Forecasting PM10 concentrations using time series models: a case of the most polluted cities in Turkey.

Abstract

Particulate matter (PM), which is one of the most important parameters in the area of air pollution, has widespread impacts on human health. Hence, the prediction of the probable concentration of PM is a highly significant subject with regard to primary warning for the protection of a population. Turkey is among the European countries with polluted air in terms of the concentration of PM with a diameter smaller than 10 μ m (PM10). The PM10 data supplies significant knowledge about how much pollution is in the air and which city is the most polluted. In this study, the values of PM10 for the most polluted cities in Turkey are forecasted using time series models, including autoregressive integrated moving average (ARIMA), error, trend and seasonal (ETS), and singular spectrum analysis (SSA). Forecast values of PM10 averaging period of 24 h for the year 2019 are obtained using SSA as the optimum time series method. The results show that the annual means of PM10 concentrations in 2019 in Hatay and Yalova, the most polluted cities, will not exceed the 50 μgm- 3 value according to air quality standards determined by the European Commission. The air quality levels of eight other cities, which are Adana, Ankara, Icel, Istanbul, Kirklareli, Sakarya, Samsun, and Sivas, will reach acceptable standards between 50 and 70 μgm- 3 for annual mean in 2019. The remaining eight cities, Amasya, Bursa, Denizli, Kahramanmaras, Kutahya, Manisa, Nigde, and Tekirdag, continue to be the most polluted cities in 2019 according to the average annual PM10 values. This study also reveals that the average PM10 value of the most polluted cities in Turkey will be 68.97 μgm- 3 for the 24-h average in 2019.