analiza aiqbench
The Observation Assimilation Transformer (OAT) represents a significant advancement in applying geometry-aware neural processes to refine numerical weather prediction fields using sparse station data. By leveraging relative positions rather than absolute coordinates, this encoder-decoder architecture achieves scalable downscaling and calibrated probabilistic outputs that outperform traditional baselines like COSMO. This work highlights the growing utility of transformer-based machine learning models for solving complex geospatial inference problems in climate science.
Observation Assimilation Transformer: Probabilistic Weather Analysis from Sparse Station Observations
arXiv:2610.28327v2 Announce Type: replace Abstract: We present the Observation Assimilation Transformer (OAT), a geometry-aware neural process that corrects and refines a prior numerical weather prediction (NWP) field into a high-resolution posterior field by assimilating sparse observations. OAT is an encoder-decoder transformer that considers a set of observed locations and their NWP prior to condition predictions at a set of arbitrary unobserved query locations. Its sparse attention mechanism makes it scalable to thousands of station observations and depends on relative positions between locations, never on the absolute coordinates. The model allows for partially missing observations, predictions at arbitrary continuous locations without grid projection, and outputs are parametric distributions. A station-randomized training algorithm is developed with continuous ranked probability score (CRPS) loss. We test OAT on two datasets - one at regional scale with hundreds of stations, and one at continental scale with thousands of stations. On these datasets OAT outperforms several baselines and shows well calibrated predictions. In an illustrative case study, OAT's downscaling capabilities are demonstrated by downscaling ERA5 (~31km) to COSMO (1km), where OAT achieves lower mean absolute error (MAE) and CRPS for temperature and precipitation on test stations compared to COSMO.
Abstract:We present the Observation Assimilation Transformer (OAT), a geometry-aware neural process that corrects and refines a prior numerical weather prediction (NWP) field into a high-resolution posterior field by assimilating sparse observations. OAT is an encoder-decoder transformer that considers a set of observed locations and their NWP prior to condition predictions at a set of arbitrary unobserved query locations. Its sparse attention mechanism makes it scalable to thousands of station observations and depends on relative positions between locations, never on the absolute coordinates. The model allows for partially missing observations, predictions at arbitrary continuous locations without grid projection, and outputs are parametric distributions. A station-randomized training algorithm is developed with continuous ranked probability score (CRPS) loss. We test OAT on two datasets - one at regional scale with hundreds of stations, and one at continental scale with thousands of stations. On these datasets OAT outperforms several baselines and shows well calibrated predictions. In an illustrative case study, OAT's downscaling capabilities are demonstrated by downscaling ERA5 (~31km) to COSMO (1km), where OAT achieves lower mean absolute error (MAE) and CRPS for temperature and precipitation on test stations compared to COSMO.
Machine Learning (cs.LG)
arXiv:2610.28327 [cs.LG]
(or arXiv:2610.28327v2 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2610.28327 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 280 cuvinte
Subiecte conexe
Efficient Real-Time Adaptation of ROMs for Unsteady Flows Using Data Assimilation
arXiv · 1 sept. 2026
Uncertainty-Aware End-to-End AI Weather Forecasting: Disentangling Observation and Model Contributions
arXiv · 1 sept. 2026
Improving Ensemble Filters with Flow Matching
arXiv · 24 sept. 2026
Learning Probabilistic Filters with Strictly Proper Scoring Rules
arXiv · 7 oct. 2026