Accurate prediction of snowmelt contribution to river discharge is essential for managing water resources in the Western US. It is well known that dust deposition events during spring ablation rapidly reduce albedo of the snowpack surface, and strongly influence the timing and magnitude of subsequent melt. Albedo is therefore a critical parameter in snowpack energy balance modeling, and is not well constrained by current modeling efforts. This is particularly relevant for the San Juan Mountains in Southwest Colorado, the first high-elevation terrain to intercept dust from the Colorado Plateau, and the southern headwaters of the Colorado River. In this region, snow darkening has been linked to snowmelt forecasting errors from operational temperature index based methods. Here, we assess the utility of representing snow albedo in a physically-based spatially distributed snowpack energy balance model. The model, iSnobal, was executed within the Automated Water Supply Model (AWSM) framework and run over two San Juan mountain watersheds, the Animas and Dolores, at 50m spatial resolution. The model ingests forcing variables from the High Resolution Rapid Refresh (HRRR) atmospheric model, and the net solar radiation is adjusted with snow albedo from MODIS. Model inputs and outputs are assessed against in situ observations and high resolution airborne maps of snow depth and albedo. This work shows that iSnobal misspecifies the timing of spring melt without realistic albedo values, and demonstrates the need to improve albedo representation in snow energy balance modeling applications throughout the cryosphere.