rioxarray clip - make sure to pass in the CRS of the geometries

[snowman2]

https://www.earthdatascience.org/courses/use-data-open-source-python/multispectral-remote-sensing/modis-data-in-python/

This is where I noticed it. I would recommend going through and making sure it is used in other clipping examples.

https://corteva.github.io/rioxarray/stable/rioxarray.html#rioxarray.raster_array.RasterArray.clip
https://corteva.github.io/rioxarray/stable/examples/clip_geom.html#Clip-using-a-GeoDataFrame

modis_clip_geometry = modis_bands_pre.rio.clip(
    fire_bound_sin.geometry,
    crs=fire_bound_sin.crs,  ## ADD CRS HERE
    all_touched=True,
    from_disk=True,
)

This ensures that the geometries are the same CRS as the raster when clipping.

[nkorinek]

Hey @snowman2 ! So in regards to this, we did that initially, but started avoiding implementing it in our lessons. We found the way that the clip function reprojects vector data added a slight tilt to the data, which in turn tilted the outcome, which wasn't what we were hoping for in our lessons. I made a gist visually representing the difference we were finding, and this happened across landsat, MODIS, and NAIP data. https://gist.github.com/nkorinek/94a69b125bbb97044c2c49a60cf94643 if there's a solution here please let us know!

[snowman2]

The reason I suggest to add the CRS is that it is a good practice to follow. Since this is a tutorial, I recommend adding it so users are used to doing it.

We found the way that the clip function reprojects vector data added a slight tilt to the data, which in turn tilted the outcome, which wasn't what we were hoping for in our lessons.

The tilt is due to the CRS of the raw raster data. MODIS shows this difference best since as it uses the sinusoidal projection. I think showing the tilt and then re-projecting the raster in the lesson will be valuable for users to see.

[snowman2]

Looks like it is discussed here: https://www.earthdatascience.org/courses/use-data-open-source-python/hierarchical-data-formats-hdf/open-MODIS-hdf4-files-python/