 2003
Negative velocity = shortening
Positive velocity = lengthening
Crater velocity history since 2003
Negative velocity = shortening
Positive velocity = lengthening
SdF velocity history since 2003 eruption
eruption
Crater velocity history since 2003 eruption
eruption
Lava flows
Feb. 2003
www.ct.ingv.it (Jan 2003)
www.ct.ingv.it (Jan 2003)
Interferogram 12’
LOS displacement (mm)
Lava flow 3 mm in 12’ (15 mm/h)
Interval: 12’ Start: Febr. 21 15:21 End: Febr. 21 15:33
Interferogram 12â&#x20AC;&#x2122;
Rapid lava flow (decorrelated)
Interval: 12â&#x20AC;&#x2122; Start: April 1 07:47 End: April 1 07:59
Interferogram 1h
Slope movement (1.5 mm/h)
LOS displacement (mm)
Rapid lava flow
Interval: 1h Start: March 1 07:36 End: March 1 08:36
Interferogram 1h
Slope movement (2.2 mm/h)
LOS displacement (mm)
Rapid lava flow
Interval: 1h Start: April 1 15:00 End: April 1 16:00
Interferogram 1h
LOS displacement (mm)
Slope movements disturbed by lava flows
Interval: 1h Start: April 27 12:05 End: April 27 13:05
Interferogram 1h
LOS displacement (mm)
Slope movements disturbed by lava flows
Interval: 1h Start: December 01 12:05 End: December 01 13:05
Interferogram 1h
Lava flows
LOS displacement (mm)
Slope movement 17 mm in 1h50â&#x20AC;&#x2122; (2.2 mm/h)
Interval: 1h 50â&#x20AC;&#x2122; Start: Febr. 21 19:10 End: Febr. 22 03:02
March 2
δ
α β
February
April, 25
Slope movements on SdF
Slope movements on the crater
Rockfalls inside the crater
Interferogram 12h
Slope movements disturbed by lava flows (phase ambiguity)
Landslides
LOS displacement (mm)
Slope movement (3 mm/ day)
Interval: 12h Start: April 13 01:20 End: April 13 13:20
Interferogram 24h
Slope movements disturbed by lava flows (phase ambiguity)
Diffused landslides
LOS displacement (mm)
Slope movement (3 mm/ day)
Interval: 24h Start: April 11 03:55 End: April 12 13:55
Interferogram 48h Rockfalls
Diffused landslides
Decorrelation due to lava flows
LOS displacement (mm)
Slope movement 6 mm in 48h (3 mm/day)
Interval: 48h Start: March 30 13:45 End: April 01 13:45
Interferogram 75h 07â&#x20AC;&#x2122; Rockfalls
Decorrelation
LOS displacement (mm)
Diffused landslides
Slope movement 7 mm in 75h (2 mm/day)
Interval: 75h 07â&#x20AC;&#x2122; Start: April 26 05:03 End: April 29 08:10
Interferogram 7d 0h 05â&#x20AC;&#x2122; Slope movement (0.9mm/d) LOS displacement (mm)
Decorrelation
Interval: 7 days 05â&#x20AC;&#x2122; Start: December 01 13:54 End: December 08 13:59
Interferogram 12d 9h 45â&#x20AC;&#x2122;
Decorrelation
LOS displacement (mm)
Slope movement (phase ambiguity)
Interval: 12 days 09h 45â&#x20AC;&#x2122; Start: April 20 19:16 End: May 05 08:10
Long period cumulated sequence Start: 26/2 16.00 End: 28/2 09.30 Total interval: 41h 30min Interval between images: 36 min Max displacement: 35 mm Max speed: 0.84 mm/h
Deformation map on DTM
Interferogram on DTM
Explosion of 5 April 2003
08.12 GMT+1
3m
Before
After
Interferogram across the explosion
LOS displacement (mm)
Decorrelation over all the interferogram
Interval: 1h 52â&#x20AC;&#x2122; Start: April 5 07:52 End: April 5 09:44
Ground-shacking effect
Interferogram after the explosion Slope movement (4 mm/h) LOS displacement (mm)
Slope movement (3.2 mm/h)
Interval: 1h 24â&#x20AC;&#x2122; Start: April 5 09:20 End: April 5 10:44
Explosion SAR sequence Interval: 06â&#x20AC;&#x2122; Phase in degrees
Start: April 5 08:07 End: April 5 08:13
2007 eruption
27th February: lava effusion from the crater
27th February vent opening (400 m a.s.l. )
Target area (2007)
Target area (2007)
11 Jan. 2007 47 days before the eruption Interval: 24h 12â&#x20AC;&#x2122; Start: 13.19 GMT 2007/01/10 End: 13.32 GMT 2007/01/11 Crater velocity: 0.04 mm/h Sciara velocity: 0.035 mm/h
26 Jan. 2007 32 days before the eruption: acceleration in the crater area Interval: 24h 07â&#x20AC;&#x2122; Start: 08.08 GMT 2007/01/25 End: 08.15 GMT 2007/01/26 Crater velocity: 0.10 mm/h Sciara velocity: 0.027 mm/h
15 Feb. 2007 12 days before the eruption: acceleration in the Sciara del Fuoco Interval: 15h 32â&#x20AC;&#x2122; Start: 02.19 GMT 2007/02/15 End: 17.51 GMT 2007/02/15 Crater velocity: 0.50 mm/h Sciara velocity: 1.50 mm/h
27 Feb. 2007 Eruption Sequence of 11’ ITF Interval: 14h 41’ Start: 00.11 GMT 2007/02/27 End: 14.52 GMT 2007/02/27
05.53 GMT
27 Feb. 2007 Power images Morphological modifications of the crater and of the upper Sciara del Fuoco Interval: 15h 12â&#x20AC;&#x2122;
21.05 GMT
Upper: 05.53 GMT 2007/02/27 Lower: 21.05 GMT 2007/02/27
Inverse velocity method Impossibile visualizzare l'immagine. La memoria del computer potrebbe essere insufficiente per aprire l'immagine oppure l'immagine potrebbe essere danneggiata. Riavviare il computer e aprire di nuovo il file. Se viene visualizzata di nuovo la x rossa, potrebbe essere necessario eliminare l'immagine e inserirla di nuovo.
