SimCast: Enhancing Precipitation Nowcasting with Short-to-Long Term Knowledge Distillation
This work addresses the need for accurate precipitation nowcasting for disaster management and other societal applications, offering an incremental improvement over existing methods.
The paper tackled the problem of precipitation nowcasting by proposing SimCast, a training pipeline with short-to-long term knowledge distillation and weighted MSE loss, which achieved mean CSI scores of 0.452 on SEVIR, 0.474 on HKO-7, and 0.361 on MeteoNet, outperforming existing approaches.
Precipitation nowcasting predicts future radar sequences based on current observations, which is a highly challenging task driven by the inherent complexity of the Earth system. Accurate nowcasting is of utmost importance for addressing various societal needs, including disaster management, agriculture, transportation, and energy optimization. As a complementary to existing non-autoregressive nowcasting approaches, we investigate the impact of prediction horizons on nowcasting models and propose SimCast, a novel training pipeline featuring a short-to-long term knowledge distillation technique coupled with a weighted MSE loss to prioritize heavy rainfall regions. Improved nowcasting predictions can be obtained without introducing additional overhead during inference. As SimCast generates deterministic predictions, we further integrate it into a diffusion-based framework named CasCast, leveraging the strengths from probabilistic models to overcome limitations such as blurriness and distribution shift in deterministic outputs. Extensive experimental results on three benchmark datasets validate the effectiveness of the proposed framework, achieving mean CSI scores of 0.452 on SEVIR, 0.474 on HKO-7, and 0.361 on MeteoNet, which outperforms existing approaches by a significant margin.