This document provides an overview of SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform. It begins with an introduction to differential SAR interferometry and examples of its applications in measuring centimeter-level ground deformations related to volcanoes, earthquakes, and landslides. It then discusses the SBAS algorithm for generating time series of surface deformation from networks of interferograms. Finally, it describes the ESA's Thematic Exploitation Platforms including the Geohazard Exploitation Platform, which provides computing resources and tools for automated DInSAR processing and analysis to support geohazards research.
SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
1. SBAS-DInSAR processing on the ESA
Geohazard Exploitation Platform
Claudio De Luca
deluca.c@irea.cnr.it
Istituto per il Rilevamento Elettromagnetico dell’Ambiente (IREA)
Consiglio Nazionale delle Ricerche (CNR),
Via Diocleziano, 328, 80124 Napoli
2. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Agenda
n Part 1: Introduction to Differential SAR Interferometry
n Part 2: ESA Platforms for Automatic Web Processing
n Part 3: New frontiers in Earth Observation research
n Open Discussion
n Part 1: Introduction to Differential SAR Interferometry
n Part 2: ESA Platforms for Automatic Web Processing
n Part 3: New frontiers in Earth Observation research
n Open Discussion
3. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
• Active sensors
• Microwave sensors
• Coherent sensors
Key points of Radar (SAR) Imaging from space
4. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
• Passive sensors use solar radiation (light) or the one emitted by the
observed object as source of illumination. Typically they operate in the
optical or infrared.
• Active sensors have their own source of illumination. Typically they
operate in the microwave.
Capability to "observe" during day and night
Key points of Radar (SAR) Imaging from space
5. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Optical image Radar (SAR) image
same moment of acquisition
Capability to "observe" even in presence of clouds
Key points of Radar (SAR) Imaging from space
6. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Coherent sensors
SAR image
Synthetic Aperture
Key points of Radar (SAR) Imaging from space
7. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Baseline
S(t1)
S(t2)
Interferogram
Differential SAR Interferometry (DInSAR)
8. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Campi Flegrei (3-7/2000)
π -π
2
2 _
λ
πϕ ≈→= losdd l
Centimetric displacements (in
some cases even sub-centimetric)
can be measured in “coherent”
areas!
losdd l _
4
λ
π
ϕ ≈
Differential SAR Interferometry (DInSAR)
LOS: Line Of Sight
ERS wavelenght: 5.6 cm
9. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
0 2.8
cm
40°30’ N
Astroni
Solfatara
Arco Felice
Agnano
Bagnoli
3/2000-7/2000
Differential SAR Interferometry (DInSAR)
φm
Phase Unwrapping
Operation
φ = φm + 2π k
Wrapped Phase (-π, π)
10. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
δϕ = −
4π
λ
r2
− r1( )= −
4π
λ
r − r1( )−
4π
λ
r2
− r( )≈ −
4π
λ
bsin !ϑ − β( )−
4π
λ
ld _los
=ϕt
+ϕd
Observations are done at different epochs and from different orbital positions
(real case)
Topographic Phase Deformation Phase
LOS: Line Of Sight
Differential SAR Interferometry (DInSAR)
11. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
DInSAR
Interferogram
Generation
(Flattening)
Coherence
Map
Topographic
Interferogram
(DEM & orbital
information are
needed)
15042009_20052009_bperp=20m
Differential SAR Interferometry (DInSAR)
12. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Small baseline DInSAR interferograms are less affected by noise effects
(decorrelation).
ERS1_01/05/1996 - ERS2_15/08/1996
baseline=950 m
ERS2_31/08/1995 - ERS2_ 15/08/1996
baseline=10 m
Napoli Bay (ERS Multi-look image)
Why Small Baseline?
13. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Spatial Decorrelation Effects depend on the perpendicular baseline of the
used SAR data pairs. To reduce such effects the use of small spatial baseline
couples is required.
Temporal Decorrelation Effects are due to temporal reflectivity changes of
the imaged scene. To reduce such effects the use of small temporal baseline
couples is required.
