Earth observation embeddings are effective sub-grid descriptors for probabilistic weather downscaling
2026-08-12 • Machine Learning
Machine Learning
AI summaryⓘ
The authors explored how to improve weather predictions at specific spots by using detailed Earth observation data instead of just broad topographic features. They combined coarse weather data with fine-scale surface information derived from TESSERA embeddings, which capture persistent land features over long time periods. This approach helped make better temperature and wind speed forecasts at individual locations across different regions. The improvements held up even when using different weather models or new observation sites. This work shows that long-term Earth observation data can enhance short-term weather predictions by accounting for local surface characteristics.
weather downscalingERA5 reanalysisTESSERA embeddingsconvolutional neural processsurface descriptorsprobabilistic forecastingsub-grid variability2 meter temperature10 meter wind speedEarth observation models
Authors
Pedro Sousa, Will Tebbutt, Sadiq Jaffer, Robin Young, Anil Madhavapeddy, Richard E. Turner
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 descriptors. We ask instead whether Earth observation foundation models can provide transferable sub-grid surface representations for probabilistic weather downscaling. We augment a convolutional conditional neural process 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 summarise surface conditions over annual timescales, they improve downscaling of instantaneous 2 m temperature and 10 m wind speed 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. We further analyse how its contribution differs by variable, finding that topography explains more of temperature's sub-grid structure, while TESSERA provides additional surface information for wind speed. These improvements persist when the coarse input is changed from ERA5 to forecasts from the Aurora AI forecasting model, and when predicting at newly deployed stations with no regional history. To our knowledge, this is the first evidence that long-timescale Earth-observation embeddings can support short-timescale weather downscaling where sub-grid departures are systematically structured by persistent surface properties.