Adding energy and water budget diagnostics (for evaluation of ML & physical models)

[davelee2804]

What problem does your feature request solve?

New diagnostics for global time series of energy (kinetic, potential, internal) budgets and exchanges, and moisture budgets, to evaluate the fidelity of these in ML (and physical models)

Describe the solution you'd like

Global and user specified regional integrals of energy and moisture components and energy exchanges as time series for ML and physical models (deterministic and ensemble)

This would involve back end common implementations for the lat-lon grid of:

• full domain integrals at each level
• gradient and divergent operators at each level
• vertical integrals and derivatives on pressure levels

And front end implementations where the above functionality is applied to compute budgets and exchanges on datasets for different sources (ie: AIFS, ERA5, IFS, LFRic, etc) over specified domains, time periods and ensemble sizes

Describe alternatives you've considered

Computing these diagnostics directly on the ML model grid (as an unstructured Delaunay/Voronoi mesh), rather than having these re-mapped to the lat-lon grid (possible future extension).

Bespoke implementation of diagnostics for the ML Hackathon:
https://github.com/JorgeBornemann/Momentum_ML_Hackathon_2025/tree/energy_diagnostics/theme_2/energy_diagnostics

Related Scores Github issue: nci/scores#995

[davelee2804]

Hi @jfrost-mo , @JorgeBornemann , Following our meeting last night, I've created the above Github issue. Please feel free to comment/edit as you see fit.

If you want make a CSET branch I could start to fill in some of the functionality here. Also we could invite the others to contribute as well (if we know their Github handles).

[jfrost-mo]

I've added yourself and @tennlee to the CSET team, so you should now be able to create a branch yourself, though a fork might be easier if you want to add more people.