Different Atmospheric
Conditions from one image to
another lead to artifacts in the
generated interferograms. Such
effects can be reduced by
properly averaging independent
interferograms (stacking).
DInSAR Limitations
14. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
VOLCANICSEISMIC
URBAN LANDSLIDES
DInSAR can play a key role in many contexts
15. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
L’AQUILA
Monday 06/04/2009,
01:00 UTC, Mw=6.3
L’Aquila 2009 earthquake: scenario
16. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
06.04
Timeline
10.04 12.04 15.04 16.0408.04
ENVISAT (ASCENDING)
COSMO-SkyMED
(DESCENDING)
TerraSAR-X
(ASCENDING)
ENVISAT (DESCENDING)
L’ Aquila
Earthquake
Mw=6.3
L’Aquila 2009 earthquake: DInSAR analysis
17. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
ENVISAT (Desc.) ENVISAT (Asc.)COSMO (Asc.)
ALOS (Asc.)
Co-seismic deformation analysis
18. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Length=12.2 ±0.4km, Width=14.1 ±0.7km
Top depth= 1.9 ±0.2km , slip 0.56 ±0.02m
rake -103°±2
47° ±1 dip
North
133°±2 strike
Source parameters: Atzori et al. 2009, GRL
Results of co-seismic interferograms modelling
Lanari et al. 2010, GRL
19. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
ERS-1
JERS
ERS-2
RADARSAT
ENVISAT
ALOS1
COSMO-SkyMed
TerraSAR-X
Sentinel-1A
Sentinel-1B
ALOS2
swath width: ≈ 100 km
revisit 2me: ≈ monthly
swath width: ≈ 40 km
revisit 2me: 4 - 11 days
swath width: ≈ 250 km
revisit 2me: 12 - 6 days
Temporal evolution of radar satellites for EO
20. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
ERS data availability: Napoli bay (Italy) example
Data Set Distribution
PerpendicularBaseline[m]
Time [year]
single deformation events generation of displacement time-series
21. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
PS SB
Persistent Scatterers (PS) vs. Small Baseline (SB)
22. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
To produce deformation times-series from a SAR data sequence:
Ÿ using interferograms characterized by a “small baseline” in order to mitigate
noise (decorrelation) phenomena;
Ÿ properly “linking” possible interferometric SAR data subset separated by large
baselines. In particular, for each coherent pixel, the deformation time-series is
computed by searching for an LS solution with a minimum norm constraint (the
SVD method is applied).
Time
Perpendicular
Baseline
Subset 2
Subset 1
Original SBAS algorithm: key idea
23. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
⎥
⎦
⎤
⎢
⎣
⎡
−
=
−
==
21
21
1
01
01
1v
M-M-
M-M-
M-
T
tt
)(t)-(t
v,, ....
tt
)(t)-(t
v
φφφφ
δφ=vB
by solving the linear system
wherein δφ is the interferometric phase vector, B the coefficient matrix
and M the number of SAR images.
To solve the linear system, we apply the SVD-method1.
Following the unwrapping operation we evaluate (for each pixel) the
displacement velocity vector v
Berardino, P., Fornaro, G., Lanari, R., Sansosti, E., 2002. A new algorithm for surface deformation monitoring based on small
baseline differential SAR interferograms. IEEE Transactions on Geoscience and Remote Sensing 40 (11), 2375–2383.
Mathematical framework of the SBAS algorithm
24. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Interferograms
Mean deformation velocity [cm/yr]
> 0.75<- 0.75
Mt. Vesuvio
Campi Flegrei
Ischia
The SBAS algorithm rationale
25. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
LOS Mean Velocity (mm/yr)
> 15<-15
Ascending Orbit
ERS-1/2, T129, F747
Descending Orbit
ERS-1/2, T222, F2853
Lundgren et al. 2004, GRL
Asc.Desc.
