Evolution Characteristics and Influencing Factors of Agricultural Drought Resilience: A New Method Based on Convolutional Neural Networks Combined with Ridge Regression

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

 

This study constructed a convolutional neural network model optimized by the Adam with weight decay algorithm (AdamW-CNN) and selected 16 agricultural drought resilience indicators. Using multi-temporal data from Qiqihar City, Heilongjiang Province, China, the drought resilience indices of four typical regions were simulated and analyzed. The results demonstrate that the AdamW-CNN model combined with the KOA-ridge model outperforms both the RMProp-CNN and CNN models in terms of fitting accuracy, stability, reliability, and evaluation precision. This research holds significant scientific value, providing a robust model for measuring agricultural drought resilience and offering a scientific basis for drought resistance studies. However, several issues require attention:

 

1. What do the different colors in the article represent?

 

2. A robust model typically requires multi-site validation. Since this study was conducted only in Qiqihar, how reliable is the model for broader applications?
3. The abbreviations in each table should be clearly defined in the table notes.
4. The abbreviations in each table should be clearly defined in the figure notes.
5. Table 4 Correction: There are no asterisks (*) in Table 4; please remove any references to them.
6. Conclusion Enhancement: The conclusion should not merely reiterate the results but instead summarize the key findings and their broader implications.

Author Response

Thank you for your comments on our manuscript. Those comments were all highly valuable and helpful for revising and improving our manuscript, and they were an important guide to our research. We have revised our manuscript, and we assure that your concerns have been addressed properly. The corrections in the paper are marked in red, and our responses are listed below.

Comments1: What do the different colors in the article represent?

Response1: Dear reviewer, we are very sorry for the confusion caused by the different colors in the article. In fact, the article has gone through a round of review process, and we have marked it with different colors to highlight the modified content.

Comments2: A robust model typically requires multi-site validation. Since this study was conducted only in Qiqihar, how reliable is the model for broader applications?

Response2: Dear Reviewer, thank you very much for your insightful comment. We fully agree that a robust and generalizable model ideally requires validation across multiple sites. In our study, we were indeed limited to data from Qiqihar due to constraints in data availability and accessibility. This limitation was carefully considered during the research design. While our models—based on Convolutional Neural Networks and Ridge Regression—are built upon well-established methodologies widely applied across various domains, we acknowledge that performance in one region does not guarantee applicability elsewhere. To address this concern, we focused on optimizing these models using the AdamW algorithm and the Kepler Optimization Algorithm, respectively, and conducted comprehensive performance evaluations through comparative analysis against baseline models. These evaluations demonstrated strong performance within our study area.

Nonetheless, we recognize that external validation is essential to assess the broader applicability of our models. As such, we have explicitly noted in the conclusion section that further validation using data from different regions is a crucial direction for future research. We sincerely appreciate your suggestion, which we believe will help strengthen future iterations of our work.

Comments3: The abbreviations in each table should be clearly defined in the table notes.

Response3: Dear reviewer, thank you for your valuable comments. We have revised, please refer to the revised manuscript.

Comments4: The abbreviations in each table should be clearly defined in the figure notes.

Response4: Dear reviewer, thank you for your valuable comments. Since the abbreviations of some figures in the manuscript have been annotated in the previous text, in order to avoid repeated annotations, the abbreviations in the figures are not annotated in the form of figure captions.

Comments5: Table 4 Correction: There are no asterisks (*) in Table 4; please remove any references to them.

Response5: Dear reviewer, thank you for your valuable comments. We have removed the asterisks (*) in the note to Table 4.

Comments6: Conclusion Enhancement: The conclusion should not merely reiterate the results but instead summarize the key findings and their broader implications.

Response6: Dear reviewer, thank you for your valuable comments. The conclusion section has been revised to delete general conclusions and highlight the important findings and practical value of the research, please refer to the revised manuscript.

