Rainfall Regression from C-Band Synthetic Aperture Radar Using Multitask Generative Adversarial Networks

Abstract This paper introduces a data-driven approach to estimate precipitation rates from synthetic aperture radar (SAR) at a spatial resolution of 200 m per pixel. It addresses previous challenges related to the collocation of SAR and weather radar data, specifically the misalignment in collocations and the scarcity of rainfall examples under strong wind. To tackle these challenges, the paper proposes a multiobjective formulation, introducing patch-level components and an adversarial component. It exploits the full NEXRAD archive to look for potential collocations with Sentinel-1 data. With additional enhancements to the training procedure and the incorporation of additional inputs, the resulting model demonstrates improved accuracy in rainfall estimates and the ability to extend its performance to scenarios up to 15 m s −1 . Significance Statement Identifying rainfall events in coastal areas is crucial for correcting meteorological models and preventing hazards. However, quantitatively measuring precipitation in open ocean and coastal areas is challenging due to the availability and range of ground-based weather radars. Satellite-based observations can provide rain information with extensive spatial coverage and high frequency but usually lack resolution. The present paper proposes to complement existing methods with synthetic aperture radar (SAR). In association with observations from weather radar, it outlines an architecture to derive a relationship between SAR information and the precipitation rate.