A new statistical model for extreme rainfall: POT-KumGP

The accurate prediction of the extreme rainfall is of great interest in geology to make provision for nature events such as landslide. The extreme value theory is widely used to predict the daily extreme rainfall probability. In this study, we introduce a new extreme value model based on the Kumaraswamy generalized-Pareto distribution by applying it to peaks-over-threshold method. This method is shortly denoted as POT-KumGP. The extreme rainfall modeling accuracy of the POT-KumGP model is compared with a POT-GP model by means of a real data modeling on the daily rainfall data of Hopa region located in Artvin province of Turkey. The empirical results show that the POT-KumGP model produces more accurate results than POT-GP model based on the model selection criteria and result of goodness-of-fit test.

Automated summary

The study introduces a new extreme value model, and through applying it to real data modeling, it is found that the POT-KumGP model produces more accurate results compared to the POT-GP model.