Strange effects in the RTC result

I used Sentinel-1 GRDH data to do the RTC processing with the following steps:
1.Apply orbits
2.Radiometric -> Calibrate
3.Multilook
4.Speckle filter
5. Radiometric > Radiometric Terrain Flattening (SRTM 1 Sec hgt auto download)
In the results (Gamma0), I saw strange effects in the upper part of the image?

You can check the SRTM by right-click > Add Elevation Band and see if there are errors, but it rather looks like a systematic bias.

Please try to run Terrain Flattening directly after Calibration. My feeling is that multi-looking might interfere

Have you tried to play with the oversampling and and overlap parameters to test if the error somehow changes?

Thank you very much for your prompt reply. I just tried to run Terrain Flattening directly after Calibration without multilooking and speckle filtering, but the effects are still there. I checked the DEM and no problem with the DEM.

then this is my only idea: