A comparative study of stochastic and deep generative models for multisite precipitation synthesis

Jorge Guevara, Dario Borges, Campbell Watson, Bianca Zadrozny

arXiv preprint · 16 July 2021

In one sentence. Compares GAN and VAE deep generative models against classical weather generators for multisite precipitation synthesis, offering guidance for future climate scenario generation.

Abstract

Future climate change scenarios are usually hypothesized using simulations from weather generators. However, there only a few works comparing and evaluating promising deep learning models for weather generation against classical approaches. This study shows preliminary results making such evaluations for the multisite precipitation synthesis task. We compared two open-source weather generators: IBMWeathergen (an extension of the Weathergen library) and RGeneratePrec, and two deep generative models: GAN and VAE, on a variety of metrics. Our preliminary results can serve as a guide for improving the design of deep learning architectures and algorithms for the multisite precipitation synthesis task.

Labels

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