Expansion of Rural Settlements on High-Quality Arable Land in Tongzhou District in Beijing, China
Round 1
Reviewer 1 Report
The paper presents interesting methods and findings by identifying rural settlements and through classification methods and demonstrates their historical impacts through expansion. However, I find the explanation of their methods in general very lacking. In addition, the writing has a very narrow case study focus and fails to deliver the important scientific impacts of their research. I suggest the authors to significantly improve their introduction and discussion while clarify their methodology.
Line 12: The logic between China registration system and rural settlement decreasing is very difficult to understand for international readers and the authors fail to provide adequate and clear explanations in the abstract or the text. I suggest removal of this sentence as it is not directly linked to the key questions of this paper.
Line 41: The last sentence of the paragraph is of great importance, but the authors fail to explain them adequately or engage previous worldwide literature. I suggest making a strong review and discussion on food safety/rural development topic (with international literature) before discussing the China case.
Line 62: For the introduction, there is a lack of comprehensive discussions on how urban sprawl, sub-center development, and rural settlement are occupying arable lands. The other 2 factors besides rural settlements are also very important factors of arable land losses. There needs to a review and a clarification of their differences in mechanisms and effects, for example:
Urban sprawl and arable land loss:
https://www.pnas.org/content/114/34/8939.short
Line 137: Data acquisition should be merged into data sources part. Also, using GIS for projection and transformations on satellite image does not need to be shown. It is more important to provide better explanations on the classification methods that identify rural settlements.
Line 164: Validation is very important and should consist of a sub-section in the method. The validation part also needs to discuss other available methods and justify the suitability the methods the authors choose. For example:
https://link.springer.com/article/10.1007/s12061-019-09296-5
https://www.tandfonline.com/doi/abs/10.1080/13658816.2017.1358814
Moreover, the validation results should be moved to results section.
Line 236: Numbers should use the same digits of decimals.
Discussions: There lacks a discussion on environmental impacts of urban sprawl and land use changes in comparison to previous international studies, for example:
https://onlinelibrary.wiley.com/doi/abs/10.1111/ele.12469
https://www.sciencedirect.com/science/article/pii/S1470160X19301682
Line 321: This part should not only include Chinese literature and neglect international literatures on similar topics.
Conclusions: The second paragraph is well-written but the first paragraph should be replaced by the broader impacts of your methods/findings in the broader scientific field. Repeating Tongzhou results again does not spark any further interests from readers.
Author Response
Point 1: The paper presents interesting methods and findings by identifying rural settlements and through classification methods and demonstrates their historical impacts through expansion. However, I find the explanation of their methods in general very lacking. In addition, the writing has a very narrow case study focus and fails to deliver the important scientific impacts of their research. I suggest the authors to significantly improve their introduction and discussion while clarify their methodology.
Response 1: Thank you for your suggestions. According to your comments, we have substantially revised our introduction section, research method section and discussion section. In brief, we added some new clarifications about the methods in the last paragraph of the Introduction section, made substantial revisions in the research methods section including the many changes in the subsection of land use type division and the subsection of the validation, and added some discussions about the research method in the second last paragraph of the Discussion section.
Point 2: Line 12: The logic between China registration system and rural settlement decreasing is very difficult to understand for international readers and the authors fail to provide adequate and clear explanations in the abstract or the text. I suggest removal of this sentence as it is not directly linked to the key questions of this paper.
Response 2: Thank you for your suggestion. In the revised version, we have removed the sentence “such as the household registration system”. In addition, we changed the sentence to “Due to some factors such as institutional problems, the total number of rural settlements in China is decreasing, while their total area is continuously increasing.”
Point 3: Line 41: The last sentence of the paragraph is of great importance, but the authors fail to explain them adequately or engage previous worldwide literature. I suggest making a strong review and discussion on food safety/rural development topic (with international literature) before discussing the China case.
Response 3: Thank you for your constructive suggestion. According to your comments, we have discussed the reasons why the expansion of rural settlements with the consumption of high-quality arable land will affect food security in the revised version. The revisions are in the first paragraph of the introduction section of the revised version. In brief, due to the huge rural baseline population, the rural settlement area in China is enormous and is in fact larger than the urban areas. Furthermore, due to some institutional factors, the areas of rural settlement in China keep increasing. The expansion of rural settlements consumes a large amount of precious cultivated land including much high-quality cultivated land. Thus, rural expansion poses a big threat on food security.
Point 4: Line 62: For the Introduction, there is a lack of comprehensive discussions on how urban sprawl, sub-center development, and rural settlement are occupying arable lands. The other 2 factors besides rural settlements are also very important factors of arable land losses. There needs to a review and a clarification of their differences in mechanisms and effects, for example: Urban sprawl and arable land loss: https://www.pnas.org/content/114/34/8939.short
Response 4: Thank you for your suggestion. We have added a review of relevant research on arable land encroachment worldwide. Prior to this paragraph, we have reviewed the different mechanisms for expansion of urban and rural settlements on cultivated land encroachment. These revisions are in the second paragraph of the Introduction section (revised version). In brief, we think that rural settlement expansion and urban expansion both caused a large arable land loss while the mechanisms are different. These differences are reflected in the occupation patterns, the causes for occupation and the degree of difficulty in controlling the occupation. The specific clarifications are shown in the second paragraph of the Introduction section.
Point 5: Line 137: Data acquisition should be merged into data sources part. Also, using GIS for projection and transformations on satellite image does not need to be shown. It is more important to provide better explanations on the classification methods that identify rural settlements.
Response 5: Thank you for your suggestion. In the revised version, we have integrated the contents of the “data source” and “data acquisition” in the original text, and to more clearly illustrate the data processing process and the classification process of the land class, the detailed contents of “data processing” are provided in the method part of this article. In addition, we have deleted the statement concerning the use of GIS for projection and transformation.
To provide better explanations on the classification methods that identify rural settlements, we have listed the sources of all the data used in this manuscript in the new Data Sources section (3.1) and illustrated the characteristics of these data sources in tabular format (Table 1 in the revised version). We have also clarified the method adopted in the land use classification for identifying rural settlements. The detailed revisions are in the first, second and third paragraphs of subsection 3.2.1 Land use classification. The detailed interpretation standards on rural settlements in 1972 and 1991 are given in Table 2 of the revised version.
