An explainable Bayesian gated recurrent unit model for multi-step streamflow forecasting

Study region: In the middle and lower reaches of the Yangtze River Basin of ChinaStudy focus: We propose an explainable Bayesian gated recurrent unit (EB-GRU) model for reliable multi-step streamflow forecasting. The proposed model introduces Bayesian inference into a gated recurrent unit (GRU) to q...

Full description

Saved in:
Bibliographic Details
Main Authors: Lizhi Tao, Yueming Nan, Zhichao Cui, Lei Wang, Dong Yang
Format: Article
Language:English
Published: Elsevier 2025-02-01
Series:Journal of Hydrology: Regional Studies
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S2214581824004907
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1832591772836954112
author Lizhi Tao
Yueming Nan
Zhichao Cui
Lei Wang
Dong Yang
author_facet Lizhi Tao
Yueming Nan
Zhichao Cui
Lei Wang
Dong Yang
author_sort Lizhi Tao
collection DOAJ
description Study region: In the middle and lower reaches of the Yangtze River Basin of ChinaStudy focus: We propose an explainable Bayesian gated recurrent unit (EB-GRU) model for reliable multi-step streamflow forecasting. The proposed model introduces Bayesian inference into a gated recurrent unit (GRU) to quantify the uncertainty of streamflow prediction, and uses SHapley Additive exPlanations (SHAP) method to analyze the importance of hydrometeorological indices on streamflow prediction. The EB-GRU is examined by forecasting the multi-step streamflow at Hukou and Qilishan stations in the middle and lower reaches of the Yangtze River Basin, and compared with the Transformer (TSF), multi-layer perceptron (MLP) and support vector machine (SVM).New hydrological insights for the region: The comparative results show that the performance of the proposed EB-GRU surpasses that of the TSF, except for the streamflow forecast at the Hukou station with a 1-day lead time. The EB-GRU outperforms the MLP and SVM at each lead time, particularly at shorter lead times, highlighting its effectiveness in capturing short-term streamflow dynamics. The analysis of uncertainty quantization shows that noise in the input data is the primary source of overall uncertainty in model prediction, whereas a notable increase is observed in the uncertainty caused by the model in the flood season. Furthermore, the application of the SHAP method reveals the critical role of water level in streamflow prediction.
format Article
id doaj-art-6832999ab2944f61910db9b4800f8980
institution Kabale University
issn 2214-5818
language English
publishDate 2025-02-01
publisher Elsevier
record_format Article
series Journal of Hydrology: Regional Studies
spelling doaj-art-6832999ab2944f61910db9b4800f89802025-01-22T05:42:11ZengElsevierJournal of Hydrology: Regional Studies2214-58182025-02-0157102141An explainable Bayesian gated recurrent unit model for multi-step streamflow forecastingLizhi Tao0Yueming Nan1Zhichao Cui2Lei Wang3Dong Yang4Key Laboratory of Poyang Lake Wetland and Watershed Research of Ministry of Education & School of Geography and Environmental Science, Jiangxi Normal University, Nanchang 330022, ChinaKey Laboratory of Poyang Lake Wetland and Watershed Research of Ministry of Education & School of Geography and Environmental Science, Jiangxi Normal University, Nanchang 330022, ChinaKey Laboratory of Poyang Lake Wetland and Watershed Research of Ministry of Education & School of Geography and Environmental Science, Jiangxi Normal University, Nanchang 330022, ChinaBeijing Fengyun Meteorological Technology Development Company Limited, Beijing 100000, ChinaJiangxi Academy of Eco-Environmental Sciences & Planning, Nanchang 330029, China; Jiangxi Provincial Key Laboratory of Environmental Pollution Control, Nanchang 330029, China; Corresponding author at: Jiangxi Provincial Key Laboratory of Environmental Pollution Control, Nanchang 330029, China.Study region: In the middle and lower reaches of the Yangtze River Basin of ChinaStudy focus: We propose an explainable Bayesian gated recurrent unit (EB-GRU) model for reliable multi-step streamflow forecasting. The proposed model introduces Bayesian inference into a gated recurrent unit (GRU) to quantify the uncertainty of streamflow prediction, and uses SHapley Additive exPlanations (SHAP) method to analyze the importance of hydrometeorological indices on streamflow prediction. The EB-GRU is examined by forecasting the multi-step streamflow at Hukou and Qilishan stations in the middle and lower reaches of the Yangtze River Basin, and compared with the Transformer (TSF), multi-layer perceptron (MLP) and support vector machine (SVM).New hydrological insights for the region: The comparative results show that the performance of the proposed EB-GRU surpasses that of the TSF, except for the streamflow forecast at the Hukou station with a 1-day lead time. The EB-GRU outperforms the MLP and SVM at each lead time, particularly at shorter lead times, highlighting its effectiveness in capturing short-term streamflow dynamics. The analysis of uncertainty quantization shows that noise in the input data is the primary source of overall uncertainty in model prediction, whereas a notable increase is observed in the uncertainty caused by the model in the flood season. Furthermore, the application of the SHAP method reveals the critical role of water level in streamflow prediction.http://www.sciencedirect.com/science/article/pii/S2214581824004907Gated recurrent unitBayesian deep learningSHapley Additive exPlanationsMulti-step streamflow forecasting
spellingShingle Lizhi Tao
Yueming Nan
Zhichao Cui
Lei Wang
Dong Yang
An explainable Bayesian gated recurrent unit model for multi-step streamflow forecasting
Journal of Hydrology: Regional Studies
Gated recurrent unit
Bayesian deep learning
SHapley Additive exPlanations
Multi-step streamflow forecasting
title An explainable Bayesian gated recurrent unit model for multi-step streamflow forecasting
title_full An explainable Bayesian gated recurrent unit model for multi-step streamflow forecasting
title_fullStr An explainable Bayesian gated recurrent unit model for multi-step streamflow forecasting
title_full_unstemmed An explainable Bayesian gated recurrent unit model for multi-step streamflow forecasting
title_short An explainable Bayesian gated recurrent unit model for multi-step streamflow forecasting
title_sort explainable bayesian gated recurrent unit model for multi step streamflow forecasting
topic Gated recurrent unit
Bayesian deep learning
SHapley Additive exPlanations
Multi-step streamflow forecasting
url http://www.sciencedirect.com/science/article/pii/S2214581824004907
work_keys_str_mv AT lizhitao anexplainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT yuemingnan anexplainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT zhichaocui anexplainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT leiwang anexplainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT dongyang anexplainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT lizhitao explainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT yuemingnan explainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT zhichaocui explainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT leiwang explainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting
AT dongyang explainablebayesiangatedrecurrentunitmodelformultistepstreamflowforecasting