GPU-Enabled Shadow Casting for Solar Potential Estimation in Large Urban Areas. Application to the Solar Cadaster of Greater Geneva

Round 1

Reviewer 1 Report

Review of the paper "GPU-Enabled shadow casting applied for solar 2 potential estimation in large urban areas"

The paper is of practical and technical interest for solar cadaster methodology since it provides a rather clear and efficient GPU approach that very significantly decreases the required time computation to produce large scale solar cadasters.

The paper is clearly written, well supported by relevant and recent bibliographic references. The figures are ok, as well.

The major concern I could have is that it is more a "technical" paper than a scientific one with a clear innovation and a scientific soundness. This scientific aspect with validations of the work has notably been published in a very nice paper by the authors (https://doi.org/10.3389/fbuil.2018.00014.)

Since the use of GPU has been already published in that domain (and cited by the author), I am wondering (without being totally certain) if the material in this paper justifies a scientific paper, since it can be seen as nice but technical follow-up of the previous paper. To my opinion, without wishing to sound "disparaging", this type of article maybe more suitable for a (very good and interesting) communications than for a scientific paper.

I have minor comments but they are more relevant for the intrinsic methodology already published (accuracy of the building footprints from DUSM, accuracy of the local slope/azimuth estimation from raster DUSM both at 50 cm resolution, model of reflected irradiance, managing of the partial shadowing in a PV system potentially composed of multiple PV modules without µinverters, etc.). And this exactly the point of my major concern.

Considering the real quality of the paper, the decision for publication depends mainly on the policy of the journal.

Author Response

We acknowledge that the scope of the paper is more technical oriented than scientific in the strict sense. The scientific background has already been published through our previous works. Therefore we checked with the editor if, according to the journal policy, there was no objection to publish this kind of paper, provided the improvements as suggested by the reviewers are satisfying. The editor answered that the manuscript fitted the journal scopes.

The changes made in the manuscript are highlighted in blue.

Reviewer 2 Report

The paper “  GPU-Enabled shadow casting applied for solar 2 potential estimations in large urban areas Research on Hydraulic safety assessment of water distribution networks” presents software for mapping out the solar potential in a large scale and a high resolution. The research seems to be interesting, but some parts of the manuscript should be better described. The manuscript requires so many changes in my opinion than I prefer to reject for allowing the authors to perform the huge changes to deserve publication in this journal. Authors should identify the gaps in current research and show the key goals obtained. Also, they should identify the limitations of their methodology, their key goals and overall, why they are making science to advance. In my opinion, Software is not properly described and the key goals are not highlighted. The Authors should have to follow the Key depicted here to have this manuscript accepted. I think this work deserves good publication (unfortunately, now I think the presentation does not meet this level now, but I think it has potential).

Key ideas to change in your manuscript:

-I think it is very necessary to quantify the effect of shadows in energy efficiency, there are much work done in that area. Does your software quantify it? Does it proportion the best location? Will it be used by any public authority? My point is that this could be a very useful tool to encourage PV panels installation, or you could add a fee if having a good location, the community does not invest in this technology.

-It is very important to analyze the input data required to run your software, and indicate the output data (in graphs or Tables, etc..)

-I would like you to compare your results with some other results provided by the software. It will highlight why your software is better.

- I have not found any link to download your software. Is it public? Is it free?

-How much time does your software requirements to get the energy maps? Is it very time-consuming? These questions should be incorporated.

- Maybe Authors should consider this manuscript as a Case study. If you think this is the point, please clarify it in the title of the manuscript. If not, you should add where you used this software (apart from Geneva region).

-The conclusions show future works that must be performed in this manuscript. Sensitivity analysis, comparison with current measurement devices,

Introduction section.

This section should be improved substantially. For the moment it is very short. Some other works should be cited as many approaches worked in this field.

The introduction section should be organised as follows:

• Background
• Gaps in knowledge
• Objective
• paper structure (Now Lines 51-57)

In the manuscript, you start explaining what you have done in Line 33. No, you should follow the previous ideas.