Point 6: Line 164: Validation is very important and should consist of a sub-section in the method. The validation part also needs to discuss other available methods and justify the suitability the methods the authors choose. For example: https://link.springer.com/article/10.1007/s12061-019-09296-5; https://www.tandfonline.com/doi/abs/10.1080/13658816.2017.1358814; Moreover, the validation results should be moved to results section.
Response 6: Thank you for your suggestion. By referring to the references mentioned in your comments, we have added the contents of subsection “validation” in the revised version.
In brief, we used the confusion matrix method to verify the classification accuracy of the images. First, we superimposed the Google Earth images for 1991 and 2015, respectively, with the land parcel classification maps of the corresponding years and established 100 random points in the Tongzhou District, using the Create Random Points tool. Subsequently, the land parcel where a sampling point was located could be viewed as the sample to be analyzed, thereby determining the corresponding land types on the base map of Google Earth images. Then we used the Confusion Matrix Using Ground Truth ROIs tool in ENVI software to generate the confusion matrix to verify the accuracy of the remote sensing image classification.
The detailed revisions are in the new subsection 3.2.2 Validation. All the materials in the new subsection are new additions.
Point 7: Line 236: Numbers should use the same digits of decimals.
Response 7: We are sorry for the mistake. In the revised version, we have kept all the numbers in two decimal places. In addition, we replaced “76.8%” with “76.83%”.
Point 8: Discussions: There lacks a discussion on environmental impacts of urban sprawl and land use changes in comparison to previous international studies, for example:
https://onlinelibrary.wiley.com/doi/abs/10.1111/ele.12469;
https://www.sciencedirect.com/science/article/pii/S1470160X19301682
Response 8: Thank you for your suggestion. Based on your comments, we have reviewed related research on the impact of expansion of construction land such as the effects of expansions of cities and rural settlements on the environment. We have added these revisions in the Discussion section of the revised version. The detailed clarifications are in the second paragraph of the Discussion section.
Point 9: Line 321: This part should not only include Chinese literature and neglect international literatures on similar topics.
Response 9: Thank you for your suggestion. According to your comments, we have carefully reviewed the related (international) literature. Unfortunately, there is only very limited literature regarding the consumption features for rural settlement expansion on high-quality farmland. Nevertheless, many researchers report the urban expansion encroachment on high-quality farmland. Therefore, in the revised version, we have reviewed mainly the related literature on the influences of urban expansion on the consumption of high-quality farmland. The specific revisions are in the second paragraph of the Discussion section.
Point 10: Conclusions: The second paragraph is well-written but the first paragraph should be replaced by the broader impacts of your methods/findings in the broader scientific field. Repeating Tongzhou results again does not spark any further interests from readers.
Response 10: Thank you for your suggestion. According to your comments, we have made substantial changes in the first paragraph of conclusion section. In brief, we included the IHC coefficient to emphasize in this paper the consumption preference for high-quality cultivated land in rural settlements. Thus, the quality of cultivated land occupied by rural settlements can be evaluated, and the mechanism regarding the preference for rural settlements expansion in occupying high-quality cultivated land is quantitatively revealed by our approach. In addition, we also provided a reference for long term land use change analysis by interpreting the decrypted military satellite remote sensing images.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
This paper is a nice and simple piece of research that analyse the process of settlement expansion in rural areas of China. Although there are many aspects that authors should reconsider, the paper as a whole is good and comply with the objectives that authors propose. Although it is not anything new, innovative and does not provide any new approach of analysis, it serves to add new insight to the understanding of the studied process in a specific region of China.
Abstract
It must focus on what the authors address in the paper. Therefore, references to some aspects that are not addressed in the paper, as the household registration system, should be avoided.
Introduction
The message that you try to give the readers is quite simple: an important dynamic in the China landscapes and territory comes from the loss of agricultural land through rural settlements expansion. The impact of these losses will depend on the quality of the agricultural land these developments take, which is a topic that has not received attention in the research community. Therefore, you aim to research this for a specific region of China.
You must summarize the introduction to provide this message, avoiding excesive repetition of the same idea (as you do now) and also the continuous references to the urban expansion process. Although it is also of interest in China, it is out of the scope of this paper and should be just mentioned once, because of its similarities with the studied processes, but not many times, like in the current form of the paper. From what I have understood, there is not any relation between urban expansion and rural settlement expansion and they are two independent processes.
Data sources and research methods
Section 3.1
Lines 112-113 are confussing. It seems like you are using land use maps that are already available, but not land use maps that you have made from satellite imagery
I do not either understand which are the imagery sources you are using, especially regarding 2015 data. Please, state clearly from which satellite you obtained each year the imagery.
Maybe it would be useful to summarize the characteristics of all employed datasets in a table.
Section 3.2
The title is not correctly written. It would be data processing, but not data acquisition. The last one refers to how you get the raw data, but no to what processing you do.
In addition, you must provide more detail and explanation about the processing you have made, focusing on the important steps of this processing. What does image registration mean? What does splicing mean?
You should state that you make all that preprocessing on the images to enhance the images and make easier the visual interpretation process.
In the end, this section is about the visual interpretation process and the definition of the categories to identify visually. Please, make this clear since the very beggining of the section. You should provide more information about who (how many, professional background...) make the image interpretation.
Regarding the validation of the image, more information is needed about how you determine the land use types according to google earth imagery. Do you also make the classification by yourself? Because this would bring another source of uncertainty and you should account for it in your analysis.
Summarize paragraph 184-189. The method you use is very simple and well known by the research community. You can just mention it and give a bit extra information and it will be enough.
If you have come up with the indices of page 8, please state it. Say that it is your proposal of indices to assess the expansion of rural settlements.
Section 4
Reword the title to just results. The analysis has been explained in the previous section. You just explain here the results. You do not do any analysis.
Section 4.1
Could you provide in tables 2 and 3 the propotions that each conversion or permanence mean? You could provide it in parenthesis or well give the propotion and then the rar ammount of land in parenthesis.
How is it possible that rural settlements transform in arable land? Would this not be a likely mapping mistake?
