Submit to this Journal Review for this Journal Propose a Special Issue

Article Menu

Share Help Cite Discuss in SciProfiles

Open AccessArticle

Peer-Review Record

Identifying Urban Road Black Spots with a Novel Method Based on the Firefly Clustering Algorithm and a Geographic Information System

Sustainability 2020, 12(5), 2091; https://doi.org/10.3390/su12052091

by Tengfei Yuan^1,*

, Xiaoqing Zeng^1,* and Tongguang Shi²

Reviewer 1: Anonymous

Reviewer 2: Anonymous

Reviewer 3: Anonymous

Sustainability 2020, 12(5), 2091; https://doi.org/10.3390/su12052091

Submission received: 30 December 2019 / Revised: 29 February 2020 / Accepted: 29 February 2020 / Published: 9 March 2020

Round 1

Reviewer 1 Report

The manuscript "Identify the Urban Road Black Spots with a Novel Method Based on Firefly Clustering Algorithm and GIS" is well-written study. The authors put a lot of effort into describing research design and the method Firefly Clustering Algorithm for identifying black spots on urban roads. The study can be labeled as methodological, therefore, link with the sustainable development in the Introduction is short, general and vague, and the Results are in place only to compare methods - no intention to present the frequency of black spots in urban areas etc. I would recommend that you use the Conclusion to make a stronger connection with sustainable development issues - to raise the reader's interest.

Minor English language and style comments:

p.2, line 38 - change into 'socio-economic'

p.3, line 66 - 'black identification method'?

p.4, line 84 - place the reference [16] at the end of the section you summarized.

p.5, line 108 - 'Figure.4' correct the style

p.15-17 - references should be organized according to instructions.

Author Response

Response to Reviewer 1 Comments

Minor English language and style comments:

p.2, line 38 - change into 'socio-economic'

p.3, line 66 - 'black identification method'?

p.4, line 84 - place the reference [16] at the end of the section you summarized.

p.5, line 108 - 'Figure.4' correct the style

p.15-17 - references should be organized according to instructions.

Thanks for your advice.

Response 1: I have modified the language and style according to the comments.

Response 2: P.2, line 38 - ‘economic’ has been changed into 'socio-economic', 'socio-economic' has a stronger connection with sustainable development;

Response 3: P.3, line 66 - 'black spots identification method' has been corrected;

Response 4: P.4, line 84 - the reference [16] has been placed at the end of the section.

Response 5: P.5, line 108 - the style of 'Figure.4' has been correct. 5. The references have been organized according to instructions, as shown below.

Blincoe, L.; Miller, T.R.; Zaloshnja, E. The economic and societal impact of motor vehicle crashes (Revised). Annals of Emergency Medicine. 2015, 66.2, 194-196. Toroyan, T. Global status report on road safety. Injury Prevention. 2009, 15(4):286. Sorensen, M.; Elvik, R. Black spot management and safety analysis of road networks. Institute of transport economics. 2007. Östen, J.; Wanvik, P.O.; Elvik, R. A new method for assessing the risk of accident associated with darkness. Accident Analysis and Prevention. 2009, 41(4), 809-15. Erdogan, S. Explorative spatial analysis of traffic accident statistics and road mortality among the provinces of Turkey. Journal of Safety Research. 2009, 40(5), 341-51.

Response 6: Besides, in order to make a stronger connection with the sustainable development issues, I have added the sentence in the conclusion, “The proposed firefly clustering algorithm based on OD cost distance can not only effectively overcome this shortcoming, but also accurately identify the black spots and purposefully provide decision for solving urban road safety problems, which is helpful for decreasing socio-economic losses and promoting the sustainable development of the city.”

Author Response File: Author Response.docx

Reviewer 2 Report

The purpose of this research is to illustrate how the firefly clustering algorithm and GIS can be used to identify the black spots in urban roads. It finds that the clustering results show that the number of identified accident point using Euclidean distance is more than the method with OD cost distance. This paper requires further revisions. There are no problem with the methods. In addition to problems with the writing, the motivation for the research is weak. The literature in not well organized. The results could be more clearly presented. The discussion section continues to provide results. This section does not connect with the literature or the motivation for the research.

Author Response

Response to Reviewer 2 Comments

In addition to problems with the writing, the motivation for the research is weak. The literature in not well organized. The results could be more clearly presented. The discussion section continues to provide results. This section does not connect with the literature or the motivation for the research.

