Estimation of Gross Primary Production and Net Primary Production of Vegetation Cover for Low Mountain Sub-Mediterranean Landscapes Using Remote Sensing and Geoinformation Modeling

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript is written clearly and presented in a logically structured way. However, there are still some aspects that need to be reviewed and eventually corrected. Correlation and trend values are indicated without any level of statistical significance. I have incorporated some suggestions into the manuscript, which I submit for your consideration.

Comments for author File: Comments.pdf

Author Response

Comments 1: This keywords have already been used in the title.

Response 1:  We've fixed the keywords.

 

Comments 2:  [400] ¿?

Response 2: We've fixed.

 

Comments 3: Where is located the Karadag Nature Reserve? Why was this territory included in the analysis?

Response 3: Based on the comments from you and the other two reviewers, in order to avoid confusion, we have decided to exclude the reserve area from the analysis in this manuscript. We will provide a more detailed analysis in future publications..

 

Comments 4: Climate values of temperature and precipitation should be included in Figure 1 or 2.

Response 4: The climate data was not output as an average annual value for the entire observation period. The analysis was carried out in the R ! In the form of data matrices. In the manuscript, we have added several geographical maps that characterize the climate characteristics of individual observation years.

 

Comments 5: Is the information in Figure 4 and Table 1 the same? If the answer is yes, there's no point in duplicating it.

Response 5:  We agree with the comment. Corrections have been made to the manuscript.

 

Comments 6: Is there any reference that helps to classify the fluctuation as moderate?

Response 6: In this case, we did not quite appropriately translate the term into English. It would be correct to say insignificant ("Significant").

 

Comments 7: Where these results are shown?

Response 7: The results obtained are presented in section 3.3 (new numbering)

 

Comments 8: Extract the repeated text

Response 8: We agree with the comment. Corrections have been made to the manuscript.

 

Comments 9: Fig. 6 and Fig. 7 are identical.

Response 9: С We agree with the comment. Corrections have been made to the manuscript.

 

Comments 10: Theil–Sen slope estimator and the Mann–Kendall test should be included in the Materials and Methods chapter. The results regarding the trend of the different vegetation covers are shown very quickly. Are the observed trends significant or not?

Response 10: We agree with the comment. Corrections have been made to the manuscript  We have added a description of the research methodology and made corrections to the text.

 

Comments 11: Is the information in Figure 9 and Table 3 the same? If the answer is yes, leave the Figure or the Table.

Response 11: We agree with the comment. Corrections have been made to the manuscript.

 

Comments 12: If the information in Figure 10 and Table 4 is the same, keep only the Figure or the Table.

Response 12: We agree with the comment. Corrections have been made to the manuscript

 

Comments 13: What significance level (p) was used to discriminate significant relationships?

Response 13: 0.05

 

Comments 14: The discussion does not consider the difference or similarity between the results obtained for southeastern Crimea and for Karadag Nature Reserve.  

Response 14: Based on the comments from you and the other two reviewers, in order to avoid confusion, we have decided to exclude the reserve area from the analysis in this manuscript. We will provide a more detailed analysis in future publications..

 

Comments 15: What is the purpose of including the Karadag Nature Reserve territory in the study?

Response 15: Based on the comments from you and the other two reviewers, in order to avoid confusion, we have decided to exclude the reserve area from the analysis in this manuscript. We will provide a more detailed analysis in future publications..

 

Comments 16: Why was the phenomenon of drought not included in the discussion, but in the conclusions?

Response 16: In this case, the translation is not entirely accurate. The corrections have been made to the manuscript.

Reviewer 2 Report

Comments and Suggestions for Authors

The paper is focused on the evaluation of interannual dynamics of GPP and NPP in a regional scale using MODIS products. The aim of improving carbon balance evaluation for different regions is still important, in particular for the former USSR area, where the net of eddy covariance measurement stations is still rather sparse. However, the present paper has important weak points and must by strongly revised.

