The Replacement of Missing Values of Continuous Air Pollution Monitoring Data Using Mean Top Bottom Imputation Technique

Air pollutants data such as PM10, carbon monoxide, sulphur dioxide and ozone concentration were obtained from automated monitoring stations. These data usually contain missing values that can cause bias due to systematic differences between observed and unobserved data. Therefore, it is important to find the best way to estimate these missing values to ensure that the data analyzed are of high precision. This paper focuses on the usage of mean top bottom imputation technique to replace the missing values. Three performance indicators were calculated in order to describe the goodness of fit of this technique. In order to test the efficiency of the method applied, PM10 monitoring dataset for Kuala Lumpur was used as case study. Three distributions that are Weibull, gamma and lognormal were fitted to the datasets after replacement of missing values using mean top bottom method and performance indicators were calculated to describe the qualities of the distributions. The results show that mean top bottom method gives very good performances at low percentage of missing data but the performances slightly decreased at higher degree of complexity. It was found that gamma distribution is the most appropriate distribution representing PM emissions in Kuala Lumpur.