Prediction of Monthly Averages of Air Pollutant Concentrations for Selected Areas in Mazovian Voivodeship

The study was carried out using long-term data, recorded at two air monitoring stations in Masovian Voivodeship. Hourly time series, obtained from the monitoring system, were averaged in calendar months to get monthly time series. The data sets, containing time series of monthly mean values from two different monitoring sites, were subjected to multivariate regression analysis. Models of multidimensional linear regression were built for the both sets of data. The obtained models describe statistical dependencies between concentrations of specified air pollutants and concentrations of other pollutants and meteorological parameters, recorded at the same monitoring station. The achieved regression equations were used to predict long-term courses of monthly concentrations. For visualization of prediction accuracy, the charts containing time series of actual and predicted monthly concentrations were prepared. The approximation precision was estimated by calculating modelling errors for each regression model. Three different measures of approximation error were applied: mean absolute error (MAE), root mean square error (RMSE), and Pearson correlation coefficient (r).