You should try tu summarize section 4.1 a bit, trying to connect the ideas of paragraph one (crosstabulation matrix) with the ones of paragraph two (spatial analysis on the maps). This would be easier to read and undertand for the readers.
Section 4.2
Sections 4.2 and 4.3 should be merged in just one, summarizing all the information they contain, which is sometimes very repetitive. They go about the same topic.
Summarize the first paragrpah. It refers to information you have already mentioned and which is in addition in the tables you have analysed before.
Could you provide in figure 5 a series of maps focusing on the changes instead of the whole map? This would make the reading and undertanding of the paper easier for the readers. It is about focusing on what changes, instead of letting readers to do this task by visual map comparison.
Section 5
Sections 5.1, 5.2 and 5.3 must be merged in just one, the dicussion. The dicussion is about the explanation of the results, but connecting what you state with the literature on the topic. It does not make sense to split these two essential parts of the discussion in separate sections, as their connection is the basis on which the discussion makes sense.
Then you can connect the conclusions and insights you get from your conclusions to explain the policy suggestions that take part of section 5.3
Abuso de punto y coma
Specific comments
Lines Lines 33-34 > Avoid repetitive expressions: With a further decrease..., with a decrease...
Lines 35-36 > Not need to say "even". If there is a growth in the area occupied by rural settlements and there was a decrease of the number of rural settlements, it is obvious that this is because some settlemenets have expanded.
Line 37 > Because... because
Lines 37-41 > I do not properly understand this. Please, reword.
Line 95 > Study area. Delete "General situacion of the"
Line 96 > Geographical positions are already on the map, so no need to provide them in the text as well.
Lines 98-99 > Area is enough information. No need to say how wide and long in the study area in km
Lines 102-103 > Could you translate the yuans to dollars? It would be easier to understand for the reader
Figure 1 > It is difficult to undertand the letters of the administrative map in grey
In the legend , DEM should be replaced by elevation.
I also miss other interesting features in the map: roads, railways, urban areas... Please incluse these in the map.
Lines 138-139 > Data cannot be adquired by visual interpretation (see comments above)
Lines 156-160 are not relevant and should be removed
Line 162. Not need to mention this tool, whose mention does not provide any extra and useful information to the reader.
Maps of figure 4 are not easy to read. Please, provide colors for each polygon, instead of using lines to identify polygons without filling.
Lines 249-251 should be removed. They are not results.
Lines 286-287 do not say anything. Remove
Author Response
This paper is a nice and simple piece of research that analyse the process of settlement expansion in rural areas of China. Although there are many aspects that authors should reconsider, the paper as a whole is good and comply with the objectives that authors propose. Although it is not anything new, innovative and does not provide any new approach of analysis, it serves to add new insight to the understanding of the studied process in a specific region of China.
Point 1: Abstract：It must focus on what the authors address in the paper. Therefore, references to some aspects that are not addressed in the paper, as the household registration system, should be avoided.
Response 1: Thank you for your suggestion. In the revised version, we removed the sentence “such as the household registration system”. In addition, we revised the sentence to “In China, due to some factors, the total number of rural settlements is decreasing, while their total area is continuously increasing.”
Point 2: Introduction：The message that you try to give the readers is quite simple: an important dynamic in the China landscapes and territory comes from the loss of agricultural land through rural settlements expansion. The impact of these losses will depend on the quality of the agricultural land these developments take, which is a topic that has not received attention in the research community. Therefore, you aim to research this for a specific region of China.
You must summarize the introduction to provide this message, avoiding excesive repetition of the same idea (as you do now) and also the continuous references to the urban expansion process. Although it is also of interest in China, it is out of the scope of this paper and should be just mentioned once, because of its similarities with the studied processes, but not many times, like in the current form of the paper. From what I have understood, there is not any relation between urban expansion and rural settlement expansion and they are two independent processes.
Response 2: We are sorry that the language in the original text is not refined enough, and the expression of the same meaning has been expressed many times.
In the revised version, we reorganized the language and deleted the repetitive expressions, striving to make our viewpoints clearer and more refined.
The inclusion of “urbanization” in the original text is intended to explain that areas with rapid urbanization development involve transfers of both the urban and rural populations, hence resulting in changes to the urban and rural populations, and changes in the layout of rural settlements. To improve clarity, we have reorganized the text, reduced the content of the urban expansion section, and focused on the expansion of rural settlements.
Point 3: Data sources and research methods
Section 3.1
Lines 112-113 are confussing. It seems like you are using land use maps that are already available, but not land use maps that you have made from satellite imagery.
I do not either understand which are the imagery sources you are using, especially regarding 2015 data. Please, state clearly from which satellite you obtained each year the imagery.
Maybe it would be useful to summarize the characteristics of all employed datasets in a table.
Response 3: We are sorry for the lack of clarity and confusion.
In the revised version, revisions have been made to improve clarity. In fact, we used three-phases of land use data in 1972, 1991 and 2015. The land use data of 1972 was obtained by interpreting KeyHole (KH) satellite imagery in-house. Also, the land use data for 1991 was obtained by interpreting the image of the Russian Land Resource Satellite System (RESURS F1) in-house. The land use data for 2015 was adopted from the Resource and Environmental Science Information Center of the Chinese Academy of Sciences. The land use data in 2015 was generated by human visual interpretation based on Landsat 8 remote sensing images, which has a high interpretation accuracy and meets our requirements.
In the revised draft, we have reorganized the text which has been checked by a native speaker. In addition, we added a new table (Table 1) to clarify the data used. The data acquisition methods were also described in detail. The characteristics of each data source were also introduced. The specific revisions are in the new subsection 3.1 Data sources and the new subsection 3.2 Research methods.
Point 4: Section 3.2
The title is not correctly written. It would be data processing, but not data acquisition. The last one refers to how you get the raw data, but no to what processing you do.
In addition, you must provide more detail and explanation about the processing you have made, focusing on the important steps of this processing. What does image registration mean? What does splicing mean?
You should state that you make all that preprocessing on the images to enhance the images and make easier the visual interpretation process.
In the end, this section is about the visual interpretation process and the definition of the categories to identify visually. Please, make this clear since the very beggining of the section. You should provide more information about who (how many, professional background...) make the image interpretation.
