1. Royal Military Academy
Communication, Information, Systems & Signals Department
Christchurch earthquakes ground motion attenuation sites analyzed with COSMO-SkyMed images
IG29 - Theory and Applications of Synthetic Aperture Radar
This work results from an international collaboration between…
Damien Closson Royal Military Academy (Belgium)
Najib Abou Karaki The University of Jordan (Jordan)
Nada Milisavljevic Royal Military Academy (Belgium)
Azzedine Bouaraba École Militaire Polytechnique (Algeria)
Paolo Pasquali Sarmap SA (Switzerland)
AOGS 2014 / IG29-D5-PM2-CA-001

2. Outline Objectives, Strategy & Hypothesis Material & Method
Results & Analysis
Conclusion
The outline of this talk is first setting up the goal, then…
AOGS 2014 / IG29-D5-PM2-CA-001

3. Objectives Identify Delineate
Interpret
potential earthquake ground motion attenuation sites in Christchurch (NZ) from VHR radar images CSK
AOGS 2014 / IG29-D5-PM2-CA-001

4. Strategy and hypothesis
Map surface deformations
Map surface changes
Repeat for several earthquakes
Compare results
Potential* ground motion attenuation sites
* Validation will be performed from soil mechanic data, superficial geology etc
Hypothesis
Spaceborne radar interferometry techniques applied to VHR radar images are able to highlight potential ground motion attenuation sites owing to a combination of differential interferograms (surface deformations), coherence, and amplitude images (surface changes).
AOGS 2014 / IG29-D5-PM2-CA-001

5. Method applied to the Canterbury EQ sequence
SLC Intermediary products Added value product
Image 1
DInSAR 23
Attenuation sites
Iteration 1
Coh12 + Coh23
Image 2
A1 + A2 + A3
EQ1
EQ-1
Amplitude change
Surface changes
Image 3
Coherence change
Potential
attenuation sites
Image 4
DInSAR 45
From img 1&2 we derive 3 different products: …
From img 2&3 we compute coherence change map
Then, from interferograms and coherence we deduce A.S.C. and from A chg and Cc chg realize a classification of the ground modifications
Ground modifications characterize A.S.C.
Contextual data are there to validate or reject the candidates
We apply the same method to several tremors and we compare the results to deduce the boundaries
Attenuation sites
Iteration 2
validation
Coh45 + Coh46
Image 5
Contextual data (geology, etc)
A4 + A5 + A6
EQ2
EQ-2
Amplitude change
Surface changes
Image 6
Coherence change
AOGS 2014 / IG29-D5-PM2-CA-001

6. Canterbury earthquake sequence
Source: the Internet – February 22nd, 2011
September 4th, 2010 : « Darfield » event
7.1 ML
Location 40 km west of Christchurch
Focal Depth 10 km
February 22nd, 2011 : « Lyttelton » event
6.3 ML
Location 10 km south of Christchurch
Focal Depth 5.9 km
June 13th, 2011 event
6.4 ML
Location 10 km east of Christchurch
Focal Depth 6.9 km A S O N D J F M
The main shock happened in september It was followed by two important aftershock in feb and June 2011.
The first aftershock was very destructive because of the shallow depth.
AOGS 2014 / IG29-D5-PM2-CA-001

7. Material: image datasetO N D J F M
2 Envisat 4 2 3
CSK time series
Earthquakes
Acquisition
Satellites
Mode
Freq.
I.A.
T.B.
N.B.
A.A.
P.S.
04-Sep-10
“Darfield”
04-Jun-10
Envisat
Asc.
5.33 23
105
114
84.5 20
17-sep-10
22-Feb-11
“ Lyttelton”
19-Feb-11
COSMO SkyMed
9.6 36 4 28
295
2.3
23-Feb-11
13-Jun-11
23-May-11
Desc. 32
357.5 26
24-Jun-11
Asc. = ascending; Desc. = descending; Freq. = frequencies (GHz); I.A. = incidence angle (degrees); T.B. = temporal baseline (days); N.B. = normal baseline (meters); A.A. = altitude of ambiguity (meters); P.S. = pixel sizes (meters)
AOGS 2014 / IG29-D5-PM2-CA-001

8. Results : differential interferograms (Linwood)
Topo
Sep10
Feb11
Jun11
This slide shows the differential interferograms for the main shock and the two aftershocks.
The black dots are GPS tracks to help you to compare the data.
The interferogram of the Feb event shows lot of noise while the two others are less perturbed.
AOGS 2014 / IG29-D5-PM2-CA-001

9. Results : differential interferograms (Linwood)
Source: Sarmap
Cosmo-SkyMed X-band satellite (ASI)
Pre-event image: 19/02/2011
Post-event image: 23/02/2011
Perp. baseline: 22 m
Fringe displacement: 1,5 cm
AOGS 2014 / IG29-D5-PM2-CA-001

10. Results : GPS data interpolation (Linwood)
Source: GNS Science
AOGS 2014 / IG29-D5-PM2-CA-001

11. Results : potential attenuation site (Linwood)
Source: Sarmap & GNS Science
Cosmo-SkyMed X-band satellite (ASI)
Pre-event image: 19/02/2011
Post-event image: 23/02/2011
Perp. baseline: 22 m
Fringe displacement: 1,5 cm
AOGS 2014 / IG29-D5-PM2-CA-001

12. Results : coherence (Linwood)
Source: Sarmap
Cosmo-SkyMed X-band satellite (ASI)
Pre-event image: 19/02/2011
Post-event image: 23/02/2011
Perp. baseline: 22 m
0=change =no change
AOGS 2014 / IG29-D5-PM2-CA-001

