Biome-Specific Estimation of Maximum Air Temperature Using MODIS LST in the São Francisco River Basin

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

kindly find the review report 

Comments for author File: Comments.pdf

Author Response

Answer to reviewer 1:

Thank you for your valuable comments and insights. We appreciate the time you took to review our work. Below are the changes we made in response to your feedback.

The research is relevant, timely, and valuable, particularly for regional climate studies and environmental policy-making in semiarid and tropical regions. The methodology is robust and the use of extensive datasets over two decades strengthens the reliability of the results. However, several areas need enhancement for better scientific communication and clarity.

Title: Slightly long and generic.

Consider simplifying to: "Biome-Specific Estimation of Maximum Air Temperature Using MODIS LST in the São Francisco River Basin".

We simplified the title as suggested. Now it is:

"Biome-Specific Estimation of Maximum Air Temperature Using MODIS LST in the São Francisco River Basin".

 Abstract: Dense and overly technical.

 Add a clearer motivation and outcome summary. Reduce technical details; highlight results and practical implications.

We revised the abstract to be clearer and more accessible, reflecting the study's motivation, results, and practical implications. Here is the updated version:

“The São Francisco River provides water for agriculture, urban areas, and hydroelectric power generation, benefiting millions of people in Brazil. Its basin supports various species, some of which are endemic and rely on its unique habitats for survival. Currently, monitoring maximum air temperature in the São Francisco River Basin is limited due to sparse weather stations. This study proposes three linear regression models to estimate maximum air temperature using satellite-derived land surface temperature (LST) from MODIS/Aqua across the basin’s three main biomes: Caatinga, Cerrado, and Mata Atlântica. With over 94,000 paired observations of ground and satellite data, the models showed good performance, accounting for 46% to 54% of temperature variation. Cross-validation confirmed reliable estimates with errors below 2.7°C. The findings demonstrate that satellite data can improve air temperature monitoring in areas with limited ground observations and suggest that the proposed biome-specific models could assist in environmental management and water resource planning in the São Francisco River Basin. This includes providing more informed policies for climate adaptation and sustainable development or analysing variations in maximum air temperature in arid and semi-arid regions to contribute to desertification mitigation strategies in the São Francisco River Basin.”

Introduction: Repetitive and wordy in places.

Remove redundancies (e.g., restating satellite capabilities). Clarify the gap in prior research earlier.

We have now (1) removed redundant statements about satellite capabilities and condensed repetitive descriptions of the São Francisco River Basin’s importance; (2) restructured the introduction to clearly identify the research gap in the third paragraph, emphasizing: The lack of biome-specific LST–Tair validation in Brazil’s policy-relevant land cover data and the limitations of prior pan-regional studies for water management applications; (3) tightened the language throughout, reducing wordiness by simplifying the climate context and merged related methodological points.

We hope to have clarified the gap in previous research by adding the following:

“Despite advances in modeling approaches-including machine learning and neural networks for air temperature estimation from satellite data [19–24] - there remains a gap: few studies have evaluated how biome-specific land cover classifications influence the LST–air temperature relationship within major river basins such as the SFRB. This is particularly relevant as Brazil’s official land cover database is increasingly used for water management and policy decisions. With the imminent completion of the São Francisco River Integration Project, designed to alleviate water shortages in semi-arid regions [25], the need for robust temperature monitoring tools has become urgent.

This study addresses this gap by developing and validating biome-partitioned linear regression models to estimate maximum air temperature from MODIS MYD21A1D LST retrievals across the three dominant biomes of the SFRB, as defined by the Brazilian environmental agency [26]. Our approach leverages the official land cover database used in policy-making, ensuring that our findings are directly applicable to regulatory and management frameworks. Additionally, by providing the first assessment of how biome- specific land cover classes modulate the relationship between satellite-derived LST and ground-based maximum air temperature in the SFRB, this work advances beyond previous studies that focused on continental or homogeneous landscapes. Our analysis quantifies the extent to which biome-partitioned models improve air temperature estimation accuracy over conventional, non-stratified approaches, thereby supporting finer-scale water resource management and climate adaptation planning. By validating these models against ground observations, we offer actionable background for integrating remote sensing into operational monitoring systems, especially in data-scarce semiarid environments.”

