Long-Term Time Series Estimation of Impervious Surface Coverage Rate in Beijing–Tianjin–Hebei Urbanization and Vulnerability Assessment of Ecological Environment Response

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Please see attached document

Comments for author File: Comments.pdf

Comments on the Quality of English Language

.

Author Response

Dear reviewers ,

 

We gratefully appreciate the editors and all reviewers for their time spend making positive and constructive comments. These comments are all valuable and helpful for revising and improving our manuscript entitled “Long-term Time Series Estimation of Impervious Surface Coverage Rate in Beijing-Tianjin-Hebei Urbanization and Vulnerability Assessment of Ecological Environment Response” (Manuscript Number: land-3763944) as well as the important guiding significance to our researches.

 

We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the revised original manuscript. The responses to the reviewer's comments are listed in the Revision Report.

 

We sincerely hope the revised version could encourage your acceptance of this paper and look forward to your comments.

 

Yours sincerely,

Authors of the manuscript land-3763944

 

 

Response to the Reviewer’s Comments

We sincerely appreciate the time taken by the three reviewers to provide us with valuable comments and constructive suggestions on our manuscript. We have made significant revisions to our manuscript, following the advice of the reviewers. Below the original comments from reviewers are copied in blue italics, followed by our point-to-point responses in roman font. Additional or altered text in the revised manuscript, when cited in this response document, is marked in red.

 

 

To reviewer #1

Reviewer: #1 (Comments to the Author)

In this paper, the authors present a case study of long-term time series estimation of the rate of impervious surface coverage in the Beijing-Tianjin-Hebei urbanization and assessment of the vulnerability of the ecological environment response using MODIS imagery and other satellite data. It is considered that the theme and the results obtained could be of interest to the readers of the journal, although the current state of the document presents some important issues that must be corrected and modified before its possible publication. Therefore, I suggest making some major corrections before proceeding to its possible publication.

My specific comments are as follows:

[Response] We appreciate the reviewer for recognizing our work. We thank the valuable comments and suggestions that helped improving our manuscript. Based on the comments and suggestions below, we have made detailed revisions to the manuscript.

 

Specific comments

1.1 Introduction: It is necessary to include the contributions of the document.

[Response] Thanks for pointing this out. The authors apologize for causing confusion to the reviewer. In the original manuscript, the last paragraph of the Introduction part indicates the contributions of the document . To offer a more explicit understanding, in the revised manuscript, the corresponding text has been reorganized. It can be located on page 2-3, line 94-114, within the last paragraph of Section 1.Introduction.

 

1.2 Section 2.2. It is essential to expand on the information that the authors convey in this section. For example, I have the following doubts about it: How many MODIS images did the authors use per year? I understand that they all correspond to the same date or close days. What is the time of passage of the sensor? What LST validation process did the authors undertake? Is the accuracy of land uses of 79.31 % for a single year or for all years, this value being the average?

[Response] Thanks for your suggestion. We now explained in more detail about the mentioned data.

Regarding the land surface temperature (LST) data used in this study, we employed the MODIS MOD11A1 Version 6 product, which provides daily daytime and nighttime LST observations at a spatial resolution of 1 km. Specifically, we used the daytime LST data (LST_Day_1km) for our analysis, which the passing time of the sensor is 10:30 a.m.. All data processing was conducted on the Google Earth Engine (GEE) platform, which enables efficient handling of large-scale remote sensing time series. The MOD11A1 dataset is a daily product, and we included all available daily observations within each year, rather than selecting images from a single or similar date. This approach ensures temporal representativeness and reduces the influence of anomalous values or data gaps.

To ensure the quality of the LST data, we leveraged the MODIS-provided quality control (QC) bands that accompany each image. These QC layers allow for pixel-level screening of cloud contamination, sensor anomalies, and other quality issues. In addition, GEE incorporates an automatic cloud masking function for MODIS data, further enhancing the reliability of the extracted temperature values.

The MOD11A1 LST product itself has been extensively validated by NASA using ground-based measurements. According to existing validation studies, the typical root-mean-square error (RMSE) ranges from approximately 1 K to 2 K, depending on surface conditions and regions. (Wan, Z., Hook, S., & Hulley, G. (2015).MOD11A1 MODIS/Terra Land Surface Temperature/Emissivity Daily L3 Global 1km SIN Grid V006[Data set]. NASA Land Processes Distributed Active Archive Center.)

The reported overall accuracy of 79.31% refers to the evaluation of the entire CLCD dataset, which includes annual land cover maps from 1990 to 2023(originally to 2019,but now updated to 2023). This accuracy was calculated based on 5,463 visually interpreted validation samples across multiple years, and thus represents the average accuracy across all years, not just a single year. As noted in Yang and Huang (2021), this evaluation was supplemented by further assessments using third-party test samples, confirming the superior performance of CLCD compared to other widely used land cover products such as MCD12Q1 and ESA-CCI.( Yang, C., & Huang, X. (2021). A Landsat-derived annual land cover dataset of China (CLCD) from 1990 to 2019. Zenodo. https://doi.org/10.5281/zenodo.4417810)

We have also add the information in our revised manuscripts. Please refer to Page 4, line160-167, line 170-176 and table 1 in the revised manuscript.

 

1.3 Fig. 4. Improve image quality. The legends are illegible.

1.4 Fig. 7. Same as the previous one. In this case, I suggest enlarging the size of the figures and keeping a single legend since they are all the same.

[Response] Thanks for pointing this out. We redesigned the layout to enhance the clarity and visibility of the image.

Figure R1 (This is Figure 4 in Chapter 2.3.5 of the main text). Spatial Distribution Maps of Normalized Factors in the Study Area.

Figure R2 (This is Figure 7 in Chapter 3.2 of the main text). EVI Classification Map (a) 2005 (b) 2010 (c) 2015 (d) 2020.

 

1.5 Section 3.3. I suggest including some more variables such as the F, P value and Prob>Chi2 statistic in order to complete the statistical analyses.

[Response]We appreciate the reviewer's suggestion to enhance the statistical analyses. In response to this comment, we have now included comprehensive statistical tests and their corresponding statistics as shown in the updated Figure R3. Our current statistical framework includes:

1. Correlation Analysis: We calculated both Pearson correlation coefficients (r = -0.3154, p = 5.231e-202) and Spearman rank correlation coefficients (ρ = -0.3092, p = 7.341e-194) to assess the linear and monotonic relationships between ΔISC and ΔEVI.
2. Group Comparison Analysis: We performed the Kruskal-Wallis H test (H = 754.7, p < 0.001) followed by post-hoc Dunn's test with Bonferroni correction to compare ΔEVI values across different ΔISC groups (Low, Medium, High).
3. Effect Size and Statistical Power: The sample size (n = 2927 for each group) provides adequate statistical power for detecting meaningful differences.

Regarding the Chi-square test specifically mentioned by the reviewer: While we acknowledge the importance of comprehensive statistical testing, the Chi-square test is primarily designed for categorical data analysis, particularly for testing independence between categorical variables or goodness-of-fit for frequency distributions. In our case, we are comparing continuous variables (ΔEVI values) across ordered groups, which makes the Kruskal-Wallis test more appropriate than the Chi-square test. The Kruskal-Wallis test is the non-parametric equivalent of one-way ANOVA and is specifically designed for comparing multiple groups of continuous or ordinal data, which aligns perfectly with our research objectives.

The statistical approach we have adopted (correlation analysis + Kruskal-Wallis + post-hoc tests) provides a robust and comprehensive framework for analyzing the relationship between impervious surface change and ecological vulnerability, ensuring both the reliability and interpretability of our findings. We believe these statistical analyses adequately address the research questions and provide sufficient evidence for the conclusions drawn in this study. We also add the results in the revise manuscript. (Page 15，L508-511, Page 16，L517-518)

Figure R3 (This is Figure 9 in Chapter 3.3 of the main text). Temporal coupling relationship between urbanization process and ecological vulnerability. (a) Trends of EVI changes in different density zones (b) Statistical comparison of EVI between high and low density ISC areas (c) Boxplot of EVI changes under ISC variation zones).

 

1.6 Fig. 11. The image quality needs to be improved. The captions are illegible, and you can't see anything in the images either.

[Response] We are grateful for the suggestion. We now modified the original Figure 11, as shown in Fig. R4.

Figure R4 (This is Figure 11 in Chapter 3.4 of the main text). Figure 11. Spatial coupling relationship between urbanization process and ecological vulnerability. (a) Pearson correlation coefficient between ISC and EVI change values during 2005–2020 (b) Representative study areas (c) The magnitude, timing and duration of impervious surface increase around Beijing Capital International Airport and Temple of Heaven Park).

 

1.7 Results and discussion. The results are clear and concise, but the authors do not discuss the results. This is essential to validate the results presented and therefore, the research itself. In this article, the sources of uncertainty (classification errors, spatial interpolation, resolution biases, etc.) that are inherent to this type of geospatial studies should be discussed.

1.8 I suggest including a section on limitations to the study.

[Response] Thank you for the valuable feedback. Following the reviewer's suggestions, we have completely restructured the Discussion section (Section 4) to provide comprehensive validation and critical analysis of our results. The revised Discussion is now organized into four distinct subsections:

In 4.1 Comparison with other global urban products section, we conducted comprehensive comparisons with established global urban products including GHSL, ESA-CCI, Zhou et al.'s dataset, and GISA across multiple urban agglomerations. In 4.2 Contributions and Limitations of ISC Data Construction section，we addressed the critical gap between binary classification and continuous coverage monitoring, particularly valuable for characterizing low-density development patterns and we systematically discussed classification errors, spatial aggregation effects, and inherited uncertainties from GAIA source data. In 4.3 Contributions and Limitations of EVI Assessment section, we highlighted the advantages of using Genetic Projection Pursuit model over subjective expert judgment approaches and discussed temporal resolution constraints (5-year intervals), fixed coefficient assumptions in HQI calculations, and potential oversimplification of ecological processes. We also addressed climate variability impacts and the challenges of capturing short-term environmental fluctuations In 4.4 Implications for Areas with Coexisting Urbanization and Rural Development section, we provided specific recommendations based on our findings, including differentiated urban growth management and location-specific approaches.

Please refer to page 19-22 section 4.

 

1.9 Spelling and grammar check.

[Response] We appreciate the reviewer's attention to spelling and grammar. In response to this comment, we have thoroughly proofread the manuscript using professional grammar-checking tools including Grammarly, supplemented by GPT-assisted language refinement. All grammatical corrections and linguistic improvements have been marked in red in the revised manuscript to clearly indicate the changes made.

 

1.10 I suggest including validation metrics (e.g. comparison with very high-resolution images or known areas) to demonstrate the accuracy of the built ISC product.

[Response] Thanks for the reviewer to point out this issue. We apologize for the vague validation part. We have implemented a comprehensive validation approach using multiple independent datasets and methods.

First, our data has its own foundation accuracy, the ISC data is derived from the mature impervious surface product GAIA through spatial aggregation processes. The underlying GAIA dataset has been extensively validated and demonstrates high accuracy with overall accuracies ranging from 89% to 97% across validation years, achieving a mean overall accuracy higher than 90% globally (Gong et al., 2020). This provides a solid foundation for the reliability of our ISC product.

