AI-based Multimodel Superensemble for Improved Weather Prediction

Recent advancements in AI-driven weather forecasting models provide faster predictions with lower computational costs compared to traditional Numerical Weather Prediction (NWP) models. However, like NWP models, data-driven models are also prone to persistent systematic biases. This study presents a multi-model ensemble technique designed to enhance forecast accuracy by leveraging the complementary strengths of state-of-the-art AI models combined using the Extreme Gradient Boosting algorithm. Our ensemble approach significantly reduced model biases, particularly improving forecast accuracy at longer lead times. These findings suggest that multi-model AI ensembles offer a promising, computationally efficient alternative to traditional methods for real-time weather forecasting, with potential applications in fields requiring high-precision weather predictions.