EnScale: Temporally-consistent multivariate generative downscaling via proper scoring rules
Maybritt Schillinger, Maxim Samarin, Xinwei Shen, Reto Knutti, Nicolai Meinshausen · 2025
Read EnScale: Temporally-consistent multivariate generative downscaling via proper scoring rules by Maybritt Schillinger, Maxim Samarin, Xinwei Shen, Reto Knutti, Nicolai Meinshausen (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-2509-26258.txt.
Description
Maybritt Schillinger and co-authors present a learning method that sharpens coarse climate forecasts, in work from 2025. World climate models run on a coarse grid, so local detail must be added later. Full regional models add that detail but cost a great deal of computing time.
Their system learns the full mapping from coarse to fine fields across many paired runs. A first step fixes large-scale mismatch, and a second step adds fine detail. Sparse local random layers keep the huge output tractable. Training uses scoring rules that reward honest spread, not just close means. Output stays consistent over time and across variables, a need for impact studies.
Catalog record
| CREATOR | Maybritt Schillinger, Maxim Samarin, Xinwei Shen, Reto Knutti, Nicolai Meinshausen |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 105 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-26258