Review Reports - Improved Modeling of Gross Primary Production and Transpiration of Sugarcane Plantations with Time-Series Landsat and Sentinel-2 Images

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

I am incredibly glad for the opportunity to review such worthwhile work. The paper is well written, indeed.

1. Remotely sensed data such as satellite imagery are widely used to regularly estimate agricultural production, monitor crop health and for optimal use of natural resources such as water and land to improve food security. In this background, the main question addressed by the research is to investigate the use of satellite data, including Landsat, Sentinel-2, and MODIS, in monitoring and understanding the phenology, Gross Primary Productivity (GPP), and transpiration of sugarcane plantations. The study aimed to assess the accuracy and potential of these data sources and models in tracking sugarcane growth and ecosystem productivity.
2. The topic is relevant in the field of remote sensing, agriculture, and environmental monitoring, as sugarcane is a major crop with economic importance. The study addresses a specific gap in the field by focusing on the application of Landsat and Sentinel-2 data alongside MODIS data to monitor sugarcane crops, especially in regions with tropical and subtropical climates. It also compares different models and data sources, adding to the existing knowledge on crop monitoring.
3. The research adds to the subject area by providing insights into the capabilities and limitations of different satellite data sources and models for monitoring sugarcane. It specifically contributes to understanding phenology, GPP, and transpiration dynamics in sugarcane plantations. By comparing different data sources and models, the study helps to identify the strengths and weaknesses of each approach in different agricultural and climatic settings.
4. The authors should consider several improvements in the methodology:
i. Further validation of the models and data sources with ground truth data to enhance the accuracy of the findings.
ii. Exploring the sensitivity of the models to variations in crop management practices and climatic conditions.
iii. Addressing the sources of error and uncertainty, such as mixed pixels in MODIS data, more comprehensively.
5. The conclusions drawn in the research article appear to be generally consistent with the evidence and arguments presented. The study provides valuable insights into the potential of Landsat, Sentinel-2, and MODIS data for sugarcane monitoring. However, it's essential to recognize the limitations and uncertainties associated with the data sources and models.
6. The references appear to be appropriate and support the claims and findings presented in the article. The authors have cited relevant studies in the fields of remote sensing, agriculture, and environmental science to contextualize their research. However, the document contains two different citation styles. e.g. APA (Celis et al., 2023) and IEEE [4]. The authors are advised to use a consistent reference style.
7. Please add legend to Figure 6.

Comments on the Quality of English Language

Minor editing during proof reading is required.

Author Response

Dear reviewer, thank you so much for your insightful feedback to help us improving our manuscript.

I am incredibly glad for the opportunity to review such worthwhile work. The paper is well written, indeed.

Remotely sensed data such as satellite imagery are widely used to regularly estimate agricultural production, monitor crop health and for optimal use of natural resources such as water and land to improve food security. In this background, the main question addressed by the research is to investigate the use of satellite data, including Landsat, Sentinel-2, and MODIS, in monitoring and understanding the phenology, Gross Primary Productivity (GPP), and transpiration of sugarcane plantations. The study aimed to assess the accuracy and potential of these data sources and models in tracking sugarcane growth and ecosystem productivity.

Response: Thank you for the nice comment and highlighting the significance of our research in the context of using remotely sensed data for agricultural monitoring and resource management.

The topic is relevant in the field of remote sensing, agriculture, and environmental monitoring, as sugarcane is a major crop with economic importance. The study addresses a specific gap in the field by focusing on the application of Landsat and Sentinel-2 data alongside MODIS data to monitor sugarcane crops, especially in regions with tropical and subtropical climates. It also compares different models and data sources, adding to the existing knowledge on crop monitoring.

Response: We appreciate your recognition of the relevance of our study in the fields of remote sensing, agriculture, and environmental monitoring. Sugarcane, indeed, is a crop of significant economic importance, and our research aims to address a crucial gap in its monitoring.

The research adds to the subject area by providing insights into the capabilities and limitations of different satellite data sources and models for monitoring sugarcane. It specifically contributes to understanding phenology, GPP, and transpiration dynamics in sugarcane plantations. By comparing different data sources and models, the study helps to identify the strengths and weaknesses of each approach in different agricultural and climatic settings.

Response: Thank you for acknowledging the contributions of our research to the field of remote sensing in agriculture.

The authors should consider several improvements in the methodology:
i. Further validation of the models and data sources with ground truth data to enhance the accuracy of the findings.

Response: Thank you for your valuable suggestion regarding the validation of our estimates. We fully recognize the importance of ground truth data in enhancing the accuracy of our findings, especially in the context of estimating Gross Primary Production (GPP) at the ecosystem and field scale.

In fact, to date there is not a direct method to measure GPP at the ecosystem scale and this is one of the biggest limitations in the field. To address this, our study relies on the GPP estimates derived from the net ecosystem exchange (NEE) data measured at the eddy covariance (EC) towers. These towers are strategically located at each of the study sites and are renowned for their high-quality data on carbon and water exchanges. The NEE data from these towers is widely regarded as the gold standard data for people to estimate GPP at the ecosystem and landscape levels, the resultant GPP_EC data are used then as a benchmark for comparison and evaluation of the accuracy of GPP estimates from process-based models and data-driven models, including our VPM model.

Exploring the sensitivity of the models to variations in crop management practices and climatic conditions.

