Using Tropical Rainfall Measuring Mission for Accurate Daily Runoff Modeling using intelligent systems: The Case of Chelif Basin (Wadi ouahrane basin in Northen Algeria)

Abstract

Predicting tools for river flow using machine learning and deep learning are considered essential for sustainable planning in water resource management. This study evaluates the effectiveness of Multilayer Perceptron (MLP), Long Short-Term Memory (LSTM), and XGBoost models in predicting daily flows using data from the Tropical Rainfall Measuring Mission (TRMM) in the Cheliff basin in northern Algeria. Our assessment, based on the Root Mean Square Error (RMSE), includes a series of scenarios involving 9 different models. Our results show that the MLP performs competitively, achieving the lowest RMSE in scenario M8 (RMSE = 2.485) when using TRMM data. Moreover, in comparison with in-situ data, both LSTM and MLP models demonstrate reasonable performance. In conclusion, this study highlights the effectiveness of LSTM in using satellite-derived precipitation data for accurate daily flow prediction