arXiv：cs.LG· Kirien Whan, Nikolaj T. M\"ucke, Karin van der Wiel·· 3 小时前AI 评分35
EC-EarthFlow：用流匹配概率模拟 EC-Earth3 每日瞬态全球气候模拟
EC-EarthFlow: Probabilistic emulation of daily transient global climate model simulations with flow matching
AI 导读
EC-EarthFlow 是一个生成式流匹配模型，用于模拟物理气候模型 EC-Earth3 的瞬态模拟结果，基于 1950-2166 年 SSP2-4.5 情景数据训练。模型仅用温度变量、以自回归方式预测次日温度场，支持一个月至更长季节的 rollout，能复现 EC-Earth3 的日变化、空间格局、年循环与长期趋势，且计算成本大幅低于物理模型，长时间推理稳定。
正文
Abstract:We introduce EC-EarthFlow, a generative flow matching model that emulates simulations from the physical climate model EC-Earth3. The model is trained on transient simulations from EC-Earth3 (1950-2166, SSP2-4.5) to predict the day ahead temperature field from the previous days temperature as well as annual mean temperature. Predictions are made auto-regressively with rollout periods of between a month and an extended season. Using only this variable of interest, we are able to reproduce the daily variability, spatial patterns, annual cycle and long-term trend from EC-Earth3 at a substantially lower computational cost than the physical model. We demonstrate that EC-EarthFlow is stable for long inference periods, and that it can learn the physical relationships as simulated in EC-Earth3.
| Subjects: | Machine Learning (cs.LG); Atmospheric and Oceanic Physics (physics.ao-ph) |
| Cite as: | arXiv:2610.09715 [cs.LG] |
| (or arXiv:2610.09715v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.09715 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Nikolaj Takata Mücke [view email]
[v1]
Wed, 7 Oct 2026 09:10:57 UTC (10,215 KB)
来源：arXiv：cs.LG · arxiv.org