Predicting Urban Expansion to Assess the Change of Landscape Character Types and Its Driving Factors in the Mountain City

Round 1

Reviewer 1 Report

Figure 2 and Table 1 can be canceled , and note the source under other tables and other figures

Author Response

Response to Reviewer 1 Comments

Figure 2 and Table 1 can be canceled, and note the source under other tables and other figures.

• Response:

Thank you very much for the valuable comments! Figure 2 illustrates the workflow of the study, parts of which are described in the main text of the manuscript and therefore there exist some overlaps. However, the figure shows the relationship between each step better than only the words. Table 1 summarizes the data sources employed in the research and, similarly, is more clear than describing them one-by-one in the manuscript. After a deep reflection by the authors, we decided to still keep Figure2 and Table1 to provide better readability for the readers.

We added a statement at the end of the manuscript indicating that all figures and tables were drawn by the authors.

Reviewer 2 Report

The work submitted for review confirms scientific considerations carried out in various research centres around the world and does not present innovative solutions. 

Author Response

Response to Reviewer 2 Comments

The work submitted for review confirms scientific considerations carried out in various research centres around the world and does not present innovative solutions. 

Response: Thank you for your suggestions. However, we are confused about the specific revisions. We would greatly appreciate if the reviewers could provide clearer suggestions for revisions.

Reviewer 3 Report

It is not quite clear how the modelling was carried out and what were the uncertainties. More explanation is needed, why these factors to determine indicators, why 10 factors, why K=8 (row 277).

Uncertainty could be evaluated if you model for 2000->2010; 2010->2020 and compared to 2000->2020. And/or if you compare modelled 2020 to actual 2020. Only after that you should apply model to see what would/could be by in 2040.

This kind of modelling assumes, that external parameters (population growth, pressure to move to cities etc) remain the same until 2040. This should be at least discussed.

Uncertainties of application of the modelling have to be addressed. Why would one believe that what you get is close to actual? How close?

Remarks on details.

Row 187-188  More thorough discussion of why the prices are high is needed.

Row 206 – 209. Presenting values of conversion probabilities (neighbourhood factor) as a table would be more perceptible (although might take more space). Why these values, needs reasoning.

Table 2. No grassland, water-body, other land used in land-use description used to determine indices. Why? Needs explanation.

Chapter 3.3. Formation of clusters is not clarified enough. Most of the chapter describes the K.Medoid method in general, only two sentences (not even a separate paragraph) (rows 235-238) say sth about this work.

Row 119   Socio-economic data is also spatial – in Table 1 the spatial resolution is provided for all data (location for POI). So being spatial is not specific for non-socio-economic data.

Rows 149-151 Unnecessary repetition of a sentence.

Pages 6-7, rows 216-218 Table should not be split between two pages

 

Author Response

Response to Reviewer 3 Comments

Thank you very much for these valuable comments! We have addressed all the issues raised one by one, as shown below. And the line number is the “simple marker” in the revision mode of Word.

 It is not quite clear how the modelling was carried out and what were the uncertainties. More explanation is needed, why these factors to determine indicators, why 10 factors, why K=8 (row 277).

• Response：

Thank you for your suggestions. We apologize for not explaining the selection basis of indexes, as well as the parameters and reliabilities of each model. We have thoroughly revised the manuscript based on your comments.

(1) The explanation of PLUS model

The PLUS model is a cellular automaton (CA) model that utilizes land expansion analysis strategy (LEAS) and multi-type random seed (CARS) to extract the driving factors of land expansion and landscape change. When compared to other models, the PLUS model demonstrates higher simulation accuracy. In Section 3.1.1, we provide a more detailed description of the advantages of the PLUS model.

