Modeling and optimizing radiological threat detection problems in urban areas is facilitated through prior characterization of background activity distributions. The background emissivity of static objects such as buildings and roads is highly dependent on the composition of these elements of urban scenes. We perform a feasibility study to determine whether truck-based (non-imaging) detector measurements can be used to estimate the relative contributions of potassium-40, uranium and thorium (KUT) to the background emissions of surfaces. To do this, we simulate a simple urban scene, and assign realistic KUT compositions to structures. The detectors on the truck are completely characterized in terms of spectral response vs. angle of incidence using Monte Carlo simulation of all three background components, with each component being expressed as a spectral basis function. The truck moves along a trajectory through this environment and measures the background activity. Assuming knowledge of the position of surfaces (as would be determined using 3D maps or optical survey data), we estimate the KUT composition. We show that the inverse problem is sufficiently well-conditioned to solve for the K-40 component, but not for reconstructing the much smaller relative activities due to U and Th.