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

Tourism-Induced Urbanization in Phuket Island, Thailand (1987–2024): A Spatiotemporal Analysis

Urban Sci. 2025, 9(3), 55; https://doi.org/10.3390/urbansci9030055

by Sitthisak Moukomla^1,*

and Wijitbusaba Marome²

Reviewer 1: Anonymous

Reviewer 2: Anonymous

Reviewer 3: Anonymous

Urban Sci. 2025, 9(3), 55; https://doi.org/10.3390/urbansci9030055

Submission received: 1 January 2025 / Revised: 9 February 2025 / Accepted: 19 February 2025 / Published: 20 February 2025

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for inviting me to review this paper. The reviewer enjoyed reading it and thinks that the research merits publication. However, the work needs some major improvements by considering the following points:

- Major concerns regarding overall structure

- According to the authors, this study “assesses the effects of tourism and economic policies on changes in land use and land cover” (L14). Judging from the title, introduction, and conclusions, this seems to be the aim of the paper. However, I do not see any existing evidence or findings that these changes in LULC are induced by tourism. The authors also mention the 1997 Asian Financial Crisis and the 2004 Indian Ocean Tsunami to explain their findings, without providing evidence to support these explanations. Due to this weakness, the explanations provided by the authors come across as ad hoc. The authors should incorporate supplementary data to make a direct connection between their findings and tourism development. For example, how did visitor numbers to Phuket Island change over the study period? Where are major tourism attractions and facilities (e.g., hotels, airports) located on the island? Do they overlap with “city cores” of urban expansion shown in Figures 5 or 6? Without such information, I cannot attribute the changes observed in this paper to tourism-driven development.

- Similar to the above point, I did not find Section 4.2 to be particularly convincing due to a lack of clear connections. The authors mention some local-level changes in industry structure, as well as global crises, but I do not see any connections to specific tourism and economic policies. For example, wouldn’t changes in land use be affected by national-level and island-level tourism/economic development plans? What were some major policies that have been implemented during the study period and which areas were affected by such policies?

- Section 4.4 Policy recommendations seem to be extremely generic rather than being linked to findings. In particular, I’m not convinced that “it is imperative to amend zoning regulations to enable the construction of a variety of mixed-use buildings in urban centers” (L443) when this study did not classify “Urban” areas into “residential, commercial, and recreational areas” (L445). I encourage authors to incorporate their findings into this section and make recommendations for Phuket based on the findings.

- In addition, here are major concerns regarding methodology/findings. Mainly, these points are avenues that require further explanation.

- I’m not entirely familiar with functions being offered in the Google Earth Engine (GEE). Hence, it seems unclear to me whether certain methodological choices were made by others or they are predetermined by the tool. Explicitly stating which part of analysis/data is from GEE would be helpful for readers to better understand the methodology.

- The authors mention a +-6 month window around the target date for each data point (L103). But which date/month is this?

- Satellite imagery from LANDSAT is being classified into six categories using a Random Forest model (L139). However, a Random Forest is a supervised algorithm, meaning that data used for training and validation must have been pre-classified. Where did the authors obtain this from? What data years had such “pre-classification” and what years were “predicted” from trained models?

- I could not follow the authors’ adaptation of the Shannon entropy index, and I do not see this use from Dou & Han (2022) that the authors are referencing. If you’re determining p_i by “dividing each year's urban area by the total urban area over all years” (L163), and summing them up for all years to get H_n, shouldn’t there be a single index over 37 years? Or does n expand each year to include previous years? Further explanations are needed.

- Table 1 mentions Low/Moderate/High levels of UEII and Entropy. How were these levels determined? In addition, the authors mention certain UEII/Entropy values as “moderate” or high” in section 3.3. But they don’t seem to match the defined Low/Moderate/High levels in Table 1. For example, shouldn’t “moderate UEII of 9.15” (L282) be “High UEII (>2%)”? Further explanations regarding how these cut-offs were determined are needed, and the interpretations of the values also need to match the levels being defined here.

