跳到正文
arXiv cs.AI· Ruian Tie, Wenbo Xiong, Zhengyu Shi, Xinyu Su, Chenyu jiang, Libo Wu, Hao Li·· 4 小时前AI 评分29

GCD:面向异构 GCM 零样本降尺度的 GCM 一致性扩散模型

GCD: GCM-consistent Diffusion for Zero-shot Downscaling across Heterogeneous GCMs

AI 导读

研究者提出 GCM 一致性扩散模型 GCD,用于异构 GCM 的零样本气候降尺度,通过地理条件化、谱滤波测量引导和蒸馏加速的共轭梯度引导,将推理从 1000 步扩散降至 6 步,同时保持与粗分辨率 GCM 输入的一致性。在五个异构 GCM 上,GCD 取得有竞争力的 99 百分位极端事件精度,并能在无需针对特定模型重训练的情况下重建合理的热带气旋高频结构。

正文 · 原文

View PDF HTML (experimental)

Abstract:Climate downscaling of General Circulation Models (GCMs) is an ill-posed inverse problem, as GCM outputs and reanalysis fields arise from different dynamical systems and lack fine-scale temporal pairing. Supervised methods thus rely on synthetic degradations or model-specific adaptation, limiting generalization across heterogeneous GCMs. Zero-shot diffusion posterior sampling (DPS) avoids paired supervision by combining a high-resolution climate prior with inference-time constraints from coarse GCM inputs, but faces consistency and efficiency challenges. We propose GCM-consistent Diffusion for Zero-shot Downscaling (GCD), which addresses these challenges with geographic conditioning, spectrally filtered measurement guidance, and distillation-accelerated Conjugate Gradient guidance. GCD reduces inference from 1,000 diffusion steps to 6 while maintaining consistency with the coarse GCM inputs. Across five heterogeneous GCMs, GCD achieves competitive 99th-percentile extreme-event accuracy and reconstructs plausible high-frequency tropical cyclone structures without model-specific retraining. Github: this https URL
Subjects: Artificial Intelligence (cs.AI)
Cite as: arXiv:2601.21760 [cs.AI]
  (or arXiv:2601.21760v3 [cs.AI] for this version)
  https://doi.org/10.48550/arXiv.2601.21760

arXiv-issued DOI via DataCite

Submission history

From: Ruian Tie [view email]
[v1] Thu, 29 Jan 2026 14:14:41 UTC (11,268 KB)
[v2] Sun, 8 Feb 2026 15:48:40 UTC (11,268 KB)
[v3] Tue, 17 Nov 2026 05:18:26 UTC (13,878 KB)

来源:arXiv cs.AI · arxiv.org