• The paper is presented rather inaccurately, with mistakes in site description, insufficient figures captions, maps with mixing English and Russian inscriptions, no description of NPP source etc. (detailed comments are in the attachment).
• It would be good to add some data about the climate of the area (e.g., long-term mean seasonal dynamics of temperature and precipitation, as well as interannual weather dynamics of the period under consideration, months of the main growing season).
• When giving units of GPP you should always notice if it is per day or per year. When describing seasonal variability I suggest to use values per day or per month, and when describing interannual variability use values per year.
• As the upper limit of GPP of 6.55 is the artifact of the database, real GPP maximums can be higher. And as this limit is reached every year, the excess of real GPP over this limit can be rather frequent events. Try to correct it somehow (e.g., using literature sources, other databases, comparison with eddy covariance measurements in similar ecosystems etc.). Or analyze seasonal dynamics and histogram to check if the contribution of these excess events can be neglected.
• The evaluations of data variability (“pronounced interannual fluctuations”, “moderate variability” etc.) should be based on statistics (at least standard deviation), which can be added to the Figs. 5 & 8.
• The analysis of NPP and GPP dependences on environmental factors is very limited. You have 8-days means, but the study is based mainly on annual totals. However, namely seasonal factors can explain your observations. E.g., steppes can be dry already in May, whereas forests remain green until October. The crop area after crop harvest remains mainly without vegetation, whereas during growing season its GPP could be high because of applying agricultural practices. Urban greenness is irrigated, which leads to it’s high GPP, etc.
• The Karadag area is very small, even in the regional scale. So, including it separately in the analysis must be clearly justified.
• I can recommend to exclude the Karadag story and limit the NPP story (as it is modeled by MODIS from GPP and includes additional sources of errors) and add a seasonal analysis

Comments for author File: Comments.pdf

Author Response

Comments 1: The paper is focused on the evaluation of interannual dynamics of GPP and NPP in a regional scale using MODIS products. The aim of improving carbon balance evaluation for different regions is still important, in particular for the former USSR area, where the net of eddy covariance measurement stations is still rather sparse. However, the present paper has important weak points and must by strongly revised.

Response 1:  Dear reviewer. We are grateful for your work done in reading the manuscript of the scientific article submitted by us and the time spent on it. Thank you for your valuable comment. We have made corrections.

 

Comments 2: The paper is presented rather inaccurately, with mistakes in site description, insufficient figures captions, maps with mixing English and Russian inscriptions, no description of NPP source etc. (detailed comments are in the attachment).

Response 2:  We have made corrections

 

Comments 3: It would be good to add some data about the climate of the area (e.g., long-term mean seasonal dynamics of temperature and precipitation, as well as interannual weather dynamics of the period under consideration, months of the main growing season).

Response 3:  We have made corrections

 

Comments 4: When giving units of GPP you should always notice if it is per day or per year. When describing seasonal variability I suggest to use values per day or per month, and when describing interannual variability use values per year.

Response 4:  We have made corrections

 

Comments 5: As the upper limit of GPP of 6.55 is the artifact of the database, real GPP maximums can be higher. And as this limit is reached every year, the excess of real GPP over this limit can be rather frequent events. Try to correct it somehow (e.g., using literature sources, other databases, comparison with eddy covariance measurements in similar ecosystems etc.). Or analyze seasonal dynamics and histogram to check if the contribution of these excess events can be neglected.

Response 5: The values of 6.55 represent artifacts that we considered as data outliers, as indicated in the graph. In turn, we note that these are single pixels that do not affect the overall calculation. To display the data more accurately and visualize it better in the figures, we removed the artifacts from the study, while the average values did not change and there was no need to adjust the data tables.

 

Comments 6: The evaluations of data variability (“pronounced interannual fluctuations”, “moderate variability” etc.) should be based on statistics (at least standard deviation), which can be added to the Figs. 5 & 8.

Response 6:  We have made corrections

 

Comments 7: The analysis of NPP and GPP dependences on environmental factors is very limited. You have 8-days means, but the study is based mainly on annual totals. However, namely seasonal factors can explain your observations. E.g., steppes can be dry already in May, whereas forests remain green until October. The crop area after crop harvest remains mainly without vegetation, whereas during growing season its GPP could be high because of applying agricultural practices. Urban greenness is irrigated, which leads to it’s high GPP, etc.

Response 7:  Unfortunately, we did not consider seasonal dynamics in this work, but we are grateful to the reviewer for a valuable idea that we can implement in future research.

 

Comments 8: The Karadag area is very small, even in the regional scale. So, including it separately in the analysis must be clearly justified. I can recommend to exclude the Karadag story and limit the NPP story (as it is modeled by MODIS from GPP and includes additional sources of errors) and add a seasonal analysis

Response 8:  Based on the comments from you and the other two reviewers, in order to avoid confusion, we have decided to exclude the reserve area from the analysis in this manuscript. We will provide a more detailed analysis in future publications.