The specific modifications are as follows:

“This approach avoids the issue of exponential growth in the number of categories associated with transformation types and retains the model's ability to analyze the mechanism of land-use change during this period while providing improved explanatory power.” (L169-171)

“The CARS module combines random seed generation and threshold decline mechanisms to enable the PLUS model to dynamically simulate the automatic generation of patches within the bounds of the development probability constraint.” (L179-182)

(2) The basis for selecting the indexes and factors

The indexes in this study including two parts: driving factors used to simulate land use changes and indexes used to characterize and cluster landscape characteristics. The indexes for clustering landscape character types are defined in three dimensions and consist of a total of 10 indexes. These indexes aim to comprehensively portray the landscape characteristics of the study area while reducing redundancy. The study area is located in a mountainous region with elevations ranging from 33m to 1303m, with rivers run through the city and the relatively flat areas along the river provide better conditions for urban expansion. Therefore, we selected three indexes - elevation, slope, and distance to water - to characterize the above geographical features.

Landscape pattern indexes can reveal the ecological risks brought by urbanization by characterizing the morphology and structural features of various patches. The LPI, AWMPED, and SHDI selected in this study to describe the landscape pattern characteristics from three aspects: dominance, regularity, and diversity, which are important dimensions for understanding landscape character types and their changes.

The land-use indexes were determined after repeated demonstrations. Urban and rural construction lands, forests, and farmlands play dominant roles in the study area, while water bodies and grasslands occupy a relatively small proportion, accounting for only 2.79% and 0.71%, respectively. As a result, the proportions of water bodies and grasslands in many units will be zero, which may affect the extraction of landscape character types. To describe the spatial distribution of water bodies We adopted the “distance to water” index instead. Grasslands were considered together with forests because they are adjacent or interlaced spatially and are usually treated as a whole when carrying out ecological conservation and restoration.

The specific modifications are as follows:

“In this study, the index system were constructed to describes the landscape characters of Chongqing city in three dimensions: topography, landscape pattern, and land-use features, aim to portray the landscape characteristics comprehensively while reducing redundancy. Unlike most large Chinese cities with flat topography, Chongqing has a highly undulating terrain with rivers run through the city. The relatively flat areas along the river provide areas for urban expansion. Therefore, we selected elevation, slope, and distance to water to characterize the above geographical features. The topography may influence the expansion patterns of construction areas, and landscape pattern indexes can reveal the ecological risks brought by urbanization by characterizing the morphology and structural features of ecological patches. Thus, the LPI, AWMPED, and SHDI selected to describe the landscape pattern characteristics from three aspects: dominance, regularity, and diversity. The land-use indexes are determined by the dominant land-use types. In the study area, urban and rural construction lands, forests, and farmlands play dominant roles, while water bodies and grasslands only accounting for only 2.79% and 0.71%, respectively. To avoid too many units with indicator values of 0 (no water bodies or grasslands), we adopted the “distance to water” index to describe the spatial distribution of water bodies. Additionally, grasslands were combined with forests because they are spatially interlaced and can be treated as a whole.” (L236-254)

(3) The explanation of K=8

Regarding the selection of K (the number of clusters), our intention is to reflect the critical characteristics of the study area with the fewest clusters while keeping the error within an acceptable range. First, we employed the Sum of Squared Error (SEE) method to observe the decreasing trend of error as K increased. The results showed that after K>3, the curve tended to flatten and the error is acceptable. However, 3 categories were unable to demonstrate the complex landscape characteristics of the study area. We tried 3-9 clusters one by one and compared the results. When K was less than 7, the integrated landscape characteristics of construction and ecological spaces in urban fringe regions were not highlighted. But if there were too many clusters, there would be information redundancy. Therefore, we clustered all samples into 8 landscape character types, which can fully express the regional characteristics of the study area without having particularly similar categories.

The specific modifications are as follows:

“Determining the value of K rationally is a key step to make the clustering results meaningful. In this study, the Sum of Squared Error (SEE) curve and the classification results were combined to determine the K-value, with the aim to reflect the critical characteristics of the study area with the fewest clusters while keeping the error within an acceptable range.” (L276-282)

“The SEE curve showed that after K>3, the curve tended to flatten and the errors are acceptable (Figure 5). However, only 3 groups were unable to demonstrate the complex landscape characteristics of the study area. We tried 3-9 groups one by one and found that when K was less than 7, the integrated landscape characteristics of construction and ecological spaces in urban fringe regions were neglected. But if there were too many clusters, there would be information redundancy.” (L318-323)

“Figure5 SEE curves for different Ks when clustering” (L342)

 

Uncertainty could be evaluated if you model for 2000->2010; 2010->2020 and compared to 2000->2020. And/or if you compare modelled 2020 to actual 2020. Only after that you should apply model to see what would/could be by in 2040.