Response 4: Thanks for your suggestion. In the revised version, we have replaced “data acquisition” with “data processing”. In addition, we have also made substantial revisions in the section. First, we explained the data processing process in detail. Second, we explained in detail the classification method of land use types and produced a table whereby the image recognition features of different terrestrial types are described and illustrated. The specific revisions are in the new subsection 3.2.1 Land use classification. These revisions include details and explanation on the processing, particularly the important steps in processing. We also mentioned that after pre-processing and image enhancement of the KH and RESURS F1 images, the spatial data of land use and rural settlements in the Tongzhou District were obtained through visual interpretation. The definition of the land use categories has also been clarified.
In addition, information about image interpretations has been added. In brief, the interpretation of the images was performed by the first author. Prior to this work, the first author has undertaken image interpretation in other projects, and has specific professional skills and image processing experience. All of the work, including image pre-processing, classification and identification of different land-use types and the acquisition of the spatial data for each land-use type, took about 8 months.
Point 5: Regarding the validation of the image, more information is needed about how you determine the land use types according to google earth imagery. Do you also make the classification by yourself? Because this would bring another source of uncertainty and you should account for it in your analysis.
Response 5: Thanks for your comment. We have added new materials about how we determine the land use types according to Google Earth imagery. We also explained the method of the land use classification in the revised version.
In addition, we have made a table to clearly explain and illustrate the interpretation marks of different land types of Google Earth images (please see Table 3 in the revised version). The classification of land use types is consistent with the classification of land classes in 2015. The verification method was performed using the confusion matrix in ENVI software.
The specific clarifications about the land classification method are in the second paragraphs in the subsection 3.2.1 Land use classification. The verification part of Google Earth image is described in detail in 3.2.2 (revised version). The accuracy assessment results are in the new 4.1 subsection Verification of image classification. In brief, we established 100 random points for the Tongzhou District, using the Create Random Points tool to verify the accuracy. Given the lack of field verification data, we used Google Earth images for 1991 and 2015 which have a very high resolution for verification. Finally, we found that the land classification accuracy rate in 1991 was 88%. Among the 100 randomly classified sample points, 12 points corresponded to the incorrect classification of the plots, and 88 plots corresponded to the correct classification of the plots (Figure 2 of the revised version). The accuracy rate of land classification in 2015 was 90%. Among the 100 randomly classified sample points, 10 points corresponded to the incorrect classification of the plots, and the plots corresponding to the other 90 points were classified correctly.
Point 6: Summarize paragraph 184-189. The method you use is very simple and well known by the research community. You can just mention it and give a bit extra information and it will be enough.
If you have come up with the indices of page 8, please state it. Say that it is your proposal of indices to assess the expansion of rural settlements.
Response 6: Thank you for your suggestion. Based on your suggestion, we have simplified the first paragraph and deleted the excessive description of the land use transfer matrix. In addition, we added a description of the source of the IHC coefficient.
Point 7: Section 4
Reword the title to just results. The analysis has been explained in the previous section. You just explain here the results. You do not do any analysis.
Response 7: Thank you for your suggestion. We have deleted “and analyses” and revised it to “Results” in the revised version.
Point 8: Section 4.1
Could you provide in tables 2 and 3 the propotions that each conversion or permanence mean? You could provide it in parenthesis or well give the propotion and then the rar ammount of land in parenthesis.
Response 8: Thank you for your suggestion. We have provided the proportions of each conversion in parenthesis after the conversion amount of each land use type in the original Tables 2 and 3. This proportion refers to the percentage of the total area of each type of land converted from other types of land. The results are shown in Table 4 and Table 5 (revised version).
Point 9: How is it possible that rural settlements transform in arable land? Would this not be a likely mapping mistake?
Response 9: We are sorry for not clearly clarifying this in the original version. In China, under certain conditions, rural settlements can be converted into arable land.
To alleviate the pressure on the reduction of cultivated land, the Chinese government try to increase the area of cultivated land through land consolidation and reclamation. Land consolidation refers to a comprehensive improvement of fields, water, roads, forests and villages through various measures to improve the quality of cultivated land, increase the effective area of cultivated land, and improve the agricultural ecological conditions and ecological environment. Land reclamation is a measure of remediation and restoration of land that has been damaged or abandoned due to damage caused by excavation, collapse, pressure, and abandoned land and natural disasters during production and construction. One of the important aspects of rural settlement consolidation is to replenish the cultivated land through the reclamation of the old house base of rural settlements.
In China, the large-scale rural population has led to a large scale of rural settlement land occupation, and the transfer of the rural population has led to derelict rural residential areas. Therefore, the derelict rural settlements are reclaimed and transformed into cultivated land through land consolidation to increase the effective arable land area. On the other hand, some old rural settlements have few ground reinforcement and strengthening measures in place excepted in the case of houses. Thus it is possible to transform many old rural settlement areas to cultivated land.
Point 10: You should try tu summarize section 4.1 a bit, trying to connect the ideas of paragraph one (crosstabulation matrix) with the ones of paragraph two (spatial analysis on the maps). This would be easier to read and undertand for the readers.
Response 10: Thank you for your suggestion. We have combined and summarized the two paragraphs in Section 4.1 of the manuscript. We first explained the amount of change in rural settlements obtained from the land use transfer matrix, and then explained the spatial distribution of this change. Based on this idea, we have divided this part into three paragraphs, indicating the changes in the quantity and space of rural settlements in 1972-1991, 1991-2015 and the entire study period from 1972-2015. The revisions are in the first, second and third paragraphs in the section 4.2 Evolution of rural settlement expansion.
Point 11: Sections 4.2 and 4.3 should be merged in just one, summarizing all the information they contain, which is sometimes very repetitive. They go about the same topic.
Summarize the first paragrpah. It refers to information you have already mentioned and which is in addition in the tables you have analysed before.
Response 11: According to your comments, we have merged sections 4.2 and 4.3, and removed the repetitive information. In the first paragraph, we have summarized the basic situation of the cultivated land in the Tongzhou District from 1972 to 2015 and its transformation direction. In the second paragraph, we have explained the changes in the quantity of different qualities of cultivated land consumed by the expansion of rural settlements in the Tongzhou District. In the third paragraph, we explain the spatial distribution of different qualities of cultivated land occupied by rural settlements between 1972 and 2015. The specific revisions are in the first, second and third paragraphs of the new subsection 4.3 Variation of arable land distribution.
