A Probabilistic U-Net Approach to Downscaling Climate Simulations

TL;DR

This paper introduces a probabilistic U-Net model for statistical downscaling of climate simulations, effectively capturing uncertainty and improving the resolution of precipitation and temperature data.

Contribution

It adapts the probabilistic U-Net architecture for climate downscaling, evaluating multiple training objectives to enhance the modeling of extremes and spatial variability.

Findings

WMSE-MS-SSIM performs well for extremes under certain settings.

afCRPS better captures spatial variability across scales.

The model effectively captures aleatoric uncertainty in climate data.

Abstract

Climate models are limited by heavy computational costs, often producing outputs at coarse spatial resolutions, while many climate change impact studies require finer scales. Statistical downscaling bridges this gap, and we adapt the probabilistic U-Net for this task, combining a deterministic U-Net backbone with a variational latent space to capture aleatoric uncertainty. We evaluate four training objectives, afCRPS and WMSE-MS-SSIM with three settings for downscaling precipitation and temperature from $16\times$ coarser resolution. Our main finding is that WMSE-MS-SSIM performs well for extremes under certain settings, whereas afCRPS better captures spatial variability across scales.