• Response：

Thank you very much for pointing out the details of the deficiency. To begin with, in our study, we collected actual land use data from 2000 and 2020, and processed relevant fundamental data to obtain nine driving factors. Subsequently, we utilized the LEAS module in PLUS v1.3.5 to extract the development probability of each land use type. Following this, we employed the CARS module in PLUS v1.3.5 to simulate the land-use distribution in 2020 based on the 2000 land-use data. The simulated 2020 land-use distribution was compared to the actual distribution, and the model's accuracy was validated using Kappa statistical tools. After multiple testing, we obtained the optimal parameter settings to ensure the stability and accuracy of the model. Finally, we employed these land use development probability grids and related parameters to simulate the land-use distribution in 2040, based on the 2020 land use data. Specific downloads of the software are available: https://github.com/HPSCIL/Patch-generating_Land_Use_Simulation_Model.

These instructions were supplemented in Section 3.1.3, with the following specific modifications:

“We simulated land-use distribution in 2020 using the CARS module in PLUS v1.3.5, based on the 2000 land-use data and the raster dataset of land-use development probability. The simulated 2020 land-use data was compared to the actual data, and the accuracy of the model was evaluated using the Kappa statistical tool. We employed multiple testing to obtain optimal parameter settings, ensuring the stability and accuracy of the model.” (L217-222)

 This kind of modelling assumes, that external parameters (population growth, pressure to move to cities etc) remain the same until 2040. This should be at least discussed.

• Response：

We are very grateful for your suggestions. The simulation of land use distribution was obtained using the CARS module in the PLUS model, based on the land use development probability dataset generated by the LEAS module, which has been previously mentioned and elaborated upon in earlier responses. The CARS module in the PLUS model is a Cellular Automaton (CA) model that incorporates a patch generation mechanism based on multiple types of random land use seeds. During the simulation process, the adaptive coefficients of each land-use type determine their competition in space, thus driving the expansion of land-use types that meet the anticipated future demand.

We have incorporated this explanation in Section 3.1.1, with the following specific modifications:

“The CARS module in the PLUS model is a Cellular Automaton (CA) model that incorporates a patch generation mechanism based on various types of random land use seeds. Throughout the simulation process, the spatial competition for each land-use type is determined by adaptive coefficients, thus driving the expansion to meet the anticipated future demand.” (L175-179)

 Uncertainties of application of the modelling have to be addressed. Why would one believe that what you get is close to actual? How close?

• Response：

Thank you very much for pointing out the details of the deficiency. As previously stated, the accuracy of the model and the suitability of related parameters were verified through multiple comparisons between simulated land use in 2020 and actual land use. The magnitude of the Kappa coefficient indicates the accuracy of the model. Relevant literature suggests that if the Kappa coefficient exceeds 0.7, the model and related parameter settings can be utilized to forecast future land use changes.

In Section 3.1.3, we have incorporated the pertinent parameter settings and explicated the significance of each parameter, which may augment its value as a reference. The following specific modifications have been made:

“In this study, patch generation refers to the attenuation coefficient of the decreasing threshold. It was set to 0.2, with a parameter range of 0 to 1, and values closer to 1 corresponded to a higher difficulty of land-use change. Expansion coefficient refers to the probability of random patch seeds. It was set to 0.7, with a parameter range of 0 to 1, and values closer to 1 corresponded to a higher probability of generating new patches.” (L223-227)

 Remarks on details.

Row 187-188  More thorough discussion of why the prices are high is needed.