The key idea you should follow is to show why your software is better than the many others now available online.  Describe the shadow casting procedures, how are they implemented in your software.

State of the art Section.

Lines 72-79. These lines should be moved to Introductions section. Some other parts should be included also in the introduction section. I recommend changing the name of this section.

Other remarks:

Figure 3: This Figure should be improved.

Lines 330-335 should be removed.

Line 344: Why do you exclude water bodies? In many research projects, PV panels are installed in structured on water bodies (even in the sea...).

Table 1: I cannot see the Table properly.  Please change it adequately.

Conclusion section.

The conclusion section should be enlarged and improved. Some points should be done in this manuscript as above mentioned.

Author Response

Below we addressed each comment of the reviewer. The changes made in the manuscript are highlighted in blue.

General comments

Authors should identify the gaps in current research and show the key goals obtained.

• Gaps in the current research were already identified after each reference in page 2. We have now added a short summary of the main gaps in L96-101. The goals of the paper and the proposed tool are also elicited better in L102-106.

Also, they should identify the limitations of their methodology, their key goals and overall, why they are making science to advance.

In my opinion, Software is not properly described and the key goals are not highlighted.

• Key goals are now better elicited in the introduction. It is also specified (in L107) that the proposed solution is represented by a set  of open source algorithms (encoded scripts) . Confusion should not be made with the common online web tools of solar cadaster. An online web tool  was also developed for the solar cadaster of the Greater Geneva, as presented in Section 4. Our paper deals with a new architecture of GPU. More concretely, the improved Figure 3 (see the comment later) hopefully enables to better describe the core tool.

Key ideas to change in your manuscript:

-I think it is very necessary to quantify the effect of shadows in energy efficiency, there are much work done in that area. Does your software quantify it? Does it proportion the best location? Will it be used by any public authority? My point is that this could be a very useful tool to encourage PV panels installation, or you could add a fee if having a good location, the community does not invest in this technology.

• All those aspects was already well described in the Section 4.1 through the online Web tools developed around the Solar cadaster of the Greater Geneva. From shadow casting and solar irradiation outputs, key indicators like solar energy production with panels are post-processed. Referring to the comment of the reviewer, the point 2 in L343-347 is improved. However such applications are not targeted by the paper.

-It is very important to analyze the input data required to run your software, and indicate the output data (in graphs or Tables, etc..)

• To address this comment, Figure 3 is improved , so as to better visualize the inputs and outputs (intermediate and final). Inputs and outputs are also described in the text.

-I would like you to compare your results with some other results provided by the software. It will highlight why your software is better.

• This comment is related to the possibility of downloading the tool, and using it to compare results with other software/tools (time calculation in particular related to our paper). See our answer to the next comment. We add a perspective in this sense in the conclusion L419-422.

- I have not found any link to download your software. Is it public? Is it free?

• The core tool as already mentioned above is not a software but a suit of open source codes. They can be made available online. In the paper, pseudo codes are given in sections 3.1 and 3.2.

-How much time does your software requirements to get the energy maps? Is it very time-consuming? These questions should be incorporated.

• Computing performance and thus execution time is the main scope of the paper. Hopefully with the goals better detailed in introduction, this will come more obvious. Calculation time was already carried out in different parts of the paper: at the end of section 2 (L179-186) and Section 4.2 that presents performance results using different types of masks. It is true that in the first version of the paper, Table 1, which displays time results, was partially masked by mistake.

- Maybe Authors should consider this manuscript as a Case study. If you think this is the point, please clarify it in the title of the manuscript. If not, you should add where you used this software (apart from Geneva region).

• We modified the title adding a second part : « Application to the solar cadaster of the Greater Geneva ». We also use the term « case study » at the end of the introduction in L110.

-The conclusions show future works that must be performed in this manuscript. Sensitivity analysis, comparison with current measurement devices,

• Unfortunately it is not possible to perform those future works within the 10 days given for the revision.

Introduction section.

This section should be improved substantially. For the moment it is very short. Some other works should be cited as many approaches worked in this field.