Comments 9: maybe "in particular"? Karadag is within SE Crimea

Response 9:  Based on the comments from you and the other two reviewers, in order to avoid confusion, we have decided to exclude the reserve area from the analysis in this manuscript. We will provide a more detailed analysis in future publications.

 

Comments 10: in which units?

Response 10:  dimensionless value

 

Comments 11:??

Response 11:  We have made corrections

 

Comments 12: It makes sense to mention also eddy covariance as a widespread method of NEE and GPP evaluation

Response 12:  We have made corrections

 

Comments 13: I think you can add more references about gas exchange measurements

Response 13:  We have made corrections

 

Comments 14: Unclear. 1. Crimean mountains are to the west, not east. 2. How the spurs of Crimean mountains can be on the north if the mountains themselves are in the west? Or the area of the study is also considered as a part of Crimean mountains? 3. As I can assume from the map, the northern border is going along the watershed, isn't it?

Response 14:  The borders are drawn along the watershed. The Crimean Mountains are shaped like an arc or a horseshoe, so they surround the region from the east and north. In the west, the mountains end near Feodosia.

Comments 15: which %?

Response 15:  We have made corrections

 

Comments 16: Can you make the map more sharp - the names of rivers are hardly readable?

Response 16:  We have made corrections

 

 

Comments 17: In the text you write "sessile oak" and in the legend "durmast oak". Please be consistent

Response 17:  We have made corrections

 

Comments 18: you didn't mention how did you get NPP

Response 18:  Данные о NPP и GPP получены с базы данных MOD17A2H с использованием Google Earth Engine

 

Comments 19: per year?

Response 19:  We have made corrections

 

Comments 20: 1. It would be better to use a background map with toponyms in English. 2. Shaded areas are settlements?

Response 20:  We have made corrections.

 

Comments 21: 1. write the axes titles in English. 2. GPP is per year? 3. What mean red dots? 4. Lower panel represents median/quartile range/non-outlier range? 5. Why do you need upper panel at all - the lower panel shows the same thing + additional statistics?

Response 21:  We have made corrections

 

Comments 22: Unclear. Every year it was absolutely the same maximum or you took the overall maximum for the whole period?

Response 22:  Every year, there was this value. This is an artifact. We have recalculated the values.

Comments 23: Unclear. If you show annual statistics, units must be per day or month. But the values are too high for daily, so I assume these are annual values. Please clarify it

Response 23:  We have made corrections

 

Comments 24: So, it is an artifact of the database, and real maximums could be higher. I suggest to write it more accurately, both in the table and in the text, that max is NOT LESS than 6.55

Response 24:  We have made corrections

 

Comments 25: pleas reformulate in order to avoid mixing up with "minus"

Response 25:  We have made corrections

 

Comments 26: annual totals

Response 26:  We have made corrections

 

Comments 27: you just showed that mean (or total)values are stable

Response 27:  We have made corrections

 

Comments 28: I think so detailed description is not necessary, because it repeats the graph. Point out the most important things!

Response 28:  We would like to keep this text in its entirety.

 

Comments 29: Not fully true: the figure shows much higher productivity of settlements area compared to natural steppe area.

Response 29:  We have made corrections

 

Comments 30: you wrote it already in Methods

Response 30:  We have made corrections

 

Comments 31: What mean black areas?

Response 31:  These are map display errors. We have made corrections

 

Comments 32: Changes were positive absolutely everywhere?

Response 32:  Yes.

 

Comments 33: Dynamics of... (otherwise one can think that it means change relatively to a certain reference)

Response 34:  We have made corrections

 

Comments 35: kgC/m2/y?

Response 35:  We have made corrections

 

Comments 36: It is not so evident, especially for precipitation.

Response 36:  We have made corrections

 

Comments 37: Which data do you mean? The satellite images cover the whole globe. Or you mean ground observations?