Thanks for your advice.

Response 1: The motivation of this research is that accurately identify the black spots, which will purposefully provide decision for solving urban road safety problems. This research will be helpful for decreasing socio-economic losses and promoting the sustainable development of the city. In order to improve the writing and emphasize the motivation, I have modified the paper seriously.

Response 2: The references have been organized according to instructions, as shown below.

Response 3: The results has been presented clearly by updating the Figure 9-11. Firstly, the results show that the intersection A, B, C are the black spots identified by the proposed method and definition. Secondly, the identification results are different by using firefly clustering algorithm with OD cost distance and Euclidean distance.

The urban road black spot is regarded as: for a road section within 500 meters or an intersection within 150 meters, there has to be a history of at least 3 casualty crashes in any one year, which means normal number of accident is 3 in 500 meters road section or 150 meters intersection at a year.

In order to connect with the motivation for the research, a paragraph has been added in the discussion section, as shown below.

The results show that the novel method combined firefly clustering algorithm and OD cost distance can identify the urban black spots and evaluate the situation of black sopts accurately. However, the method based on Euclidean distance can overestimate the number of black spots especially in the intersections. Therefore, the proposed method based on firefly clustering algorithm and GIS can not only contribute to identify urban road black spots, but also play an auxiliary role in evaluating the situation of black sopts, which will help reduce the urban road crashes and maintain the urban sustainable development.

Author Response File: Author Response.docx

Reviewer 3 Report

The article is on an interesting topic such as the hot-spot identification. It is generally well structured and written, even if it could be improved in some parts. General comment on the methodology and related results/discussion. A comparison is made between different techniques to identify black spots (i.e., different distances/algorithms). The conclusion is that "the accident search method based on Euclidean distance can overestimate the number of black spots". However, this is not explicitly coherent with the discussion section. For instance, in Figures 10 and 11 it is evidently displayed that there are 3 black spots in both cases. However, the number of crashes for each black spot is different. I am not convinced of concluding that the OD cost distance is better by only looking at these spatial plots. To understand which method is better, I will suggest to study in deep the single crashes which have occurred to the sites A, B, C (as based on the example, or anyway in a similar fashion). In this way, for example, it could be excluded that the crashes 4, 5 and 6 for site A, 6 for site B, 5, 6 and 7 for site C are relevant to the intersection and so that the OD cost distance method is more accurate in identifying the boundary of the black spot. Otherwise, the suggested conclusions are in my opinion not evidently inferrable from the study. Other comments: a) In the abstract, a measure to reduce the number of urban road crashes is mentioned twice. However, this study suggest a measure to identify black spots, not to provide countermeasures. b) lines 46-47 "In the urban road circumstance, there are some traffic accidents occurring on similar road sections or intersections during a period of time, which are called black spot". This definition is inaccurate. c) I will suggest to refer to other comprehensive sources to update the taxonomy of the possible measures to identify black spots summarized in Table 1 (e.g. the Highway Safety Manual). Moreover, I will change accident "number" with accident "frequency". The matrix analysis method should be better defined. d) Lines 96-98. So, which guidelines were followed to define black spots? Otherwise, it seems that an arbitrary threshold was chosen. e) Please clarify sentences at lines 107-112. f) I will reword the sentence at line 116 ("As known to all..") g) Figure 6. Maybe something is missing in the box of the outputs. (Output the black spot... should be reworded) h) Why severity is abruptly mentioned at line 250? Is it the right word? i) Flyover crossings are in the conclusions. However, the matter of flyover crossings is not specifically discussed in the results/discussion. j) Please check format of references.

Author Response

Response to Reviewer 3 Comments

The conclusion is that "the accident search method based on Euclidean distance can overestimate the number of black spots". However, this is not explicitly coherent with the discussion section. For instance, in Figures 10 and 11 it is evidently displayed that there are 3 black spots in both cases. However, the number of crashes for each black spot is different. I am not convinced of concluding that the OD cost distance is better by only looking at these spatial plots. To understand which method is better, I will suggest to study in deep the single crashes which have occurred to the sites A, B, C (as based on the example, or anyway in a similar fashion). In this way, for example, it could be excluded that the crashes 4, 5 and 6 for site A, 6 for site B, 5, 6 and 7 for site C are relevant to the intersection and so that the OD cost distance method is more accurate in identifying the boundary of the black spot. Otherwise, the suggested conclusions are in my opinion not evidently inferable from the study.

