A Sensitivity Analysis of the Treatment of Area Sources by the Climatological Dispersion Model

Abstract

The area-source contributions calculated by the Climatological Dispersion Model (CDM) are sensitive to the size of the grid square used to specify the area source emissions. The authors recommend estimation of highest long term concentrations be accomplished by specifying the emissions with as small a grid size as convenient in that portion of the region having the largest area-source emissions.

These conclusions are based on a sensitivity study employing the SO₂ emissions inventory developed as part of the St. Louis Regional Air Pollution Study. This inventory specifies the SO₂ emissions over a 200-km (east-west) by 140-km (north-south) region. The CDM was used to estimate annual average SO₂ concentrations from area-source emissions for 13 locations in the St. Louis metropolitan area for 1976. Although the area source emissions constituted only 3.5 % of total area and point source emissions, estimated concentrations from area sources ranged from 14 to 67% of the total concentration estimated at the 13 sites. The median estimated impact from area sources was 32% of the total from both point and area sources. We examined the variation in concentration estimates due to variation in the degree of detail with which the spatial changes in the emissions were specified and due to variation in the areal extent of the emissions modeled. We observed that, for the receptor locations nearest the larger area-source emissions, the concentration estimates were quite sensitive to the detail with which the spatial variation in emissions was specified. Variations in calculated impact of greater than 30% were observed depending on whether a grid size of 1, 2, 5, 10 or 20 km was selected. We observed that the area source concentration estimates were dominated by the portion of the emissions inventory having the majority of emissions. Neglecting 35 % of the area source emissions on the outer fringes of the region and thereby reducing the rfiodeled inventory to a 50 by 50-km square, gave nearly identical results to modeling the entire inventory. Based on the wide variability in calculated concentrations attributable to user-selected parameters, the authors strongly recommend that sources be modeled as point sources and only minor indeterminate emissions be modeled as area sources.