Modeling and analysis of spectrum demand in large scale urban regions is key to effective Dynamic Spectrum Access (DSA) in the next generation networks. Some of the recent work on characterizing cellular traffic has been done by empirical analysis of proprietary cellular provider data, or by the use of stochastic processes that can match aggregate characteristics of the real networks. However, these approaches have limited use in DSA and network planning in realistic and dynamic settings. In this paper, the authors describe a first principles based modeling environment for spectrum demand and mobile call graphs at a highly detailed spatio-temporal scale. Their approach involves the integration of a diverse set of public and commercial datasets and computational models into a common architecture for data exchange.