Beavers are of increasing interest as a mechanism for ecosystem restoration. However, the physical watershed characteristics that allow for successful beaver re‐establishment are not consistent across habitat suitability models. We delineated the extent of beaver impounded surface water in the Pole Mountain Area (PMA) of south‐eastern Wyoming with six National Agricultural Imagery Program acquisitions over an 18‐year time period via a random forest classification algorithm. We computed beaver pond area for first‐ and second‐order watersheds within the PMA and related the impounded pond area to climatic metrics such as maximum annual snow water equivalent and snowmelt and rain input into the ground. There was no apparent relationship between annual climate metrics and beaver pond area. Further, we developed a random forest regression model to correlate beaver pond area with topographically derived watershed characteristics (e.g. hillslope size and riparian area size) to explore physical controls on impounded area at the watershed scale. Controls differed for first‐ and second‐order watersheds, with pond area in smaller watersheds largely being controlled by lateral water contributions from the hillslopes and in the larger watersheds by water from upstream and available space in the riparian areas. Interestingly, vegetation height did not emerge as an important predictor for pond area, likely because of generally good availability of forage material in all watersheds. This study improves our understanding of watershed characteristics that contribute to beaver habitat suitability and can be used to guide beaver reintroduction efforts through analysis targeted at identifying watersheds most appropriate to beaver pond construction.