Modeling generalized statistical distributions of PM 2.5 concentrations during the COVID-19 pandemic in Jakarta, Indonesia

Understanding the probabilistic or statistical behavior of air concentrations is necessary for the
effective management of air pollution, such as PM
2.5
. Failure to consider the appropriateness of
the model can lead to making inferences that are not supported by scientific evidence. The main
focus of this article is to find the best statistical distribution in fitting PM
2.5
concentrations in the
periods of February–June 2018 and February–June 2019 (the periods without COVID-19) and
in the period of February–June 2020 (the period with COVID-19) in Jakarta, Indonesia. This
article considers making an assessment of the performance of both generalized distributions (e.g.,
generalized gamma, generalized extreme value, and generalized log-logistic [GLL]) and classical
distributions (such as lognormal [LN], gamma, Weibull, log-logistic, and Gumbel) in modeling
daily concentrations of PM
2.5
in the period of February–June 2020, or the period during which
the COVID-19 pandemic is present, in Jakarta. For comparison purposes, this study also
analyzed PM
2.5
concentrations in the periods of February–June 2018 and February–June 2019.
The comparative evaluation of the models of each period of data uses graphical analyses and
goodness-of-fit statistics. The results of applications indicate that the generalized distributions
fit the data better than do the classical distributions. Particularly, compared with the classical
distributions, including the LN model, the GLL distribution is the most appropriate model in
fitting PM
2.5
concentrations in the periods without and during the period with COVID-19 in
Jakarta, Indonesia.