Identifying the geographic factors that control feedbacks between soil, climate, and biota is one of the oldest lines of ecological inquiry; however, until recently this inquiry has involved extrapolation from individual study sites. Currently, large spatial databases of watershed fluxes, soil geochemical properties, forest inventories, and remotely sensed vegetation indices can reveal feedbacks between ecosystems, soil nutrient status, and climate. This session addresses ways in which large geographic datasets can be combined with statistical and process-based modeling approaches to evaluate continental- to global-scale feedbacks between soil, climate, and vegetation. We pose two main questions: (1) how can spatial datasets be used to test geographic hypotheses that link soil properties, climate, and ecosystem function? (2) What modeling approaches (e.g. statistical versus process-based) are best suited to integrating spatial data with theory while overcoming fundamental limitations, including scale mismatch, missing information, and spatial bias? This session welcomes perspectives from early-career scientists across disciplines who apply novel approaches to spatial datasets, testing hypotheses at the intersection of ecological, geological, and hydrologic sciences.