Second, we conduct a multi-dataset cross-validation using four additional independent datasets:

• GHSL (Global Human Settlement Layer): A well-established global dataset for built-up area mapping.
• GISA (Global Impervious Surface Area): A comprehensive impervious surface dataset (available through 2019).
• ESACCI (European Space Agency Climate Change Initiative): A global land cover dataset.
• Zhou et al. (2018): Urban dynamics dataset from "A global record of annual urban dynamics (1992–2013) from nighttime lights".

The three histograms in Fig R4 presented Pearson correlation coefficients between GHSL and ISC values calculated from sample points across three major urban agglomerations for seven time periods (1985, 1990, 1995, 2000, 2005, 2010, and 2015):

Beijing-Tianjin-Hebei region: μ = 0.8218, σ = 0.1750；Yangtze River Delta: μ = 0.8899, σ = 0.1221；Pearl River Delta: μ = 0.8943, σ = 0.1171. The red dashed lines represent the mean correlation values (μ), while σ indicates standard deviation. These high correlation coefficients (>0.82 across all regions) demonstrate strong agreement between ISC and GHSL data, indicating high temporal accuracy of the ISC product.

The bottom panels in Fig R5 show validation against multiple independent datasets. Panel (a) presents a time series comparison of annual impervious surface area trends from four datasets (ISC, ESACCI, Zhou et al. 2018, and GISA) over 1985-2019, demonstrating that ISC captures similar temporal patterns and magnitudes to all three established global products. Panel (b) focuses on the dataset showing the most consistent trends with ISC - GISA - using 700 random sample points (20 points per year from 1985-2019) for detailed correlation analysis between ISC and GISA impervious surface areas. The strong correlation (r = 0.897) between ISC and GISA, combined with the consistent temporal trends observed across all four datasets, may provides robust evidence for the accuracy and reliability of our ISC product across the entire study period. We have incorporated this validation analysis into the " Comparison with other global urban products " section in the discussion for reviewers' reference.

Fig. R4(This is Figure 12 in Chapter 4.1 of the main text). Pearson correlation coefficient distributions between ISC and GHSL data across three major urban agglomerations (Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta) for seven time periods (1985-2015).

 

Fig. R5(This is Figure 13 in Chapter 4.1 of the main text). Comparison of ISC with other global urban products over past years. (a) Time series comparison of annual impervious surface area trends (b) Correlation analysis of ISC with GISA from 700 random sampling points (1985-2019).

 

1.11 Add a validation section of the EVI model and more reproducible details.

[Response] We apologize for the vague details and we appreciate the reviewer's request for enhanced validation of our EVI model and additional reproducible details. We have made the following improvements to address these concerns. While acknowledging that no completely identical EVI products exist for direct validation, we have implemented several validation strategies:

• Cross-reference with Existing Studies: Our EVI results show strong consistency with the ecological vulnerability patterns reported by (Fu Gang et al. 2016) in "Ecological Vulnerability Assessment in Beijing Based on GIS Spatial Analysis". Both studies identify similar high-vulnerability zones in Beijing's urban core and low-vulnerability areas in the northern mountainous regions, providing confidence in our methodology's validity.
• Internal Consistency Validation: We conducted sensitivity analysis by varying the GPP parameters and confirmed that our results remain stable across different parameter sets, demonstrating model robustness.
• Spatial Logic Validation: Our EVI patterns align with known ecological gradients and urbanization intensity distributions in the Beijing-Tianjin-Hebei region, providing face validity for our results.

We have also added a comprehensive discussion section (Section 4.3) that thoroughly examines both the contributions and limitations of our EVI assessment approach. This section provides critical evaluation of our methodology's strengths, acknowledges potential uncertainties, and discusses the implications of our findings for broader applications.

We acknowledge that further validation approaches could strengthen our methodology. Future research will focus on developing more comprehensive validation frameworks and refining the EVI model to enhance its applicability and accuracy across different regional contexts.

 

Reference

FU Gang, BAI Jia-de, QI Yue, et al. Ecological Vulnerability Assessment in Beijing Based on GIS Spatial Analysis[J]. Journal of Ecology and Rural Environment, 2018, 34(9): 830-839.

Wan, Z., Hook, S., & Hulley, G. (2015).MOD11A1 MODIS/Terra Land Surface Temperature/Emissivity Daily L3 Global 1km SIN Grid V006[Data set]. NASA Land Processes Distributed Active Archive Center.

Yang, C., & Huang, X. (2021). A Landsat-derived annual land cover dataset of China (CLCD) from 1990 to 2019. Zenodo. https://doi.org/10.5281/zenodo.4417810

 

 

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript presents a comprehensive study on urban sprawl and its impact on ecological vulnerability in the Beijing-Tianjin-Hebei (BTH) region. The methodology is sound and the results are relevant to sustainable urban planning. However, there are areas that require clarification and improvement. The topic is worthy of research; however, it has several mayor changes that need to be addressed before it can be considered for publication.

General comments

C1. The main concern is that the novelty of the research is not fully clear. If such novelty is not clearly highlighted, the risk is that the manuscript looks more a simple case study rather than a research paper.

C2. What are the major contributions of this study? should be carefully mentioned in the discussion section.

C3. Reorganize the introduction to link ISC → Ecological indicators → EVI from the beginning.

C4. The scope and objectives of the paper should be also clearly presented at the end of the introduction section.

C5. The work is disproportionately focused on the EVI (methods, results and discussion), while the construction of the ISC, although well executed, acts as a secondary input. This generates a disconnect between the title, the stated objectives and the core of the analysis.

C6. The conclusion should offer practical recommendations based on the findings of the study.

C7. It seems that the authors did not read the guide for authors. The references in the text are not properly cited.

*The answer to these questions should be reflected in the manuscript.

Specific comments

Line 35: It is not clear what knowledge gap this study addresses until the final paragraph. Make the hypothesis or central question explicit earlier.

Line 95: Add a sentence contrasting why GAIA needs aggregation to 1 km. Why not another resolution? How does it affect heterogeneity?

Line 145: Delete.

Line 177-187: Indicate how temporal differences were handled (MODIS, GAIA, population do not match in exact years).

Line 216-219: Justify choice of 1 km (vs. 500 m or other resolutions) and discuss possible aggregation bias.

Line 240: The standard deviation ranking method is appropriate, but indicate why this method was chosen instead of quantiles or Jenks.

Line 259-288: For expansion and centroid analysis: Explain how administrative boundary changes were handled and indicate whether centroid location error analysis was performed.

Line 290-301: Lacking detail of software used and libraries or tools implemented.

Line 303-369: The vulnerability model is complex. The authors should: a) Explain why genetic search projection (GPP) was chosen, b) Include complete EVI equation and expected range, and c) Justify advantages and limitations of the GPP.

Line 371-379: Missing statistics of annual change and significance (e.g., mean rates and standard deviation).

Line 424-452: Variability in vulnerability is mentioned, but confidence intervals and uncertainty analysis are not shown.

Line 469-476: Discuss possible causes of declining EVI in low density areas (effects of conservation policies?).

Line 481-502: Figure 9 shows ISC-EVI relationships, but a formal statistical test (linear regression, Pearson correlation, p-value) is missing.

Line 545-546: Correlation is mentioned (+0.6, -0.5) without indicating significance level.

Author Response

Dear reviewers ,

 

We gratefully appreciate the editors and all reviewers for their time spend making positive and constructive comments. These comments are all valuable and helpful for revising and improving our manuscript entitled “Long-term Time Series Estimation of Impervious Surface Coverage Rate in Beijing-Tianjin-Hebei Urbanization and Vulnerability Assessment of Ecological Environment Response” (Manuscript Number: land-3763944) as well as the important guiding significance to our researches.

 

We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the revised original manuscript. The responses to the reviewer's comments are listed in the Revision Report.

 

We sincerely hope the revised version could encourage your acceptance of this paper and look forward to your comments.

 

Yours sincerely,

Authors of the manuscript land-3763944

 

 

Response to the Reviewer’s Comments

We sincerely appreciate the time taken by the three reviewers to provide us with valuable comments and constructive suggestions on our manuscript. We have made significant revisions to our manuscript, following the advice of the reviewers. Below the original comments from reviewers are copied in blue italics, followed by our point-to-point responses in roman font. Additional or altered text in the revised manuscript, when cited in this response document, is marked in red.

 

To reviewer #2

Reviewer: #2 (Comments to the Author)

The manuscript presents a comprehensive study on urban sprawl and its impact on ecological vulnerability in the Beijing-Tianjin-Hebei (BTH) region. The methodology is sound and the results are relevant to sustainable urban planning. However, there are areas that require clarification and improvement. The topic is worthy of research; however, it has several mayor changes that need to be addressed before it can be considered for publication.

[Response] We thank the reviewer for the comments about our study. The reviewer provided us with many pertinent and useful suggestion. We realized that the manuscript was still lacking and revised it. In particular, we have described the highlights of this study more clearly in the Introduction and Discussion Chapter. Hopefully, our revisions will make improvement to the manuscript. We are keen to answer these questions in detail as follows.

 

General comments

C1. The main concern is that the novelty of the research is not fully clear. If such novelty is not clearly highlighted, the risk is that the manuscript looks more a simple case study rather than a research paper.

[Response] Thank you for your valuable feedback, and we apologize for not sufficiently highlighting the research novelty in the initial manuscript structure. We greatly appreciate this important concern as it helps clarify the fundamental contributions of our work.

The novelty of this research extends significantly beyond a simple case study through several key methodological and conceptual innovations. First, we developed a novel spatial aggregation methodology that converts 30m binary impervious surface data into 1km continuous coverage data, addressing a critical limitation in existing urbanization monitoring approaches. As stated in Section 2.3.1 and the Introduction: "To address this limitation, we constructed Impervious Surface Coverage (ISC) data generated from multi-temporal remote sensing imagery... Compared to traditional binary classification results, this method can more accurately reflect the degree of imperviousness within pixels, demonstrating higher expressive capability and application value in urban impervious intensity assessment, expansion process characterization, and spatial heterogeneity analysis." This methodological advance bridges the gap between micro-scale data and macro-scale analysis needs, solving the scale matching problem that has constrained previous urbanization studies.

Second, we introduced an objective weighting system using Genetic Projection Pursuit modeling to eliminate subjective bias in ecological vulnerability assessments. As detailed in Section 2.3.5: "Using Genetic Projection Pursuit modeling, we eliminate subjective weighting bias while establishing quantitative ISC-EVI relationships across different urbanization intensity zones... introducing the projection pursuit model to calculate the weights of ecological sensitive factors, avoiding the deviation caused by subjective weighting and improving the objectivity of evaluation results." This algorithmic, data-driven approach represents a significant advancement over current vulnerability assessments that rely on subjective expert judgment or arbitrary weighting schemes.

Third, we established a comprehensive analytical framework that integrates "impervious surface dynamics - multi-factor ecological response" paradigm, creating quantitative relationships between urbanization intensity gradients and ecosystem responses. As emphasized in the Introduction: "This complete process can be directly transferred to other high-density urban agglomerations such as the Yangtze River Delta and the Pearl River Delta, providing standardized tools for analyzing the urbanization ecological effects in similar regions and establishing a normalized ecological vulnerability early warning mechanism." Most existing studies examine urbanization and ecological vulnerability as separate phenomena, while our framework establishes their coupled dynamics.

