Evaluating Cross-region Generalization for Wavelet-Diffusion Precipitation Downscaling
Summary
This study evaluates the cross-region and cross-event generalization of the Wavelet Diffusion Model for precipitation downscaling using U.S. regional data, showing that models trained on limited regions can perform competitively in unseen areas.
View Cached Full Text
Cached at: 09/25/26, 09:36 AM
# Evaluating Cross-region Generalization for Wavelet-Diffusion Precipitation Downscaling Source: [https://arxiv.org/abs/2609.28749](https://arxiv.org/abs/2609.28749) [View PDF](https://arxiv.org/pdf/2609.28749) > Abstract:Diffusion models have shown strong potential for kilometer\-scale precipitation downscaling, but their performance in geographically unseen regions and event regimes remains insufficiently understood\. Building on the wavelet diffusion model \(WDM\) framework, this study evaluates cross\-region and cross\-event generalization\. Six 3 x 3 deg U\.S\. regions represent convective, winter, tropical, and atmospheric\-river precipitation regimes\. Low\-resolution inputs are generated by block averaging NOAA Multi\-Radar/Multi\-Sensor \(MRMS\) composite reflectivity fields\. A WDM trained only on Oklahoma \(OK\) samples and a WDM trained on all six regions are compared with nearest\-neighbor and Bicubic interpolation\. Model performance is evaluated using three metric families that measure image\-domain reconstruction, spectral and distributional fidelity, and bin\-wise precipitation detection\. The OK\-trained WDM remains competitive outside OK\. Although the all\-region WDM delivers the best and most consistent overall image\-domain and detection performance, its gains are uneven across precipitation intensities\. Bin\-wise critical success index \(CSI\) over 5\-dBZ reflectivity bins shows that WDM improvements concentrate in localized higher\-reflectivity structures, which image\-domain metrics partly obscure\. In addition, the performance differences among samples are strongly associated with the spatial organization of the precipitation field, quantified by Moran's I as the spatial autocorrelation of each reflectivity bin\. The sample\-level Moran's I\-CSI correlation stratified by sample intensity reaches 0\.901 in all six regions, including regions unseen during training\. Overall, these findings support future efforts to transfer downscaling models to regions with limited local training data and to generate globally consistent, high\-resolution precipitation products\. ## Submission history From: Weikang Qian \[[view email](https://arxiv.org/show-email/20b4e91b/2609.28749)\] **\[v1\]**Wed, 23 Sep 2026 19:47:31 UTC \(2,061 KB\)
Similar Articles
Multi-Quantile Regression for Extreme Precipitation Downscaling
Introduces Q-srdrn, a multi-quantile super-resolution network using pinball loss to improve extreme precipitation downscaling, achieving dramatic detection rate gains for heavy rainfall events while maintaining overall accuracy.
Transferable Dual-Stream Representations for Mesoscale-Preserving Sea Surface Temperature Downscaling
This paper introduces EddyFlow, a deep learning framework for kilometer-scale sea surface temperature downscaling that balances predictive accuracy, scale-dependent structure, and regional generalization. It achieves strong zero-shot performance and near-ideal spectral fidelity across multiple ocean regions.
Domain-Adaptive Climate Downscaling Under Temporal Distribution Shift
This paper investigates temporal out-of-distribution shift in deep-learning-based climate downscaling and proposes a domain-adaptive framework that combines supervised reconstruction with domain alignment to improve high-resolution climate projections under non-stationary conditions.
SimCast-S2S: An Efficient Generative Model for Subseasonal Precipitation Forecasting via Transfer Learning from Climate Simulations
SimCast-S2S is a generative latent-diffusion framework for probabilistic subseasonal precipitation forecasting that leverages transfer learning from climate simulations to outperform deep learning baselines and compete with operational systems.
Iterative Refinement Diffusion for Super-Resolved Data Assimilation of Multiscale Physical Systems
This paper introduces an iterative refinement framework for data assimilation that combines neural operators and diffusion models to super-resolve multiscale physical systems, outperforming baselines on benchmarks like Kraichnan turbulence.