Open Data Cube Algorithms

Xarray and Dask friendly EO Processing Algorithms and Utilities.

Note: This package only contains algorithms. If you want to use them for processing EO data, you'll find using odc-stats much simpler.

• Cloud Masking
• Geometric Median
• Percentiles
• Dask Aware Raster Reprojection
• Efficiently generating and saving Cloud Optimized GeoTIFFs to S3
• Reshaping Dask Arrays for Efficient Computation
• Converting between Floats with NaNs and Ints with nodata values

Installation

pip install odc-algo

Usage

Building

1. Install the Python build tool. python -m pip install build
2. Build this package. python -m build

Development

1. Follow build instructions
2. Install as dev pip install -e .[dev]

Alternatively, install with the whl file.

Tasks

• Decide whether to use pixi, or uv. I think pixi, it handles Rust stuff.
• Document the Geomedian API we're trying to expose. See odc-stats
• Document the Percentile API we're exposing.
• Regresssion Tests instead of installing old dependencies.
• Update GitHub Actions to build and test Rust backend
• Update GitHub Actions to build and test against multiple Python versions
• Consider what type of binary wheels to build. [abi3/multi-python version compatible is tempting]
• Update Rust dependencies.
• De-duplicate with odc-geo. It includes COG and Warp functionality that's better maintained, but may not be identical...
• Consider vendoring the skimage morphology functions we're using, instead of depending on the whole thing.