Influence of metrological factors on variations of particulate matter (PM10) concentration during haze episodes in Malaysia
Abstract
Particulate matter (PM10) is one of the major problems of air pollution around the world due to its ability to cause the adverse effect to environment and human health. Therefore, it is very important to develop PM10 prediction model for the future to give the early warning to the public. The hourly measurement data of PM10 concentration together with the
meteorological parameters were monitored from the three study areas in Malaysia (Klang,Muar and Seremban). The observations were obtained over four year’s period in year 2004,
2005, 2006 and 2009 that experienced the haze. The descriptive statistics was used to
explain the statistical characteristics for the data sets. While, monthly average and time series plot were used to describe the trend of PM10 concentration during the whole time frame. The finding proved that the haze event occur during June to October due to the
summer season monsoon with the concentration above the concentration above the
Malaysian Ambient Air Quality Guidelines (MAAQG - 50μg/m3). By using the multiple
linear regression analysis (MLR), PM10 prediction model for each study area were
developed based on the PM10 concentration and also weather parameters such as wind
speed, temperature, humidity, SO2, NO2 and CO. There are two types of model which are
MLR and PCA-MLR. Each of the study areas has their own significant meteorological
parameters that have relationship with the PM10 concentration. From these significant, the
PCA-MLR model was developed. Performance indicators analysis was carried out based on
the normalize error (NAE), prediction accuracy (PA), coefficient of determination (R2),
root mean squared error (RMSE) and index of agreement (IA) and the best fit models for
all three study areas were determined. From the study, the PCA-MLR is better than the
MLR model.