STC-ViT: Spatio Temporal Continuous Vision Transformer for Weather Forecasting
This work addresses weather forecasting for operational systems by proposing an incremental hybrid method that combines transformers with continuous modeling and physics constraints.
The paper tackles the problem of weather forecasting by developing STC-ViT, a transformer model that incorporates continuous time layers and physics-informed loss to learn spatio-temporal features, achieving competitive performance with computational efficiency compared to state-of-the-art data-driven models.
Operational weather forecasting system relies on computationally expensive physics-based models. Recently, transformer based models have shown remarkable potential in weather forecasting achieving state-of-the-art results. However, transformers are discrete and physics-agnostic models which limit their ability to learn the continuous spatio-temporal features of the dynamical weather system. We address this issue with STC-ViT, a Spatio-Temporal Continuous Vision Transformer for weather forecasting. STC-ViT incorporates the continuous time Neural ODE layers with multi-head attention mechanism to learn the continuous weather evolution over time. The attention mechanism is encoded as a differentiable function in the transformer architecture to model the complex weather dynamics. Further, we define a customised physics informed loss for STC-ViT which penalize the model's predictions for deviating away from physical laws. We evaluate STC-ViT against operational Numerical Weather Prediction (NWP) model and several deep learning based weather forecasting models. STC-ViT, trained on 1.5-degree 6-hourly data, demonstrates computational efficiency and competitive performance compared to state-of-the-art data-driven models trained on higher-resolution data for global forecasting.