Statistical Performance of Several Mesoscale Atmospheric Dispersion Models

Abstract

Seventeen mesoscale dispersion models were statistically evaluated with data from krypton-85 emissions from the Savannah River Plant, Aiken, SC. Widely accepted statistical parameters were used for the statistical evaluation. The models were able to predict concentrations more accurately for longer periods of time (yearly) than for shorter periods (twice-daily). The correlations between predictions from different models were approximately the same as correlations between predictions and measurements.