Dear authors,

Thank you for the detailed response to the reviewers comments. I have received further comments from reviewer 2, who raised a few issues. Please see review below. I think it is critical that these concerns are addressed and further clarifications are incorporated into the manuscript. Once these are addressed the manuscript will be ready for publication in ACP.

Best,
Sally

----
The added supplement information is very clear that the AM3 mechanism may have larger yields for wrong reasons. The additional figures are also very helpful. The new supplement makes it a much better paper. There are still three issues I think the authors need to correct. The changes will not require much additional work.

One of my review questions is why there is a strong emphasis on the higher yield under low-NOx condition. It is still unanswered. Since this study and Li et al. both suggested a higher yield under lower NOx conditions (for different reasons), I suppose there must be observation evidence requiring it. But one of the authors’ responses stated “there are only 3 observations with NOx < 100 pptv”. Then there were no observations under really low-NOx conditions. Why would two studies focus on an issue that has not observation support? Can the authors make a clear statement in the paper as to whether or not a higher yield under low-NOx condition (than MCM) is required by the observations? It needs to be stated in the paper’s conclusions. Note that Fig. 6 is not the evidence if most of the missing CHOCHO in the lowest NOx bin is from pinene oxidation. In Fig. 6, the observations do not show an increase R(GF) with NOx (if the lowest NOx bin is not included); in contrast, the model shows the increase for the whole NOx range. The statement in Line 31 (P. 7) “In both the model and observations there is a subset of low-NOx points with higher RGF values (0.03-0.06)” is very misleading since the model did not simulate the observations of RGF>0.035, which are the majority of 0.03-0.06 data points.

P. 6, Line 31-32, P.7 Line 1-2. If shallow convection is the reason, why does it not affect MVK+MACR in the observations (Fig. S8)? The model CO bias is larger at 3-4 km than 1-3 km. That cannot be an indication of shallow convection. I think it’s better to acknowledge that the reasons for measured CHOCHO at 2-3 km are not understood.

In the response, the authors stated that “The HCHO scaling was based on a validation of OMI HCHO observations using SEAC4RS HCHO observations by Zhu et al. (2016). The reasons for the bias are presently unknown, and we do not claim that the CHOCHO retrievals are not subject to similar error sources.” This is a fair statement and should be noted in the conclusions of this paper. It will then put a statement like “The HCHO satellite data are better validated…” in the appropriate context.

Dear Authors,

Thank you for your response to Reviewer 2’s further comments. Most of the comments have been addressed adequately. I have some further comments about Figure 6. It appears that even with the higher yields (than MCM), GEOS-CHEM still cannot capture the magnitude of the high RGF values in the observed data. Regarding comment 2 from the reviewer, I agree that 1) page 7 line 31 should be modified to more accurately reflect the difference in observation and model results, 2) an explicit statement on whether a higher yield (than MCM) under low-NOx condition is required by the observations should be included in the manuscript (and further justifications can be included if necessary). These will improve the clarity of the manuscript. Once these are considered and addressed, the manuscript will be accepted for publication in ACP.

Best,
Sally