Dear Dr. Alexander and co-authors,

Thanks for the revised manuscript and the response to the referees, which is satisfactory.
Your manuscript is accepted for publication in Ocean Science, provided you account for the last technical issues as given below.

Comments from referee #1
Page 5, Line 1: What do you mean by “interannual standard deviations”? I assume you apply the Lanczos filter as described, deseasonalize the data which yields time series that retain only anomalies on interannual time scales, right? From these time series, you compute standard deviations and use them to normalize the response? I guess the term “interannual standard deviations” got me confused and you might want to clarify this passage, as the normalization of all the data is key to interpreting your results.
YOUR ANSWER: The standardized anomalies during February-March-April (FMA) are computed starting with monthly data. A monthly mean climatology is subtracted for each month creating a monthly mean time series of anomalies. The Lanczos time filtering is now applied, which removes variability on time scales longer than 10 years. FMA averages of time filtered anomalies are created for each year. The inter-annual standard deviation of the FMA anomalies is computed and used to standardize (normalize) the FMA time-filtered anomalies.
EDITOR: Your explanation is fine, but I would expect it in the manuscript, not just in the response. If the referee doesn’t understand it correctly, most other readers will likely neither.

Comment from referee #1
Model Evaluation
The authors demonstrate that ESM2.6 shows an improved coastal ENSO response compared to ESM2M. They also compare their model results to a data-assimilative ROMS hindcast. Can the authors elaborate on the reason of using the data assimilative regional model for means of comparison? Would e.g. the long (though not as high-frequent) CalCOFI records provide an additional opportunity for a model independent evaluation of the presented response?
YOUR ANSWER: Our main reason for using the ROMS reanalysis was that it is the best available spatio-temporally resolved estimate of the physical ocean state off the US West Coast. Individual data sources provide reasonably well resolved information on portions of the water column (e.g., satellite data for the surface), or poorly resolved information on the full water column (e.g., CalCOFI, Argo, etc.). The ROMS reanalysis assimilates these data (including CalCOFI), so it combines the strengths of the observations with the strengths of an unconstrained (no data assimilation) ocean model, to give a product that is more useful than either of those alone.
EDITOR: As above, please modify the manuscript text to include such deliberations.

Comment from referee #2
-Line 24: This sentence makes one wonder: What are the drift issues in the carbonate chemistry? It would be useful to read a line or two about this, to keep the reader from wondering if there is anything wrong with the model and better justify the need of a filter. If the drift happens only at the beginning of the simulation as stated, why not consider those years as “spinup time” and remove them from the analysis? Please explain more or re-word the sentence.
YOUR ANSWER: As global carbonate chemistry equilibrium timescales are on the order of thousands of years with respect to water masses and on the order of tens of millennia with respect to river inputs and sediment burial, running the model to equilibrium was not an option in this case.
Figure 1 below shows the temporal evolution of dissolved inorganic carbon (DIC) for the 52 simulation years. The unfiltered DIC is shown in red, and the Lanczos filtered DIC is shown in blue. The drift affects the first 10 to 20 years of the simulation. The authors opted not to remove these affected years, but rather to filter the data, in order to retain as many years as possible, given the relative brevity of the simulation.
EDITOR: as above, please include this kind of information in the manuscript. The figure may be added to the supplementary material.

Please check all the data sets used in this study and add relevant references, and/or correctly acknowledge the use of these datasets.

Figure 1 In the caption please use sea surface temperature not SST, and sea level pressure not SLP, etc. Also in other captions.
Figure 2 Please add the unit
Figure 3 Please add the unit of chl
Figure 4, 5, 6, 7 Please add units of all parameters

P24, L17 reference Di Lorenzo et al is not correct: North Pacific Gyre Oscillation links ocean climate and ecosystem change, Geophys. Res. Lett., 35, L08607, doi:10.1029/2007GL032838.
P25, L5 Frischknecht et al reference not complete
P25, L7 Garcia et al reference not complete
P25, L9 Garcia et al reference not complete
P25, L22 Hoerling et al reference not complete
P25, L24 Hoerling et al reference not complete
P26, L10 Lewis and Wallace reference is not complete
Please check all references carefully, e.g. there are many volume numbers missing. I have indicated some issues above, but there are more.

Thanks and best wishes
Mario Hoppema