Mt. Etna (Italy): 1992-2000 deformation analysis
26. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Vertical East-West
[mm/yr]
> 15<-15
Down Up [mm/yr]
> 20<-20
West East
Mt. Etna (Italy): 1992-2000 deformation analysis
Lundgren et al. 2004, GRL
27. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Meanvelocity[mm/yr]
> 20
<-20
Fernandina
Isabela
Sierra Negra
Wolf
ENVISAT Ascending (2003-2007)
Galápagos
28. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
> 30<-30
Mean Velocity (mm/yr)
ENVISAT descending (2002-2009)
σ ~ 1 cm
Δ: SAR
* : GPS
Kilauea
Mauna Loa
Mauna Loa and Kilauea Volcanoes (Hawaii)
29. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Mean velocity (mm/yr)
< -10 0 > 10
Santa Ana
Basin
Pomona
Newport –
Inglewood
fault
GPS Network
(SCIGN)
N
Los Angeles Bay
Lanari et al. 2004, GRL
ERS-1/2 Descending data
(1995-2002)
30. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Mean velocity (mm/yr)
< -10 0 > 10
c
d
e
Santa Ana Basin: DInSAR vs. GPS
31. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
San Francisco Bay
Meanvelocity[mm/yr]
> 6
<-6
ERS-1/2 Descending (1992-2000)
32. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
EGU General Assembly 2011
April 03-08, 2011 Vienna, Austria
Mean velocity (mm/yr)
< -6 0 > 6
Hayward fault
33. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Creep event (Feb 1996)
* = Alignment arrays
Average boxes both sides
Difference to relative
motion
Lanari et al. 2007, RSE
Hayward fault: deformation time series
34. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Ÿ exploiting interferograms characterized by a “small
baseline” in order to limit the noise (decorrelation)
phenomena, thus maximizing the number of
investigated pixels;
Ÿ using no a priori model on the investigated
deformation signal.
The SBAS approach allows to produce “long term”
deformation times-series by:
SBAS approach: key points
35. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
SBAS vs. Alignment array:
San Francisco Bay
SBAS vs. GPS: Los Angeles
LOS
<-10
> 10
mm/year
SBAS vs. Leveling: Napoli Bay
σdispl ~ 5-10 mm
σvel ~ 1-2 mm/year
SBAS-DInSAR result accuracy
Casu et al. 2006, RSE
36. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Agenda
n Part 1: Introduction to Differential SAR Interferometry
n Part 2: ESA Platforms for Automatic Web Processing
n Part 3: New frontiers in Earth Observation research
n Open Discussion
n Part 1: Introduction to Differential SAR Interferometry
n Part 2: ESA Platforms for Automatic Web Processing
n Part 3: New frontiers in Earth Observation research
n Open Discussion
37. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
The Thematic Exploitation Platforms goals are:
à Facilitate use & processing of large datasets (including non-space data) by a
large number of users (science and non-science).
à Processing services, software (e.g. toolboxes, etc.) and computing
resources.
à Provide an environment for services development, integration and
exploitation.
à Federate user communities around common scientific & thematic objectives.
à Promote shared science objectives & better use of satellite EO.
à Collaboration tools (e.g. knowledge base, open publications, social
networking).
The Thematic Exploitation Platform (TEP)
38. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Data Archives
- ERS
- ENVISAT
- Sentinels
- CSK
- TSX
- ALOS
SBAS, NSBAS, ROI_PAC, StaMPS, DIAPASON, DORIS, …
- G-POD
- Cloud
- Federated
resources
ESA Geohazard Exploitation Platform (GEP)
39. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
ESA Geohazard Exploitation Platform (GEP)
G-POD (legacy platform, from SSEP):
- Tool integration need interaction with “integration” team
- GRID based, even if could be instantiated on Cloud
GEP (under development/validation, pre-operations start in
2017):
- Tool integration easier (Cloud sandboxes)
- Cloud based and cost-effective resource provisioning
- Contains several Pilot Projects providing Geohazards and InSAR-
related services (e.g. SBAS, DIAPASON, StaMPS, …)
- Allows results, sharing, publication (e.g. via Zenodo, DOI),
reproducibility, discovery and other collaboration services
- Interested users (Early Adopters) should submit an User
Request Form (URF) to ESA for evaluation
40. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
http://gpod.eo.esa.int - eo-gpod@esa.int
G-POD Home Page
41. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
http://gpod.eo.esa.int - eo-gpod@esa.int
G-POD Home Page
50. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Reference Point Selection
Reference point must be on land and possibly in
an expected stable area
52. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Archive Querying and Data Selection
53. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Archive Querying and Data Selection
G-POD Archive
VA4 Archive
VA4 - Only for Processing
54. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Archive Querying and Data Selection
55. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Keep the same illumination geometry:
select the same Track!!!
59. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Archive Querying and Data Selection
Some acquisitions could be acquired the same
day but at slightly different times
60. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Processed Area
Area of Interest (AOI)
The algorithm merges adjacent frames of the same track to totally cover
the selected AoI.
Data Selection and Optimization
61. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Processed Area
Area of Interest (AOI)
Large Raw Data are cut around the Area of Interest
Data Selection and Optimization
62. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Processed Area
Area of Interest (AOI)
Automatic data rejection of acquisitions that do not cover the AoI
Data Rejected
Data Selection and Optimization
65. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
To be modified by expert users.
We warmly suggest to keep them unchanged!
Final settings and Job submission
66. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Processing Parameter Selection
• Noise Filtering
• Temporal and Spatial
Baseline
• Doppler centroid
parameters
• Time Series Generation
Spatial Network
• Atmospheric Filtering
• Multi-temporal/
interferogram analysis
67. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Baseline Constrain Selection
• Temporal and Spatial
Baselines have as default
values 1500 days and
400 m, respectively.
• These two parameters
have a strong impact
when multi-temporal
analysis is performed.
• Too small values have the
consequence to reduce
the SAR dataset, too big
values limit the success
of Phase Unwrapping
procedure.
69. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Interf. Pair Selection: Delaunay Triangulation
70. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Interf. Pair Selection: Large Baseline Removing
Maximum Spatial Baseline: 300 m
71. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Interf. Pair Selection: Large Baseline Removing
Maximum Spatial Baseline: 50 m
72. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
G.P.D. and Doppler Selection
• Ground Pixel dimension is
the value used for the
spatial multilook of the
interferograms generated
with ERS and ENV sensors
• Max Allowed Delta-Doppler
is the minimum azimuth
b a n d w i d t h o v e r l a p
between master and slave
images .
• Max Allowed Doppler
Centroid has a strong
impact for ERS-2 images
acquired after February
2000.
73. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Common Band Filtering Selection
Common Band Filtering
f e a t u r e s i s a i m e d a t
increase the interferograms
coherence.
This filter removes the
u n c o r r e l a t e d s p e c t r a l
contributions to perform
the interferogram using the
c o m m o n b a n d w i d t h
between master and slave
images
74. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Coherence threshold
• This parameter permits to
identify the (spatial)
network of points that are
subsequently unwrapped.
• It is strongly related to
the algorithm used for
Phase Unwrapping:
Extended Minimum Cost
Flow (EMCF).
• Default value of 0.7 is
suitable for a large part
of case studies.
• This in not a threshold to
select the final spatial
distribution of points
75. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
SBAS Phase unwrapping: the Extended-MCF
Algorithm
The EMCF algorithm (Pepe at al, 2006, TGRS) allows to unwrap a sequence
of M multi-temporal differential interferograms that generate a triangulation
in the Temporal/Perpendicular Baseline Plane.
Time [year]
PerpendicularBaseline[m]
1992 1999 2007
76. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
A Delaunay triangulation involving the coherent points in the Azimuth/Range
domain is subsequently built.
Aziumth
Range
The i-th edge of the triangulation corresponds to the i-th differential
interferogram:
( ) ( ), 1,..., ,i z ga r i M a r A RΦ = ∈ ×
Extended-MCF Algorithm: Rationale
“Temporal” network “Spatial” network
77. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
A Delaunay triangulation involving the coherent points in the Azimuth/Range
domain is subsequently built. The spatial network can be defined
through one of the inputs of the GUI.
