Source Region Identification Using Kernel Smoothing

Abstract

As described in this paper, nonparametric wind regression is a source-to-receptor source apportionment model that can be used to identify and quantify the impact of possible source regions of pollutants as defined by wind direction sectors. It is described in detail with an example of its application to SO2 data from East St. Louis, IL. The model uses nonparametric kernel smoothing methods to apportion the observed average concentration of a pollutant to sectors defined by ranges of wind direction and speed. Formulas are given for the uncertainty of all of the important components of the model, and these are found to give nearly the same uncertainties as blocked bootstrap estimates of uncertainty. The model was applied to data for the first quarter (January, February, and March) of 2003, 2004, and 2005. The results for East St. Louis show that almost 50% of the average SO2 concentration can be apportioned to two 30 degrees wide wind sectors containing a zinc smelter and a brewery; a nearby steel mill did not appearto have a significant impact on SO2 during this period.