Methods: Some subsections are overly descriptive and technical.

Consider summarizing the satellite data characteristics more concisely and moving details to supplementary material.

To address the reviewer’s comment about overly descriptive and technical content in the Methods section, we revised version of the MODIS LST data and Data processing subsections, which are the most technical. These edits condense the content and suggest what was moved to the recently created section of Supplementary Material. They are now written as follows:

MODIS LST data

We used land surface temperature (LST) data from the Aqua satellite's MODIS sensor, specifically the MYD21A1D product, which provides daily global LST retrievals at 1 km spatial resolution. The product is based on the MOD21 algorithm, designed to improve LST estimates in arid and semiarid regions through enhanced atmospheric correction and emissivity separation.

We selected Aqua/MODIS over Terra/MODIS due to its 13:30 local overpass time, which is more representative of peak solar heating and thus closer to the daily maximum LST. The MYD21A1D product uses observations from thermal infrared bands and includes only cloud-free pixels with high-quality LST estimates. We used version 6.1 of the product, covering the period from July 4, 2002, to the present.

Details on MODIS spectral bands, algorithm characteristics, and quality filtering criteria are provided in the Supplementary Material (Section S1).

Data processing

We extracted daily LST data from MYD21A1D at the locations of 48 INMET weather stations within the São Francisco River Basin (SFRB). MODIS data were converted from digital counts to degrees Celsius using standard scaling factors. Only cloud-free, high-quality observations were retained. We aligned LST data with daily maximum air temperature records from weather stations, removing pairs with missing values.

The process of tile selection, mosaicking, and handling of missing MODIS values is detailed in the Supplementary Material (Section S2).

Supplementary Material

S1. MODIS LST Data Product Details

The MYD21A1D product is a level 3 daily global dataset generated from the Aqua satellite’s MODIS sensor. It is based on the MOD21 algorithm, which employs:

Temperature Emissivity Separation (TES): Separates surface temperature from surface emissivity.

Water Vapor Scaling Correction: Reduces cold bias, especially over dry or semiarid regions like the SFRB.

Key characteristics:

Sensor: MODIS aboard Aqua

Spectral bands used: Bands 29, 31, and 32 (thermal infrared)

Spatial resolution: 1 km

Temporal resolution: Daily, daytime granules

Overpass time: ~13:30 local time

Product version: Collection 6.1

File format: Sinusoidal gridded tiles

Data source: NASA’s LP DAAC archive

This product was chosen over Terra/MODIS (which overpasses at 10:30 local time) to better match the timing of maximum solar irradiance and likely maximum LST.

S2. MODIS Data Extraction and Preprocessing

To extract MYD21A1D data over the São Francisco River Basin, we performed the following:

 

Tile Selection: Five MODIS tiles covering the SFRB were identified: h13v11, h13v10, h14v10, h13v9, h14v9.

Mosaicking: Tiles were merged into a single raster mosaic for each day.

Quality Control: MOD21A1 processing includes filters for: Cloud contamination; Poor LST or emissivity accuracy; Tiles with <15% valid coverage; Missing data handling: Cells with missing or low-confidence data were excluded.

Conversion: MODIS LST values were scaled using the factor 0.02 (to Kelvin) and then converted to Celsius by subtracting 273.15.

Daily LST values were extracted at the coordinates of 48 INMET weather stations. These series were matched with corresponding air temperature records for subsequent regression analysis. All records with missing values in either dataset were discarded.

 

Results: Text is largely descriptive of tables/figures.

Add more interpretation of what these results mean in practical or ecological terms.

We have substantially revised the Results section to incorporate more ecological and practical interpretation of our findings. Specifically, we have clarified how vegetation structure, canopy density, and climatic variability across biomes influence the strength of the LST–air temperature relationship and discussed how biome-specific regression performance reflects surface energy balance dynamics and surface–atmosphere coupling.

Discussion: Lacks in-depth exploration of implications.

