Evaluation of photochemical grid models (UAM-IV, UAM-V, and the ROM/UAM-IV couple) using data from the lake michigan ozone study (LMOS)

Data from two ozone episodes from the 1991 Lake Michigan Ozone Study (LMOS) have been used to evaluate two commonly used photochemical grid models-UAM-IV and UAM-V. To assure a fair comparison, the models were run on the same domain and grid size, with the same inputs of source emissions, diagnostic meteorology, and initial and boundary conditions. In addition, UAM-V was run in nested-grid mode with input meteorology taken from a prognostic model (CALRAMS), and UAM-IV was run in regulatory mode with boundary conditions taken from ROM. Little difference was found between the UAM-IV and UAM-V daily peak ozone predictions, which showed mean relative biases of ± 5 to 10%. The use of diagnostic vs prognostic meteorological model inputs had little effect on UAM-V performance statistics, although the spatial location of the broad regional ozone plume was better with the prognostic meteorological model. The ROM/UAM-IV couple tended to consistently underpredict peak ozone by about 20%, primarily due to ROM underpredictions of ozone on the upwind domain boundaries. It was found that all models tended to underpredict ozone precursors (VOCs and NOx) by as much as a factor of two, although the UAM-V underpredictions were not as large. Furthermore, all models consistently underpredicted ozone and precursors at large distances (200–500 km) from the source regions, where relatively high ozone concentrations (150–180 ppb) were sometimes observed. The relative responses of the models to 50% reductions in anthropogenic VOC and NOx) emissions were also evaluated. The directions and magnitudes of the predicted ozone changes were roughly consistent for the models. The best agreement occurred for the 50% anthropogenic VOC emissions reductions, where all models predicted about a 15% decrease in peak ozone concentrations. When a 50% anthropogenic NOx) emissions reduction was imposed, UAM-IV and UAM-V/CALRAMS predicted about a 3 or 4% increase in peak ozone concentration and UAM-V/CALMET and the ROM/UAM-IV couple predicted about a 5–20% increase in peak ozone.