| Summary: | Particulate matter (PM10) is the main air pollutant during high particulate event (HPE) or also known as haze in Southeast Asia specifically in Malaysia. PM10 emanation is believed to cause the strongest harm to public health and environment during this time. Therefore, it is very important to develop good PM10 prediction model during these event that can be used to give the early warning to the public. A database with hourly PM10 concentration together with other trace gases and weather parameters were obtained from Department of Environment (DOE) Malaysia. The dataset was obtained from 2012 to 2016 at two study areas located in Klang Valley, namely, Petaling Jaya and Shah Alam. Three predictive models namely Multiple Linear Regression (MLR), Principle Component Regression (PCR) and Artificial Neural Network (ANN) were developed to predict the concentration of PM10 for the next-day, next-two-day and next-three-day. The predicted values were evaluated using several performance indicators i.e. Normalised Absolute Error (NAE), Root Mean Squared Error (RMSE), Prediction Accuracy (PA), Coefficient of Determination (R2) and Index of Agreement (IA). ANN was selected as the best prediction model for PM10 concentration during HPE with the smallest average error (NAE = 0.11; RMSE = 9.69) and highest agreement with the observed values with the average of performances of R2 = 0.97. © 2022 Universitatea "Alexandru Ioan Cuza" din Iasi. All rights reserved.
|
|---|