Artefacts in S1TBX radiometric terrain flattening

[Guillaume]

The TF tool is renaming the Beta_Nought into Gamma_Nought, so I considered gamma_GRD_noTF and gamma_GRD_TF. Maybe one should compare beta_GRD_noTF and gamma_GRD_TF?

[eyeinsky]

Yes, after TF, the output is terrain-flattened gamma nought, so that is correct. But as an input to TF, one should provide a radar geometry image in the backscatter convention beta nought, not gamma nought.

[ChrisBoedinger]

Hi @mengdahl, hi all,
I am trying to correct for terrain using the Terrain Flattening operator and Range Doppler Terrain Correction. However results after Terrain Flattening are highly unsatisfactory (s. Screeshots). I wonder if this is (still) a issue of the algorithm or if there is a way to improve the output.
I´m working on Win10 64bit, Snap 6.0 Beta, Example dataset: S1B_IW_GRDH_1SDV_20170110T095447_20170110T095512_003784_00681B_B4AB.SAFE (Chilean Andes)
I use a gap filled TanDEM-X DEM (doesnt cover the whole radar scene) so I doubt that the issue is within the DEM (Projection UTM 19S, resampled to 10m with bilinear; also tried WGS84 Lat/long).
I applied (and varied) the following pre-processing steps: Thermal Noise Removal >Apply Orbit File > Calibrate to beta > (Subset) > Speckle Filter >Terrain Flattening > Terrain Correction

– Calibrated Image –

– DEM (same extent as TC&TF below) –

– Terrain Flattened and Terrain Corrected (without additional normalization) –

– Terrain Corrected –
Range Doppler with radiometric normalization set performs well in referencing the image to the source DEM but correction for terrain is weak and shadows largely persist (as well as bright areas)
Similar posts:

Terrain Flattening Black Areas

dear @lveciI I have images from TerraSAR 3m resolution And would like to apply terrain flattening and then Terrain Correction. Which DEM will do the job better? 3 or 1 sec? Kind Regards

Terrain correction setting for S1 IW-GRD data

Hello all, I’m working on a multi-temporal S1 IW-GRD pair. I tried applying Terrain Correction on the SIGMA0 calibrated image for mitigating the SAR distortions. I both considered Range Doppler TC and SAR Simulation TC. For both, I used the SRTM1Sec DEM (I guess it should take into account the gravitational model of the Earth too) ; the DEM interpolation was conducted with either BiSinc or Biinear, while resampling with either Bilinear or NN, respectively. The output corrected map, anyway, s…

Best
Chris

[mengdahl]

Could you fix the orientation of the images - it’s very difficult to compare the results for the moment. The terrain flattened scene looks good IMO, I don’t know what is going on with the terrain corrected version.

[ChrisBoedinger]

Hi @mengdahl,
I edited the screenshots, hope its now easier to compare. I´m not sure with the TF image, the blurriness and the NaN holes (in the TF&TC image) are much different from what I know of corrected images from MapReady (which of course uses a different algorithm and doesnt support S1). Let me know if I can provide you with any additional info