Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff models

The primary merit of data-driven-based runoff models is their capability to handle various inputs, including hydrological, land use, and geographical data, allowing for flexibility regarding different environmental conditions and landscapes. Physics-based models provide a comprehensive framework for...

Full description

Saved in:
Bibliographic Details
Main Authors: Pin-Chun Huang, Kwan Tun Lee
Format: Article
Language:English
Published: Taylor & Francis Group 2025-12-01
Series:Geomatics, Natural Hazards & Risk
Subjects:
Online Access:https://www.tandfonline.com/doi/10.1080/19475705.2025.2516725
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849425748744470528
author Pin-Chun Huang
Kwan Tun Lee
author_facet Pin-Chun Huang
Kwan Tun Lee
author_sort Pin-Chun Huang
collection DOAJ
description The primary merit of data-driven-based runoff models is their capability to handle various inputs, including hydrological, land use, and geographical data, allowing for flexibility regarding different environmental conditions and landscapes. Physics-based models provide a comprehensive framework for understanding runoff processes, offering physical realism and transferability advantages. In contrast, they may require more expertise and complicated numerical operations compared to data-driven models. The present study aims to improve the predictive capability of data-driven models by including the advantages of physics-based models in the model’s structure and preprocessing input features. To achieve this goal, associated environmental factors adopted in theoretical models, having more rigorous physical interpretation for runoff predictions, are thoroughly examined, especially for the features associated with topographic descriptors. The topological distribution inherent in the input data space is analyzed to improve predictive accuracy. The proposed artificial intelligence (AI) model, which incorporates a classification algorithm for preprocessing input features prior to training a model based on the recurrent neural network, exhibits outstanding performance in runoff discharge prediction. The main contribution of this study is to establish a robust runoff model that retains the original superiority of the data-driven model while extending its capability to capture hydrological processes and underlying physical influences in predicting hydrological responses from river basins.
format Article
id doaj-art-9fbc1f6da3a04466863d37456c75a3e6
institution Kabale University
issn 1947-5705
1947-5713
language English
publishDate 2025-12-01
publisher Taylor & Francis Group
record_format Article
series Geomatics, Natural Hazards & Risk
spelling doaj-art-9fbc1f6da3a04466863d37456c75a3e62025-08-20T03:29:39ZengTaylor & Francis GroupGeomatics, Natural Hazards & Risk1947-57051947-57132025-12-0116110.1080/19475705.2025.2516725Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff modelsPin-Chun Huang0Kwan Tun Lee1Department of Harbor and River Engineering, National Taiwan Ocean University, Keelung, TaiwanDepartment of Harbor and River Engineering, National Taiwan Ocean University, Keelung, TaiwanThe primary merit of data-driven-based runoff models is their capability to handle various inputs, including hydrological, land use, and geographical data, allowing for flexibility regarding different environmental conditions and landscapes. Physics-based models provide a comprehensive framework for understanding runoff processes, offering physical realism and transferability advantages. In contrast, they may require more expertise and complicated numerical operations compared to data-driven models. The present study aims to improve the predictive capability of data-driven models by including the advantages of physics-based models in the model’s structure and preprocessing input features. To achieve this goal, associated environmental factors adopted in theoretical models, having more rigorous physical interpretation for runoff predictions, are thoroughly examined, especially for the features associated with topographic descriptors. The topological distribution inherent in the input data space is analyzed to improve predictive accuracy. The proposed artificial intelligence (AI) model, which incorporates a classification algorithm for preprocessing input features prior to training a model based on the recurrent neural network, exhibits outstanding performance in runoff discharge prediction. The main contribution of this study is to establish a robust runoff model that retains the original superiority of the data-driven model while extending its capability to capture hydrological processes and underlying physical influences in predicting hydrological responses from river basins.https://www.tandfonline.com/doi/10.1080/19475705.2025.2516725Rainfall-runoff modelgeomorphological factorsdata-driven modelmachine learninghydrological process
spellingShingle Pin-Chun Huang
Kwan Tun Lee
Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff models
Geomatics, Natural Hazards & Risk
Rainfall-runoff model
geomorphological factors
data-driven model
machine learning
hydrological process
title Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff models
title_full Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff models
title_fullStr Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff models
title_full_unstemmed Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff models
title_short Developing an alternative data-driven model to resemble geomorphologic rainfall-runoff models
title_sort developing an alternative data driven model to resemble geomorphologic rainfall runoff models
topic Rainfall-runoff model
geomorphological factors
data-driven model
machine learning
hydrological process
url https://www.tandfonline.com/doi/10.1080/19475705.2025.2516725
work_keys_str_mv AT pinchunhuang developinganalternativedatadrivenmodeltoresemblegeomorphologicrainfallrunoffmodels
AT kwantunlee developinganalternativedatadrivenmodeltoresemblegeomorphologicrainfallrunoffmodels