Enhancing Urban Land Use Identification Using Urban Morphology

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The article discusses issues regarding the classification of urban land, which are important in the aspect of responsible space management and future spatial planning. Proposed method of land classification is a kind of innovative and combines some particular features of land use and land morphology not used together so far. The paper is very intersting, but in many fields too superficially covers particular aspects of proposed method, especially in chapter "3. Methodology":

1. "The reference land area for different types of POIs is determined based on the study area to ensure the accuracy of the preliminary identification results" - how?

2. I assume that FAR value is calculated by taking some assumptions ("a theoretical floor 377 height of 3m was assumed") and it is not "a real" gross floor area to size of land, but calculated floor area to size of block ratio? It should be clearly stated in this text.

3. "The thodL can be dynamically adjusted to suit specific conditions in different cities or regions" - how? Isn't it mainly affect the identifications results? Isn't it important? It aslso refers to thodNUM and thodH. It is partially analyzed in section 5.1 but it should have some initial assumptions.

4. "w1 and w2 represent the respective weight values, and it is assumed that w1 + w2 = 1" - how are the weights evaluated?

5. I think that section 3.3 should be reorganized and figure 3 should be placed after definitions of VI and HI to better recognition of the algorithm. The Authors should also inform what kind of komputer software were used to this anlysis.

6. Also in figures 6 and 7 identification results show that many blocks as classified as "mixed-use". Isn't it is less precise or is it better to classify as mixed-use? Maybe, instead of actual mixed-use identification, the algorith should try to classify the block based on secondary land use type retrived from secondary POIs proportion?

7. As to redaction issues, I think that Figure 11 should be divided into 4 separate "bigger and clearly readable" maps with it's own legends as it shows a whole region of land identification.

Best regards.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This paper talks about a new method, called the Urban MorpHology based Land Use Classification (UMH-LUC) model, to better figure out what different parts of a city are used for, like homes or shops, by looking at the city's shape and size from above and using special data points. By testing this method in the Pearl River Delta, they found it works really well, especially for finding where people live, mixed-use areas, and lands on the edge of the city. The UMH-LUC model achieved an 81.7% accuracy in identifying urban land uses and a Kappa coefficient of 77.6%, which is an 11.9% improvement over methods that don't use urban morphology for identification.

 

The study results rely heavily on the quality and availability of data, which can affect the accuracy of land use identification, making it challenging to apply in other areas with limited data sources. These validation process involves random sampling from the identified blocks, and a confusion matrix is used to assess several key indicators, including Producer's Accuracy, User's Accuracy, Overall Accuracy, and Kappa coefficient.

 

There is a subjective element in verifying the accuracy of the model's results, which means that different exercise might interpret the results in slightly different ways, adding a layer of uncertainty. Setting up the model requires choosing specific settings for its parameters, which can be time-consuming and might not always lead to the best results for every situation. Additionally, the method used is already obsolete. Various authors have developed more robust methods based on the work "Death to Kappa: birth of quantity disagreement and allocation disagreement for accuracy assessment" by Pontius & Millones (2011).

 

In summary, the method is novel and well founded. The research design uses a combination of Points of Interest (POI) data and urban morphology metrics such as floor area ratio (FAR) and building footprint dissimilarity to identify urban land use, which is a novel approach that leverages both social sensing data and urban physical characteristics to improve accuracy. By incorporating a dynamic reclassification process based on FAR measurements and area and perimeter metrics, the design allows for refinement of initial classification results, addressing the limitations of relying solely on POI data.

 

However, the validation of the model presents methodological deficiencies that must be strengthened. These observations may change the results, but the paper will be more robust in the use of different data sets and indicator systems designed specifically for marginal and central cities to improve the accuracy of land use identification.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

First of all thank to the Authors for full response to my remarks and adding a lot of major explanatory paragraphs to the text. It was hard to evaluate and reconsider because of the complexity of the assessed article. But now the article is more clear to understand.

As I see (and was almost sure), some key parameters of the model are once difficult to set and twice have a crutial impact on the final results. Maybe you should try to automatically adjust this parameters based on some sample data from analysed area.

In my opinion it could be divided into at least two separate articles: first describing general method, source data (how to obtain them and prepare them for analyses) and tools (methods, formulas, some assumptions, special software) for completing the work, with final results; and the second one - containing a deeper look at issues related to the selection of weights, used coefficients and determination of land use classes and related problems in the appropriate setting of model parameters. But, as I wrote abowe, the article is as complete as it could be in that form and volume, so I put it to Editors decision.

