In this study we use the NASA GEOS-5 Earth System Model (ESM) and Data Assimilation System (DAS) to examine the direct and semi-direct impacts of absorbing biomass burning (smoke) aerosol over southern Africa. We perform an ensemble of short-term (5 day) forecasts during the peak of the burning season (August). Constrained by initial conditions imposed by the assimilation, the short-term nature of the forecast serves to separate the direct effects of aerosol absorption (and atmospheric heating) from the changes in circulation that can accrue over longer free-running GCM simulations. We are thus able to more accurately estimate direct aerosol forcing and resulting changes in cloud radiative forcing, a measure of the semi-direct effect.
One of the unique features of the GEOS-5 DAS is the Incremental Analysis Update (IAU) technique in which the observational constraint is introduced gradually, preventing shocks in model physical parameterizations. The IAU is defined as the difference between the model forecast and analysis in a data assimilation cycle. It represents the complex combination of all model errors due to inadequate representation of physical processes (model parameterizations), numerical errors, and processes that have been omitted from the model. The difference in the IAU generated in model run with and without biomass burning aerosols, or with varying aerosol parameters, is a useful metric with which to assess the effect of aerosols on climate and cloud distributions. We perform a series of sensitivity simulations in which we vary biomass burning aerosol properties (e.g. increased emissions, increased absorption), and we examine the impact of changes in these properties on clouds and the IAU. The set of aerosol properties which tend to minimize the IAU (i.e. the model error) and agree with available observations (e.g. from satellite sensors) are expected to yield the best estimate of both the biomass burning direct and semi-direct effects. Based on these sensitivity studies, we present our best estimate of the aerosol direct radiative forcing and the aerosol semi-direct effect, quantified as a difference in cloud radiative forcing between a simulation with biomass burning aerosols and one without.