Economic geography can be viewed as a large and growing network of interacting activities. This fundamental network structure and the large size of such systems makes the complex network approach an attractive model for growth dynamics modeling. In this paper the authors propose the use of complex networks for geographical modeling and demonstrate how such an application can be combined with a cellular model to produce output that is consistent with large-scale regularities such as power laws and fractality. Complex networks can provide a stringent framework for growth dynamic modeling where concepts from, for example, spatial interaction models and multiplicative growth models, can be combined with the flexible representation of land and behaviour found in cellular automata and agent-based models. In addition, there exists a large body of theory for the analysis of complex networks that have direct applications in urban geographic problems. The intended use of such models is twofold: (1) to address the problem of how the empirically observed hierarchical structure of settlements can be explained as a stationary property of a stochastic evolutionary process rather than as equilibrium points in a dynamic process, and, (2) to improve the predictive quality of applied urban modeling.