Evaluation of the Level of Farmland Infrastructure Based on High-Resolution Images of UAV

Round 1

Reviewer 1 Report

I agree that it is a key to promote to modernize the agricultural infrastructure to secure food supply around the world. To do that, to monitor the agricultural infrastructure efficiently in timely fashion is very important. So, using remote sensing to identify the objective of infrastructure is good research topic. Even if, I believe UAV is still very limited to monitor such infrastructure because of stability and of continuously and uniformity of observation data quality. So, I have a couple of questions and suggestion about your research.

 

1.      Can you explain and do any comparison between UAV based result and GeoEYE or other high resolution optical satellite based data such as Skysat?

2.      Can you please specify what is priority of infrastructure which you would like to detect. Such as drain system or irrigation. Which is more important or high priority? And what is your research detection accuracy about them?

3.      eCongnition is famous categorize software. What is your research uniqueness even if you use eCongition? How about the difference if you use random forest or other AI/ML technology instead of eCongition?

4.      In your study, please specify the limitation of your method. I believe even if UAV data, there are several limitation to detect infrastructure.

5.      Chapter 3.3.2 is very difficult to understand because you explain one by one without total picture.

6.      Same as about No 2, can you please define the meaning of farmland fields, … shown in table 2? And how to set grade? Do you have any standard definition of grade shown in table 3?

7.      How to define excellent, good and poor grade with 1-3 grade? Can you also define it clearly?

English text is fine

Author Response

Dear Reviewer,

     Thanks very much for taking your time to review this manuscript. On behalf of my co-authors, we thank you very much for giving us an opportunity to revise our manuscript, and we also appreciate you very much for your positive and constructive comments and suggestions on our manuscript entitled “Evaluation of the level of farmland infrastructure based on high-resolution images of UAV” (Manuscript Number: 2533765). We have addressed the comments raised by the reviewers, and the amendments are highlighted in red in the revised manuscript. Please find my itemized responses in below and my revisions/corrections in the re-submitted files We hope that the revision is acceptable, and I look forward to hearing from you soon.

1. Can you explain and do any comparison between UAV based result and GeoEYE or other high resolution optical satellite based data such as Skysat?

    The study first chooses to use UAV mapping to acquire high-resolution images because UAVs have fewer requirements for the geographic airspace of the flight as well as the meteorological conditions, the study area experiences more rainfall in the summer, the weather changes quickly, and UAVs are characterized by a rapid response to mapping with a preparation time for lift-off of only about 15 minutes. Traditional high-resolution satellite image data has limits due to slow data collecting and poor timeliness, making it difficult to get the information needed right away. Furthermore, because of the size of China and its complicated terrain and climate, the quality of mapping is necessarily impacted by limitations in the collection of satellite remote sensing data due to topography and cloud cover.

    Remote sensing photos and drone photographs both have their own benefits and shouldn't be used in place of one another rather than as a complement. The drone doesn't have the advantage of remote sensing and mapping in a large area range because of the battery factor, and if it guarantees the accuracy and range at the same time, it must increase the workload and financial investment. Despite the drone image having high resolution, good timeliness, and being less affected by weather conditions, the image obtained is specific time, and can't complete the continuous observation for a long time. The inadequacies can be made up for by the satellite picture, which also provides observation data that is more extensive and macroscopic. It can also be continually monitored for a long period, which can satisfy the demand for remote sensing across a broad area. This deficiency can be made up for by the observation data provided by satellite imaging, which is more extensive and macroscopic, capable of long-term continuous monitoring and able to meet the needs of remote sensing across a wide area.

2. Can you please specify what is priority of infrastructure which you would like to detect. Such as drain system or irrigation. Which is more important or high priority? And what is your research detection accuracy about them?

   Farmland ditches are the most significant feature when establishing the weights of the indicators, followed by farmland fields, farmland roads, and finally farmland forest belts, according to the priority ranking found in chapter 2.3.3 of the article. Regular fields, well-constructed roads, and canals have good extraction effects in the actual extraction process, whereas weeds affect drainage ditches, branch canals, and forest belts. To ensure accuracy, after completing the classification using the eCognition software, the corrections will be further confirmed through the visual interpretation method; the sample area is not large, and the accuracy of the results obtained by using the visual interpretation method will be further confirmed.

