Multi-Criteria High Voltage Power Line Routing—An Open Source GIS-Based Approach

Round 1

Reviewer 1 Report

Dear Authors:

I believe that the topic is interesting and well addressed. The approach for power lines deployment through GIS, considering additional parameters too, sound very good to me.

Congratulations for your work.

Author Response

Dear Sir or Madam,

We thank you for your kind review of our submission and for your positive feedback.

Kind regards,

Hendrik Scharf

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper presents an approach for determining the best route for powerlines. In this context, the paper presents a cost function. The paper main contribution is to combine information from different sources rather than the approach as it uses well-known techniques for data clustering, and path finding.

My main concerns is that the paper does not clearly state the novelty when compared to other approach. The only claim in this regard which I found is that “Previous papers currently cover only smaller regions, often lying their focus on only a few of the criteria while neglecting the remaining ones.” I strongly think that authors should add a reference to this or better described related works.

Regarding the open-access online platform, it is difficult to determine what is. On one hand, there are a set of derived files for the Case study. On the other hand, there is a git with the scripts for used. I think that authors should present a Python modules, with a generic API that allows other used. With the current scripts, if someone want to use the script for other scenario, it is necessary to mess with the complete script. Moreover, as the propose method could easily add other layers of information and weights, the module should support such configuration. Hence, the Python notebooks should be refactored into a generic modules, and the notebooks for the current case-studies.

Author Response

Dear Sir or Madam,

We thank you for your review of our submission and for your positive and useful feedback.

Below are our responses to your remarks:

Point 1: My main concerns is that the paper does not clearly state the novelty when compared to other approach. The only claim in this regard which I found is that “Previous papers currently cover only smaller regions, often lying their focus on only a few of the criteria while neglecting the remaining ones.” I strongly think that authors should add a reference to this or better described related works.

Response:

Our work distinguishes itself from others in several ways. In contrast to other papers, we make all scripts and data publicly available in order to promote increased transparency to this discussion of high public interest. While other works lay their focus on a smaller project-specific region, often focusing only on a few of the criteria, we provide a wide range of data for the whole of Europe. In our case study, we concentrate on analyzing the interactions between different dimensions and their sensitivity rather than finding “good” weights (which is common in several in literature). Appendix B gives a detailed overview on related works. We have stress these aspects in the introduction of our revised manuscript so that the research contribution of our work and the differences from other studies becomes clearer.

Point 2: Regarding the open-access online platform, it is difficult to determine what is. On one hand, there are a set of derived files for the Case study. On the other hand, there is a git with the scripts for used. I think that authors should present a Python modules, with a generic API that allows other used. With the current scripts, if someone want to use the script for other scenario, it is necessary to mess with the complete script. Moreover, as the propose method could easily add other layers of information and weights, the module should support such configuration. Hence, the Python notebooks should be refactored into a generic modules, and the notebooks for the current case-studies.

Response: While it is possible to use a modules-based code structure for providing the scripts, jupyter notebook was chosen instead. The reason is because it allows for simultaneous documentation and code. We think this would improve the readability/modifiability of the code.  This is especially important if the future studies have a very different structure. Module based code would be easy to use if the structure of the studies is not be changed. However, if significant code change is required, it would be easier to do it in jupyter notebooks. Providing an API is beyond the scope of our work at this stage; however, this can be explored further if significant interest is observed in using the provided data and scripts. In order to avoid any possible misunderstanding, we avoid the use of the wording’s “platform” or “tool” in our revised manuscript.

We modified the manuscript as described above and hope that you find our responses satisfactory.

Kind regards,

Hendrik Scharf

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper presents a project results related to planning large-scale infrastructure projects such as power transmission lines. Authors develop an open-access online platform which allows for transparent and replicable route determination, tracing and assessment.

Generally, this paper is very well-written and clearly states the problem, used data and methods implemented in the open-access platform. The paper explains underlying methods of data collection, processing and algorithms of data preparation, route generation and assessment. Also, for demonstration purposes, two case studies of planned DC lines are examined and analyzed with the platform and geodata set provided.

The paper is technically correct and with content relevant to the journal readers.

My recommendation to authors is to add details that relate to the criteria for combining, calculating and evaluating used dimensions. As authors noted, broad range of criteria needs to be included to cope for this complex setting. 