Discuss implications for climate monitoring, water resource planning, and potential application in other basins or biomes.

In the revised version of the manuscript, we have substantially strengthened the Conclusions section to clarify the broader implications of our findings. Specifically, we now emphasize: the operational suitability of MODIS LST as a proxy for daily maximum air temperature (Tₘₐₓ), particularly in data-scarce or climatically vulnerable regions, such as part of the SFRB, the policy relevance of these results for climate monitoring, drought risk assessment, water resource planning, and biome-specific conservation strategies.

 

Conclusion: Could benefit from a stronger take-home message.

Emphasize policy relevance and the suitability of using LST as a proxy for Tmax.

In the revised version of the manuscript, we have substantially strengthened the Conclusions section to clarify the broader implications of our findings. Specifically, we now emphasize: the operational suitability of MODIS LST as a proxy for daily maximum air temperature (Tₘₐₓ), particularly in data-scarce or climatically vulnerable regions, such as part of the SFRB, the policy relevance of these results for climate monitoring, drought risk assessment, water resource planning, and biome-specific conservation strategies.

 

Figures & Tables: Figure 1 caption is too technical.

Simplify and explain better for non-specialist readers. Clearly mark regression lines and equality lines.

Done.

Final Recommendation.

With moderate revisions, this paper can make a meaningful contribution to the fields of remote sensing, climatology, and regional planning. The technical foundation is solid, but clearer communication and tighter structuring will significantly enhance its impact.

We sincerely thank the reviewer for their constructive assessment. We have revised the manuscript to improve clarity and structure, especially in the Results, Discussion, and Conclusion sections.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript reports the linear regression modeling of air temperature measurements in the São Francisco River Basin in Brazil, at three different biomes, using remote sensing MODIS land surface temperature. Results show a positive and relevant relationship between both temperatures, evidencing the possibility of using land surface temperature as a proxy for air temperature in the region. Nevertheless, the manuscript could benefit from some improvements. Comments and suggestions are attached to this message.

Comments for author File: Comments.pdf

Comments on the Quality of English Language

The english is overall good. The text only needs small corrections.

Author Response

Answer to reviewer 2:

Title: The title can’t be in all caps; therefore, it should be corrected to the proper formatting.

We would like to express our gratitude for your comments and insights. We deeply appreciate the time and effort you invested in reviewing our work.

We have corrected the formatting in accordance with the journal's editorial standards.
Additionally, the title was reformulated based on Reviewer 1’s suggestion, aiming to improve its clarity and conciseness.

"Biome-Specific Estimation of Maximum Air Temperature Using MODIS LST in the São Francisco River Basin."

Abstract: The abstract should include a brief introduction of the context and motivation of the study, as well as a brief conclusion and look into the future. Furthermore, the language should be more formal (passive writing) and avoid first-person writing. This applies to the rest of the manuscript.

Based on your feedback — as well as complementary comments from the other reviewers — we have revised the abstract to improve its clarity, and structure. Specifically, we added a brief contextual introduction to highlight the motivation and relevance of the study, clarified the main findings.

Keywords: Keywords should not be present in the abstract and should have all the same formatting.

We add the keywords using MDPI’s latex template.

1. Introduction: The introduction should describe what biomes are and their respective importance.

“To validate the well-established relationship between remote sensed LST and ground-based air temperature for the policy-making land cover database over the SFRB.”

Why is this validation necessary if the relationship is well-established?

Please provide more references as well, preferably more international ones.

Other observations:

• Line 34: spelling mistake “inclunding”
• Line 46: spelling mistake “build”

We thank the reviewer for pointing out these spelling errors. The mistakes on lines 34 (“inclunding”) and 46 (“build”) have been corrected in the revised manuscript.

“and automatic weather stations to measure several meteorological variables, including".

“heat flux, LST and air temperature, however neither relationships were built across the”.

