Review for “Prescribing the aerosol effective radiative forcing in the Simple Cloud-Resolving E3SM Atmosphere Model v1” by Mahfouz and co-authors.

Mahfouz and co-authors present a new simplified framework for representing aerosol effects on radiation and clouds (SPA; a simplified prescribed aerosol scheme) that maintains some degree of fidelity regardless of model spatial resolution. This is timely given the move towards kilometre-scale earth system models that complement the coarser resolution models traditionally used for CMIP-style experiments. Having a framework that is consistent across model configurations provides an important tool for quantifying the role of aerosols in the earth system.

In this study, the authors produce aerosol climatologies for pre-industrial and present-day climates using a coarse-scale (~100s km) version of E3SMv3 that includes a detailed aerosol microphysics model. The climatologies are used to prescribe time-varying 3D fields of aerosol optical properties in the higher-resolution (10s km) version of E3SMv3, thus representing aerosol-radiation interactions associated with anthropogenic activity. For aerosol-cloud interactions, the authors couple the aerosol climatologies to the cloud microphysics scheme via a function that relates the aerosol concentration to a cloud droplet number concentration. Sensitivity tests are performed to find a setup of SPA that best reproduces the global ERFaero that the coarse-resolution version of E3SMv3 produces. The SPA scheme and its associated methodology is an excellent addition to the community’s ability to represent aerosol processes in an idealized/prescribed framework.

I thoroughly enjoyed reading this manuscript, and believe it is well suited for publication in ACP. However, I have some minor comments I would like to be addressed before recommending publication.

General comments.

This study implicitly assumes that E3SM-v3 is accurately representing aerosol distributions (throughout the industrial era) and Nc distributions. Have these been evaluated? If these evaluations are yet to be performed and published it may undermine this study. Can the authors include these evaluations themselves? Given that Nc ends up being an important contributor to the sensitivity can the authors compare to the Nc observations from Grosvenor et al 2018? If these evaluations cannot be included/referenced I recommend the authors explicitly state these as potential caveats/limitations of the study.

It is recognised that an aerosol perturbation in one region may have non-local effects associated with large scale circulation. This in turn can impact subsidence rates, moisture transport etc and represent non-local aerosol feedback. I would expect the nudging of the U/V wind components to prevent, or at the least dampen, these effects. Do the authors recognise this as a limitation?

One of the primary benefits of using kilometre-scale configurations for examining aerosol processes is the opportunity to study regional-scale responses of the atmosphere/clouds/radiation to the aerosol perturbation. By focusing on the global ERFaero, the regional-scale fidelity is sacrificed. Is this appropriate? Would a region-to-region comparison (or series of limited domains) be a better suited test?

Throughout the manuscript, the authors discuss the robust ERFaer signal that is achieved through the nudging. Are all grids included in the global magnitudes or only those that provide a robust signal? If there is a lot of heterogeneity in the statistical significance of the grid points I suggest the authors include stippling/masking to demonstrate this in the global plots – otherwise include a note in the manuscript to make the reader aware that all grid points can be considered significant/robust etc.

Specific comments.

Line 3 “we quantify the forcing due to anthropogenic aerosol changes using a simplified prescribed aerosol scheme…”. I think it is more accurate to state that you are assessing the sensitivity of the forcing. In reality, the aerosol forcing that has already been quantified by E3SMv3 and is now being used to constrain SCREAM via a series of sensitivity experiments.

Line 5. “Nudged simulations at 3 km and 12 km horizontal grid spacings reveal a more negative aerosol forcing than the reference 100-km 5 E3SM v3..” This would imply that model-based estimates of the global aerosol forcing are dependent on resolution. This may well be the case, but I don’t think this is supported by the study.  The sensitivity experiments suggest that there is considerable impact from a combination of the Nc limiter in the reference simulation and an inadequate representation of ACI that (initially) fails to capture the coarser-scale configuration. Between these two effects it is impossible to say there is resolution-dependence between the 100 km and 3/12 km configurations. I suggest this is rewritten to better capture the study outcomes.

Line 6 “…exhibits little overall resolution sensitivity”. As a global ERFaero I would agree, but there is much more sensitivity at the regional scale – as highlighted by the authors on line 210. I would argue that differences between these two configurations actually do demonstrate resolution dependence (unlike 100km vs 12km). The only difference between the 3 and 12 km configurations is the resolution – the ACI/ARI treatments in this case are identical and therefore a better comparison. At the least, I suggest that the authors expand this sentence to highlight the spatial differences between the 3 and 12km configurations.

Line 75. Is this also an assumption explored in Caldwell et al 2021? If so, I recommend adding the reference at the end of this sentence for clarity.

Line 80/106. How is Nc in the P3 scheme coupled to the radiation scheme? Do changes in Nc influence the droplet effective radius in RRTMGP?

Line 124. In SCREAM v1 all clouds (I presume) are represented by P3, therefore the SPA scheme can influence all clouds. Is this also the case in the coarse configuration of E3SMv3? Is the aerosol microphysics scheme coupled to cloud microphysics in deep convection? / all clouds? If not, then are you comparing like for like? Perhaps expand this to fully describe how aerosols influence different clouds in the E3SMv3 configuration.

Line 172. Why do you choose to nudge all levels rather than only the top 70 as in Zhang et al. 2022a?

Line 174. Do you mean ERA5 instead of MERRA-2?

Line 176. Was 2009-2010 a similarly weak el-nino year?

Table 1. For clarity, the authors could add Nc >= 20 cm-3 to the reference case. Also, the footnote should be moved to the methodology when defining/explaining the Nc limiter (see comment further down)

Line 195. Is the 12 km configuration able to sufficiently represent convection? I thought this was in the gray zone?

Line 218. Typo respectiely

Line 251. Table 2 should be Table 1?

Line 251. The Nc limiter. My understanding is that the standard E3SMv3 configuration uses a Nc >= 20 cm-3 limiter – is this correct? This isn’t defined/explained in the methodology – please include or make clear. It took me a while to (hopefully correctly) realise that this is the default setting.

Line 252. Are the frequent occurrence of low Nc values anonymously low with respect to observations? I presume this is why the default reference configuration limits Nc. If this is the case, then why do the authors work towards reproducing the global ERFaero from the non Nc-limited configuration?

Line 256. By fitting a linear (log-log) function the gradient of dNc/dNccn is constant – yet as the authors note this is not reflected by the E3SMv3 data. For the phase space that represent most of the data, the resulting gradient is too shallow (centre of the plot) – thereby dampening the sensitivity of Nc to Nccn.  I wonder whether the best function is a three-regime function that would better capture the changing gradients. This wouldn’t add much computational burden and would better reflect the variable gradients.

Line 263. Is this Section reference correct?

Line 264. Please can you add the reference simulation to Figure 4 to demonstrate the good agreement?

Line 291. Main conclusion. The aim is to reproduce E3SM-MAM with E3SM-SPA. This is achieved when comparing E3SM-MAM that is not limited by Nc, with E3SM-SPA using the Nccn-Nc function derived from E3SM-MAM data when limited by Nc. Does this not point to a fundamental discrepancy? None of the sensitivity simulations are able to reproduce the E3SM-MAM reference ERFaero value of -0.74 Wm-2. Do you think this demonstrates an ability to constrain the model across different resolutions? I would say this is the only the case when Nc is not limited – but then don’t you have an unrealistic model configuration?