Evaluation of Optimum Methods for Predicting Pollution Concentration in Gis Environment

Air pollution is overflowing in big cities, especially in areas where pollution sources and the human population are concentrated. Economic growth and industrialization caused the increasing emissions of air polluting. Then, the quantities of polluting have increased dramatically; the evaluation of a suitable method for predicting and monitoring the pollution will be very important. To prevent or minimize damages of atmospheric pollution, optimum predicting methods are urgently needed which can rapidly and reliably detect and quantify air quality. One of the important spatial analyses for this application in GIS environment is surface simulation using Gostatistical methods. These lead us through creating a statistically valid surface which subsequently is used in GIS models for optimum decision making. Analysis create predicted surface for unmeasured points (Which we have not enough information on them) in study area. For This purpose, Ground stations and MODIS image of Tehran are used for collecting online air pollution information. Then, different geostatistical methods have been used for finding out the optimum prediction method for air pollution, based on received observations. These methods are performed based on spatial relationships (spatial similarity) among the measured points. In here, we use fractal and simple semivariogram for calculating correlation between points and determining which one of them is better for our application. We tested that the fractal dimension which measured by the spatial correlation length is more reliable based on autocorrelation and structural analysis. After that, we proved co-Kriging interpolation is more accurate by producing and evaluating prediction standard error maps.