Runoff forecasting is crucial for water resources management, flood risk assessment, and ecological protection, as accurate basin runoff prediction can provide a scientific basis for the rational allocation of water resources and advance flood prevention. This study aims to improve the accuracy of monthly runoff prediction by proposing a novel method based on the combination of variational modal decomposition (VMD) and adaptive particle swarm optimization bi-directional short-term and long-term memory network (APSO-BiLSTM), addressing the limitations of traditional models. By analyzing data from the Beibei Hydrological Station of the Jialing River, this study compared the performance of four models: VMD-APSO-BiLSTM, VMD-APSO-LSTM, APSO-BiLSTM, and APSO-LSTM. The results demonstrate that the VMD-APSO-BiLSTM model outperforms the other models during both the training and testing periods, achieving Nash-Sutcliffe efficiencies (NSEs) of 0.95 and 0.93, and root mean square errors (RMSEs) of 516.75 and 992.15, respectively. These findings indicate that the proposed model effectively captures the nonlinear characteristics and long-term dependencies of hydrological data, highlighting its potential for enhancing runoff forecasting accuracy.

Jianming Zhang, Shubo Zhang, Kelin Wang, Jiahao Bai, Ran Tao. Modeling of Monthly Runoff Prediction Based on VMD-APSO-BiLSTM Model. Academic Journal of Computing & Information Science (2025), Vol. 8, Issue 4: 86-91. https://doi.org/10.25236/AJCIS.2025.080410.

[1] Smith J D, Brown L. Hydrological modeling for water resource management: A review[J]. Journal of Hydrology, 2019, 563: 175-183.  

[2] Wang X, Chen Y. Challenges in streamflow forecasting: Data scarcity and model complexity[J]. Water Resources Research, 2020, 56, (7): e2019WR026123.  

[3] Zhang L, Liu Y. Performance evaluation of ARIMA and SARIMA models in the context of streamflow forecasting[J]. Water, 2018, 10, (11): 1634.  

[4] Feng Y, Liu D. A comparative study of machine learning and ARIMA in predicting river flow[J]. Journal of Hydrology, 2021, 601: 126693.

[5] Huang G, Li Q. Enhanced river flow forecasting using hybrid models: A critical review[J]. Water Research, 2022, 198: 117195.

[6] Xiong Y, Zhou J Z, Sun N, et al. Monthly runoff prediction based on self-adaptive variational mode decomposition and long short-term memory network[J]. Journal of Hydraulic Engineering, 2023, 54, (2): 172-183.  

[7] Wang X N, Chang J B, Jin H, et al. Research on annual runoff prediction model based on adaptive particle swarm optimization–long short-term memory with coupled variational mode decomposition and spectral clustering reconstruction[J]. Water, 2024, 16, (8): 1179.  

[8] Chen H, Lu T, Huang J, et al. An improved VMD-LSTM model for time-varying GNSS time series prediction with temporally correlated noise[J]. Remote Sensing, 2023, 15, (14): 3694.  

[9] Anonymous. New findings from Xi'an University of Technology describe advances in information technology (A comparative study of VMD-based hybrid forecasting model for nonstationary daily streamflow time series)[J]. Information Technology Newsweekly, 2020, (341).  

[10] Zhang X , Qiao W , Huang J ,et al.Flow prediction in the lower Yellow River based on CEEMDAN-BiLSTM coupled model[J].Water science & technology: Water supply, 2023(1/2):23.DOI:10.2166/ws.2022.426. 

[11] Chen K, Keqiao C. APSO-LSTM: An improved LSTM neural network model based on APSO algorithm[J]. Journal of Physics: Conference Series, 2020, 1651, (1): 012151.

[12] Guo Wenxian, Jiao Xuyang, Zhou Haotong, et al. Changes in the Hydrological Situation of the Jialing River and Their Impacts on Fish [J]. Resources and Environment in the Yangtze River Basin, 2022, 31(04): 805 - 813.