Furthermore, we developed ISC as a multi-dimensional indicator capable of characterizing temporal rhythm, intensity, and duration of urbanization processes, as demonstrated in Section 3.4: "ISC is not only a spatial marker of urbanization intensity but can also reflect the temporal rhythm and spatial pathways of urbanization processes... three change dimension layers of ISC... change magnitude, change year, and change duration, to characterize the intensity, timing, and process characteristics of urban construction." This advances beyond traditional urban monitoring that focuses on static snapshots or simple change detection.

Finally, our methodological framework demonstrates transferability and robustness through validation across multiple urban agglomerations (Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta), as shown in Figure 12 and discussed in Section 4.1. This validation confirms that our contributions provide both theoretical advancement and practical tools applicable globally to similar urban contexts, establishing this work as a methodological contribution rather than a regional case study.

 

C2. What are the major contributions of this study? should be carefully mentioned in the discussion section.

[Response] Thank you for highlighting the importance of clearly articulating the major contributions of our study. We appreciate this guidance and would like to direct the reviewer's attention to where these contributions are comprehensively presented in our manuscript.

Research objectives and contributions are explicitly stated in the final paragraph of the Introduction section, where we outline our four main research aims and the methodological innovations that constitute our primary contributions to the field. Also in the Discussion section, we have systematically addressed our contributions through two dedicated subsections:

In Section 4.2 "Contributions and Limitations of ISC Data Construction" presents our methodological contributions to urbanization monitoring. Our development of continuous ISC coverage data addresses critical gaps in existing urbanization monitoring approaches by providing sub-pixel continuity that traditional binary classification methods cannot achieve. The temporal continuity maintained throughout our 35-year dataset (1985-2020) provides unprecedented insights into urbanization dynamics, enabling detection of subtle year-to-year variations in urban expansion rates and intensity changes that discrete time-point analyses miss. Furthermore, our methodology demonstrates enhanced capability for characterizing low-density development and gradual expansion patterns that are increasingly common in contemporary urban growth, while showing superior performance in revealing urbanization intensity gradients compared to conventional binary approaches.

In Section 4.3 "Contributions and Limitations of EVI Assessment" details our analytical contributions to ecological vulnerability assessment. The integration of Genetic Projection Pursuit modeling represents a significant advancement by eliminating subjective weighting bias that affects traditional vulnerability assessments, ensuring that indicator weights reflect actual data structure and relationships rather than expert judgment. Our multi-dimensional indicator system captures complex urbanization-ecology interactions by incorporating both natural and anthropogenic factors, providing insights that single-factor analyses cannot achieve. The temporal analysis framework we developed reveals dynamic vulnerability patterns and threshold effects that static assessments miss, enabling identification of when and where urbanization pressures exceed ecosystem resilience capacities. Most importantly, we established quantitative ISC-EVI relationships across different urbanization zones, providing a robust foundation for evidence-based urban planning and ecological management strategies.

Additionally, Section 4.4 discusses the practical contributions through policy implications and transferability to other urban agglomerations.

 

C3. Reorganize the introduction to link ISC → Ecological indicators → EVI from the beginning.

C4. The scope and objectives of the paper should be also clearly presented at the end of the introduction section.

[Response] We sincerely thank the reviewer for this excellent guidance on improving the logical structure and clarity of our introduction.

We have comprehensively reorganized the Introduction section to address both concerns:

1. Restructured Logical Flow (C3): We have rebuilt the introduction following the suggested ISC → Ecological indicators → EVI progression:

• Paragraph 1: Established the global urbanization context and the critical need to understand urban expansion-ecological vulnerability relationships
• Paragraph 2: Introduced impervious surfaces as key indicators for monitoring urbanization processes, highlighting the limitations of binary approaches and the advantages of our ISC methodology
• Paragraph 3: Transitioned to ecological indicators and their integration for vulnerability assessment, emphasizing the need for comprehensive multi-factor approaches
• Paragraph 4: Presented the EVI framework and our Genetic Projection Pursuit methodology as an objective solution to traditional subjective weighting problems
• Paragraph 5: Established our technical framework's contributions and replicability for regional applications

2. Clear Objectives and Scope Statement (C4): We have added a concluding paragraph that explicitly presents our research objectives in Paragraph 5 of the introduction part. Whereas study area scope is comprehensively addressed in Section 2.1, where we detail the geographical, administrative, and policy context of the Beijing-Tianjin-Hebei region.

 

C5. The work is disproportionately focused on the EVI (methods, results and discussion), while the construction of the ISC, although well executed, acts as a secondary input. This generates a disconnect between the title, the stated objectives and the core of the analysis.

[Response] We appreciate the reviewer's observation regarding the balance between ISC and EVI components. We would like to clarify that this balance is intentional and scientifically justified for the following reasons:

1. Methodological hierarchy: The ISC construction serves as the foundational data layer that enables the subsequent EVI analysis. While ISC methodology is well-established (spatial aggregation), our contribution lies in demonstrating its application for continuous urbanization monitoring and its integration with ecological assessment.
2. Scientific novelty: The primary scientific advance is in establishing quantitative relationships between urbanization intensity gradients (measured via ISC) and ecological vulnerability patterns. The EVI methodology using Genetic Projection Pursuit addresses existing gaps in objective vulnerability assessment.
3. Policy relevance: Urban planners and policymakers are ultimately interested in understanding ecological impacts of urbanization, which requires both accurate urbanization measurement (ISC) and comprehensive vulnerability assessment (EVI). The integrated approach provides actionable insights that neither component could deliver independently.
4. Scope alignment: The Beijing-Tianjin-Hebei case study specifically requires this integrated approach because the region exhibits complex urbanization-ecology interactions that necessitate both continuous urbanization monitoring and multi-factor vulnerability assessment.

The title reflects the methodological foundation (ISC) while the analysis delivers the policy-relevant outcomes (vulnerability assessment). We believe this integration represents the paper's core contribution to sustainable urban development research.

 

C6. The conclusion should offer practical recommendations based on the findings of the study.

[Response] Thank you for this valuable suggestion regarding the inclusion of practical recommendations based on our study findings. We completely agree that translating research results into actionable policy guidance is essential for the practical impact and societal relevance of our work.

We have addressed this concern by significantly strengthening Section 4.4 "Implications for Areas with Coexisting Urbanization and Rural Development" in the Discussion, where we now provide comprehensive practical recommendations based on our findings.

 

C7. It seems that the authors did not read the guide for authors. The references in the text are not properly cited.

[Response] We sincerely apologize for the oversight regarding the proper citation format according to the journal's Guide for Authors. This was an inexcusable error on our part, and we take full responsibility for not carefully following the journal's formatting requirements in our initial submission.

We have thoroughly reviewed the Guide for Authors and systematically corrected all in-text citations throughout the manuscript to ensure they now conform to the journal's specific citation style requirements.

 

*The answer to these questions should be reflected in the manuscript.

 

Specific comments

2.1 Line 35: It is not clear what knowledge gap this study addresses until the final paragraph. Make the hypothesis or central question explicit earlier.

[Response] Thank you for this important feedback regarding the clarity and organization of our introduction. We sincerely apologize for the unclear structure in the original version and greatly appreciate your constructive suggestions.

We have substantially reorganized the introduction section to address your concerns about the delayed presentation of knowledge gaps and central research questions. The new introduction structure is as follows:

Paragraph 1: Opens with the central research question and introduces the urbanization-ecological vulnerability relationship challenge

Paragraph 2: Reviews impervious surface monitoring approaches and existing data products, establishing the technical context

Paragraph 3: Focuses specifically on GAIA data characteristics and explicitly identifies Knowledge Gap 1: "Existing binary approaches cannot adequately characterize the continuous nature of urbanization processes, especially low-density development and urban fringe dynamics"(Page 2, L 66-69)

Paragraph 4: Introduces our ISC methodology as the solution to address the first knowledge gap

Paragraph 5: Transitions to urbanization-environment interactions and identifies Knowledge Gap 2: "Current ecological vulnerability assessments often rely on subjective weighting schemes and fail to establish quantitative relationships between urbanization intensity gradients and ecosystem responses"(Page 2-3, L 85-91)

Paragraph 6: Presents our integrated research framework linking ISC data construction → ecological indicator integration → EVI assessment, with three explicit research objectives. Please refer to the revised manuscript in the Introduction section.

 

2.2 Line 95: Add a sentence contrasting why GAIA needs aggregation to 1 km. Why not another resolution? How does it affect heterogeneity?

2.5 Line 216-219: Justify choice of 1 km (vs. 500 m or other resolutions) and discuss possible aggregation bias.

[Response] Thanks for the reviewer’s attention to this point. We chose 1 km aggregation resolution for several interconnected reasons that balance analytical needs with data consistency. First, to ensure spatial consistency across all variables in our EVI assessment framework, we standardized to 1 km resolution to match our MODIS-derived indicators (FVC, LST) and population density data, which are natively available at this resolution. Second, 1 km represents an optimal scale for capturing regional urbanization patterns while maintaining computational efficiency for the 35-year time series analysis across the 216,000 km² study area. Third, this resolution effectively captures the spatial heterogeneity of urbanization processes at the city-to-regional scale, which is appropriate for policy-relevant analysis in mega-urban agglomerations.

Regarding heterogeneity effects, the 1 km aggregation inevitably smooths fine-scale variations within pixels, potentially underestimating urban fragmentation in areas with highly dispersed settlement patterns. However, this scale preserves the essential spatial gradients needed to distinguish between different urbanization intensity zones while reducing noise from isolated small patches that may not be ecologically or policy-relevant at the regional scale. We acknowledge this trade-off between fine-scale detail and regional-scale pattern detection, and note that our validation against GHSL data (showing correlations >0.82) confirms that the aggregated ISC effectively captures urbanization dynamics at the intended analytical scale.

Regarding aggregation bias, we acknowledge several potential issues and their implications. The aggregation process may introduce edge effects where pixels at urban-rural boundaries are over- or under-represented depending on the spatial configuration of impervious surfaces within the 1km grid. Additionally, heterogeneity smoothing occurs in areas with mixed development patterns, such as suburban areas with scattered buildings, where important spatial detail that could be ecologically significant may be lost. Threshold effects present another concern, as small but ecologically important impervious patches smaller than 1 hectare may be diluted in the aggregation process, potentially missing critical habitat fragmentation patterns.

To address these concerns, we implemented several mitigation measures. Our validation against existing products demonstrates that aggregation bias does not significantly compromise data reliability, as evidenced by the strong correlations (R² > 0.8) with GHSL and GISA shown in Figures 12-13. We conducted multi-scale analysis through centroid analysis at both regional and local scales to capture patterns that might be lost in uniform aggregation. Furthermore, we applied quality control measures by implementing mask processing to invalid values and edge areas before aggregation to minimize edge effects.