Response 37:  Yes, ground observations

Reviewer 3 Report

Comments and Suggestions for Authors

1. The claim of novelty rests heavily on being the "first" systematic 25-year assessment for this specific area, but the paper doesn't really engage with how methodologically distinct this is from similar regional MODIS-based GPP/NPP studies elsewhere in the Mediterranean basin. What is actually new here beyond the geographic location?
2. Given that MODIS pixels are 500 m and the Karadag reserve is fairly small and topographically complex, how confident are the authors that vegetation-community-level breakdowns aren't smeared by mixed-pixel effects? This seems like it could meaningfully bias the community-type comparisons, especially at boundaries between forest and cultivated land.
3. The discussion acknowledges the coarse resolution limitation but then still draws fairly specific conclusions about individual vegetation types and even settlement-level correlation reversals. There's a bit of a disconnect between the caveats stated and the confidence of the conclusions drawn from the same data.
4. The correlation analysis with temperature and precipitation is described as pixel-wise but no information is given on autocorrelation handling or the effective degrees of freedom, which matters a lot for spatial time series like this. Were significance levels adjusted at all?
5. CUE is computed simply as NPP/GPP without any uncertainty propagation from the underlying MODIS QA flags, so how much of the CUE variability across vegetation types is signal versus retrieval noise?
6. The Theil-Sen and Mann-Kendall trend analysis appears only briefly for vegetation types and isn't applied consistently across GPP, NPP, and CUE series, so it's hard to tell whether the "no long-term trend" conclusion in the abstract holds up rigorously across all metrics rather than just being an eyeballed impression from the figures.

Author Response

Comments 1: The claim of novelty rests heavily on being the "first" systematic 25-year assessment for this specific area, but the paper doesn't really engage with how methodologically distinct this is from similar regional MODIS-based GPP/NPP studies elsewhere in the Mediterranean basin. What is actually new here beyond the geographic location?

Response 1: For the first time, an assessment was conducted for a region where no such studies had been conducted. Our study focuses on the Black Sea region, which, although close to the Mediterranean region, has its own characteristics, including serving as a border for the distribution of certain Mediterranean species and landscapes. For the first time, GPP, NPP, and CUE were assessed separately for nine vegetation formations in this region, and a pixel-by-pixel correlation analysis was conducted, highlighting the novelty of the study.

 

Comments 2: Given that MODIS pixels are 500 m and the Karadag reserve is fairly small and topographically complex, how confident are the authors that vegetation-community-level breakdowns aren't smeared by mixed-pixel effects? This seems like it could meaningfully bias the community-type comparisons, especially at boundaries between forest and cultivated land.

Response 2:  Based on the comments from you and the other two reviewers, in order to avoid confusion, we have decided to exclude the reserve area from the analysis in this manuscript. We will provide a more detailed analysis in future publications.

 

Comments 3: The discussion acknowledges the coarse resolution limitation but then still draws fairly specific conclusions about individual vegetation types and even settlement-level correlation reversals. There's a bit of a disconnect between the caveats stated and the confidence of the conclusions drawn from the same data.

Response 3: Based on the comments from you and the other two reviewers, in order to avoid confusion, we have decided to exclude the reserve area from the analysis in this manuscript. We will provide a more detailed analysis in future publications. For the territory of South-Eastern Crimea, the effect of pixel size is less noticeable and is offset by the large area occupied by vegetation types.

Comments 4: The correlation analysis with temperature and precipitation is described as pixel-wise but no information is given on autocorrelation handling or the effective degrees of freedom, which matters a lot for spatial time series like this. Were significance levels adjusted at all?

Response 4: We have corrected the geographical maps and rebuilt them based on your feedback.

 

Comments 5: CUE is computed simply as NPP/GPP without any uncertainty propagation from the underlying MODIS QA flags, so how much of the CUE variability across vegetation types is signal versus retrieval noise?

Response 5:  We fully agree that the use of ready-made MOD17 products, despite their high level of processing, does not guarantee the absence of errors. We are aware that uncertainties are present at all stages, from atmospheric correction to the productivity calculation algorithm itself. However, we would like to point out that some pixels along the coast were excluded during the study, as they cannot provide an objective assessment of the situation on land. Nevertheless, we acknowledge that it is impossible to completely eliminate all errors. We have added a discussion of this limitation to the "Discussion" section, noting that uncertainties in the MODIS input data may contribute to the observed variability of CUE, and our findings should be interpreted with this in mind.

Comments 6: The Theil-Sen and Mann-Kendall trend analysis appears only briefly for vegetation types and isn't applied consistently across GPP, NPP, and CUE series, so it's hard to tell whether the "no long-term trend" conclusion in the abstract holds up rigorously across all metrics rather than just being an eyeballed impression from the figures.

Response 6: These characteristics were calculated only for vegetation types for the NPP indicator.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Most of the objections raised against the first version of the paper have been addressed. However, some formal aspects still require review and, more importantly, it is essential to accurately indicate the significance levels used to distinguish trends and correlations, in order to support the statistical analysis of the results. In the new version, the reference to the territory of the Karadag Nature Reserve should be removed from the conclusions. In the revised manuscript I left comments and suggestions for you to take into account.

