Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat
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.
Related
- Efficient Analog-Initialized Latent Transport for Spatially Coherent Probabilistic Downscaling from Global to Kilometer Scales
- Kilometer-Scale AI Downscaling of Atlantic Hurricanes with Generative Ensembles
- SwAIther-Precip: Lead-Time-Aware Bias Correction Enables Kilometer-Scale Downscaling of Global AI Precipitation Forecasts over Switzerland
- Spatial Heterogeneity and Inhomogeneity of Cirrus Microphysical Properties on Kilometer Spatial Scales
- Diffusion-Based Refinement for Kilometer-Scale Probabilistic Precipitation Nowcasting
BibTeX
@misc{marey2026less,
title = {{Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat}},
author = {Ahmed Marey and Henry Lu and Abhishek Gaur and Sherif Goubran and Malek Aloui and Theodore Potsis and David Rolnick and Alex Hernandez-Garcia and Liangzhu Leon Wang},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2609.40140},
eprint = {2609.40140},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.40140},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Less is more: error-distance scaling relation for data-efficient kilometer-scale downscaling of extreme heat
AU - Ahmed Marey
AU - Henry Lu
AU - Abhishek Gaur
AU - Sherif Goubran
AU - Malek Aloui
AU - Theodore Potsis
AU - David Rolnick
AU - Alex Hernandez-Garcia
AU - Liangzhu Leon Wang
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2609.40140
UR - https://arxiv.org/abs/2609.40140
AB - 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.
ER -