Personally, I undestand the information at pixels affected by shadowing as lost. It cannot be restored from both radiometric or geometric terrain correction.
What Terrain Flattening is able to do is to brighten the backscatter from slopes facing away from the sensor (and to reduce the backscatter of slopes facing towards the sensor) based on a DEM. But this only works for pixels which produced backscatter to the sensor which is not the case for shadow areas.
many thx for your kind usual answer.
I’m a bit confused regarding the concept of “lost information” related to the shadows areas. I’ve understood that, in SAR, shadows is an area where, for looking angle reason, the signals can’t arrive (for this reason there is no backscatter).
A couple of questions to better understand:
1-every pixel that change values in TerrainFlatteningGamma0 (comparing with Sigma0), shouldn’t be considered shadows?
2-Which values the deep shadows areas, should take? Maybe NaN? Maybe the lowest of the range values?
This is really hard to tell because the DEM used for Terrain Flattening is only an approximation of the actual surfaces. The radar satellite itself does not know if its signal has reached ground or not. It simply puts together an image from the returned signal. Accordingly, these pixels do not exist (before terrain correction).
You can estimate shadow areas (based on SRTM) with the SAR Simulation operator.