Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models

Accurate streamflow prediction is essential for water resource management and ecological conservation. With climate change and human activities intensifying extreme weather events, the risks associated with floods have grown, threatening both socioeconomic robustness and ecological integrity. Conven...

Full description

Saved in:
Bibliographic Details
Main Authors: Ningchang Kang, Zhaocai Wang, Anbin Zhang, Hang Chen
Format: Article
Language:English
Published: Elsevier 2025-12-01
Series:Ecological Informatics
Subjects:
Online Access:http://www.sciencedirect.com/science/article/pii/S1574954125003000
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849233598694031360
author Ningchang Kang
Zhaocai Wang
Anbin Zhang
Hang Chen
author_facet Ningchang Kang
Zhaocai Wang
Anbin Zhang
Hang Chen
author_sort Ningchang Kang
collection DOAJ
description Accurate streamflow prediction is essential for water resource management and ecological conservation. With climate change and human activities intensifying extreme weather events, the risks associated with floods have grown, threatening both socioeconomic robustness and ecological integrity. Conventional prediction methods, such as physical and statistical models, often struggle to capture the complex nonlinear and nonstationary characteristics of streamflow. To address this challenge, this study presents a vectorized hybrid STL-LSTM-GRU-Transformer model designed to enhance prediction accuracy and stability. The approach begins by applying seasonal-trend decomposition using loess (STL) to separate streamflow data into trend, seasonal, and residual components. These components are then modeled independently: long short-term memory (LSTM) and convolutional neural networks (CNN) handle trend and seasonal patterns, while gated recurrent units (GRU) and Transformer process residual fluctuations. Furthermore, the model incorporates the Runoff process vectorization (RPV) method alongside vectorization techniques to improve sensitivity to extreme events. Evaluated on 2010–2022 data from six Jialing River stations, the model achieves 0.9991 (NSE), outperforming 12 benchmarks. SHAP analysis identifies dew point temperature (26.7 % contribution) and solar radiation (15.7 %) as key drivers, while kernel density estimation provides reliable probabilistic forecasts (PICP = 0.90 at 95 % CI). Demonstrating robust performance in flood-drought transition prediction (NSE > 0.9983), this approach contributes valuable insights for advancing flood early warning systems and hydro-ecological security.
format Article
id doaj-art-4e0603035ee8469dbe1d595b1b643604
institution Kabale University
issn 1574-9541
language English
publishDate 2025-12-01
publisher Elsevier
record_format Article
series Ecological Informatics
spelling doaj-art-4e0603035ee8469dbe1d595b1b6436042025-08-20T05:05:35ZengElsevierEcological Informatics1574-95412025-12-019010329110.1016/j.ecoinf.2025.103291Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning modelsNingchang Kang0Zhaocai Wang1Anbin Zhang2Hang Chen3College of Information, Shanghai Ocean University, Shanghai 201306, PR ChinaCorresponding author.; College of Information, Shanghai Ocean University, Shanghai 201306, PR ChinaCollege of Information, Shanghai Ocean University, Shanghai 201306, PR ChinaCollege of Information, Shanghai Ocean University, Shanghai 201306, PR ChinaAccurate streamflow prediction is essential for water resource management and ecological conservation. With climate change and human activities intensifying extreme weather events, the risks associated with floods have grown, threatening both socioeconomic robustness and ecological integrity. Conventional prediction methods, such as physical and statistical models, often struggle to capture the complex nonlinear and nonstationary characteristics of streamflow. To address this challenge, this study presents a vectorized hybrid STL-LSTM-GRU-Transformer model designed to enhance prediction accuracy and stability. The approach begins by applying seasonal-trend decomposition using loess (STL) to separate streamflow data into trend, seasonal, and residual components. These components are then modeled independently: long short-term memory (LSTM) and convolutional neural networks (CNN) handle trend and seasonal patterns, while gated recurrent units (GRU) and Transformer process residual fluctuations. Furthermore, the model incorporates the Runoff process vectorization (RPV) method alongside vectorization techniques to improve sensitivity to extreme events. Evaluated on 2010–2022 data from six Jialing River stations, the model achieves 0.9991 (NSE), outperforming 12 benchmarks. SHAP analysis identifies dew point temperature (26.7 % contribution) and solar radiation (15.7 %) as key drivers, while kernel density estimation provides reliable probabilistic forecasts (PICP = 0.90 at 95 % CI). Demonstrating robust performance in flood-drought transition prediction (NSE > 0.9983), this approach contributes valuable insights for advancing flood early warning systems and hydro-ecological security.http://www.sciencedirect.com/science/article/pii/S1574954125003000Streamflow predictionDeep learningFeature extractionMode decompositionRunoff process vectorization
spellingShingle Ningchang Kang
Zhaocai Wang
Anbin Zhang
Hang Chen
Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models
Ecological Informatics
Streamflow prediction
Deep learning
Feature extraction
Mode decomposition
Runoff process vectorization
title Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models
title_full Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models
title_fullStr Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models
title_full_unstemmed Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models
title_short Improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models
title_sort improving the prediction of streamflow in large watersheds based on seasonal trend decomposition and vectorized deep learning models
topic Streamflow prediction
Deep learning
Feature extraction
Mode decomposition
Runoff process vectorization
url http://www.sciencedirect.com/science/article/pii/S1574954125003000
work_keys_str_mv AT ningchangkang improvingthepredictionofstreamflowinlargewatershedsbasedonseasonaltrenddecompositionandvectorizeddeeplearningmodels
AT zhaocaiwang improvingthepredictionofstreamflowinlargewatershedsbasedonseasonaltrenddecompositionandvectorizeddeeplearningmodels
AT anbinzhang improvingthepredictionofstreamflowinlargewatershedsbasedonseasonaltrenddecompositionandvectorizeddeeplearningmodels
AT hangchen improvingthepredictionofstreamflowinlargewatershedsbasedonseasonaltrenddecompositionandvectorizeddeeplearningmodels