Modelling of extreme streamflow using copula

This study explores applying various copula models to estimate the dependencies between streamflow and stage data for the Kahang River in Kluang, Johor. Using daily streamflow and stage data, we compared the performance of several copula parameter estimation methods: Maximum Pseudo-Likelihood Estima...

Full description

Bibliographic Details
Main Authors: Buliah, Nur Amirah, Ling, Wendy Shin Yie
Format: Article
Language:English
Published: Universiti Putra Malaysia 2024
Online Access:http://psasir.upm.edu.my/id/eprint/120285/
http://psasir.upm.edu.my/id/eprint/120285/1/120285.pdf
Description
Summary:This study explores applying various copula models to estimate the dependencies between streamflow and stage data for the Kahang River in Kluang, Johor. Using daily streamflow and stage data, we compared the performance of several copula parameter estimation methods: Maximum Pseudo-Likelihood Estimator (MPLE), Inference Functions for Margins Estimator (IFME), Method-of-Moments Estimator (MoM), Empirical Copula Estimation, and Robust Estimation by Maximum Mean Discrepancy Minimization (MMD). Our findings indicate that different copula performed best for different estimation methods. Specifically, the Student t-copula best fits IFME, the Frank copula for Kendall’s tau, Spearman’s rho, and the most recent method, MMD. Also, the Joe copula is best for the MPLE and the empirical copula estimation method. The Jackknife interval method produced narrower and more precise confidence intervals across multiple methods, making it the best interval estimator. This comprehensive analysis improves hydrological modelling, facilitating effective water resource management and flood risk assessment.