Description

Adds new submodule earthkit.meteo.regimes with classes to define a weather regime classification based on patterns and functions to project anomaly fields onto these patterns to compute a regime index, following the approach of Michel and Rivière (2011).

The implementation aims to cater to users at three levels:

• High-level: Users that want to apply a regime classification, i.e., compute a regime index, using a preconfigured instance of RegimePatterns from elsewhere. These users only need functions project and standardise.
• Medium-level: Users that create a new regime classification by instantiating one of the provided subclasses of RegimePatterns (i.e., the required pattern generation is already implemented).
• Low-level: Users that create a new regime classification and need to implement a new pattern generation scheme in a class inheriting from RegimePatterns.

Classes to represent fixed patterns (ConstantRegimePatterns) and base patterns modulated by a custom scalar function (ModulatedRegimePatterns) are provided with this first implementation. An example notebook shows how to use the data of Grams to recreate their 7-pattern regimes classification and apply it to test data.

Implementations are at array-level with dictionaries used to label outputs based on their association with individual regimes.

Features to consider in follow-up work

• Higher-level interface compatible with FieldList and xarray
• Support for grids other than lat-lon
• Automatic regridding and cropping of input data based on the gridspec in the pattern class with earthkit.regrid
• Automatic generation of area weights based on the gridspec
• Tools to extract regime patterns from data (EOFs and clustering methods are most commonly used)

Contributor Declaration

By opening this pull request, I affirm the following:

• All authors agree to the Contributor License Agreement.
• The code follows the project's coding standards.
• I have performed self-review and added comments where needed.
• I have added or updated tests to verify that my changes are effective and functional.
• I have run all existing tests and confirmed they pass.