Modeling dispersion at distances of meters from urban sources

Abstract This paper describes the evaluation and improvement of dispersion models for estimating ground-level concentrations in the vicinity of small sources located in urban areas. The models were evaluated with observations from a tracer study conducted at the University of California, Riverside. This experiment simulated a non-buoyant release from the top of a small source in an urban area. The tracer, SF6, was sampled at several receptors within 20 m from the source. Several receptors were located upwind of the dominant westerly wind direction. Model estimates from ISC-PRIME and AERMOD-PRIME were evaluated with hourly observed concentrations. The evaluation indicated that the highest concentrations were overestimated by these models. At the same time, the lower range of concentrations was underestimated. A diagnostic study with a simple Gaussian dispersion model that incorporated site specific meteorology indicated that these errors could be reduced by accounting for the lateral meandering caused by increased horizontal turbulence in urban areas. While AERMOD incorporates lateral meandering, it switches it off in the near field affected by PRIME estimates. This study found that using onsite turbulence information in a simple model for meandering can lead to adequate estimates of observed concentrations even when downwash effects are not modeled explicitly.