1.1.0

• added MLX array backend support
• added GRIT support, and added precomputed grit segments network
• changed precomputed river network storage location from github to sites.ecmwf.int
  • added precomputed cama_01min, hydrosheds_30sec
• various bugfixes:
  • device handling issue for torch
  • fixed jax array backend
  • xarray metadata issues in some edge cases
• security: upgraded pyo3 and numpy crate versions 0.19 -> 0.26

1.0.0

Complete refactoring of the code with many new features:

• 2-level API: high-level based on xarray and low-level based on standard array (typically Numpy)
• Support for multiple standard array backends (Numpy, Cupy, Torch, Tensorflow, JAX)
• Support for bifurcations
• Extractment of subnetworks and cropping
• Addition of coordinate information to river networks
• New documentation on readthedocs
• Improved API
• Downstream statistics as well as upstream statistics

Full Changelog: 0.1.4...1.0.0

0.1.4

• speed up network creation using Rust (for topological label computation)
  • set up mixed Rust-Python project tooling
• add merit hydro reader
• add ability to read files in pcraster .map format

0.1.3

• Adds variance and standard deviation as possible metrics
• Adds distance and length calculations

0.1.2

Make compatible for inclusion in pip install earthkit.

0.1.1

Ensure compatibility with pypi.

0.1.0

First release with the following features:

• Support for PCRaster, CaMa-Flood and HydroSHEDS river networks
• Computing upstream statistics and statistics over catchments and subcatchments
• Finding catchments and subcatchments
• Calculation of upstream or downstream fields
• Handle arbitrary missing values
• Handle N-dimensional fields