I processed ALOS-2 PALSAR-2 Dual pol Fine (SM3) from L 1.1 CEOS. The resulting calibrate, multi-looked, speckle filtered, terrain flattened and terrain corrected image (see screenshot: Gamma0, HV polarisation) present very high gamma0 values in steep terrain. Is there any way this can be improved at all?
a comparison with terrain corrected Sigma0 would be interesting. My suspicion is that the Terrain Flattening introduced these werid lines, either because the DEM is not good enough or the topography is too extreme.
Is this a highly mountainous area?
Yes, you are right the terrain flattening before the terrain correction introduces those lines. Below the sigma0 HV without terrain flattening. I use the SRTM 1sec HGT DEM with a resolution which is lower than the resolution of the output so that could also influence things. The topography is also quite extreme with slopes of 50 degrees.
If it is indeed the terrain flattening implementation in SNAP are you aware of other open source tools that I could test in order to do try an alternative terrain flattening? It might be exactly the same with other tools given the rugged terrain though.
ASF MapReady had an excellent terrain normalization, but to my knowledge it does not support ALOS-2 products, as it was not further developed.
But I know that the Terrain Flattening in SNAP is currently being evaluated and improved. Can’t tell however when it will be finished.