Also, my dilemma is about weights. Authors noted that to calculate the three dimensions (Acceptance, Environmental and Infrastructure), the criteria within the dimension are combined assigning equal weights to the criteria. Why and how you explain this decision?

How the value for Economic dimension is calculated? It is enough to say “the evaluation of the distance of a path indicate 1.4 mEUR for each kilometre of a DC line”?

Also, there remains a significant question about trust in open data. How authors comments this issue?

Author Response

Dear Sir or Madam,

We thank you for your review of our submission and for your positive and useful feedback.

Below are our responses to your remarks:

Point 1: My recommendation to authors is to add details that relate to the criteria for combining, calculating and evaluating used dimensions. As authors noted, broad range of criteria needs to be included to cope for this complex setting.

Response: The selection of criteria is derived from the results of our literature research. Comparing our criteria with the ones stated in table A2, it turns out that we consider nearly all of those criteria. Please consider that Corine data itself includes many types of terrain. Several other studies take those into account as many single criterion (e.g. wetland, forests, recreation, etc.).

Point 2: Also, my dilemma is about weights. Authors noted that to calculate the three dimensions (Acceptance, Environmental and Infrastructure), the criteria within the dimension are combined assigning equal weights to the criteria. Why and how you explain this decision?

Response: The criteria within environment and infrastructure dimensions are equally weighted. This is a neutral perspective. We have not analyzed this issue in detail. For infrastructure, the rationale behind this is that the advantageousness of following such existing infrastructure is the same for each type of infrastructure. For the environmental dimension, besides from the Corine data set, protected areas and river courses has been into account separately. Here, the finding of taking equal weights is also derived from results which come from our investigations on varying the proportion of weighting factors. With the extra weight on rivers, we avoid multiple river crossings of line routes. For finer adjustment of these weights, further expert knowledge would be required. These weights can also be seen as exemplary proposals. For further in-depth analysis it possible to change the weighting factors easily within the provided scripts.

Point 3: How the value for economic dimension is calculated? It is enough to say “the evaluation of the distance of a path indicate 1.4 mEUR for each kilometre of a DC line”?

Response: On the basis of our literature research, it is extremely challenging to make accurate statements on financial estimations. The value above is certainly very rough. As publicly available geo data for the economical dimension, we identified line length itself and the slope of a terrain. However, for accurate estimations, further data such as public and private land owning, specific technology type cost, spatial correctors for construction costs (regional factors), multipliers for different kinds of land, etc. would be required. Length only is a very simple indicator which should be, however, not completely disregarded during rough planning. We think that economic evaluation is a very detailed topic in itself, big enough for another study. We do not analyze this issue as we focus on the non-monetary dimensions. However, in the introduction and subsection 2.1, we stress that our approach alone is not sufficient for good monetary estimations and that it would require further information.

Point 4: Also, there remains a significant question about trust in open data. How authors comments this issue?

Response: In our eyes, we chose good reliable open sources (European Environment Agency, UN Environment World Conservation Monitoring Centre, and European Commission Joint Research Centre) for geo data. For Open Street Map, several sources conform a good validity of data. Furthermore, if there are some reliability issues in the data at high resolution level, it would be reduced in the process of reducing the resolution to 500 m. Also, this work can be seen as first estimates w.r.t route planning. When more detailed plan is to be required, it would be important to collect high accuracy and high-resolution data. Besides, we consider it a value in itself that we only use open data as this topic is of high public interest involving multiple stakeholders. 

We modified the manuscript as described above and hope that you find our responses satisfactory.

Kind regards,

Hendrik Scharf

Author Response File: Author Response.pdf

Reviewer 4 Report

First, I would like to thank the authors for conducting this very interesting and useful research and making the toolset publicly available. I thoroughly enjoyed reading the article, and I would like to see this published after addressing a few concerns of mine.

I am also attaching a scanned version of the reading copy I worked on in case it is further useful.

Main Comments

1) Language: While generally well written, I think the paper would benefit tremendously from a copyeditor whose native language is English. There are lots of grammatical mistakes that need to be corrected before the publication. It wouldn’t look good neither for the journal, nor for the authors. I noticed some parts that can be substantially shortened or omitted, as well. Here are some examples (also check out the scanned version):

Pg2-Line44: provide -> provided

Pg2-Line47: Our also -> Also, our

Pg2-Line52: such as e.g. -> such as 

Pg2-Line52: and e.g. -> and

1.1) There are instances of inattentive writing: For example, it looks like your final sentence (page 15 lines 417&418) on future work was written for the sake of writing it.

