Design flood estimation for global river networks based on machine learning models

<p>Design flood estimation is a fundamental task in hydrology. In this research, we propose a machine-learning-based approach to estimate design floods globally. This approach involves three stages: (i) estimating at-site flood frequency curves for global gauging stations using the Anderson–Da...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	G. Zhao, P. Bates, J. Neal, B. Pang
Formato:	article
Lenguaje:	EN
Publicado:	Copernicus Publications 2021
Materias:	Technology T Environmental technology. Sanitary engineering TD1-1066 Geography. Anthropology. Recreation G Environmental sciences GE1-350
Acceso en línea:	https://doaj.org/article/cc686fbb87c74b4680dbf28a08067fb8
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:cc686fbb87c74b4680dbf28a08067fb8
record_format	dspace
spelling	oai:doaj.org-article:cc686fbb87c74b4680dbf28a08067fb82021-11-22T09:30:22ZDesign flood estimation for global river networks based on machine learning models10.5194/hess-25-5981-20211027-56061607-7938https://doaj.org/article/cc686fbb87c74b4680dbf28a08067fb82021-11-01T00:00:00Zhttps://hess.copernicus.org/articles/25/5981/2021/hess-25-5981-2021.pdfhttps://doaj.org/toc/1027-5606https://doaj.org/toc/1607-7938<p>Design flood estimation is a fundamental task in hydrology. In this research, we propose a machine-learning-based approach to estimate design floods globally. This approach involves three stages: (i) estimating at-site flood frequency curves for global gauging stations using the Anderson–Darling test and a Bayesian Markov chain Monte Carlo (MCMC) method; (ii) clustering these stations into subgroups using a <span class="inline-formula"><i>K</i></span>-means model based on 12 globally available catchment descriptors; and (iii) developing a regression model in each subgroup for regional design flood estimation using the same descriptors. A total of 11 793 stations globally were selected for model development, and three widely used regression models were compared for design flood estimation. The results showed that (1) the proposed approach achieved the highest accuracy for design flood estimation when using all 12 descriptors for clustering; and the performance of the regression was improved by considering more descriptors during training and validation; (2) a support vector machine regression provided the highest prediction performance amongst all regression models tested, with a root mean square normalised error of 0.708 for 100-year return period flood estimation; (3) 100-year design floods in tropical, arid, temperate, cold and polar climate zones could be reliably estimated (i.e. <span class="inline-formula"><i><</i>±</span>25 % error), with relative mean bias (RBIAS) values of <span class="inline-formula">−</span>0.199, <span class="inline-formula">−</span>0.233, <span class="inline-formula">−</span>0.169, 0.179 and <span class="inline-formula">−</span>0.091 respectively; (4) the machine-learning-based approach developed in this paper showed considerable improvement over the index-flood-based method introduced by Smith et al. (2015, <a href="https://doi.org/10.1002/2014WR015814">https://doi.org/10.1002/2014WR015814</a>) for design flood estimation at global scales; and (5) the average RBIAS in estimation is less than 18 % for 10-, 20-, 50- and 100-year design floods. We conclude that the proposed approach is a valid method to estimate design floods anywhere on the global river network, improving our prediction of the flood hazard, especially in ungauged areas.</p>G. ZhaoP. BatesP. BatesJ. NealJ. NealB. PangCopernicus PublicationsarticleTechnologyTEnvironmental technology. Sanitary engineeringTD1-1066Geography. Anthropology. RecreationGEnvironmental sciencesGE1-350ENHydrology and Earth System Sciences, Vol 25, Pp 5981-5999 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Technology T Environmental technology. Sanitary engineering TD1-1066 Geography. Anthropology. Recreation G Environmental sciences GE1-350
spellingShingle	Technology T Environmental technology. Sanitary engineering TD1-1066 Geography. Anthropology. Recreation G Environmental sciences GE1-350 G. Zhao P. Bates P. Bates J. Neal J. Neal B. Pang Design flood estimation for global river networks based on machine learning models
description	<p>Design flood estimation is a fundamental task in hydrology. In this research, we propose a machine-learning-based approach to estimate design floods globally. This approach involves three stages: (i) estimating at-site flood frequency curves for global gauging stations using the Anderson–Darling test and a Bayesian Markov chain Monte Carlo (MCMC) method; (ii) clustering these stations into subgroups using a <span class="inline-formula"><i>K</i></span>-means model based on 12 globally available catchment descriptors; and (iii) developing a regression model in each subgroup for regional design flood estimation using the same descriptors. A total of 11 793 stations globally were selected for model development, and three widely used regression models were compared for design flood estimation. The results showed that (1) the proposed approach achieved the highest accuracy for design flood estimation when using all 12 descriptors for clustering; and the performance of the regression was improved by considering more descriptors during training and validation; (2) a support vector machine regression provided the highest prediction performance amongst all regression models tested, with a root mean square normalised error of 0.708 for 100-year return period flood estimation; (3) 100-year design floods in tropical, arid, temperate, cold and polar climate zones could be reliably estimated (i.e. <span class="inline-formula"><i><</i>±</span>25 % error), with relative mean bias (RBIAS) values of <span class="inline-formula">−</span>0.199, <span class="inline-formula">−</span>0.233, <span class="inline-formula">−</span>0.169, 0.179 and <span class="inline-formula">−</span>0.091 respectively; (4) the machine-learning-based approach developed in this paper showed considerable improvement over the index-flood-based method introduced by Smith et al. (2015, <a href="https://doi.org/10.1002/2014WR015814">https://doi.org/10.1002/2014WR015814</a>) for design flood estimation at global scales; and (5) the average RBIAS in estimation is less than 18 % for 10-, 20-, 50- and 100-year design floods. We conclude that the proposed approach is a valid method to estimate design floods anywhere on the global river network, improving our prediction of the flood hazard, especially in ungauged areas.</p>
format	article
author	G. Zhao P. Bates P. Bates J. Neal J. Neal B. Pang
author_facet	G. Zhao P. Bates P. Bates J. Neal J. Neal B. Pang
author_sort	G. Zhao
title	Design flood estimation for global river networks based on machine learning models
title_short	Design flood estimation for global river networks based on machine learning models
title_full	Design flood estimation for global river networks based on machine learning models
title_fullStr	Design flood estimation for global river networks based on machine learning models
title_full_unstemmed	Design flood estimation for global river networks based on machine learning models
title_sort	design flood estimation for global river networks based on machine learning models
publisher	Copernicus Publications
publishDate	2021
url	https://doaj.org/article/cc686fbb87c74b4680dbf28a08067fb8
work_keys_str_mv	AT gzhao designfloodestimationforglobalrivernetworksbasedonmachinelearningmodels AT pbates designfloodestimationforglobalrivernetworksbasedonmachinelearningmodels AT pbates designfloodestimationforglobalrivernetworksbasedonmachinelearningmodels AT jneal designfloodestimationforglobalrivernetworksbasedonmachinelearningmodels AT jneal designfloodestimationforglobalrivernetworksbasedonmachinelearningmodels AT bpang designfloodestimationforglobalrivernetworksbasedonmachinelearningmodels
_version_	1718417821009969152

Design flood estimation for global river networks based on machine learning models

Ejemplares similares