Modeling the "human terrain" is important for effective training of non-kinetic missions in urban terrain. It is equally important for planning responses to epidemics of infectious diseases, where combinations of vaccines, antiviral drugs, and social distancing are the tools to be employed. Social distancing methods include quarantine; closing schools, workplaces, and community centers; and travel restrictions. Social distancing methods are relevant to both non-kinetic missions and responses to epidemics.

This paper describes the development of behavior models for large-scale agent-based models to support planning for responses to potential epidemics such as avian flu (Cooley et al., 2008). In these models, each person in an urban area is represented as an agent. Models have been constructed for several municipal areas, including Chicago, Illinois (Ferguson, Longini); Portland, Oregon (Eubanks); Pittsburgh, Pennsylvania (Cooley): and the Research Triangle Park, North Carolina (Cooley). These behavior models must represent the interaction patterns of the people in the area being modeled. This requires fusing data from multiple sources about the social networks and movement patterns of the population. This process starts with an allocation of agents to the households in the area that is consistent with available demographic information including age, gender, and socio-economic status. Algorithms assign these agents to their schools, workplaces, shopping malls, and hospitals in ways that are consistent with available source data. Additional algorithms estimate daily travel patterns and interaction times at these locations based on these assignments and additional source data on travel times.