Modeling the Characteristics of Unhealthy Air Pollution Events Using Bivariate Copulas

Abstract

Investigating the dependence structures among the characteristics of the current unhealthy air pollution events is a valuable endeavor to understand the pollution behavior more clearly and determine the potential future risks. This study determined the characteristics of air pollution events based on their duration, severity, and intensity. It focused on modeling the dependence structures for all the possible pairs of characteristics, which were (duration, intensity), (severity, intensity), and (duration, severity), using various parametric copula models. The appropriate copula models for describing the behavior of the relationship pairs of the (duration, intensity), (severity, intensity), and (duration, severity) were found to be the Tawn type 1, 180°-rotated Tawn type 1, and Joe, respectively. This result showed that the dependence structures for the pairs were skewed and asymmetric. Therefore, the obtained copulas were appropriate models for such non-elliptical structures. These obtained models can be further extended in future work through the vine copula approach to provide a more comprehensive insight into the tri-variate relationship of the duration–intensity–severity characteristics.