We use the Global Modeling Initiative (GMI) chemistry and transport model with Modern-Era Retrospective Analysis for Research and Applications (MERRA) meteorological fields to quantify heterogeneous chemical ozone loss in Arctic winters 2005-2015. Comparisons to Aura Microwave Limb Sounder N2O and O3 observations show the GMI simulation credibly represents the transport processes and net heterogeneous chemical loss necessary to simulate Arctic ozone. We find that the maximum seasonal ozone depletion varies linearly with the number of cold days and with wave driving (eddy heat flux) calculated from MERRA fields. We use this relationship and MERRA temperatures to estimate seasonal ozone loss from 1993-2004 when inorganic chlorine levels were in the same range as during the Aura period. Using these loss estimates and the observed March mean 63-90°N column O3, we quantify the sensitivity of the ozone dynamical resupply to wave driving, separating it from the sensitivity of ozone depletion to wave driving. The results show that about 2/3 of the deviation of the observed March Arctic O3 from an assumed climatological mean is due to variations in O3 resupply and 1/3 is due to depletion. Winters with a stratospheric sudden warming (SSW) before mid-February have about 1/3 the depletion of winters without one and export less depletion to the midlatitudes. However, a larger effect on the spring midlatitude ozone comes from dynamical differences between warm and cold Arctic winters, which can mask or add to the impact of exported depletion.