Response: This a great comment and we appreciate it. Our study specifically contrasts model performance between sites in Louisiana, USA, and Sao Paolo, Brazil. This comparison is crucial due to the significant differences in climatic conditions, particularly Brazil's higher cloud coverage, which affects the quality of optical data. This situation is a common challenge across tropical regions, where sugarcane is extensively grown. Additionally, we have integrated data from eddy covariance (EC) towers to assess the impact of these varying conditions on the model performance. This approach allows us to effectively evaluate and demonstrate the robustness of our models in diverse agricultural environments.

iii. Addressing the sources of error and uncertainty, such as mixed pixels in MODIS data, more comprehensively.

Response: Thank you for the suggestion. We have tried to improve the clarity of this and mentioned it in sections 4.1., 4.2., and 4.3.

The conclusions drawn in the research article appear to be generally consistent with the evidence and arguments presented. The study provides valuable insights into the potential of Landsat, Sentinel-2, and MODIS data for sugarcane monitoring. However, it's essential to recognize the limitations and uncertainties associated with the data sources and models.

Response: Thank you for the suggestion. We have included this in section 4.3. and in the conclusions section.

The references appear to be appropriate and support the claims and findings presented in the article. The authors have cited relevant studies in the fields of remote sensing, agriculture, and environmental science to contextualize their research. However, the document contains two different citation styles. e.g. APA (Celis et al., 2023) and IEEE [4]. The authors are advised to use a consistent reference style.

Response: Thank you for this comment, we have reviewed the document and ensure that all the references are consistent in the IEEE format.

Please add legend to Figure 6.

Response: We appreciate the suggestion. Nonetheless, in this figure we did not include a legend as the information displayed are the scatter plots between the temperature (daily mean and daily daytime) against the GPP data from the EC tower and for this reason there is no legend on this figure.

Reviewer 2 Report

Comments and Suggestions for Authors

In this paper, Landsat and Sentinel-2 time series images are used to study the gross primary production and transpiration of sugarcane plantations, which has certain reference value for sugarcane planting fields. The paper gives a comprehensive and detailed introduction to the model, data and calculation method, and it is a good experimental report.

However, there is no impressive original work in this paper, and the difference between the model calculation results and the ground observation data is obvious (as shown in Table 4, Table 5 and Figure 9), which can not explain that the model observation results based on remote sensing images can replace or make up for the deficiency of ground observation. In addition, this paper does not put forward any ideas of improving models or methods.

Author Response

Response: Thank you for your thoughtful feedback and insights. We acknowledge your observation and concern regarding the differences that might exist between ground data and satellite-based estimates. Below there are some specific points that address your comments regarding the data differences and the contributions of our work.

Addressing Data Discrepancies: The differences noted in Figure 9 between the evapotranspiration (ET) data from the tower and the transpiration values estimated by our model are expected. In sugarcane agroecosystems, ET typically also includes components like soil evaporation, which are not captured in transpiration-only estimates. Figure 9 is used to represent the seasonal dynamics of the water flux using the ET measured by the EC tower, which is well captured by the T_VTM estimates, and it successfully represents a significant portion of the total ET rate through transpiration estimates as it is expected in this agroecosystem.
Novelty in Model Evaluation: This study introduces, for the first time, the evaluation of the VPM model using a time-series of combined optical data from Landsat and Sentinel-2 for sugarcane monitoring. This novel approach is a significant step forward in remote sensing applications for sugarcane agriculture.
Innovative Use of Site-Specific Optimal Temperature (Topt): We have adopted a novel approach by utilizing site-specific Topt calculated from local data, rather than general biome-specific Topt. This method aligns well with the Topt estimate derived from EC GPP data, demonstrating its effectiveness and contributing significantly to the modeling community. This advancement contributes to the testing and validation of a new version of the VPM model that utilizes site-specific Topt.
Pioneering Application of the VTM Model: Our study is the first to apply the Vegetation Transpiration Model (VTM) to sugarcane crops. This innovative model provides a simple approach to represent this water flux component and its seasonal dynamics aligns well with the seasonal dynamics of ET data as measured at the eddy flux tower sites.
Complementing Ground Observations: While our study does not propose replacing ground observations with satellite-based data, it highlights the capabilities of the VPM and VTM models in providing reliable estimates that effectively capture the crop's carbon and water flux seasonal dynamics.

Our study aimed to provide empirical evidence specific to sugarcane plantations, which have unique characteristics and challenges compared to other croplands. For example, sugarcane is a perennial crop and it is possible to have different ratooning crops in adjacent fields, especially in small plots such as Louisiana.

While it's commonly suggested that higher spatial-resolution satellite images consistently produce more accurate datasets, this isn't always the case—particularly in regions like Brazil, where frequent cloud cover compromises the quality and continuity of high spatial resolution data. Our research underscored the advantages of combining Landsat and Sentinel-2 data to monitor sugarcane crop phenology in locations such as Louisiana (USA) and Sao Paulo (Brazil). Our in-depth analysis reveals that High Spatial Resolution (HSR), especially when combining Landsat and Sentinel-2 datasets, captures the nuances of sugarcane fields more effectively than Moderate Spatial Resolution (MSR) images. This works for small sugarcane fields like in the Louisiana site as well for larger fields like it was in the case of Brazil.

