Document Type : Research Paper
Department of Spatial Information Systems, Faculty of Surveying Engineering and Geospatial Information, College of engineering, University of Tehran, Tehran, Iran.
Today, air pollution has become one of the most critical problems in densely populated cities, which causes many city residents to suffer from lung problems every year and can have irreparable effects on citizens' health. Air pollution recording devices in cities record pollution hourly. The technical issues of these devices sometimes cause some of the important data not to be recorded, and as a result, fixed values are created in the data. In this study, fixed values have been estimated. For this purpose, the study of air pollution events in Tehran including the concentration of PM2.5, PM10, SO2, NO2, O3 and CO pollutants was conducted. The LANN algorithm, used in the estimation and forecasting of single-variable time series, has been implemented and compared for all pollutants. Also, in another part of the study, other environmental pollutants have been considered in the estimation of fixed values, and by using the neural network method, the estimation of fixed values for all pollutants has been done. RMSE index was also used to check and compare algorithms. The value of RMSE in the LANN method was lower than other simpler models including mean, linear regression and LOCF, so its value was 30 to 50% lower, depending on the type of pollutant. Also, the neural network algorithm had lower RMSE than other methods in estimating PM2.5 values and its value was 7.78.