EnScale: Temporally‐Consistent Multivariate Generative Downscaling via Proper Scoring Rules
- 1ETH Zurich Zurich Switzerland
- 2Swiss Data Science Center EPFL and ETH Zurich Zurich Switzerland
- 3University of Washington Seattle WA USA
Abstract
The practical use of future climate projections from global circulation models (GCMs) is often limited by their coarse spatial resolution, requiring downscaling to generate high‐resolution data. Regional climate models (RCMs) provide this refinement, but are computationally expensive. To address this issue, machine learning (ML) models can learn the downscaling function, mapping coarse GCM outputs to high‐resolution fields. Among these, generative approaches aim to capture the full conditional distribution of RCM data given coarse‐scale GCM data, which is characterized by large variability and thus challenging to model accurately. We introduce EnScale , a generative ML framework emulating the full GCM‐to‐RCM map by training on multiple pairs of GCM and corresponding RCM data. It first adjusts large‐scale mismatches between GCM and coarsened RCM data, followed by a super‐resolution step to generate high‐resolution fields. To efficiently model the high‐dimensional output, the super‐resolution step employs a novel class of sparse local stochastic layers. Both steps employ generative models optimized with the energy score, a proper scoring rule. Compared to state‐of‐the‐art ML downscaling approaches, our setup reduces computational cost by about one order of magnitude. EnScale jointly emulates multiple variables—temperature, precipitation, solar radiation, and wind—spatially consistent over Central Europe. In addition, we propose a variant EnScale‐t that enables temporally consistent downscaling. We establish a comprehensive evaluation framework across various categories including calibration, spatial and temporal structure, extremes, and multivariate dependencies. Comparison with diverse benchmarks demonstrates EnScale(‐t) 's competitive performance and computational efficiency, offering a promising approach for accurate and temporally consistent RCM emulation.
Related
- Generative modelling for multivariate spatio-temporal downscaling of monthly climate fields
- Kilometer-Scale AI Downscaling of Atlantic Hurricanes with Generative Ensembles
- Stress-Testing Dynamical and Generative Downscaling Using Subseasonal Extreme Precipitation Forecasts
- Sea surface salinity downscaling using deep generative diffusion models
- A temporally consistent global 500 m-resolution monthly VIIRS-like nighttime light dataset (1992–2024)
BibTeX
@article{schillinger2026enscale,
title = {{EnScale: Temporally‐Consistent Multivariate Generative Downscaling via Proper Scoring Rules}},
author = {Maybritt Schillinger and Maxim Samarin and Xinwei Shen and Reto Knutti and Nicolai Meinshausen},
journal = {Journal of Advances in Modeling Earth Systems},
year = {2026},
month = aug,
doi = {10.1029/2025ms005593},
url = {https://doi.org/10.1029/2025ms005593},
note = {CC BY},
}
RIS
TY - JOUR
TI - EnScale: Temporally‐Consistent Multivariate Generative Downscaling via Proper Scoring Rules
AU - Maybritt Schillinger
AU - Maxim Samarin
AU - Xinwei Shen
AU - Reto Knutti
AU - Nicolai Meinshausen
PY - 2026
DA - 2026/08/01
JO - Journal of Advances in Modeling Earth Systems
DO - 10.1029/2025ms005593
UR - https://doi.org/10.1029/2025ms005593
AB - The practical use of future climate projections from global circulation models (GCMs) is often limited by their coarse spatial resolution, requiring downscaling to generate high‐resolution data. Regional climate models (RCMs) provide this refinement, but are computationally expensive. To address this issue, machine learning (ML) models can learn the downscaling function, mapping coarse GCM outputs to high‐resolution fields. Among these, generative approaches aim to capture the full conditional distribution of RCM data given coarse‐scale GCM data, which is characterized by large variability and thus challenging to model accurately. We introduce EnScale , a generative ML framework emulating the full GCM‐to‐RCM map by training on multiple pairs of GCM and corresponding RCM data. It first adjusts large‐scale mismatches between GCM and coarsened RCM data, followed by a super‐resolution step to generate high‐resolution fields. To efficiently model the high‐dimensional output, the super‐resolution step employs a novel class of sparse local stochastic layers. Both steps employ generative models optimized with the energy score, a proper scoring rule. Compared to state‐of‐the‐art ML downscaling approaches, our setup reduces computational cost by about one order of magnitude. EnScale jointly emulates multiple variables—temperature, precipitation, solar radiation, and wind—spatially consistent over Central Europe. In addition, we propose a variant EnScale‐t that enables temporally consistent downscaling. We establish a comprehensive evaluation framework across various categories including calibration, spatial and temporal structure, extremes, and multivariate dependencies. Comparison with diverse benchmarks demonstrates EnScale(‐t) 's competitive performance and computational efficiency, offering a promising approach for accurate and temporally consistent RCM emulation.
SN - 1942-2466
ER -