Spatial Analysis of Urban Historic Landscapes Based on Semiautomatic Point Cloud Classification with RandLA-Net Model—Taking the Ancient City of Fangzhou in Huangling County as an Example
Round 1
Reviewer 1 Report
Comments and Suggestions for Authors
The article represents a valuable contribution to the innovative methods of analyzing the spatial features of historic urban landscape, the symbiosis between built-up structures and landscape, and factors causing its spatial degradation. The methodology uses low-altitude UAV photogrammetry to obtain 3D point data, classified using the RandLA-Net model, enabling quantitative spatial analysis of the urban historical landscape. Spatial characteristics of topography, street space, and architectural structures are examined to quantify the spatial differences between historical and modern structures and identify the disturbances. The method is reproducible and can support decision-making on the preservation and development of urban heritage areas.
The main question addressed by the research is to contribute to methods of analyzing and identifying the spatial disturbances causing the degradation of the values of historical urban environments. The gap in the field of heritage protection is the lack of sufficiently accurate, quantifiable methods, and this research addresses this specific gap.
The methodology proposed by the authors does not aim to cover all the aspects needed in decision-making to preserve the values of historic urban environments. It addresses a specific aspect of quantifying their spatial characteristics. This method enables the gain of relatively accurate quantified data on spatial characteristics of landscape and built-up structures typical for historic environments, and quantitative identification of the characteristics of those structures that cause the spatial disturbance.
The discussion and conclusion sections are consistent with the evidence and arguments presented in the results section, which shows that the main visual conflicts between historical and modern architectural structures are mainly in heights, roof slopes, and the width of streets.
I recommend adding examples of how the results of this methodology could be used in practice by authorities for guidance on the further preservation and development of the area.
The references are appropriate.
The tables and figures need to be edited to be easy to read in the PDF version of the article.
Only minor revisions (adding missing references) are needed.
Author Response
Please see the attachment.
Author Response File: 
Author Response.docx
Reviewer 2 Report
Comments and Suggestions for Authors
This is an interesting article, which does need some major revision to enhance its readability. Here are some items that need to be addressed:
1) You haven't mentioned the sensor (UAV/camera) that was used, nor provided its technical specs. These should be added to the research.
2) Also, the photogrammetric process as a whole was not thoroughly explained. Some of the variables you list affect directly the quality of the point cloud. You must provide a more detailed workflow and also ponder (in the discussion section) if pre-processing could have been improved to help the algorithm enhance its classification.
3) You mention "Experiments show that the model achieves the optimal performance at 50 rounds". What was the early-stop criterion? This section needs more explanations, otherwise it sounds too empirical (more than AI already is).
4) F1, precision, recall, accuracy and other methods should be summarized in a table, for better understanding.
5) Regarding the results, nothing to add to 3.1.1. 3.1.2, however, could benefit from improving the quality of the images, since so much is corroborated by them. The context maps on the left are difficult to understand, and the profiles are not quite clear from the map.
6) The space syntax theory explanations should go to a previous section, not straight as part of the results. When explaining the methodology, the steps done after classifying the point cloud should be listed and explained there. The results must focus mainly on what was found, not introducing a theoretical concept to the reader. In a certain way, most of the text in 3.2 could have been explained earlier, and 3.2 would be mostly a way to show the results applied to that specific context.
7) Section 3.3 contains several broken links. 3.3.2 mentions "point-to-point distances between two building categories and the Xuanyuan Street point cloud layer, generating a distance heatmap illustrating historical buildings’ proximity to modern constructions." How was this done? Was it an algorithm such as Hausdorff distance? Did it measure distances between individual points or individual buildings? The results can be quite misleading, if you can't individualize point coulds as specific buildings.
8) The discussion and the conclusion sections are very short and fail to properly elaborate on the results. THey merely mention what was said before. A more thorough discussion section should address comparative analyses to other scenarios, possibilities for improvement and discuss on the benefit of the proposed study as opposed to what has been done elsewhere.