Point 13: Could you provide in figure 5 a series of maps focusing on the changes instead of the whole map? This would make the reading and undertanding of the paper easier for the readers. It is about focusing on what changes, instead of letting readers to do this task by visual map comparison.
Response 13: Thank you for your suggestion. To better represent the changes between the two periods, we have replaced the current map for different years with the map of arable land reduction between the two periods. The revised figure is Figure 5 (revised version).
Point 13: Section 5
Sections 5.1, 5.2 and 5.3 must be merged in just one, the dicussion. The dicussion is about the explanation of the results, but connecting what you state with the literature on the topic. It does not make sense to split these two essential parts of the discussion in separate sections, as their connection is the basis on which the discussion makes sense.
Then you can connect the conclusions and insights you get from your conclusions to explain the policy suggestions that take part of section 5.3
Response 13: Thank you for your suggestion. According to your suggestion, in the revised version, we have reorganized the text and merged the contents of 5.1 and 5.2 in the original into a single section which discusses the findings of this study and also makes comparisons with key literature.
In brief we divided the Discussion into two main paragraphs. In the first paragraph, we compared the conclusions drawn in this paper with the research results of other scholars. In the second paragraph, we focused mainly on the actual situation for the Tongzhou District, and discussed the reasons for the preferential expansion of rural settlements whereby high-quality cultivated land in the two different time periods examined in this paper are occupied. After that, we discussed the research methods used in the study and their significance. Finally, based on the results of this paper, we proposed policy recommendations. The specific revisions are in the first, second, third and fourth paragraphs in the Discussion section.
Specific comments
Point 14: Lines Lines 33-34 > Avoid repetitive expressions: With a further decrease..., with a decrease...
Response 14: Thank you for your suggestion. We revised the text as follows:
In recent decades, the number of rural settlements in China declined significantly from nearly 3.77 million in 1990 to 2.73 million in 2010, with a further decrease to 2.65 million in 2015 [10], the number of rural settlements in China having decreased by 42.5% in the past 20 years.
Point 15: Lines 35-36 > Not need to say "even". If there is a growth in the area occupied by rural settlements and there was a decrease of the number of rural settlements, it is obvious that this is because some settlemenets have expanded.
Response 15: Thank you for your suggestion. We have deleted the word “even”.
Point 16: Line 37 > Because... because
Response 16: We are sorry for the inappropriate description. We have revised this sentence to: During the expansion of rural settlements, arable land is generally occupied because of its fertile soil and beneficial geographic location [15-17]. Given that the production capacity of arable land is reflected in the quantity and quality of arable land and the yield per unit area, when more high-quality arable land is consumed, the greater is the threat to the production capacity of the arable land and the more severe the food security problem will be.
Point 17: Lines 37-41 > I do not properly understand this. Please, reword.
Response 17: We are sorry for the inappropriate description. We have re-phrased this sentence as follows: Given that the production capacity of arable land can be reflected in the quantity of arable land and the yield per unit area (namely the quality of arable land), when the same amount of arable land is consumed, the more high-quality arable land is consumed; hence the greater the threat to the production capacity of the arable land, and the more severe the food security problem will be.
Point 18: Line 95 > Study area. Delete "General situacion of the"
Response 18: Thank you for your suggestion. We have deleted “General situation of the” and revised it to “Study area” in the revised version.
Point 19: Line 96 > Geographical positions are already on the map, so no need to provide them in the text as well.
Response 19: Thank you for your suggestion. We have deleted “(north latitude of 39°36'~40°02', east longitude of 116°32'~116°56')” in the revised version.
Point 20: Lines 98-99 > Area is enough information. No need to say how wide and long in the study area in km
Response 20: Thank you for your suggestion. We have deleted “with a width of 36.5 km from east to west and a length of 48 km from north to south.” in the revised version.
Point 21: Lines 102-103 > Could you translate the yuans to dollars? It would be easier to understand for the reader
Response 21: Thank you for your suggestion. We have replaced “59.54 billion yuan” with “8.62 billion $” and replaced “21,648 yuan” with “3,134 $” in the revised version.
Point 22: Figure 1 > It is difficult to undertand the letters of the administrative map in grey
In the legend , DEM should be replaced by elevation.
I also miss other interesting features in the map: roads, railways, urban areas... Please incluse these in the map.
Response 22: Thank you for the suggestions. The administrative map in grey represents the district levels in Beijing. In the revised version, we have capitalized the initial names for each district, we have replaced “DEM” with “Elevation”, and added roads, railways and urban areas in the map. See Figure 1 in the revised version.
Point 23: Lines 138-139 > Data cannot be adquired by visual interpretation (see comments above)
Response 23: Thank you for your suggestion. We have revised the sentence in the revised version. This question has been addressed in Point 4.
Point 24: Lines 156-160 are not relevant and should be removed
Response 24: Thank you for your suggestion. We have removed these sentences in the revised version.
Point 25: Line 162. Not need to mention this tool, whose mention does not provide any extra and useful information to the reader.
Response 25: Thank you for your suggestion. We have removed “In ArcGIS” in the revised version.
Point 26: Maps of figure 4 are not easy to read. Please, provide colors for each polygon, instead of using lines to identify polygons without filling.
Response 26: Thank you for your suggestion. We have revised Figure 4 in the revised version (See Figure 4 of the revised version).
Point 27: Lines 249-251 should be removed. They are not results.
Response 27: Thank you for your suggestion. We have removed this sentence in the revised version.
Point 28: Lines 286-287 do not say anything. Remove
Response 28: Thank you for your suggestion. We have removed this sentence in the revised version.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The revisions addressed reviewer comments. I suggest the authors to go through the reference list again for better formatting as there are some loose ends existing.
Author Response
Point 1: The revisions addressed reviewer comments. I suggest the authors to go through the reference list again for better formatting as there are some loose ends existing.
Response 1: We are sorry for the mistakes in the reference list. According to your comments, we have double checked the references and corrected the mistakes of them.
Author Response File: Author Response.pdf
Reviewer 2 Report
Authors have improved the initial draft of the paper and I really thank them for keeping in considerations most of the comments I have made. However, the paper is still far to accomplish the requirements of a lead and high impact international journals.
