10m Offset

I have been using the same Lidar DEM for the SAR orthorectification in SNAP, and getting the water mask out of it. I am wondering whether any one knows by any chance why there is an offset of 10 m when I overlap my SAR-derived water mask raster outputs?

I had similar issue between optical and SAR-derived water masks, but for the case above I am using the same orthorectification algorithm,using the same SAR processing approach. What doesn’t go right?

Thanks