Comparisons on different raster features from different Python libraries based on xarray, with or without dask.
This repository contains a series of Jupyter notebooks comparing how open-source Python libraries handle several raster-processing tasks.
The following libraries have been considered :
rasteriorioxarrayodc-geogeoutilsxarray-spatial
- opening raster files
- reading bands and metadata
- accessing GCPs and RPCs
- saving with different settings : locally, to COG and to Zarr formats
- reading a dataset by window
- decimating a raster
- reprojecting rasters to a new CRS (without RPC)
- reproject_match a DEM with invalid CRS to a reference raster
- creating spatial masks based on geometry
- cropping rasters to vector polygons or bounding boxes
- read multiple raster tiles
- combine them into a single raster file
- rasterizing vector geometries into rasters
- vectorizing raster regions into geometries
- how nodata values are handled between different libraries
- pros and cons between NaNs and masked arrays
- nodata interpolation to fill missing values
- usecase on reprojecting a raster and performance comparison with or without dask