Towards Kriging-informed Conditional Diffusion for Regional Sea-Level Data Downscaling
Subhankar Ghosh, Arun Sharma, Jayant Gupta, Aneesh Subramanian, Shashi Shekhar
TL;DR
This work tackles regional climate downscaling by integrating geostatistics with diffusion probabilistic modeling. The Ki-CDPM framework conditions a conditional diffusion process on a Kriging-derived high-resolution input $\mathbf{y}$, with a variogram-based regularizer $\mathcal{R}_V$ that aligns the generated spatial structure with observed high-resolution variability. Empirical results on sea-level elevation and eddy kinetic energy demonstrate that Ki-CDPM outperforms state-of-the-art baselines in RMSE, MAE, PCC, and CRPS, while providing probabilistic outputs. The approach offers improved spatial fidelity and uncertainty quantification, enabling more reliable regional climate impact assessments and coastal management.
Abstract
Given coarser-resolution projections from global climate models or satellite data, the downscaling problem aims to estimate finer-resolution regional climate data, capturing fine-scale spatial patterns and variability. Downscaling is any method to derive high-resolution data from low-resolution variables, often to provide more detailed and local predictions and analyses. This problem is societally crucial for effective adaptation, mitigation, and resilience against significant risks from climate change. The challenge arises from spatial heterogeneity and the need to recover finer-scale features while ensuring model generalization. Most downscaling methods \cite{Li2020} fail to capture the spatial dependencies at finer scales and underperform on real-world climate datasets, such as sea-level rise. We propose a novel Kriging-informed Conditional Diffusion Probabilistic Model (Ki-CDPM) to capture spatial variability while preserving fine-scale features. Experimental results on climate data show that our proposed method is more accurate than state-of-the-art downscaling techniques.