A pivotal aspect of our study was the evaluation of both HSR and MSR-derived EVI to ascertain optimal temperature (Topt). Topt is a crucial parameter for various biophysical processes and is typically generalized by biome type. This generalization poses significant challenges in crop modeling, especially considering that C3 and C4 crops follow different photosynthetic pathways. Moreover, a single crop, such as sugarcane, can exhibit varying Topt values depending on its environmental context as shown in our study. By leveraging eddy covariance data, our study determined the growth optimal temperature and used this benchmark to compare with the Topt derived from EVI at varying spatial resolutions. Notably, our findings are groundbreaking: for both sugarcane sites, the Topt deduced using HSR EVI aligned closely with the data from the eddy covariance tower. This approach could be used at sugarcane plantations in different locations.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript entitled “Improved modeling of gross primary production and transpiration of sugarcane plantations with time series Landsat and Sentinel-2 images” presents a study on the application of moderate (MSR) and high spatial resolution (HSR) satellite images for estimating gross primary production (GPP) and transpiration (T) in sugarcane plantations. The utilization of the Vegetation Photosynthesis Model (VPM) and Vegetation Transpiration Model (VTM) integrated with satellite and on-ground data is worthy, and the scope of the manuscript and the topic are suitable for the journal. However, there are points in which the manuscript should be improved prior to publication:

The main concern is that the manuscript includes unnecessary details but misses an in-depth discussion on time series analysis in precision agriculture, essential for understanding the time-related changes and seasonal effects on GPP and T estimates. For instance, the MODIS data section (lines 490-565) covers well-known information. It's unclear why the authors emphasize these non-new findings, especially since MODIS isn't mentioned in the title.

In general, the introduction and discussion sections need a better focus on time series analysis, as this is the main topic of the paper but is currently weak. The paper should detail image acquisition frequency, gap handling in data, and how models account for seasonal and yearly changes. Moreover, citations should be relevant, not just numerous (for example, in line 61: [30]–[35]), and clearly linked to the content, explaining the relationship between the citation and the article. In addition, the article lacks relevant background on using Landsat or Sentinel 2 time series in precision agriculture:

· “Recognition and Changes Analysis of Complex Planting Patterns Based Time Series Landsat and Sentinel-2 Images in Jianghan Plain, China” https://doi.org/10.3390/agronomy12081773.

· “Potential of Functional Analysis Applied to Sentinel-2 Time-Series to Assess Relevant Agronomic Parameters at the within-Field Level in Viticulture” https://doi.org/10.1016/j.compag.2022.106726.

· “LSFuseNet: Dual-Fusion of Landsat-8 and Sentinel-2 Multispectral Time Series for Permutation Invariant Applications” https://doi.org/10.1109/DSAA60987.2023.10302483

· “Medium-Resolution Multispectral Satellite Imagery in Precision Agriculture: Mapping Precision Canola (Brassica Napus L.) Yield Using Sentinel-2 Time Series” https://doi.org/10.1007/s11119-022-09874-7.

The Materials and Methods section details the study sites, data collection, and models for sugarcane analysis in Brazil and the USA. It includes useful site-specific environmental and operational information. However, the model explanation (specifically time series analysis) is confusing, which is important for understanding the data's temporal dynamics and model accuracy.

In the Results section, the findings on climate, vegetation, carbon fluxes, and their relationships in sugarcane plantations are well-presented with figures and tables.

The Discussion section analyses vegetation indices and the Vegetation Transpiration Model (VTM). Yet, it lacks depth in analysing and discussing the results. A clearer explanation of how time variations were handled and comparisons with other precision agriculture studies is necessary. Furthermore, sections 4.1 and 4.2 are redundant and unclear. Instead, a more detailed discussion is needed on time series analysis and its impact on the accuracy of GPP and transpiration model estimates.

The Conclusions section does not show crucial numerical results. The authors should include these, making the conclusions brief and in bullet points.

Lastly, about the format, the references are not in the journal's required style. For example, “F. Gascon et al.,” is the wrong format. The authors should use the journal's guidelines. Also, they sometimes write “Fig” and other times “Figure” (like “Fig. 3a” and “Figure 4”). They should use the same style throughout the paper.

Specific comments

Line 32

“The introduction lacks sufficient background on time series analysis.”

Reviewer’s comment.

Line 59-61

“However, because of high … plantations [30]–[35].”

This statement could be expanded to explain how the scarcity of EC sites impacts the generalizability of the study's findings. A brief discussion on the representativeness of the chosen sites in light of this limitation would be beneficial.

Line 75

“These GPP data products at MSR are useful, but still too coarse for precision agriculture,”

Indeed. Therefore, it remains unclear why the authors dedicated a significant portion of the paper to discussing MODIS data.

Line 77-84

"Note that most of … available."

Elaborating on how field sizes specifically influence the applicability and accuracy of HSR imagery in GPP estimation would strengthen the argument.

Line 105-114

"The performance of the … and global scales."

This section is critical but lacks specifics about the time series analysis aspect.

Line 154

“24 days after planting.”

Is there a specific reason for this?

Line 182

“into 1-day and 8-day data”

What is the rationale for choosing 1-day and 8-day data intervals and how do these intervals affect the time series analysis, particularly in terms of capturing the phenological changes in sugarcane?

Line 192-219

"2.3. Land Surface Reflectance and Vegetation Index data... reported in previous studies.”

And how does this combination impact the time series analysis? particularly in terms of temporal resolution and data consistency.