We have revised the manuscript under your guidance, and we hope it will be met with approval.

Best regards,

Authors

Reviewer 2 Report

Comments and Suggestions for Authors

 

The history of mankind is a history of struggle against disasters. Using the macro statistical data, the author pays attention to the agricultural flood resilience and influencing factors in Heilongjiang Province. In general, the research topic has certain research significance, and the argumentation process is relatively solid. In order to better improve the quality of the paper, several suggestions for reference:

(1) In the introduction, the definition of core concepts should be clearly given, which is the key to the construction of the subsequent measurement index system and the selection of methods.

(2) The research review on agricultural flood resilience is not enough, and the marginal contribution of the research is not so convincing, and the review needs to be strengthened.

(3) Many methods are used in the research, why these methods are used, and where the logical correlation between methods and methods needs to be further explained.

(4) The selection of variables needs to have a more adequate theoretical basis.

Author Response

Thank you for your comments on our manuscript. Those comments were all highly valuable and helpful for revising and improving our manuscript, and they were an important guide to our research. We have revised our manuscript, and we assure that your concerns have been addressed properly. The corrections in the paper are marked in red, and our responses are listed below.

Comments1: In the introduction, the definition of core concepts should be clearly given, which is the key to the construction of the subsequent measurement index system and the selection of methods.

Response1: Dear reviewer，thank you for your valuable comments. I completely agree with your constructive suggestion. In lines 80-89 of the introduction, a clear concept and definition of agricultural drought resilience are given.

Comments2: The research review on agricultural flood resilience is not enough, and the marginal contribution of the research is not so convincing, and the review needs to be strengthened.

Response2: Dear reviewer，thank you for your valuable comments. This paper elaborates the research progress of agricultural drought resilience from the aspects of research background, influencing factors and research methods, and points out the shortcomings of existing research methods. In this regard, we try to construct a convolutional neural networks model based on the Adam optimization algorithm based on weight decay method. The potential contribution of this study is to improve the accuracy of agricultural drought resilience measurement, which provides a more reliable new method for research in this field. Please refer to lines 93-153 in the revised manuscript.

Comments3: Many methods are used in the research, why these methods are used, and where the logical correlation between methods and methods needs to be further explained.

Response3: Dear reviewer，thank you for your valuable comments. I completely agree with your constructive suggestions. In lines 183-194 of the introduction, the application of the methods and the logical correlation between methods and methods needs has been explained, and the flowchart of the paper is drawn to make it easier for the reader to understand.

Comments4: The selection of variables needs to have a more adequate theoretical basis.

Response4: Dear reviewer，thank you for your valuable comments. I completely agree with your constructive suggestions. For the selection of parameter variables of the AdamW-CNN model in this manuscript, the references cited in lines 326-328 as theoretical basis. For the selection of parameter variables of the KOA-Ridge model in this manuscript, the references cited in lines 410-412 is used as a theoretical basis. For the indicators selected for DPSIR conceptual model, the reasons for selecting these indicators are added and supplement some references in lines 450-473.

We have revised the manuscript under the guidance of the editor/reviewers, and we hope it will be met with approval.

Best regards,

Authors

Reviewer 3 Report

Comments and Suggestions for Authors

I have had the pleasure of reviewing your manuscript titled "

 Evolution characteristics and influencing factors of agricultural drought resilience: A new method based on convolutional neural networks combined with ridge regression." I must commend your comprehensive and robust approach to this pertinent issue in the realm of sustainable agricultural science.

This manuscript tackles a crucial issue in sustainable agriculture using sophisticated techniques. The methodological development is MS’s main strength. However, to be suitable for Sustainability, it needs substantial revision to better justify the complex methods chosen.