Authors must deeply revise the way the paper is written, which makes confussing the reading a lot of times, especially regarding the introduction.
Introduction
This section still needs to be deeply summarized, linking all ideas you are addressing. You go once and other time arround similar points along the whole introduction, so the reader feels usually lost between similar information that is not really connected. Authors must clearly structure this section and make clear to the reader which is the problem to be addressed, its extent and importance and the objectives of the paper.
I find very confusing all ideas and points addressed in lines 55-63. I do not understand what you are trying to say or the process you are trying to explain. Authors must reword this paragraph. In addition, the new paragraph that has been added (46-63) it is not clearly connected with the rest of the text, especially with the next paragraph, which goes again about rural settlement expansion globally.
Data sources and research methods
In Table 1, when you are talking about data sources, the column about data acquisition method seems confusing. If the data sources are just satellite imagery (like in 1972 and 1991) that then you classificate through the methods that you describe in the next section, the data acquisition method is the method to obtain the satellite imagery, but not the one by means you obtain the land use classificated map and that you explain in section 3.2.
Lines 154-156 must be relocated at the beggining of section 3.2, when you talk about preprocessing and image enhancement processing.
Regarding section 3.2.1 you must also provide information about who makes the interpretation: is it just one person or are them several people? In that case, is there a coordinator or how do you assure the consistence between the work made by these different people? You have given me some of this information on your response to my comments, but I do not see it in the text.
In the validation section, you must specify the characteristics of the Google Earth imagery you use as reference. You just say high resolution, but what this means respect to the resolution of the dataset you are exploying? Also I find really important the missing validation of 1971 images. I know this may be not possible due to the lack of reference data, but could you come with something that at least could give the reader an idea of the quality of that interpretation?
I also agree with the other reviewer in the need of giving more importance to this section and discussing about different validation approaches available and the pros and cons of the one you have selected.
In addition, when you do change analysis from different land use maps, you must not only focus on the accuracy of the datasets as a whole, but on the accuracy of the changes they measure. Generally, the dataset may have an accuracy of 80%, but if the changes only represent 5% of the area, this means that all measured changes may be wrong. Therefore, I recommend the authors to validate and visually compare with reference data the changes measured by their series of maps. Especially because they will come with high sources of uncertainty: you are comparing maps that come from very different sources and which are made through different procedures.
Finally, you must explain clearly in section 3.2.3 how you carry out the analysis that you show in section 4.3. You explain the obtation of quantitative data in tables, but do not say anything about the spatial analysis that you make by means of overlying maps.
Results
Figure note of 202 and 206 must be avoided and the information integrated in the figure itself. The first point is already addressed by the legend of the figure. In the other cases, you may state which land use is each land parcel by putting a text below each image.
Same applies to note of figure 3.
In addition, section 4.1 lacks of a lot of information about quality and accuracy of dataset. Given you have been working with them, is it not possible to inform the reader qualitatively about the accuracy of the dataset? You may also bring here issues related to accuracy and uncertainty of measured changes, the uncertainties related to the validation process you have followed, etc.
Table 4 > to improve the understanding of the table, please include a sum up of the percentage also in the last row, so the reader can understand that the percentages refer to the proportions of changes for each class in the later map.
Figure 4 would be easier to read if you just provide the footprint of land uses, especially rural settements, for the three different years in the three different maps (one map with uses for 1972, another one for 1991 and a last one for 2015). In the current form, it is quite confussing to interpret the evolution of rural settlement expansion through the years.
Reword the caption of figure 5 as in the current form the reader cannot underastand what it represent. Is it about losses?
Discussion
The disucssion is now improved, but authors have included a few new paragrpahs that do not directly connect whith what is said in the paper. Plase, in the discussion, discuss the results, but not talk in general about land use changes and setllement expansion in China according to a literature review, because you could even do this without carrying out any analysis. Therefore, authors must connect all these ideas with the results of their analysis.
I miss proper discussion about uncertainty of the performed analysis. Authors are not critic about the methods and data they have employed, which come with many sources of uncertainty that could have deeply influenced the results thwy have got.
Conclusions
Please, focus on what ideas the paper provides, but not make again a summary of the results of your analysis (first paragraph)
Author Response
Authors have improved the initial draft of the paper and I really thank them for keeping in considerations most of the comments I have made. However, the paper is still far to accomplish the requirements of a lead and high impact international journals.
Authors must deeply revise the way the paper is written, which makes confusing the reading a lot of times, especially regarding the introduction.
Point 1: Introduction: This section still needs to be deeply summarized, linking all ideas you are addressing. You go once and other time arround similar points along the whole introduction, so the reader feels usually lost between similar information that is not really connected. Authors must clearly structure this section and make clear to the reader which is the problem to be addressed, its extent and importance and the objectives of the paper.
Response 1: We are sorry for the confusing expressions in the introduction section. According to your comments, we have made a major revision on introduction section. One paragraph that was not closely related to the main ideas we would like to address was dropped. In the other paragraphs, we reorganized them and deleted a lot of materials that was not closely related to our research. After revision, we think the new introduction section are clearer now.
Point 2: I find very confusing all ideas and points addressed in lines 55-63. I do not understand what you are trying to say or the process you are trying to explain. Authors must reword this paragraph. In addition, the new paragraph that has been added (46-63) it is not clearly connected with the rest of the text, especially with the next paragraph, which goes again about rural settlement expansion globally.
Response 2: According to the comments of point 1 and point 2, we almost rewrote the introduction section. After revision, most of the materials from line 55 to 63 in the former version were dropped due to the low correlations with our main ideas of conducting this research. After revision, we think the introduction section is clearer now.
Point 3: Data sources and research methods：In Table 1, when you are talking about data sources, the column about data acquisition method seems confusing. If the data sources are just satellite imagery (like in 1972 and 1991) that then you classificate through the methods that you describe in the next section, the data acquisition method is the method to obtain the satellite imagery, but not the one by means you obtain the land use classificated map and that you explain in section 3.2.
