Generative climate downscaling enables high-resolution compound risk assessment by preserving multivariate dependencies
For climate risk assessment, it addresses the critical problem of preserving multivariate dependencies in downscaled climate projections, which is essential for accurate compound hazard analysis.
This work introduces a diffusion-based multivariate generative downscaling framework that preserves inter-variable correlations under a 50x resolution increase, reducing correlation errors by over fourfold compared to baselines and improving drought detection accuracy.
Physics-based climate projections using general circulation models are essential for assessing future risks, but their coarse resolution limits regional decision-making. Statistical downscaling can efficiently add detail, yet many methods treat variables independently, degrading inter-variable relationships that govern compound hazards such as heat stress, drought, and wildfire. Here we show that a diffusion-based multivariate generative framework, combined with bias correction, recovers degraded inter-variable correlations even under a 50$\times$ increase in linear resolution. When applied to five meteorological variables over Japan, the framework reduces inter-variable correlation errors by more than fourfold relative to existing baselines while improving both univariate and spatial accuracy, leading to more accurate detection of severe drought. These results demonstrate that multivariate generative downscaling improves the reliability of compound risk assessment under large resolution gaps.