3. eCongnition is famous categorize software. What is your research uniqueness even if you use eCongition? How about the difference if you use random forest or other AI/ML technology instead of eCongition?

   In this study, the eCognition software is used to implement the object-oriented classification method of segmentation followed by classification, which overcomes the limitation of traditional remote sensing software to classify images based solely on spectral information and significantly increases the accuracy of automatic identification of high spatial resolution data. Classification techniques like random forests collect typical information in terms of individual pixels. This leads to a situation where the localisation is overemphasized while the geometric structure of the surrounding patch as a whole is neglected. The object-oriented supervised classification method of eCognition software is the most appropriate classification method for classifying farmland infrastructure, which consists of farmland plots, farmland roads, farmland ditches, and farmland forests. This method ensures the completeness of each type of infrastructure and the high accuracy of the classification at the same time.

4. In your study, please specify the limitation of your method. I believe even if UAV data, there are several limitation to detect infrastructure.

   First of all, the limitations of UAV mapping have been explained in NO.1, and the UAV mapping used in this study obtained high-resolution images, and due to the limited flight range of the UAV and the vast area of Kenli District, the six representative sample areas selected can reflect the basic farmland infrastructure situation in Kenli District, but they are not 100% representative.

    The study also neglected to examine the evaluation indexes and evaluation methods applicable to various types of landforms, different crops, various climates, and various land types, as well as the evaluation index system of the farmland infrastructure. This study is a preliminary exploration of the evaluation of the level of farmland infrastructure, and farmland infrastructure has a strong geographical, complex, and dynamic nature.

.5. Chapter 3.3.2 is very difficult to understand because you explain one by one without total picture.

Part 3.3.2 of the original text elaborates on the quantitative formulas for a total of 10 factors included in the aforementioned four areas. Part 3.3.1 of the original text is a segmented discussion of farmland fields, farmland roads, farmland ditches, and farmland forest networks. I have restructured the original text for readability by combining the two parts in the hope of expressing a clearer picture. (In the amended section, certain details have been underlined.)

6. Same as about No 2, can you please define the meaning of farmland fields, … shown in table 2? And how to set grade? Do you have any standard definition of grade shown in table 3?

farmland fields：A field is a section of land that has been divided into smaller sections with precise measurements and shapes for the same purpose or for many uses, eventually forming a contiguous area of land. According to varied needs and goals, fields can be divided into a variety of sizes and shapes, including square, rectangular, triangular, etc.

farmland roads：The final level of the road network is the farmland roads, which are the infrastructures constructed to link fields and roads for the transportation of agricultural products, labor, farming equipment, and other production activities.

farmland ditches：A ditch is a modest to large depression used to control the direction of water flow. Ditches can be used to channel water from farther-off sources for plant irrigation or to drain water from low-lying regions, roadways, or fields.

farmland forestry networks：A farmland forestry network is a protective forest created by planting a single row or more than two rows of trees or shrubs in the shape of a narrow forest belt or small grid in accordance with specific spacing, width, structure, and direction design around the roadside, canal, ridge, and farmland.

The three classification criteria listed in table 3 were developed based on our reflection after looking through pertinent data and incorporating findings from related studies. The usage of variables with already established criteria is chosen, such as the regularity of fields and the agricultural plot's slope. However, no unique standard was found for density, perimeter ratio, or area ratio, therefore we used the total indicator data of the six study sample areas as the foundation, dividing it into three grades of varying degrees, each with its own quantitative standard. Different levels represent varying degrees of infrastructure service to agriculture and infrastructure functional strengths. Grade 1 denotes excellent service function, Grade 2 mediocre service function, and Grade 3 poor service function.

7. How to define excellent, good and poor grade with 1-3 grade? Can you also define it clearly?

We integrated the assessment results graphics to characterize three grades:

Excellent grade is very complete infrastructure，fields are flat and regular, concentrated and continuous; roads are well constructed and convenient for transportation; irrigation canals and drainage ditches are complete, neat, and clearly visible; and forest belts are neatly arranged and moderately dense.