P 5 L 138. Separate the subheading from the paragraph.

P12 L 325-326. Are the trends significant or not?

P12 L343. Remove these references.

P14 L372-374. Are the correlations significant? What significance level (p) was used to discriminate significant relationships?

P15 L442. I don't understand the purpose of keeping this reference. This reference should be excluded.

P16 L460-462. Why sub-Mediterranean low-mountain landscapes of Crimea remain sensitive to climate change? I don't understand this sentence, considering that you didn't find any vegetation trends.

Author Response

Comment 1: Most of the objections raised against the first version of the paper have been addressed. However, some formal aspects still require review and, more importantly, it is essential to accurately indicate the significance levels used to distinguish trends and correlations, in order to support the statistical analysis of the results. In the new version, the reference to the territory of the Karadag Nature Reserve should be removed from the conclusions. In the revised manuscript I left comments and suggestions for you to take into account.

Response 1: Dear reviewer, we are grateful to you for taking the time to read the manuscript of the scientific article. Your valuable comments are very important to the authors.

Comment 2: P 5 L 138. Separate the subheading from the paragraph.

Response 2: done

Comment 3: P12 L 325-326. Are the trends significant or not?

Response 3: Yes

Comment 4: P12 L343. Remove these references.

Response 4: done

Comment 5: P14 L372-374. Are the correlations significant? What significance level (p) was used to discriminate significant relationships?

Response 5: The obtained values ​​are significant; we used a p-value < 0.05.

Comment 6: P15 L442. I don't understand the purpose of keeping this reference. This reference should be excluded.

Response 6: done

Comment 7: P16 L460-462. Why sub-Mediterranean low-mountain landscapes of Crimea remain sensitive to climate change? I don't understand this sentence, considering that you didn't find any vegetation trends.

Response 7: Corrected

 

Reviewer 3 Report

Comments and Suggestions for Authors

The revised version is clearer in several places and the regional focus is relevant, but some of the earlier concerns are only partly addressed and a few important methodological issues remain unresolved.

• There are two figure 2 and from figure 3 jump to figure 5.
• The novelty is still framed mainly around this being the first assessment for the study area. The manuscript should explain more clearly what methodological or conceptual insight is gained beyond applying established MODIS productivity products to a previously unassessed region.
• The conclusion that there is no long-term productivity trend is stronger than the analysis presented. Trend testing appears to be reported only for NPP by vegetation type, while equivalent formal results for regional GPP and CUE are not shown.
• The interpretation of higher CUE in agricultural, steppe, and urban areas as an ecological consequence of simplified structure is too confident for a ratio derived from satellite products. This needs to be presented as a tentative interpretation rather than a demonstrated mechanism.
• The methods need to state unambiguously how annual NPP was obtained. The workflow describes processing of the GPP product, but the source product, temporal aggregation, scaling, and any gap-filling applied to NPP are not sufficiently documented.
• The manuscript says that spatial autocorrelation was accounted for through effective degrees of freedom, but provides no implementation details or results. How were the effective sample sizes estimated for each pixel time series, and were temporal autocorrelation and multiple testing across pixels also considered?
• The climate analysis implicitly treats annual temperature and precipitation as direct and independent controls on productivity. In this dry and topographically variable setting, temperature, moisture availability, elevation, radiation, and land-cover composition are likely confounded. Pairwise correlations alone cannot support the causal interpretations currently made.
• Pixel-wise correlations for settlement areas are attributed to anthropogenic activity without land-use or management information. Could these patterns instead reflect mixed pixels, irrigation, impervious-surface contamination, or the different spatial support of the climate datasets relative to MODIS?
• Check the spelling of vegetation-community names in legends.
• Define the treatment of missing and masked pixels in the methods.

Author Response

Comment 1:The revised version is clearer in several places and the regional focus is relevant, but some of the earlier concerns are only partly addressed and a few important methodological issues remain unresolved.

Response 1: Dear reviewer, we are grateful to you for taking the time to read the manuscript of the scientific article. Your valuable comments are very important to the authors.

Comment 2:There are two figure 2 and from figure 3 jump to figure 5.

Response 2: We have corrected the inaccuracies and adjusted the numbering of the figures to match..

Comment 3:The novelty is still framed mainly around this being the first assessment for the study area. The manuscript should explain more clearly what methodological or conceptual insight is gained beyond applying established MODIS productivity products to a previously unassessed region.

