Generalizing Local weather Forecast to Large-grained Temporal Scales by Physics-AI Hybrid Modeling
Authors: Wanghan Xu, Fenghua Ling, Wenlong Zhang, Tao Han, Hao Chen, Wanli Ouyang, Lei Bai
Abstract: Data-driven artificial intelligence (AI) fashions have made vital developments in local weather forecasting, notably in medium-range and nowcasting. Nonetheless, most data-driven local weather forecasting fashions are black-box packages that focus on learning info mapping fairly than fine-grained bodily evolution inside the time dimension. Consequently, the restrictions inside the temporal scale of datasets cease these fashions from forecasting at finer time scales. This paper proposes a physics-AI hybrid model (i.e., WeatherGFT) which Generalizes local weather forecasts to Finer-grained Temporal scales previous teaching dataset. Significantly, we make use of a fastidiously designed PDE kernel to simulate bodily evolution on a small time scale (e.g., 300 seconds) and use a parallel neural networks with a learnable router for bias correction. Furthermore, we introduce a lead time-aware teaching framework to promote the generalization of the model at completely totally different lead events. The burden analysis of physics-AI modules signifies that physics conducts important evolution whereas AI performs corrections adaptively. Intensive experiments current that WeatherGFT expert on an hourly dataset, achieves state-of-the-art effectivity all through a variety of lead events and shows the potential to generalize 30-minute forecast