Scholarlink

<p>Reliable medium- to long-term runoff forecasting is of critical importance for water resource management and regional planning. However, the high nonlinearity and non-stationarity of hydrological processes continue to pose significant challenges. To address this issue, this study proposes an integrated forecasting framework termed SVPsEC, which embeds multi-module collaboration and cross-subsequence spatiotemporal coupling within a unified forecasting system. In this framework, Variational Mode Decomposition (VMD) decomposes the original runoff series into intrinsic mode functions with reduced complexity, effectively preserving multi-scale hydrological dynamic characteristics. Parallel CNN–GRU modules collaboratively extract local spatial features and long-term temporal dependencies of each component, achieving early fusion of heterogeneous spatiotemporal information. The Sparrow Search Algorithm (SSA) is employed to optimize the hyperparameters of the decomposition, feature extraction, and forecasting modules, further enhancing the collaborative capability of the framework. Additionally, an error correction strategy is introduced to progressively correct residuals and improve forecasting stability. Evaluations at three hydrological stations demonstrate that SVPsEC consistently produces highly accurate and stable forecasts, with NSC values all exceeding 0.98, R above 0.99, and significantly reduced RMSE compared to benchmark methods. Peak flow forecasting performance is also notably improved. Overall, the decomposition-based multi-module collaboration strategy, combined with coupled spatiotemporal representation, effectively enhances runoff forecasting capability and provides a reliable tool for watershed-scale water resource planning and hydrological risk management. </p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Hybrid Multi-Strategy Monthly Runoff Forecasting Model Based on Parallel CNN-GRU Architecture, SSA Optimization, and Error Correction Mechanisms

  • 期刊论文

  • 发表日期：

    2026年1月10日

查看全文
  • Liyuan Wang,
  • Xuebin Wang,
  • Jianxia Chang,
  • Xuejiao Meng,
  • Yimin Wang,
  • Chengqing Ren,
  • Junhao Zhang

摘要

Reliable medium- to long-term runoff forecasting is of critical importance for water resource management and regional planning. However, the high nonlinearity and non-stationarity of hydrological processes continue to pose significant challenges. To address this issue, this study proposes an integrated forecasting framework termed SVPsEC, which embeds multi-module collaboration and cross-subsequence spatiotemporal coupling within a unified forecasting system. In this framework, Variational Mode Decomposition (VMD) decomposes the original runoff series into intrinsic mode functions with reduced complexity, effectively preserving multi-scale hydrological dynamic characteristics. Parallel CNN–GRU modules collaboratively extract local spatial features and long-term temporal dependencies of each component, achieving early fusion of heterogeneous spatiotemporal information. The Sparrow Search Algorithm (SSA) is employed to optimize the hyperparameters of the decomposition, feature extraction, and forecasting modules, further enhancing the collaborative capability of the framework. Additionally, an error correction strategy is introduced to progressively correct residuals and improve forecasting stability. Evaluations at three hydrological stations demonstrate that SVPsEC consistently produces highly accurate and stable forecasts, with NSC values all exceeding 0.98, R above 0.99, and significantly reduced RMSE compared to benchmark methods. Peak flow forecasting performance is also notably improved. Overall, the decomposition-based multi-module collaboration strategy, combined with coupled spatiotemporal representation, effectively enhances runoff forecasting capability and provides a reliable tool for watershed-scale water resource planning and hydrological risk management.