A fundamental problem in homeland security is predicting transport and dispersion of a chemical, biological, nuclear, or radiological (CBNR) release. The current transport and dispersion models require specific information about the source and about the local meteorological conditions to make accurate predictions. Sometimes these data are not readily available or measurements are sparse. There are several techniques of data assimilation that may prove applicable in this problem. Although data assimilation is a very developed field in NWP, it has not been specifically applied to transport and dispersion models. In NWP some of the frequently used techniques include optimal interpolation, 3D-Var, extended Kalman filtering, ensemble Kalman filtering, 4D-Var, and data nudging. These methods have to be newly evaluated in the context of transport and dispersion modeling. Two aspects of dispersion assimilation make it different from NWP: 1. the observations are likely to be sparser and 2. the contaminant plume is localized. Assumptions and the specifics of the techniques in data assimilation have been tuned for NWP. The impact of these differences on the implementation and success of traditional data assimilation techniques is assessed through a series of numerical experiments.