Currently there is no specific methods in xsarsea to get a NRCS (aka sigma0 denoised) value from a Sentinel-1 product as a Level-2 like. For instance, WV products computing the average value of xsarsigma0 values lead to values up to 3 dB below compare to the value provided in official ESA Level-2 product (oswNrcsNezCorr variable).
Looking at the ESA processor it seems that some steps are potentially missing in xsarsea.
Typically what ESA processor do to compute the sigma0 is:
1) read digital number
2) apply calibration (dividing by the sigma0_lut²)
3) low pass filter the image
4) keep intensity multiplying by <abs(image)²>
5) remove platform (I suppose it is related to https://en.wikipedia.org/wiki/Apodization technics.
6) final average sigma0 value is `mean((image np.conj(image)).real)`
I don't know if we need to be able to reproduce the sigma0 given by ESA
Currently there is no specific methods in
xsarseato get a NRCS (aka sigma0 denoised) value from a Sentinel-1 product as a Level-2 like. For instance, WV products computing the average value ofxsarsigma0values lead to values up to 3 dB below compare to the value provided in official ESA Level-2 product (oswNrcsNezCorrvariable). Looking at the ESA processor it seems that some steps are potentially missing inxsarsea. Typically what ESA processor do to compute the sigma0 is: 1) read digital number 2) apply calibration (dividing by the sigma0_lut²) 3) low pass filter the image 4) keep intensity multiplying by <abs(image)²> 5) remove platform (I suppose it is related to https://en.wikipedia.org/wiki/Apodization technics. 6) final average sigma0 value is `mean((image np.conj(image)).real)`