Aziumth
Range
The i-th edge of the triangulation corresponds to the i-th differential
interferogram:
( ) ( ), 1,..., ,i z ga r i M a r A RΦ = ∈ ×
Extended-MCF Algorithm: Rationale
78. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Coherence Threshold Selection
Coh: 0.7Coh: 0.8
79. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
• APS Smoothing Time
Window represents the
window of the temporal
filter in the APS removing
procedure.
• Larger values generate
more smoothed time
series.
• Default value is 200
days; useful values lie in
the range 100 to 400
days.
Displacement Time Series Generation and
APS Filtering
80. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Hawaii T200 case study
>6<-6
cm/yr
Mauna Loa
Kilauea
Est Rift
zone
81. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Final settings and Job submission
To be modified by expert users.
We warmly suggest to keep them unchanged!
90. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
On a pixel by pixel basis, the SBAS algorithm allows the generation of the
surface deformation time-series
[ ]10, ,..., NΦ ≡ Φ Φ
Subsequently, once the deformation time-series are retrieved, the quality
of the inversion is checked by comparing the original interferograms to the
ones reconstructed from the obtained time-series, by means of the so-
called Temporal Coherence factor:
( )1
exp i i
M
i IM IS
i
M
=
⎡ ⎤ΔΦ − Φ − Φ
⎣ ⎦
Γ =
∑
Small Baseline Subset SBAS algorithm
91. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Temporal Coherence is a quality index of our phase reconstruction; two
different aspects may lead to a decrease of the temporal coherence
values:
Ø Time-Inconsistent Phase Unwrapping Errors;
Ø Time-inconsistencies among the generated multilook (wrapped)
interferograms, due to the fact that multi-look operations are
independently carried out on each single SAR data pair.
INDEED …
Temporal Coherence Factor
92. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
April 27,
2008
July 26,
2009
April 12, 2009
ΔΦ 1
ΔΦ 1
Azimuth
Range
ΔΦ 2 ΔΦ 3
ΔΦ 2
ΔΦ 3
Azimuth
Range
Azimuth
Range
Multi-look (Wrapped) Phase Time-Inconsistency
93. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
April 27,
2008
July 26,
2009
April 12, 2009
ΔΦ 1
ΔΦ 2
ΔΦ 3
Accordingly, multilook (wrapped) interferograms are not time-consistent. It means
that phase acquisitions are not known while a redundant set of M multi-look
interferograms is generated.
We extend the temporal coherence idea to wrapped interferograms, to have time-
consistent noise filtered SB multilook interferograms before any phase
unwrapping operation.
Azimuth
Range
1 2 3 0.r ≡ ΔΦ + ΔΦ + ΔΦ ≠
π
-π
Multi-look (Wrapped) Phase Time-Inconsistency
94. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
We search for the solution of the following maximization problem:
( )
1
1
exp
arg max
k k
M
k k IM IS
k
M
k
k
jξ
ξ
=
Φ
=
⎧ ⎫
⎡ ⎤⎪ ⎪ΔΦ − Φ + Φ
⎣ ⎦⎪ ⎪⎪ ⎪
Φ = ⎨ ⎬
⎪ ⎪
⎪ ⎪
⎪ ⎪⎩ ⎭
∑
∑
)
Basically, we maximize a weighted version of the temporal coherence
factor; the weights, representing our confidence on the phase stability of
the generated M multilook DInSAR phases, are set by using the pixel
spatial coherence
( )( )
( )
2 2
2 2
1
exp , 1,...,
1 1
A R
A R
N N
k k
A R h N p N
j x h y p k M
N N
ξ
=− =−
⎡ ⎤= ΔΦ + + ∀ =⎣ ⎦+ + ∑ ∑
Phase Inversion Algorithm
95. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
We search for the solution of the following maximization problem:
( )
1
1
exp
arg max
k k
M
k k IM IS
k
M
k
k
jξ
ξ
=
Φ
=
⎧ ⎫
⎡ ⎤⎪ ⎪ΔΦ − Φ + Φ
⎣ ⎦⎪ ⎪⎪ ⎪
Φ = ⎨ ⎬
⎪ ⎪
⎪ ⎪
⎪ ⎪⎩ ⎭
∑
∑
)
Basically, we maximize a weighted version of the temporal coherence
factor; the weights, representing our confidence on the phase stability of
the generated M multilook DInSAR phases, are set by using the pixel
spatial coherence
( )( )
( )
2 2
2 2
1
exp , 1,...,
1 1
A R
A R
N N
k k
A R h N p N
j x h y p k M
N N
ξ
=− =−
⎡ ⎤= ΔΦ + + ∀ =⎣ ⎦+ + ∑ ∑
Phase Inversion Algorithm
96. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Improved SBAS Inversion
Ø Once the phase acquisitions are retrieved, a noise-filtered version of
the original interferograms is generated.
