First forcing estimates from the future CMIP6 scenarios of anthropogenic aerosol optical properties and an associated Twomey effect

We present the first forcing interpretation of the future anthropogenic aerosol scenarios of CMIP6 with the simple plumes parameterisation MACv2-SP. The nine scenarios for 2015 to 2100 are based on anthropogenic aerosol emissions for use in CMIP6 (Riahi et al., 2017; Gidden et al., 2018). We use the emissions to scale the observationally informed anthropogenic aerosol optical properties and the associated effect on the cloud albedo of present-day (Fiedler et al., 2017; Stevens et al., 2017) into the future. The resulting scenarios in MACv2-SP are then ranked according to their strength in forcing magnitude and spatial asymmetries for anthropogenic aerosol. All scenarios, except SSP3-70 and SSP4-60, show a decrease in anthropogenic aerosol by 2100 with a range from 108% to 36% of the anthropogenic aerosol optical depth in 2015. We estimate the radiative forcing of anthropogenic aerosol from high- A nd lowend scenarios in the mid-2090s by performing ensembles of simulations with the atmosphere-only configuration of MPIESM1.2. MACv2-SP translates the CMIP6 emission scenarios for inducing anthropogenic aerosol forcing. With the implementation in our model, we obtain forcing estimates for both the shortwave instantaneous radiative forcing (RF) and the effective radiative forcing (ERF) of anthropogenic aerosol relative to 1850. Here, ERF accounts for rapid atmospheric adjustments and natural variability internal to the model. The ERF of anthropogenic aerosol for the mid-2090s ranges from-0.15Wm -2 for SSP1-19 to-0.54Wm -2 for SSP3-70, i.e. the mid-2090s ERF is 30 %-108% of the value in the mid-2000s due to differences in the emission pathway alone. Assuming a stronger Twomey effect changes these ERFs to-0.39 and-0.92Wm -2 , respectively, which are similar to estimates obtained from models with complex aerosol parameterisations. The year-to-year standard deviations around 0.3Wm -2 associated with natural variability highlight the necessity to average over sufficiently long time periods for estimating ERF; this is in contrast to RF that is typically well constrained after simulating just 1 year. The scenario interpretation of MACv2-SP will be used within the framework of CMIP6 and other cutting-edge scientific endeavours