Line 216

“with similar reflective wavelengths [87], [89], [90]”

It is not clear why the authors need 3 citations for this statement.

Line 222-224

“VIs were combined together and … (maximum NDVI) as in past studies”

This sentence is unclear and difficult to understand.

Line 278-293

"The Brazil site... (34 mol m-2 day-1 in June) (Fig. 2b)."

How do these seasonal variations influence the interpretation of the GPP and EVI data? The authors should elaborate further on the discussion section.

Line 298-311

"At the Brazil site (Fig. 3c)... with a growing season length of 7 months."

A more detailed discussion is needed on the time series approach for determining the growing season's start and end, and its impact on the study's conclusions.

Line 314-331

"In the Brazil site (Fig. 3a)... demarcated with vegetation indices (Fig. 3d)."

What is the rationale behind “the GPP > 1 g C m-2 day-1 criterion”? Moreover, how anomalies or outliers were handled and what impact this has on the interpretation of the data?

Line 391-407

It is not clear to this reviewer what the authors mean by “relationships”. In this paragraph, there is no single correlation value (r, R2…) that justifies the information stated. Moreover, in Figure 7 where are the relationships? There is no single correlation between data in the figures.

Line 411

“Fig, 8” should be “Fig. 8”.

Line 461

“until August (Tables 4 and 5).”

Why tables 4 and 5 are cited before table 3 (line 469) in the text? The citation of figures and tables should follow a logical order.

Line 489

Since time series analysis is a critical tool that the authors are using in this article (even the word is included in the title), the authors should include a subsection in the discussion section to discuss the implications of using time series in precision agriculture and they should compare their results with the results of other researchers.

Line 496

“and interannual dynamics of vegetation canopies [29], [106]–[109].”

Consider revising the citation style to make it clearer what information is drawn from each referenced work.

Line 502

“of vegetation canopy [110]–[113].”

Similar to the previous comment, clarifying the contribution of each cited article to the statement would be better, especially since the reviewed articles have distinct differences.

Line 604

“exhibited lower values for the”

Please provide R2 values in the same way as you have provided for TVTM_EC

Line 633-660

Why aren't the numbers from the results section included in the discussion? It's important to summarize the most important data from the tables in the results section. The authors should add these to show the full picture of the findings. Also, the conclusions are too long and repetitive. They could be made shorter and clearer, maybe by using bullet points.

Comments on the Quality of English Language

The English in the paper is mostly fine, but there are a few small mistakes. For instance, there are some punctuation errors (like in line 219, "studies. [91].", and in line 411, it should be "Fig. 8" not "Fig, 8"). In addition, some longer sentences need commas to be easier to read.

Author Response

Dear reviewer, thank you so much for all your comments and suggestions to make our manuscript, clearer, and stronger.

The manuscript entitled “Improved modeling of gross primary production and transpiration of sugarcane plantations with time series Landsat and Sentinel-2 images” presents a study on the application of moderate (MSR) and high spatial resolution (HSR) satellite images for estimating gross primary production (GPP) and transpiration (T) in sugarcane plantations. The utilization of the Vegetation Photosynthesis Model (VPM) and Vegetation Transpiration Model (VTM) integrated with satellite and on-ground data is worthy, and the scope of the manuscript and the topic are suitable for the journal. However, there are points in which the manuscript should be improved prior to publication:

Response: Thank you for your valuable feedback and pointing out this concern. Regarding the emphasis on moderate spatial resolution (MSR) data, our intention was to provide a comprehensive context for understanding the current state of GPP modeling in agriculture, where the available data products are available at 500m spatial resolution.

In our study, the USA site was representative of a plantation with multiple small fields, while Brazil site represents a location with frequent cloud coverage but larger fields. This helped us evaluate the potential of HSR for this specific type of cropland. Our findings reveal that, despite the challenges posed by cloud coverage in Brazil, the combined time series of Landsat and Sentinel-2 data effectively compensated for these limitations. This was evident in the strong linear relationship observed between the Enhanced Vegetation Index (EVI) and GPP data from the eddy covariance (EC) tower at the Brazil site. This outcome is particularly noteworthy as it contrasts with the usual preference for MSR data in such environments.

We believe these insights contribute significantly to the understanding of remote sensing applications in agriculture, particularly in varying crop and climatic contexts. Our study demonstrates the potential of HSR data in providing detailed and continuous time series information, even in challenging environments, thereby enhancing the precision and accuracy of crop monitoring and modeling.

In general, the introduction and discussion sections need a better focus on time series analysis, as this is the main topic of the paper but is currently weak. The paper should detail image acquisition frequency, gap handling in data, and how models account for seasonal and yearly changes.

Response: Thank you for your constructive feedback emphasizing the need for a stronger focus on time series analysis in the introduction and discussion sections of our paper. We acknowledge the importance of this aspect as it forms the core of our study. In response to your comment, we have enriched the introduction and discussion sections with additional insights into time series analysis.

In Section 2.3, we have detailed the temporal resolution of the various datasets used, including aspects of quality control. To maintain the integrity of our data, we consciously avoided gap filling, a decision we now elaborate on in this section. This approach ensures that our model's inputs are based on actual satellite observations, enhancing the reliability of our findings.

