Evaluation of the HPAC dispersion model with the Joint Urban-2003 (JU2003) tracer observations

Results are presented of a comprehensive evaluation of the Hazard Prediction and Assessment Capability (HPAC) model, using data from the Joint Urban 2003 (JU2003) Field Experiment in Oklahoma City. This paper focuses on Intensive Operating Periods (IOPs) 3 through 10, where the first four are during the daytime and the last four are during the nighttime. There were three separate continuous releases (of 30 minute duration) of SF6 tracer each day from a point source near ground level, in or immediately upwind of the built-up downtown area. Tracer was sampled from over 100 locations at distances ranging from 0.1 to 4 km from the source. Two alternate urban configurations of HPAC are tested using four optional meteorological inputs. The two urban configurations are the Urban Dispersion Model (UDM) model and the Urban Canopy (UC) model. The four meteorological input options are basic default National Weather Service data (BDF), a single averaged wind over the domain (SGL), an upwind anemometer and radiosonde (UPWIND), and detailed three-dimensional winds from a mesoscale meteorological model, MM5 (MEDOC). A large number of types of model outputs are being evaluated, including dosages and 15, 30, and 60-minute averaged concentrations. Maximum concentrations on downwind distance arcs are being evaluated as well as concentrations paired in time and space at all samplers. This paper gives results for the maximum 30-minute averaged concentrations at six downwind distance arcs. It is found that the HPAC configuration using UDM and using the MM5 meteorological inputs (MEDOC) performed better than the others, with a Fractional Bias of 0.06 and a Normalized Mean Square Error (NMSE) of about 3.0. The other options tended to have more overpredictions during the night and underpredictions during the day.