I have some rasters that contain spectral data, and some spatially identical rasters that have the nadir information from the acquisition for each pixel. I'm looking for a way to mosaic these together, preferentially taking pixels with the lowest nadir for overlapping areas (i.e. keep the most directly underneath the drone pixels).
This is hyperspectral data, so it's quite big, some of the rasters are >200GB.
The best I've found with lots of googling is this: SAGA-GIS Module Library Documentation (v2.2.0)
This function has the ability to select a decision criteria (e.g. minimum), but it's poorly documented and I can't tell how it would deal with multiple bands - it looks like its designed for a single band raster (grid) eg a DEM. I suspect it isn't capable of mapping the results of the decision rule to the other bands. But I'm not able to find much info or many examples, so I'm not entirely sure.
I've been trying to use Xarray to do this work (joining the nadir and spectral data together as separate data arrays in the same dataset because it ties in with Dask and that should make it easier to deal with memory management and hopefully multithreading once I get it working. I'm going to post a question with some demo code separately because that is doing my head in.
At this point I'm keen for anything that will work (preferably open source).
Are there any GDAL functions or other programs that I've glossed over in my reading that might let me do this easily?