Estimation of R-Vine Copula Parameters in Multivariate Flood Frequency Analysis Using Arithmetic Optimization Algorithm and Comparing the Performance with Genetic Algorithm

This study aims to compare the performance of the Arithmetic Optimization Algorithm (AOA) and Genetic Algorithm (GA) in parameter estimation of R-Vine models in multivariate flood frequency analysis. Three hydrometric stations on the Karkhe River, including Hamidiyeh, Abdolkhan, and Paye-Pol, were considered as the case study. The R-Vine models (D-Vine and C-Vine) were considered for 4-variable flood frequency analysis (peak discharge, flood volume, base time, and peak time). Results showed that AOA outperformed GA for both D-Vine and C-Vine models in all three stations. AOA in the D-Vine model represented better performance with loglikelihood values ​​of 1802.96, 1644.67, and 1609.84, respectively, for Hamidiyeh, Abdolkhan, and Paye-Pol, compared to GA with loglikelihood values ​​of 1700.51, 1581.73 and 1524.48. Similarly, in the C-Vine model, AOA was able to achieve a loglikelihood value of 1782.67 vs. 1667.03 by GA for Hamidiyeh, 1648.82 vs. 1437.38 for Abdolkhan, and 1622.83 vs. 1472.96 for Paye-Pol stations. Generally, it was concluded that AOA outperformed GA, particularly in the exploitation phase, for estimating the optimal answer with higher accuracy. The findings of this research can assist managers and decision-makers in the water resources management sector. Some of this research’s main practical applications include flood control planning, estimating flood damages, and reservoir operation management. Graphical Abstract