Diffusion models for probabilistic precipitation generation from atmospheric variables
Michael Aich, Sebastian Bathiany, Philipp Hess, Yu Huang, Niklas Boers · 2025
Read Diffusion models for probabilistic precipitation generation from atmospheric variables by Michael Aich, Sebastian Bathiany, Philipp Hess, Yu Huang, Niklas Boers (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-00307.txt.
Description
Michael Aich and co-authors generate rainfall fields with diffusion models in a 2025 climate paper. Earth system models do not compute rain directly; they rely on parameterized column physics that misses local detail and carries notable biases, especially for extremes like floods and droughts.
Their conditional diffusion model, built on a UNet, takes a handful of large-scale atmospheric variables and produces high-resolution global daily precipitation. Trained on the ERA5 reanalysis, it also transfers to climate model data it never saw. Because the method samples many outcomes, it yields probabilistic forecasts that carry their own uncertainty estimates.
Catalog record
| CREATOR | Michael Aich, Sebastian Bathiany, Philipp Hess, Yu Huang, Niklas Boers |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-00307