Mosaic, Clip and Create DEM

I have four image tiles of Sentinel-1A GRD. I have to mosaic it then clip the image by vector file and then create DEM. I tried to mosaic using SAR Mosaic tool in RADAR tab. After mosaic I tried to create DEM using Rang Doppler Terrain Correction tool but I found error which says

GRD-data is not suitable for DEM-generation, you’ll need SLC:s. Please be aware that the quality of the generated DEMs is likely to be low compared against publicly available ones.

1 Like

@mitkotecha Allow me to hop in and recommend my work on the retrieval of digital elevation models from Sentinel-1 radar data - open applications, techniques and limitations


thank you for your reply. I get SLC data. Now I have to generate DEM. I have four image tiles so first I have to mosaic images and then generate DEM or I have to generate DEM and then mosaic it? Please give me suggestion.

From a technical perspective you probably first create DEMs from all tiles and then mosaic the final result. Otherwise you would have to combine TOPS Merge and Slice Assembly at once which could be challenging

But 4 tiles sound like a massive area…
May I ask why you need such a large DEM? Creating a DEM from Sentinel-1 can work out to get some detailed topographic features for a specific region, but the quality for such a large area will probably not be better than from an openly available DEM, such as SRTM or AW3D30.
The chances for mosaicing errors, unwrapping errors, atmospheric contributions and others are just too high over this extent.

Actually I am going in Geoinformatics. My thesis topic is Flash flood susceptibility using Machine Learning. My study area is Luni river basin Rajasthan, India. For flood susceptibility I need DEM to analyze Flow direction, Flow Accumulation, Slope, Aspect and all hydrological parameters.
I think you have more experience so can you suggest something? As you told I am facing too much errors in each and every steps.

For hydrological analysis you need a consistent DEM without errors. The openly available ones (Copernicus 30, ALOS World 3D 30, SRTM…) provide this quality. It makes little sense to me to create one yourself from Sentinel-1 over such large areas just to have a higher spatial resolution, but with drastically more errors.

No supervisor should actually urge you to create a DEM yourself from Sentinel-1 without making sure that this is feasible.

Why not start with a small area (1 sub-swath and 3-4 bursts) to test how the result could look like, then compare it to openly available DEMs and if the quality is sufficient, you can try to increase the area.

1 Like