Dear authors,
Thanks you for carefully responding to all referee comments and for your proposed improvements to the manuscript. All referees see a great value in your work but also raised numerous points to address. I believe that if the referees comments are addressed adequately, as indicated in most of your responses, the paper will be indeed a valuable puzzle piece for understanding surface-atmosphere water exchange. In addition to the changes you proposed, could you consider the following?

1) Referee 1 critices that the mechanisms resulting in modelled NWL were not explained, while reference is made to an alleged under-estimation of nocturnal stomatal conductance in the models without going into details. Even though a detailed analysis of how NWL is simulated in the different models may be out of scope for this paper, it would be helpful if you provided NWL results for each model in addition to the multi-model mean values, or at least a ranking of the models, so that the readers could verify for themselves what might cause the spread in the simulations. The CMIP5 data does acutally distinguish between water_evaporation_flux_from_soil, water_evaporation_flux_from_canopy and transpiration_flux, so you could at least verify what proportion of simulated NWL comes from transpiration as opposed to evaporation in the models. I think that this would be within the scope of the study, which is to provide a "general overview of NWL across the globe from observations and climate models".

2) The explanation of the process underlying NWL should include an energy balance consideration. You mention the influence of air temperature, VPD and wind speed on NWL, but what about sensible heat flux, soil heat flux (also soil temperature), and longwave radiation?

3) Please make sure that all your responses to the referees are also reflected in the manuscript, as the referees' comments likely reflect your future readers' thoughts. For example, your response to Referee #1 that the data does not account for LE storage might also be an important information for the reader.

4) I agree with Referee #1 that Figure 7 does not add much value to the paper and could be removed, unless you would like to make a strong point that the models do not only disagree about the magnitude of NWLf but also its trend.

5) In your response to Referee #3, you propose that possible ecological advantages of NWL include capacitance refilling, embolism removal and hydraulic redistribution, among others. Hydraulic redistribution, capacitance refilling and embolism removal may explain nocturnal sap flow, but not water loss. The difference between sap flow and NWL sensed by EC towers should be emphasised more prominently throughout the text.

6) Your response to Referee #3 about P7L1: It should be easy to verify if these sites include more gap-filled data than the average, and to mention this in the text.

7) Data availability and reproducibility of results: Thank you for providing the original lysimeter data. However, for your analysis to become reproducible, it would be important to also provide the scripts that were used to analyse both the Fluxnet and the CMIP5 data. In addition, when I opened the link given for the CMIP5 data, I had to click on CMIP5 in the list of data, create an account and eventually landed on a search page for data. Could you please provide accurate instructions on how to access the exact data you used for the study? Is there a specific search query that would take the reader to the right data? Same for the Fluxnet data: which 99 stations did you use, and which years of each station?

Thank you for all your efforts that will make this paper more useful for the readership.

Best regards,
Stan Schymanski

Dear authors,

Thanks very much for the thorough revision and for using all our comments to further improve the manuscript. Both referees are happy with the revision, but one of them requested some final technical corrections, which I would like you to consider before publication.
Could you also consider the following final remarks from my side?

1) It is still not made very clear that NWL and sap flow are not necessarily the same thing. If sap flow serves predominantly capacitance re-filling, there can be nocturnal sap flow without any NWL. For example on P 13L5-15, NWL and sap flow are used almost interchangeably without clarification of this potential difference.

2) Thank you for the much more detailed description of how to obtain the underlying data, very helpful! I disagree, though, that the scripts used for data selection, analysis and plotting are not relevant. First of all, if someone had the idea to repeat your analysis but change one step or plot a different set of variables, they might be able to save tremendous amounts of time by using your scripts instead of re-coding everything again. Secondly, if someone fails to reproduce your plots using the same data, it would be much easier and faster to find out where things deviate if your scripts were available. I hope you agree that providing your scripts along with the data would tremendously improve the utility and impact of your paper in the future.

Best regards,
Stan