Comparison of Two Transport Descriptors within a Hybrid Receptor Modeling System

ENSCO is developing systems for imputing source locations using transport and dispersion modeling in combination with sample records of pollutant concentrations. These data are analyzed within the general framework commonly referred to as hybrid receptor modeling in which a statistic (an estimate of conditional probability) is calculated for each grid square based on the relative strength of association between the passage of air parcels over that grid square and the arrivals of those air parcels at the receptor during times of elevated pollutant concentrations.

Conditional probability surfaces can be estimated using both forward and backward descriptors of transport. In both approaches, we use a statistical bootstrapping algorithm to obtain confidence intervals on the conditional probability estimates. In the forward-transport approach, the conditional probability field is estimated from simulated airborne concentrations at a receptor arising from hypothetical plumes arriving from each point of a grid of hypothetical sources. The simulated airborne concentrations are produced by the transport and dispersion model SLAM, which is a Lagrangian Gaussian-puff and trajectory model that can ingest a wide variety of meteorological data and formats. In the backward modeling approach, the conditional probabilities are estimated from frequency counts of backward-trajectory segment endpoints in the various cells of the grid. When used with backwards trajectories, it is similar to Hopke's PSCF model used with bootstrapping (Hopke et al., 1995).

We have used the new model in both modes to analyze airborne concentration and trajectory data over the Southwestern United States during the late 1980's using data from various projects which investigated sources of haze in Grand Canyon National Park. These results are compared and contrasted with previously published results.

Hopke, P. K. , Li, C. L., Landsberger, S. 1995. The use of bootstrapping to estimate conditional probability fields for source locations of airborne pollutants, Chemometrics and Intelligent Laboratory Systems 30, 69-79.