.png){kind=icon}
.png){kind=icon}
Return
Mixture copula parameter estimation with metaheuristic algorithms, comparative study under hydrological context
.png){kind=icon}
PRE
AI
DOI：10.1007/s00477-025-02914-4.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Hydrological events are often described by several dependent characteristics, such as peak, volume, and duration for floods. They can be studied in a multivariate framework where copulas are a powerful tool to capture the dependence structure. A multivariate Hydrological frequency analysis is generally based on homogeneity, serial-independence, and stationarity. However, the homogeneity assumption is not often fulfilled due to several reasons, such as climate change, human activities, and the mixture nature of multiple generating processes, potentially affecting the copula selection. As a result, using a single copula (non-mixture) may not be appropriate, and a mixture of copulas is needed. A growing number of studies have recently been conducted on parameter estimation for mixture copulas. However, the hydrological literature on mixture copulas is still in its infancy. Furthermore, the developed estimation methods have numerous optimization-related drawbacks. Due to their successful application for several optimization tasks, we consider metaheuristic algorithms for mixture copulas' parameter estimation. A simulation study is performed to evaluate and compare the effectiveness of these algorithms under hydrological constraints. Simulation results show similar performance in terms of relative errors, whereas a considerable difference is denoted concerning the needed time to converge to optimal parameters of mixture copula models. Moreover, two applications to real-world datasets are provided.
Keywords:
Hydrologic frequency analysis
Homogeneity
Mixture copulas
Parameter estimation
Metaheuristic algorithms
Maximum pseudo-likelihood
Monte Carlo simulation
Journal
IF:
3.6
Papers:
3.5K
Citations:
6.9K