We acknowledge the limitations inherent in our approach. The aggregation to 1km resolution may underestimate urban fragmentation in areas with dispersed settlement patterns, particularly in the rural-urban fringe, which could potentially affect our vulnerability assessment in transition zones. However, for our regional-scale analysis focused on major urbanization trends and policy implications, the benefits of consistent resolution and computational feasibility outweigh these limitations. Future studies could benefit from multi-resolution approaches or adaptive aggregation methods that preserve important spatial details while maintaining computational efficiency.

We have revised this in section 2.3.1(Page 6, L211-216) and add section 4.1,4.2 in the discussion part for ISC comparison and limitation discuss.

 

2.3 Line 145: Delete.

[Response] Thanks for pointing this out. We now deleted the redundant sentence.

 

2.4 Line 177-187: Indicate how temporal differences were handled (MODIS, GAIA, population do not match in exact years).

[Response] Thank you for this important clarification request. We acknowledge that our original manuscript did not sufficiently detail the temporal alignment procedure between different datasets. We revised Section 2.3.5 to emphasize temporal synchronization in EVI calculations

Clarification of our approach:

• EVI calculations used strictly temporally aligned data: For each target year (2005, 2010, 2015, 2020), we extracted corresponding annual data from each dataset
• MODIS and population data coverage: Both datasets provide annual data covering our study period, allowing exact year matching
• GAIA data: Annual impervious surface maps were available for all target years within the 1985-2020 coverage period
• No temporal interpolation or misalignment: Each EVI calculation uses only data from the corresponding year, ensuring temporal consistency

We have now addressed this concern in Section 2.3.5(Page 9, L 307-310).

 

2.6 Line 240: The standard deviation ranking method is appropriate, but indicate why this method was chosen instead of quantiles or Jenks.

[Response] Thanks for the reviewer to point out this issue. Traditional methods (such as equal interval division) may be affected by extreme values. To objectively classify ISC density into grades, the mean-standard deviation method is used for density segmentation instead of alternative approaches such as quantiles or Jenks natural breaks classification.

The mean-standard deviation method was selected for several reasons: (1) Theoretical foundation: Unlike quantiles that rely purely on data distribution ranking or Jenks that focuses on minimizing within-class variance, the mean-standard deviation method is grounded in statistical theory and provides meaningful thresholds based on data dispersion characteristics (Chen et al., 2019); (2) Comparability across regions: The method enables consistent classification criteria that can be applied across different study areas and time periods, whereas quantiles produce relative classifications that vary with each dataset; (3) Ecological interpretation: The standard deviation-based thresholds correspond to statistically significant deviations from the regional mean, which better reflects the ecological significance of urbanization intensity gradients; (4) Robustness to outliers: While sensitive to extreme values, this method provides more stable classifications for time-series analysis compared to Jenks natural breaks, which can produce inconsistent temporal comparisons when break points shift between time periods.

This approach has been successfully applied in similar impervious surface studies, particularly in the analysis of temporal and spatial variation of impervious surface area and its relationship with thermal environment. The mean-standard deviation method is based on data distribution characteristics and can more adaptively distinguish high, medium, and low density areas while maintaining statistical rigor and reproducibility.

 

2.7 Line 259-288: For expansion and centroid analysis: Explain how administrative boundary changes were handled and indicate whether centroid location error analysis was performed.

[Response] Thanks for the reviewer’s attention to this point. To ensure temporal consistency throughout the 35-year study period, we adopted a fixed administrative boundary approach based on the 2020 administrative divisions of the Beijing-Tianjin-Hebei region. All historical impervious surface data (1985-2020) were processed within these consistent boundaries to avoid artificial centroid shifts caused by administrative boundary changes rather than actual urban development patterns. We verified that major administrative boundary changes during the study period (such as the establishment of Xiongan New Area in 2017 and several district mergers in Beijing and Tianjin) did not significantly affect the overall regional analysis, as our focus was on impervious surface distribution patterns rather than administrative unit-based statistics. This approach follows established practices in long-term urban expansion studies (Xu et al., 2018) and ensures that observed centroid movements reflect genuine urbanization processes rather than administrative adjustments.

The centroid calculation method is well-established in urban expansion studies and has been widely validated in similar research contexts. The weighted average approach effectively captures the spatial distribution characteristics of impervious surfaces, and the resulting centroid coordinates provide reliable indicators of urban development directionality and intensity changes over time.

To determine the different center directions of urban development at different levels, time series impervious surface images from six periods were used to calculate impervious surface centroids at both overall and local regional scales: (1) 1990, (2) 1995, (3) 2000, (4) 2005, (5) 2010, (6) 2015, (7) 2020.

 

2.8 Line 290-301: Lacking detail of software used and libraries or tools implemented.

[Response] We are grateful for the suggestion. All landscape pattern calculations were performed using FRAGSTATS 4.2, a widely-used spatial pattern analysis program for categorical maps. The annual land use/land cover raster data (30m resolution) from CLCD were directly input into FRAGSTATS as GeoTIFF format files. We utilized the patch-level and class-level metrics modules within FRAGSTATS to compute landscape indices. The software was configured with 8-neighbor connectivity rules to identify patches and calculate spatial metrics. All calculations were performed using the software's standard algorithms without additional customization.

We have add additional information in the revised manuscript(Page 9, L 297-299).

 

2.9 Line 303-369: The vulnerability model is complex. The authors should: a) Explain why genetic search projection (GPP) was chosen, b) Include complete EVI equation and expected range, and c) Justify advantages and limitations of the GPP.

[Response] Thanks for the reviewer’s attention to this point. In our study, The Genetic Projection Pursuit (GPP) model was selected for weight determination in EVI calculation due to its ability to objectively optimize indicator weights without subjective bias. Unlike traditional vulnerability assessments that rely on expert judgment or arbitrary weighting schemes, GPP employs genetic algorithms to search for optimal projection directions that maximize the projection index function , where  represents the standard deviation and  represents the local density of projected values. This approach ensures that indicator weights reflect the actual data structure and relationships within our multi-dimensional dataset, thereby enhancing the objectivity and reproducibility of vulnerability assessments.

The final EVI calculation equation is:

 = normalized indicator values

n = number of indicators (5 in this study: FVC, HQI, LST, slope, population density)

The EVI values range from 0 to 1, where higher values indicate greater ecological vulnerability. Based on our analysis across four study years (2005-2020), the actual EVI range was 0.15-0.78, which was classified into three categories using natural breaks: low vulnerability (EVI < 0.25), moderate vulnerability (0.25 < EVI < 0.50), and high vulnerability (EVI > 0.50).

The explanation also connects well with a more detailed discussion of GPP contributions and limitations in section 4.3, providing a complete picture of why this method was chosen and its limitations, please refer to section4.3 in page 21, Line 682-710.

 

2.10 Line 371-379: Missing statistics of annual change and significance (e.g., mean rates and standard deviation).

[Response] We thank the reviewer for this valuable suggestion to strengthen the quantitative foundation of our urbanization analysis. We have addressed this comment by adding comprehensive statistical data. Including mean annual expansion rates for each major city/region in the Beijing-Tianjin-Hebei area, calculated from our 35-year GAIA dataset (1985-2020) and statistical significance tests comparing expansion rates between the rapidly developing Beijing-Tianjin corridor and the slower-developing northern cities (Chengde and Zhangjiakou), with appropriate p-values.(Page 11, L 385-389.)

 

2.11 Line 424-452: Variability in vulnerability is mentioned, but confidence intervals and uncertainty analysis are not shown.

[Response] We sincerely appreciate the reviewer's important observation regarding the explicit presentation of uncertainty analysis and confidence intervals in our EVI assessment. This is indeed a crucial aspect of robust vulnerability analysis that deserves clear articulation.

We acknowledge that while our EVI methodology incorporated multiple forms of uncertainty assessment, we failed to explicitly present these as formal uncertainty quantification in the vulnerability discussion. Our approach actually included several uncertainty assessment components specifically for the EVI analysis: (1) Statistical significance testing with p < 0.001 confidence levels for EVI differences between different ISC density zones (Figure 9b), demonstrating 99.9% confidence in our vulnerability classifications; (2) Temporal consistency validation across the 15-year study period (2005-2020) showing robust EVI pattern stability; (3) Methodological uncertainty evaluation through the Genetic Projection Pursuit model's inherent parameter optimization process, which provides uncertainty bounds by testing multiple weight combinations before converging on optimal solutions; and (4) Statistical validation of the quantitative relationship between urbanization intensity and EVI changes (Figure 9c, p < 0.001).

 

2.12 Line 469-476: Discuss possible causes of declining EVI in low density areas (effects of conservation policies?).

[Response] Thank you for raising this important question about the underlying causes of declining EVI in low-density areas. We appreciate the reviewer's interest in understanding the drivers behind this concerning trend.

We would like to direct the reviewer's attention to our added discussion of these causes in the Conclusions section (Page 22, Lines 752-754), where we address the multiple stressors contributing to the 4.2% decline in low-vulnerability mountainous areas. Specifically, we identify several key factors:

Climate change impacts on montane ecosystems - affecting ecosystem stability and resilience in mountainous regions.

Tourism and infrastructure development pressure - increasing anthropogenic stress even in low-density areas.

Policy implementation challenges - where conservation measures may be insufficient relative to development pressures.

 

2.13 Line 481-502: Figure 9 shows ISC-EVI relationships, but a formal statistical test (linear regression, Pearson correlation, p-value) is missing.

[Response] Thank you for this valuable feedback, We have addressed this concern by incorporating comprehensive statistical analysis directly into the figure and corresponding text. Based on the revised Figure 9, we have added the following statistical elements:

1. Linear Regression Analysis (Figure 9a):

• High-density areas: y = 0.0031x + 5.7001, R² = 0.7047
• Low-density areas: y = -0.0004x + 1.1172, R² = 0.1077
• These regression equations quantify the temporal relationships between urbanization and ecological vulnerability

2. Statistical Significance Testing (Figure 9b):

• Added formal statistical comparisons between high and low density areas for each study year
• All comparisons show highly significant differences (p < 0.001, indicated by ***)
• This provides 99.9% confidence levels for our density-based vulnerability classifications

3. ANOVA and Multiple Comparisons (Figure 9c):

• Formal statistical testing of EVI changes across different ISC increment groups
• Significant differences between all groups (p < 0.001, indicated by ***)
• Sample sizes clearly indicated (n = 2927 for each group)
• This validates the quantitative relationship between impervious surface expansion and ecological vulnerability changes

In the corresponding text, we have added:

• Explicit reporting of regression coefficients and R² values
• Statistical significance levels (p < 0.001) for all major relationships
• Confidence interval interpretations for policy applications

Figure R1 (This is Figure 9 in Chapter 3.3 of the main text). Temporal coupling relationship between urbanization process and ecological vulnerability. (a) Trends of EVI changes in different density zones (b) Statistical comparison of EVI between high and low density ISC areas (c) Boxplot of EVI changes under ISC variation zones).

