Filtering changes the value of sigma0_db to strange big numbers!

Hello all!

I am new to Radarsat 2 data using the Sentinel Toolbox. I used to process Sentinel 1 data without any problem, but right now I get the odd pixel values after filtering.

  1. Multilooking (# R 1, #Az 2, GR 10.7)
    2.Calibration (saved as db- for Radarsat2 it sounds like both sigma0 and then db conversion are done together)

  2. Single Product Speckle Filter (I dont know which one works better for Radarsat2 (Lee Sigma, Lee, Refined Lee). Using any of those the db values will be changed to higher pixel values which is not correct! Practically, the db values will be destroyed! However, filtering is supposed to reduce the noise, not change the pixel values dramatically like that!

  3. Terrain correction (Doopler TC)

Hope sb. can help me with that!
Thanks

The filter probably assumes it will see linear values. In general dB-conversion should be the last step as most operations (like mean) do not produce a correct result if performed on a dB-image.

1 Like

Yes, the filters are not expecting dB data. We’ll add a warning message for this. Conversion to dB should be done last.

1 Like

thanks a lot for your reply! I didnt pay attention to that!

I have found similar problem when I run Radarsat2 Calibration in SNAP, where there is no option to select only output in sigma naught! I have sigma naught_db that I dont want besides gamma and beta naught.

Thanks a lot

The default output is sigma0. You don’t need to select any checkbox in the UI.

1 Like