Spatial Analysis to Predict PM۱۰ Pollutant in City of Tehran

Air pollution, which is becoming more severe as cities grow and develop, has forced the government to control and monitor pollutants. This essay focuses on the spatial prediction of the monthly average of PM۱۰ using Geographic Information science (GIS). This was done using an MLP Neural Network and a Random Forest on a grid of ۵۰۰m * ۵۰۰m and ۳۰۰۰m * ۳۰۰۰m. The results revealed that the MLP in the Grid ۳۰۰۰m * ۳۰۰۰m with an RMSE of ۲.۴۹۲ outperformed the Grid ۵۰۰m * ۵۰۰m with an RMSE of ۳.۵۹۲. Therefore, PM۱۰ modelling succeeded at higher spatial resolution.

Keywords:

: Air pollution , Neural Network , Random Forest , Spatial prediction , GIS