 

2.14 Line 545-546: Correlation is mentioned (+0.6, -0.5) without indicating significance level.

[Response] Thank you for pointing out the missing significance levels for the correlation coefficients mentioned in our spatial coupling analysis. We have addressed this important statistical detail in the revised manuscript. The large sample sizes inherent in our grid-based analysis (covering thousands of 1km pixels across the 216,000 km² study area over multiple time periods) provide robust statistical power for correlation analysis. All correlation coefficients reported in our study achieve high significance levels (p < 0.001), which we have now explicitly stated in the revised text to ensure complete statistical transparency.(Page 17-18, Line 564-567)

 

Reference

 

Chen, Y. (2019). Temporal and spatial variation of the impervious surface area in Zhengzhou City and its relationship with thermal environment.

Xu, J., Zhao, Y., Zhong, K., Zhang, F., Liu, X., & Sun, C. (2018). Measuring spatio-temporal dynamics of impervious surface in Guangzhou, China, from 1988 to 2015, using time-series Landsat imagery. Science of the Total Environment, 627, 264–281.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

This study used GAIA impervious surface data to get its coverage dataset to estimate the urban expansion processes and landscape pattern indices as well as the spatiotemporal response relationships of ecological environment changes in the Beijing-Tianjin-Hebei region. Then combining the natural and social factors to assess the vulnerability of urbanization development. The overall structure of the article is clear, the methods are appropriate, the results are reliable, the chart information is sufficient, but there are some unclear expressions in the text, the chart design needs to be optimized, and the format of the article needs to be adjusted to ensure its scientificity and readability.

1. It is necessary to specify the accuracy of the impervious surface data. The creation of this data is mentioned as a key point in the text, which indicates its significance.
2. The abstract should include some quantitative expressions to support the reliability of the results.
3. Why these specific natural and social factors were chosen？
4. Lines 165-167: some contents is missing.
5. The font differences in Table 2 are too significant.
6. Line 273, two same citing references.
7. Why choose CLCD data for calculating the area of land cover types instead of using the MODIS product with the same resolution, in order to reduce the differences between the sensors?
8. 4, The font size in the picture could be made larger.
9. Line 338, font size is too large.
10. What is the basis for determining the EVI threshold?
11. In Figure 5, what is the black distribution area in the left part? The left image shows the change in the area of the impervious surface over the years at the beginning and the end?
12. The font sizes in the picture need to be standardized to enhance clarity and readability.
13. In the left picture of Figure 9, what do the dotted lines represent?
14. The title of Figure 10 is not accurate enough and needs to be combined with the specific information shown in the figure. “ab patch density cd patch area”: The expression of the subheadings of the pictures can be made clearer. This occurs multiple times in the text.
15. The results are integrated with the discussion in the text. However, the discussion section in the current text is not sufficient and needs to be enhanced to demonstrate scientificity.
16. At present, are there any other data on the distribution of impervious surfaces that can be compared with the data in the text to demonstrate the reliability of the data presented herein?
17. The obtained ecological vulnerability statistics and other results in the text can they be compared with the actual situation to better verify the methodology?
18. The literature discussion is relatively scarce. The analysis of the current situation in the introduction needs to be strengthened with reference to the literature.

Author Response

Dear reviewers ,

 

We gratefully appreciate the editors and all reviewers for their time spend making positive and constructive comments. These comments are all valuable and helpful for revising and improving our manuscript entitled “Long-term Time Series Estimation of Impervious Surface Coverage Rate in Beijing-Tianjin-Hebei Urbanization and Vulnerability Assessment of Ecological Environment Response” (Manuscript Number: land-3763944) as well as the important guiding significance to our researches.

 

We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the revised original manuscript. The responses to the reviewer's comments are listed in the Revision Report.

 

We sincerely hope the revised version could encourage your acceptance of this paper and look forward to your comments.

 

Yours sincerely,

Authors of the manuscript land-3763944

 

 

Response to the Reviewer’s Comments

We sincerely appreciate the time taken by the three reviewers to provide us with valuable comments and constructive suggestions on our manuscript. We have made significant revisions to our manuscript, following the advice of the reviewers. Below the original comments from reviewers are copied in blue italics, followed by our point-to-point responses in roman font. Additional or altered text in the revised manuscript, when cited in this response document, is marked in red.

 

To reviewer #3

Reviewer: #3 (Comments to the Author)

This study used GAIA impervious surface data to get its coverage dataset to estimate the urban expansion processes and landscape pattern indices as well as the spatiotemporal response relationships of ecological environment changes in the Beijing-Tianjin-Hebei region. Then combining the natural and social factors to assess the vulnerability of urbanization development. The overall structure of the article is clear, the methods are appropriate, the results are reliable, the chart information is sufficient, but there are some unclear expressions in the text, the chart design needs to be optimized, and the format of the article needs to be adjusted to ensure its scientificity and readability.

[Response] Thank you for this general summary. We thank him/her for valuable comments and suggestions that inspire us a lot. We have carefully revised the manuscript following the detailed comments and suggestions

 

Specific comments

3.1 It is necessary to specify the accuracy of the impervious surface data. The creation of this data is mentioned as a key point in the text, which indicates its significance.

[Response] Thanks for your valuable suggestion. You are absolutely correct that specifying the accuracy of the GAIA impervious surface data is crucial, as it forms the foundation for our ISC construction methodology. Based on the GAIA dataset documentation and validation studies (Gong et al., 2020), The GAIA dataset demonstrates consistently high accuracy with overall accuracies (OA) ranging from 89% to 97% across different validation years (1985, 1990, 1995, 2000, 2005, 2010, and 2015). Mean overall accuracy is higher than 90% across the validation period, and 89% overall accuracy using the improved algorithm specifically designed for arid and semi-arid regions. The Accuracy assessment was conducted using independent validation sample units with manual interpretation from Landsat images referenced to high-resolution Google Earth imagery. We have add this accuracy specification to our methodology section to provide full transparency about the data quality underlying our ISC construction approach. Please refer to page 6-7, line 206-209,222-224, within section 2.3.1.

 

3.2 The abstract should include some quantitative expressions to support the reliability of the results.

[Response]Thank you for the valuable suggestion regarding the inclusion of quantitative expressions in the abstract. We have substantially revised the abstract to incorporate specific numerical data that demonstrate the reliability and significance of our findings. The revised abstract now includes: (1) precise impervious surface area measurements showing an increase from 5,272.3 km² to 32,171.1 km² with an 88.5% growth rate; (2) specific urban expansion rates during peak periods (1,079.7 km²/year around 1990 and 1,300 km²/year during 2010-2015); (3) quantified forest area changes from 2,500 km² to 7,000 km² in high-density areas; and (4) measurable ecological vulnerability changes with a 4.2% decrease in low vulnerability areas. These quantitative metrics provide concrete evidence supporting our conclusions about urbanization patterns and ecological responses in the Beijing-Tianjin-Hebei region, thereby enhancing the scientific rigor and credibility of our research findings.

Please refer to page 1, line 17-20,22,28-30, within Section Abstract.

 

3.3 Why these specific natural and social factors were chosen？

[Response]We thank the reviewer for this insightful question regarding our factor selection methodology.

The selection of five specific factors (HQI, LST, FVC, population density, and slope) for calculating ecological vulnerability index (EVI) was based on their proven effectiveness in comprehensively representing different dimensions of ecosystem health and human impact. This multi-factor approach aligns with established remote sensing ecological assessment frameworks.

Rationale for Factor Selection:

1.Vegetation Environment Indicators:

Fractional Vegetation Coverage (FVC) was selected as it directly reflects vegetation prosperity and growth status within a region. FVC is calculated using the pixel binary method, where NDVI values with cumulative probabilities of 0.5% and 99.5% represent pure soil and pure vegetation respectively.

Habitat Quality Index (HQI) was chosen to characterize the ecological environment quality that supports species persistence. Following the "Technical Standards for Ecological Environment Assessment" promulgated by China's Ministry of Environmental Protection (MEP-2015), where A_f, A_g, A_w, A_c, A_i, and A_b represent forest, grassland, water body, cultivated land, impervious surface, and bare land areas respectively, with weights reflecting their ecological importance. Notably, the impervious surface area (A_i) used in HQI calculation is derived from our constructed ISC dataset, ensuring methodological consistency throughout the study.

2.Climate and Topographic Factors:

Land Surface Temperature (LST) serves as an indicator of climate change and urban heat island effects, directly reflecting the thermal environment impact of urbanization.

Slope represents topographic constraints and natural vulnerability to environmental changes, particularly important in mountainous regions of the study area.

3.Anthropogenic Pressure:

Population density quantifies direct human pressure on ecosystems, representing the intensity of anthropogenic activities and resource consumption.

On the other hand, this factor selection is consistent with the widely-used Remote Sensing Ecological Index (RSEI), which similarly incorporates vegetation coverage, land surface temperature, and anthropogenic factors to assess ecological quality. The RSEI framework has demonstrated that these indicators effectively capture the multidimensional nature of ecological health, including biological productivity, thermal environment, and human disturbance.

The combination of these five factors provides a comprehensive assessment framework that captures: (1) biological health through vegetation indicators, (2) environmental stress through temperature and topographic factors, and (3) anthropogenic pressure through population density. This multi-dimensional approach ensures that the ecological vulnerability assessment reflects both natural ecosystem conditions and human-environment interactions, which is essential for understanding urbanization impacts in rapidly developing regions like Beijing-Tianjin-Hebei.

 

• Lines 165-167: some contents is missing.

[Response]Thank you for pointing this out. We sincerely apologize for this oversight. The missing content in lines 165-167 resulted from an incorrect paragraph division during the manuscript formatting process. This section should have been connected with the subsequent paragraph, as they form a continuous discussion of the same topic. We will correct this formatting error to ensure the content flows properly and there are no gaps in the text.

 

3.5The font differences in Table 2 are too significant.

[Response]Thank you for bringing this to our attention. We apologize for the inconsistent font formatting in Table 2. We have now revised the table to ensure uniform font style and size throughout, maintaining consistency with the rest of the manuscript. The corrected version will be included in our revised submission.

 

• Line 273, two same citing references.

[Response] Thank you for pointing out this error. We sincerely apologize for the duplicate citing references on line 273. This was an oversight during the manuscript preparation process. We have carefully reviewed line 273 and identified the redundant citation. We will remove the duplicate reference and ensure that only the appropriate citation remains. We have also conducted a thorough check of the entire manuscript to identify and correct any other potential duplicate citations. (Page 8，L274)

 

• Why choose CLCD data for calculating the area of land cover types instead of using the MODIS product with the same resolution, in order to reduce the differences between the sensors?

[Response] Thank you for this important question regarding our choice of CLCD data over MODIS products. We chose CLCD (China Land Cover Dataset) data for calculating land cover area for the following reasons:

First, CLCD provides significantly higher spatial resolution (30 m) compared to commonly used MODIS land cover products (500 m resolution), which offers much greater accuracy and detail for land cover classification and area calculation. This higher resolution is particularly important for our study area, as it allows us to capture fine-scale land cover patterns and transitions that would be missed at coarser resolutions.

Second, CLCD data was not only used for calculating land cover areas but also served as the foundation for computing landscape indices in subsequent analyses. Using a consistent, high-resolution dataset throughout our methodology ensures internal consistency and reduces potential errors that could arise from mixing datasets with different spatial resolutions and classification schemes. Also, CLCD was specifically developed for China and demonstrates better applicability in our study region compared to global MODIS products. The dataset is optimized for Chinese landscapes and land cover characteristics.

