用生成模型提升气候极端事件预测精度，解决传统模型分辨率低、计算贵的问题。
Generative artificial intelligence improves projections of climate extremes
- 基于流匹配与分布对齐的生成式深度学习框架，实现多变量降尺度。
- 在多种排放情景下保持高精度与稳定性，可模拟复合极端事件。
- 适合气候建模、灾害风险评估等领域的研究者使用。
气候变化正加剧极端事件，威胁生物多样性、人类健康和粮食安全。全球气候模型（GCMs）虽是未来气候预测的关键工具，但其分辨率粗糙且计算成本高，难以准确表征极端气候。本文提出FuXi-CMIPAlign，一种用于降尺度CMIP输出的生成式深度学习框架。该模型结合流匹配（Flow Matching）进行生成建模，并通过最大均值差异（MMD）损失实现领域自适应，以对齐训练数据与推理数据的特征分布，从而缓解输入差异，提升模型在不同排放情景下的准确性、稳定性和泛化能力。该方法可实现空间、时间及多变量降尺度，更真实地模拟复合极端事件（如热带气旋）。
原文摘要 · Abstract (English)
Climate change is amplifying extreme events, posing escalating risks to biodiversity, human health, and food security. GCMs are essential for projecting future climate, yet their coarse resolution and high computational costs constrain their ability to represent extremes. Here, we introduce FuXi-CMIPAlign, a generative deep learning framework for downscaling CMIP outputs. The model integrates Flow Matching for generative modeling with domain adaptation via MMD loss to align feature distributions between training data and inference data, thereby mitigating input discrepancies and improving accuracy, stability, and generalization across emission scenarios. FuXi-CMIPAlign performs spatial, temporal, and multivariate downscaling, enabling more realistic simulation of compound extremes such as TCs.
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。