Note that Space Adaptive Multilook operations could be also
preliminarly applied to the original interferograms at the
expenses of the overall computation time
97. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Azimuth
Range
February 1, 2009 – August 15, 2010
Perpendicular Baseline ~300 m
The Abruzzi area case-study
98. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Azimuth
Range
May 8, 2005 – April 12, 2009
Perpendicular Baseline ~800 m
The Abruzzi area case-study
111. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
LOS Unit Vectors
Data Field
112. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
Deformation Time series [cm]
Data Field
113. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Ask for an Account to:
eo-gpod@esa.int
sbas-help@irea.cnr.it
114. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
eo-gpod@esa.int
EO-SSO ID: eduusr03
Password: Educational_03
EO-SSO ID: eduusr05
Password: Educational_05
EO-SSO ID: eduusr06
Password: Educational_06
G-POD Educational Account Credentials
115. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Agenda
n Part 1: Introduction to Differential SAR Interferometry
n Part 2: ESA Platforms for Automatic Web Processing
n Part 3: New frontiers in Earth Observation research
n Open Discussion
n Part 1: Introduction to Differential SAR Interferometry
n Part 2: ESA Platforms for Automatic Web Processing
n Part 3: New frontiers in Earth Observation research
n Open Discussion
116. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
SAR Data Scenario: Satellites
1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 2018 2020
ERS-1
JERS
ERS-2
RADARSAT
ENVISAT
ALOS
COSMO-SkyMed
TerraSAR-X
Sen2nel-1A
Sen2nel-1B
Swath Width ≈ 100 km
Revisit 2me ≈ monthly
Swath Width≈ 40 km
Revisit Time: 4 - 11 days
Swath Width ≈ 250 km
Revisit Time: 12 - 6 days
TerraSAR-X
COSMO-SkyMed
ALOS-2
117. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
SENTINEL1-A
ENVISAT
CSK
40 km
100 km
250 km
SAR Data Scenario: Coverage Comparison
118. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Sen3nel-1: ~ 6 TByte data per day
ERS & ENVISAT data over world tectonic regions: 70+ Tera on line; about 10 days of S1-A acquisi2ons!
Sentinel-1 monthly coverage
Towards a Big Data scenario
119. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Sentinel-1: Small Baseline System
1995 2000 2005 2010 2015
Time [year]
-1.0
-0.5
0.0
0.5
1.0
PerpendicularBaseline[km]
ERS and ENVISAT
Perpendicular baseline distribution
120. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Sentinel-1: Small Baseline System
Perpendicular baseline distribution
ERS and ENVISAT
1995 2000 2005 2010 2015
Time [year]
-1.0
-0.5
0.0
0.5
1.0
PerpendicularBaseline[km]
Sentinel-1
121. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
ESA’s GEP web site
https://geohazards-tep.eo.esa.int
122. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
TEP Scenario 1 – EO Data Exploitation,which allows a user to discover/select
data and pre-existing processing service; process data; and visualize/analyse or
select and apply data manipulation tools to the result.
TEP Scenario 2 – New EO Service Development, which allows a user to
discover/select a data sample and software components; engineer (or upload)
and validate an application (such as a processor); and deploy the application on
the platform for use also by other users.
