A model-data inversion is applied to an extensive observational dataset collected in the Southern Ocean north of the Ross Sea during late autumn to early winter, producing estimates of the frequency-dependent rate of dissipation by sea ice. The modeling platform is WAVEWATCH III® which accounts for non-stationarity, advection, wave generation, and other relevant processes. The resulting 9477 dissipation profiles are co-located with other variables such as ice thickness to quantify correlations which might be exploited in later studies to improve predictions. An average of dissipation profiles from cases of thinner ice near the ice edge is fitted to a simple binomial. The binomial shows remarkable qualitative similarity to prior observation-based estimates of dissipation, and the power dependence is consistent with at least three theoretical models, one of which assumes that dissipation is dominated by turbulence generated by shear at the ice-water interface. Estimated dissipation is lower closer to the ice edge, where ice is thinner, and waveheight is larger. The quantified correlation with ice thickness may be exploited to develop new parametric predictions of dissipation.