Response 3: We are sorry for these mistakes and confusions. According to your comments, we have revised Table 1. In the revised version, we changed the “Data acquisition method” to “Data sources” and the “Data source introduction” to “Brief descriptions”. In addition, the website of acquiring the sataellite images and land use map in 2015 were also added in the table to better present the data sources.
Point 3: Lines 154-156 must be relocated at the beggining of section 3.2, when you talk about preprocessing and image enhancement processing.
Response 3: Thanks for your suggestion. We have moved this sentence to the beginning of section 3.2.1. In addition, we also deleted some materials in the first paragraph that could duplicate with this sentence.
Point 4: Regarding section 3.2.1 you must also provide information about who makes the interpretation: is it just one person or are them several people? In that case, is there a coordinator or how do you assure the consistence between the work made by these different people? You have given me some of this information on your response to my comments, but I do not see it in the text.
Response 4: We are sorry for not clarifying this clearly. According to your comments, we added the information about the visual interpretations. The revisions are as follows:
The interpretation of the images was performed by the first author. Prior to this work, the first author has undertaken image interpretation in other projects, and has specific professional skills and image processing experience. All of the work, including image pre-processing, classification and identification of different land-use types and the acquisition of the spatial data for each land-use type, took about 8 months.
The newly added materials can be found in the last paragraph in section 3.2.1.
Point 5: In the validation section, you must specify the characteristics of the Google Earth imagery you use as reference. You just say high resolution, but what this means respect to the resolution of the dataset you are exploying?
Also I find really important the missing validation of 1971 images. I know this may be not possible due to the lack of reference data, but could you come with something that at least could give the reader an idea of the quality of that interpretation?
I also agree with the other reviewer in the need of giving more importance to this section and discussing about different validation approaches available and the pros and cons of the one you have selected
In addition, when you do change analysis from different land use maps, you must not only focus on the accuracy of the datasets as a whole, but on the accuracy of the changes they measure. Generally, the dataset may have an accuracy of 80%, but if the changes only represent 5% of the area, this means that all measured changes may be wrong. Therefore, I recommend the authors to validate and visually compare with reference data the changes measured by their series of maps. Especially because they will come with high sources of uncertainty: you are comparing maps that come from very different sources and which are made through different procedures.
Response 5: According to your comments, we thoroughly revised our approach on accuracy assessment.
First, we agreed with your inquiry about the resolution of Google Earth images utilized for validation. Since the resolution of Google Earth image in 1991 is as low as 15m which is lower that of the images (3m) that we have utilized to map land use in 1991, we dropped the idea of using Google Earth imagery for 1991 as reference data. Considering that the reference data in 1972 was also lack, we developed an alternative solution to generate reference data for accuracy assessment. In brief, the approach can be described as follows:
For the lack of available field survey data and high-resolution image data, we developed an alternative approach to generate the reference data. The land use of Tongzhou in 1972 and 1991 were classified by the first author herself using visual interpretation approach. In general, the visual interpretation of a high-resolution satellite image can accurately identify the true land use to a large degree. To further increase the accuracy of visual interpretation, we invited another two experts to help creating reference data. The two experts both have rich experiences in mapping land use of Beijing. Specifically, when the random points utilized to verify accuracy were created, three experts including the first author started identifying the real land use of these points by visually interpreting the satellite images adopted to map land use in this research. Only when the judgements of the three experts are consistent, the land use of the sample point was determined and be utilized as reference data. Using this approach, we created the reference data of Tongzhou in 1972 and 1991. For the validation of the land use map in 2015, we utilized the historical Google Earth images for 2015 with a spatial resolution of up to 0.23 m.
To create reference data, we established 100 random points in Tongzhou using the Create Random Points tool of ENVI software. Then the satellite images for 1972 and 1991 and Google Earth images for 2015 were overlaid with the random points. After the visual interpretations by the three experts, the land use of the random points was determined. If the land use classification of a certain point is not unanimous among the three experts, the point was dropped from the reference data. Then another random point was created as an alternative point for reference data. Finally, 100 random reference data of 1972, 1991 and 2015 were created under the efforts of the three experts.
Second, according to your comments, we reviewed the main different validation approaches. In brief, Accuracy is was assessed by the agree degree of produced map with the reference classification [49]. Accuracy assessment usually requires a sample-based validation [50]. In summary, the accuracy assessment witnessed a change from qualitative assessment to quantified assessment [51, 52]. In the first stage, the accuracy assessment was conducted by visual judgement which have a great deal of uncertainties. In the second stage, three sub-stages were divided. In the first sub-stage, the area ratio of produced map was compared with that of reference data to assess the accuracy. The limitation of this method is lack of the validation of the location. In the second sub-stage, the accuracy assessment focused on the comparison of site specific of land use type and accuracy metrics. In the followed stage, error matrix was proposed to assess the accuracy. Confusion matrix has several statistics and thus can assess the accuracy from several aspects. In general, by using error matrix, we can calculated the overall accuracy, Kappa coefficient, producer accuracy and user accuracy. The overall accuracy and Kappa coefficient describe the overall accuracy of classification of all types while the producer accuracy and user accuracy describe the accuracy of classification of a single type [53, 54]. Nowadays, error matrix was widely adopted to assess accuracy of land use classification.
In spite of error matrix, several other approaches have been proposed for accuracy assessment [55], such as the approach of fuzzy sets [56], an error model developed by Michael et al.[57] and a weighted analysis of variance adjustment approach [58]. It can be seen that there is no standard method for accuracy assessment [55]. In consideration of the wide adoption of error matrix, we assessed the accuracy of our maps based on error matrix [59-61].
Last, according to your comments, we verified the accuracy of mapped changes. To validate the changes, we extracted the changes from 1972 to 1991 and from 1991 to 2015. Within these changes, we created 50 random points to verify their accuracy. The generation of the reference data was the same with the approach utilized to verify the maps of 1972, 1991 and 2015. Finally, the accuracy of the changes was assessed by calculating the proportion of corrected conversion samples to total conversion samples.
These revisions can be found in the section of “3.2.2 Validation”.
Point 6: Finally, you must explain clearly in section 3.2.3 how you carry out the analysis that you show in section 4.3. You explain the obtation of quantitative data in tables, but do not say anything about the spatial analysis that you make by means of overlying maps.