2) This study is more of a utility rather than novel research, which is fine. However, I would expect more rigorous analyses accordingly. One thing that the paper would benefit substantially is to conduct some sort of a validation analysis, such as examination of existing power lines, and to critique those existing power lines with the same logic you applied to the hypothetical lines. For example, you could take an existing high voltage power line, and place alternative routes based on the weighting factors you used in Figure A1 (Acceptance, Infrastructure, Environment, Equally Weighted) with the same Start/End points on a map. Such comparison maps would allow evaluating the existing lines as well as the proposed lines, and to see and discuss potential shortcomings of existing lines or your methodology (which is fine).

3) I see that in the conclusion the authors make a point about the cost analysis, and that the length of the line being a proxy to the cost. I think a more comprehensive cost estimation that would incorporate other factors such as the elevation, slope, public/private ownership would make sense for the general reader, and it would make a whole lot more sense for the policymakers who might want to check out your work. 3.1.) On page 15, lines 391-392 you mention how provided data allows estimating monetary aspects, but that’s not accurate.

4) Raster cell size: The authors lowered the resolution (e.g. 25 meters to 500 meters). This logic is not obvious to me. You could resample the datasets with 500 meters to 25 meters so that you could capture all the information in the high-resolution imageries, while losing nothing in the low-resolution imagery (you would have a lot of identical cells in the originally low-resolution datasets). Please address this or discuss why this would be a bad idea.

Minor comments:

1) It would be beneficial to indicate that this is a macro scale solution (not readily applicable intracity), therefore it would be helpful to rename the title so that it would reflect that (e.g. Multi-Criteria High Voltage Power Line …)

2) After studying Figure 7 and then looking at the appendix D (Figure A1) I was confused because I was expecting to see the labels A, B and C on the figure A1. I would either suggest removing A, B, C from Figure 7, or create another map for the appendix that draws corresponding lines on the map.

3) I don’t see why you need to showcase two case studies. I don’t think the second one adds much to the main manuscript. I can instead be added to the supplementary material or completely omitted.

4) Page 15, line 408-409: I don’t think using raster datasets is a big problem. As far as I understand this is a decision assistance tool and should be used to generate macro level recommendations. I imagine transmission planners and engineers work at finer scales (even 1:1000) to work through physical obstacles or private plots of lands.

5) As far as I know, we use jenks classification to present the data meaningfully to human eyes. Within the context of multicriteria decision making, equal range classification makes more sense to me as the difference between 2 and 1 and the difference between 7 and 6 are the same, while this is not the case for the Jenks classification. Please address this.

6) Figure 4: I can only see a few colors on the map, it looks quite uniform. Can you try a different color scheme to identify colors better?

Comments for author File: Comments.pdf

Author Response

Dear Sir or Madam,

We thank you for your review of our submission and for your positive and useful feedback.

Below are our responses to your remarks:

Point 1: Language: While generally well written, I think the paper would benefit tremendously from a copyeditor whose native language is English. There are lots of grammatical mistakes that need to be corrected before the publication. It wouldn’t look good neither for the journal, nor for the authors. I noticed some parts that can be substantially shortened or omitted, as well.

Response: We have now used English language proof-reading services to improve the script. Moreover, we revised the text with regard to parts which may bloat the text unnecessarily.

Point 2: This study is more of a utility rather than novel research, which is fine. However, I would expect more rigorous analyses accordingly. One thing that the paper would benefit substantially is to conduct some sort of a validation analysis, such as examination of existing power lines, and to critique those existing power lines with the same logic you applied to the hypothetical lines. For example, you could take an existing high voltage power line, and place alternative routes based on the weighting factors you used in Figure A1 (Acceptance, Infrastructure, Environment, Equally Weighted) with the same Start/End points on a map. Such comparison maps would allow evaluating the existing lines as well as the proposed lines, and to see and discuss potential shortcomings of existing lines or your methodology (which is fine).

Response: For the German case study, we added the route which has been recently proposed by the responsible transmission system operators to the map in the appendix. Furthermore, we plotted the overlapping of our determined routes and the planned one with another ternary diagram.

