AdaWeather: Adaptively Mixing Probabilistic Weather Forecasts with Logarithmic Regret

2026-06-01

AI Summary

This work addresses the instability and lack of robustness in existing single-model weather forecasting approaches across spatiotemporal domains. To overcome these limitations, the authors propose AdaWeather, a novel framework that adaptively combines multiple probabilistic weather forecast models through online learning, dynamically generating a unified probabilistic prediction via a mixture-of-experts strategy. Notably, AdaWeather achieves, for the first time, a logarithmic regret bound relative to the best fixed mixture of experts, outperforming conventional methods that only compete against the single best expert. Experimental results on temperature forecasting demonstrate that AdaWeather significantly surpasses current state-of-the-art approaches, confirming its effectiveness and robustness in real-world forecasting scenarios.

Abstract

Recent advances in machine learning have produced probabilistic weather forecasting models comparable to state-of-the-art numerical weather predictors. But no model consistently dominates spatio-temporally, and relative performance is highly context-dependent. This motivates adaptive methods for combining multiple forecasts to obtain improvements and robustness. While combined forecasts have been proposed in the literature, these are achieved either through supervised learning or through prediction with expert advice methods. We introduce AdaWeather, an adaptive framework that combines many probabilistic forecasts using both machine learning as well as mixture of experts to arrive at a unified improved probabilistic forecast. While traditional expert methods develop the regret bounds with respect to the best single expert in hindsight, we extend the algorithm and analysis to show our method has logarithmic regret compared to the best static mixture of experts in hindsight. Empirically, we focus on forecasting temperature, and observe improvements over existing methods.

Problem

• probabilistic weather forecasting
• forecast combination
• adaptive methods
• spatio-temporal variability
• model robustness

Innovation

• adaptive forecasting
• probabilistic weather prediction
• mixture of experts
• logarithmic regret
• forecast combination