Other comments:

a) In the abstract, a measure to reduce the number of urban road crashes is mentioned twice. However, this study suggest a measure to identify black spots, not to provide countermeasures. b) lines 46-47 "In the urban road circumstance, there are some traffic accidents occurring on similar road sections or intersections during a period of time, which are called black spot". This definition is inaccurate. c) I will suggest to refer to other comprehensive sources to update the taxonomy of the possible measures to identify black spots summarized in Table 1 (e.g. the Highway Safety Manual). Moreover, I will change accident "number" with accident "frequency". The matrix analysis method should be better defined. d) Lines 96-98. So, which guidelines were followed to define black spots? Otherwise, it seems that an arbitrary threshold was chosen. e) Please clarify sentences at lines 107-112. f) I will reword the sentence at line 116 ("As known to all.”) g) Figure 6. Maybe something is missing in the box of the outputs. (Output the black spot... should be reworded) h) Why severity is abruptly mentioned at line 250? Is it the right word? i) Flyover crossings are in the conclusions. However, the matter of flyover crossings is not specifically discussed in the results/discussion. j) Please check format of references.

Reply: Thanks for your advice.

Response 1: To understand which method is better, a paragraph has been added in the discussion section, as shown below.

The number and location are key parameters of black spots. So the novel method combined firefly clustering algorithm and OD cost distance can void the phenomenon that classify some unrelated accident points as the same black spots. In this way, we can deal with black spots in different way according to the situation of black spots.

Response 2: The main purpose of this research is to improve the accuracy of black spot identification, and this identification method will be helpful to reduce the number of urban road crashes. Due to the countermeasures isn’t included in this research, so the abstract has been modified by this comment.

Abstract: With the rapid development of urban road traffic, there is a certain number of black spots in urban road network. Therefore, it is important to create a method to identify the urban rood black spots effectively, quickly and accurately to ensure the safety of resident and maintain the sustainable development of city.

Response 3: Lines 46-47, this is common definition, so this research has researched the definition of black spot in section 2.1. Finally the urban road black spot is regarded as: for a road section within 500 meters or an intersection within 150 meters, there has to be a history of at least 3 casualty crashes in any one year, which means normal number of accident is 3 in 500 meters road section or 150 meters intersection at a year.

Response 4: The taxonomy of the possible measures to identify black spots summarized in Table 1 has been updated. Besides I have changed accident "number" with accident "frequency". And the matrix analysis method has been be better defined.

Table 1. Summary of Black Spot Identification Method

Method	Principle	Advantages	Disadvantages	Suitable conditions
Accident frequency method	Identify and sort the accident according to the frequency of accident	It considers the length and traffic of road section.	It doesn’t consider the regression effect of accidents	It is suitable for the road section or intersection which condition is similar and traffic is not heavy [4].
Matrix analysis method	Identify the accident according to the accident number and frequency	Its evaluation result is accurate and flexible	Its identification criteria is subjective.	It is suitable for the road section or intersection which condition is similar and traffic is not heavy [5].
Accident rate method	Identify the accident according to the Accident rate	It considers many factors about accidents.	It needs a lot of accident data and neglects randomness of accident.	It is suitable for describing the regional accident situation [6].
Equivalent accidents number method	Identify the accident according to the equivalent accidents number	It considers many factors about accidents.	It needs a lot of accident data and it's difficult to determine the weight value.	It is suitable for urban roads or highways with the similar conditions [7].
Quality control method	Identify the accident according to the set threshold.	It considers the traffic conditions and its evaluation result is accurate	It requires a lot of traffic data and classification work.	It applies to the road section with low traffic flow [8].
Cumulative frequency method	Identify the accident according to accident number and accident rate per kilometer	It uses a lot of basic traffic data.	It does not take into account the situation of the accident.	It applies to roads with widely varying accident conditions [9].
Regression analysis method	It considers a lot of factors of accident.	It considers different factors of accident.	There are high requirements for the model parameters and basic data.	It applies to the regional accident quantification [10].
Fuzzy evaluation method	It considers a lot of factors of accident.	Its mathematical model is simple and suitable for the multi-level problems.	Its index weight is subjective.	It's widely used in many conditions [11].
Expert experience method	Identify the accident according to accident number.	It can estimate the result quickly and easily.	It is too subjective.	It applies to roads that lack basic data [12].
BP neural network	It considers a lot of factors of accident.	It can evaluate the accident comprehensively	Its indicator is not directly related to the accident.	It applies to the highway [13].

