Estimation of the daily flow in river basins using the data-driven model and traditional approaches: an application in the Hieu river basin, Vietnam

Abstract

Abstract This paper presents the long short-term memory (LSTM), rating curve (RC), and rainfall–runoff model that can be used for estimating the daily flow at the outlet of river basins. The Hieu river basin in Vietnam is selected as an example for demonstrating the ability of multiple approaches. Hydro-meteorological data at the Quy Chau station are collected over a long period from 1/1/1991 to 31/12/2020. Multiple approaches mentioned above are implemented and used for calculating the daily flow in the studied river basin. The coefficients and modeling parameters in each approach are then carefully determined based on five statistical error indexes. The results revealed that the RC using either one or two segments and the LSTM model using water elevation as input data represented the observed daily flow very well, with the values of dimensional errors (i.e. mean error, mean absolute error and root mean square error) equal only to about of 1% of the observed magnitude of the flow in the studied river basin, while Nash–Sutcliffe efficiency and correlation coefficients are greater than 0.95. Impacts of different types of input datasets on estimated values of the daily flow are also presented when the LSTM model is applied.