Huge ensembles part I design of ensemble weather forecasts with spherical Fourier neural operators; Huge ensembles part II properties of a huge ensemble of hindcasts generated with spherical Fourier neural operators

Studying low-likelihood high-impact extreme weather events in a warming world is a significant and challenging task for current ensemble forecasting systems. While these systems presently use up to 100 members, larger ensembles could enrich the sampling of internal variability. They may capture the long tails associated with climate hazards better than traditional ensemble sizes. Due to computational constraints, it is infeasible to generate huge ensembles (comprised of 1,000-10,000 members) with traditional, physics-based numerical models. In this two-part paper, we replace traditional numerical simulations with machine learning (ML) to generate hindcasts of huge ensembles. In Part I, we construct an ensemble weather forecasting system based on Spherical Fourier Neural Operators (SFNO), and we discuss important design decisions for constructing such an ensemble. The ensemble represents model uncertainty through perturbed-parameter techniques, and it represents initial condition uncerta..., , , # Huge ensembles part I design of ensemble weather forecasts with spherical Fourier neural operators; Huge ensembles part II properties of a huge ensemble of hindcasts generated with spherical Fourier neural operators

## Changelog

The April version of the files includes code changes to calculate the scores on diagnostics (reliability diagrams, CRPS, skill of best member) on new variables: cold temperature extremes, 10m wind speed, and the heat index. These changes are reflected in the `score_extreme_diagnostics_v2morevars.py` and `extreme_scoring_utils_v2morevars.py` files in the codebase and the tar files for the figures. These updated files reflect the code and data analysis versions as of April 2025.

## HENS

This codebase is a fork of the earth2mip repository and the modulus-makani codebase, developed by NVIDIA. It is used to run huge ensemble (HENS) weather forecasts with the SFNO architecture. It serves as the codebase for the following two papers:

\"Huge Ensembles Part II: Pro...,