Sensitivity of mesoscale modeling of smoke direct radiative effect to the emission inventory: a case study in northern sub-Saharan African region

An ensemble approach is used to examine the sensitivity of smoke loading and smoke direct radiative effect in the atmosphere to uncertainties in smoke emission estimates. Seven different fire emission inventories are applied independently to WRF-Chem model (v3.5) with the same model configuration (excluding dust and other emission sources) over the northern sub-Saharan African (NSSA) biomass-burning region. Results for November and February 2010 are analyzed, respectively representing the start and end of the biomass burning season in the study region. For February 2010, estimates of total smoke emission vary by a factor of 12, but simulated regional (15°W–42°E, 13°S–17°N) and monthly averages of column PM2.5 loading differed by only a factor of 7. Simulations of surface PM2.5 concentration, aerosol optical depth (AOD), smoke radiative forcing at the top-of-atmosphere and at the surface, and air temperature at 2 m and at 700 hPa showed similar or smaller variations. The smaller differences in these simulated variables may reflect the atmospheric diffusion and deposition effects that dampen the large difference in smoke emissions that are highly concentrated in areas much smaller than the regional domain of the study. Indeed, for individual grid cells, large differences (up to a factor of 33) persist in simulated smoke-related variables and radiative effects including semi-direct effect. Similar results are also found for November 2010, despite differences in meteorology and fire activity. Hence, biomass burning emission uncertainties have a large influence on the reliability of model simulations of atmospheric aerosol loading, transport, and radiative impacts, and this influence is largest at local and hourly-to-daily scales. Conversely, the potential of atmospheric observations to constrain source magnitudes is greater at fine scales, and less effective for regional/seasonal averages. Accurate quantification of smoke effects on regional climate and air quality requires further reduction of emission uncertainties, particularly for regions of high fire concentrations such as NSSA.