While we acknowledge that using MODIS products might reduce sensor differences, the substantial improvement in spatial accuracy provided by the 30 m CLCD data outweighs this consideration. The higher resolution is crucial for accurately quantifying land cover changes and landscape patterns, which are central to our research objectives.

We have added a brief explanation of this rationale to the methodology section to clarify our data selection process. (Page 5，L170-176)

 

• 4, The font size in the picture could be made larger.

[Response] Thanks for pointing this out. We redesigned the layout to enhance the clarity and visibility of the image.

Figure R1 (This is Figure 4 in Chapter 2.3.5 of the main text). Spatial Distribution Maps of Normalized Factors in the Study Area.

 

• Line 338, font size is too large.

[Response] Thanks for pointing this out. We have adjust the font size in the revised manuscript.(Page 10，L343-344)

3.10 What is the basis for determining the EVI threshold?

[Response]Thank you for this important question regarding the determination of EVI thresholds. The EVI threshold determination was based on the natural breaks (Jenks) classification method implemented in ArcMap, combined with the actual EVI calculation results from four study years (2005, 2010, 2015, and 2020). The Jenks natural breaks algorithm identifies breakpoints in the data distribution by minimizing within-class variance while maximizing between-class variance. This statistical approach ensures that the classification boundaries occur at natural gaps in the data distribution rather than arbitrary equal intervals. The natural breaks approach maximizes the contrast between different vulnerability classes, making the temporal and spatial changes in ecological vulnerability more evident and interpretable. This method ensures that the three categories (low: <0.25, moderate: 0.25-0.50, high: >0.50) capture the most significant variations in ecological vulnerability patterns. At the same time, using the combined four-year dataset ensures that the thresholds are representative of the full range of ecological conditions observed in the Beijing-Tianjin-Hebei region during the study period, providing robust classification boundaries for vulnerability assessment.

We have added this methodological explanation to the EVI assessment section to clarify our threshold determination process. (Page 11，L373-376)

 

3.11 In Figure 5, what is the black distribution area in the left part? The left image shows the change in the area of the impervious surface over the years at the beginning and the end?

[Response] Thank you for this question regarding Figure 5.

The black areas in the left image represent non-impervious surface regions that remained unchanged throughout the study period. The figure displays the cumulative overlay of annual impervious surface expansion from 1985 to 2020, effectively showing the multi-year expansion pattern of impervious surfaces in the Beijing-Tianjin-Hebei region.

Specifically, this visualization technique layers all annual impervious surface data from 1985 to 2020, with the color gradient (from light orange to dark orange) indicating the temporal sequence of impervious surface development. The lighter orange areas represent earlier impervious surface development (closer to 1985), while the darker orange areas represent more recent development (closer to 2020). The black areas indicate regions that remained as non-impervious surfaces (such as forests, water bodies, agricultural land, and undeveloped areas) throughout the entire 35-year study period.

We add a clearer explanation of this visualization method in the figure caption to avoid confusion. (Page 11，L380-381,383-384)

 

3.12 The font sizes in the picture need to be standardized to enhance clarity and readability.

[Response] Thank you for this important suggestion regarding figure clarity and readability. We have carefully reviewed all figures in the manuscript and standardized the font sizes and layouts according to your recommendations to enhance clarity and readability. The modified figures with improved typography, consistent font sizes, and better visual presentation have been updated throughout the manuscript and are highlighted in red in the revised version.

Figure R2 (This is Figure 7 in Chapter 3.2 of the main text). EVI Classification Map (a) 2005 (b) 2010 (c) 2015 (d) 2020.

Figure R3 (This is Figure 11 in Chapter 3.4 of the main text). Figure 11. Spatial coupling relationship between urbanization process and ecological vulnerability. (a) Pearson correlation coefficient between ISC and EVI change values during 2005–2020 (b) Representative study areas (c) The magnitude, timing and duration of impervious surface increase around Beijing Capital International Airport and Temple of Heaven Park).

 

3.13 In the left picture of Figure 9, what do the dotted lines represent?

[Response] Thank you for this question regarding the representation in Figure 9a.

In the left panel of Figure 9, the dots represent the mean values of EVI calculated from all rasterized pixels for each respective year (2005, 2010, 2015, and 2020). The dotted lines represent the linear regression trend lines fitted to show the temporal changes in EVI over the study period. The regression equations show:

High-density ISC areas: y = 0.0031x - 5.9201 (R² = 0.7047), indicating a positive trend

Low-density ISC areas: y = -0.0004x + 1.1172 (R² = 0.1077), indicating a slight negative trend

This visualization approach helps distinguish between the actual observed values and the statistical trend analysis, providing both empirical data points and trend interpretation for the temporal coupling relationship between urbanization and ecological vulnerability.

 

3.14 The title of Figure 10 is not accurate enough and needs to be combined with the specific information shown in the figure. “ab patch density cd patch area”: The expression of the subheadings of the pictures can be made clearer. This occurs multiple times in the text.

[Response] Thank you for this suggestion. We have thoroughly revised all figure titles and subplot labeling throughout the manuscript to make them more accurate and informative. Specifically: We have changed all subplot references from the unclear format (e.g., "ab patch density cd patch area") to a clearer standardized format using individual subplot identifiers like (a), (b), (c), (d), etc. This makes it much easier for readers to identify and reference specific parts of multi-panel figures.

 

3.15 The results are integrated with the discussion in the text. However, the discussion section in the current text is not sufficient and needs to be enhanced to demonstrate scientificity.

[Response] We sincerely apologize for the insufficient discussion in the original manuscript. Thank you for the valuable feedback. Following the reviewer's suggestions, we have completely restructured the Discussion section (Section 4) to provide comprehensive validation and critical analysis of our results.

The revised Discussion is now organized into four distinct subsections that systematically address the scientific rigor and validity of our findings:

Section 4.1 Comparison with Other Global Urban Products

We conducted comprehensive comparisons with established global urban products including GHSL, ESA-CCI, Zhou et al.'s dataset, and GISA across multiple urban agglomerations. These comparisons provide essential benchmarking against internationally recognized datasets and validate the accuracy and reliability of our methodology through quantitative assessments.

Section 4.2 Contributions and Limitations of ISC Data Construction

We addressed the critical gap between binary classification and continuous coverage monitoring, particularly valuable for characterizing low-density development patterns. We systematically discussed classification errors, spatial aggregation effects, and inherited uncertainties from GAIA source data. This section provides transparent acknowledgment of methodological limitations while highlighting the unique contributions of our approach.

Section 4.3 Contributions and Limitations of EVI Assessment

We highlighted the advantages of using the Genetic Projection Pursuit model over subjective expert judgment approaches and discussed temporal resolution constraints (5-year intervals), fixed coefficient assumptions in HQI calculations, and potential oversimplification of ecological processes. We also addressed climate variability impacts and the challenges of capturing short-term environmental fluctuations, providing a balanced assessment of both strengths and limitations.

Section 4.4 Implications for Areas with Coexisting Urbanization and Rural Development We provided specific, evidence-based recommendations derived from our findings, including differentiated urban growth management strategies and location-specific approaches. This section bridges the gap between scientific findings and practical applications, demonstrating the real-world relevance of our research.

The enhanced Discussion section now spans four pages (pages 19-22) and provides the comprehensive scientific validation and critical analysis requested. Please refer to the revised Section 4 (pages 19-22) for the complete enhanced discussion.

 

3.16 At present, are there any other data on the distribution of impervious surfaces that can be compared with the data in the text to demonstrate the reliability of the data presented herein?

[Response] Thanks for the reviewer to point out this issue. We apologize for the vague validation part. We have implemented a comprehensive validation approach using multiple independent datasets and methods.

First, our data has its own foundation accuracy, the ISC data is derived from the mature impervious surface product GAIA through spatial aggregation processes. The underlying GAIA dataset has been extensively validated and demonstrates high accuracy with overall accuracies ranging from 89% to 97% across validation years, achieving a mean overall accuracy higher than 90% globally (Gong et al., 2020). This provides a solid foundation for the reliability of our ISC product.

Second, we conduct a multi-dataset cross-validation using four additional independent datasets:

• GHSL (Global Human Settlement Layer): A well-established global dataset for built-up area mapping.
• GISA (Global Impervious Surface Area): A comprehensive impervious surface dataset (available through 2019).
• ESACCI (European Space Agency Climate Change Initiative): A global land cover dataset.
• Zhou et al. (2018): Urban dynamics dataset from "A global record of annual urban dynamics (1992–2013) from nighttime lights".

The three histograms in Fig R4 presented Pearson correlation coefficients between GHSL and ISC values calculated from sample points across three major urban agglomerations for seven time periods (1985, 1990, 1995, 2000, 2005, 2010, and 2015):

Beijing-Tianjin-Hebei region: μ = 0.8218, σ = 0.1750；Yangtze River Delta: μ = 0.8899, σ = 0.1221；Pearl River Delta: μ = 0.8943, σ = 0.1171. The red dashed lines represent the mean correlation values (μ), while σ indicates standard deviation. These high correlation coefficients (>0.82 across all regions) demonstrate strong agreement between ISC and GHSL data, indicating high temporal accuracy of the ISC product.

The bottom panels in Fig R5 show validation against multiple independent datasets. Panel (a) presents a time series comparison of annual impervious surface area trends from four datasets (ISC, ESACCI, Zhou et al. 2018, and GISA) over 1985-2019, demonstrating that ISC captures similar temporal patterns and magnitudes to all three established global products. Panel (b) focuses on the dataset showing the most consistent trends with ISC - GISA - using 700 random sample points (20 points per year from 1985-2019) for detailed correlation analysis between ISC and GISA impervious surface areas. The strong correlation (r = 0.897) between ISC and GISA, combined with the consistent temporal trends observed across all four datasets, may provides robust evidence for the accuracy and reliability of our ISC product across the entire study period. We have incorporated this validation analysis into the "Result Validation and Uncertainty Analysis" section in the discussion for reviewers' reference.

Fig. R4(This is Figure 12 in Chapter 4.1 of the main text). Pearson correlation coefficient distributions between ISC and GHSL data across three major urban agglomerations (Beijing-Tianjin-Hebei, Yangtze River Delta, and Pearl River Delta) for seven time periods (1985-2015)

 

Fig. R5(This is Figure 13 in Chapter 4.1 of the main text). Comparison of ISC with other global urban products over past

years. (a) Time series comparison of annual impervious surface area trends (b) Correlation analysis of ISC with GISA from 700 random sampling points (1985-2019)

 

3.17 The obtained ecological vulnerability statistics and other results in the text can they be compared with the actual situation to better verify the methodology?

[Response] We greatly appreciate the reviewer's suggestion to compare our ecological vulnerability statistics and results with actual conditions to better verify our methodology. We have enhanced our results validation by comparing our EVI findings with documented real environmental conditions in the Beijing-Tianjin-Hebei region:

Our identified high-vulnerability expansion areas align with regions where environmental protection policies have been intensified, such as the ecological red line zones and areas subject to industrial relocation policies. Our EVI patterns show consistency with official air quality monitoring data, water quality assessments, and biodiversity surveys published by local environmental protection agencies.

