Batch

 processing  and  stacking  in  

GMTSAR  
Xiaopeng  Tong  and  David  Sandwell  

•  Introduc9on  
•  Overview  
•  Batch  processing  with  an  example  
•  Stacking  with  an  example  
•  Summary  

InSAR  workshop  2013,  Boulder,  CO  

stacking for time series

 combined high-resolution!
velocity!

http://topex.ucsd.edu/pub/saf_models/insar/ALOS_ASC_masked.kmz

standard deviation!

[Tong et al., 2013]"

3

Parkfield SAF "
red 10 mm/yr"
blue -10mm/yr  
Parkfield SAF "
red 10 mm/yr"
blue -10mm/yr  
Creeping SAF "
red 10 mm/yr"
blue -10mm/yr  
Creeping SAF "
red 10 mm/yr"
blue -10mm/yr  
 Introduc9on  
•  Automa9c  processing  of  a  stack  of  SAR  data  to  
generate  interferograms    
•  Average  mul9ple  interferograms  to  es9mate  
velocity  due  to  inter-­‐seismic,  post-­‐seismic,  
volcano  sources  
•  Can  be  integrated  into  9me-­‐series  analysis  
tools,  such  as  SBAS,  PSInSAR  
 Introduc9on  -­‐  con9nued  
•  Automa9c  data  processing  has  advantages:    
–  Objec9ve  
–  Reproducible  
–  Less  prone  to  errors  
•  WriQen  in  shell  and  is  easy  to  modify  
•  Advanced  user  can  explore  customized  
features  
 Overview  
•  Batch  processing:  
–  Preprocessing  without  a  master  image  
pre_proc_init.csh  
–  Preprocessing  with  a  master  image  
pre_proc_batch.csh  
–  Align  a  stack  of  SAR  data  align_batch.csh  
–  Form  a  stack  of  interferograms  in2_batch.csh  
•  Stacking:  
–  Average  the  phase  stack_phase.bash  
–  Average  the  correla9on  stack_corr.bash  
–  Compute  standard  devia9ons  std.bash  
–  GPS/InSAR  integra9on  stack.csh  
 Batch processing: pre_proc_init.csh

•  Func9on:    
–  preprocess  a  stack  of  SAR  data  using  default  
parameters  (earth  radius,  Doppler  centroid,  near  
range)  
–  Generate  baseline-­‐9me  plot  to  choose  master  
images,  alignment  strategy,  interferometric  pairs    
 Batch processing: pre_proc_init.csh
For  example:    
 Batch  processing:  pre_proc_init.csh  
For  example:    
 Baseline-­‐9me  plot:  stacktable_all.ps  
12285
5000
11614 25705
10943

23692
09601 23021
4000
22350

08930
08259
21008
baseline (m)

06246 07588
04233
3000 19666

05575
12956
03562 18995
18324

02891 16311
2000

04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

year Track  213  Frame  660  

 Baseline-­‐9me  plot:  stacktable_all.ps  
12285
5000
11614 25705
10943

23692
09601 23021
4000
22350

08930
08259
21008
baseline (m)

06246 07588
04233
3000 19666

05575
12956
03562 18995
18324

Super  master  
02891 16311
2000

04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

year Track  213  Frame  660  

 Batch processing: pre_proc_batch.csh

•  Func9on:    
–  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: pre_proc_batch.csh
1.  Modify  data.in  file  

Super  master  

2.  Delete  old  PRM  and  raw  files  

 Batch processing: pre_proc_batch.csh
3.  Modify  batch.config  file  and  run  pre_proc_batch.csh  
 
Stop  here  to  look  at  the  batch.config  file  
 Batch processing: align_batch.csh
•  Func9on:  
–  Focus  SAR  images  to  form  Single  Look  Complex  
(SLC)  data    
–  Align  (image  registra9on)  a  stack  of  SLC  data  using  
2D  cross-­‐correla9on  within  sub-­‐pixel  (<10m)  
accuracy  
 
Baseline  9me  plot   Track  213  Frame  660  
12285
5000
11614 25705
10943

23692
09601 23021
4000
22350

08930
08259
21008
baseline (m)

06246 07588
04233
3000 19666

05575
12956
03562 18995
18324

2000 02891
Super  master   16311

04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

year
 “Leap  frog”  method  to  align  SAR  images  
primary match
12285
5000 secondary match
11614 25705
tertiary match 10943

09601
Slave   23692
23021
4000
22350

08930
08259
21008
baseline (m)

06246 07588
04233
3000 19666

05575
12956
03562 18995
18324

02891 16311
2000
master  
04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

year
 “Leap  frog”  method  to  align  SAR  images  
primary match
12285
5000 secondary match
11614 25705
tertiary match 10943

23692

4000 Surrogate   09601 23021

master   22350

08930
08259
21008
baseline (m)

