You are on page 1of 26

Batch

processing

•  Automa'c processing of a large


volume (~TB) of SAR data to
generate hundreds to thousands
of interferograms

•  Analyze interferograms to
es'mate ground deforma'on and
its 'me-varia'ons to study
–  Earthquake cycle deforma'on
–  volcano unrest
–  Glacier flow
–  Urban and infrastructure
–  hydrology
–  landslide


Batch processing for time series
Overview

•  Batch processing:
1.  Preprocessing without a master image
pre_proc_init.csh
2.  Preprocessing with a master image
pre_proc_batch.csh
3.  Align a stack of SAR data align_batch.csh
4.  Form a stack of interferograms in2_batch.csh
•  Example data set
–  GMTSAR website Envisat stack example
Batch processing step 1: pre_proc_init.csh

•  Func'on:
–  preprocess a stack of SAR data using default
parameters (earth radius, Doppler centroid, near
range)
–  Generate baseline-4me plot to choose master
images, alignment strategy, interferometric pairs
Baseline-'me plot: stacktable_all.ps

400

200
baseline (m)

07968
0
19491

12477

−200

Super master

−400

−600

2004 2005 2006


year
Batch processing step 2:
pre_proc_batch.csh
•  Func'on:
–  preprocess a stack of SAR data using uniform
parameters (earth radius, Doppler centroid, near
range) to make them geometrically consistent
with one single image (super master)
Batch processing step 2:
pre_proc_batch.csh
1. Modify data.in file

Super master

2. Delete old PRM and raw files



3. Edit batch.config file
Batch processing step 3: align_batch.csh
•  Func'on:
–  Focus SAR images to form Single Look Complex
(SLC) data
–  Align (image co-registra'on) a stack of SLC data
using 2D cross-correla'on within sub-pixel (<10m)
accuracy

Co-registra'on SAR images (Alignment)

400

200
baseline (m)

07968
0
19491

Slave 12477

−200
Slave
master

−400

−600

2004 2005 2006


year
Batch processing step 3: align_batch.csh
1. Edit align.in file

Master Slave Super master

2. Then run align_batch.csh

Time-consuming part of the processing .. take a break here ..


Batch processing step 4: intf_batch.csh

•  Func'on:
–  Convert Digital Eleva'on Model into radar
coordinates
–  Form interferograms using two SLC data
–  Remove phase due to earth curvature and
topography
–  Plot amplitude, correla'on, phase using GMT
–  Unwrap using SNAPHU
–  Geocode and make Google Earth KML files
Choose interferogram pairs

400

200

A
baseline (m)

07968
0
19491

B
C
12477

−200

−400

−600

2004 2005 2006


year
Batch processing step 4: intf_batch.csh
1. Edit in^.in file to choose inteferogram pairs

reference repeat
Batch processing step 4: intf_batch.csh
2. Make Digital Eleva'on Model (DEM): dem.grd file and put
it inside topo/ directory
Batch processing step 4: intf_batch.csh
3. Check/modify batch.config file

Time-consuming part of the processing .. take a break here ..


Batch processing: results
•  Interferograms are stored in individual folders in in^/
•  The folder can be named aaer either date (default) or orbital
number
•  Each interferogram folder contains the following files:
1.  Amplitude, phase, correla4on, unwrapped phase, filtered
phase image files in NetCDF format “.grd”
2.  Corresponding files aaer geocoding with subfix “_ll.grd”
3.  Postscripts plots: “.ps”
4.  Google Earth “.kml” and “.png”
More interferograms …
InSAR 'me-series methods
•  Stacking (average mul'ple interferograms)
•  Small-Baseline Subset (SBAS)
–  Download a test case from GMTSAR website
•  Persistent Scaderer InSAR (PSInSAR)
–  StaMPS (Stanford Methods for PS InSAR)
–  Wriden in MATLAB
•  Giant (Generic InSAR analysis toolbox)
–  Mul'ple 'me-series analysis tools
–  Wriden in Python
Math of stacking (Appendix C)

φdef
′ =
∑ φ (x, Δt)
i
res Unwrapped phase
∑ Δt i

Time span

Recovered deforma'on
re + h r +h
∑ φ (x, Δt) 4π
i ∑ Bi perp × ρbsin θ
− ∑ Bi perp Bi para × 2e
ρ bsin θ
φdef
′ = φdef
′ (x) + −
atm
dh
∑ Δt λ i
∑ Δt i

True deforma'on atmosphere topography


Stacking example with the complete data set
Track 213 Frame 660

Land subsidence near Coachella Valley, California


baseline time plot
3000

12285

2000 11614
10943
25705

09601 23692
23021
22350
1000
perpendicular baseline (m)

08930
08259 21008
06246
04233 07588
19666
0 05575
12956
03562 18995
18324

02891 16311

−1000
04904 14969
15640

14298

−2000

02220 13627

−3000
0 200 400 600 800 1000 1200 1400 1600 1800
number of days
InSAR ;me-series methods
Time-series show Point B
seasonal varia;on and
linear trend

Point A
Small Baseline Subset (SBAS)
Small Baseline Subset (SBAS) methods

⎡ 1 1 0 ... β B1 ⎤ ⎡ m1 ⎤ ⎡ d1 ⎤
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ 0 1 1 ... β B2 ⎥ ⎢ m2 ⎥ ⎢ d2 ⎥
W ⎢ ... ... ... ... ... ⎥ ⎢ ... ⎥=W ⎢ ... ⎥
⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ λ − λ 0 ... 0 ⎥ ⎢ ms ⎥ ⎢ 0 ⎥
⎢⎣ 0 λ − λ ... 0 ⎥⎦ ⎢⎣ Δh ⎥
⎦ ⎢⎣ 0 ⎥⎦
W = diag{γ 1 , γ 2 , γ 3 ,!, γ n }
di -- Unwrapped phase
Bi -- Perpendicular baseline
γ i -- Coherence
mi -- Incremental range change
Δh -- DEM error

[Berardino et al., 2002; Schmidt and Burgmann, 2003]


Demo

You might also like