Dynamic Simulation of the Crown Net Photosynthetic Rate for Young Larix olgensis Henry Trees
Round 1
Reviewer 1 Report
The Authors have expanded upon previous photosynthetic light-response models directed toward single leaves and applied their simulations to the entire crown area of an economically important tree species within their region of study. Their model takes radiation transfer within the crown into account and has successfully used this parameter in simulating the net seasonal dynamic photosynthetic rate within a crown multilayer. The model has provided a basis for ultimate accurate estimation of long-term CO2 uptake by a plantation of this species and is of value in directing the artificial pruning for newly planted forests.
Suggested corrections for typographical and syntax errors:
Line 33: an L. olgensis plantation
Line 49: “developed” is redundant here
Line 95: “in northeastern China” is also redundant and can be replaced by: located in this region
Line 139: during measurements of the data for…..
Line 160: Following this, all branches
Line 161: Thereafter, in each
Line 218: omit “and VPD” should this be: Amax?
Line 235: quality of fit..
Line 272, 276: “
Author Response
see the word file.
Reviewer 2 Report
The content of the paper in principle dont bring new research ideas. The modelling of crown photosynthesis has long-term tradition and presented paper dont bring somethink new. I was not able didn find any original reasons why the papepr will be usefull for readers
Author Response
see the word file.Author Response File: 
Author Response.docx
Reviewer 3 Report
The experiment provided by the authors attempted to link the variability of parameters of a leaf level light response curve model for photosynthesis (PLR) with an array of independent variables from meteorology and leaf traits. Thus the manuscript’s content is in the scope of the journal and containing interesting point to discuss.
The effort provided by the authors to explain the variability of PLR model parameters is appreciable and potentially drives to interesting insight. However, significant improvement are needed and unfortunately some analysis for scientifically validating the experimental results are missing for now. For that reason I encourage the authors to improve the weak points of their manuscript before resubmitting.
The authors should be more convincing to the concrete possibility to apply their modeling approach in order to scale leaf level photosynthesis model to canopy scale. Low it is discussed the risk of scaling apparent quantum yield (AQY), the maximum photosynthesis (Amax) and dark respiration Rd to less sensical values (e.g. negative AQY).
Also if the aim of the manuscript is proposing a model for scaling An from leaf level to canopy scale, I see essential carrying out a cross-validation experiment, maybe splitting the dataset in two parts (e.g by years, by trees or by both combined) and using one part of data for tuning the model parameter and the remaining data for validating the trained model. The scientific validity of this manuscript have to pass for a comprehensive cross-validation analysis. Also the validation step should be extended to the equation 9 (or to the rescaled PAR by eq 4, 5 and 9)
More details should be provided about the modeling strategy applied to arrive at the eq. 8: Which algorithm and how it is used for fitting the coefficient in table 3? The coefficients in table 3 have been fitted all together (fitting the equation 8) or separated for each parameter (fitting one separated equation for each parameter)? I guess the parameters have been fitted all together and only one R2 and RMSE are reported in table 3. Fitting all coefficient in table 3 by one equation could be risky for the equifinality of retrieved coefficients, for which, functional relationship into eq 8 could be less consistent with the preliminary findings in figure (1). Look for example Rd: in eq 8 Rd is inverse correlated with Tleaf, while by figure 1 it appears a direct correlation.
In addition, the authors should discuss which is the impact of using VPD, RDINC, LMA and Tleaf for estimating AQY, Amax and Rn on the covariance matrix of retrieved parameters, and if this pattern is consistent with the covariance pattern among the PLR parameters as reported in figure 1.
Finally the authors could compare the output by eq.8 to the An estimates without constraining AQY, Amax and Rd by meteorology and leaf traits dynamic.
Significant improvements are required also in the manuscript organization and methods description. PAR measurements are reported section 2.4 and 2.6 but it is not clear how each measurement has been used. Also it is not clear if PAR measurements are acquired in the same moment of photosynthesis measurements (section 2.2) or if PAR and photosynthesis were measured as separated.
Equation 8 and 9 could fit better into Material and Methods sections, maybe in a sub-section focused on the modeling approach. Also, eq. 9 is not consistent with eq. 5.
Minor things:
Ln 106 “in this region are -2.0 °C”. Maybe 2.0 °C?
Ln 283 “according to model 6”. Maybe 8?
Author Response
see the word file.
Author Response File: Author Response.docx
Round 2
Reviewer 2 Report
The content of the paper is not very innovative. Crown photosyntehsis models are many. What I missing is the clear statement WHY and GAINS.
Author Response
see word file.
Author Response File: Author Response.docx
Reviewer 3 Report
The author revised the manuscript by adding small contents focused on validating their modeling approach about An scaling and K estimates. A short description of the methods applied to acquire data for validation is provided in the new section 2.3, while results are shortly shown in Table 4 and 5. In Tables 4 and 5, the metrics used for data validation and data fitting have to be the same, while in the manuscript there are different metrics for the two dataset.
The authors should better clarify in section 2.3 if data used for validation (the ones acquired in natural state) are coming from the same 5 trees of the previous sections or not. I guess the trees are the same, as reported in the caption of Table 2 (also the range of acquired variables that are reported in Table 2 is the same both about the dataset used for fitting the model and data for validation). Also I invite the authors to specify better the period of interest for acquiring the validation data and if it matches with the fitting data acquisition.
If in validation data and fitting data the sample trees and period of interest are the same, the validation results are quite weak because data used for the analysis are not independent (same tree, maybe same branches and same period of interest) and I did not see it as a robust validation. Conversely, a stronger evaluation of the model capability (more convincing about the scientific validity of the proposed method) would be derived by a more robust validation analysis e.g. splitting the dataset by trees (or by the years of study or both) in two samples one for training and the other for validation. The best is implementing a comprehensive cross-validation analysis on the overall dataset e.g. by the leave-one-tree-out method, consisting in leaving out a tree from model parametrization for using it for validation and repeating the operation for all the tree of the sample.
About the model parameter, (AQY, Amax and Rd), I see a conceptual problem in the manuscript development. The manuscript starting point is a semiempirical light response curve for photosynthesis (PLR) model, which parameter should be into a specific range to be sensical from the physiological point of view. Then the correlation between retrieved parameters and meteorological drivers, leaf traits and canopy properties was investigated. The insights emerging by this analysis are used for revisiting the model and replacing the PLR parameters with empirical relationships based on meteorological, leaf traits and canopy properties drivers. However, the output of the equations used for replacing PLR’s parameters have to be consistent with the parameters itself, otherwise is less sensical referring to the PLR equation (despite it is modified). If the aim of the manuscript if maximizing the An prediction, why do not use a machine learning? Instead, if the author want to demonstrate that their method scales An from leaf level to canopy scale, by scaling the parameters of a leaf level PLR curves, the authors have to demonstrate that the scaled parameters are consistent with the ones retrieved at leaf level by fitting the PLR curves, looking also the covariance pattern among PLR parameters and drivers used for scaling.
Ln 257. If this model (eq 10) is not a finding of this experiment and already used in another paper, it should not be among the experiment results but in the methods.
Author Response
see Word file.
Author Response File: Author Response.docx