- Throughout the paper, the authors mainly rely on a 5-year timeframe in their findings and results. Although, I’ve noticed several instances where the authors use a shorter 1-year timeframe (e.g., L255) or a longer 10-year timeframe (e.g., section 3.4). Hopping to a shorter 1-year timeframe would be particularly problematic since readers cannot simply “interpolate” 1-year data that’s not being presented in the paper. For example, if “The city area surged from 85 sq km to around 160 sq km during the period of 2005-2006” (L255), how is this land use comparable to 2003 or 2004 levels? I suggest either providing these more granular data as yearly line graphs or unifying the temporal unit of analysis being used throughout the paper.

- Some minor points that need attention

- In Figure 4, the colors for Agriculture and Mangrove seem to be identical and hence cannot be distinguished in the panel (a).

- In Figure 6, the authors used discrete colors to distinguish the urban expansion areas. I would instead recommend graduated/continuous colors to easily distinguish urban expansion over time.

- I suggest adding row-wise percentages in Table 2 for each land use category. Section 3.1 is describing these percentages and would be helpful for readers to see them side by side in the table.

- Authors use phrases like “increased intricacy in layouts” (L275) or “maintained level of complexity” when explaining specific (L285). I am failing to see how entropy can indicate complexity or intricacy of land use layout (wouldn’t it be simply an indicator of equal dispersion of urban areas regardless of specific layout/structure?). Hence, I suggest authors clarify the meanings.

Comments on the Quality of English Language

I’ve noticed a few instances of incomplete sentences/grammatical errors. For instance, L199 needs to be “The agricultural use of land” instead of just “The use of land”. I recommend authors to revisit the manuscript after the revision.

Author Response

- Major concerns regarding overall structure

Comments 1: According to the authors, this study “assesses the effects of tourism and economic policies on changes in land use and land cover” (L14). Judging from the title, introduction, and conclusions, this seems to be the aim of the paper. However, I do not see any existing evidence or findings that these changes in LULC are induced by tourism. The authors also mention the 1997 Asian Financial Crisis and the 2004 Indian Ocean Tsunami to explain their findings, without providing evidence to support these explanations. Due to this weakness, the explanations provided by the authors come across as ad hoc. The authors should incorporate supplementary data to make a direct connection between their findings and tourism development. For example, how did visitor numbers to Phuket Island change over the study period? Where are major tourism attractions and facilities (e.g., hotels, airports) located on the island? Do they overlap with “city cores” of urban expansion shown in Figures 5 or 6? Without such information, I cannot attribute the changes observed in this paper to tourism-driven development.

Response 1: We appreciate the reviewer's feedback and have revised our manuscript to strengthen the connection between tourism, economic policies, and land use/land cover (LULC) changes. Specifically, we have made the following improvements:

Clarification of Methodology (L13-16):

We explicitly state that we integrate Landsat satellite images with sophisticated analytical methods to assess the effects of tourism and economic policies on LULC changes.
We also specify that we utilize Google Earth Engine (GEE) for cloud-based data processing and Random Forest (RF) models for classification. This methodological clarification reinforces the robustness of our analysis.

Inclusion of Visitor Data (Figure 1, L58-65):

To directly link tourism to LULC changes, we provide Figure 1 as supplementary data illustrating visitor numbers to Phuket Island over the study period.
We analyze the relationship between visitor trends and revenue, demonstrating how tourism-dependent economic fluctuations correspond to external crises such as the 1997 Asian Financial Crisis, the 2004 Indian Ocean Tsunami, the 2020 COVID-19 pandemic, and the 2021 countrywide lockdown.

Spatial Analysis of Tourism Attractions (Figures 2 & 5):

We have incorporated spatial data on major tourism attractions in Figure 2, alongside the LULC changes in Figure 5.
This addition provides a clearer visualization of how urban expansion overlaps with key tourism hubs, strengthening the argument that tourism plays a role in shaping land development patterns.

These modifications address the reviewer's concern regarding the attribution of LULC changes to tourism by incorporating supplementary data and improving spatial contextualization. We believe these revisions enhance the rigor of our study and clarify the connection between tourism and land use transformation.

Comments 2. Similar to the above point, I did not find Section 4.2 to be particularly convincing due to a lack of clear connections. The authors mention some local-level changes in industry structure, as well as global crises, but I do not see any connections to specific tourism and economic policies. For example, wouldn’t changes in land use be affected by national-level and island-level tourism/economic development plans? What were some major policies that have been implemented during the study period and which areas were affected by such policies?