• Response：

Thank you very much for pointing out the deficiency. We have added details on the reasons for the high prices in Chongqing. Firstly, Chongqing is a city with a great natural environment in China. It has low air pollution, a large number of forests and rivers, z which has attracted a large number of non-local migrants. Secondly, the rapid government-led urbanisation of recent years has seen rural residents migrate to the city. The conflict between limited land supply and growing demand has led to a steady rise in urban property prices.

The specific modifications are as follows:

“The most notable phenomenon in the city of Chongqing is the price of real estate. Influenced by rapid urbanisation and Chongqing's natural environment[42, 43], large numbers of migrants are moving into the city. The conflict between limited land supply and growing demand has led to a steady rise in urban property prices. High prices are accompanied by a good provision of infrastructure in the surrounding area[44].” (L201-205)

 Row 206 – 209. Presenting values of conversion probabilities (neighbourhood factor) as a table would be more perceptible (although might take more space). Why these values, needs reasoning.

• Response：

Thank you very much for pointing out the details of the deficiency. In our study, the parameter 'neighborhood factor' specifically refers to neighborhood weights within the PLUS v1.3.5 software. To prevent confusion, we have revised this terminology to 'neighborhood weights'. This parameter mainly reflects the influence of different land-use types (pixels) on the neighborhood, with higher values indicating a greater impact on the neighborhood. The neighborhood weight of each land-use type is determined by its proportion of changing area, and their cumulative value is 1. To further clarify this parameter, we have included supplementary explanations and tables in Section 3.1.3.

The following specific modifications have been made:

“Neighborhood weights refers to the influence of pixels on neighborhood, with a parameter range of 0 to 1, higher values corresponded to a greater effect on neighborhood. Each neighborhood weight could be calculated by the ratio of the change area of this type to the total change area.” (L228-231)

“Table 2 Neighborhood weights of land-use types” (L236)

Table 2. No grassland, water-body, other land used in land-use description used to determine indices. Why? Needs explanation.

• Response：

Thank you for pointing out the unclear parts in our manuscript. As the response to the first comment, urban and rural construction lands, forests, and farmlands play dominant roles in the study area, while water bodies and grasslands occupy a relatively small proportion, accounting for only 2.79% and 0.71%, respectively. As a result, the proportions of water bodies and grasslands in many units will be zero, which may affect the extraction of landscape character types. To describe the spatial distribution of water bodies We adopted the “distance to water” index instead. Grasslands were considered together with forests because they are adjacent or interlaced spatially and are usually treated as a whole when carrying out ecological conservation and restoration. We have added the relevant contents to the manuscript.

The specific modifications are as follows:

“The land-use indexes are determined by the dominant land-use types. In the study area, urban and rural construction lands, forests, and farmlands play dominant roles, while water bodies and grasslands only accounting for only 2.79% and 0.71%, respectively. To avoid too many units with indicator values of 0 (no water bodies or grasslands), we adopted the “distance to water” index to describe the spatial distribution of water bodies. Additionally, grasslands were combined with forests because they are spatially interlaced and can be treated as a whole.” (L249-255)

 Chapter 3.3. Formation of clusters is not clarified enough. Most of the chapter describes the K.Medoid method in general, only two sentences (not even a separate paragraph) (rows 235-238) say sth about this work.

• Response：

As you mentioned, we did not provide detailed information about the clustering method and the determination of K-value in the manuscript. We have added these details to the methods (section 3.3) and results (section 4.2) respectively.

The specific modifications are as follows:

“Determining the value of K rationally is a key step to make the clustering results meaningful. In this study, the Sum of Squared Error (SEE) curve and the classification results were combined to determine the K-value, with the aim to reflect the critical characteristics of the study area with the fewest clusters while keeping the error within an acceptable range.” (L276-280)

“The SEE curve showed that after K>3, the curve tended to flatten and the errors are acceptable (Figure 5). However, only 3 groups were unable to demonstrate the complex landscape characteristics of the study area. We tried 3-9 groups one by one and found that when K was less than 7, the integrated landscape characteristics of construction and ecological spaces in urban fringe regions were neglected. But if there were too many clusters, there would be information redundancy.” (L318-323)

“Figure5 SEE curves for different Ks when clustering” (L342)

 Row 119   Socio-economic data is also spatial – in Table 1 the spatial resolution is provided for all data (location for POI). So being spatial is not specific for non-socio-economic data.