We have verified our temporal EVI trends against documented environmental changes. The EVI improvements in certain high-density areas correspond with the implementation of major green infrastructure projects and urban ecological restoration programs. Our EVI results show strong consistency with the ecological vulnerability patterns reported by Fu Gang et al. (2016) in "Ecological Vulnerability Assessment in Beijing Based on GIS Spatial Analysis". Both studies identify similar high-vulnerability zones in Beijing's urban core and low-vulnerability areas in the northern mountainous regions, providing confidence in our methodology's validity.

We have also substantially expanded our discussion section (Section 4.3) to provide comprehensive analysis of how our results align with actual environmental conditions, strengthening the empirical validation of our methodology. Future studies will incorporate more extensive field validation and real-time environmental monitoring data to further enhance the accuracy and reliability of our vulnerability assessment framework.

 

3.18 The literature discussion is relatively scarce. The analysis of the current situation in the introduction needs to be strengthened with reference to the literature.

[Response] We sincerely apologize for the insufficient literature discussion in the original introduction section. We acknowledge that the analysis of the current research situation was not adequately supported by comprehensive literature review.

In response to this valuable feedback, we have undertaken the following revisions. We have substantially revised and expanded the introduction to provide a more thorough analysis of the current research landscape. The revised introduction now includes: A more comprehensive review of existing impervious surface mapping methodologies and their limitations; Enhanced discussion of current ecological vulnerability assessment approaches; Better contextualization of our research within the broader field of urban-ecological research; Clearer identification of research gaps that our study addresses. And the newly added reference to the literature are marked red in the text.

 

Reference

Xu, H., Wang, M., Shi, T., Guan, H., Fang, C., & Lin, Z. (2018). Prediction of ecological effects of potential population and impervious surface increases using a remote sensing based ecological index (RSEI). Ecological Indicators, 93, 730–740. https://doi.org/10.1016/j.ecolind.2018.05.055

FU Gang, BAI Jia-de, QI Yue, et al. Ecological Vulnerability Assessment in Beijing Based on GIS Spatial Analysis[J]. Journal of Ecology and Rural Environment, 2018, 34(9): 830-839.

 

Author Response File: Author Response.docx

Reviewer 4 Report

Comments and Suggestions for Authors

The manuscript presents a long-term time-series estimation of impervious surface coverage in the Beijing-Tianjin-Hebei (BTH) region and analyzes the vulnerability of the ecological environment response. While it holds moderate research value, it lacks significant innovation points. The main concerns are outlined below:

1. Data Calculation Clarification Needed: The reported increase of impervious surface area from 5,272.3 km² in 1985 to 32,171.1 km² in 2018, with an implied average annual growth rate of 88.5%, appears mathematically inconsistent. Please provide the exact calculation formula used. The standard formula [(Final Value - Initial Value) / Initial Value] / Time Span does not yield this result. Clarification of the calculation methodology is essential.
2. Insufficient Attribution Analysis in Conclusions: Conclusions regarding "continuously intensifying grassland fragmentation" and the "fluctuating downward trend of grassland area in low-density construction zones," as well as the "continued expansion of high-vulnerability zones in the Beijing-Tianjin core areas" and the "significant decline in the proportion of low-vulnerability zones in northern mountainous areas," lack thorough differentiation between the potential driving factors. A more nuanced attribution analysis is required.
3. Underdeveloped Policy Recommendations: The policy suggestions are notably brief. Expanding these with concrete, actionable recommendations specific to the findings would significantly enhance the applied value and practical significance of the research.
4. Missing Limitations and Future Work: The manuscript omits a critical discussion of the study's limitations and does not propose clear directions for future research. A dedicated section addressing these aspects is necessary.
5. Justification for Case Study Selection: The rationale for selecting the BTH region as the focus of this study, particularly within the context of global urbanization trends, needs strengthening. Explicitly justify why this specific region warrants in-depth analysis.
6. Structural Recommendation - Separate Discussion: The discussion section should be distinctly separated from the results and presented as an independent chapter to improve clarity and structural organization.

Author Response

Dear reviewers ,

 

We gratefully appreciate the editors and all reviewers for their time spend making positive and constructive comments. These comments are all valuable and helpful for revising and improving our manuscript entitled “Long-term Time Series Estimation of Impervious Surface Coverage Rate in Beijing-Tianjin-Hebei Urbanization and Vulnerability Assessment of Ecological Environment Response” (Manuscript Number: land-3763944) as well as the important guiding significance to our researches.

 

We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the revised original manuscript. The responses to the reviewer's comments are listed in the Revision Report.

 

We sincerely hope the revised version could encourage your acceptance of this paper and look forward to your comments.

 

Yours sincerely,

Authors of the manuscript land-3763944

 

 

Response to the Reviewer’s Comments

We sincerely appreciate the time taken by the three reviewers to provide us with valuable comments and constructive suggestions on our manuscript. We have made significant revisions to our manuscript, following the advice of the reviewers. Below the original comments from reviewers are copied in blue italics, followed by our point-to-point responses in roman font. Additional or altered text in the revised manuscript, when cited in this response document, is marked in red.

 

 

To reviewer #4

Reviewer: #4 (Comments to the Author)

General comments

The manuscript presents a long-term time-series estimation of impervious surface coverage in the Beijing-Tianjin-Hebei (BTH) region and analyzes the vulnerability of the ecological environment response. While it holds moderate research value, it lacks significant innovation points.

[Response] We thank the reviewer for the comments about our study. The reviewer provided us with many pertinent and useful suggestion. We realized that the manuscript was still lacking and revised it. In particular, we have restructured the article by separating the Results and Discussion sections, enhanced the content of the Discussion section, and clarified unclear descriptions throughout the manuscript to improve overall clarity and readability. Hopefully, our revisions will make improvement to the manuscript. We are keen to answer these questions in detail as follows.

 

Specific comments

4.1 Data Calculation Clarification Needed: The reported increase of impervious surface area from 5,272.3 km² in 1985 to 32,171.1 km² in 2018, with an implied average annual growth rate of 88.5%, appears mathematically inconsistent. Please provide the exact calculation formula used. The standard formula [(Final Value - Initial Value) / Initial Value] / Time Span does not yield this result. Clarification of the calculation methodology is essential.

[Response] We apologize for the calculation error in the original manuscript. The reported "year-on-year growth of 88.5%" was mathematically incorrect. We have recalculated using the proper formulas:

Calculation Formulas:

Total growth rate = [(Final Value - Initial Value) / Initial Value] × 100%

Annual average increase = (Final Value - Initial Value) ÷ Number of Years

Corrected Results:

According to calculations, the impervious surface area in the Beijing-Tianjin-Hebei region increased from 7,579.3 km² in 1985 to 37,484.0 km² in 2020, representing a total growth of 394.5% over the 35-year period, with an average annual increase of 854.4 km²/year.

Calculation Details:

Total growth rate = [(37,484.0 - 7,579.3) / 7,579.3] × 100% = 394.5%

Average annual increase = (37,484.0 - 7,579.3) ÷ 35 = 854.4 km²/year

We have corrected this calculation error in the manuscript and ensure all growth rate calculations follow these standard formulas. (Page 11，L395-398)

4.2 Insufficient Attribution Analysis in Conclusions: Conclusions regarding "continuously intensifying grassland fragmentation" and the "fluctuating downward trend of grassland area in low-density construction zones," as well as the "continued expansion of high-vulnerability zones in the Beijing-Tianjin core areas" and the "significant decline in the proportion of low-vulnerability zones in northern mountainous areas," lack thorough differentiation between the potential driving factors. A more nuanced attribution analysis is required.

[Response] We sincerely appreciate the reviewer's insightful comment regarding the insufficient attribution analysis in our conclusions. We acknowledge that our current conclusions lack thorough differentiation of the potential driving factors behind the observed phenomena. We have revised the conclusions section to provide a more nuanced attribution analysis for the key findings mentioned. (Page 22，L745-747, L749-754).

4.3 Underdeveloped Policy Recommendations: The policy suggestions are notably brief. Expanding these with concrete, actionable recommendations specific to the findings would significantly enhance the applied value and practical significance of the research.

[Response] We sincerely apologize for the insufficient policy recommendations in the original manuscript. Thank you for this constructive feedback. Following the reviewer's suggestion, we have significantly expanded and enhanced the policy recommendations to provide concrete, actionable guidance that directly stems from our research findings.

The comprehensive policy recommendations have been incorporated into Section 4.4 Implications for Areas with Coexisting Urbanization and Rural Development. This section now provides detailed, evidence-based policy guidance including:

• Differentiated urban growth management strategies tailored to specific development patterns identified in our analysis
• Location-specific approaches for managing the urbanization-rural development interface
• Practical guidelines for balancing urban expansion with ecological preservation

These expanded policy recommendations are directly derived from our ISC data construction results and EVI assessment findings, ensuring they are scientifically grounded and practically applicable.

Please refer to Section 4.4 (pages 21-22) for the complete expanded policy recommendations.

4.4 Missing Limitations and Future Work: The manuscript omits a critical discussion of the study's limitations and does not propose clear directions for future research. A dedicated section addressing these aspects is necessary.

[Response] We sincerely apologize for the insufficient discussion of limitations and absence of future research directions in the original manuscript. Thank you for this important feedback.

Following the reviewer's suggestions, we have restructured the manuscript by separating the original combined Results and Discussion sections to allow for more comprehensive analysis. The study limitations and future research directions are now systematically addressed as follows:

Study Limitations: The limitations discussion has been integrated into the Discussion section, with dedicated coverage for each methodological component:

Section 4.2 addresses the limitations of ISC data construction methodology

Section 4.3 discusses the limitations of EVI assessment approach

This structure ensures that limitations are discussed in direct context with their respective methods, providing readers with a clear understanding of the constraints and scope of each analytical component. Please refer to page 20-21, line 648-710 in the revised manuscript.

Future Research Directions: The future work discussion has been incorporated into the Conclusion section at the end of the manuscript, where we propose directions for advancing this research area based on the comprehensive limitations analysis presented in the Discussion section. Please refer to pages 22-23 line 767-784 in the revised manuscript.

4.5 Justification for Case Study Selection: The rationale for selecting the BTH region as the focus of this study, particularly within the context of global urbanization trends, needs strengthening. Explicitly justify why this specific region warrants in-depth analysis.

[Response] We appreciate the reviewer's request for stronger justification of our case study selection. The Beijing-Tianjin-Hebei (BTH) region was selected as our focus area based on multiple compelling factors that make it an ideal laboratory for studying urbanization-ecology interactions within the global urbanization context.

The BTH region represents one of the world's most significant urbanization laboratories, encompassing 110 million people across 216,000 km² and accounting for ~10% of China's GDP. The region's 394.5% increase in impervious surface area over our 35-year study period exemplifies the dramatic urban transformations characteristic of 21st-century global development. BTH exhibits the "leaping, edge-sparse, and discontinuous" expansion patterns increasingly common in developing economies worldwide, while its polycentric development model (Beijing-Tianjin dual cores plus satellite cities) mirrors urbanization patterns observed in other major global regions such as the Pearl River Delta, São Paulo metropolitan area, and emerging African urban corridors.