Recognizing that seasonal variability unfolds over time, we ensured a minimum of one high-quality observation per week. This frequency aligns with the temporal resolution of commonly used MODIS data products and is crucial for capturing seasonal dynamics into our model through inputs like the Enhanced Vegetation Index (EVI) and the Land Surface Water Index (LSWI).

Furthermore, the high spatial resolution (HSR) data occasionally provided additional high-quality observations within a week, thereby enriching our monitoring capabilities. This increased frequency of quality data inputs not only enhances the monitoring process but also introduces greater variability into the model, allowing for a more nuanced understanding of seasonal and annual changes in sugarcane plantations.

Moreover, citations should be relevant, not just numerous (for example, in line 61: [30]–[35]), and clearly linked to the content, explaining the relationship between the citation and the article.

In addition, the article lacks relevant background on using Landsat or Sentinel 2 time series in precision agriculture:

“Recognition and Changes Analysis of Complex Planting Patterns Based Time Series Landsat and Sentinel-2 Images in Jianghan Plain, China” https://doi.org/10.3390/agronomy12081773.
“Potential of Functional Analysis Applied to Sentinel-2 Time-Series to Assess Relevant Agronomic Parameters at the within-Field Level in Viticulture” https://doi.org/10.1016/j.compag.2022.106726.
“LSFuseNet: Dual-Fusion of Landsat-8 and Sentinel-2 Multispectral Time Series for Permutation Invariant Applications” https://doi.org/10.1109/DSAA60987.2023.10302483
“Medium-Resolution Multispectral Satellite Imagery in Precision Agriculture: Mapping Precision Canola (Brassica Napus L.) Yield Using Sentinel-2 Time Series” https://doi.org/10.1007/s11119-022-09874-7.

Response: Thank you for your valuable feedback regarding the use of citations in our manuscript. We have added some of the references you recommended and few more relevant to precision agriculture in our introduction and discussion sections. We understand the importance of ensuring that each citation is not only relevant but also clearly linked to the content, elucidating the relationship between the cited work and our study.

Regarding the specific citations (30 to 35) mentioned in line 61, these references were included to highlight studies that have utilized eddy covariance (EC) data in sugarcane plantations. Given the limited number of studies in this specific area, each of these citations was chosen to underscore the existing body of work related to our research focus.

In light of your feedback, we have carefully reviewed our manuscript to ensure that all citations are directly pertinent to the work presented. Also, we appreciate the recommendations of manuscripts that you suggested that use Landsat and Sentinel-2 in the context of precision agriculture, this and other relevant manuscripts have been added across the manuscript to enhance our article.

The Materials and Methods section details the study sites, data collection, and models for sugarcane analysis in Brazil and the USA. It includes useful site-specific environmental and operational information. However, the model explanation (specifically time series analysis) is confusing, which is important for understanding the data's temporal dynamics and model accuracy.

Response: Thank you for your feedback regarding the Materials and Methods section of our manuscript. We appreciate your acknowledgment of the detailed site-specific information provided. However, we understand your concern about the clarity of the model explanation, particularly regarding time series analysis, which is crucial for comprehending the temporal dynamics of the data and the accuracy of our models.

In response to your comment, we will revise the Materials and Methods section to provide a clearer and more comprehensive explanation of the collection, processing, and use of the time series of optical data utilized in our manuscript. Our aim is to enhance the reader's understanding of how we utilized time series data to analyze the temporal dynamics of sugarcane growth (growing season length), productivity, but more importantly the applicability of them to estimate vegetation carbon uptake.

In the Results section, the findings on climate, vegetation, carbon fluxes, and their relationships in sugarcane plantations are well-presented with figures and tables.

Response: Thank you for your suggestion, we think that this is very important to help us improve our manuscript. We have tried to incorporate your feedback into the discussion section. Our section 4.1. focuses on the vegetation indices from MSR and HSR in tracking the land surface phenology of sugarcane over both sites. On the other hand, section 4.2. focuses on the GPP estimates at both HSR and MSR. Following your suggestion, we added some statements connecting the influence of the VIs time variation into the model precision.

The Conclusions section does not show crucial numerical results. The authors should include these, making the conclusions brief and in bullet points.

Response: This is a great suggestion, we have added the more relevant outcomes of our study into the conclusion section in bullet points, but to avoid being repetitive or increasing too much the extent of the section, we tried to focus on the take home messages from our study.

Response: Thank you for this comment, we have reviewed the different citations in the manuscript and corrected all of the ones that were in the wrong format. Please note that we used across the manuscript the abbreviated term across the document (Fig. #), but for the sections where we were expanding deeper on the figure, we used the full expression.

Specific comments

Line 32. “The introduction lacks sufficient background on time series analysis.”

Response: Thank you for pointing this out and we agree that our introduction could benefit on further expanding on this. Following your advice some additional information was added into the introduction, highlighting the following elements of time series analysis done in our study.

