---
title: "Earth Observation Embeddings as Effective Sub-grid Descriptors for Probabilistic Weather Downscaling"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.12271v1"
permalink: "https://feed.astralor.com/items/itm_6e69ae71482ceb8438d83c3b"
published: "2026-08-12T17:10:42.000Z"
score: 78
channel: "ai"
tags: ["Aurora","ERA5","TESSERA","Earth Observation Foundation Models","Probabilistic Downscaling"]
---

# Earth Observation Embeddings as Effective Sub-grid Descriptors for Probabilistic Weather Downscaling

## AI commentary

> What stands out is the elegant combination of static surface information with transient atmospheric prediction, replacing hand-crafted features with foundation model embeddings. For teams working on operational weather AI, especially those needing station-level forecasts in unobserved regions, the transferability results are highly relevant.

## At a glance

This study proposes using embeddings from the Earth observation foundation model TESSERA as local surface descriptors to augment a convolutional conditional neural process for probabilistic downscaling of coarse ERA5 reanalysis fields at ~25 km resolution. Across five climatically diverse regions, the method significantly improves predictions of 2 m temperature and 10 m wind speed, with overall CRPS skill gains of 11.5% and 6.2%, respectively. Improvements persist when the coarse input is changed to forecasts from the Aurora AI 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.

## Key points

- TESSERA embeddings compress 10 m resolution surface information as sub-grid descriptors to enhance probabilistic downscaling models.
- CRPS skill improves by 11.5% for 2 m temperature and 6.2% for 10 m wind speed across five climatically diverse regions.
- Downscaling improvements remain robust when the coarse input is switched to Aurora AI forecasts or when predicting at newly deployed stations.
- Topography explains more of temperature's sub-grid structure, while TESSERA provides additional surface information for wind speed.

## Full text

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.

---

Verify against the original before citing; curated by Astral Feed: https://feed.astralor.com/items/itm_6e69ae71482ceb8438d83c3b