MULTIVARIATE STATISTICAL ANALYSIS FOR NEXT HOUR PM10 CONCENTRATION FORECASTING DURING EPISODIC TRANS-BOUNDARY HAZE EVENTS IN MALAYSIA

Abstract

Episodic trans-boundary haze events had happens caused higher particulate matter (PM10) concentration in Malaysia's atmosphere. The high concentration of PM10 had exceeded the New Malaysia Ambient Air Quality Standard (NMAAQS) and caused a higher Air Pollutant Index (API) reading. Hence, this situation prompts an adverse impact on human health and also the country's economic sector. All the meteorological factors (ambient temperature, relative humidity, and wind speed) data and other gaseous pollutants (carbon monoxide, sulphur dioxide, nitrogen oxide, nitrogen dioxide, and ozone) that involved in this study were acquired from the Department of Environment Malaysia from the year 2005 until 2014 based on the haze chronology information. Thus, the correlation among the influenced factors led to multicollinearity problems, especially in model development. The aim of this study is to develop Multiple Linear Regression (MLR) and Principal Component Regression (PCR) models. The multicollinearity problem has occurred among the independent variables, and the Principal Component Analysis (PCA) has been introducing in the PCR model.