Rows 149-151 Unnecessary repetition of a sentence.

• Response：

Thank you very much for pointing out the deficiency. Firstly, we have modified the presentation of the data. There is no longer a classification of spatial data, as all data used in this study are spatial. Secondly, we have deleted the last sentence in 3. Methodology (which was repeated).

The specific modifications are as follows:

“A total of five types of data were used in this study, including the land-use, the digital elevation model (DEM), transportation, the Point of Interest (POI) and the socio-economic data.” (L124-126)

 Pages 6-7, rows 216-218 Table should not be split between two pages

• Response：

Thank you very much for pointing out the deficiency. We modified all the tables so that they are not split in 2 pages.

The specific modifications are as follows:

“Table 3 Details of the index system to cluster LCTs” (L257)

Reviewer 4 Report

The manuscript by Mou, Chen and Huang represents an interesting contribution on the relationship between urban sprawl in mountainous cities and landscape changes. Rapid urbanization has had large impacts on the landscape, ecosystems and people's quality of life. Rapid urbanization also makes people more vulnerable to climate change impacts, as seen in recent decades. The mistakes of the past, which have occurred in various cities of the world, provide evidence of unsustainable practices that should not be adopted. 

Based on these considerations, this research attempts to predict and identify Landscape Character Types in different periods using the PLUS model and K-Medoids clustering algorithm. The methodology is applied to Chongqing, a city located in south-central China with over 30 million inhabitants, to reveal differences in the influence of driving factors on Landscape Character Types.

The manuscript is full of tables and graphic elaborations.

In order to improve the manuscript quality I suggest some minor revisions: 

1) Please check the manuscript carefully because there are some typos. For example line 15 (K-Mediods instead of K-Medoids), lines 633 and 636 (uppercase instead of lowercase), line 529 (delete 'p') and so on.

2) Some international cases should be mentioned in the introduction (lines 27-91). Only Paris region is mentioned (line 51). See also point number 4.

3) Conclusions (lines 491-510) are too short to represent the interesting content of the paper and the results of the research.

4) References are adequate to the study and updated, however, it is suggested to include more international references about urban sprawl (in Europe, in the Americas, in Africa). See also point number 2.

 

 

Author Response

Response to Reviewer 4 Comments

Thank you very much for these valuable comments! We have addressed all the issues raised one by one, as shown below. And the line number is the “simple marker” in the revision mode of Word.

The manuscript by Mou, Chen and Huang represents an interesting contribution on the relationship between urban sprawl in mountainous cities and landscape changes. Rapid urbanization has had large impacts on the landscape, ecosystems and people's quality of life. Rapid urbanization also makes people more vulnerable to climate change impacts, as seen in recent decades. The mistakes of the past, which have occurred in various cities of the world, provide evidence of unsustainable practices that should not be adopted. 

Based on these considerations, this research attempts to predict and identify Landscape Character Types in different periods using the PLUS model and K-Medoids clustering algorithm. The methodology is applied to Chongqing, a city located in south-central China with over 30 million inhabitants, to reveal differences in the influence of driving factors on Landscape Character Types.

The manuscript is full of tables and graphic elaborations.

In order to improve the manuscript quality I suggest some minor revisions:

1) Please check the manuscript carefully because there are some typos. For example line 15 (K-Mediods instead of K-Medoids), lines 633 and 636 (uppercase instead of lowercase), line 529 (delete 'p') and so on.

• Response：

Thank you very much for pointing out the deficiency. We double-checked the entire manuscript and corrected all typos.

