A novel unit Garima-generalized extreme value distribution for stationary and non-stationary modeling: Application to PM2.5 maxima in Thailand

Abstract

This presented an extended generalized extreme value (GEV) distribution called the unit Garima-generalized extreme value (UGa-GEV) model for block maxima (BM) analysis. Its properties, including survival, hazard, quantile, and return level functions, are proposed. The parameters of the proposed models were estimated by the maximum likelihood method, which was applied to analyze the maximum monthly particulate matter with a diameter of less than 2.5 microns (PM2.5) in Thailand. Climate change has undermined the reliability of stationary models to accurately record extreme environmental events over the last few decades with exploration now focusing on nonstationary factors like trends and shifts to better reflect changing environmental conditions. To address this problem, this study proposed non-stationary maxima models using covariates in meteorology (PM10, ozone, carbon monoxide, nitrogen dioxide, and sulfur dioxide) and time. The optimal model between the GEV and UGa-GEV was compared using real data analysis in Bangkok and Chiang Rai, Thailand with our results demonstrating that the proposed UGa-GEV model provided more flexibility than the traditional GEV model.