Seasonal ground level ozone prediction using multiple linear regression (MLR) model

To assess the surface ozone concentration (O3), there is a need to establish relationship between air pollutants and meteorological parameters. The study was conducted on variation of air pollutants, viz. O3, nitrogen oxides (NOX = NO2 + NO) and carbon monoxide (CO) along with meteorological parameters like temperature (Temp), relative humidity (RH), solar radiation (SR) and wind speed (WS). The precursor gases were recorded in Hyderabad at Tata Institute of Fundamental Research-National Balloon Facility (TIFR-NBF; 17.47° N, 78.58° E). Correlation analysis is done on hourly averaged trace gases concentration and metrological data for the entire year 2016. O3 is in negative correlation with NOX, CO and RH. NOX which is one of the precursor gases plays a major role in formation of O3 by photo-chemical reaction (PCR). The increase in O3 concentration is in proportion with the decrease in NOX concentration. O3 correlated positively with Temp, SR and WS. Two sets of four models were constructed with multiple linear regression (MLR) representing the data for the three seasons (summer, winter and monsoon) and for the total year as well. The adjusted R2 was determined and found to be in the range of 0.6 to 0.9 for the models using precursor gases and 0.9 by meteorological parameters. The models were validated by various performance indicators, viz. root mean square error (RMSE), mean absolute error (MAE) and mean biased error (MBE).