Point 1: The use of a CNN, typically excelling in spatial data or sequential data with local patterns, for a set of 16 aggregated, potentially time-series, indicators derived from the DPSIR framework needs stronger justification. While CNNs can be applied to tabular data, it's not their primary strength. Why was CNN chosen over other potentially more suitable machine learning models for time-series or regression tasks on tabular data (e.g., LSTMs, Gradient Boosting Machines, Random Forests, even simpler NNs)? The text mentions CNN advantages like feature sharing and local connection (Lines 139-140), but how these specifically benefit this type of aggregated indicator data needs explicit explanation

Point 2: The introduction motivates the study well regarding the impacts of drought but could more explicitly connect agricultural drought resilience to the core principles of sustainable agriculture (e.g., long-term productivity, resource conservation, economic viability, social equity)

Point 3: The discussion of key driving factors (Precipitation, investment, grain output/area, per capita land, irrigation) is currently quite brief (Lines 573-582, 752-753). From a sustainable agriculture perspective, this needs significant expansion. Why are these factors dominant in the specific context of Qiqihar's farming systems (dryland, black soil, temperate monsoon)? How do they interact?

Point 4: The spatiotemporal analysis (Fig 10) identifies regional differences. This section needs more interpretation grounded in the agricultural realities of those specific counties/districts. What are the differences in farming systems, socio-economic conditions, or water access that might explain these resilience patterns?

Author Response

Thank you for your comments on our manuscript. Those comments were all highly valuable and helpful for revising and improving our manuscript, and they were an important guide to our research. We have revised our manuscript, and we assure that your concerns have been addressed properly. The corrections in the paper are marked in red, and our responses are listed below.

Comments1: The use of a CNN, typically excelling in spatial data or sequential data with local patterns, for a set of 16 aggregated, potentially time-series, indicators derived from the DPSIR framework needs stronger justification. While CNNs can be applied to tabular data, it's not their primary strength. Why was CNN chosen over other potentially more suitable machine learning models for time-series or regression tasks on tabular data (e.g., LSTMs, Gradient Boosting Machines, Random Forests, even simpler NNs)? The text mentions CNN advantages like feature sharing and local connection (Lines 139-140), but how these specifically benefit this type of aggregated indicator data needs explicit explanation

Response1: Dear reviewer, thank you for your valuable comments. We have added clear explanations as follows: There is a progressive and cyclic relationship between DPSIR indicators. When applying CNN to process aggregated time series indicators based on the DPSIR framework, its core advantage is that CNN can automatically capture the coordinated changes of indicator subsets in specific time segments through the local connection weight sharing mechanism. Compared with recursive network models such as LSTMs, CNN is more suitable for extracting indicator features with hierarchical correlation and local temporal causal chains under the DPSIR framework, and is more conducive to avoiding overfitting and maintaining model interpretability in multi-dimensional indicator scenarios.

Comments2: The introduction motivates the study well regarding the impacts of drought but could more explicitly connect agricultural drought resilience to the core principles of sustainable agriculture (e.g., long-term productivity, resource conservation, economic viability, social equity)

Response2: Dear reviewer, thank you for your valuable comments. In the Introduction section, we have added relevant content as follows: Improving agricultural drought resilience is inextricably linked to the core principles of sustainable agriculture. Its essence lies in achieving multi-dimensional synergy of stabilizing agricultural productivity, improving resource utilization efficiency, maintaining economic sustainability and promoting social equity by enhancing the adaptability and resistance of agricultural systems to drought.

Comments3: The discussion of key driving factors (Precipitation, investment, grain output/area, per capita land, irrigation) is currently quite brief (Lines 573-582, 752-753). From a sustainable agriculture perspective, this needs significant expansion. Why are these factors dominant in the specific context of Qiqihar's farming systems (dryland, black soil, temperate monsoon)? How do they interact?

