Assessment of Air Pollution Modelling using Spatial Interpolation Techniques

Abstract

Air pollution has become the growing worldwide threat. Effective assessment of air pollution modelling depends on significant, wide-spread and distinguished instrumental data which are not possible for a developing country like India to have established an extensive network. This shortcoming on the capability to precisely envisage the pollutant concentration at unmonitored stations can be accomplished by spatial interpolation. Spatial Interpolation is a process, in which a parameter can be predicted and enumerated from a limited number of sample data points which is used to model the data for air pollutants like NO2, SO2 and RSPM. This study establishes an assessment of interpolation techniques to produce fine-scale air quality data for Tamil Nadu, India. The interpolation techniques which are used to evaluate the air quality data are Kriging, Inverse Distance Weighting, Local Polynomial Interpolation, Global Polynomial Interpolation, Radial Basis Function, Kernel Interpolation, Diffusion Interpolation, Natural Neighbour and Spline. The accuracy assessment was done with the help of statistical error metrics such as Root Mean Square Error, Mean Average Error, Mean Average Percentage Error, Mean Average Relative Error and Index of Agreement. Cross-Validation is executed by considering 20% of the data points as test data and remaining 80% as training data respectively. Spline, Local Polynomial Interpolation and Global Polynomial Interpolation technique showed enhanced resemblance between the observed and estimated values for the pollutants NO2, SO2 and RSPM respectively. This study embarks the association between the air pollutant and their distribution through spatial interpolation using Geographical Information System.