From a methodological perspective, BTH offers exceptional advantages for testing our ISC methodology and vulnerability assessment framework. The region provides excellent remote sensing data coverage and quality, comprehensive long-term datasets enabling robust 35-year time series analysis, and encompasses diverse ecological zones from megacity cores to pristine mountainous areas. This environmental gradient diversity, combined with multiple climate zones and topographic conditions, enhances the transferability of our findings to other regions. Additionally, BTH represents one of the world's most environmentally stressed urban regions, with air quality challenges, water scarcity, and ecosystem degradation providing extreme test conditions for vulnerability assessment methods that are representative of challenges facing many rapid-growth urban areas globally. The research findings and methodological framework developed through this BTH case study directly address critical knowledge gaps in understanding urbanization-environment linkages in mega-regions, aligning with UN SDG 11 research priorities. Our 35-year study period captures complete cycles of policy implementation, economic transition, and environmental change, covering critical periods in global urbanization from pre-reform (1980s) through sustainable development transition (2010s-2020s).

We have revised add the justify for case study in the 2.1 study area Section, please refer to the revised manuscript Page 4, L142-151.

4.6 Structural Recommendation - Separate Discussion: The discussion section should be distinctly separated from the results and presented as an independent chapter to improve clarity and structural organization.

[Response] We sincerely apologize for the insufficient discussion in the original manuscript. Thank you for the valuable feedback. Following the reviewer's suggestions, we have completely restructured the Discussion section (Section 4) to provide comprehensive validation and critical analysis of our results.

The revised Discussion is now organized into four distinct subsections that systematically address the scientific rigor and validity of our findings:

Section 4.1 Comparison with Other Global Urban Products

We conducted comprehensive comparisons with established global urban products including GHSL, ESA-CCI, Zhou et al.'s dataset, and GISA across multiple urban agglomerations. These comparisons provide essential benchmarking against internationally recognized datasets and validate the accuracy and reliability of our methodology through quantitative assessments.

Section 4.2 Contributions and Limitations of ISC Data Construction

We addressed the critical gap between binary classification and continuous coverage monitoring, particularly valuable for characterizing low-density development patterns. We systematically discussed classification errors, spatial aggregation effects, and inherited uncertainties from GAIA source data. This section provides transparent acknowledgment of methodological limitations while highlighting the unique contributions of our approach.

Section 4.3 Contributions and Limitations of EVI Assessment

We highlighted the advantages of using the Genetic Projection Pursuit model over subjective expert judgment approaches and discussed temporal resolution constraints (5-year intervals), fixed coefficient assumptions in HQI calculations, and potential oversimplification of ecological processes. We also addressed climate variability impacts and the challenges of capturing short-term environmental fluctuations, providing a balanced assessment of both strengths and limitations.

Section 4.4 Policy Recommendations for Areas with Coexisting Urbanization and Rural Development

We provided specific, evidence-based recommendations derived from our findings, including differentiated urban growth management strategies and location-specific approaches. This section bridges the gap between scientific findings and practical applications, demonstrating the real-world relevance of our research.

The enhanced Discussion section now spans four pages (pages 19-22) and provides the comprehensive scientific validation and critical analysis requested. Please refer to the revised Section 4 (pages 19-22) for the complete enhanced discussion.

 

Author Response File: Author Response.docx

Reviewer 5 Report

Comments and Suggestions for Authors

In Figure 1...explain which source they used to generate the DEM.

Table 1...the authors mention that they used MODIS to calculate LST at 30 meters...which is incorrect. There is no MODIS LST product at 30 meters...please clarify this aspect...and if the pixel was resampled...it would be a potential error because the smallest pixel in LST-MODIS is 500 meters, and resampling it at 30 meters would be an underestimation or overestimation.

Add GRID to Figure 3.

Improve the resolution of the images in Figure 4.

Important observation...

The authors need to add a comparison of LST versus SST values ​​(Po Hai Bay) to their analysis because...

The scientific literature indicates that there is a relationship between LST versus SST in urban areas, due to ocean-atmosphere exchange. It is one of the main factors influencing temperature changes in the coastal zone.

Authors can consult the following article to obtain and analyze the LST versus SST relationship:

Sandoval, S., Escobar-Flores, J. G., & Badar Munir, M. (2023). Urbanization and its impacts on land surface temperature and sea surface temperature in a tourist region in Mexico from 1990 to 2020. Remote Sensing Applications: Society and Environment, 32, 101046.

Moazzam, M. F. U., Lee, B. G., & Kim, S. (2025). Impact of Sea Surface Temperature on City Temperature near Warm and Cold Ocean Currents in Summer Season for Northern Hemisphere. Atmosphere, 16(1), 54. https://doi.org/10.3390/atmos16010054

 

And the following website to obtain SST data:

https://giovanni.gsfc.nasa.gov/giovanni/

 

Comments on the Quality of English Language

nothing

Author Response

Dear reviewers and editors,

 

We gratefully appreciate the editors and all reviewers for their time spend making positive and constructive comments. These comments are all valuable and helpful for revising and improving our manuscript entitled “Long-term Time Series Estimation of Impervious Surface Coverage Rate in Beijing-Tianjin-Hebei Urbanization and Vulnerability Assessment of Ecological Environment Response” (Manuscript Number: land-3763944) as well as the important guiding significance to our researches.

 

We have studied comments carefully and have made correction which we hope meet with approval. Revised portion are marked in red in the revised original manuscript. The responses to the reviewer's comments are listed in the Revision Report.

 

We sincerely hope the revised version could encourage your acceptance of this paper and look forward to your comments.

 

Yours sincerely,

Authors of the manuscript land-3763944

 

 

Response to the Reviewer’s Comments

We sincerely appreciate the time taken by the three reviewers to provide us with valuable comments and constructive suggestions on our manuscript. We have made significant revisions to our manuscript, following the advice of the reviewers. Below the original comments from reviewers are copied in blue italics, followed by our point-to-point responses in roman font. Additional or altered text in the revised manuscript, when cited in this response document, is marked in red.

 

 

To reviewer #5

Reviewer: #5 (Comments to the Author)

[Response] Thank the reviewer for recognizing the important value of our work. We appreciate him/her for valuable comments and suggestions. Following the comments and suggestions, we have carefully revised the manuscript as follows.

 

5.1 In Figure 1...explain which source they used to generate the DEM.

[Response] The DEM (Digital Elevation Model) shown in Figure 1b was generated using NASA's global 30-meter resolution SRTM (Shuttle Radar Topography Mission) elevation data. As mentioned in our methodology and Table 1, we utilized elevation data from the L-band Phased Array Synthetic Aperture Radar (PALSAR) of the Advanced Land Observing Satellite (ALOS), specifically using the NASA global 30m SRTM elevation DEM dataset to extract elevation, topographic relief, and slope information for the Beijing-Tianjin-Hebei region.

We have add this specific data source information to the Figure 1 caption to provide clear attribution.

 

5.2 Table 1...the authors mention that they used MODIS to calculate LST at 30 meters...which is incorrect. There is no MODIS LST product at 30 meters...please clarify this aspect...and if the pixel was resampled...it would be a potential error because the smallest pixel in LST-MODIS is 500 meters, and resampling it at 30 meters would be an underestimation or overestimation.

[Response] Thank you for pointing out this important error in Table 1. We sincerely apologize for the mistake in our description.

You are absolutely correct that there is no MODIS LST product available at 30 meters resolution, and resampling from 500m to 30m would indeed introduce significant estimation errors. We have corrected this error in the revised manuscript.

We used 1000m resolution MODIS data to calculate both FVC and LST. All five indicators used in our EVI assessment (FVC, LST, HQI, slope, and population density) were unified to 1000m resolution to maintain consistency across all datasets. Specifically, MODIS data (FVC and LST) is used at 1000m native resolution - no inappropriate resampling was performed. HQI is calculated from 30m CLCD data and then resampled to 1000m using spatial aggregation methods. Slope is derived from 30m SRTM DEM and resampled to 1000m. Population density is Originally at 1000m resolution.

We have corrected Table 1 to accurately reflect that MODIS data was used at 1000m resolution, not 30m.

 

5.3 Improve the resolution of the images in Figure 4.

[Response]Thanks for pointing this out. We redesigned the layout to enhance the clarity and visibility of the image.

Figure R1 (This is Figure 4 in Chapter 2.3.5 of the main text). Spatial Distribution Maps of Normalized Factors in the Study Area.

 

5.4 The authors need to add a comparison of LST versus SST values ​​(Po Hai Bay) to their analysis because...

The scientific literature indicates that there is a relationship between LST versus SST in urban areas, due to ocean-atmosphere exchange. It is one of the main factors influencing temperature changes in the coastal zone.

[Response]We appreciate the reviewer's reference to the interesting study on sea surface temperature (SST) impacts on urban temperature. However, we would like to clarify the scope and objectives of our research, which differ fundamentally from the cited coastal city study:

1. Study focus and geographic context: Our research focuses on the Beijing-Tianjin-Hebei (BTH) region, which is primarily an inland continental area with limited direct oceanic influence. While the region borders the Bohai Sea to the east, the majority of our study area (approximately 85%) consists of inland plains and mountainous regions where continental climate factors dominate over oceanic influences.
2. LST as an ecological indicator vs. climate driver: In our study, LST serves as an ecological vulnerability indicator rather than a climate variable requiring oceanic explanations. We use LST to represent thermal stress on ecosystems and urban heat island effects caused by land cover changes, urbanization intensity, and vegetation loss - all terrestrial processes that are well-documented in continental regions.
3. Methodological consistency: Our EVI framework integrates five terrestrial factors (FVC, HQI, LST, slope, population density) that collectively represent ecosystem stress from urbanization pressures. Adding SST would introduce a marine climate variable that is not directly relevant to the terrestrial ecological processes we are investigating and would compromise the internal consistency of our indicator system.
4. Scale and resolution considerations: Our analysis operates at 1km resolution across a 216,000 km² primarily inland region. The spatial extent and inland location make SST influences negligible compared to local land use/land cover changes, topographic effects, and urbanization intensity - which are the primary drivers of ecological vulnerability in this continental setting.
5. Validation of terrestrial approach: Our results showing clear relationships between urbanization intensity gradients and ecological vulnerability patterns, landscape fragmentation changes, and vegetation dynamics confirm that terrestrial factors adequately explain the observed ecological responses without requiring oceanic variables.

We sincerely apologize if our manuscript did not adequately explain the rationale for focusing on terrestrial factors in this continental study context. While SST analysis would be valuable for coastal urban studies, our continental focus on urbanization-ecology interactions in the BTH region is appropriately addressed through terrestrial indicators that directly capture the land use change impacts we are investigating.

Nevertheless, we greatly appreciate the reviewer's thoughtful suggestion and acknowledge the importance of ocean-atmosphere interactions in coastal regions. In our future research, particularly when extending our methodology to coastal urban agglomerations such as the Pearl River Delta or when studying the eastern coastal portions of the BTH region in greater detail, we will certainly consider incorporating SST variables to capture the marine climate influences on urban ecological vulnerability. This could provide valuable insights for comprehensive regional assessments that span both terrestrial and coastal environments.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

.

Reviewer 4 Report

Comments and Suggestions for Authors

I have no further comments on this manuscript.