An important challenge for seismic monitoring of nuclear explosions at low magnitude to verify a nuclear-test-ban treaty is the development of techniques that use regional phases for detection, location, and identification. In order to use such phases, region-specific earth models and tools are needed that accurately predict features such as travel times, amplitudes, and spectral characteristics. In this paper, we present our efforts to use two-dimensional finite-difference modeling to help develop and validate regional earth models for the Middle East and North Africa and to develop predictive algorithms for identifying anomalous regional phases. To help develop and validate a model for the Middle East and North Africa, we compare data and finite-difference simulations for selected regions. We show that the proposed three-dimensional regional model is a significant improvement over standard one-dimensional models by comparing features of broadband data and simulations and differences between observed and predicted features such as narrow-band group velocities.