We have now (1) removed redundant statements about satellite capabilities and condensed repetitive descriptions of the São Francisco River Basin’s importance; (2) restructured the introduction to clearly identify the research gap in the third paragraph, emphasizing: The lack of biome-specific LST–Tair validation in Brazil’s policy-relevant land cover data and the limitations of prior pan-regional studies for water management applications; (3) tightened the language throughout, reducing wordiness by simplifying the climate context and merged related methodological points.

2. Data and methods: I think the section would benefit from a figure with photographs showcasing the different biomes.

We do not possess original photographs of the biomes and, to avoid copyright issues, we have chosen not to use third-party images. Instead, we opted to provide descriptive text to characterize the biomes.

Figure 2 should be Figure 1 and be included in the section for direct view.

The MDPI’s latex template has been used. Cross-referencing should now be functioning correctly.

Other observations:

• Line 78: spelling mistake “Ambiante”
• Line 88: spelling mistake “fall”
• Line 94: “Despite Mata Atlântica is under the same tropical climate than Cerrado”

Please correct English in this phrase

• Line 99: Acronym LST should have been introduced before
• Line 192: “Then we validated the regression model, trained on one subset, on the other two subsets and on itself (for reference). Another reference used in the cross-validation is the validation of the regression model trained on all subsets.”

Please provide a clearer explanation.

We thank the reviewer for the detailed observations, which significantly contributed to the improvement of the manuscript. All spelling corrections have been implemented in the revised version of the manuscript. The acronym LST (Land Surface Temperature) has been properly introduced at its first mention, in accordance with scientific writing standards.

“Geografia e Estatistica (IBGE) and the Ministerio do Meio Ambiente (MMA). To elaborate”

“that usually falls scattered over the area assigned to Caatinga in the MBB.”

“Although the Mata Atlântica is under the same tropical climate as the Cerrado, it is wetter, which favors its richness of biological variation”

“Land surface temperature (LST) is currently measured, with global coverage, by a”

“Cross-validation was performed by dividing the data into three subsets, each corresponding to a biome. For each regression model trained on one subset, predictions were validated on the other two subsets as well as on the subset used for training (as a reference). Additionally, a model was trained on all subsets combined, and its predictions were also validated as an additional reference. Model performance was evaluated using the Root Mean Squared Error (RMSE) and Mean Absolute Error (MAE).”

3. Results: Figure 2 and tables should be inserted in the section for direct view. Furthermore, the subplots should be bigger and have smaller labels. This would increase clarity in the data interpretation. In any case, at the lower end of the subplots (LST between 10 - 30) the datasets (Air Temp vs LST) seem to have a more spurious relationship. Have the authors looked into this? Furthermore, although there is a positive relationship and trend in all Air Temp vs LST datasets and the RMSE and MAE having values are around two, looking at the subplots, the distance between the linear regression lines and actual values seems considerable (>10). How do the authors explain this? Is this a visual question because most points are close to the models?

Yes, we appreciate the reviewer’s observation. We have now discussed in both the Results and Discussion sections, cooler surface temperatures often coincide with conditions — such as increased cloud cover, higher soil moisture, and dense canopy shading — that weaken the coupling between land surface and air temperatures. By working with large datasets (up to ~95 000 paired observations for the full basin), these localized and temporal deviations are largely averaged out in the regression analysis, resulting in robust model fits and highly significant coefficients (p < 0.01).

Other observations:

• Line 199: The BDMEP and INMET were already defined before. No need to define them again
• Line 211: It should be Figure 2
• Table captions are too long. Please shorten them and move some of the text to the body.
• Add point (2) in the Table 3 caption and correct the word “metris”
• Please explain in the captions what the grey lines represent in the subplots

 

“After the data processing, we selected 48 weather stations from BDMEP/INMET for April 2002 -present.”

“Figure 2 illustrates the distribution of the three biomes and the INMET weather stations in the São Francisco River basin.”

Caption explained.

4. Discussion: The discussion lacks references. References are essential to validate the arguments raised by the authors. Please provide recent and relevant references to the discussion.

Do the results satisfy the initial objectives of validating the well-established relationship between remote-sensed LST and ground-based air temperature? How close are the results to those obtained by other authors?

Furthermore, why do the authors believe the R2 values are good enough and not insufficient?