TEP Scenario 3 – New EO Product Development, which allows a user to
Authenticate; alternatively upload and deploy a new processor; discover/select
data; process the data; and eventually publish the resulting product. Note that
this scenario implies the implementation of effective, scalable, and cost-
optimized strategies for infrastructure provisioning for processing of very big
datasets.
GEP Pilot Project Scenarios
geohazards-tep@esa.int
If interested in submitting a Pilot Project according to the 3 depicted scenarios, contact:
123. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
ESA’s GEP web site
https://geohazards-tep.eo.esa.int
151. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
TEP Scenario 1 – EO Data Exploitation,which allows a user to discover/select
data and pre-existing processing service; process data; and visualize/analyse or
select and apply data manipulation tools to the result.
TEP Scenario 2 – New EO Service Development, which allows a user to
discover/select a data sample and software components; engineer (or upload)
and validate an application (such as a processor); and deploy the application on
the platform for use also by other users.
TEP Scenario 3 – New EO Product Development, which allows a user to
Authenticate; alternatively upload and deploy a new processor; discover/select
data; process the data; and eventually publish the resulting product. Note that
this scenario implies the implementation of effective, scalable, and cost-
optimized strategies for infrastructure provisioning for processing of very big
datasets.
GEP Pilot Project Scenarios
geohazards-tep@esa.int
If interested in submitting a Pilot Project according to the 3 depicted scenarios, contact:
160. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
cm/year
>6<-6
Campi Flegrei Caldera
Systematic
S1 P-SBAS
processing
Sen2nel-1
12/6 days
repeat cycle
Community Geoportal
2015 2016
Time [year]
-2
0
2
4
6
8
10
Displacement[cm]
2015 2016
Time [year]
-2
0
2
4
6
8
10
Displacement[cm]
2015 2016
Time [year]
-2
0
2
4
6
8
10
Displacement[cm]
2015 2016
Time [year]
-2
0
2
4
6
8
10
Displacement[cm]
GEP: Sentinel-1 P-SBAS systematic processing
161. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Sentinel-1 P-SBAS results at large scale
162. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
cm/year
>6<-6
Time Interval: October 2014 – March 2016
Sentinel-1 P-SBAS results at large scale
163. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
cm/year
>6<-6
Time Interval: October 2014 – March 2016
Sentinel-1 P-SBAS results at large scale
164. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
cm/year
>6<-6
Time Interval: October 2014 – March 2016
Sentinel-1 P-SBAS results at large scale
165. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Campi Flegrei caldera
2015 2016
Time [year]
-2
0
2
4
6
8
10
Displacement[cm]
Sentinel-1 P-SBAS results at large scale
166. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Gargano Region
Sentinel-1 P-SBAS results at large scale
167. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Gargano Region
2015 2016
Time [year]
-6
-4
-2
0
2
Displacement[cm]
2015 2016
Time [year]
-6
-4
-2
0
2
Displacement[cm]
Sentinel-1 P-SBAS results at large scale
168. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
cm/year
>6<-6
2015 2016
Time [year]
-2
0
2
4
6
Displacement[cm]
2015 2016
Time [year]
-10
-8
-6
-4
-2
0
2
4
Displacement[cm]
Etna Volcano
Sentinel-1 P-SBAS results at large scale
169. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Ask for an Account to:
geohazards-tep@esa.int
sbas-help@irea.cnr.it
170. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
Credits
CNR-IREA
M. Bonano, F. Casu, C. De Luca, R. Lanari, M. Manunta, I. Zinno,
M. Manzo, A. Pepe
ESA RSS
R. Cuccu, J.M. Delgado Blasco, G. Rivolta
ESA
H. Laur, S. Loekken, P. Bally, S. Pinto, A. Marin, A. Cuomo
Terradue
F. Brito, F. Pacini, E. Mathot,, H. Caumont
171. SBAS-DInSAR processing on the ESA Geohazard Exploitation Platform
12-days Interferometric
Coherence map of almost
the en3re Europe
300 slices
June-July 2015
Coverage: 3,200,000 km2.
In coopera2on with ESA G-POD
Thank you!
geohazards-tep@esa.int