Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation

Precipitation is a key driving factor of hydrologic modeling for impact studies. However, there are challenges due to limited long-term data availability and complex parameterizations of existing stochastic weather generators (SWGs) due to spatiotemporal uncertainty. We introduced state-of-the-art G...

Full description

Bibliographic Details
Main Authors: Ji, Hong Kang, Mirzaei, Majid, Lai, Sai Hin, Dehghani, Adnan, Dehghani, Amin
Format: Article
Published: Elsevier Ltd 2024
Online Access:http://psasir.upm.edu.my/id/eprint/105818/
_version_ 1848864616044560384
author Ji, Hong Kang
Mirzaei, Majid
Lai, Sai Hin
Dehghani, Adnan
Dehghani, Amin
author_facet Ji, Hong Kang
Mirzaei, Majid
Lai, Sai Hin
Dehghani, Adnan
Dehghani, Amin
author_sort Ji, Hong Kang
building UPM Institutional Repository
collection Online Access
description Precipitation is a key driving factor of hydrologic modeling for impact studies. However, there are challenges due to limited long-term data availability and complex parameterizations of existing stochastic weather generators (SWGs) due to spatiotemporal uncertainty. We introduced state-of-the-art Generative Adversarial Network (GAN) as a data-driven multi-site SWG and synthesized extensive hourly precipitation over 30 years at 14 stations. These samples were then fed into an hourly-calibrated SWAT model for streamflow generation. Results showed that the well-trained GAN improved rainfall data by accurately representing spatiotemporal distribution of raw data rather than simply replicating its statistical characteristics. GAN also helped display authentic spatial correlation patterns of extreme rainfall events well. We concluded that GAN offers a superior spatiotemporal distribution of raw data compared to conventional methods, thus enhancing the reliability of flood frequency evaluations.
first_indexed 2025-11-15T13:51:38Z
format Article
id upm-105818
institution Universiti Putra Malaysia
institution_category Local University
last_indexed 2025-11-15T13:51:38Z
publishDate 2024
publisher Elsevier Ltd
recordtype eprints
repository_type Digital Repository
spelling upm-1058182024-03-29T03:28:42Z http://psasir.upm.edu.my/id/eprint/105818/ Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation Ji, Hong Kang Mirzaei, Majid Lai, Sai Hin Dehghani, Adnan Dehghani, Amin Precipitation is a key driving factor of hydrologic modeling for impact studies. However, there are challenges due to limited long-term data availability and complex parameterizations of existing stochastic weather generators (SWGs) due to spatiotemporal uncertainty. We introduced state-of-the-art Generative Adversarial Network (GAN) as a data-driven multi-site SWG and synthesized extensive hourly precipitation over 30 years at 14 stations. These samples were then fed into an hourly-calibrated SWAT model for streamflow generation. Results showed that the well-trained GAN improved rainfall data by accurately representing spatiotemporal distribution of raw data rather than simply replicating its statistical characteristics. GAN also helped display authentic spatial correlation patterns of extreme rainfall events well. We concluded that GAN offers a superior spatiotemporal distribution of raw data compared to conventional methods, thus enhancing the reliability of flood frequency evaluations. Elsevier Ltd 2024 Article PeerReviewed Ji, Hong Kang and Mirzaei, Majid and Lai, Sai Hin and Dehghani, Adnan and Dehghani, Amin (2024) Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation. Environmental Modelling and Software, 172. art. no. 105896. pp. 1-14. ISSN 1364-8152; ESSN: 1873-6726 10.1016/j.envsoft.2023.105896
spellingShingle Ji, Hong Kang
Mirzaei, Majid
Lai, Sai Hin
Dehghani, Adnan
Dehghani, Amin
Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation
title Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation
title_full Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation
title_fullStr Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation
title_full_unstemmed Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation
title_short Implementing generative adversarial network (GAN) as a data-driven multi-site stochastic weather generator for flood frequency estimation
title_sort implementing generative adversarial network (gan) as a data-driven multi-site stochastic weather generator for flood frequency estimation
url http://psasir.upm.edu.my/id/eprint/105818/
http://psasir.upm.edu.my/id/eprint/105818/