High-Resolution Probabilistic Data-Driven Weather Modeling with a Stretched-Grid

2025-11-28

AI Summary

This work addresses the challenge of generating high-resolution, probabilistic, and spatially coherent multi-variable (87 variables) weather forecasts with arbitrary lead times and ensemble sizes. Methodologically, it employs a stretched-grid discretization to achieve regional adaptivity—resolving critical areas at 2.5 km while maintaining a global mean resolution of 31 km—and introduces a novel pointwise continuous ranked probability score (CRPS) loss that jointly operates in physical and spectral domains; notably, it is the first to incorporate spectral-domain CRPS to explicitly enforce spatial consistency. The model is trained using a stochastic encoder–decoder architecture. Experimental evaluation against the operational MetCoOp ensemble forecasting system demonstrates significant improvements in both spatial coherence and forecast accuracy on ground-based observations, particularly for meso- and microscale atmospheric structures.

Technology Category

• Machine Learning: Ensemble Methods
• Search and Optimization: Mixed Discrete/Continuous Search
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

We present a probabilistic data-driven weather model capable of providing an ensemble of high spatial resolution realizations of 87 variables at arbitrary forecast length and ensemble size. The model uses a stretched grid, dedicating 2.5 km resolution to a region of interest, and 31 km resolution elsewhere. Based on a stochastic encoder-decoder architecture, the model is trained using a loss function based on the Continuous Ranked Probability Score (CRPS) evaluated point-wise in real and spectral space. The spectral loss components is shown to be necessary to create fields that are spatially coherent. The model is compared to high-resolution operational numerical weather prediction forecasts from the MetCoOp Ensemble Prediction System (MEPS), showing competitive forecasts when evaluated against observations from surface weather stations. The model produced fields that are more spatially coherent than mean squared error based models and CRPS based models without the spectral component in the loss.

Problem

• Develops probabilistic weather model with high-resolution stretched grid
• Ensures spatial coherence using spectral loss in training
• Compares performance against operational numerical weather prediction systems

Innovation

• Stretched-grid with 2.5 km regional resolution
• Stochastic encoder-decoder architecture for ensembles
• Spectral-space CRPS loss ensuring spatial coherence