3000 04233
06246 07588
19666
Slave  
05575
12956
03562 18995
18324

02891 16311
2000

04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

year
 “Leap  frog”  method  to  align  SAR  images  
primary match
12285
5000 secondary match
11614 25705
tertiary match 10943

23692
09601 23021
4000
22350

08930
08259
21008
baseline (m)

06246 07588
04233
3000 19666

05575
12956
03562 18995
18324

02891 16311
2000

04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

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

Master  or  Surrogate   Slave   Super  master  

master  

2.  Then  run  align_batch.csh  

Time-­‐consuming  part  of  the  processing  ..  take  a  break  here  ..  
 Batch processing: intf_batch.csh

•  Func9on:    
–  Convert  Digital  Eleva9on  Model  into  radar  
coordinates  
–  Form  interferograms  using  two  SLC  data  
–  Remove  phase  due  to  earth  curvature  and  
topography  
–  Plot  amplitude,  correla9on,  phase  using  GMT  
–  Unwrap  using  SNAPHU  
–  Geocode  and  make  Google  Earth  KML  files    
 Choose  interferograms  pairs  
primary match
12285
5000 secondary match
11614 25705
tertiary match 10943

23692
09601 23021
4000
22350

08930
08259
21008
baseline (m)

06246 07588
04233
3000 19666

05575
12956
03562 18995
18324

02891 16311
2000

04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

year
 Batch processing: intf_batch.csh
1.  Edit  inc.in  file  to  choose  inteferogram  pairs  
 Batch processing: intf_batch.csh
2.  Make  dem.grd  file  and  put  it  inside  topo/  directory  
 Batch processing: intf_batch.csh
2.  Make  dem.grd  file  and  put  it  inside  topo/  directory  
 Batch processing: intf_batch.csh
3.  Check/modify  batch.config  file  

Time-­‐consuming  part  of  the  processing  ..  take  a  break  here  ..  
 Batch processing: results
•     All  interferograms  are  in  different  folders  in  inc/  
•     The  folder  can  be  named  aeer  either  date  or  orbital  number  
ü     Modify  batch.config  to  choose  among  date  or  orbital  number  
•     Each  interferogram  folder  contains  the  following  files:    
1.  Amplitude,  phase,  correla>on,  unwrapped  phase,      filtered  
phase  image  files  in  GMT/NetCDF  format  “.grd”    
2.  Corresponding  files  aeer  geocoding  with  subfix  “_ll.grd”  
3.  Postscripts  plots:  “.ps”    
4.  Google  Earth  “.kml”  and  “.png”        
 Batch processing: results
Phase  of  the  3  interferograms    

primary match
12285
5000 secondary match
11614 25705
tertiary match 10943

23692
09601 23021
4000
22350

08930
08259
21008
baseline (m)

06246 07588
04233
3000 19666

05575
12956
03562 18995
18324

02891 16311
2000

04904 14969
15640

14298
1000

02220 13627
0

2006 2007 2008 2009 2010 2011

year
 Overview  
•  Batch  processing:  
–  Preprocessing  without  a  master  image  
pre_proc_init.csh  
–  Preprocessing  with  a  master  image  
pre_proc_batch.csh  
–  Align  a  stack  of  SAR  data  align_batch.csh  
–  Form  a  stack  of  interferograms  in2_batch.csh  
•  Stacking:  
–  Average  the  phase  stack_phase.bash  
–  Average  the  correla9on  stack_corr.bash  
–  Compute  standard  devia9ons  std.bash  
–  GPS/InSAR  integra9on  stack.csh  
 Stacking:  stack_phase.bash    

•  Func9on:  
–  Average  unwrapped  phase  
–  Convert  the  phase  (radius)  to  velocity  (mm/yr)  
•  Note:  it’s  necessary  to  check  the  unwrapped  phase  
before  stacking  or  9me-­‐series  analysis  because  
unwrapping  from  SNAPHU  may  give  errors,  which  
will  corrupt  results  
•  Complex  processing  techniques  (e.g.  filtering,  
detrending,  GPS/InSAR  integra9on)  can  be  
incorporated  along  with  stacking  
complete data set

 complete data set

Track  213  Frame  660    

Land  subsidence  near  Coachella  Valley,  California  

 Summary  
•  Batch  processing  shell  scripts  provide  automa9c  
InSAR  data  processing  (preprocess,  alignment,  
form  interferogram)  
•  Stacking  shell  scripts  (under-­‐development)  
provide  methods  to  es9mate  mean  velocity  and  
its  standard  devia9ons.  
•  Advanced  user  can  develop  custom  scripts  using  
tools  inside  GMT  and  GMTSAR.  
•  InSAR  9me-­‐series  analysis  can  be  developed  in  
the  next  step  of  processing.  
•  Any  ques9ons?