Predicting Particulate Matter (PM₁₀) during High Particulate Event (HPE) using Quantile Regression in Klang Valley, Malaysia

Abstract

Particulate matter (PM10) is the key indicator of air quality index (API) during high particulate event (HPE). The presence of PM10 is believed to have an adverse effect on human health and environment. Therefore, the prediction of future PM10 concentration is very important because it can aid the local authorities to implement precautionary actions to limit the impact of air pollution. This study aims to compare the performances of two predictive models, which include Multiple Linear Regression (MLR) and Quantile Regression (QR) in predicting the next-day PM10 concentration during HPE. The hourly dataset of PM10 concentration with other trace gases and weather parameters at Kelang and Petaling Jaya from the year of historic haze event in Malaysia (1997, 2005, 2013 and 2015) were obtained from Department of Environment (DOE) Malaysia. Three performance measures namely Mean Absolute Error (MAE), Normalised Absolute Error (NAE) and Root Mean Squared Error (RMSE) were calculated to evaluate the performances of the predictive models. From the results, QR model at quantile 0.3 and 0.6 was chosen as the best predictive tools for predicting the next day PM10 concentration during haze event in Kelang and Petaling Jaya, respectively. showed better performance for the prediction of next-day PM10 concentration in Kelang. These results indicate that QR can be used as one of predictive tool to forecast air pollution concentration especially during unusual condition of air quality.