Response 2: We appreciate the reviewer's insightful comments and acknowledge the need for a clearer connection between tourism, economic policies, and land-use changes in Section 4.2. In response to these concerns, we have extensively revised this section to explicitly outline the role of national and local policies in shaping land-use transformations in Phuket. The key improvements include:

Explicit Link Between Tourism and Land Demand:

We provide a clearer discussion on how the surge in international tourism demand has driven up land prices, particularly in high-end tourist locations on Phuket’s western coast.
We elaborate on the social consequences of this transformation, such as residential displacement of local residents to more affordable areas in the eastern part of the island, reinforcing the link between tourism growth and urban migration patterns.

Incorporation of Key National and Local Policies:

We integrate a discussion on major tourism and economic development policies that influenced land-use dynamics. Specifically:

Visit Thailand Year (2015) and Amazing Thailand (1998), which spurred investment in luxury resorts and condominiums.
The easing of foreign ownership laws, which facilitated international real estate investment and accelerated urban expansion.
The Smart City Initiative (2015), which shifted development priorities towards sustainable urban planning and mixed-use inland developments.

Integration of Crisis-Response Policies and Their Spatial Effects:

We expand on how past crises, including the 1997 Asian Financial Crisis, the 2004 Indian Ocean Tsunami, and the 2020 COVID-19 pandemic, affected land-use patterns:

The 1997 crisis led to a decline in foreign investment, delaying tourism infrastructure projects, while the Thai government introduced economic rehabilitation policies such as tax incentives and tourism subsidies.
The 2004 tsunami prompted large-scale reconstruction efforts, international aid, and the introduction of new zoning restrictions, albeit with inconsistent enforcement. This influenced urban expansion differently across locations (e.g., high-rise development in Kamala Beach vs. conservation zones in other areas).
The COVID-19 pandemic caused underutilization of tourism infrastructure, leading to land-use repurposing and a shift toward more flexible urban design strategies.

Case Study: The Phuket Sandbox Program (2021):

We introduce the Phuket Sandbox program, a government initiative aimed at revitalizing tourism by allowing vaccinated visitors to enter the island without quarantine restrictions.
This policy not only provided a temporary boost to the tourism economy but also altered land-use objectives, as numerous hotels and resorts were repurposed for domestic tourism and alternative uses, marking a shift toward more resilient urban planning.

Additional Figures for Policy-Tourism-Land Use Nexus:

We now supplement our analysis with Figure 1, illustrating fluctuations in tourist arrivals in response to global and national crises.
Figures 2 and 5 highlight major tourism attractions and land-use changes, making it easier to visualize how policies and external shocks influenced spatial development.

These revisions strengthen the section by establishing a clear and direct link between tourism-driven economic policies, crisis responses, and land-use transformations. We believe these changes comprehensively address the reviewer’s concerns and significantly enhance the clarity and impact of our study. We revised Section 4.2. as follow;

Phuket has transformed into a sanctuary for international visitors. Consequently, the demand for land is surging resulting in land prices soaring, particularly in excellent tourist locations. In the western part of the island, where high-end hotels and resorts are situated, has encountered a higher cost of living. The service sector, propelled by tourism, necessitates a considerable workforce; nevertheless, numerous local employees find residence in these expensive regions unaffordable. Consequently, residential displacement has compelled local residents to congregate in the eastern region of the island, especially in Phuket Downtown and adjacent neighborhoods, where housing and living expenses are comparatively more economical. Phuket's geographical limitation as an island, which restricts expansion into adjacent provinces, has resulted in land scarcity that necessitates the transformation of agricultural regions into urban areas to meet population growth and economic needs. This urban evolution underscores the island's distinctive developmental challenges, as alterations in land use are influenced by the convergence of tourism dynamics, housing affordability, and geographical constraints.

