Author(s): Nguyen Duc Hanh; Le Huu Minh Quan; Nguyen Thi Phuong Anh; Nguyen Tien Giang; Dao Ba Huy; Nguyen Que Chi
Linked Author(s): Tien Giang Nguyen
Keywords: Reservoir inflow forecasting machine learning uncertainty analysis multi-step ahead forecasting
Abstract: The Da River Basin is a transboundary river basin, covering an area of 53,000 km2, in which, 49% of its upstream area locates in China and the remainder locates in Vietnam. Numerous large reservoirs have been constructed in this basin, mainly for electricity generation. In Vietnam, there are five major reservoirs namely Lai Chau, Hoa Binh, Son La, Ban Chat, and Hoi Quang. Among these, Lai Chau and Ban Chat are the two uppermost reservoirs located on the main upstream branches of this interconnected reservoir system. Accurate inflow forecasts for these two upstream reservoirs are crucial for efficiently operating the entire system. Given the limited availability of observational data from the upstream region, forecasting approaches based on traditional statistical methods and, more recently, machine learning techniques are deemed appropriate. This study focuses on developing and selecting forecasting models for inflow prediction to these two reservoirs. The models considered include the traditional statistical ARIMA model and several widely used machine learning algorithms namely Random Forest (RF), LightGBM (LGBM), Support Vector Regression (SVR), Artificial Neural Networks (ANN), and Long Short-Term Memory (LSTM). The performance of the models was assessed through standard evaluation metrics (NSE, RMSE, MAE, KGE, and Average) complemented by an in-depth uncertainty analysis. The results indicate that for Lai Chau Reservoir, the ANN model provided the most accurate prediction for one-day-ahead forecasts (NSE = 0.88, RMSE = 172, MAE = 106, and KGE = 0.91). Meanwhile, for Ban Chat Reservoir, the LSTM model achieved the best one-day-ahead prediction (NSE = 0.75, RMSE = 87, MAE = 35, and KGE = 0.76). When applying these best-performing models (ANN for Lai Chau and LSTM for Ban Chat) to multi-step-ahead forecasting, the results reveal a decline in performance as the lead time increases. Generally, forecasts become less reliable (NSE < 0.6, KGE < 0.7) when the lead time exceeds seven days.
Year: 2025