Generalization of Linear and Area Features Incorporating a Shape Measure

Round 1

Reviewer 1 Report

[1]       The order of sections should be well adjusted. There are two sections numbered 3 and two sections numbered 4.

[2]       The section of Discussion is too simply. More details of the experiments can be provided.

[3]       I cannot find any quality analysis in the experiments of cartographic generalization. Only the generalization results are generated.

[4]       No contrast experiments can be found. Some contrast experiments should be performed to show the advantages of the proposed method.

[5]       The innovation of the article needs to be highlighted, and the conclusion is too broad. Specific improvements of the proposed method need to be identified.

Author Response

Dear reviewer,

Thank you for your constructive comments. All of them were taken into account and they have been incorporated in the manuscript as follows:

1. The order of sections should be well adjusted. There are two sections numbered 3 and two sections numbered 4.

• Sections numbers are adjusted and sections are revised accordingly.

2. The section of Discussion is too simply. More details of the experiments can be provided.

• In section 3 a flowchart was added (figure 4) for the analytical presentation of the selection method. Discussion section is fully revised. Major changes are highlighted in yellow.

3. I cannot find any quality analysis in the experiments of cartographic generalization. Only the generalization results are generated.

• Results section is revised to include the requested quality analysis. Major changes are highlighted in yellow.

4. No contrast experiments can be found. Some contrast experiments should be performed to show the advantages of the proposed method.

• In section 2.2 shape measures referred to in past research like Hausdrorff distance, modified Hausdorff distance, discrete Frechet distance, turning function distance, Fourier descriptors distances are tested, and results in tables are provided along with conclusions.

5. The innovation of the article needs to be highlighted, and the conclusion is too broad. Specific improvements of the proposed method need to be identified.

• The results and discussion section are enhanced, and the innovation of the research is clarified in the discussion section 6. Future work on improvements of the proposed method is included in the conclusions section 7 (highlighted in yellow).

Besides the above, other improvements have been made in the manuscript like: the addition of map at the original scale in order to be used for comparison with the generalized one, the additional references and their relationship to the methodology developed, etc.

Reviewer 2 Report

In this article, the automatic cartographic generalization of linear and area features was carried out, and the evaluation model was developed for the assessment and verification of the generalization results based on the consideration of shape preservation. The authors comprehensively considered the processes of cartographic generalization. However, I think there are some places in the article that need further explanation. Here are some of my questions and suggestions:

1.     Can shape measurement give a quantitative value for the generalization result finally? In this way, different cartographic generalization methods can be compared with each other.

2.     Add a flow chart to describe the selection of the generalized features in Section 3. 

3.  What is the meaning of numbers in Figure 5-7?

 4.  The data at 1:250000 could be added into Figure 8 for comparison the generalization result. 

5. The author should emphasize the innovation of the article, and more latest literatures needs to cite.

Author Response

Dear reviewer,

Thank you for your constructive comments. All of them were taken into account and they have been incorporated in the manuscript as follows:

1. Can shape measurement give a quantitative value for the generalization result finally? In this way, different cartographic generalization methods can be compared with each other.

• In the revised discussion section 6 it is clarified that the selection method was developed due to the fact that existing approaches cannot lead to an adequate value of the generalization result to be used for portrayal.

2. Add a flow chart to describe the selection of the generalized features in Section 3. 

• In section 3 a flowchart was added (figure 4) for the analytical presentation of the selection methodology.

3. What is the meaning of numbers in Figure 5-7?

• Numbers in figures 5-7 are lines/ polygons id numbers. They are removed to avoid any ambiguity. We used them in the framework of the research in order to apply and evaluate the methodology developed to a number of over 50 lines with different shapes

4. The data at 1:250,000 could be added into Figure 8 for comparison with the generalization result.

• A figure of the road network at scale 1: 250,000 is added now in section 4 (figure 9).

5. The author should emphasize the innovation of the article, and more latest literatures needs to cite.

• New approaches regarding cartographic generalization were added in the introduction section and past research references regarding shape measures were added in section 2.1. List of References is revised accordingly. Results and discussion sections are differentiated now into two different sections (sections 5,6). The results and discussion are enhanced and fully revised. The innovation of the research is better clarified in the discussion section 6.
• The most important changes to the manuscript are highlighted in yellow in order to assist in the review reading. Other minor changes have been made also like improvements to the language etc.