Response3: Dear reviewer, thank you for your valuable comments. We have added relevant discussion content as follows: It can be seen that Precipitation, Grain output per unit cultivated area, Investment in primary industry, Per capita cultivated land area, and Proportion of effective irrigation area are the key constraints on the spatiotemporal changes in agricultural drought resilience in the study area. These five key factors are all positive indicators, and they all play an important role in promoting agricultural drought resilience in the study area. It is precisely because of the size differences of this type of indicators at different temporal and spatial scales and their combined effects that dominate the spatiotemporal evolution pattern of agricultural drought resilience in the study area. In the pressure layer, Per capita cultivated land area and Proportion of effective irrigation area are the bearing factors of drought resilience. The per capita cultivated land area reflects the amount of cultivated land resources available to each population. The larger the value, the more food is obtained through cultivation. Due to the large amount of food obtained in the early stage, it is easier to resist the risk of food production reduction when drought occurs. The Proportion of effective irrigation area reflects the ratio of farmland area that can be irrigated normally through the irrigation system in normal years. The larger the value, the larger the area that can be alleviated by the irrigation system when drought occurs. Precipitation, as the most direct natural factor indicator in the driving force system, is the main inducing factor for the occurrence of drought. Investment in primary industry is the most important indicator of the impact of economic factors on agriculture. The more investment in agricultural economy, the more conducive it is to improve the agricultural irrigation system, thereby enhancing the agricultural system's ability to resist drought and thus improve its resilience. Qiqihar, a specific research area, is affected by the temperate monsoon climate and belongs to a semi-arid area. Its agricultural type is dryland farming. Insufficient precipitation and consecutive spring droughts have become the most important factors restricting the sustainable development of local agriculture, and have also become the main factors affecting the drought resilience of local agriculture. This also forces the Proportion of effective irrigation area, an indicator that represents the effectiveness of the irrigation system, to become the main controlling factor of local drought resilience. In addition, Qiqihar is located in the black soil region of Northeast China and is an important grain production base in China. This also leads to the fact that Grain output per unit cultivated area and per capita cultivated land area in Qiqihar become typical factors affecting agricultural drought resilience. Investment in primary industry in Qiqihar is mainly used to improve local irrigation facilities. Regional differences in investment also directly affect the level of agricultural drought resilience.

Comments4: The spatiotemporal analysis (Fig 10) identifies regional differences. This section needs more interpretation grounded in the agricultural realities of those specific counties/districts. What are the differences in farming systems, socio-economic conditions, or water access that might explain these resilience patterns?

Response4: Dear reviewer, thank you for your valuable comments. We have added relevant content as follows: Among the spatial changes in agricultural drought resilience levels in different regions of Qiqihar in three stages, the drought resilience level of Kedong County in-creased less. The main reason is that its average annual precipitation is lower than that of other regions, the number of irrigation facilities is also smaller, and the investment in the primary industry is lower, which makes it easy to encounter drought and unable to recover in time. Therefore, the drought resilience of Kedong County in the third stage only increased from Level I to Level II. In the third stage, although Keshan County and Tailai County were at the III level, their drought resilience index was still at a relatively low level of level III. Although their investment in primary industry and per capita cultivated land area were at a relatively high level, due to the impact of continuous low rainfall in 2015-2016, their drought resilience index was at a relatively low level of level III. Longjiang County and Gannan County in the northwest are adja-cent to the urban area of Qiqihar. Against the background of high investment in the primary industry, although the precipitation is at a medium level in the whole region, their per capita cultivated land area and effective irrigation area have increased sig-nificantly. Therefore, the drought resilience of Longjiang County and Gannan County is at a relatively high level, both of which have been upgraded to Level IV.

We have revised the manuscript under your guidance, and we hope it will be met with approval.

