Improving Medium Range Severe Weather Prediction through Transformer Post-processing of AI Weather Forecasts

2025-05-16

AI Summary

To address insufficient accuracy in medium-range (1–8 day) severe weather forecasting, this paper proposes a decoder-only Transformer-based AI post-processing method. It innovatively models forecast lead times as sequential tokens, explicitly learning the temporal evolution of atmospheric states. Taking Pangu-Weather forecasts as input, the approach achieves purely data-driven mesoscale severe weather correction without incorporating explicit convection parameterization schemes. Compared to conventional fully connected network post-processors, it significantly improves key skill scores—including the Threat Score (TS) and Brier Score (BS). When initialized from high-resolution analysis fields, the method achieves state-of-the-art accuracy and reliability for heavy precipitation and thunderstorm forecasting. This work establishes the efficacy of temporal tokenization for numerical weather prediction (NWP) post-processing and introduces a novel paradigm for AI-enhanced mesoscale forecasting.

Technology Category

• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Reasoning under Uncertainty: Relational Probabilistic Models

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Improving the skill of medium-range (1-8 day) severe weather prediction is crucial for mitigating societal impacts. This study introduces a novel approach leveraging decoder-only transformer networks to post-process AI-based weather forecasts, specifically from the Pangu-Weather model, for improved severe weather guidance. Unlike traditional post-processing methods that use a dense neural network to predict the probability of severe weather using discrete forecast samples, our method treats forecast lead times as sequential ``tokens'', enabling the transformer to learn complex temporal relationships within the evolving atmospheric state. We compare this approach against post-processing of the Global Forecast System (GFS) using both a traditional dense neural network and our transformer, as well as configurations that exclude convective parameters to fairly evaluate the impact of using the Pangu-Weather AI model. Results demonstrate that the transformer-based post-processing significantly enhances forecast skill compared to dense neural networks. Furthermore, AI-driven forecasts, particularly Pangu-Weather initialized from high resolution analysis, exhibit superior performance to GFS in the medium-range, even without explicit convective parameters. Our approach offers improved accuracy, and reliability, which also provides interpretability through feature attribution analysis, advancing medium-range severe weather prediction capabilities.

Problem

• Enhancing medium-range severe weather prediction accuracy
• Using transformer networks to post-process AI forecasts
• Comparing AI and traditional models for weather guidance

Innovation

• Transformer post-processes AI weather forecasts
• Treats forecast lead times as sequential tokens
• Enhances accuracy and reliability with interpretability