Open Source Rasters Comparison

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 :

• rasterio
• rioxarray
• odc-geo
• geoutils
  
  

Notebooks Overview

1. 📖 io_and_attributes.ipynb

• opening raster files
• reading bands and metadata
• accessing GCPs and RPCs
• saving with different settings : locally, to COG and to Zarr formats

2. 🪟 windows_and_decimation.ipynb

• reading a dataset by window
• decimating a raster

3. 📽️ reprojection.ipynb

• reprojecting rasters to a new CRS (without RPC)
• reproject_match a DEM with invalid CRS to a reference raster

4. 🎭 mask_and_crop.ipynb

• creating spatial masks based on geometry
• cropping rasters to vector polygons or bounding boxes

5. 🔗 merge.ipynb

• read multiple raster tiles
• combine them into a single raster file

6. 🗺️ rasterization_and_vectorization.ipynb

• rasterizing vector geometries into rasters
• vectorizing raster regions into geometries

7. 🗄️ nodata_management.ipynb

• how nodata values are handled between different libraries
• pros and cons between NaNs and masked arrays
• nodata interpolation to fill missing values

8. 🎯 usecase_reproj_dask.ipynb

• usecase on reprojecting a raster and performance comparison with or without dask