9) Also, and perhaps the most important point: you have developed this sophisticated CNN to classify point clouds, but these point clouds you classified were later turned into raster formats for some applications. Hence, what is the purpose of working with clouds? Also, wouldn't a proper photogrammetric stereo-restitution (which gathers building footprints and other surfaces as simplified meshes) be more suitable for the kind of spatial analyses you did? It would certainly individualize buildings by their shape and attribute table, and also represent other terrain features. This can also be done from photogrammetric UAV imagery. So, you must make the case for using point clouds as your primary source of data, despite surrounding evidence that 3D CAD/BIM/GIS models are simpler to process and more accurate in terms of spatial analyses.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 3 Report
Comments and Suggestions for Authors
This manuscript focused on historical spatial degradation in Fangzhou Ancient City (China) using modern approaches. While the topic is relevant to current research and practice, the manuscript shows serious pitfalls that need to be addressed. The main concerns of this reviewer regard the findings, discussion of the findings, and conclusions.
In Section 3, the findings are mixed with the method, personal opinions and/or discussion of the results. As an example, please see: Lines 322-323: “We recommend strict height controls for areas exceeding +20 m residuals (9.7% of samples) to preserve traditional "terrain-adaptive construction" principles”; Lines 343-345: “It is recommended to strengthen the principle of terrain adaptation in the subsequent conservation planning, so as to restore the visual order of "loess-urban silhouette".”; etc. I would suggest the Authors clearly describe the findings avoiding personal opinions and reporting the discussion in the proper section.
Example of findings mixed with method: Lines 374-376: “Subsequently, high-precision polylines are generated along road boundaries via 2D polygon fitting, constructing a topological street network boundary model [27]”; Lines 377-379: “Further, centerlines of the street network are extracted through polyline axial fitting in AutoCAD, generating vectorized midline data to resolve spatial syntax’s dependency on linear topological inputs”; etc. I would suggest the Authors clearly describe software and methodologies in the appropriate section of the manuscript.
In addition, what is the meaning of the quotation marks “” in “making the best use of the situation”, “city-adapted-to-plateau”, etc.? Please consider removing the unnecessary “” as they cause confusion if not used properly (e.g., in citations). Furthermore, please define each acronym included in the manuscript (e.g., what does KNN mean?) and fix the references: several references are missing (see the message “[Error! Reference source not found]”).
Figure 12, 13, 16, 17, and 18 are unclear.
In Section 4, the discussion of the findings is missing. The Authors did not discuss the results in light of the findings of previous international research and/or strategies and/or practices. I would suggest the Authors focus on the critical discussion of the results and the implications of the findings. Part of the discussion is currently in Section 5. Conclusions.
Section 5 needs improvement. It should be a summary of the key findings but also include: implications to theory and practice; limitations of the research; future research recommendations.
In my opinion, Sections 3, 4, and 5 need to be rewritten.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 2 Report
Comments and Suggestions for Authors
The article has significantly improved and, I believe the authors have properly addressed all points I've raised. Congratulations!
Author Response
Thank you for your valuable comments, which contributed greatly to improving the readability and scientific quality of the revised manuscript.
Reviewer 3 Report
Comments and Suggestions for Authors
The Authors have worked hard on the manuscript, which has been significantly improved. This reviewer asks for minor revisions. Please see the detailed report below.
Lines 204-206: “Its 4K imaging resolution and ±0.1m hovering accuracy meet scientific aerial photography requirements”. Please add references.
Acronyms: please define the full set of acronyms (e.g., PSO, VR/AR, …).
Lines 706-709: “In Lyon, 706 France, the 3D modeling of historic urban districts underscores the principle of preserving digital traces for heritage, positioning digital representations of urban spatial structures as decision-making references and communication tools”. Please add references.
Lines 709-713: “The integration of point cloud technology with BIM for modeling has also gained international prominence. In a collaborative project between Autodesk and Brazil’s Ibirapuera Museum, laser-scanned buildings and surrounding parks were used to construct comprehensive BIM models for heritage conservation, operational management, and expansion planning”. Please add references.
Author Response
Please see the attachment.
Author Response File: Author Response.docx
