Climatic or environmental change is not only driving distributional shifts in species today, but it has also caused distributions to expand and contract in the past. Inferences about the geographic locations of past populations, especially regions that served as refugia (i.e., source populations) and migratory routes are a challenging endeavor. Refugial areas may be evidenced from fossil records or regions of temporal stability inferred from ecological niche models. Genomic data offer an alternative and broadly applicable source of information about the locality of refugial areas, especially relative to fossil data, which are either unavailable or incomplete for most species. Here we present a pipeline we developed (called x-origin) for statistically inferring the geographic origin of range expansion using a spatially explicit coalescent model and an Approximate Bayesian Computation testing framework. In addition to assessing the probability of specific latitudinal and longitudinal coordinates of refugial or source populations, such inferences can also be made accounting for the effects of temporal and spatial environmental heterogeneity, which may impact migration routes. We demonstrate x-origin with an analysis of genomic data collected in the Collared pika that underwent post-glacial expansion across Alaska, as well as present an assessment of its accuracy under a known model of expansion to validate the approach.