Collection of Sequential Data: Our study involves gathering time-ordered satellite data from MODIS and Landsat-Sentinel. This data is a series of observations recorded at consistent time intervals.
Analysis of Temporal Patterns: By evaluating the EVI derived from each satellite image over time, we analyzed how this index changes throughout the sugarcane growing season. This is a key aspect of time series analysis, where we looked for patterns, trends, or seasonal variations in the data.
Comparison with Ground Data: We compared the time series of satellite-derived EVI data with GPP data from EC towers. This comparison helps in understanding how well the satellite data captures the real-time changes happening in the sugarcane fields, which is crucial for validating the accuracy of remote sensing.
Model Input and Performance Evaluation: Using the satellite data as input into for our models (VPM and VTM) and comparing the model outputs with ground observations is another form of time series analysis done in our manuscript. Here, we assessed the performance of the models over time, checking how well they simulate the GPP and transpiration of the sugarcane plantations across different years.
Understanding Seasonal and Yearly Changes: Our approach of ensuring at least one good quality observation per week from the satellite data is a method to capture the seasonal dynamics in the plantations. This frequent observation helps in understanding how GPP and transpiration rates vary with seasonal changes, which is a critical aspect of time series analysis in agriculture.

In summary, our study utilizes time series analysis by collecting, analyzing, and modeling satellite data over time to understand and predict the temporal dynamics of GPP and transpiration in sugarcane plantations. This approach is essential for precision agriculture, where understanding the temporal changes in crop conditions is key to effective monitoring and management. Some of these details were added in L113-124.

Line 59-61. “However, because of high … plantations [30]–[35].” This statement could be expanded to explain how the scarcity of EC sites impacts the generalizability of the study's findings. A brief discussion on the representativeness of the chosen sites in light of this limitation would be beneficial.

Response: Thank you for your suggestion to expand on the statement regarding the scarcity of eddy covariance (EC) sites and its impact on the generalizability of our study's findings. We recognize the importance of addressing this limitation and its implications for the representativeness of our chosen sites. Following your recommendation, we have added some additional details.

Line 75. “These GPP data products at MSR are useful, but still too coarse for precision agriculture,” Indeed. Therefore, it remains unclear why the authors dedicated a significant portion of the paper to discussing MODIS data.

Response: Thank you for mentioning this element. MODIS data product is still some of the most used datasets and to date it is still widely being used for agricultural applications, despite the number of limitations. The current NASA MOD17 GPP data product is in fact at 500m spatial resolution. In addition, while it's commonly suggested that higher spatial-resolution satellite images consistently produce more accurate datasets, this isn't always the case—particularly in regions like Brazil, where frequent cloud cover compromises the quality and continuity of high spatial resolution data. Our research underscored the advantages of combining Landsat and Sentinel-2 data to monitor sugarcane crop phenology in locations such as Louisiana (USA) and Sao Paulo (Brazil). Our in-depth analysis reveals that High Spatial Resolution (HSR), especially when combining Landsat and Sentinel-2 datasets, captures the nuances of sugarcane fields more effectively than Medium Spatial Resolution (MSR) images. This for small sugarcane fields like in the Louisiana site as well for larger fields like it was in the case of Brazil, where the pixel size was closer to MODIS spatial resolution.

Line 77-84. "Note that most of … available."

Elaborating on how field sizes specifically influence the applicability and accuracy of HSR imagery in GPP estimation would strengthen the argument.

Response: This is a great suggestion; we have added a short sentence to make the argument more compelling and stronger.

Line 105-114. "The performance of the … and global scales."

This section is critical but lacks specifics about the time series analysis aspect.

Response: Thank for this great comment, we have added different elements specific to the time series generation and application relevant of this process in the study.

Line 154. “24 days after planting.” Is there a specific reason for this?

Response: Thank you for this query. The inclusion of this detail is significant for several reasons:

Data Accuracy and Context: Mentioning the installation timing provides context for the data collected. It helps readers understand the stage of crop growth during which the flux measurements began, which is crucial for interpreting the data accurately.
Methodological Transparency: Detailing the timing of the flux tower installation enhances the transparency of our methodology. It allows other researchers to understand the conditions under which the data was collected, which is essential for replicating the study or comparing its results with another research.
Understanding Crop Development Stages: The early growth stage of sugarcane, which begins shortly after planting, is a critical period for carbon and water flux dynamics. By specifying that the flux tower was installed 24 days after planting, we highlight that our study captures this important phase, providing valuable insights into the crop's developmental processes.
Baseline for Analysis: This timing sets a baseline for our analysis of the flux data. It indicates the starting point of our observations in relation to the sugarcane planting cycle, which is important for any temporal analysis or modeling work based on this data.

In summary, mentioning the specific timing of the flux tower installation is relevant as it adds clarity to our research methodology, ensures transparency, and aids in the accurate interpretation of the data collected.

Line 182. “into 1-day and 8-day data”. What is the rationale for choosing 1-day and 8-day data intervals and how do these intervals affect the time series analysis, particularly in terms of capturing the phenological changes in sugarcane?

Response: Thank you for your insightful comment regarding our choice of 1-day and 8-day data intervals in the time series analysis. The rationale behind selecting these specific intervals is closely tied to the temporal resolution of the satellite data and the objective of capturing phenological changes in sugarcane effectively.

Matching Temporal Resolution of MSR Data: The moderate spatial resolution (MSR) data with an 8-day temporal resolution. To align with this, we used 8-day averages for the other inputs and our analysis.
Accommodating HSR Data Frequency: High spatial resolution (HSR) data from Landsat and Sentinel-2 have a more irregular temporal frequency, ranging between 4 to 8 days, depending on quality conditions. To utilize this data effectively and maintain temporal consistency, we calculated estimates using daily values. This approach allows us to match the actual observation frequency of the HSR data, ensuring that our analysis is as close to the real-time acquisition as possible.

