Forecasting Daytime Ground-Level Ozone Concentration in Urbanized Areas of Malaysia Using Predictive Models

Ground-level ozone (O3) is one of the most significant forms air pollution around world due to its ability cause adverse effects on human health and environment. Understanding variation association O3 level with precursors weather parameters important for developing precise forecasting models that are needed mitigation planning early warning purposes. In this study, hourly data (O3, CO, NO2, PM10, NmHC, SO2) (relative humidity, temperature, UVB, wind speed direction) covering a ten year period (2003–2012) in selected urban areas Malaysia were analyzed. The main aim research was model band greatest solar radiation meteorology using proposed predictive models. Six developed which Multiple Linear Regression (MLR), Feed-Forward Neural Network (FFANN), Radial Basis Function (RBFANN), three modified models, namely Principal Component (PCR), PCA-FFANN, PCA-RBFANN. performances evaluated four performance measures, i.e., Mean Absolute Error (MAE), Root Squared (RMSE), Index Agreement (IA), Coefficient Determination (R2). Surface best described linear regression (MLR) smallest calculated error (MAE = 6.06; RMSE 7.77) highest value IA R2 (0.85 0.91 respectively). non-linear (FFANN RBFANN) fitted observed well, but slightly less accurate compared MLR. Nonetheless, all unmodified (MLR, ANN, RBF) outperformed modified-version (PCR, PCA-RBFANN). Verification done pollutant 2018. MLR dataset 2018 very well predicting daily specified range values 0.85 0.95. These indicate can be used as reliable methods predict daytime Malaysia. Thus, it tool by authority forecast high concentration providing population.