Point 3: I see that in the conclusion the authors make a point about the cost analysis, and that the length of the line being a proxy to the cost. I think a more comprehensive cost estimation that would incorporate other factors such as the elevation, slope, public/private ownership would make sense for the general reader, and it would make a whole lot more sense for the policymakers who might want to check out your work. 3.1.) On page 15, lines 391-392 you mention how provided data allows estimating monetary aspects, but that’s not accurate.

Response: On the basis of our literature research, it is extremely challenging to make accurate statements on financial estimations. The value above is certainly too rough. As publicly available geo data for the economical dimension, we identified line length itself and the slope of a terrain it has to overcome with. However, for more accurate estimations, further data such as public and private land owning, that you already mentioned, specific technology type cost, spatial correctors for construction costs (regional factors), multipliers for different types of land, etc. would be required. Length is only a very simple indicator which should be, however, not completely disregarded during rough planning. We think that economic evaluation is a very detailed topic in itself, big enough for another study. We do not analyze this issue as we focus on the non-monetary dimensions. However, in the introduction and in subsection 2.1, we stress that our approach alone is not sufficient for good monetary estimations and that it would require further information.

Point 4: Raster cell size: The authors lowered the resolution (e.g. 25 meters to 500 meters). This logic is not obvious to me. You could resample the datasets with 500 meters to 25 meters so that you could capture all the information in the high-resolution imageries, while losing nothing in the low-resolution imagery (you would have a lot of identical cells in the originally low-resolution datasets). Please address this or discuss why this would be a bad idea.

Response: Using higher resolution data would definitely improve the accuracy of the results. However, file sizes and computational requirement increase significantly. Currently, size of our provided data set is 2.6 GB. If we increase resolution to 250 m, it would be 12.8 GB. Thus, we think that a 500 meter rastering is a good compromise between file size and level of detail. Since we present our data sources and processing steps in detail, if smaller areas are to be analyzed, it would be possible and feasible to apply higher-resolution rasters using the stated sources and our scripts. There are associated papers with far smaller regions which have even lower resolutions. Papers which use higher resolutions assess a specific project and thus cover comparatively small regions, i.e. in most cases with areas in the double-digit square kilometer range.

Point 5: Minor comments.

1)      It would be beneficial to indicate that this is a macro scale solution (not readily applicable intracity), therefore it would be helpful to rename the title so that it would reflect that (e.g. Multi-Criteria High Voltage Power Line …)

Response: We renamed the title so that it becomes clearer that it is a macro-perspective.

2)      After studying Figure 7 and then looking at the appendix D (Figure A1) I was confused because I was expecting to see the labels A, B and C on the figure A1. I would either suggest removing A, B, C from Figure 7, or create another map for the appendix that draws corresponding lines on the map.

Response: We replaced the routes in the map accordingly.

3)      I don’t see why you need to showcase two case studies. I don’t think the second one adds much to the main manuscript. I can instead be added to the supplementary material or completely omitted.

Response: The additional value to our work is considered to be rather small, but still we think that another case study better verifies our approach and helps to show the various results for different terrains throughout Europe. That is why we would like to have the second case study in the main text.

4)      Page 15, line 408-409: I don’t think using raster datasets is a big problem. As far as I understand this is a decision assistance tool and should be used to generate macro level recommendations. I imagine transmission planners and engineers work at finer scales (even 1:1000) to work through physical obstacles or private plots of lands.

Response: The critiques on rasters have been removed from the manuscript.

5)      As far as I know, we use jenks classification to present the data meaningfully to human eyes. Within the context of multicriteria decision making, equal range classification makes more sense to me as the difference between 2 and 1 and the difference between 7 and 6 are the same, while this is not the case for the Jenks classification. Please address this.

Response: We decided to use jenks classification after experimenting with linear scaling. When we worked with equal classes, the results were unsatisfactory, as most of the points were underrepresented because of extreme high values in few raster cells.

6)      Figure 4: I can only see a few colors on the map, it looks quite uniform. Can you try a different color scheme to identify colors better?

We changed the color palette to visually emphasize the differences.

We modified the manuscript accordingly, also considering your comments on the scanned document, and hope that you find our responses satisfactory.

Kind regards,

Hendrik Scharf

Author Response File: Author Response.pdf

Round 2

Reviewer 4 Report