I have been trying to understand the functionality “Terrain flattering” in SNAP. I am trying to use the function to ensure I measure the same land spot, from the somewhat same incidence angle. For a full season of winter wheat crops.
I have been trying to do the operation on a GRD IW product without success. It get the following error when I attempt to run the operation
I get that I do multi-looking and in this case, I needed set the range to 4, so I get 40m pixels instead of the normal 10m for this product. I derived at this value by trial and error. But is there some way to determine the demanded resolution of the DEM should be so that I could run with this GRD product directly. In my case, the multi-looking is removing too much information. Also if I used an external DEM, what are the demands?
I think my main challenge is that I do not completely understand, what the function is doing. Also, I was a bit surprised the output was the gamma_0 band for both VV and VH since it demanded the beta_0 as input. Is there some detailed description of the function so that I could get this understanding.
I know of sigma, beta, and gamma from this figure, but I would like to relate this to terrain flattening.
I hope to get a few hints.
to resolve the error, uncheck the “re-grid” box.
A bit of explanation can be found here:
Thank very much for the response, I think it explains my current problem.
I will reattempt to run the scan again a see what I get in terms of output.
I am still trying to investigate options for external DEM models.
In Denmark, we have a complete model for the whole country, with a 0.25m grid if I remember correctly.
This model is updated yearly by overflying with LIDAR.
But I am not sure if this DEM would be valid for sentinel-1 data and if it would provide an advantage.
I was hoping to get even better Terrain Corrections for each scan and perhaps improved Terrain Flattening.
Another thing that puzzles me is the need to download the DEM models each time. Would it not be possible to download all DEM models for my country and then redirect SNAP the select the correct one?
I think the better your DEM the higher the chances for good correction. But maybe 0.25 m will create smaller artefacts and you will have to smooth it a bit before using it.
Actually, if a DEM of an area was required once, snap will store it for later use under C:\Users\username\snap\auxdata\dem
I do not think using a higher resolution DEM than the SAR data is beneficial. I would recommend you downsample your DEM to your target pixel size.
Then I have must have a bug in my SNAP configuration, if the intention was that all downloaded DEM are stored. But I will check what is stored in my Ubuntu snap folder and or if there is a size limit parameter. I recently redid a seqence GRD scans, and both SNAP and GPT wanted to download many of DEM again. I did not check if it was all DEM models.
So something like agrid of 5x5 meters would be better ?
I assume it should be the double of the value of the GRD IW scan.
GRD resolution is 20m with 10m pixel size -> use 10x10 meters