Best regards,

Authors

Reviewer 4 Report

Comments and Suggestions for Authors

11 April 2025

Manuscript ID: sustainability-3574807

Title: Evolution characteristics and influencing factors of agricultural drought resilience: A new method based on convolutional neural networks combined with ridge regression

This manuscript presents a novel and methodologically interesting approach to evaluating agricultural drought resilience by combining a convolutional neural network (CNN) optimised with the AdamW algorithm and a ridge regression framework. The integration of machine learning techniques demonstrates a multi-dimensional strategy for assessing drought resilience at the regional scale. The focus on Qiqihar City, Heilongjiang Province, provides a well-defined case study, and the temporal span of the analysis (2000–2021) allows for meaningful insights into long-term trends. Overall, the topic is timely and relevant to the scope of Sustainability, and the manuscript holds potential to contribute to the scientific discourse on climate resilience in agriculture. However, there are several aspects that require clarification and revision to enhance the manuscript’s clarity, technical precision, and accessibility to a broader readership.

Please find below specific comments for the authors’ consideration:

1. While the manuscript offers a valuable contribution to the assessment of agricultural drought, there are some acronyms introduced in the abstract that are not defined upon first use, which may make it difficult for readers (who are not familiar with these terms) to understand the first approach to the manuscript. For instance, DPSIR should be expanded as Driving Forces-Pressures-State-Impact-Response (assuming that this is what it means) when first mentioned. Similarly, KOA in “KOA Ridge Regression” requires clarification, as it does not correspond to a widely recognised method and may need to be defined as a bespoke or domain-specific approach. Additionally, RMProp-CNN appears to be either a variant or a misspelling of RMSProp-CNN, and should be corrected or explicitly defined if it is indeed distinct. Including these definitions in the abstract will improve accessibility and ensure greater transparency of the methodology. It is also recommended that brief descriptions be provided in the methods section to reinforce understanding.
2. In the Abstract section please also include the area of Qiqihar City, it could be the exact area or an approximate value.
3. The keywords listed should appear in the abstract using the exact phrasing presented. For instance, “Adam weight decay optimizer” is not mentioned in the abstract in that exact form, and “Drought resilience evaluation” could be more explicitly reflected (there are some instances where “drought resilience” appears in the abstract but the key word “drought resilience evaluation” never appears in the abstract, please correct me if I am wrong. Please ensure consistency between the keywords and the terms used in the abstract.
4. In the first line of the Introduction, drought by nature is already a negative phenomenon, so saying it has a "negative impact" might feel redundant. Please correct the opening phrase. The phrase could say something, for instance: In recent years, drought has posed an increasingly serious threat to regional sustainable development and human society.
5. The authors mention the Palmer Drought Severity Index (PDSI) from 1965, but they do not address the Standardised Precipitation Index (SPI), introduced in 1993, or the Standardised Precipitation Evapotranspiration Index (SPEI), introduced in 2009. Please provide a brief discussion of these indices and how they relate to the assessment of drought.
6. A major issue with the paper is the lack of a clear definition of agricultural drought. It is important to ensure that readers fully understand what is meant by agricultural drought in the context of this study. This could be introduced in the first paragraph for clarity. For instance, for severe agricultural drought, SPEI values below -1.5 are typically associated with significant soil moisture deficits, leading to substantial agricultural impacts, such as reduced crop yields or crop failure. However, if a different definition of agricultural drought is intended, please clarify and provide a clear explanation in the paper. Kindly address this matter as thoroughly as possible.
7. Line 92: Please defined DPSIR.
8. Lines 151-158: The way the objectives are currently written seems more like a description of the methodology rather than clearly defined research objectives. Research objectives should typically be action-oriented, outlining the intended outcomes or goals of the study. So far they look like a description only.
9. Please add the following reference to the manuscript (specifically when approaching some previous drought analyses): Espinosa, L.A.; Portela, M.M.; Matos, J.P.; Gharbia, S. Climate Change Trends in a European Coastal Metropolitan Area: Rainfall, Temperature, and Extreme Events (1864–2021). Atmosphere 2022, 13, 1995. https://doi.org/10.3390/atmos13121995
10. In the Study Area Section, please add the area of Qiqihar City.
11. Figure 3 is not an original figure (DPSIR). I understand that it has been adapted for the context of drought, and this should be clearly stated in the figure caption, for example, by adding “adapted from…”. Furthermore, the reference for the DPSIR framework is somewhat invalid or incorrect (given that this framework is an important part of the methodology). It would be more appropriate to cite a reference closer to the creation of the framework, such as Smeets, E., & Weterings, R. (1999). Environmental indicators: Typology and overview. The DPSIR framework was developed by the European Environment Agency (EEA) in the late 1990s, and this should be reflected in the citation.
12. In Equation 18, is the notation at the top of the X_i correct?
13. Please define MSE.
14. In Section 4.3, there is a lack of explanation regarding the figures. Since colours and sizes are used in the plot (Figure 7), please make sure to utilise these elements to better explain the results.

