An improved dust emission model – Part 2: Evaluation in the Community Earth System Model, with implications for the use of dust source functions

Abstract. The complex nature of mineral dust aerosol emission makes it a difficult process to represent accurately in weather and climate models. Indeed, results in the companion paper indicate that many large-scale models underestimate the dust flux's sensitivity to the soil's threshold friction velocity for erosion. We hypothesize that this finding explains why many dust cycle simulations are improved by using an empirical dust source function that shifts emissions towards the world's most erodible regions. Here, we both test this hypothesis and evaluate the performance of the new dust emission parameterization presented in the companion paper. We do so by implementing the new emission scheme into the Community Earth System Model (CESM) and comparing the resulting dust cycle simulations against an array of measurements. We find that the new scheme shifts emissions towards the world's most erodible regions in a manner that is strikingly similar to the effect of implementing a widely used source function based on satellite observations of dust source regions. Furthermore, model comparisons against aerosol optical depth measurements show that the new physically based scheme produces a statistically significant improvement in CESM's representation of dust emission, which exceeds the improvement produced by implementing a source function. These results indicate that the need to use an empirical source function is eliminated, at least in CESM, by the additional physics in the new scheme, and in particular by its increased sensitivity to the soil's threshold friction velocity. Since the threshold friction velocity is affected by climate changes, our results further suggest that many large-scale models underestimate the global dust cycle's climate sensitivity.