Abstract
We compared observed and model-simulated surface concentration on a spatial wale of 1100 km and a temporal scale of 36 h. The Eulerian tracer model calculated advection by the mean winds and gradient transport for subgrid-scale turbulent transfer. The simulations were evaluated by the maximum concentrations, spatial correlations between concentrations, plume sizes, and trajectory errors. We examined the following inputs and model parameters: 1) different meteorologies, including simulated, observed, and combinations of the two; 2) the spatial and temporal resolution of the observations; 3) the spatial resolution of the meteorological model; and 4) the spatial resolution of the tracer model. The best meteorology was observed horizontal winds, enhanced with additional, nonstandard rawinsondes and model-simulated eddy diffusivities. The spatial resolution of the tracer model was more important than that of the meteorological model. Meteorology from a mesoscale model could be competitive with that from standard observations. We found consistency between different measures of tracer simulation quality.
