High-Resolution Climate Projections Using Diffusion-Based Downscaling of a Lightweight Climate Emulator

2026-02-13

AI Summary

This study addresses the limitation of the LUCIE lightweight climate simulator, which, despite stably reproducing large-scale climate statistics, operates at a coarse resolution of approximately 300 km—insufficient for regional impact assessments. To overcome this, the authors introduce, for the first time, a probabilistic diffusion generative model for climate downscaling, integrating a conditional and posterior sampling framework to enhance LUCIE’s output from ~300 km to ~25 km resolution while preserving physical consistency. The method is built upon the Spherical Fourier Neural Operator and trained and validated using ERA5 reanalysis data. Experimental results demonstrate that the downscaled fields effectively retain the original dynamical structures, as evidenced by metrics including RMSE, power spectra, and empirical orthogonal functions (EOFs), while substantially improving the representation of regional-scale features.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Machine Learning: Deep Generative Models & Autoencoders
• Natural Language Processing: Generation

Application Category

• Web Mining and Content Analysis: Web data generation and simulation
• Search and Retrieval-Augmented AI: Large language models for search
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

The proliferation of data-driven models in weather and climate sciences has marked a significant paradigm shift, with advanced models demonstrating exceptional skill in medium-range forecasting. However, these models are often limited by long-term instabilities, climatological drift, and substantial computational costs during training and inference, restricting their broader application for climate studies. Addressing these limitations, Guan et al. (2024) introduced LUCIE, a lightweight, physically consistent climate emulator utilizing a Spherical Fourier Neural Operator (SFNO) architecture. This model is able to reproduce accurate long-term statistics including climatological mean and seasonal variability. However, LUCIE's native resolution (~300 km) is inadequate for detailed regional impact assessments. To overcome this limitation, we introduce a deep learning-based downscaling framework, leveraging probabilistic diffusion-based generative models with conditional and posterior sampling frameworks. These models downscale coarse LUCIE outputs to 25 km resolution. They are trained on approximately 14,000 ERA5 timesteps spanning 2000-2009 and evaluated on LUCIE predictions from 2010 to 2020. Model performance is assessed through diverse metrics, including latitude-averaged RMSE, power spectrum, probability density functions and First Empirical Orthogonal Function of the zonal wind. We observe that the proposed approach is able to preserve the coarse-grained dynamics from LUCIE while generating fine-scaled climatological statistics at ~28km resolution.

Problem

• climate downscaling
• high-resolution climate projections
• lightweight climate emulator
• spatial resolution
• regional impact assessment

Innovation

• diffusion-based downscaling
• lightweight climate emulator
• Spherical Fourier Neural Operator
• high-resolution climate projection
• probabilistic generative modeling