Response 3: The novelty of the work lies in the creation of the first 25‑year series (2001–2025) of spatially distributed data on gross (GPP) and net (NPP) primary production for this territory. The use of remote sensing methods and geoinformation modeling has made it possible for the first time to conduct a systematic assessment of the gross and net primary production of the vegetation cover in Southeastern Crimea. Prior to this, systematic calculations of these indicators for Southeastern Crimea using modern satellite data and regionally adapted models were only fragmentary. The sub‑Mediterranean landscapes of Crimea serve as an important reference point for understanding the responses of productivity and carbon use efficiency (CUE) of transitional ecosystems located at the edge of their range to increasing aridification. Regional assessments of GPP and NPP remain insufficiently studied for such ecologically sensitive areas. The novelty is also supported by the conceptual value of the comparison of energy costs and carbon use efficiency (CUE) between natural mesophytic forests and anthropogenically simplified landscapes located within the same complex terrain. At the same time, the authors acknowledge that further fieldwork is required to obtain and compare satellite data and field research data. We have supplemented the discussion section with this information.

Comment 4:The conclusion that there is no long-term productivity trend is stronger than the analysis presented. Trend testing appears to be reported only for NPP by vegetation type, while equivalent formal results for regional GPP and CUE are not shown.

Response 4: Yes, we confirm this..

Comment 5:The interpretation of higher CUE in agricultural, steppe, and urban areas as an ecological consequence of simplified structure is too confident for a ratio derived from satellite products. This needs to be presented as a tentative interpretation rather than a demonstrated mechanism.

Response 5: Thank you for your valuable comment; we have added the corresponding explanation to the manuscript.

Comment 6:The methods need to state unambiguously how annual NPP was obtained. The workflow describes processing of the GPP product, but the source product, temporal aggregation, scaling, and any gap-filling applied to NPP are not sufficiently documented.

Response 6:  To obtain NPP data, the GEE platform was used: https://developers.google.com/earth-engine/datasets/catalog/MODIS_061_MOD17A3HGF?hl=ru#bands. We received already processed annual data.

Comment 7: The manuscript says that spatial autocorrelation was accounted for through effective degrees of freedom, but provides no implementation details or results. How were the effective sample sizes estimated for each pixel time series, and were temporal autocorrelation and multiple testing across pixels also considered?

Response 7: For each pixel, using long‑term series, we:

• calculate the Pearson correlation coefficient;
• build a linear regression and obtain the residuals;
• calculate the first‑order autocorrelation of the residuals;
• determine the effective number of observations (effective degrees of freedom) using the Bailey–Hammersley formula.
• Then, the significance of the correlation is checked using the t‑statistic with a correction for neff: and the adjusted p‑value is calculated.

All calculations are performed pixel by pixel using the app() function from the terra package.

Multiple pixel‑by‑pixel checks were not performed.

 

Comment 8:The climate analysis implicitly treats annual temperature and precipitation as direct and independent controls on productivity. In this dry and topographically variable setting, temperature, moisture availability, elevation, radiation, and land-cover composition are likely confounded. Pairwise correlations alone cannot support the causal interpretations currently made.  Pixel-wise correlations for settlement areas are attributed to anthropogenic activity without land-use or management information. Could these patterns instead reflect mixed pixels, irrigation, impervious-surface contamination, or the different spatial support of the climate datasets relative to MODIS?

Response 8: We thank you for the insightful and substantive comment. We fully agree that pairwise correlations do not allow us to draw direct cause-and-effect conclusions, especially in such a complex and mosaic territory as Crimea, where many factors act simultaneously (elevation, temperature, radiation balance, type of vegetation, soil conditions, anthropogenic load). Our analysis is intended as the first stage for identify the spatial aspects of the relationship between productivity and climatic indicators. We deliberately limited ourselves to pairwise correlation in order to identify the areas where precipitation and temperature may play a leading role, which then require additional field studies.

Comment 9:Check the spelling of vegetation-community names in legends.

Response 9: Done

Comment 10:Define the treatment of missing and masked pixels in the methods.

Response 10: Missing and masked pixels were excluded from the study and replaced with NA..

Round 3

Reviewer 3 Report

Comments and Suggestions for Authors

Dear authors, thank you for the careful and thorough revision, all my previous concerns have been addressed satisfactorily and the manuscript reads much stronger now. I have no further comments, just a couple of closing suggestions. Please do one more pass over the full text for language and grammar polish, confirm that all abbreviations are defined consistently at first use and not redefined later, and double check that table and figure numbering lines up correctly throughout.