search

ppolabs / jlinda

jLinda - Java Library for Interferometric Data Analysis
9 stars 5 forks source link

Topo Phase removal Op; error for external dem #20

Open AndreaMinchella opened 11 years ago

AndreaMinchella commented 11 years ago

Version used: NEST 5.16 OS: Linux Centos 64 bit

Processing a tandem pair over the Lazio Area in Italy, I was able successfully to remove the topography employing the SRTM and ASTER Dems but I got an error message (see attachment) using one external DEM. The external dem is wgs84 ll coordinates. screenshot-1

pmar commented 11 years ago

This issue is under further investigation.

pmar commented 11 years ago

Context and some background on this issue:

External DEM's, are being parsed differently than the standard (eg, ASTER, SRTM, etc) ones in core functionality of NEST. This can cause conflicts with InSAR functionality.

The external DEM's are handled directly, parse what you have and that's it.

The main difficulty with this is when an extra DEM information around tile is required, while the external DEM is not sufficiently large enough. This extra information, is needed in order to guarantee the smooth and reliable results. These extents are computed for ever tile separately, and depend on the actual topographic signal in the tile. In a case that 'external' DEM does not contain enough coverage, including this dynamically computed extent, the errors like reported would be triggered.

Not doing any fiddling with external DEM, was a design choice. Making sure that SAR product is smaller than the external DEM would resolve the issue. Usually few km's should be enough, unless the area of interest is eg Himalayas. However, this should be better reported to the user.

Possible solutions:

  1. More robust check on overlap between external DEM and SAR product. In case of not sufficient overlap trigger an error, explain where and what is the problem.
  2. Fill extents needed with values estimated from available DEM, doing some stats, and filling extent needed with the values coming from some smart statistics. Also, warn user that the results might be unreliable.
pmar commented 11 years ago

Also @zbiper proposed resolving this issue in following way:

if your DEM is smaller that SAR crop it would be nice whether we can automatically mask the non overlapping area and continue.