Predicting CO Concentrations Levels Using Probability Distributions

Full Text PDF PDF
Author(s) A. S. Yahaya | N.A. Ramli | A.Z Ul-Saufie| H. A. Hamid| H. Ahmat| Z.A Mohtar
Pages 900-905
Volume 3
Issue 9
Date September, 2013
Keywords Carbon monoxide, probability distributions, performance indicators

Abstract

In Malaysia, air pollutant emissions were monitored all over the country to detect any significant change which may cause harm to human health and the environment. This research is focused on carbon monoxide (CO) concentration as it is known to cause severe health impact to human as well as environment. Therefore a well developed system need to be used in order to analyze the trending of all of the pollutants emission inventories. In this research, seven theoretical distributions that are Weibull, gamma, lognormal, Laplace, Rayleigh, log-logistic and inverse Gaussian distributions were developed. It is used to verify and simulate the trend of the monitoring data for CO emission in Kuala Lumpur, which is the capital of Malaysia, in the form of probability density function and hence can be used as a prediction tool. The method of maximum likelihood estimates (MLE) was used for estimating the parameters of the distributions. The best distribution was determined using the plots for the cumulative distribution functions (cdf) and performance indicators. Five performance indicators used are the root mean square error (RMSE), index of agreement (IA), prediction accuracy (PA) and coefficient of determination (R2). From the performance indicators, it was found that the best distribution to represent the CO concentration level in Kuala Lumpur for 2002 is the inverse Gaussian distribution. Based on the prediction using the inverse Gaussian distribution, it can be concluded that the CO concentration level in Kuala Lumpur does not exceed the Malaysian Ambient Air Quality Guidelines of 9 parts per million (ppm).

< Back to September Issue