Performance comparison of IHACRES, random forest and copula-based models in rainfall-runoff simulation

Abstract In this study, two models of Random Forest and copula-based simulation were used to evaluate the accuracy efficiency IHACRES rainfall-runoff model in simulating daily discharge Siminehroud River south Lake Urmia basin, Iran. A trivariate was created using discharge, rainfall temperature data on a scale period 1992–2018. Vine family their conditional densities implement model. By calibrating also selecting tree sequence accordance with data, simulations performed study area. The studied evaluated RMSE NSE criteria, violin plot Taylor diagram. results comparing error rate area showed that vine-based reduces statistics by about 14.5 16.5%, respectively, compared models. According presented diagrams, certainty are acceptable. While does not have acceptable has good performance due unique as well involving marginal distributions fitted data. Although increased 5% model, it is significant mathematical complexity