Buoyancy fluxes and glacial melt rates at vertical ice-ocean interfaces are commonly parameterized using theories derived for unbounded free plumes. A Large Eddy Simulation is used to analyze the disparate dynamics of free plumes and wall-bounded plumes; the distinctions between the two are supported by recent theoretical and experimental advances and demonstrate that unbounded plume theory does not adequately represent plume/boundary layer dynamics at ice-ocean interfaces. Modifications to parameterizations consistent with these simulations are tested and compared to results from numerical and laboratory experiments of meltwater plumes. These modifications include 50\% weaker entrainment and a distinct plume-driven friction velocity in the shear boundary layer up to 8 times greater than the externally-driven friction velocity. Using these modifications leads to 40 times the ambient melt rate predicted by commonly used parameterizations at vertical glaciers faces, which is consistent (and necessary for consistency) with observed melt rates at LeConte Glacier, Alaska.