A Geospatial Framework for Spatiotemporal Crash Hotspot Detection Using Space–Time Cube Modeling and Emerging Pattern Analysis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Overall, this seems to be part of a technical report or the paper was written very quickly, which impacted the quality of the paper. Although I acknowledge the effort put to review and filter the crashes, the paper can not be published in its current state.

The objectives stated in the paper look as steps or part of the methodology followed. The research gap is not clear. The contribution of the paper is not mentioned clearly. One reason of these issues is the absence of a proper literature review. This area of research has been discussed in tens of research and the paper only discussed 5 relevant previous research, which is not acceptable. 

The method followed in the paper are concentrated on patter recognition in the data. 

One example of quickly jumping to conclusions without proper evidence is the following statement in Line 213 "This crash distribution highlights the importance of spatial investigation in prioritizing safety interventions." It is not clear how the paper presented "safety interventions" prioritization in response to the spatial investigation. Simply presenting crash stats with the corresponding location is not spatial investigation. 

The discussion is only one paragraph!!

Comments on the Quality of English Language

The entire paper needs to be proof read. There is lots of redundant sentences such as repeating the use of ArcGIS pro 2.5 and the study period between 2014 - 2017 so many times.

The GIS abbreviation was spelled out after the mention of GIS multiple time in the manuscript and it was mentioned twice in the same page "Geographical information system (GIS)"

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1. The number of references cited in this paper is insufficient. Only six articles are cited in Section 2, which is insufficient to achieve the purpose of a literature review.
2. The third section should highlight the generalizability of the proposed method. It should have the potential to be extended from the I-75 case to other cities or countries.
3. The authors should explain the applicability of the research findings to today’s urban environments, given that the study is based on data from nearly ten years ago.
4. Most of the figures are of low resolution.
5. Figure 12 needs to be redesigned. The current image is confusing: the map is not visible, and the meaning of the dense bar charts is unclear.
6. The discussion section is very brief. Either expand the content and length of the discussion, or consider merging it into the Results section.
7. The conclusion section is generally not presented in the form of a table. It is recommended that the authors refer to widely accepted formats for writing the conclusion section.
8. The formatting of the reference list needs to be standardized. For references 1 and 2, the responsible institution or publisher is not indicated, making it difficult for readers to access the corresponding references.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This manuscript presents a timely and well-structured geospatial framework for crash hotspot detection using space-time cube (STC) modeling and emerging pattern analysis. The integration of GIS-based tools and spatiotemporal analytics is commendable and offers practical insights for transportation safety planning. However, the following issues should be addressed to improve clarity, transparency, and scientific rigor:

1. Improve Clarity in Table 1

The phrasing in the third column of Table 1 is occasionally vague and inconsistent (e.g., “Crash data, including locations, dates, and times, monthly”). To enhance clarity and readability, the descriptions of required data should be revised using more precise, structured language. For example:

• “Crash data, including locations and times” → “Monthly crash records with geolocation (latitude, longitude), timestamp, and roadway ID”

• “Crash locations, temporal and spatial data” → “Crash records with spatial coordinates and temporal attributes”

Such improvements will make the table more useful for readers aiming to reproduce or adapt the framework.

2. Over-Reliance on GIS Software Without Mathematical Detail

The methodology presented relies heavily on ArcGIS Pro’s built-in tools for visualisation and analysis, including Space-Time Cube generation, Moran’s I calculation, and Emerging Hotspot Analysis. While these tools are powerful, the manuscript does not provide sufficient mathematical or algorithmic descriptions of the core methods employed. This limits both transparency and reproducibility, particularly for researchers using alternative platforms.

Recommendations:

• Introduce mathematical formulations for key components:

  • The Moran’s I statistic for spatial autocorrelation

  • The Getis-Ord Gi* statistic used in hotspot detection

  • The spatial-temporal binning structure of the space-time cube (STC)

• Explain how parameters such as spatial bin size, temporal resolution, and spatial weight matrices are defined and influence the analysis.

• If proprietary tools are essential, consider offering open-source analogs or outlining the algorithmic workflow in pseudocode.

Inclusion of these mathematical foundations will significantly improve the scientific robustness of the study and ensure its broader applicability beyond the ArcGIS ecosystem

Comments on the Quality of English Language

Minor revisiosns:

Line 16

Replace “interpreting” with “analysing” for formal tone.

Line 56

“Major corridors, such as Interstate 75 (I-75)” – already stated earlier. Trim repetition.

Line 89–109

Combine bullet points where overlapping; e.g., points 1 and 3 can be merged.

Table 1

Align row formatting and ensure terminology is consistent (e.g., “Spatiotemporal analysis” vs. “Space-time analysis”).

Figure Captions

Follow IEEE or MDPI figure caption style (e.g., "Figure X. [Description]").

Figure 5

Typo in caption: “from 2014 to 207” → “from 2014 to 2017”

Table 2

“Charotte” should be corrected to “Charlotte”

Table 4

“Moran' I” → “Moran’s I”

Figures 10–12

Consider increasing label sizes and adding legends if not already present.

Reference [9]

Include access date in full format: “Accessed on July 22, 2025.”

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

1. For readers to quickly catch your contribution, please include all of the abbreviations.
2. Does the abstract sufficiently quantify the study’s contributions and novelty compared to existing research?
3. Can the authors clarify how this study advances prior research that also uses GIS-based crash analysis?
4. The literature review section, in its current form, is relatively weak and should be strengthened with more details and justifications. You may add a comparative table to the previous research.
5. How does this study’s approach compare with machine learning models used for crash prediction in other recent works?
6. Were there any limitations in the crash data collected from SSOGIS (e.g., underreporting, missing variables)?
7. Did the authors perform any sensitivity analysis to determine how hotspot classification changes with bin size or spatial resolution?
8. How were roadway design characteristics or environmental conditions considered or excluded in hotspot analysis?
9. Could the authors provide more statistical support (e.g., confidence intervals or effect sizes) for the trend classifications?
10. How does the spatial distribution of hotspots correlate with land use or population density?
11. What are the implications of newly emerging hotspots—were any of these areas previously identified as low risk?
12. How adaptable is the proposed framework for real-time crash monitoring or integration with smart traffic systems?
13. Can the model be directly applied to other corridors or regions without significant retraining or calibration?
14. What are the authors specific plans for expanding the model to include crash severity, weather conditions, or driver behavior?
15. How will the incorporation of real-time data streams (e.g., from IoT or traffic sensors) enhance the existing model?

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

THe authors did a good job improving the paper. However, I am still can not see a significant contribution of the paper. The literature already has well-established methods that identify hot spots using GIS. There are two main questions here:

• Why the interstate collision investigation is so important? Can a regional model be used instead?
• why the proposed method can be better than the other methods used in the literature? Even if these methods are not developed based on inter-state highways. Can previously development methods be transferred to the inter-state condition? 

There are other studies that were not reviewed in the literature such as:

• Predictive Models and GIS for Road Safety: Application to a Segment of the Chone–Flavio Alfaro Road.
• A new integrated GIS-based analysis to detect hotspots: A case study of the city of Sherbrooke.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Some figures are still in low resolution. Authors should fix them during proof.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report

Comments and Suggestions for Authors

Thanks!

Author Response

No changes were requested.

The authors wish to thank the reviewer for his/her time and contribution towards improving our manuscript.  We appreciate it.

 

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

.