In my opinion, the paper should undergo a major revision, not due to a lack of knowledge, but rather to improve the balance of the content. Some parts of the methodology are too lengthy, particularly when the entire formula is explained, whereas a single reference could suffice. Additionally, Figure 3 should be redone: there is no need for arrows, but the letters should be placed in the correct positions, and the outliers should be clearly explained in the figure caption. Please ensure that the diagram presented in the introduction is better placed in the methodology section. The current structure of the methodology is difficult to follow.

Furthermore, my main concern is that the results section lacks sufficient explanation. Although the results seem interesting, and it is clear that the authors spent considerable time preparing the figures, the section would benefit from more detailed explanations to achieve a better balance throughout the paper.

That said, I believe the work has strong potential, and with these revisions, the paper could be greatly enhanced. Keep up the good work, and I’m confident the final version will be even more impactful!

Author Response

Thank you for your comments on our manuscript. Those comments were all highly valuable and helpful for revising and improving our manuscript, and they were an important guide to our research. We have revised our manuscript, and we assure that your concerns have been addressed properly. The corrections in the paper are marked in red, and our responses are listed below.

Comments1: While the manuscript offers a valuable contribution to the assessment of agricultural drought, there are some acronyms introduced in the abstract that are not defined upon first use, which may make it difficult for readers (who are not familiar with these terms) to understand the first approach to the manuscript. For instance, DPSIR should be expanded as Driving Forces-Pressures-State-Impact-Response (assuming that this is what it means) when first mentioned. Similarly, KOA in “KOA Ridge Regression” requires clarification, as it does not correspond to a widely recognised method and may need to be defined as a bespoke or domain-specific approach. Additionally, RMProp-CNN appears to be either a variant or a misspelling of RMSProp-CNN, and should be corrected or explicitly defined if it is indeed distinct. Including these definitions in the abstract will improve accessibility and ensure greater transparency of the methodology. It is also recommended that brief descriptions be provided in the methods section to reinforce understanding.

Response1: Dear reviewer, thank you for your valuable comments. We have revised, please refer to the revised manuscript.

Comments2: In the Abstract section please also include the area of Qiqihar City, it could be the exact area or an approximate value.

Response2: Dear reviewer, thank you for your valuable comments. We have added the area of Qiqihar City in the Abstract section.

Comments3: The keywords listed should appear in the abstract using the exact phrasing presented. For instance, “Adam weight decay optimizer” is not mentioned in the abstract in that exact form, and “Drought resilience evaluation” could be more explicitly reflected (there are some instances where “drought resilience” appears in the abstract but the key word “drought resilience evaluation” never appears in the abstract, please correct me if I am wrong. Please ensure consistency between the keywords and the terms used in the abstract.

Response3: Dear reviewer, thank you for your valuable comments. We have revised, please refer to the revised Abstract and Keywords section.

Comments4: In the first line of the Introduction, drought by nature is already a negative phenomenon, so saying it has a "negative impact" might feel redundant. Please correct the opening phrase. The phrase could say something, for instance: In recent years, drought has posed an increasingly serious threat to regional sustainable development and human society.

