We are currently witnessing a leap in provisioning wireless networks using satellites, enabled by the recent reduction in the launching costs and the advent of the Internet-of-Things. Understanding the path-loss between satellites and ground users is vital for the proper planning and design of such networks. This paper presents a systematic framework for modeling satellite-to-ground path-loss in urban environments. The modeling approach is semi-analytic, where the line-of-sight probability is obtained using tractable tools from stochastic geometry, while the shadowing is captured using a Gaussian mixture model that can be trained using measurements collected from global navigation satellite system (GNSS) receivers. The presented modeling framework balances simplicity and accuracy as illustrated with the results of the conducted measurements.