Good grade indicates a relatively full infrastructure, with regular fields, undulations within the fields, clear roads that suit transportation demands, somewhat intact irrigation canals and drainage ditches, and a slightly broken but overall well-organized forest belt. 

Poor grades include fields with weeds and wasteland, roads with damaged potholes and slightly problematic access, ditches with breaks or weeds, and forest belt deficits or no forest belts.

（Because of the vast number of evaluation elements and their division into distinct levels, the definition emphasizes a qualitative description, which is slightly more difficult to define quantitatively.）

Reviewer 2 Report

1.    Authors should change the keyword, and use a short key word

2.    Please add a theatricals background title or use the following title: Research methodology and theatricals background

3.    Is better to use appendix to present some information

4.    Please use results and discussion instead results and analysis

5.    Discussion part must be revised, in this part authors only descripted study (it is the same as  conclusion)  (is not a discussion) and authors already use analysis in the results and analysis part

6.    Authors should explain the results heterogeneity ( excellent, good and poor)

7.    Authors’ presents a lot of important information, but the paper is not good structured. Please try to improve paper structure

1.    Authors should change the keyword, and use a short key word

2.    Please add a theatricals background title or use the following title: Research methodology and theatricals background

3.    Is better to use appendix to present some information

4.    Please use results and discussion instead results and analysis

5.    Discussion part must be revised, in this part authors only descripted study (it is the same as  conclusion)  (is not a discussion) and authors already use analysis in the results and analysis part

6.    Authors should explain the results heterogeneity ( excellent, good and poor)

7.    Authors’ presents a lot of important information, but the paper is not good structured. Please try to improve paper structure

Authors use a sample and good English language 

Author Response

Dear Reviewer,

    Thanks very much for taking your time to review this manuscript. On behalf of my co-authors, we thank you very much for giving us an opportunity to revise our manuscript, and we also appreciate you very much for your positive and constructive comments and suggestions on our manuscript entitled “Evaluation of the level of farmland infrastructure based on high-resolution images of UAV” (Manuscript Number: 2533765). We have addressed the comments raised by the reviewers, and the amendments are highlighted in red in the revised manuscript. Please find my itemized responses in below and my revisions/corrections in the re-submitted files .We hope that the revision is acceptable, and I look forward to hearing from you soon.

1. Authors should change the keyword, and use a short key word

    The keywords have been modified by changing the original keywords: UAV high-resolution imagery; extraction of feature information; the level of farmland infrastructure ; evaluation system to UAV images; extraction of features; farmland infrastructure ; evaluation system (The modified contents have been highlighted in the manuscript).

2. Please add a theatricals background title or use the following title: Research methodology and theatricals background

    Concerning the background material content, such as the study area selection, data gathering, and processing, we merged it with the Methods section, which has been moved to the Materials and Methods section, so that it can also better comply with the required format of the paper. (The modified contents have been highlighted in the manuscript).

3. Is better to use appendix to present some information

    The paper contains more figures and tables, and using an appendix for presenting would be more concise. However, on the one hand, because the figures and tables bridge the content of the article and are easily read, and on the other hand, to comply with the journal's manuscript format, we still choose not to change the position of the figures and tables. we hope to receive your understanding.

4. Please use results and discussion instead results and analysis

    We have corrected it,We have deleted analysis from the title and changed the title to Results, and the content of Discussion is presented separately as Part 4. And the content has been modified and optimized (The modified contents have been highlighted in the manuscript).

5. Discussion part must be revised, in this part authors only descripted study (it is the same as  conclusion)  (is not a discussion) and authors already use analysis in the results and analysis part

    The analysis and conclusion content was removed from the Discussion section, and new content was added. This new content discusses the benefits of using drone surveying and ecognition software for classifying land objects in the article as well as the areas where the current farmland infrastructure evaluation system needs to be improved and refined in the future. It also discusses the possibility of studying large-scale inversion research through the eval system in greater detail. (The modified contents have been highlighted in the manuscript).

