发表机构
IBM Research; Fathom; STFC Hartree Centre（IBM研究院; Fathom; STFC哈特雷中心）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对降水降尺度问题，探究三种扩散模型条件策略，发现交叉注意力条件优于通道拼接，且Prithvi WxC基础模型在数据有限场景下性能良好。
AI 中文摘要
高分辨率降水场对水文影响评估至关重要，但全球气候模型输出过于粗糙且存在偏差，无法直接使用。基于AI的扩散模型统计降尺度是一种有前景的方法，但大尺度大气预测因子如何条件生成的机制仍未被充分探究。我们研究了应用于日降水降尺度的去噪扩散概率模型的三种条件策略：上采样粗分辨率预测因子的通道拼接、带可学习卷积编码器的交叉注意力条件、带预训练Prithvi WxC天气基础模型的冻结编码器的交叉注意力条件。所有策略在相同条件下，针对科罗拉多河流域，与无条件基线进行评估，使用概率、分布、频谱和极端事件指标。拼接条件实现了最低的逐点CRPS和MSE，但易产生过度平滑的场，抑制高强度事件。相比之下，交叉注意力条件提供了显著更好的分布真实性和适度的频谱保真度提升，对极端事件的改进最大：Prithvi-WxC条件模型保留了超过一半的>100mm/天事件，但因样本有限估计存在不确定性。在完整数据集上训练时，可学习卷积模型的表现与基础模型条件方法相似，且所需计算资源更低；不过，Prithvi-WxC条件模型仅用5年训练数据即可达到相当性能。这些结果表明，交叉注意力条件在概率降水降尺度方面优于简单拼接，且预训练基础模型表示在数据受限场景中可能带来益处。
英文摘要
High-resolution precipitation fields are essential for hydrological impact assessment, yet global climate model outputs are too coarse and biased for direct use. AI-based statistical downscaling with diffusion models offers a promising approach, but the mechanism by which large-scale atmospheric predictors condition generation remains largely unexplored. We investigate three conditioning strategies for a denoising diffusion probabilistic model applied to daily precipitation downscaling: channel concatenation of upsampled coarse predictors, cross-attention conditioning with a learned convolutional encoder, and cross-attention conditioning with the frozen encoder of the pretrained Prithvi WxC weather foundation model. All strategies are evaluated against an unconditioned baseline under identical conditions using probabilistic, distributional, spectral, and extreme-event metrics for the Colorado River Basin. Concatenation conditioning achieves the lowest point-wise CRPS and MSE, but tends to produce over-smoothed fields that suppress high-intensity events. In contrast, cross-attention conditioning provides substantially better distributional realism and modest improvements in spectral fidelity. Improvements are greatest for extremes: the Prithvi-WxC conditioned model retains over half of >100mm/day events, although estimates are uncertain due to limited samples. When trained on the full dataset, the learned convolutional model performs similarly to the foundation model-conditioned approach while requiring lower computational resources. However, the Prithvi-WxC-conditioned model achieves comparable performance with only five years of training data. These results indicate that cross-attention conditioning offers advantages over simple concatenation for probabilistic precipitation downscaling, and that pre-trained foundation model representations may offer benefits in data-limited settings.