Response 6: We are sorry for not clearly clarifying this. According to your comments, we have added detailed descriptions about the process of spatial analysis. Totally, five steps were adopted to carry out the analysis. The detailed revisions are as follows:
According to the equations of calculating IHC, five steps of spatial analysis were performed to acquire IHC. The maps of land use in 1992 and 2015 were resampled in a spatial resolution of 2m. Second, rural settlements expansion were mapped by overlapping land use maps of any two periods. Third, the arable land losses resulted by rural settlements expansion were mapped by overlapping the expanded rural settlement with the arable land of the land use map in the start of the research period. Fourth, the high-quality arable land of Tongzhou were identified using the followed equation [47]:
AQR=AQp/AQa (4)
where, AQR is the arable land quality rank; AQp is the land quality score of p parcel; and AQa is the average quality score of the total arable land. The low, medium and high quality arable land of Tongzhou were determined according to AQR with the value ranges 0-0.8, 0.8-1.2 and 1.2+ [47], respectively. Thus, the high-quality land map of Tongzhou was generated according to AQR.
Last, the map of lost arable land due to rural settlement expansion was overlapped with the high-quality land map of Tongzhou. Thus, the high-quality arable land loss map can be generated. According to equation (1), (2) and (3), the IHC can be calculated.
These revisions can be found in the last part of section 3.2.3.
Point 7: Results：Figure note of 202 and 206 must be avoided and the information integrated in the figure itself. The first point is already addressed by the legend of the figure. In the other cases, you may state which land use is each land parcel by putting a text below each image.
Same applies to note of figure 3.
Response 7: Thanks for your suggestion. We have revised Figures 2 and 3 according to your comments. Specially, we deleted the "note" in the texts and explained the related contents at the bottom of each figure.
Point 8: In addition, section 4.1 lacks of a lot of information about quality and accuracy of dataset. Given you have been working with them, is it not possible to inform the reader qualitatively about the accuracy of the dataset? You may also bring here issues related to accuracy and uncertainty of measured changes, the uncertainties related to the validation process you have followed, etc.
Response 8: According to your comments of points 5, 6 and 8, we revised our accuracy assessment approach thoroughly. In summary, we developed an approach to generate reference data and an approach to assess the accuracy of mapped changes. Using the generated reference data, we calculated the error matrix to assess the accuracy. In the revised section 4.1, we presented the accuracy results of these assessment. Considering that comment of point 13 asked us to discuss the uncertainty of the performed analysis in discussion section, we added the discussions of uncertainties of the accuracy assessment to section of “5. Discussion”.
Point 9: Table 4 > to improve the understanding of the table, please include a sum up of the percentage also in the last row, so the reader can understand that the percentages refer to the proportions of changes for each class in the later map.
Response 9: Thank you for your suggestion. Based on your comments, we have added the sum of percentages to the last line of tables 4 and 5 in the revised version.
Point 10: Figure 4 would be easier to read if you just provide the footprint of land uses, especially rural settements, for the three different years in the three different maps (one map with uses for 1972, another one for 1991 and a last one for 2015). In the current form, it is quite confussing to interpret the evolution of rural settlement expansion through the years.
Response 10: Thank you for your suggestion. According to your suggestion, we utilized three maps in the revised draft to show the use of rural settlement in 1972, 1991 and 2015, respectively.
Point 11: Reword the caption of figure 5 as in the current form the reader cannot underastand what it represent. Is it about losses?
Response 11: We are sorry for the confusions. In the revised version, we have changed the caption of figure 5 from “Quality of arable land and its reduction from 1972-1991 (a), and 1991-2015 (b).” to “Figure 5. The quality of lost arable land due to rural settlements expansion from 1972-1991 (a), and 1991-2015 (b).”.
Point 12: Discussion：The disucssion is now improved, but authors have included a few new paragrpahs that do not directly connect whith what is said in the paper. Plase, in the discussion, discuss the results, but not talk in general about land use changes and setllement expansion in China according to a literature review, because you could even do this without carrying out any analysis. Therefore, authors must connect all these ideas with the results of their analysis.
Response 12: According to your comments, we conducted a major revision on discussion section. In the revision version, we have focused on discussing the results conclude from the paper. A lot of materials talking in general about land use changes and settlement expansion in China according to literature review were dropped. In addition, some other materials that was not closely related to our results were dropped too. Considering this the comments of point 12 and point 13, we added a paragraph to discussion about uncertainty of the performed analysis.
Point 13: I miss proper discussion about uncertainty of the performed analysis. Authors are not critic about the methods and data they have employed, which come with many sources of uncertainty that could have deeply influenced the results thwy have got.
Response 13: According to your comments, we added a new paragraph in the discussion section to discuss about uncertainty of the performed analysis. The revisions are as follows:
In this paper, the decrypted military remote sensing images were used to obtain the spatial data of rural settlements in 1972 and 1991. The images can provide a reference for obtaining long-term land use change data at a relatively high resolution. However, the validation of these maps is very difficult due to the lack of reference data. We proposed an approach to create the reference data. Using these data, the over accuracy of the land use classification for 1972 and 1991 are 93% and 90%, respectively. We also proposed an approach to assess the accuracy of mapped land use changes. It was found that the accuracy of changes from 1972 to 1991 and from 1991 to 2015 was 90% and 88%, respectively. It can be seen that the accuracy of these maps and mapped changes is high. However, there are many uncertainties for the accuracy assessment due to the creation of reference data. Here, the reference data was created by the visual interpretation consistency test among three experts. However, even if the opinions of the three experts are unanimous, the classification of reference data still may be wrong. Considering this problem, the actual accuracy of the land use map for 1972 and 1991 could be lower than that of our assessment. In addition, when assessing the change accuracy from 1971 to 1992 and from 1992 to 2015, we resampled the land use map of 1992 and 2015 to the map of 2m resolution and 30m resolution, respectively. These resamples could also generate some uncertainties for change accuracy assessment.
Point 14: Conclusions：Please, focus on what ideas the paper provides, but not make again a summary of the results of your analysis (first paragraph)
Response 14: Thanks for your suggestion. Based on your comments, we revised the conclusion section. Some materials that was over describing the results in the first paragraph was dropped. In addition, the other parts of conclusion were also modified according to the changes of result and discussion section.