Other observations:

• Line 242: Figure 1 should be Figure 2
• Line 272: “(2) the surroundings of the weather stations at which observations of Air Tmax were sampled for the biome Mata Atlântica are long under the effect of urbanization, which lead the range of LST and air temperature to be near to the range of the observations of semiarid regions of the biomes Cerrado and Caatinga.”

Was the influence of urbanization properly quantified/estimated/modelled? Are there references for this?

In the revised version of the manuscript, the Results and Discussion sections were rewritten to include studies that support the findings presented here.

5. Conclusion: “The regression models explained, at least, 46% of the variation within the observations of maximum air temperature, which means that LST MYD21A1D could be used as key variable to estimate Air Tmax in the São Francisco River basin.”

Do the authors plan to improve the models in the future? What other variables could explain the remaining variation of Air Temp?

Other observations:

• What were the main limitations of the study in the opinion of the authors?
• Line 285: “key” missing an article

In the updated version of the manuscript, the Conclusion section was revised to incorporate the study's limitations and address the comments raised by other peers, as pointed out below:

“Despite these findings, several limitations should be noted. First, cloud contamination can introduce data gaps, most notably during the wet season in humid biomes. Second, the under-representation of Mata Atlântica stations limits the robustness of its biome-specific model. Third, the use of simple linear regressions does not capture non-linear or lagged surface–air interactions (e.g., canopy buffering or soil-moisture feedbacks) that may influence extreme temperature events.

Looking forward, the adoption of more sophisticated modeling approaches could further improve predictive accuracy and capture complex environmental interactions. Future work should explore machine-learning algorithms. Incorporating these intelligent methods, along with additional predictors (e.g., vegetation indices, topography, soil moisture), will likely yield more reliable estimates of near-surface air temperature from satellite LST.”

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

I have read the reviewed manuscript and I have the following comments:

Abstract is good but abbreviations should be removed. This is done later in the article.

The introduction is good, although minors errors are still present like missing spaces before citations. Just make sure that when you introduce an abbreviation like LST that you don't repeat it again in the following sections. This happens in the results, for example (Line 175).

The method section is alright, but Figure 2 is introduced before Figure 1. Please correct the numbering.

The results seem sound but figure 1 has not been changed it seems. I really think it should be improved as I recommended initially. Furthermore, some of the paragraphs in the results sections are discussing the results already (for example, lines 185-191). Any discussion of the results should be done in the discussion section.

The discussion section should be begin with a brief description of what was done for the article and then yes, enter into the discussion of the results. But more importantly, references are still missing... The author said that references were included but I can't see none. I can't stress enough how important it is to include references from which the arguments raised by the authors can be validated by comparison with results and methodology from other peers. For example:

"These statistical outcomes support the reliability of MYD21A1D 220
LST retrievals as predictors of maximum air temperature in the São Francisco River Basin 221
(SFRB). In practical terms, this strengthens the case for incorporating satellite-derived LST 222
as a cost-effective and scalable tool for regional temperature monitoring and for improving 223
early warning systems, particularly in areas vulnerable to heat stress and drought, such as 224
the semi-arid Caatinga."

"These processes decouple LST from near-surface air temperatures, reducing the reliability of simple linear models for forested areas. This outcome confirms that vegetation type and local climatic conditions significantly influence temperature dynamics, validating the need for biome-specific models. More importantly, it reinforces the value of forest cover as a regulator of microclimate."

Are you able to provide references that validate these claims? 

Please address this issue.

As for the conclusion, it is well written, but it includes a discussion of the limitations of the study, which is alright if done briefly, but it should be in the discussion section as well and contain more detail, referecing other studies if possible.

Last request: "These results reveal that LST can serve as a proxy for maximum air temperature with moderate confidence for the Cerrado, Caatinga and Mata Atlântica biomes - critical in areas
where station coverage is sparse. However, any operational system using LST to estimate 
Tmax should incorporate vegetation indices (e.g., NDVI) to compensate for canopy fraction 
changes [31] in a way that one can confidently parameterize sensible and latent heat fluxes  based on LST retrievals, for example. This could improve estimates of evapotranspiration 
rates, soil-moisture depletion, and ultimately streamflow."

