Home > Geospatial Application Papers > Environment > Air Pollution
Abstract

Multi-Dimensional Interpolation Using Artificial Neural Networks: An Urban Air Pollution Case Study


Saeed Nadi
Department of Surveying and Geomatic Eng., Engineering Faculty, Islamic Azad University, Meybod Branch, Yazd, Iran,
Iran
Email: snadi@engineer.com


Interpolation is one of the most important analyses for air pollution study and many other applications in science and engineering. Nowadays, the fast growth of geospatial information systems (GIS) facilitates the applications of spatial interpolation which has attributed much attention recently.Although there are a number of renowned methods for spatial interpolations such as Inverse Distance Weighting (IDW), spline or kriging, one of the difficulties researchers almost encountered when using these methods are their limitations for multi-dimensional interpolations. The spatial interpolations are vital analysis in so many applications such as air pollution analysis and monitoring in which pollutant parameters such as CO, SO2, O3, NO, NO2 and PM (particulate matters) must be interpolated under different space-time positions.In this paper some tests have been provided using a data set of 1832 control points and 38 check points for interpolating X, Y, Z using IDW, kriging, spline and multi layer perceptron (MLP) neural network with two and three layer each contains from one to five neurons and different configurations of PURELIN, LOGSIG and TANSIG activation functions in each layers. The RMSE for control points and check points are computed and compared for each method.Using the results of this study, measurements of five air pollutant parameters including CO, O3, NO2, SO2 and PM for seven air quality monitoring stations over Tehran metropolitan in Iran for last two years are interpolated. For this, a multi dimensional interpolations using MLP neural network are used to monitor air pollution for any time in any place. Furthermore, the extrapolation of pollutions for near future provides very interesting information which can be used in many critical decisions such as by aged or ill peoples to avoid dangerous places where and when the pollution reaches a special threshold.