Convert points to S1 projection?


Has anyone done or seen the use case, where surface location is converted into S1 file projection? To illustrate, let’s say one has a fixed area on the surface and needs data only about that area. Avoiding terrain correction would save quite some computational resources.


You could try projecting the geographical coordinates of your area directly onto the S1’s system. Just be aware that skipping terrain correction might affect accuracy, especially in uneven terrains.

Yes, but I do not have a good idea, of where to start from. I understand the problem with uneven terrain, it can be ignored initially. But even for that, I guess it should be possible to figure out the correction for each orbit and these should remain constant.

GDAL provides different methods for coordinate transformation between CRS coordinates and image coordinates, even when the image is only georeferenced using ground control points, like Sentinel-1 GRD products are.

Have a look at gdaltransform. Most precise results will be obtained using thin plate splines to interpolate the GCPs, this is the -tps option of gdaltransform. GDALs GCP based transformers are also available in Python using the OSGeo Python bindings or using the rasterio library (where TPS will also be available in the upcoming release).


Does gdaltrasform handle DEMs as well? DEMs are necessary when dealing with geocoding of SAR data.

No, gdaltransform will only interpolate the given set of GCPs, which in case of Sentinel-1 will be the geolocation grid generated by the IPF. So if the IPF does not take a DEM into account when generating the geolocation grid (does it?), then of course this method will not be very accurate if there’s a lot of elevation in the AOI.

The S-1 IPF does not perform geocoding with a DEM.

1 Like