Mapping the dynamics of Vcmax at global scale is important for the improvement of the model performance in predicting GPP and to understand the driving factors for its spatial and temporal variations. Recent studies have developed multiple methods to retrieve Vcmax based on satellite observations. This paper by Chen et al. summarized these approaches and provide a direct comparison between these datasets, the one predicted by optimality theory (EOT) as well as in situ observations. The satellite-based datasets generally show good consistency with the EOT and observations. The authors also evaluate the difference between the satellite observations and EOT and suggest that the difference can be explained by irrigation, soil PH, and nitrogen content. This is a solid paper and the developed datasets should be published. However, I still have some comments for the improvement of the manuscript.

In the abstract, the authors mentioned that they use a data assimilation technique to combine the SIF generated Vcmax and LCC generated Vcmax to get an optimized Vcmax, I did not find the description of this data assimilation method. Later in the results, I feel that the authors are referring the TROPOMI SIF based Vcmax as the assimilated Vcmax. If this is the case, the presentation in the abstract should be revised. In the abstract, the authors suggest that the data assimilation technique is to combine "two types” of remote sensing dataset, one is SIF based, the other is LCC based. Clearly, TROPOMI SIF Vcmax, based on its names, should still be considered as SIF based. This naming system is misleading to the readers. I would suggest the authors to reconsider this naming system or revise the abstract.

The authors suggested that irrigation may be the reason to explain the difference between satellite observed Vcmax and EOT predicted ones. I would argue that the improvement in the crop industry ("green revolution"), mostly better seeds, fertilization usages to be the plausible cause. This is based on the fact that the difference in satellite and EOT predicted Vcmax is large over all cropland regions, no matter it is irrigated or not (e.g., irrigation cannot explain the difference in Africa and south America). Second, irrigation would provide enough water which tends to reduce Vcmax based on the optimality theory, this is different than what we see in this comparison.

The manuscript mostly focuses on the comparison of the spatial variation of Vcmax from different datasets. Based on my understanding, all three remote sensing-based Vcamx have seasonal variations. Previous studies have highlighted the importance of correctly representing the seasonal variation of Vcmax to the improvement of seasonal GPP simulations. This seems to be an advantage of the dataset. But I did not see much stress on this temporal variation throughout the manuscript, this is also no cross comparison of these datasets at temporal scales.

Detailed comments:

L31, why three? LCC, SIF and the optimized one?

L32, the link provides two SIF based Vcmax, which is not described here.

L48, it would be good to briefly describe how Vcmax can be derived from SIF, you did this for LCC later but not here.

L64, and SIF is quite noisy.

L69, … to produce a global Vcmax time series dataset? Single time series may refer to only one vector.

L98, the SIF-photosynthesis relationship is only linear at longer time scales (weekly or monthly), you may want to specify this. This sentence can be misleading considering you use “instantaneous”.

L100, “sunlit leaves are the predominant sources of SIF” a reference would be helpful here.

L150, were these obtained from sunlit leaves only? The remote sensing datasets are for the sunlit leaves, right?

L165, I was expected to see the equation here.

L224, also plant genetic engineering. I think this may be a more plausible reason to explain the difference between TROPOMI and EOT. Human selections are producing much more productive crops that the optimality theory cannot predict. It happens that the much of croplands have irrigation. In Fig. 4b, the different is obvious in almost all croplands across the globe.

L228, but the optimality theory predicts lower Vcmax at regions with abundant water resource.

L258, I think you mean biome level Vcmax here.

L259, not sure if TROPOMI is the dataset obtained from data assimilation. This needs to be clarified in the method.

Title: Global Datasets of Leaf Photosynthetic Capacity for Ecological and Earth System Research

Author(s): Jing M. Chen, Rong Wang, Yihong Liu, Liming He, Holly Croft, Xiangzhong Luo, Han Wang, Nicholas G. Smith, Trevor F. Keenan, I. Colin Prentice, Yongguang Zhang, Weimin Ju, and Ning Dong

MS No.: essd-2022-136

If we accept the Farquhar-von Caemmerer-Berry  photosynthetic model as the dominant paradigm for computing leaf and ecosystem photosynthesis and to apply it to the challenge of assessing photosynthesis everywhere and all the time, we will need to assess such key parameters as Vcmax, at a reference temperature.  Chen and colleagues have been leading the way in developing a means to do this and here is their global dataset.  It profits from the sharing of data by many through the TRY plant traits dataset (>3700 datasets) and use of optimization theory by many of the co authors and inferences with information from satellite remote sensing to upscale information.

For the upscaling the authors use two multiple constraints and plausible means, SIF and leaf chlorophyll information deduced from plant reflected spectra.  These are useful and defensible.  Though I do worry about SIF as the signal is small and many show that it represents absorbed light more.  But I don’t see this as a fatal flaw and it is worth exploring

Dechant, B., et al. (2020). "Canopy structure explains the relationship between photosynthesis and sun-induced chlorophyll fluorescence in crops." Remote Sensing of Environment 241: 111733.

My other words of wisdom, having spent time with books on the ground assessing Vcmax is that we know there is lots of seasonality in this parameter, with changes in leaf allocation of N and effects of soil moisture deficits.  But this request may be beyond the scope of this work. But I strongly argue for future efforts to create seasonal maps of Vcmax. My other experience is to find vertical variations in Vcmax with depth in deciduous forests, as there is much light acclimation and strong vertical gradients in leaf N that affect Vcmax.  This complication, too, is beyond the scope of this work. 

In the methods, I am glad to see the authors consider clumping and sun and shade leaves.  This is an effort I would insist upon if one is working on a specific canopy.  Though for global assessments I worry that by doing so it may introduce error in Vcmax as we may not now these other factors with enough precision. 

With regards to inverting information derived from leaf chlorophyll I am satisfied to see them using a state of art radiative transfer model, PROSPECT, for this inversion.  It is the best way to proceed in my mind. Yes, one may use simple empirical algorithms instead, but are they good enough? Nor may they be mechanistic enough.

As noted above using 3700 datasets on A/Ci brings the remote sensing inversion to reality. Can’t ask for a better way to do this.

Temperature normalization is always the trickiest as we see lots of temperature acclimation in the field. But don’t know what else to suggest. Better than nothing.

Results

While it is nice to see computations compared with ground based measurements, realize that the model is fitted with information from the ground. So a bit circular. Would be better to reserve a subset of data for model testing.  It probably wont change things because with 3700 data points there is over sampling, especially given the scaling work of Reich and others showing that 80% of variances in leaf photosynthesis scales with only a few factors, leaf N, specific leaf weight and age.  Maybe comparing your results to this economic leaf scaling result may be a reasonable alternative.

Glad to see a section on response to drivers. Useful. The issue on irrigation is interesting and could be a scale emergent property from this work.  Remember irrigated fields are also fertilized so they will stand out compared to native vegetation.

Discussion

Looking at your maps I see high Vcmax in desert and semiarid areas (Africa, India, Australia and the Cerrado of Brazil) In my early work on stress, I looked a lot at Park Nobel’s work on desert species and indeed did see among the higher Vcmax values. Thinking about Prentice optimization theory I think it makes sense. They need to acquire enough carbon to outpace respiration. But they have a short growing season due to low water supply and high demand.  The only way they can make the economics work is to achieve very high rates of photosynthesis, which comes at the cost of high Vcmax and N. I find this interesting and the authors may want to discuss this a bit

The quality of the figures is good enough. Looks like they are generated by Matlab and have nice color gradients.