Brown carbon (BrC) is an absorbing organic aerosol, primarily emitted through biomass burning (BB), with a large degree of variability in observed chemical and microphysical properties. This makes model representation of the species difficult. Retrievals of BrC from measured radiance fields have the potential to constrain model schemes and improve estimates of BrC radiative effects. We used a retrieval of BrC optical depth and mass from AERONET sites in BB regions to evaluate GISS ModelE Earth system model’s BrC module. We approached this evaluation by comparing BrC properties defined in ModelE with underlying properties prescribed by the AERONET retrieval, then constrained our scheme to match those assumptions. Our analysis showed that, just with this initial harmonization, model bias was decreased, and performance, relative to the retrieval, was improved. This allowed for an estimate of global average BrC radiative effect, 0.03 Wm-2, grounded in the measured radiance fields represented by the retrieval. Through this work, we demonstrate the necessity of harmonizing model scheme parameters with a speciated retrieval’s assumptions, even if such assumptions are not necessarily more physically correct, to ensure an “apples-to-apples” comparison and ultimately improve the estimate of individual aerosol direct effects within a climate model.