用简单方法实现顶尖天气预报，训练快10倍、推理快10倍。
U-Cast: A Surprisingly Simple and Efficient Frontier Probabilistic AI Weather Forecaster
- 基于标准U-Net，先用MAE预训练再用CRPS微调，仅需简单流程。
- 在1.5°分辨率下性能超GenCast和IFS ENS，训练计算量减少10倍以上。
- 适合想低成本部署前沿概率天气模型的研究者与团队。
基于AI的天气预报已媲美传统物理集合预报，但当前最先进模型依赖专用架构与巨大算力，门槛高。我们证明此类复杂性对顶尖性能并非必需。提出 extit{U-Cast}，一个基于标准U-Net的概率预报器，采用简单训练流程：先在平均绝对误差（MAE）上进行确定性预训练，再用蒙特卡洛丢弃实现随机性的连续排名概率评分（CRPS）短时微调。结果表明，该模型在1.5°分辨率下性能达到或超越GenCast和IFS ENS，训练算力低于领先CRPS模型10倍以上，推理延迟低于扩散模型10倍以上。U-Cast训练耗时不足12个H200 GPU天，可在3秒内生成15天集合预报。这些结果表明，通用可扩展架构结合高效训练方案，可在极低成本下实现复杂领域专用设计的顶尖表现，为更广泛社区提供前沿概率天气建模的新可能。
原文摘要 · Abstract (English)
AI-based weather forecasting now rivals traditional physics-based ensembles, but state-of-the-art (SOTA) models rely on specialized architectures and massive computational budgets, creating a high barrier to entry. We demonstrate that such complexity is unnecessary for frontier performance. We introduce \ours, a probabilistic forecaster built on a standard U-Net backbone trained with a simple recipe: deterministic pre-training on Mean Absolute Error followed by short probabilistic fine-tuning on the Continuous Ranked Probability Score (CRPS) using Monte Carlo Dropout for stochasticity. As a result, our model matches or exceeds the probabilistic skill of GenCast and IFS ENS at $1.5^\circ$ resolution while reducing training compute by over $10\times$ compared to leading CRPS-based models and inference latency by over $10\times$ compared to diffusion-based models. U-Cast trains in under 12 H200 GPU-days and generates a 15-day ensemble forecast in 3 seconds. These results suggest that scalable, general-purpose architectures paired with efficient training curricula can match complex domain-specific designs at a fraction of the cost, opening the training of frontier probabilistic weather models to the broader community.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。