The recent widespread availability of sub-meter satellite imagery has revolutionized multiple disciplines in the geosciences. We developed an automated, open-source workflow to generate 2-m digital elevation models (DEMs) from commercial stereo imagery, with <0.5-1.0 m horizontal and vertical accuracy after co-registration. We are producing 1000s of DEMs in a high-performance computing environment, and assembling dense time series and regional mosaics for West Antarctica, High Mountain Asia (HMA), and the contiguous US (CONUS). In Antarctica, these DEMs provide precise measurements of the spatial and temporal evolution of grounding zones, sub-shelf ocean cavity geometry, and basal melt rates beneath vulnerable ice shelves. We document connections between these observables and upstream ice sheet dynamics, and evaluate melt rate parameterizations used in prognostic ice flow models to constrain future sea level rise. We are also using these DEM records to study the long-term (~50-year), decadal, interannual, and seasonal mass balance of mountain glaciers on a continental scale. These products provide a systematic assessment of regional climate change, and offer basin-scale assessments of snow water equivalent and snow/ice melt runoff contributions for downstream water resource applications. In addition, these observations offer measurements of dynamic landscape evolution (e.g., landslides, sediment redistribution by rivers) that can be used for both hazard assessment and rapid response to natural disasters. The sub-meter commercial stereo image archive spans nearly 10 years, and ongoing scientific data acquisition efforts have expanded dramatically since ~2013. These resources offer a nearly global, on-demand, low-cost capability that approaches the quality of airborne lidar data, and they will undoubtedly be an essential component of future Earth observing programs.