Use load_stac to read in rasterized extractions

[VincentVerelst]

After the extractions we will have 4 overarching STAC collections. (S1, S2, Agera5 and ground truth).

• [x] Make sure load_stac is able to load in each STAC collection. The assets are netcdf's with time series
• [x] Make sure the backend and Python client have the band information (needed for feature computation later)
• [x] Merge the 4 collections with merge_cubes

This is a subissue of https://github.com/Open-EO/openeo-gfmap/issues/4

[VincentVerelst]

Currently, we get the following error message: Java.lang.IllegalArgumentException: GDAL Could not retrieve time values from netcdf.

Code example:

import openeo
c = openeo.connect('openeo.vito.be').authenticate_oidc()
test = c.load_stac('/data/users/Public/vincent.verelst/world_cereal/extractions_bys2tile/stac/collection.json')
test.execute_batch('test_load_stac.nc')

Job ID: j-240409358b0841e7b8e48dfb4e7df3e8 backend: openeo.vito.be

[jdries]

@VincentVerelst make sure to try on openeo-dev. I fixed a very similar issue for you last week I believe, but not sure if we promoted the fix to prod already.

[VincentVerelst]

Confirmed, load_stac works on openeo-dev. Two important notes:

• Not working on prod yet
• Also need to wait for next release of the Python client to have the band information also on the client side. Currently need to use a local checkout of the python client master branch.

[VincentVerelst]

Currently getting geotrellis.raster.gdal.MalformedDataTypeException: Unable to determine NoData value. GDAL Exception Code: 4. when trying to load the ground truth data (netCDF without time dimension)

Loading the S2 data (netCDF with time dimension) works.

Both on backend openeo-dev.vito.be

[VincentVerelst]

Currently blocked by https://github.com/Open-EO/openeo-geotrellis-extensions/issues/282

[VincentVerelst]

with https://github.com/Open-EO/openeo-gfmap/pull/99 it should be able to also read in the ground truth netCDF with load_stac