Technical Abstract: Vegetation mapping via remote sensing tools is a useful and cost-effective approach for understanding current land use and species habitat patterns, but has limited utility in predicting the future state of vegetation. We describe an approach that couples data-supported, conceptual state-and-transition models with soil-geomorphic mapping and remote sensing of vegetation states to provide maps of potential vegetation and their likely future state under different management scenarios. Soil maps are critical components of this approach that are often unattainable, so we describe an approach to classifying geomorphology-based land types based on landforms and remote-sensed imagery in arid to semiarid zones. The resulting maps reflect management hazards and opportunities for restoration that are not addressed in conventional vegetation maps.