Modeling and Forecasting Air Pollution of Tehran Application of Autoregressive Model with Long Memory Properties

Environmental pollution modeling is one of the essential requirements in the field of air quality monitoring which with using the output of the model, improvement of future situation can be possible. The existing literature of the modeling of environmental pollution –especially air pollutants- could be divided to two whole categories. First, those researches that in addition of pollutants data, they used some factors such as temperature, wind direction, wind speed and humidity. The second one –which this study belong to- with using time series regression models and by usage of the existing data about each pollutant, the future situation was forecasted.
In this study, we forecast future pollutants (CO,PM10,NO2,SO2,O3,PM2.5) status with ARIMA, ARFIMA and ARIMA-GARCH models with Box-Jenkins approach, then the best model is determined with MSE, RMSE, MAE and MAPE.
Findings: Results indicate that the assumption of existence of long-memory is acceptable but the hypothesis that always ARFIMA models prepare the best forecast is rejected.
Discussion and This study proves the application of econometric models to predict the pollutants state. Based on the high social costs of pollutant emissions, it is recommended that using these models, identify the pollutants affecting the future of the city and reduce the level of their dissemination of efficiency plans