My understanding of the main objective of this manuscript is to explain why the global aerosol direct radiative forcing can exhibit a different trend than aerosol emissions in some climate models. The authors use the MPI-ESM1.2 model to quantify the historical direct, indirect and clear-sky aerosol forcings, and they find that the efficiency of aerosol direct radiative forcing is enhanced in the low-latitude regions such as the South and East Asia, where cloud masking is less and aerosol residence time is longer, compared to mid-latitude regions like Europe and North America. The work is interesting and falls within the journal’s scope, but there are major concerns on the methodology employed in the current manuscript.       

1. The modeling approach to estimating the historical aerosol direct and indirect forcing from emissions used in MPI-ESM1.2 for this study is different from many CMIP6 models that represent aerosol effects through more explicit aerosol-cloud-radiation processes. This raises the concern of accuracy and consistency issues of the present study. The emissions of major anthropogenic species such as sulfur compounds and black carbon have been changing differently since 1980s across major source regions. However, the prescribed aerosol optical properties in the MPI-ESM1.2 simulations are based on measurements for 2005. I don’t think the year-2005 aerosol properties are representative of the past few decades, especially for this research focused on the time evolution of aerosol forcing. Given that the key conclusion on the impact of changes in the aerosol residence time between mid-latitude and lower-latitude emission regions, I wonder whether the assumptions and model representation of aerosol optical properties reflect such changes. How large is the uncertainty in aerosol forcing comparing to the magnitude of direct and indirect forcing?
2. The magnitude of present-day aerosol direct radiative forcing is larger than the CMIP model mean (Bellouin et al., 2020) and in the AR6 assessment. Is it because only sulfur aerosol is considered in the MPI-ESM1.2 model estimates? What other anthropogenic aerosol components (or precursor gases) are considered in the simulations? Figure 1 highlights the global anthropogenic SO2 emissions trend, but according to Hoesly et al. (2018), many other important components (e.g., BC, OC, NH3, NOx) had global increasing trends and very different regional trends in the past few decades. BC and OC are particularly important in aerosol direct forcing over South and East Asia, where this study emphasizes an increase in the sulfur emissions and direct forcing. Please explicitly evaluate the impact of BC and OC changes on the direct forcing trends.
3. The discussion of aerosol SSA got me confused, as there is not sufficient information on the other aerosol species and why biomass aerosol is relevant to the focus of anthropogenic forcing of this study. I also wonder whether the SSA is set to a spatiotemporally uniform value in the simulations, which comes back to my major concern on the methodology.
4. Although I don’t expect the MPI-ESM1.2 results to be similar to other CMIP6 models, it would be nice to see how comparable the aerosol forcing estimates of this study to other models or observational analysis. 

Reference:

Hoesly, R. M., Smith, S. J., Feng, L., Klimont, Z., Janssens-Maenhout, G., Pitkanen, T., Seibert, J. J., Vu, L., Andres, R. J., Bolt, R. M., Bond, T. C., Dawidowski, L., Kholod, N., Kurokawa, J.-I., Li, M., Liu, L., Lu, Z., Moura, M. C. P., O'Rourke, P. R., and Zhang, Q.: Historical (1750–2014) anthropogenic emissions of reactive gases and aerosols from the Community Emissions Data System (CEDS), Geosci. Model Dev., 11, 369–408, https://doi.org/10.5194/gmd-11-369-2018, 2018.

This short study uses the MPI-ESM climate model to suggest that aerosol direct radiative forcing has not followed the same decreasing trend as aerosol emissions. The authors explain that decoupling by an increase in aerosol radiative forcing efficiency (radiative forcing per unit emitted mass or per unit aerosol optical depth) due to a regional shift in aerosol distributions.

It is nice to see the possibility of a change in aerosol direct radiative forcing efficiency from changes in emission regions. That possibility has been mentioned in the past, including in the conclusion of the Bellouin et al. (2020) review paper cited by the authors, but I am not aware of a publication dedicated to the subject. The other results discussed by the authors (reduction in aerosol direct radiative forcing due to cloudiness, aerosol optical depth is more correlated than emissions with direct radiative forcing) have been discussed several times, especially in AeroCom papers, but they provide the context needed to explain the main result.

However, the study needs to go deeper when explaining the reasons for the change in radiative forcing efficiency, as I comment below. For this reason, I recommend additional analyses to clarify the drivers of aerosol direct radiative forcing efficiency in MPI-ESM.

Main comments:

The headline result that aerosol direct forcing is decoupled from aerosol emissions because of a change in radiative forcing efficiency is in many ways “built-in” the MAC-v2SP aerosol prescription used in MPI-ESM. The shapes and slopes of the curves shown on Figure 2 are a consequence of the choices made in MAC-v2SP in terms of aerosol plume properties and relative change in cloud droplet number. So the sensitivity to those choices needs to be explored more critically, specifically:

• There is a disconnect in the study between SSA and direct radiative forcing efficiencies that I do not understand. In Figure 2 and 4, why is direct radiative forcing efficiency in South Asia so strong? South Asia aerosols are known to be strongly absorbing, so I would have expected their direct radiative forcing to be less negative for a given AOD compared to regions dominated with more scattering aerosols. Moreover, the suggestion in section 3.5 that SSA does not significantly drive direct radiative forcing efficiency seems to go against AeroCom findings (e.g., the large differences in normalised radiative forcing shown between different aerosol types in the Tables of Myhre et al. (2013) https://doi.org/10.5194/acp-13-1853-2013 ) and I do not understand why. Perhaps MACv2-SP aerosols are too dominated by sulfate?
• The link between residence times (different removal rates depending on region) discussed in Section 3.4 and MAC-v2SP is unclear. MAC-v2SP is a combination of global aerosol modelling outputs and satellite retrievals, so it may implicitly account for different residence times, but that would not influence radiative forcing efficiencies, since MACv2-SP simply scales plumes up and down with emission rates. Unless that Section 3.4 discussion is of a speculative nature?

Other comments:

• Lines 21-23: Note that Forster et al. (2021), cited just a few sentences earlier, prefers the aerosol-radiation and aerosol-cloud terminology instead of direct and indirect.
• Line 29: Quaas et al. (2022) https://doi.org/10.5194/acp-22-12221-2022, and especially their Section 5, seems a very relevant reference here, and elsewhere in the paper as well.
• Line 73: “as a two-sided version”. What does that mean?
• Line 103: How are those “Tg of SO2 equivalent” calculated?
• Line 128: “absorption prevails in the presence of clouds” – only if the aerosols are above clouds.
• Line 151-152: Do you use the latest CEDS emissions for the MACv2-SP plume scaling, dated 21 April 2021? There have been significant changes in aerosol emission trends, especially over China, that would affect the results presented here.