Previously, we have modeled radar signatures, involving large (-10-15 dI3) variations in radar cross-section (RCS), that have been observed at strongly convergent Ocean fronts and at an estuarine front. In each of these cases, we obtained quantitative agreement with measurement but only by including wave-breaking (WB) effects in an approximate manner. However, in each case, we used the composite scattering (CS) model at a frequency, where this model may be deficient. For this reason, questions remain concerning the importance of WB effects in these simulations. In the present study, we monitor the sensitivity of the simulations with respect to this CS model approximation by comparing the results from three different radar model simulations of RCS, derived from a common wave spectrum. The spectrum is calculated using a full-spectral treatment of wave-current interaction. The resulting simulations are used to model the radar signature of the buoyant plume associated with the efflux of fresh water from the Chesapeake Bay that was observed during the COPE-2 experiment. In each case, it is possible to simulate this signature, in quantitative agreement with experiment, but only by including WB effects. However, CS model predictions for the behavior of the signature do not agree with the comparable predictions from the two remaining models. Additional simulations indicate these differences result from higher order terms that are not included in the Kirchoff approximation.