Emission Source Location Errors as a Determinant of Concentration Estimate Inaccuracies

摘要

We report our experience in identifying errors in the reported locations of industrial emission sources that influence the actual - and the estimated - concentrations of air toxics in the Greenpoint-Williamsburg neighborhoods in Brooklyn, NY. Using GIS, we placed the sources within and around the study area. We found that, of 87 TRI air emissions sources in Brooklyn, about half of the TPI-supplied latitudes and longitudes placed the facilities within 150 meters of their correct locations. Other facilities -- over 15% of them -- were misplaced by between I and 7 kilometers from their actual locations. We found that, overall, location errors contributed modestly to the fairly wide uncertainty bounds around the model's predictions. However, the primary impact of these errors takes place at the local level: The incorrect locations of two facilities, for instance, when corrected, would move a predicted "hot spot" for one of the toxic substances from an area with high population density to one with fewer people, but higher percentages of minority and poor residents. We also document that readily-available sources of receptor site locations such as schools also suffers from similar errors. We discuss methods for efficiently identifying and correcting location errors.