|This paper is well written and the results will be of interest to the model development community. Particularly, the use a model like spCAM5, which better resolves convective processes, to learn how we might improve convective parameterizations. Overall, the authors have responded well to many of the reviewer comments, but have not necessarily reflected the responses as edits to manuscript. Additionally, the manuscript would likely have a larger impact if some of the remaining comments were addressed directly. I outlined some remaining issues below that I believe would make the work more impactful and leave it to the editor to decide if these are within the scope of this work.|
1. I think some of the responses to both reviewers' comments add important insight and could be better incorporated into the main manuscript. In particular, clarifying the how the results of this work that are new and discussing how they might be applied to guide future model development would be useful. I also suggest incorporating the “additional figures” into the main manuscript or as supplemental material, which will likely be of interest to many readers.
2. The additional figures using total instead of convective precipitation do not completely alleviate the issues about how "convective precipitation" is defined. It is not enough to show that the percent defined as convective is similar in spCAM5 as in CanAM4.3, or that the total and convective precipitation in spCAM5 shows a similar result. These do not necessarily indicate the offline calculation of "convective precipitation" means the same thing it does in a parameterized model. Would it be possible to calculate the "convective precipitation" from "total precipitation" in CanAM4.3 using the same method used for spCAM5, then compare to what CanAM4.3 internally defines as “convective precipitation”? If an offline calculation of "convective precipitation" gives the same result as the modeled "convective precipitation" in CanAM4.3, it would support its use with spCAM5.
Additionally, there are differences for CanAM4.3 in the relationship between “mean peak precipitation” and “event length” when "total" is used instead of "convective" precipitation (Figure 1d vs. Additional Figure 1d). In particular, the total precipitation shows an increase with event length. Does this imply that large-scale precipitation plays an important role for longer events?
For additional Figure 2 and 3, it would a be useful to show CanAM4.3 total precipitation in order to assess consistency with the convective precipitation results alongside spCAM5.
3. It makes sense that spCAM5 could be a proxy "in the absence of high spatial resolution and sub-hourly observations". However, it would be useful to show that spCAM5 can capture the relationships seen in nature at the space-time resolutions that observations are available. It would make a much stronger case to include some comparison to these relationships, even at lower temporal resolution, to demonstrate that how well spCAM5 can match observations.
4. I think the paper would be stronger if CAM5 results were included in some way to address the unconstrained differences between spCAM5 and CanAM4.3. If running CAM5 is too difficult, I would encourage the authors to look for a similar simulation available in archives (ESGF or NCAR) of CAM5 data to enable at least a qualitative comparison. In the absence of including CAM5 results, it is important to clarify in more detail what all the relevant differences between spCAM5 and CanAM4.3 might be. I do not think it is sufficient to simply provide a reference, a summary of the differences, including microphysics and aerosol, would be helpful.