Technical Note: Temporal Disaggregation of Spatial Rainfall Fields with Generative Adversarial Networks

Abstract

Abstract. Creating spatially coherent rainfall patterns with high temporal resolution from data with lower temporal resolution is necessary in many geoscientific applications. From a statistical perspective, this presents a high- dimensional, highly under-determined problem. Recent advances in machine learning provide methods for learning such probability distributions. We test the usage of Generative Adversarial Networks (GANs) for estimating the full probability distribution of spatial rainfall patterns with high temporal resolution, conditioned on a field of lower temporal resolution. The GAN is trained on rainfall radar data with hourly resolution. Given a new field of daily precipitation sums, it can sample scenarios of spatiotemporal patterns with sub-daily resolution. While the generated patterns do not perfectly reproduce the statistics of observations, they are visually hardly distinguishable from real patterns. Limitations that we found are that providing additional input (such as geographical information) to the GAN surprisingly lead to worse results, showing that it is not trivial to increase the amount of used input information. Additionally, while in principle the GAN should learn the probability distribution in itself, we still needed expert judgment to determine at which point the training should stop, because longer training leads to worse results.

Adjusting spatial dependence of climate model outputs with cycle-consistent adversarial networks

“…Due to the stochastic and high-dimensionality nature of many physical processes of the Earth system, GANs and conditional GANs are particularly appealing for atmospheric science problems. Recently, they have been used for various Earth-science related applications: for instance for statistical downscaling Wang et al 2021), temporal disaggregation of spatial rainfall fields (Scher and Peßenteiner 2020), sampling of extreme values (Bhatia et al 2020), modelling of chaotic dynamical systems (e.g., Xie et al 2018;Wu et al 2020), classification of snowflake images , weather forecasting (Bihlo 2020) and stochastic parameterization in geophysical models (Gagne II et al 2020).…”

Stochastic Super-Resolution for Downscaling Time-Evolving Atmospheric Fields With a Generative Adversarial Network

“…Regardless, the uncertainty aspect has been largely ignored in earlier attempts at improving the resolution of climate fields using deep learning even when employing GANs for this problem [16] or for other super-resolution applications related to climate or remote sensing [17]- [19] although a few studies have used GANs to represent uncertainty in other atmospheric data problems [20], [21]. Moreover, while GANs have been recently also used to model the time evolution of atmospheric fields [22], few studies using deep learning have investigated modeling the uncertainty of the generated high-resolution image in a manner consistent with the time evolution of atmospheric fields-a problem analogous to video super-resolution, which has also been studied using GANs [23], [24].…”

Adjusting spatial dependence of climate model outputs with Cycle-Consistent Adversarial Networks

“…Due to the stochastic and high-dimensionality nature of many physical processes of the Earth system, GANs and conditional GANs are particularly appealing for atmospheric science problems. Recently, they have been used for various Earthscience related applications: for instance for statistical downscaling Wang et al 2021), temporal disaggregation of spatial rainfall fields (Scher and Peßenteiner 2020), sampling of extreme values (Bhatia et al 2020), modelling of chaotic dynamical systems (e.g., Xie et al 2018;Wu et al 2020), classification of snowflake images , weather forecasting (Bihlo 2020) and stochastic parameterization in geophysical models (Gagne II et al 2020).…”