We use gpt to batch process large numbers of S-1 GRD files. We updated SNAP to 9.0 and we are now getting unwanted zeros over water bodies. I compared an image processed with an earlier version of SNAP (6 or 7 I think) with those produced with the newer version and compared the two. It looks to me like the application of a water mask, as opposed to the setting of very low values to a nodata value of 0 due to thermal noise removal or something like that. This is because it is leaving similar low values near shores, my guess is this is because of an imperfect water mask near the coasts.
We have not changed our python script that calls gpt. We are running the gpt commands Apply-Orbit-File, Calibration, Terrain-Flattening, and Terrain-Correction. Sometimes we don’t apply Terrain-Flattening and still get the problem.
Any input on where/how to turn off application of the water mask (if that is indeed the problem) would be appreciated.