1 α −1
v f = ∞ ⇒ v1i = [A(α − 1)]
if α = 2 then: 1/v = A(tf-t)
Fukuzono (1985)
(t f − t )
1 α −1
ERUPTION AND LANDSLIDES
Inverse velocity plot – 27 February
Landslides of Feb.27
10.57 - 13.37 GMT
LOS displacement (mm)
14.30 - 14.41 GMT
2007/02/27
13.48 - 14.09 GMT
14.41 - 14.52 GMT
71
18.46 - 18.57 GMT
LOS displacement (mm)
cumulated 18.57- 19.29 GMT
2007/02/27
19.08- 19.19 GMT
19.40 -19.51 GMT
72
19.51 - 20.01 GMT
20.44 - 20.55 GMT
20.33 - 20.44 GMT
21.37 - 21.48 GMT
73
8-9 March: lava effusion from new vent
8-9 March 2007 Opening of a 2nd vent
Sequence of 1h ITF Interval: 32h 13â&#x20AC;&#x2122; Start: 11.26 GMT 2007/03/08 End: 19.39 GMT 2007/03/09
8-9 March 2007: Opening of new vent
Time interval of 11 minutes (11.17-11.28 UT 9 March 2007) velocity greater than 300 mm/h
09/03/2007 15.07
0,002
09/03/2007 14.24
0,003
09/03/2007 13.40
09/03/2007 12.57
09/03/2007 12.14
09/03/2007 11.31
09/03/2007 10.48
09/03/2007 10.04
09/03/2007 09.21
09/03/2007 08.38
09/03/2007 07.55
09/03/2007 07.12
inverse of velocity 1/(mm/h)
Inverse velocity plot 9 March 0,01
0,009
0,008
0,007
0,006
0,005
0,004
vent opening and landslides
0,001
0
Thermal camera (750m a.s.l.) on 2007/03/15 20.40 GMT Geophysics Laboratory â&#x20AC;&#x201C; Department of Earth Sciences
79
15 March 2007 Explosion (paroxism) Sequence of 10’ ITF Interval: 1h 47’ Start: 18.49 GMT 2007/03/15 End: 20.36 GMT 2007/03/15
9 March effusion
27 Feb. effusion and landslides
SdF velocity in 2007
15 March explosion
27 Feb. effusion and landslides
Crater velocity in 2007
Conclusions RADAR monitoring works in any time/weather condition but spatial resolution is strongly dependent on antenna dimensions and distance
The SAR technique allows to collect radar images from large distances with high spatial resolution
Images acquired in different times can be used if the elapsed time is short compared to surface modifications rate The differential phase contains information on path modification occurred along the LOS between two acquisitions A surface deformation corresponds to a distance (range) variation
Needs a coherent radar
phase is stable in time & space Interferometry is possible
Deformation Maps can be obtained
GB SAR fast acquisition
Conclusions • Ground-based In-SAR technique is used for real-time and early-warning for forecasting eruptions, landslides and related tsunami • The tecnique permits a costant and continuous monitoring in all weather and environmental conditions • The inverse velocity method (Fukuzono) is used as model for forecasting the failure time
• From space the knowledge of the orbital parameters, the registration of coherent echo, and e.m. fields reconstruction algorithms, make available MW images for RS use. • Phase information from pair of images makes available submillimetric sensitivity to surface range variations. Great hopes arose on DinSAR from space for: topography (DEM generation), glaciers dynamics, landslides monitoring, subsidence studies • ... but in real world accuracy is a little far from expected value: main drawback: revisiting time too long for many applications: temporal decorrelation (repeat pass)
Ground Based DInSAR can offer down to 10 minutes “repeat pass” !
Synthetic Aperture Radar techinque can provide MW images with adequate spatial resolution Some SAR images from space
Oil spill detection -> <- Flooded areas mapping Different areas
To perform it we need: • spatial/temporal diversity: a motion as the orbit or a linear scan (GB SAR) • a coherent radar