As to some final editional remarks I suggest to include all "specifications and standards" with web adresses which you provide in Author Response into "References", as all of it contain some source informations on which the calculations are based. Of course all these sources should be cited in the main text at the appropriate place.

Best regards.

Author Response

First of all thank to the Authors for full response to my remarks and adding a lot of major explanatory paragraphs to the text. It was hard to evaluate and reconsider because of the complexity of the assessed article. But now the article is more clear to understand.

Response: Thank you for your suggestions regarding the article. After making targeted revisions based on your feedback, the exposition of the principles in the article becomes clearer, and the readability is also improved.

As I see (and was almost sure), some key parameters of the model are once difficult to set and twice have a crutial impact on the final results. Maybe you should try to automatically adjust this parameters based on some sample data from analysed area.

Response: Regarding your suggestion to automate the adjustment of recognition parameters based on example data, we have indeed sought the optimal parameters through typical case studies in some research areas. However, the process is still not fully automated; the presence of human factors means that more prior knowledge is required, increasing both the time cost and the difficulty of disseminating the method. Going forward, our focus will be on developing a more automated approach.

In my opinion it could be divided into at least two separate articles: first describing general method, source data (how to obtain them and prepare them for analyses) and tools (methods, formulas, some assumptions, special software) for completing the work, with final results; and the second one - containing a deeper look at issues related to the selection of weights, used coefficients and determination of land use classes and related problems in the appropriate setting of model parameters. But, as I wrote abowe, the article is as complete as it could be in that form and volume, so I put it to Editors decision.

Response: Your suggestion to divide the article into two separate papers is an intriguing idea. Indeed, while writing this paper, we also realized that the content was too voluminous, requiring significant space just to clarify the principles and perform accuracy validation. Consequently, the initial discussion on parameter settings was not very extensive or in-depth. Even now, there are many aspects that could be further explored and merit deeper investigation. Initially, our paper also included a module on using this method to identify urban villages, which was removed due to word count limitations. This indirectly confirms that your suggestion is constructive.

 As to some final editional remarks I suggest to include all "specifications and standards" with web adresses which you provide in Author Response into "References", as all of it contain some source informations on which the calculations are based. Of course all these sources should be cited in the main text at the appropriate place. Best regards.

Response: We have cited the 'norms and standards' referenced in this paper in references 39、57 and 63.

References

39. Zhang X; Du S; Zhou Y; Xu Y. Extracting physical urban areas of 81 major Chinese cities from high-resolution land uses. Cities. 2022, 131, 104061. [10.1016/j.cities.2022.104061]

57. China MOHURD. Code for classification of urban land use and planning standards of development land. 2012. Available at: https://www.planning.org.cn/law/uploads/2013/1383993139.pdf.

63. China MOHURD. The ministry of housing and urban-rural development has issued the national standards announcement of the general code for civil buildings. 2022. Available at: https://www.mohurd.gov.cn/gongkai/zhengce/zhengcefilelib/202208/20220824_767703.html.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for taking the time to read the reference included in the review and so quickly incorporate a greater number of more robust metrics for your work. I would only include in the conclusions the importance of this type of more robust validation and how it helps to highlight the quality of their work.

Author Response

Thank you for taking the time to read the reference included in the review and so quickly incorporate a greater number of more robust metrics for your work. I would only include in the conclusions the importance of this type of more robust validation and how it helps to highlight the quality of their work.

Response: Thank you for your valuable suggestion! We have incorporated the verification results of the methods you suggested into the conclusion and abstract, providing more robust support for the conclusions of the paper.

 

Abstract Line 25-26 has been revised as follow:

 

Moreover, the overall disagreement for UMH-LUC is 0.183, a reduction of 0.099 compared to LUC without urban morphology and 0.19 compared to EULUC-China.

 

Section 6 Line 815-821 has been revised as follow:

 

The quantity disagreement in UMH-LUC is 0.046, which is significantly lower than those observed in the other two methods, suggesting that this model's classification results most closely project real-world proportions. The allocation disagreement is 0.137, which is similar to that of the EULUC-China method but lower than that observed in LUC without urban morphology. Overall, the overall disagreement for UMH-LUC is 0.183, a reduction of 0.099 compared to LUC without urban morphology and 0.19 compared to EULUC-China.

Author Response File: Author Response.pdf