Central fitting distributions and extreme value distributions for prediction of high PM 10 concentration

Central fitting distribution such as lognormal distribution can give a good result for fitting the mean concentration of air pollutants data. However, many researches find that it cannot precisely fit the high concentration region. Therefore, this research compares central fitting distribution (cfd) that is the lognormal distribution with the extreme value distributions (evd) that are Frechet and Gumbel distribution to fit the maximum daily PM 10 concentration in 2002, 2003 and 2004 data in Seberang Perai, Penang. Method of moments was used to estimate the distributions' parameters. Goodness of fit criteria was used in order to select distribution that gives the best fit to the data. Furthermore, the exceedences of a critical PM 10 concentration over the Malaysian Ambient Air Quality Guidelines (MAAQG) were estimated using the best distribution. The results show that the extreme value distribution gives better fit to the actual high PM 10 concentration than the central fitting distribution. The exceedences over the MAAQG were also successfully predicted by using this method.