Fluid Modeling and the Evaluation of Inherent Uncertainty

Abstract

Atmospheric dispersion models are widely used to estimate the impact of nonreactive pollutant releases into the atmosphere. Most such models predict the ensemble-average concentration field, the average over a large collection of releases under similar externally imposed conditions. The stochastic nature of pollutant dispersion in the atmospheric boundary layer (ABL) makes pollutant concentrations vary greatly from one sampling period to another under the same meteorological conditions, however. Therefore, if the sampling time is not sufficiently long, the values predicted by even a perfect model will inevitably differ from mean values measured in the atmosphere. The ratio of their root-mean-square difference and the ensemble-average concentration is called the inherent uncertainty. This work investigates the relationship between inherent uncertainty in laboratory and ABL flows. This relationship provides a way to use laboratory measurements to estimate the inherent uncertainty in ABL flows. For a given averaging time, it is shown that the inherent uncertainty in laboratory flows is smaller than in the ABL under the same stability and statistical conditions. This result is illustrated with the Willis?Deardorff convection tank data to show that the values of the inherent uncertainty in pollutant dispersion from a near-surface source in the ABL exceed 50% for an averaging time on the order of 1 h.