The specific modifications are as follows:

“…based on the PLUS model and the K-Medoids algorithm…” (L15)

“LU, Y.; XU, S.; LIU, S.; WU, J. An approach to urban landscape character assessment: linking urban big data and machine learning. Sustain Cities Soc 2022, 83, 103983, doi:10.1016/j.scs.2022.103983.” (L708)

“Affairs., U.N.D.O. The World at Six Billion. In United Nations Secretariat: Washington, D.C, 2000; Vol. 26, 841.” (L591)

“GORMUS, S.; OĞUZ, D.; EŞBAH TUNÇAY, H.; CENGİZ, S. Using Landscape Character Analysis to Assess The Relationship Between Protected and Nonprotected Areas: The Case of The Küre Mountains National Park. Tarım Bilimleri Dergisi 2021, doi:10.15832/ankutbd.640159.” (L705-707)

2) Some international cases should be mentioned in the introduction (lines 27-91). Only Paris region is mentioned (line 51). See also point number 4.

• Response：

Thank you very much for pointing out the deficiency. We have added studies of land use simulations in different countries and cities (especially in North America and Africa) to show the breadth and scientific validity of the application of these models.

The specific modifications are as follows:

“Onuwa Okwuashi et al. presented a novel integration of support vector machine, Markov chain and cellular automata for urban change modelling in Nigeria, Africa’s most populous city[16]. In Paris, A. Lemonsu et al. employed a land-use transport interaction socio-economic model and an urban climate model to simulate urban temperatures under five urban expansion scenarios[17]. There is also a study that use the Multi-layer perceptron (MLP) based Artificial Neural Network-Markov Chain (ANN-Markov) model to simulate three future urban growth scenarios in Miami Metropolitan Area. It assesses the future flood risk under each scenario[18].” (L49-56)

 3) Conclusions (lines 491-510) are too short to represent the interesting content of the paper and the results of the research.

• Response：

Thank you very much for pointing out the deficiency. We have rewritten the conclusions. Summarising the interesting elements of the study results. We have detailed the important and interesting results of this study in “Conclusions”.

The specific modifications are as follows:

“The results are as follows: (1) If the urban development trend from 2000 to 2020 continues until 2040, urban construction land in the central city of Chongqing will encroach on a large amount of farmland (79.6%) and a small amount of forest (10.1%). There also has been some expansion of urban construction land around some rivers and forest belts.

(2) From 2000 to 2040, there is an encroachment of the LCTs (dominated by farmland) around the built-up area by another LCTs (dominated by urban construction land), as well as an expansion of villages away from the built-up area.

(3) The driving factors contribute to the conversion of all land use types from high to low: the nighttime light, POI, elevation and distance to trunk roads. The distance to water bodies mainly influences the conversion of water bodies and urban construction land. Population density and distance to trunk roads mainly influence the conversion of rural construction land.

(4) There are also differences in the main driving factors affecting changes in LCTs. The nighttime light has the highest contribution to Types 1, 4, 5, 7, and 8. The elevation has the highest contribution to Types 3 and 6.” (L549-563)

 4) References are adequate to the study and updated, however, it is suggested to include more international references about urban sprawl (in Europe, in the Americas, in Africa). See also point number 2.

• Response：

Thank you very much for pointing out the deficiency. We have added relevant research in North America (Miami Metropolitan Area) and Africa (Nigeria, Africa’s most populous city).

The specific modifications are as follows:

“Onuwa Okwuashi et al. presented a novel integration of support vector machine, Markov chain and cellular automata for urban change modelling in Nigeria, Africa’s most populous city[16]. In Paris, A. Lemonsu et al. employed a land-use transport interaction socio-economic model and an urban climate model to simulate urban temperatures under five urban expansion scenarios[17]. There is also a study that use the Multi-layer perceptron (MLP) based Artificial Neural Network-Markov Chain (ANN-Markov) model to simulate three future urban growth scenarios in Miami Metropolitan Area. It assesses the future flood risk under each scenario[18].” (L49-56)

Round 2

Reviewer 3 Report

Usage of dots and spaces in the Figure and Table titles needs to be rechecked, it is not consistent (and should follow the journal style)