Effect of varying mesoscale model grid spacing on dispersion model results

Mesoscale models are increasingly being used to provide input meteorological data for dispersion model analyses. Modelers often struggle with developing a mesoscale model configuration that produces the most representative meteorological data with the most efficient use of computer resources. Larger grid spacing (e.g. > 4 km in horizontal) is desirable to conserve computer resources but will not always produce adequate wind, temperature, and turbulence fields required by the dispersion models. Grid spacing is especially important in areas of non-homogeneous terrain.

We conducted a study to compare the effect of various mesoscale model grid spacing on dispersion model results. The Regional Atmospheric Modeling System (RAMS) was used in the study. RAMS was run with a nesting configuration where horizontal grid spacing was set at 48 km, 12 km, 4 km, and 1 km. The vertical spacing was 25-m near the surface on the inner grid and 100 m near the surface on the outer three grids. The dispersion model used in the study was the Short-range Layered Atmospheric Model (SLAM). SLAM is a trajectory puff model that uses the three-dimensional gridded wind fields from RAMS and other mesoscale models to produce trajectories and calculate the dispersion. The SLAM model uses the RAMS output parameters u- and v-wind component, temperature, dew point, pressure, precipitation, cloud fraction, friction velocity, surface roughness, surface heat flux and terrain/land use.

RAMS was run for an area of relatively smooth terrain and for an area of moderately rough terrain for a two-day period. The dispersion model was then run for these cases and the predicted concentrations were obtained. Ground level concentrations from the different model runs were compared.

The results indicate that each nest of RAMS, which is a two-way nested model, produced slight variations in the predicted wind fields. The trajectory and dispersion model predictions made from the data from these different nests were sensitive to the varying wind fields. Comparison of the location and magnitudes of the predicted ground-level concentrations are presented in this paper. Comparisons with field study observations are also included.