Optimized Atmospheric Forecasting Model: Integrating Informer-GRU on IGRA Data

With the advancement of artificial intelligence, atmospheric numerical prediction is increasingly shifting from traditional physical models to deep learning models to address complex data dependencies.This study introduces a novel prediction method that integrates an Informer-GRU model.The model leverages long-distance dependencies in time series through a multi-head attention mechanism and positional encoding to enhance prediction accuracy.Experimental results demonstrate that our model significantly outperforms traditional models in predicting multi-dimensional atmospheric data, achieving approximately 10% reduction in mean squared error.This study not only demonstrates a robust integration of Informer-GRU with Bayesian optimization but also sets a new benchmark for atmospheric numerical prediction, potentially enhancing decision-making in weathersensitive sectors.