I calibrated Sentinel 1A and 1B products. Consequently, if I apply Back Geo Coding and Terrain correction the resulting slave product shows a lot of speckling as can be seen in the enclosed intensity image. The entire sequence of steps that are involved are as follows:
It probably has something to do with the BackGeocoding step because only the slave is affected and because the same results are obtained when excluding Terrain Correction from the processing chain. It does not have anything to do with Sentinel 1B images, as purely 1A products produce the same “NaN” valued speckles. Have you seen this issue before? Thank you in advance for the help.
This is a problem on how we handle 0 and the no data value with calibration. The issue is https://senbox.atlassian.net/browse/SITBX-408. I don’t know yet how best to solve it since 0 for most products is the nodata values coming from the product but can also be valid data.
For now you should be able to just right click on the bands, go to band properties and disable the no data value and then process your data.
I tried to process the same files with disabled “no data values” in the properties of the bands however this does not resolve the issue.
I disabled the no data values before calibration, and afterwards again, before Backgeocoding the calibrated data. Any other thoughts maybe? I will update you when I know more.
I am sorry to keep bothering you guys. I am still figuring out this issue completely. As you said it should be possible to turn off using NaN values in the image. I did this however still a lot of speckling exists. I just figured out that when leaving Terrain Correction out of the equation no speckling is visible. When including the Terrain Correction the speckling becomes visible.
The speckling occurse in case the following steps are taken:
Step 1:
"Unticking the NaN-values used in the origininal SAR image zip file in the SNAP gui and saving as BEAMDIMAP "
Step 2:
Read-Calibration-TopsarSplit-Write
Step 3:
(“Unticking the NaN Value used in SNAP gui in the resulting dim files of Step 2”)
Step 4:
ProdSetRead-BackGeoCoding-BandMaths(To deramp the master band)-Subset(To get the slave bands from the BackGeoCoding)-Merge(Subset&BandMaths)-TopSarDeburst-TerrainCorrection-Subset(region of interest)-Write
In case this clears up my issue for you guys, please let me know.
In fact strictly speaking calibration is only valid for the intensity image. Personally I think that I & Q should not be touched. One workaround is to make a separate stack of calibrated intensity images.
