When working with Sentinel-1 data over an ocean, which ellipsoid correction is preferred? I guess the Geolocation Grid since there is no terrain effects to be corrected. Can someone clarify when to use the Average Height Range Doppler and when to use the Geolocation Grid?
In which case should the Range doppler terrain correction be used instead of the two ones previously mentioned?
whenever topographic variation could lead to geometric distortions in your SAR backscatter RD Terrain Correction can help to reduce them. For ocean bodies the tool won’t work in SNAP as no elevation data can be downloaded over water areas.
Thanks for the answer.
What about the two options for ellipsoid correction?
I cannot answer that to a sufficient degree, sorry.
Waiting then for some community feedback
I don’t think there should be much difference between the two, if I remember correctly the recommended one is Average Height Zero Doppler.
I’ve been testing both options in different scenarios (for my ocean application) and not significant differences are founded. I guess my question is more theoretical than practical.
Thanks for the feedback !
Unless things have changed in recent versions of SNAP, Range Doppler (RD) TC with S1 images should be very accurate, whether over land or over water, so long as accurate height values are used.
I don’t know exactly how ellipsoid correction (EC) works or how the data used by EC is obtained. When I tested it in the past, EC was not as accurate as RD TC over water.
See thread1 and thread2 for more info.
Based on the latter thread, I’d say that:
- RD TC with SRTM
- RD TC with EGM08 with ‘Apply Earth Gravitational Model’ unticked
should give similar very accurate geometric results over water. This is so because RD TC with the standard DEMs (eg SRTM) incorporates a gravitational model (which is similar to EGM08, I assume).
I guess that in your case it shouldn’t make no difference.
The S-1 geolocation grid is computed using a (coarse) DEM. Over the ocean the average height will be 0. Meaning that if using the geolocation grid it is equivalent as using a flat earth i.e. an ellipsoid.
I guess that the Average Height Range Doppler function of SNAP (that I confess I don’t know) will do a similar thing: estimating an average height using a DEM over the scene. If the scene is over the ocean then the height should be 0 … hence falling back to the first case. I will people knowing this function to confirm or correct my understanding.