Relevance of ENSO indices as predictors in seasonal river discharge forecasting in the Srepok basin

Abstract Flow irregularities seriously affect agricultural production and ecosystems, so early forecasting of river discharge contributes to minimizing its impacts. This study analyzes the relationship between El Niño–Southern Oscillation (ENSO) and river discharge in the Srepok basin, thereby building a forecast model using the stepwise linear regression method. The Srepok Basin located in the Central Highlands of Vietnam, is an sub-basin of the Mekong River system. In this study, the dependent variable is monthly mean river discharge, and the independent variables are sea surface temperature (SST) in Niño regions, Southern Oscillation Index (SOI), river discharge in the previous month. These data cover the period from 1982 to 2022. Analysis results show that the independent variables have a good relationship with the dependent variable in the period from February to May and the dependent variable is taken from 1 to 6 months in advance. The evaluation results show that it is possible to forecast monthly average river discharge based on ENSO variables and river discharge in the previous month. With forecast periods from 1 to 6 months, the mean absolute percentage error ranges from 17.5% to 27.4% and the Nash-Sutcliffe efficiency ranges from 0.70 to 0.85. Compared to results from other reports, the forecast quality has improved. ENSO variables participating in forecast models only include SOI and SST in Niño 4 regions with time scale from 3 to 4 months. These are also the variables that have the best relationship with river discharge. The relative contribution of these variables to the forecasting equation is from 25% to 49%.