Comparison of the IFDM building downwash model predictions with field data

The building downwash parameterization for plume models proposed in Cosemans et al. (2012) is integrated in a plume model (IFDM, Immission Frequency Distribution Model) and the predictions of this model are compared with real-life field data. The field data are time series for four monitoring sites of daily arsenic (As) concentrations measured over a two year period (Jan. 2010-Dec. 2011) near a plant that emits some As. In-stack emission measurements as well as measured As concentrations in ambient air show considerable time-dependent variations which prohibited a straightforward comparison of measured and computed concentrations using average emissions. Therefore, two model evaluations have been done. First, the time-variability was considered small enough during twenty months (out of the 24 months in the two-year period) to compare the central moving averages of the measured concentrations with those computed using constant emissions. Time series compare well, and none of the measured annual average concentrations has a bias larger than 50% while 7 out of 8 year averages (4 monitoring sites, two years) have a bias lower than 30%. During the four months not used in this evaluation, measured daily concentrations are about four times higher than during the other twenty months, resulting in yearly average increase up to 80%. Using a time-varying emission scenario, using information from daily As-measurements in the recirculation zone on top of the building with the sources, the model is able to reproduce the annual mean concentrations at the different locations including all the months. Finally, not only the yearly averages, but also the time series of measured and computed concentrations, pollutant roses and Q-Q plots are evaluated.