Using the bivariate approach to spatial estimation of air pollution by ozone

Improved techniques for analyzing data and spatial interpolation of the concentration of pollutants in urban air, obtained from monitoring stations, it is important to study the problems of smog formation and to determine the urban areas where a high concentration of pollutants may affect the health of its inhabitants, nature conservation and preservation of property. The most common techniques used so far considered the methods based on the inverse of the distance between measured points and points to consider IDW, and the geostatistical methods ordinary kriging KO, and universal kriging KU, which account for stationary and nonstationary data respectively. To determine the effect of non-stationarity on the mean of the data used in this study initially KO and KU methods are considered for interpolation of data from the peak of ozone and monthly average concentration of ozone in the atmosphere of the Mexico City. A comparison of estimation accuracy achieved by these two models allows us to judge the importance of stationarity in the estimation. Cross-validation is used to compare the performance of the two methods of interpolation. Next we consider bivariate estimate, which incorporates variables that are highly correlated with the concentration of ozone.The cokriging method is used to incorporate the temperature, the concentration of NO2 and SO2 concentration as secondary variables in the spatial prediction of Ozone. A cross-validation procedure is used for statistical evaluation of the results and to compare the predictive ability of variables used in the cokriging method for estimating the concentration of ozone. To this end we evaluate the root mean squared error RMSE between the observed and estimated values at each station for each of the methods under consideration.© 2010 Published by Elsevier Ltd. Selection and/or peer-review under responsibility of Centro de Investigación en Geografía y Geomática “Ing. Jorge L. Tamayo”