Response 5: Lines 96-98. Combined with previous definition research, this research mainly refers the rules of black spots identification promulgated by China in 2001. Ultimately the urban road black spot is regarded as: for a road section within 500 meters or an intersection within 150 meters, there has to be a history of at least 3 casualty crashes in any one year, which means normal number of accident is 3 in 500 meters road section or 150 meters intersection at a year. In addition, previous definition research mainly contain American Transportation engineering handbook, and Road safety engineering guide.

Response 5: Lines 107-112. The main purpose of this paragraph is to illustrate that distance-measure can impacts on the identification of black spot. Meanwhile, this paragraph is to prove that Euclidean distance will overestimate the accident situation.

Response 7: Line 116. I have modified "As known to all,” to “As is known to all,”.

Response 8: I have modified Figure 6, as shown below. “Output the black spot of ...” has been reworded to “Output the accident points of ...”

Response 9: Line 250. I have modified “severity” to “situation”.

Response 10: Due to the matter of flyover crossings is not specifically verified in the results/discussion, so I have deleted the flyover crossings. This research mainly aims at black spots identification in urban road, especially at intersection.

Response 11: The references have been organized according to instructions, as shown below.

Author Response File: Author Response.docx

Round 2

Reviewer 2 Report

The discussion section of this manuscript is much improved.

The manuscript still needs some revision and editing. I will provide some examples.

The phrase "urban road circumstance" is frequently used. Please define this. I am not clear if this is referencing a physical/environmental or human situation. An example would be helpful.

Here is an example of a revision that would benefit the readability of the manuscript. This is from line 49-52 and is my attempt to simplify the text.

Priority should be give to remediation of the black spots that would reduce traffic accidents by improving road safety, and promote greater social and economic benefits. Therefore, the focus of this study is how to better identify black spots.

Line 117: "As is known to all, it happens randomly for a single traffic accident." The meaning of this sentence is not clear to me.

These example are not meant to be all the revisions necessary. I would suggest sending this manuscript to someone who can improve its readability.

Author Response

Response to Reviewer 2 Comments

The discussion section of this manuscript is much improved.

The manuscript still needs some revision and editing. I will provide some examples.

The phrase "urban road circumstance" is frequently used. Please define this. I am not clear if this is referencing a physical/environmental or human situation. An example would be helpful.

Here is an example of a revision that would benefit the readability of the manuscript. This is from line 49-52 and is my attempt to simplify the text.

Priority should be given to remediation of the black spots that would reduce traffic accidents by improving road safety, and promote greater social and economic benefits. Therefore, the focus of this study is how to better identify black spots.

Line 117: "As is known to all, it happens randomly for a single traffic accident." The meaning of this sentence is not clear to me.

These example are not meant to be all the revisions necessary. I would suggest sending this manuscript to someone who can improve its readability.

Thanks for your advice.

Response 1: The urban road circumstance has been defined. There are many similar intersections and road sections, as well as the traffic conditions in the urban road circumstance, so there are some traffic accidents occurring on similar road sections or intersections during a period of time, which are called black spot.

Response 2: Line 49-52 has been simplified. Identifying the black spots should be paid attention which will help reduce traffic accidents, improve road safety, and promote greater socio-economic benefits. Therefore, the focus of this study is how to better identify black spots.

Line 43-45 has been simplified. Meanwhile, urban road traffic accidents account for a high proportion in the road traffic accidents. They not only cause incalculable economic losses to the society, but also have a serious negative impact on urban sustainable development.

Line 77-80 has been simplified. The purpose of this research is to illustrate how the firefly clustering algorithm and GIS can be used to identify the black spots in urban roads. The method is expected to provide a reference guide for the mitigation of accident in complex urban road circumstance.

Line 119-121 has been simplified. But when several accidents occur continuously in one place of the urban road within a certain period, they must be attracted by some external factors. So this phenomenon of aggregation is very similar to the firefly clustering phenomenon.

Line 195-198 has been simplified. Therefore, this research adopts the OD cost distance to indicate the shortest distance between the accident points. Because OD cost distance not only means the least-cost or shortest path from a chosen destination to the source point, but also signifies additional factors beyond the cost surface to account for actual travel distance over the terrain.

Response 3: Line 117 has been modified. According to the distribution characteristics of traffic accident point, it happens randomly for a single traffic accident.

