A Predictive SARIMA Model for PM10 and PM2.5 levels in Mashhad based on traffic flow and metrological data

Air pollution is known as a major cause of health and environment damages.Various factors are involved in increasing air pollution, among which road trafficis one of the main sources of these issues. Therefore, in this study, the effects oftraffic flow and metrological parameters on the PM10 and PM2.5 levels areinvestigated by a predictive model. Time series analysis is used to predict futuredaily levels of PM10 and PM 2.5 in Mashhad, based on predicted daily traffic flowon the major highways of Mashhad, temperature, wind speed and humidity. Themajor innovation of this paper is that the air pollution time series is modeledbased on another time series (traffic volume). In other words, the time seriesmodel for air pollution contains some time series variables (as exogenousvariables). These time series variables have some effects on each other, which areconsidered by Cross-Correlation Function (CCF). For each variable,Autocorrelation Function (ACF) and Partial Autocorrelation Function (PACF)are calculated. ACF defines the seasonal patterns of the observations, and PACFremoves dependence of internal lags for each variable. A predictive SARIMAmodel, which estimates the future levels of PM10 and PM 2.5 is a result of thisstudy. The R-Square of the proposed model is 0.714 and 0.676; and RSME of itis 8.667 and 9.374 for PM10 and PM 2.5; respectively.

کلیدواژه ها:

SARIMA model ، a ir pollution ، traffic flow ، PM10 and PM 2.5 levels