The effect of diffusion on tracer puffs simulated by a regional scale Eulerian model

Abstract

On a spatial scale of up to 1,100 km and a temporal scale of up to 36 hours, we used Eulerian models to simulate the transport and diffusion of a passive tracer released from a point source near the surface. An Eulerian tracer model was driven by meteorological analysis of 6‐hour rawinsonde data supplemented by the simulations of an Eulerian mesoscale meteorological model. The tracer model simulations were compared with 6‐hour integrated samples collected at 86 surface stations during the Cross‐Appalachian Tracer Experiment (CAPTEX) for six different releases. We examined the effect of the parameterizations of vertical and horizontal sub‐grid scale turbulent transport and the finite difference advection scheme on the simulations of concentration and the trajectories of the center of mass of the tracer at the surface. We found that the simulated trajectory error was sensitive to the vertical eddy diffusivity but insensitive to the advection scheme and hence insensitive to (horizontal) numerical diffusion. Using 70‐km horizontal resolution, only the most sophisticated advection scheme tested here (Prather, 1986) simulated a tracer puff with a horizontal area that was not too large during the first 36 hours after a point source release. The other schemes tested (Russell and Lerner, 1981; a simplified version of Smolarkiewicz, 1983) had enough numerical diffusion that the simulated horizontal tracer puffs were larger than the observed puffs. There were considerable differences in the capability of the tracer models to simulate individual releases, particularly for quantitative comparisons of simulated and observed concentrations.