The dispersion of particles from wildfires, volcanic eruptions, dust storms, and other aerosol sources can affect many environmental factors downwind, including air quality. Aerosol injection height is one source attribute that mediates downwind dispersion, as wind speed and direction can vary dramatically with elevation. Using plume heights derived from space-based, multi-angle imaging, we examine the impact of initializing plumes in the NOAA Air Resources Laboratory's Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model with satellite-measured vs. nominal (model-calculated or VAAC-reported) injection height on the simulated dispersion of six large aerosol plumes. When there are significant differences in nominal vs. satellite-derived particle injection heights, especially if both heights are in the free troposphere or if one injection height is within the planetary boundary layer (PBL) and the other is above the PBL, differences in simulation results can arise. In the cases studied with significant nominal vs. satellite-derived injection height differences, the HYSPLIT model can represent plume evolution better, relative to independent satellite observations, if the injection height in the model is constrained by hyper-stereo satellite retrievals.