RainAI -- Precipitation Nowcasting from Satellite Data

TL;DR

RainAI introduces a novel spatiotemporal learning approach using a 2D U-Net for high-resolution precipitation nowcasting from satellite data, outperforming baseline models through dataset refinement, loss function improvements, and advanced upsampling techniques.

Contribution

The paper presents a simple yet effective 2D U-Net based method for precipitation forecasting, emphasizing dataset quality, loss function optimization, and probabilistic outputs, surpassing previous baselines.

Findings

Outperforms 3D U-Net baseline in accuracy and efficiency

Dataset refinement significantly improves model performance

Alternative loss functions enable probabilistic forecasting

Abstract

This paper presents a solution to the Weather4Cast 2023 competition, where the goal is to forecast high-resolution precipitation with an 8-hour lead time using lower-resolution satellite radiance images. We propose a simple, yet effective method for spatiotemporal feature learning using a 2D U-Net model, that outperforms the official 3D U-Net baseline in both performance and efficiency. We place emphasis on refining the dataset, through importance sampling and dataset preparation, and show that such techniques have a significant impact on performance. We further study an alternative cross-entropy loss function that improves performance over the standard mean squared error loss, while also enabling models to produce probabilistic outputs. Additional techniques are explored regarding the generation of predictions at different lead times, specifically through Conditioning Lead Time. Lastly, to generate high-resolution forecasts, we evaluate standard and learned upsampling methods. The code and trained parameters are available at https://github.com/rafapablos/w4c23-rainai.