Climate model adds km scale detail faster with exact probabilities
Climate downscaling with likelihood functional flow matching
Machine Learning
Summary
Local decisions about weather need detailed information, but climate models typically show details only at large scales. The authors created a new method that gives very fine, kilometer-level climate details with a model you can ask for any location or size. Their approach makes generating these details much faster and more accurate than previous methods, and it can also show how uncertain the predictions are exactly. They tested it on temperature and precipitation data over northern Europe and it worked better than other models even ones much bigger.
What this means in practice
- •For weather forecasters: Produce highly detailed local temperature and precipitation forecasts by enhancing large-scale climate model outputs with faster and more precise downscaling.
- •For renewable energy planners: Generate fine-scale daily climate ensembles to improve risk assessments for heat, frost, and flooding impacting energy infrastructure investments.
Authors
Zhi-Song Liu, Helmi Toropainen, Michael Boy
Abstract
Decisions about heat, frost and flooding are made locally, yet reanalyses and climate models resolve tens of kilometres. Generative downscalers add kilometre-scale detail with uncertainty, but are tied to a grid, costly to sample and lack a tractable likelihood. Here we introduce CD-FFM, a functional flow-matching model that treats the high-resolution field as a continuous random field that one model can query at any location and resolution. Its velocity acts pointwise, so spatial variability comes from a noise prior fitted to the observed spectrum and every observation has an exact likelihood. Downscaling ERA5 to 1 km daily temperature over Fennoscandia, CD-FFM lowers the continuous ranked probability score by 27-31% against regression and 12-18% against a forty-times larger neural operator in each of three test years, and outperforms a CorrDiff-style diffusion model while sampling about fifty times faster. The same model transfers to precipitation and makes large kilometre-scale ensembles affordable.