PSRGAN: Generative Adversarial Networks for Precipitation Downscaling

Zhuang Li, Zhenyu Lu, Yuhao Zhang, Yizhe Li

Abstract: This study proposes an innovative generative adversarial network (GAN)-based downscaling model for precipitation, named PSRGAN, which aims to enhance the spatial resolution of meteorological data using deep learning techniques. The PSRGAN model integrates a multiscale feature fusion module (Rception), a kernel attention module (KAM), and the generator–discriminator framework of GANs to address challenges such as data sparsity and spatiotemporal correlations that traditional precipitation super-resolution (SR) methods struggle with. By extracting multiscale spatial features, PSRGAN improves the model’s ability to detect key precipitation regions and enhances the accuracy of predicting extreme precipitation events. The model is trained and tested using low-resolution (LR) and high-resolution (HR) simulated datasets based on regional climate models (RCMs), with performance evaluated through various metrics. The experimental results demonstrate that PSRGAN achieves strong performance in the precipitation downscaling task.