In order to understand the dynamics of emergence and spreading of socio-technical innovations and population moves it is important to determine the place of origin of these populations. Here we focus on the role of geographical factors, such as land fertility and mountains in the context of human population evolution and distribution dynamics. We use a constrained diffusion-based computational model, computer simulations and the analysis of geographical and land-quality data. Our analysis shows that successful human populations, i.e. those which become dominant in their socio – geographical environment, originate from lands of many valleys with relatively low land fertility, which are close to areas of high land fertility. Many of the homelands predicted by our analysis match the assumed homelands of known successful populations (e.g. Bantus, Turkic, Maya). We also predict other likely homelands as well, where further archaeological, linguistic or genetic exploration may confirm the place of origin for populations with no currently identified urheimat. Our work is significant because it advances the understanding of human population dynamics by guiding the identification of the origin locations of successful populations.