Open pr2718 opened 1 year ago
The TDS01 ERS1 example dataset has range size 5000 for current development version, but baseline has 4991, hence a 5 pixel padding exclusion.
TDS09 has 5128 vs 5141, hence a 6/7 padding.
Focussing quality should also improve for the areas not in the middle of picture (maybe need to modify usage of Vr as well).
Echo metadata might need to be propagated as well, because the parser does not know a) how much of an extra lines should be parsed (besides sensing start and end) b) when windowing range and azimuth, the metadata shall be updated
There is something interesting happening regarding SWST for ERS here - https://github.com/gmtsar/gmtsar/blob/master/preproc/ERS_preproc/ers_line_fixer/ers_line_fixer.c#L434
One option would be to simply pad left and right for the range as it is done here - https://github.com/isce-framework/isce2/blob/main/components/isceobj/Sensor/ERS_EnviSAT.py#L557 This would be instead of cutting range.
Investigate sar_metadata.h CalcKa() and CalcAperturePixels(). Further on investigate L1 datasets which L0 has not a long sensing time -> therefore try to find how much was cut by the reference processor (because it cannot include more pixels in the beginning and in the end and then should publish less data than requested by the sensing end and start?!). Try to correlate these findings with CalcAperturePixels()
Results windowing is implemented with this pull request - https://github.com/cgi-estonia-space/asar-focus/pull/33
Range compression windowing is implemented by this PR - https://github.com/cgi-estonia-space/asar-focus/pull/34
Range/Azimuth compression windowing