Author Response File: Author Response.docx

Reviewer 3 Report

The article has slightly been improved with respect to the previous version. However, I still think that there are problems to fix, specifically regarding the first comment from my previous report.

I will try to explain my point in other words.

Method 1 (Euclidean distance): it highlights a given number of crashes in hotspot locations

Method 2 (proposed): it highlights less crashes in hotspot locations.

The conclusion is that the Euclidean distance method overestimates crashes at hotspots locations.

I definitely think that it is not possible to say "overestimates" if one do not really know if the crash can be really associated or not to the intersection as based (for example) on crash reports. If crashes are wrongly associated to intersections trough the application of the Euclidean distance method, but their patterns and dynamics are more segment-related, then it could be possible to say, in my opinion, that the Euclidean distance method "overestimates" crashes in hotspots. Otherwise, this is not evident.

Other minor comments:

I personally don't like the form "As all know" or similar versions. It would be better to say "It is well known.."

I don't understand the concept of "situation" instead of severity. Is it the right word?

Author Response

Response to Reviewer 3 Comments

The article has slightly been improved with respect to the previous version. However, I still think that there are problems to fix, specifically regarding the first comment from my previous report.

I will try to explain my point in other words.

Method 1 (Euclidean distance): it highlights a given number of crashes in hotspot locations

Method 2 (proposed): it highlights less crashes in hotspot locations.

The conclusion is that the Euclidean distance method overestimates crashes at hotspots locations.

I definitely think that it is not possible to say "overestimates" if one do not really know if the crash can be really associated or not to the intersection as based (for example) on crash reports. If crashes are wrongly associated to intersections through the application of the Euclidean distance method, but their patterns and dynamics are more segment-related, then it could be possible to say, in my opinion, that the Euclidean distance method "overestimates" crashes in hotspots. Otherwise, this is not evident.

Other minor comments:

I personally don't like the form "As all know" or similar versions. It would be better to say "It is well known.."

I don't understand the concept of "situation" instead of severity. Is it the right word?

Thanks for your advice.

Response 1: If the crash cannot be really associated with the intersection, the Euclidean distance method will overestimate crashes at hotspots locations. But the proposed method can avoid this phenomenon, so I have modified discussion according to advice.

Due to the Euclidean distance method can overestimate the number of accident points, so this research should discuss whether the identified accident points can be really associated with the intersection by comparing the accident reports. As described in the accident reports, some accident points’ location record are not intersection, such as the accident point 1, 5, 6 of black spot A, accident point 6 of black spot B and accident point 1, 5, 7 of black spot C. Therefore, the Firefly Clustering Algorithm and Euclidean distance method can overestimate the number of accident points. But the Firefly Clustering Algorithm and OD cost distance method can identify the black spot effectively.

Table 1. Accident points’ location record of Black spot A

Euclidean distance	OD cost distance	Location Record
1	—	Chunxuan Road section
2	1	Chunxuan and Feiyue Intersection
3	2	Chunxuan and Feiyue Intersection
4	3	Chunxuan and Feiyue Intersection
5	—	Chunxuan Road section
6	—	Chunxuan Road section

Table 2. Accident points’ location record of Black spot B

Euclidean distance	OD cost distance	Location Record
1	1	Chunbo and Feiyue Intersection
2	2	Chunbo and Feiyue Intersection
3	3	Chunbo and Feiyue Intersection
4	4	Chunbo and Feiyue Intersection
5	5	Chunbo and Feiyue Intersection
6	—	Feiyue Road section

Table 3. Accident points’ location record of Black spot C

Euclidean distance	OD cost distance	Location Record
1	—	Shiji Road section
2	1	Chunhui and Shiji Intersection
3	2	Chunhui and Shiji Intersection
4	3	Chunhui and Shiji Intersection
5	—	Chunhui Road section
6	4	Chunhui and Shiji Intersection
7	—	Shiji Road section

Response 2: "As is known to all, it happens randomly for a single traffic accident." The meaning of this sentence is not clear to me. So I have modified it to “According to the distribution characteristics of traffic accident point, it happens randomly for a single traffic accident.”

Response 3: I have modified “situation” to “condition”. The condition of accident mainly indicate the number of accident in this research.

Author Response File: Author Response.docx

Article Menu

Identifying Urban Road Black Spots with a Novel Method Based on the Firefly Clustering Algorithm and a Geographic Information System

Further Information

Guidelines

MDPI Initiatives

Follow MDPI