Response4: Dear reviewer, thank you for your valuable comments. We have revised, please refer to the revised Introduction section.

Comments5: The authors mention the Palmer Drought Severity Index (PDSI) from 1965, but they do not address the Standardised Precipitation Index (SPI), introduced in 1993, or the Standardised Precipitation Evapotranspiration Index (SPEI), introduced in 2009. Please provide a brief discussion of these indices and how they relate to the assessment of drought.

Response5: Dear reviewer, thank you for your valuable comments. We have supplemented the introduction of SPI and SPEI indices and briefly discussed the application of these indices in drought assessment as follows: In 1993, McKee T B et al. proposed the Standardized Precipitation Index (SPI), which has attracted widespread attention due to its advantages such as convenient calculation and flexible time measurement. However, it ignores factors such as evapotranspiration, which makes it have certain limitations in drought monitoring. In 2010, Vicente-Serrano S M et al. considered the evapotranspiration factor based on the SPI index and proposed the Standardized Precipitation Evapotranspiration Index (SPEI). This index can effectively reflect the water loss on a long time scale and has strong applicability in drought monitoring in different regions.

Comments6: A major issue with the paper is the lack of a clear definition of agricultural drought. It is important to ensure that readers fully understand what is meant by agricultural drought in the context of this study. This could be introduced in the first paragraph for clarity. For instance, for severe agricultural drought, SPEI values below -1.5 are typically associated with significant soil moisture deficits, leading to substantial agricultural impacts, such as reduced crop yields or crop failure. However, if a different definition of agricultural drought is intended, please clarify and provide a clear explanation in the paper. Kindly address this matter as thoroughly as possible.

Response6: Dear reviewer, thank you for your valuable comments. We have revised, please refer to the revised Introduction section.

Comments7: Line 92: Please defined DPSIR.

Response7: Dear reviewer, thank you for your valuable comments. We have given the definition of the DPSIR model as follows: The Driver-Pressure-State-Impact-Response (DPSIR) model is a conceptual framework for analyzing the interaction between environmental and social systems. It is widely used in sustainable development, ecological management and policy evaluation research. The model can reveal the dynamic relationship between human activities and environmental changes through causal chains.

Comments8: Lines 151-158: The way the objectives are currently written seems more like a description of the methodology rather than clearly defined research objectives. Research objectives should typically be action-oriented, outlining the intended outcomes or goals of the study. So far they look like a description only.

Response8: Dear reviewer, thank you for your valuable comments. We have revised as follows: (1) Construct a regional agricultural drought resilience evaluation index system based on the DPSIR conceptual model. (2) Accurately evaluate the regional agricultural drought resilience characteristics using the AdamW-CNN model. (3) Identify the key driving factors of regional agricultural drought resilience, es-tablish the KOA-Ridge regression equation, and analyze the future evolution trend of resilience. (4) Verify the performance of AdamW-CNN and KOA-Ridge models.

Comments9: Please add the following reference to the manuscript (specifically when approaching some previous drought analyses): Espinosa, L.A.; Portela, M.M.; Matos, J.P.; Gharbia, S. Climate Change Trends in a European Coastal Metropolitan Area: Rainfall, Temperature, and Extreme Events (1864–2021). Atmosphere 2022, 13, 1995. https://doi.org/10.3390/atmos13121995

Response9: Dear reviewer, we have added this reference.

Comments10: In the Study Area Section, please add the area of Qiqihar City.

Response10: Dear reviewer, we have added the area of Qiqihar City, please refer to the revised Study Area section.

Comments11: Figure 3 is not an original figure (DPSIR). I understand that it has been adapted for the context of drought, and this should be clearly stated in the figure caption, for example, by adding “adapted from…”. Furthermore, the reference for the DPSIR framework is somewhat invalid or incorrect (given that this framework is an important part of the methodology). It would be more appropriate to cite a reference closer to the creation of the framework, such as Smeets, E., & Weterings, R. (1999). Environmental indicators: Typology and overview. The DPSIR framework was developed by the European Environment Agency (EEA) in the late 1990s, and this should be reflected in the citation.

