Evaluation of HPAC-Urban and JEM with U2000, JU2003, MSG05, and MID05 tracer observations

An overview is given of our experience over the past two years in evaluating HPAC-Urban and JEM with tracer data from the major U.S. urban field experiments over the past ten years in Salt Lake City (Urban 2000 or U2000), Oklahoma City (Joint Urban 2003 or JU2003), and Manhattan (Madison Square Garden 2005 or MSG05 and Midtown 2005 or MID05). HPAC-Urban includes SCIPUFF with the Urban Canopy (UC), Urban Dispersion Model (UDM), and MicroSwift-Spray (MSS) urban module options; and JEM includes only the UC and UDM options. In each set of evaluations, a few likely options are tested for meteorological inputs (e.g., closest NWS airport site, representative downtown building rooftop, upwind vertical profile from remote sounder, NWP outputs). The paper summarizes the results of the evaluations, including use of model acceptance criteria in the JEM evaluations. In general we find that there is not a single urban model option or a single type of meteorological input that provides clear improvements over any other. For maximum concentrations on downwind distance arcs, most of the time the relative mean bias has a magnitude less than 0.6 and the relative RMSE is less than about three.