"June 1998." Includes bibliographical references (pages 49-53).

This report summarizes a verification of the SCIPUFF model as described in the draft report "PC-SCIPUFF Version 0.2 Technical Documentation" by Sykes et al. (1996, ARAP Report 712, Titan Corp). The verification included a scientific review of the model physics and parameterizations described in the report, and checks for their internal usage and consistency with current practices in atmospheric dispersion modeling. This work is intended to examine the scientific basis and defensibility of the model for the intended application. A related task is an assessment of the model's general capabilities and limitations. A line-by-line verification of the computer source code was not possible; however, the code was checked with a commercial code analyzer. About 4 7 potential coding inconsistencies were identified. The sensitivity of SCIPUFF to changes in input parameters was examined. These tests indicated that SCIPUFF behaved in a rational way to these changes. Comparisons of SCIPUFF output with time-averaged tracer gas plumes showed that, overall, SCIPUFF can be said to overpredict by less than a factor of two, but there is a greater than factor of two scatter between individual pairs of observed and predicted values