By employing these two intervals, we aim to leverage the strengths of both MSR and HSR data, ensuring a comprehensive and accurate representation of the sugarcane's phenological stages and growth patterns.

Line 192-219. "2.3. Land Surface Reflectance and Vegetation Index data... reported in previous studies.” And how does this combination impact the time series analysis? particularly in terms of temporal resolution and data consistency.

Response: Thank you for your question regarding the impact of combining land surface reflectance and vegetation index data from different sources on our time series analysis. In fact, the temporal consistency of the HSR was one of the elements that our study aimed to demonstrate.

Our approach in this regard was significantly informed by the quality control processes and methodologies reported in previous studies, particularly those focusing on the combined use of Landsat and Sentinel-2 data. The quality control processes for both moderate spatial resolution (MSR) and high spatial resolution (HSR) data, as established in prior research, provided a robust foundation for our study. By adopting these tried-and-tested methods, we ensured the reliability and accuracy of our data processing.

The combined used of both Sentinel-2A/B provides a 5-day temporal resolution and the added combination of Landsat data, when excluding overlapping days can improve it to 2 to 4 temporal resolutions. However, because of the data quality, there was a consistent trend of a 5 to 8-day temporal resolution, where there was for every week at least 1 day with good quality observations.

The insights gained from previous studies on the combined use of Landsat and Sentinel-2 were instrumental in guiding our data processing and combination techniques. These studies demonstrated effective ways to integrate data from these two sources, which was pivotal for our analysis.

Also, because our study focused on evaluating estimates within the tower footprint. Given the similarities in spectral and radiometric resolution between Landsat and Sentinel-2, we were able to treat observations from both datasets as if they were from a single source. This was feasible because one Sentinel pixel typically falls within one Landsat pixel, which in turn falls within the footprint of the tower.

In regards the impact on time series analysis, the combination of these datasets allowed us to maintain a high level of temporal resolution and data consistency. By effectively merging Landsat and Sentinel-2 observations, we could capture more frequent and detailed snapshots of the sugarcane plantations, enhancing our ability to monitor and analyze their phenological changes over time.

The successful application of this integrated approach in previous studies provided the scientific basis and confidence for us to employ it in our research. This methodology has proven to be effective in ensuring a comprehensive and accurate time series analysis, particularly in terms of capturing the temporal dynamics of GPP and transpiration in sugarcane plantations.

Line 216. "with similar reflective wavelengths [87], [89], [90]”. It is not clear why the authors need 3 citations for this statement.

Response: Thank you for your query regarding the use of three citations to support the statement about similar reflective wavelengths in Landsat and Sentinel-2 data. The rationale behind multiple citations is to provide a robust and well-substantiated basis for our methodology, particularly in the context of integrating data from these two satellite sources.

Line 222-224. “VIs were combined together and … (maximum NDVI) as in past studies”

This sentence is unclear and difficult to understand.

Response: Thank you for pointing this out, we have improved the writing in this section.

Line 278-293. "The Brazil site... (34 mol m-2 day-1 in June) (Fig. 2b)."

How do these seasonal variations influence the interpretation of the GPP and EVI data? The authors should elaborate further on the discussion section.

Response: Thank you for your insightful query. The seasonal variations observed at the Brazil site indeed play a crucial role in interpreting the Gross Primary Production (GPP) and Enhanced Vegetation Index (EVI) data. In the discussion section, we will elaborate on this aspect as follows:

The seasonal variations, characterized by changes in temperature, sunlight, and rainfall, directly influence the photosynthetic activity of sugarcane. For instance, higher GPP values during peak growth months reflect increased photosynthesis due to optimal growing conditions. Conversely, lower GPP values in other months indicate reduced photosynthetic activity, which could be due to factors like lower sunlight or temperature variations. Understanding these seasonal patterns is essential for interpreting GPP data in the context of the crop's physiological responses to environmental conditions.

Line 298-311. "At the Brazil site (Fig. 3c)... with a growing season length of 7 months."

A more detailed discussion is needed on the time series approach for determining the growing season's start and end, and its impact on the study's conclusions.

Response: Thank you for pointing out, we have added several complementary statements into the discussion in section 4.1. to address the growing season length using the VI-based method

Line 314-331. "In the Brazil site (Fig. 3a)... demarcated with vegetation indices (Fig. 3d)."

What is the rationale behind “the GPP > 1 g C m-2 day-1 criterion”? Moreover, how anomalies or outliers were handled and what impact this has on the interpretation of the data?

Response: Thank you for your questions regarding our methodology at the Brazil site. The rationale behind using the criterion of "GPP > 1 g C m-2 day-1" and our approach to handling anomalies are as follows:

Rationale for GPP Criterion: The criterion of "GPP > 1 g C m-2 day-1" was established as a threshold to identify active photosynthetic periods of the sugarcane crop. This specific value was chosen based on prior research and field observations indicating that having 3 consecutive GPP values above this threshold typically correspond to the starting period of vegetative growth and active carbon assimilation. By setting this criterion, we aimed to focus our analysis on the key phases of crop growth, ensuring that our interpretations and conclusions are centered around the most productive and biologically active periods.
Handling Anomalies and Outliers: In our dataset, anomalies or outliers in GPP were identified using the net ecosystem exchange (NEE) data. These anomalies could arise due to various factors, such as sensor errors, atmospheric conditions, or unusual environmental events. One outlier was identified for the sugarcane 2018 period. The site manager re-processed data and the anomaly was still present. Since the EC estimates are the standard reference, this was considered in the analyses and mentioned in the discussion as it introduces some uncertainty for that specific year. The information were retained and analyzed as part of the natural variability in the dataset.

