Analysis of RADM gas concentration predictions using OSCAR and NEROS monitoring data

Abstract

The ability of the Regional Acid Deposition Model (RADM) to produce realistic atmospheric trace gas concentration patterns for a variety of atmospheric conditions is explored by comparing model results with surface monitoring data. Two frontal storm system episodes, monitored in 1981 as part of the Oxidation and Scavenging Characteristics of Aptil Rains (OSCAR) experiment, and one relatively dry high-pressure-system case from the 1979 Northeast Regional Oxidant Study (NEROS) were studied. The temporal and spatial variations of SO 2, O 3 and N species monitored at selected locations throughout the eastern U.S. and Canada are compared to the corresponding RADM grid-average predictions. The analysis is complicated by the fact that grid averages are nor necessarily representative of point measurements when the measurements are strongly influenced by local subgrid-scale phenomena. The comparison database also is limited by gaps in the hourly monitoring data and ambiguity in the N species measurements. To reduce these difficulties, observational sites located in rural and suburban areas which are less influenced by large emission sources were chosen for the comparisons, sharp hourly spikes in the observational data were omitted from the analysis and nitrogen species comparisons were limited to special observations where the measurement ambiguities were less pronounced. Spatial distributions of SO 2 and O 3 daytime (10 a.m.–4 p.m.) and night-time (10 p.m.–4 a.m.) RADM concentration averages were compared to the observational averages to confirm that the model and data concentration patterns agreed over the domain. Intercomparisons of individual grid-point pair comparisons were facilitated using fractional differences, which are defined as the difference in the observations and predictions divided by the sum of the observations and predictions. Predictions and observations showed similar patterns of high and low concentrations throughout the region for all three of the events for the daytime comparisons; the lower night-time agreement improved when night-time stability adjustments were applied to the model predictions. Summarizing over all of the individual grid-point, day-total comparisons for all three events, it was found that, for O 3, all of the pairs were in agreement to better than a factor of two and 50% of them were in agreement to a factor of 1.1 or better; for SO 2, 88% of the pairs were in agreement to better than a factor of two and 25% of the pairs were in agreement to a factor of 1.1 or better. Predicted N species concentrations in the 1–4 and 5–7 ppb ranges for two OSCAR cases were comparable to the levels measured at a rural Indiana site during the OSCAR I and OSCAR IV events. Predictions of NEROS NO x in the 3–11 ppb range were comparable with the concentrations measured at a rural Pennsylvania site for a summer 1986 case meteorologically similar to the 1979 NEROS I event. In general, higher agreement is associated with well-mixed conditions in areas where local characteristics of the observational site more closely resemble the grid-average characteristics.