Reviewer 3 Report

This paper explores different generalisation techniques for cartographic lines and polygons. The study is clear about its aims and techniques and presents a solid discussion of some accuracy measures. Figures 4-7 give a good idea about the results obtained after applying the different algorithms discussed. In general, this work is well presented. However, I failed to see its relevance as the discussion section is quite short.

Moreover, I don't know what the contribution of the authors is. The paper replicates the cited algorithms (without providing any code, so we don't know how the results were produced). Authors need to clarify what the contribution is and how their proposal differs from the available commercial options. QGIS has all these features, for example. This is why the discussion section needs major improvements. I don't know if IJGI asks for the computer codes, but providing those is fundamental for an open science approach and ensuring the results are accurate as a reviewer. 

One detail is that the pdf is not rendered correctly on page 9. As a result, I cannot read tables 6 and 7. 

Author Response

Dear reviewer,

Thank you for your constructive comments. All of them were taken into account and they have been incorporated in the manuscript as follows:

1. This paper explores different generalisation techniques for cartographic lines and polygons. The study is clear about its aims and techniques and presents a solid discussion of some accuracy measures. Figures 4-7 give a good idea about the results obtained after applying the different algorithms discussed. In general, this work is well presented.

However, I failed to see its relevance as the discussion section is quite short.

2. As you pointed out the basic concept of the article is the identification of shape measures for linear and polygonal features and a methodology for the selection of the appropriate generalized  feature for portrayal (sections 2,3). The quality model presented in section 4 contributes to the quantitative assessment of the features selected for portrayal based on the methodology described. The results and discussion sections are revised and enhanced, and the innovation of the research is clarified in the discussion section 6. 
3. Moreover, I don't know what the contribution of the authors is. The paper replicates the cited algorithms (without providing any code, so we don't know how the results were produced). Authors need to clarify what the contribution is and how their proposal differs from the available commercial options. QGIS has all these features, for example. This is why the discussion section needs major improvements. I don't know if IJGI asks for the computer codes, but providing those is fundamental for an open science approach and ensuring the results are accurate as a reviewer

• Python code developed in the framework of the research is not required for inclusion in the submission but they are available. Shapely and scipy Python modules available online and ArcGIS simplification algorithms together with a number of other functions developed have been used and referred in section 2.2. QGIS offers a number of features for generalization but not for shape description and assessment of the generalization results. The innovation of the research is documented through the results from the tested similarity measures on a set of over 50 lines with different shapes.
• Other improvements were made to the manuscript like the addition of a map at the original scale (1:250,000) in order to be used for comparison with the generalized one, the additional references and their relationship to the methodology developed, etc.

The most important changes to the manuscript are highlighted in yellow in order to assist in the review reading. Other minor changes have been made also like improvements to the language etc.

Round 2

Reviewer 2 Report

Dear authors:

I'm glad to see the detailed revision. There are still two suggestions:

(1) adjust the Figure 4 to make it more clear and readability；

(2) Consider combining Figure 9，Figure10 and Figure 11 into one figure by using different colors so as to directly see the contrast results before and after generalization.

Author Response

Dear reviewer,

Thank you for your constructive comments. All of them were taken into account and they have been incorporated in the manuscript as follows:

Figure No 4 has been modified and adjusted in order to be more readable.

Due to the small scale of the generalized maps and the constraints imposed, an overlay at the resulting scale would not show the differences between the initial and the resulting lines/polygons. In order to respond to your suggestion we added a new figure #12  showing a section of the area enlarged to highlight  the differences. 

Reviewer 3 Report

I appreciate the effort put in by the authors to accommodate all my suggestions and clarify the relevance of their work. As I mentioned, this is a solid work worth publishing.
After reading again, I would further suggest removing Table 1, or at least short it, as is confusing in its present form and does not add any real value to the overall work.
Also, the pdf is still ill-formatted. Tables 1 and t have the line number overlapping, making reading hard. Also, there is a type at Table 7 it says TaTable
Figure 4 is messy. You should not mix the orientation of the text on a single figure (Vertical at the top, horizontal everywhere else). Moreover, the text on the decisions is all over the place. Figure 4 needs much extra work to make it useful.

Author Response

Dear reviewer,

Thank you for your constructive comments. All of them were taken into account and have been incorporated in the manuscript as follows:

Following your suggestion Table 1 is sorted to include only the information that is useful to the reader.

The typo in the legend of Table 7 has been corrected

Figure 4 has been redesigned using the same orientation throughout in order to be clear/useful.

In order to improve the presentation of the results a new figure (#12) has been added showing a section of the area enlarged to highlight the differences before and after generalization.