In the manuscript, we will elaborate on these aspects to provide a clearer understanding of our methodology and its implications for the interpretation of GPP and vegetation index data.

Line 391-407. It is not clear to this reviewer what the authors mean by “relationships”. In this paragraph, there is no single correlation value (r, R2…) that justifies the information stated. Moreover, in Figure 7 where are the relationships? There is no single correlation between data in the figures.

Response: Thank you for your feedback and for pointing out the need for clarity in our discussion of the relationships between vegetation indices and air temperature. I understand your concern regarding the absence of specific correlation values (like r or R²) in the paragraph and the figures.

In the manuscript, when we refer to "relationships," in this specific section, we are discussing the observed trends and patterns between the Enhanced Vegetation Index (EVI) from both Landsat-Sentinel-2 (EVI_LS-S2) and MODIS (EVI_MOD), and air temperature (T_DT and T_DA). These relationships were explored as a potential method to calculate the optimal air temperature (Topt) for sugarcane growth. We did not calculate the correlation values as the scatter plots in figure 7 are intended to illustrate the temperature value range at which the cluster of highest EVI values take place. This was used to determine the site-specific optimal growth temperature.

Line 411. “Fig, 8” should be “Fig. 8”.

Response: Thank you for highlighting this, we have corrected it.

Line 461. “until August (Tables 4 and 5).” Why tables 4 and 5 are cited before table 3 (line 469) in the text? The citation of figures and tables should follow a logical order.

Response: This is absolutely right, we have addressed this by mentioning table 3 appropriately in this section before tables 4 and 5.

Line 489. Since that the authors are using in this article (even the word is included in the title), the authors should include a subsection in the discussion section to discuss the implications of using time series in precision agriculture and they should compare their results with the results of other researchers.

Response: We agree that time series analysis is a great tool, but we just focused on putting the time series data together to use as input, we gap filled the data, and made sure to keep a temporal consistency, but we didn’t delve into the quantitative analysis of time series. This is a valuable comment, and we appreciate the suggestion. We have expended in the discussion some of the implications in precision agriculture as recommended. On the other hand, it is important to note that one of our main goals was to evaluate the capabilities of a modeling-based tool like the VPM model. In this context, since this specific model have been previously compared with other modeling approaches (e.g. Zhang et al., 2017) and results from other researchers, we focused on evaluating the changes in the performance when fed with vegetation indices at different spatial resolutions, over two representative sugarcane sites. One of the contributions for precision agriculture is that the optimal temperature value, which is widely consider as biome specific, using the eddy covariance data, we were able to display that EVI could be used as a successful driver to estimate the site-specific optimal crop growth temperature, which can be used in the VPM and any other model that uses this information in the parametrization. Section 4.1. further discusses the analysis and the results of the vegetation time series used as input in the model.

Line 496

“and interannual dynamics of vegetation canopies [29], [106]–[109].”

Consider revising the citation style to make it clearer what information is drawn from each referenced work.

Response: We agree that a different citation style might make it clearer to follow. However, per journal guidelines we have to adhere in using the IEEE style.

Line 502

“of vegetation canopy [110]–[113].”

Similar to the previous comment, clarifying the contribution of each cited article to the statement would be better, especially since the reviewed articles have distinct differences.

Response: This is a good suggestion, we tried to make it easier to follow across existing references and some the newly added.

Line 604

“exhibited lower values for the”

Please provide R2 values in the same way as you have provided for TVTM_EC

Response: This information has been added as suggested.

Line 633-660

Response: Thank you for your suggestion, we have incorporated some of the main numbers from the results into our discussion section. Also, we added the bullet points into the conclusion per your suggestion.

Comments on the Quality of English Language

Response: Thanks for the recommendations, we have addressed and checked the above-mentioned punctuation errors.

Reviewer 4 Report

Comments and Suggestions for Authors

There are references with the wrong format in the introduction, for example check lines 98, 100 and 106. Review the whole text.

line 257 fix CO2 typo same as line 260 review across the whole text.

Figure 8 improve visualization of the legend in the left plots.

Check format for the font type and size in all figures and tables.

Author Response

Dear reviewer, thank you so much for your feedback!

Comments and Suggestions for Authors

There are references with the wrong format in the introduction, for example check lines 98, 100 and 106. Review the whole text.

Response: Thank you for pointing this out. We have addressed the references mismatch across the manuscript.

line 257 fix CO2 typo same as line 260 review across the whole text.

Response: Thanks for highlighting this, we have addressed these typos.

Figure 8 improve visualization of the legend in the left plots.

Response:

Check format for the font type and size in all figures and tables.

Response: Thanks for highlighting this, we have reviewed the font types across the document.

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The author gave a satisfactory answer to the questions, and at the same time revised the paper, perfecting the previous shortcomings.

Reviewer 3 Report

Comments and Suggestions for Authors

The authors have responded appropriately to all prior feedback and questions.

Comments on the Quality of English Language

Authors are advised to thoroughly revise the English language used throughout the manuscript. For instance, in line 357, they have written "EVI reached below 0.1," which is awkward. A more appropriate phrasing would be "EVI fell to below 0.1."