Could you talk more about this is in discussion? I think it would be important.

Author Response

Thank you for your thorough evaluation and constructive suggestions. Below, we address each of your comments in turn, indicating the revisions made in the manuscript.

Abstract abbreviations
Comment: “Abstract is good but abbreviations should be removed. This is done later in the article.”
Response: “abbreviations were replaced from the abstract.”

Introduction – citation formatting & abbreviation use
Comment: “The introduction is good, although minor errors are still present like missing spaces before citations. Just make sure that when you introduce an abbreviation like LST that you don’t repeat it again in the following sections. This happens in the results, for example (Line 175).”
Response: “Right. We deleted all instances of land surface temperature that were coming before the abbreviation. We did the same for SFRB, MBB, INMET.”

Methods – figure numbering
Comment: “The method section is alright, but Figure 2 is introduced before Figure 1. Please correct the numbering.”
Response: “Numbering has been corrected.”

Results – figure quality and placement of discussion
Comment: “The results seem sound but figure 1 has not been changed it seems. I really think it should be improved as I recommended initially. Furthermore, some of the paragraphs in the results sections are discussing the results already (for example, lines 185–191). Any discussion of the results should be done in the discussion section.”
Response: “The discussion of the results have been moved to Discussion. We tried to change the figure as you suggested, but all of the attempts resulted in an overall decrease of the quality of the figure.”

Discussion – structure and missing references
Comment:

“The discussion section should be begin with a brief description of what was done for the article and then yes, enter into the discussion of the results. But more importantly, references are still missing... The author said that references were included but I can’t see none. I can’t stress enough how important it is to include references from which the arguments raised by the authors can be validated by comparison with results and methodology from other peers. For example:

“These statistical outcomes support the reliability of MYD21A1D LST retrievals as predictors of maximum air temperature in the São Francisco River Basin (SFRB). In practical terms, this strengthens the case for incorporating satellite-derived LST as a cost-effective and scalable tool for regional temperature monitoring and for improving early warning systems, particularly in areas vulnerable to heat stress and drought, such as the semi-arid Caatinga.”

“These processes decouple LST from near-surface air temperatures, reducing the reliability of simple linear models for forested areas. This outcome confirms that vegetation type and local climatic conditions significantly influence temperature dynamics, validating the need for biome-specific models. More importantly, it reinforces the value of forest cover as a regulator of microclimate.”

Are you able to provide references that validate these claims? Please address this issue.
Response: “As we moved some discussion from the Results to the Discussion section, more references came along with the Discussion. Also we provided the references that were missing for the statements given by the in the Comment above.”

Conclusion – relocation of limitations
Comment: “As for the conclusion, it is well written, but it includes a discussion of the limitations of the study, which is alright if done briefly, but it should be in the discussion section as well and contain more detail, referencing other studies if possible.”
Response: “We moved to the discussion section and added references.”

Final discussion paragraph enhancement
Comment:

“These results reveal that LST can serve as a proxy for maximum air temperature with moderate confidence for the Cerrado, Caatinga and Mata Atlântica biomes—critical in areas where station coverage is sparse. However, any operational system using LST to estimate Tmax should incorporate vegetation indices (e.g., NDVI) to compensate for canopy fraction changes [31] in a way that one can confidently parameterize sensible and latent heat fluxes based on LST retrievals, for example. This could improve estimates of evapotranspiration rates, soil-moisture depletion, and ultimately streamflow.”
Response:
“We hope by adding the following to the paragraph we have addressed this issue.

‘This is because by accounting for canopy fraction and phenological changes, coupled models capture drought‑induced declines in ET more accurately than LST alone, particularly in pasture‑dominated areas of the Cerrado. Late‑fall satellite soil moisture estimates, when informed by both LST and vegetation signals, is closely linked with subsequent spring streamflow, offering lead time for hydrological forecasting. This combined remote sensing approach could improve representation of pre‑wet‑season moisture deficits and runoff generation in seasonally dry environments.’”