Earth observation embeddings are effective sub-grid descriptors for probabilistic weather downscaling
Abstract
Global weather reanalyses and forecasts resolve the evolving atmospheric state on coarse grids, but site-specific applications require predictions at arbitrary locations where near-surface conditions also depend on unresolved terrain and land-surface properties. Existing probabilistic downscalers address this gap using hand-crafted topographic and surface descriptors. We ask instead whether Earth observation foundation models can provide transferable subgrid surface representations for probabilistic weather downscaling. We augment a convolutional conditional neural process (ConvCNP) that downscales coarse ERA5 reanalysis fields at ~25 km resolution with a learned local surface descriptor, obtained by compressing a patch of TESSERA embeddings at 10 m resolution. Although these embeddings summarize annual surface conditions, they improve downscaling by encoding persistent surface properties that capture a location's departure from the coarse-grid atmospheric state. Across five climatically diverse regions, the embedding improves point and probabilistic skill at stations held out in both space and time, overall improving CRPS skill by 11.5% for 2 m temperature and 6.2% for 10 m wind speed relative to a topography-only ConvCNP baseline. A hand-crafted descriptor incorporating richer surface information than topography alone captures comparable persistent subgrid signal but yields far smaller predictive gains than the learned embedding representation. These improvements persist when forecasts from the Aurora AI model replace ERA5 reanalysis fields and when predicting at newly deployed weather station networks. To our knowledge, this is the first evidence that long-timescale Earth observation embeddings can support short-timescale weather downscaling where subgrid departures are systematically structured by persistent surface properties.
Related
- Efficient Analog-Initialized Latent Transport for Spatially Coherent Probabilistic Downscaling from Global to Kilometer Scales
- Emergent Sparsity: How Many Effective Variables Survive Coarse-Graining in AlphaEarth Embeddings?
- Deep learning-based downscaling of PM2.5, NO2 and O3 over Europe across emission-reduction scenarios
- Statistical Downscaling of Daily Precipitation from ERA5 Atmospheric Fields to IMERG over West Africa
- Spherical Physics-informed Neural Operator with Multi-scale Coupling for Meteorological Downscaling
BibTeX
@misc{sousa2026earth,
title = {{Earth observation embeddings are effective sub-grid descriptors for probabilistic weather downscaling}},
author = {Pedro Sousa and Will Tebbutt and Sadiq Jaffer and Robin Young and Anil Madhavapeddy and Richard E. Turner},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2608.12271},
eprint = {2608.12271},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2608.12271},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Earth observation embeddings are effective sub-grid descriptors for probabilistic weather downscaling
AU - Pedro Sousa
AU - Will Tebbutt
AU - Sadiq Jaffer
AU - Robin Young
AU - Anil Madhavapeddy
AU - Richard E. Turner
PY - 2026
DA - 2026/09/03
PB - arXiv
DO - 10.48550/arxiv.2608.12271
UR - https://arxiv.org/abs/2608.12271
AB - Global weather reanalyses and forecasts resolve the evolving atmospheric state on coarse grids, but site-specific applications require predictions at arbitrary locations where near-surface conditions also depend on unresolved terrain and land-surface properties. Existing probabilistic downscalers address this gap using hand-crafted topographic and surface descriptors. We ask instead whether Earth observation foundation models can provide transferable subgrid surface representations for probabilistic weather downscaling. We augment a convolutional conditional neural process (ConvCNP) that downscales coarse ERA5 reanalysis fields at ~25 km resolution with a learned local surface descriptor, obtained by compressing a patch of TESSERA embeddings at 10 m resolution. Although these embeddings summarize annual surface conditions, they improve downscaling by encoding persistent surface properties that capture a location's departure from the coarse-grid atmospheric state. Across five climatically diverse regions, the embedding improves point and probabilistic skill at stations held out in both space and time, overall improving CRPS skill by 11.5% for 2 m temperature and 6.2% for 10 m wind speed relative to a topography-only ConvCNP baseline. A hand-crafted descriptor incorporating richer surface information than topography alone captures comparable persistent subgrid signal but yields far smaller predictive gains than the learned embedding representation. These improvements persist when forecasts from the Aurora AI model replace ERA5 reanalysis fields and when predicting at newly deployed weather station networks. To our knowledge, this is the first evidence that long-timescale Earth observation embeddings can support short-timescale weather downscaling where subgrid departures are systematically structured by persistent surface properties.
ER -