13. Analysis: specific ground conditions - Burwood
Background: topo map 1:50,000
Source:
Burwood Landfill [source: Dr Leonid Itskovich]
90 hectares
Up to 25 m deep
Up to 264,000 tonnes of refuse per year
60% organic waste
Operated 1984 – 2005 (inactive)
AOGS 2014 / IG29-D5-PM2-CA-001

14. Analysis: classification of surface changes
SLC product product product 3
Image 1
Coh12
Coh changes
Coh23
Before -> after
AL -> AH
AH -> AL
AL -> AL
AH -> AH
Image 2
Validation with aerial photographs
Surface modifications A1
EQ-1
From img 1&2 we derive 3 different products: …
From img 2&3 we compute coherence change map
Then, from interferograms and coherence we deduce A.S.C. and from A chg and Cc chg realize a classification of the ground modifications
Ground modifications characterize A.S.C.
Contextual data are there to validate or reject the candidates
We apply the same method to several tremors and we compare the results to deduce the boundaries A2
A changes
Before -> after
CL -> CH
CH -> CL
CL -> CL
CH -> CH
Image 3 A3
AOGS 2014 / IG29-D5-PM2-CA-001

15. AOGS 2014 / IG29-D5-PM2-CA-001 (A1H, A2H, C1L, C2L)
Diffuse surface remaining
diffuse (e.g., non-flat
roofs not being
affected by the
earthquake)
roof, garden, vertical walls, waste
roof tilted towards the E and W
(A1H, A2H, C1L, C2L)
Diffuse surface remaining diffuse
(e.g., non-flat roofs not being
affected by the earthquake)
water
(A1L, A2L, C1L, C2L)
Specular surface such
as water, mud flat areas
(A1L, A2L, C1H, C2H)
Specular surface such as a
roof or low-traffic smooth streets
not affected by the earthquake
(A1L, A2L, C1L, C2L)
Specular surface such
as water, mud flat areas
AOGS 2014 / IG29-D5-PM2-CA-001
mud
Parking place

16. + + Conclusion earthquakes attenuation Soil amplification
Linwood
earthquakes
attenuation
Soil amplification
Ground shaking
inventory
Economic value
Physical damage
demography
Economic loss
casualties
input
output
Model:
Source: selena.sourceforge.net +
AOGS 2014 / IG29-D5-PM2-CA-001

17. Prospect for the future
We expect experiment this approach in other areas prone to earthquakes and we are looking for partnerships
The cost of the research should decrease in the near future owing to the SENTINEL-1 data (pixel 5x5m, C-band, revisit time 24 days then 11 days when two satellites) – first image forecasted in September/October 2014
… questions?
AOGS 2014 / IG29-D5-PM2-CA-001

18. AOGS 2014 / IG29-D5-PM2-CA-001 Colour Meaning Possible situations
(40.3%)
A1L, A2L, C1L, C2L
Specular surface such as water, mud flat areas, trees, shrubs
32445 (3.5%)
A1L, A2L, C1H, C2L
Specular surface such as a roof or low-traffic smooth streets affected by the earthquake (e.g., by liquefaction)
6963 (0.7%)
A1L, A2L, C1L, C2H
Specular surface such as a roof or high-traffic smooth streets affected by the earthquake
49778 (5.3%)
A1L, A2L, C1H, C2H
Specular surface such as a roof or low-traffic smooth streets not affected by the earthquake
29749 (3.2%)
A1H, A2L, C1L, C2L
Diffuse surface becoming specular due to liquefaction, but remaining highly active
12042 (1.3%)
A1H, A2L, C1H, C2L
Diffuse surface becoming specular due to liquefaction, not active before the earthquake, active after it (e.g., low traffic road becoming active due to mud cleaning operations)
1625 (0.2%)
A1H, A2L, C1L, C2H
Diffuse surface becoming specular due to liquefaction, active before the earthquake, not active after it (high-traffic rough road becoming smooth due to mud, not used because of the mud)
22277 (2.4%)
A1H, A2L, C1H, C2H
Rough surface becoming specular with no activity before nor after (low traffic road affected by liquefaction, e.g.)
60881 (6.5%)
A1L, A2H, C1L, C2L
Smooth surface becoming rough and being active before and after (such as a smooth high-traffic street becoming rough due to fractures, and being used before and after)
12498 (1.3%)
A1L, A2H, C1H, C2L
Specular surface such as a roof becoming rough (due to a collapse of chimney, e.g.) and not being active before while it loses coherence as it keeps on collapsing
3770 (0.4%)
A1L, A2H, C1L, C2H
Specular surface becoming rough and being active before while not being active after (smooth high-traffic street becoming rough due to fractures and not being used after)
24060 (2.6%)
A1L, A2H, C1H, C2H
Specular surface such as a roof becoming rough (due to a collapse of chimney, e.g.) and not being active before nor after
67808 (7.2%)
A1H, A2H, C1L, C2L
Diffuse surface remaining diffuse and being active before and after (e.g., non-flat roofs not being affected by the earthquake)
39555 (4.2%)
A1H, A2H, C1H, C2L
Diffuse surface such as a rough road affected by liquefaction
7217 (0.8%)
A1H, A2H, C1L, C2H
Surface remained rough, but the angle of the illuminated surface changed so that the coherence increased
(20.1%)
A1H, A2H, C1H, C2H
Surface remained rough, not being affected by the earthquake
AOGS 2014 / IG29-D5-PM2-CA-001