The introduction section should be organised as follows:

• Background
• Gaps in knowledge
• Objective
• paper structure (Now Lines 51-57)

In the manuscript, you start explaining what you have done in Line 33. No, you should follow the previous ideas.

• We carefully followed the relevant advices of the reviewer by merging the former sections 1 (Introduction) and 2 (State of the art) into one section (Introduction), eliciting in particular the main gaps from the state of the art and the goals of the paper. The common thread of this section seems logical now.

The key idea you should follow is to show why your software is better than the many others now available online. 

• Again the core tool presented in the paper is not a software like the online tools we find in the usual web tools mapping solar potential, or the Web interface we present in Section 4. Comparing online tools would lead to an another topic and thus another paper.

Describe the shadow casting procedures, how are they implemented in your software.

• The main principles of the shadow casting procedure is summarized in the L246-251. More details are given in the reference [1] as mentioned in this part. However in order to support the understanding a new figure (Figure 4) adapted from [1] and illustrating the concept of shadow casting is added. Code listing 1 describes how the concept is implemented.

State of the art Section.

Lines 72-79. These lines should be moved to Introductions section. Some other parts should be included also in the introduction section. I recommend changing the name of this section.

• This was done by merging the former sections Introduction and State of the art as mentioned above.

Other remarks:

Figure 3: This Figure should be improved.

• Yes, right. Figure 3 is completely redesigned. In addition and related to this Figure, the formula equation of the final irradiation calculation is added in page 7, based on [1].

Lines 330-335 should be removed.

• Done: moved to Appendix B.

Line 344: Why do you exclude water bodies? In many research projects, PV panels are installed in structured on water bodies (even in the sea...).

• This is just an example illustrating of how using masks can reduce calculation. In urban tiles we focus only on building roofs. In rural tiles in this example, water bodies correspond to small rivers (this is specified in L373). But of course the reviewer comment is entirely justified: PV installations on large water surfaces is very promising. We add a mention to this aspect in L393.

Table 1: I cannot see the Table properly.  Please change it adequately.

• Yes a part of the tab was accidentally masked in the manuscript version provided by the editor.

Conclusion section.

The conclusion section should be enlarged and improved. Some points should be done in this manuscript as above mentioned.

• As already said above the time limit for revising the paper is not enough to implement some perspectives like sensitivity analysis, which is indeed important. The conclusion has been enlarged and improved particularly in the first and third paragraph.

Round 2

Reviewer 2 Report

The paper “  GPU-Enabled shadow casting applied for solar potential estimation in large urban areas. Application to the solar cadaster of the Greater Geneva" describes a software solution for mapping out the solar potential in a large scale and a high resolution. The manuscript has improved in this second round. However, I feel that although the manuscript is close to publication, some changes are still required. I am going to choose minor changes as they can be performed in a short period of time.

The introduction:

Parragraph 67-95. This should be shortened. It shows key ideas in one manuscript for each paragraph. This must be improved. Many studies have been developed.

Key goals obtained. Authors have written them. OK

Limitations of their methodology. I haven't found them. Please add them. 

Rev: I like the section3.1, the workflow has been improved and formulas are shown here. 

Author Response

Reviewer 2. Paragraph 67-95. This should be shortened. It shows key ideas in one manuscript for each paragraph. This must be improved. Many studies have been developed.

Answer: this part has been shortened. In particular, the two following sentences have been removed, as they did not address totally the purpose of the paper:

Previously line 72-73: "...and use a leaf area index in order to take into account the varying foliage that can cause obstruction at different times of the year".

Previously line 92-93: "Walch et al. [11] have developed a very elaborate model of solar mapping, that uses machine learning techniques for dealing with uncertainty, and they use conventional shadow casting for direct illumination and for computing the sky view factor."

The first paragraph of this part has been restructured.

Reviewer 2. Limitations of their methodology. I haven't found them. Please add them.

Answer. These limitations are presented in the Conclusion (Section 5, lines 406-413). The perspectives of improvement (lines 414-421) also indirectly address the limitations and the way to overcome them. Therefore, we think there's no need to add anything more.