6. Authors should explain the results heterogeneity ( excellent, good and poor)

Based on the figures and tables of the evaluation results, We have summarized their definitions,

      The level of excellence is that the infrastructure is very complete, the fields are flat and orderly, and concentrated and contiguous; The roads are well constructed and the transportation is convenient; Irrigation channels and drainage ditches are intact, neat, and clearly visible, with neatly arranged forest belts and moderate density. A good level is that the infrastructure is relatively complete, the fields are neat, there are undulations in the fields, the roads are clear, and can meet transportation needs. The irrigation and drainage channels are relatively complete, and the forest belt is slightly broken but arranged in an orderly manner. The poor level refers to slightly complete infrastructure, with weeds and wasteland in the fields, damaged and pitted roads, slightly inconvenient transportation, damaged or weedy ditches, and a shortage or absence of forest belts. (The modified contents have been highlighted in the manuscript）

7. Authors’ presents a lot of important information, but the paper is not good structured. Please try to improve paper structure

    We have restructured the structure of the article ,the “2.Area of Study, data collection and processing”and “3.Research Methodology” have been combined into the current “2. Materials and Methods” due to changes in the article's structure. The original “Results and analysis” has been changed to “3.Results” and the “4.Discussion” has been redone. A more streamlined and logical structure has been achieved by reorganizing the article and making certain content changes. (The modified contents have been highlighted in the manuscript）

Round 2

Reviewer 1 Report

Thank you for your reply but your reply for my question are not well incorporate to your paper yet. Such as comparison with AI/ML of No3, No 6 of definition of type, and No 7. Please add some text for those answers. 

Author Response

Dear Reviewer,

    Thank you very much for your positive and constructive comments and suggestions on our manuscript entitled “Evaluation of the level of farmland infrastructure based on high-resolution images of UAV” (Manuscript Number: 2533765). We have addressed the comments raised by the reviewers, and the modification are highlighted in green in the revised manuscript. Please find my responses in below and my corrections in the re-submitted files.We hope that the revision is acceptable, and I look forward to hearing from you soon.

1.Thank you for your reply but your reply for my question are not well incorporate to your paper yet.Such as comparison with AI/ML of No3,No 6 of definition of type,and No7.Please add some text for those answers.

We've added relevant content to the article.

• The discussions of AI/ML in No3 have been added to section 2.3.1
• The definitions of four farmland infrastructure types in No6 have been added to section 2.3.2
• The definitions of the three levels in No7 have been added to the end of 2.3.4

In addition, due to the restructuring of the content, The discussion on the comparison of UAV remote sensing imagery with other high-resolution imagery has been reorganized to the end of section 2.2.The description of the reasons for using the ecognition software for categorization rather than other methods has been adjusted to the beginning of 2.3.1.

（All of the above has been highlighted in green in the text）

Reviewer 2 Report

In the 2.3.1 part, please don’t use numbering in the beginning of each paragraph

The same remark in the 2.3.2 part, and please use numbering; like a; b; c;……; for each indexes

Authors should improve the discussion part, they must discuss and explain each finding result, and not give various general information relating to this study, it is better to integrate the discussion part in the results part

good english

Author Response

Dear Reviewer,

     Thank you very much for your positive and constructive comments and suggestions on our manuscript entitled “Evaluation of the level of farmland infrastructure based on high-resolution images of UAV” (Manuscript Number: 2533765). We have addressed the comments raised by the reviewers, and the modification are highlighted in green in the revised manuscript. Please find my responses in below and my corrections in the re-submitted files.We hope that the revision is acceptable, and I look forward to hearing from you soon.

1.In the 2.3.1 part.please don’t use numbering in the beginning of each paragraph

We have removed the serial numbers of each paragraph in 2.3.1 part.

2.The same remark in the 2.3.2 part,and please use numbering:like a;b;c;……;for each indexes

We have changed the serial numbers of each index in 2.3.2 part.

3.Authors should improve the discussion part,they must discuss and explain each finding result,and not give various general information relating to this study,it is better to integrate the discussion part in the results part

The revisions to the discussion section are complete. The results of the study and their significance are discussed first, followed by the innovations, shortcomings and future directions of the article（Changes have been highlighted in green in the Discussion section of the text）

Round 3

Reviewer 1 Report

No further comment