Generative Modeling of High-resolution Global Precipitation Forecasts

James Duncan, Shashank Subramanian, Peter Harrington

arXiv preprint · 22 October 2022 · University of California, Berkeley, Lawrence Berkeley National Laboratory / NERSC

In one sentence. A GAN-enhanced FourCastNet better captures fine-scale structure and extreme global precipitation percentiles, matching NWP skill at 1-2 day leads.

Abstract

Forecasting global precipitation patterns and, in particular, extreme precipitation events is of critical importance to preparing for and adapting to climate change. Making accurate high-resolution precipitation forecasts using traditional physical models remains a major challenge in operational weather forecasting as they incur substantial computational costs and struggle to achieve sufficient forecast skill. Recently, deep-learning-based models have shown great promise in closing the gap with numerical weather prediction (NWP) models in terms of precipitation forecast skill, opening up exciting new avenues for precipitation modeling. However, it is challenging for these deep learning models to fully resolve the fine-scale structures of precipitation phenomena and adequately characterize the extremes of the long-tailed precipitation distribution. In this work, we present several improvements to the architecture and training process of a current state-of-the art deep learning precipitation model (FourCastNet) using a novel generative adversarial network (GAN) to better capture fine scales and extremes. Our improvements achieve superior performance in capturing the extreme percentiles of global precipitation, while comparable to state-of-the-art NWP models in terms of forecast skill at 1--2 day lead times. Together, these improvements set a new state-of-the-art in global precipitation forecasting.

Labels

Labels and summary are generated by Claude and reviewed weekly. Wrong? Edit this entry.