This resource has been updated. Click here to move to the new site.

InSAR Processing: Questions and Answers
  1. Should I be concerned by the error message " ** ERROR **: only 0 bytes are read out of 11812 requested" ?
  2. Should I be concerned by the error message "Note: Nonstandard floating-point mode enabled, See the  Numerical Computation Guide, ieee_sun(3M)"?
  3. What is wrong when the processing aborts following the message "Too few points left after culling: 2 left"?
  4. How much disk space is required to process one SAR pair?
  5. Why doesn't dgx.pl work?
  6. What should I do if nothing seems to work?
  7. How can I transfer the interferogram into an image file?
  8. Why did the processing fail to unwrap the interferogram?
  9. What is an acceptable maximum perpendicular baseline?
  10. How can I determine if the raw data file is corrupted?
  11. What has gone wrong when the processing pauses with "p must be at least 2 in moment"?
  12. Why does the unwrapping process fail in the program trees.c?
  13. Why does icu fail while trying to unwrap the image?
  14. Why does the amplitude band appear out of focus or blurred?


 
1)  Should I be concerned by the error message " ** ERROR **: only 0 bytes are read out of 11812 requested" ?
This error message is fairly common and can be ignored.
 
2)  Should I be concerned by the error message "Note: Nonstandardfloating-point mode enabled. See the Numerical Computation Guide ieee_sun(3M)"?
This warning is printed by all the programs that are compiled with the "-fast" optimization and can be ignored.
 
3) What is wrong when the processing aborts following the message "Too few points left after culling: 2 left"? 
This error occurs when the SAR processor is unsuccessful at determining the offset between the two SAR scenes.  Look at the file fitoff_ampcor.out.  If the "number of points remaining" is less that 50, then the processing will abort.  If the number is just under 50, you can try to reduce the threshold by running "fitoff" manually.  Look at the perl script make_offset.pl for the correct syntax.  If the number of points remaining is significantly less than 50, then the orbit or timing information for the two scenes may be incorrect resulting in a poor initial guess for the offset between scenes.  You may need to manually specify the offset between the two *slc files by including x_start and y_start in the *proc file.  Alternatively, check question 14 for another possibility.
 
4)  How much disk space is required to process one SAR pair?
You will need about 5 GB of disk space to run through the whole processing sequence.
 
5)  Why doesn't dgx.pl work?
The base program, dgx.f, was not included in the ROI_PAC distribution.  If you do have this program, then you may have to reduce the bit depth to 8 on your monitor or switch to a monitor on a different machine.  For SunOS, you can use the UNIX program "m64config" to change the bit depth.
 
6)  What should I do if nothing seems to work?
Often the processing routines fail if previous output files exist.  Try deleting intermediate files and start the process again.  If problems continue to arise, then delete all files and begin anew.
 
7)  How can I transfer the interferogram into an image file?
While viewing an image using dgx.pl you can use the left mouse button on the three buttons at the bottom of the main DGX window to select the image amplitude (Amp), height or phase (Hgt), or a combination of amplitude and height (Mix). If you press the middle mouse button on the image part of the window, you get a "Zoom" window that can zoom in or out and print what is visible in the window. If you want to see the full image, you can zoom out 4-5 times. Pressing the right mouse button on the "Print" button will write the image to a JPEG file.
 
8)  Why did the processing fail to unwrap the interferogram?
If the unwrapping process begins in a region with no coherence, then the unwrapping procedure will fail.  The point where this process begins can be set by defining the point in the input parameter file (date1-date2.proc) using the parameters unw_seedx and unw_seedy.   A seed point can be found by looking at any filtered, wrapped interferogram, such as "filt*.int".  Use DGX to view the file and choose a point in the image where the phase is smooth and coherent   The coordinates can be found on the top border by clicking the left mouse button while moving the pointer in the region of interest.
 
9)  What is an acceptable maximum perpendicular baseline? 
There is no set value for what perpendicular baseline is acceptable.  The maximum threshold will depend on what you consider to be an acceptable phase error resulting from the overall coherence of the image and the caliber of the DEM.   Any topographic signature remaining in the image may suggest that the perpendicular baseline is too large for the given DEM.  The baseline information for a processed tandem pair can be found in the file date1_date2_baseline.rsc.
 
10) How can I determine if a raw data file is corrupted?
In general, the size of the imagery file should be 300Mb (+/- 25 Mb).  The SARLEADER file should be about 28 Kb and the VDF file about 2 Kb.  Large deviations from these sizes might suggest a problem.  To test the file you can run the script make_raw.pl and examine the file date_parse_line.out.  This file will list lines in the file that have bad data values or missing data.  Several lines of missing data should not be a concern.  ROI_PAC simply duplicates the previous line.  If dozens or hundreds of adjacent lines are missing from a section of the file, then the file may be corrupted.  This problem could have originated while the data was downloaded from the satellite (for which you have no control over) or while you were transferring files on your local disk.
 
11)  What has gone wrong when the processing pauses with "p must be at least 2 in moment"?
The processing does not have enough information to calculate the statistical information necessary to determine the offset between two SAR scenes.  Type "go" to resume the processing, which should subsequently fail.  Refer to question #3 above for further instructions.
 
12) Why does the unwrapping process fail in the program trees.c? 
The program trees.c is inefficient at processing large files.  Try using the icu unwrapping method.
 
13) Why does icu fail while trying to unwrap the image?
Try running icu manually.  If the programs quits with an error message about the azimuth buffer size, you will need to lower the value in the icu input file.  If it complains about the range buffer size, the icu program is not setup for the RAM configuration on your computer.  You will need to edit the file sizes.inc in the ROI_PAC source directory for the icu package and recompile the program.
 
14) Why does the amplitude band appear out of focus or blurred? 
Occasionally the software incorrectly calculates the doppler ambiguity, resulting in a blurred image observed in the *.slc image.   The scripts will often fail during make_offset.pl.  This problem is most prevalent for data collected by ERS2 after February of 2000 due to the loss of one of its gyroscopes.  Check the DOPPLER_RANGE0 value in the *raw.rsc for both SAR scenes.  The value should range between 0.5 and -0.5.  If the value is outside this range, subtract 1 and replace the value with the result.  This value should be similar to values calculated along the same latitude, or previously processed SAR scenes of the same frame.


Answers complied with assistance from Eric Fielding.