Technical Abstract: To accurately simulate watershed hydrology after forest harvest using SWAT, it is important to understand the factors that potentially make certain sites more sensitive to disturbance. The growth model in SWAT has been modified to provide a more precise description of forest growth dynamics, by integrating a multi-species growth model, ALMANACBF. We observed significant differences in percent cover of vegetation, biomass and leaf area index among sites representing differing pre-harvest forest stands and associated with topography, between hummocks and depressions. Our results demonstrate that on the hummocky terrain of the Boreal Plain, there are important variations in vegetation recovery after disturbance among different ecosites and among landscape positions. LIDAR imagery and targeted sampling of obvious landscape variations was effective in correcting vegetation cover estimates to account for differences in cover observed between depressions and hummocks in a variety of sites.