Response11: Dear reviewer, thank you for your valuable comments. We have revised, please refer to the revised manuscript.

Comments12:  In Equation 18, is the notation at the top of the Xi correct?

Response12: Dear reviewer, the system encountered an error when converting the original manuscript to PDF. We have checked the original Word version and there is no problem. If you still see similar problems in the revised manuscript, you can check the Word version of the revised manuscript.

Comments13:  Please define MSE.

Response13: Dear reviewer, the definition of MSE has been given in the revised manuscript.

Comments14: In Section 4.3, there is a lack of explanation regarding the figures. Since colours and sizes are used in the plot (Figure 7), please make sure to utilise these elements to better explain the results.

Response14: Dear reviewer, thank you for your valuable comments. We have revised as follows: The blue color in Fig. 7. represents negative correlation, the red color represents positive correlation, and the size of the circle represents the strength of the correlation. Fig. 7 shows that there are both positive and negative correlations between the selected indicators, but the overall Kendall Tau-b correlation coefficients are small, ranging from -0.6 to 0.6, indicating that the collinearity between the data is low, which can ensure the high accuracy of the model operation.

Comments15: In my opinion, the paper should undergo a major revision, not due to a lack of knowledge, but rather to improve the balance of the content. Some parts of the methodology are too lengthy, particularly when the entire formula is explained, whereas a single reference could suffice. Additionally, Figure 3 should be redone: there is no need for arrows, but the letters should be placed in the correct positions, and the outliers should be clearly explained in the figure caption. Please ensure that the diagram presented in the introduction is better placed in the methodology section. The current structure of the methodology is difficult to follow.

Furthermore, my main concern is that the results section lacks sufficient explanation. Although the results seem interesting, and it is clear that the authors spent considerable time preparing the figures, the section would benefit from more detailed explanations to achieve a better balance throughout the paper.

Response15: Dear reviewer, thank you for your valuable comments. Considering that this article focuses on methodological research, the methods involved in the manuscript have specific application scenarios in the following text. In order to facilitate readers' understanding, this part of the content retains its integrity. However, in order to improve the balance of the content, Sections 4.5 and 4.6 were supplemented with a lot of in-depth analysis of the results, which also provided a more comprehensive explanation of the results section. The Figure 3 you mentioned should refer to Figure 4 in the manuscript. We have revised it to address the relevant issues you raised and also provided a clear explanation for the outliers. The figure in the introduction section is a general flowchart of the study based on the suggestions of the academic editor, mainly to allow readers to have a clearer understanding of the overall structure of the article. In addition, in the methodology section, the application flowchart of the AdamW-CNN model and KOA-Ridge regression model is drawn, which will make it easier for readers to understand.

We have revised the manuscript under your guidance, and we hope it will be met with approval.

Best regards,

Authors

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

After careful examination of the revised manuscript, I am pleased to report that you have thoroughly addressed my previous comments and implemented substantial improvements. The revisions have significantly enhanced the quality and clarity of their work. your research makes a valuable contribution to the field of sustainable agricultural science, particularly in the areas of drought resilience.

Thank you for considering my assessment.

Reviewer 4 Report

Comments and Suggestions for Authors

Dear Authors and Assistant Editor,

Thank you for the thorough revision of the manuscript. The authors have successfully addressed all of my previous comments, including clarifications of acronyms, improvements in the abstract and keywords, corrections to citations, and enhancements in the explanation of the methodology and results. The revised version now presents a clearer, more balanced, and accessible study, with strengthened methodological transparency and structure. I am satisfied with the changes made, and from my perspective, I consider the manuscript suitable for publication, provided there are no further concerns from the other reviewers.

Best regards,

Reviewer