Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River Basin

The sharp decline in streamflow prediction accuracy with increasing lead times remains a persistent challenge for effective water resources management and flood mitigation. In this study, we developed a coupled deep learning model for daily streamflow prediction in the Hanjiang River Basin, China. T...

Full description

Saved in:
Bibliographic Details
Main Authors: Jianze Huang, Jialang Chen, Haijun Huang, Xitian Cai
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Hydrology
Subjects:
Online Access:https://www.mdpi.com/2306-5338/12/7/168
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849733770056302592
author Jianze Huang
Jialang Chen
Haijun Huang
Xitian Cai
author_facet Jianze Huang
Jialang Chen
Haijun Huang
Xitian Cai
author_sort Jianze Huang
collection DOAJ
description The sharp decline in streamflow prediction accuracy with increasing lead times remains a persistent challenge for effective water resources management and flood mitigation. In this study, we developed a coupled deep learning model for daily streamflow prediction in the Hanjiang River Basin, China. The proposed model integrates self-attention (SA), a one-dimensional convolutional neural network (1D-CNN), and bidirectional long short-term memory (BiLSTM). The model’s effectiveness was assessed during flood events, and its predictive uncertainty was quantified using kernel density estimation (KDE). The results demonstrate that the proposed model consistently outperforms baseline models across all lead times. It achieved Nash-Sutcliffe Efficiency (<i>NSE</i>) scores of 0.92, 0.86, and 0.79 for 1-, 3-, and 5-days, respectively, showing particular strength at these extended lead time predictions. During major flood events, the model demonstrated an enhanced capacity to capture peak magnitudes and timings. It achieved the highest NSE values of 0.924, 0.862, and 0.797 for the 1-, 3-, and 5-day forecasting horizons, respectively, thereby showcasing the strengths of integrating CNN and SA mechanisms for recognizing local hydrological patterns. Furthermore, KDE-based uncertainty analysis identified a high prediction interval coverage in different forecast periods and a relatively narrow prediction interval width, indicating the strong robustness of the proposed model. Overall, the proposed SA-CNN-BiLSTM model demonstrates significantly improved accuracy, especially for extended lead times and flood events, and provides robust uncertainty quantification, thereby offering a more reliable tool for reservoir operation and flood risk management.
format Article
id doaj-art-4733aa77a2274cb1921e596b7d64bfc5
institution DOAJ
issn 2306-5338
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Hydrology
spelling doaj-art-4733aa77a2274cb1921e596b7d64bfc52025-08-20T03:07:58ZengMDPI AGHydrology2306-53382025-06-0112716810.3390/hydrology12070168Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River BasinJianze Huang0Jialang Chen1Haijun Huang2Xitian Cai3School of Civil Engineering, Sun Yat-sen University, Guangzhou 510275, ChinaSchool of Civil Engineering, Sun Yat-sen University, Guangzhou 510275, ChinaSchool of Civil Engineering, Sun Yat-sen University, Guangzhou 510275, ChinaSchool of Civil Engineering, Sun Yat-sen University, Guangzhou 510275, ChinaThe sharp decline in streamflow prediction accuracy with increasing lead times remains a persistent challenge for effective water resources management and flood mitigation. In this study, we developed a coupled deep learning model for daily streamflow prediction in the Hanjiang River Basin, China. The proposed model integrates self-attention (SA), a one-dimensional convolutional neural network (1D-CNN), and bidirectional long short-term memory (BiLSTM). The model’s effectiveness was assessed during flood events, and its predictive uncertainty was quantified using kernel density estimation (KDE). The results demonstrate that the proposed model consistently outperforms baseline models across all lead times. It achieved Nash-Sutcliffe Efficiency (<i>NSE</i>) scores of 0.92, 0.86, and 0.79 for 1-, 3-, and 5-days, respectively, showing particular strength at these extended lead time predictions. During major flood events, the model demonstrated an enhanced capacity to capture peak magnitudes and timings. It achieved the highest NSE values of 0.924, 0.862, and 0.797 for the 1-, 3-, and 5-day forecasting horizons, respectively, thereby showcasing the strengths of integrating CNN and SA mechanisms for recognizing local hydrological patterns. Furthermore, KDE-based uncertainty analysis identified a high prediction interval coverage in different forecast periods and a relatively narrow prediction interval width, indicating the strong robustness of the proposed model. Overall, the proposed SA-CNN-BiLSTM model demonstrates significantly improved accuracy, especially for extended lead times and flood events, and provides robust uncertainty quantification, thereby offering a more reliable tool for reservoir operation and flood risk management.https://www.mdpi.com/2306-5338/12/7/168streamflow predictiondeep learninguncertainty analysis
spellingShingle Jianze Huang
Jialang Chen
Haijun Huang
Xitian Cai
Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River Basin
Hydrology
streamflow prediction
deep learning
uncertainty analysis
title Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River Basin
title_full Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River Basin
title_fullStr Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River Basin
title_full_unstemmed Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River Basin
title_short Deep Learning-Based Daily Streamflow Prediction Model for the Hanjiang River Basin
title_sort deep learning based daily streamflow prediction model for the hanjiang river basin
topic streamflow prediction
deep learning
uncertainty analysis
url https://www.mdpi.com/2306-5338/12/7/168
work_keys_str_mv AT jianzehuang deeplearningbaseddailystreamflowpredictionmodelforthehanjiangriverbasin
AT jialangchen deeplearningbaseddailystreamflowpredictionmodelforthehanjiangriverbasin
AT haijunhuang deeplearningbaseddailystreamflowpredictionmodelforthehanjiangriverbasin
AT xitiancai deeplearningbaseddailystreamflowpredictionmodelforthehanjiangriverbasin