In case of a chemical, biological, radiological, or nuclear (CBRN) release, emergency managers need to draw on all the available data to create the most accurate forecast possible. Traditionally a numerical weather prediction (NWP) model provides the bridge between raw meteorological observations and the forcing for a transport and dispersion model. This approach has two shortcomings: sparse meteorological observations can lead to inaccurate wind fields and the source information of the contaminant can be either unavailable or inaccurate. These two problems can be addressed together by assimilating concentration and wind observations into a coupled NWP and dispersion model system. Assimilating wind observations into the model is straightforward using traditional data assimilation methods such as a Kalman Filter, Nudging, or 4D-Var. Two aspects of the assimilation of concentration data into the model complicate the process however: 1) the observations are likely to be sparser because the contaminant is localized, and 2) the dispersing concentration field is not coupled fully to the meteorology. In particular, weather affects concentration, but the reverse is not true. The impact of these differences on the implementation and success of traditional data assimilation techniques is assessed through a series of numerical experiments with a two-dimensional shallow water equation forced dispersion model. The results are compared and data needs are considered.