Enhancing Long-Term Reservoir Inflow Forecasting: An Integrated Approach Combining Switch Prediction Method, Ensemble Rainfall Forecasts, and Machine Learning Techniques

Abstract. This study makes a unique contribution by evaluating the effectiveness of the Switch Prediction Method (SPM) in integrating ensemble rainfall forecasts, significantly improving the accuracy of long-term inflow forecasting for reservoirs—a crucial aspect of hydrological forecasting. The proposed approach combines Numerical Weather Prediction (NWP) data with advanced machine learning techniques, specifically Support Vector Machine (SVM) and Long Short-Term Memory (LSTM) models, to develop a robust forecasting framework. The study utilizes comprehensive datasets from the ShihMen Reservoir in Taiwan to assess the performance of the proposed model. The SPM dynamically integrates multiple meteorological forecasts to reduce uncertainty and improve rainfall input accuracy. These enhanced inputs are then used in multi-step forecasting (MSF) models with a 72-hour lead time. The results demonstrate that the LSTM-based model, combined with SPM-integrated forecasts, delivers accurate and stable inflow predictions. For instance, in the case of Typhoon Soudelor—the test event with the highest observed peak inflow (5,634.1 cms)—the proposed SPM-LSTM-MSF model achieved a Mean Absolute Error (MAE) of 178.8 cms and a Coefficient of Efficiency (CE) of 0.87, demonstrating superior accuracy and temporal stability compared to the SVM-based approach. These findings highlight the potential of SPM and machine learning techniques in enhancing reservoir management and flood control strategies, offering a robust and adaptable solution for complex hydrological forecasting tasks.