Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat

2026-09-30

AI Summary

This study addresses the high training costs and unclear data requirements associated with kilometer-scale downscaling simulations of extreme heat by proposing the CASPER model. Methodologically, CASPER employs a U-Net architecture, a structure-preserving loss function, and reanalysis data downscaling techniques. Crucially, it reveals a linear scaling relationship between prediction error and climatic distance, demonstrating that the model can achieve comparable accuracy using only a few simulated months. Experimental results indicate that during heatwaves, CASPER maintains errors below 1.8 K while effectively preserving fine-grained spatial structures. By substantially reducing both training data demands and computational barriers, this work establishes an efficient data utilization paradigm for high-resolution climate simulations.

Abstract

Extreme heat is where urban adaptation needs kilometer-scale data the most, but the simulations training a downscaler can cost more than they save, and how much is needed has not been identified. We measured it with CASPER, a U-Net with a structure-preserving loss downscaling 32 km reanalysis to 1 km temperature, humidity and wind, across 24 configurations of one to eight months. Held-out error grows linearly with climatological distance to the training data, RMSE = 0.83 + 2.95 d, explaining 90% of its variance against 7% for volume and predicting unseen months in advance. On held-out extreme summer weeks CASPER preserves the fine-scale structure and cross-variable physics that matched-budget baselines degrade, and matches station observations during documented heat waves to within 1.8 K. Transfer to a new region degrades geographically; 11 days of local simulation cuts Vancouver's held-out error from 3.8 to 1.3 K. Training periods should span the target climate: the same accuracy for four times less simulation, putting kilometer-scale downscaling of extreme heat within reach of groups without large computing facilities.

Problem

• extreme heat
• kilometer-scale downscaling
• data efficiency
• error-distance scaling
• climate transferability

Innovation

• downscaling
• U-Net
• structure-preserving loss
• error-distance scaling
• data efficiency