An accidental or intentional release of hazardous chemical, biological, radiological, or nuclear material into the atmosphere obligates responsible agencies to model its transport and dispersion in order to mitigate the effects. That modeling requires input parameters that may not be known and must be estimated given sensor measurements of the resulting concentration field. The genetic algorithm (GA) method used here has been successful in back-calculating source characteristics and meteorological parameters necessary to predict the contaminant transport and dispersion. Previous validation studies utilized identical twin experiments wherein the synthetic validation data were created using the same transport and dispersion models used for the back-calculations. To evaluate different models a database of concentration fields was needed. For that reason, the Fusing Sensor Information from Observing Networks (FUSION) Field Trial (FFT 07) was conducted to evaluate sensor data fusion algorithms under controlled environmental conditions. This study extends the validation by using instead synthetic data generated with a time-dependent computational fluid dynamics (CFD) model, the same model that was used to configure the setup for FFT07. Such data inherently include time-dependent behavior unique to each contaminant episode rather than the ensemble average predicted by the transport and dispersion model used in previous studies. The genetic algorithm coupled with the sheared Gaussian puff model then back-calculates both source characteristics and mean meteorological data. The solution is optimized by the GA and further tuned with the Nelder-Mead downhill simplex algorithm, known as the Hybrid GA. We also use the Hybrid GA with preliminary data from FFT07 to characterize and estimate the source.