Additionally, Phuket's development has been shaped by many national and local policy measures aimed at promoting economic growth and tourism, alongside these worldwide problems. In the early 2000s, these laws catalyzed a significant increase in investment, especially in areas such as Patong Beach and Karon, where agricultural land was converted into commercial zones and resorts. A pivotal moment in the island's urban growth occurred during the 2004 tsunami, which inflicted significant damage on Phuket's coastline. International assistance and governmental initiatives were employed to facilitate reconstruction operations, specifically targeting the rehabilitation of tourism infrastructure. The implementation of new zoning restrictions aimed to limit growth in high-risk coastal regions; yet, enforcement was irregular. In specific areas, like Kamala Beach, post-tsunami rehabilitation led to the establishment of high-rise hotels and rapid urban growth, but other regions were designated for conservation to mitigate future dangers. This paradox underscores the tension between environmental preservation and economic recovery that has defined much of Phuket's development.

The island had a substantial rise in the development of resorts and condominiums due to initiatives like Visit Thailand Year (2015) and Amazing Thailand (1998), which provided financial incentives for tourism investment. The expansion of luxury home projects in sought-after locations like Laguna and Kata Beach was greatly impacted by these policies. The alteration of Phuket's land-use patterns was expedited by the easing of foreign ownership laws for real estate, drawing international purchasers and developers. Furthermore, the Smart City Initiative, launched in 2015, marked a pivotal shift in Phuket's developmental path by emphasizing sustainable urban planning and digital innovation. The island's economy has diversified with investments in technological parks, transportation facilities, and upscale residential complexes, thus reducing its reliance on conventional tourism. These alterations have also impacted land-use objectives, placing increased focus on urban densification and mixed-use developments in inland areas such as Kathu and Chalong.

Numerous events, both global and local, impact tourism in Phuket. The 1997 Asian Financial Crisis, the 2004 Indian Ocean Tsunami, and the 2020 COVID-19 pandemic all led to substantial declines in tourist arrivals and corresponding alterations in land-use dynamics (for detailed information, please refer to Figure 1). The 1997 financial crisis led to a significant decline in foreign investment and the deferral of various tourism-related infrastructure initiatives. The Thai government enacted economic rehabilitation measures, including tax incentives for international investors and subsidies for domestic tourism operators. In 2020, the COVID-19 pandemic posed unique issues, since the collapse of the worldwide tourism industry led to widespread economic upheavals and the underutilization of tourism infrastructure [59].

In 2021, the Thai government created the Phuket Sandbox program to mitigate these repercussions. This program allowed immunized international passengers to visit the island without quarantine requirements. The tourism sector had a temporary enhancement due to this policy, which also prompted a reevaluation of land-use objectives. A multitude of properties, including hotels and resorts, were repurposed for domestic tourism or alternative uses, signifying a shift towards more flexible and resilient urban design techniques. Consequently, land-use decisions have been markedly affected by environmental and zoning restrictions, notwithstanding their erratic implementation.

Comments 3. Section 4.4 Policy recommendations seem to be extremely generic rather than being linked to findings. In particular, I’m not convinced that “it is imperative to amend zoning regulations to enable the construction of a variety of mixed-use buildings in urban centers” (L443) when this study did not classify “Urban” areas into “residential, commercial, and recreational areas” (L445). I encourage authors to incorporate their findings into this section and make recommendations for Phuket based on the findings.

Response 3: We sincerely appreciate the reviewer’s feedback regarding Section 4.4 and acknowledge the need for stronger alignment between our findings and policy recommendations. To address these concerns, we have significantly revised this section to ensure that our recommendations are directly informed by the study’s results, particularly regarding the impacts of tourism-driven urban expansion, environmental constraints, and economic resilience in Phuket.

Key Revisions and Improvements:

Zoning Regulations Aligned with Findings

We recognize the reviewer's concern about the lack of classification between residential, commercial, and recreational areas.
To improve specificity, we have tailored zoning recommendations based on Phuket’s distinct urban dynamics.

We propose geographically specific zoning that safeguards fragile coastal ecosystems while directing compact, mixed-use development toward designated urban centers (e.g., Phuket Downtown and Chalong).
Our recommendations incorporate urban growth boundaries to mitigate uncontrolled expansion into agricultural lands and environmentally vulnerable areas, ensuring alignment with our study’s land-use change findings.

Housing Affordability and Tourism-Induced Displacement

The housing affordability crisis in Phuket, exacerbated by tourism-driven gentrification, is directly linked to findings on residential displacement from high-cost western coastal areas to eastern inland regions.
We now propose affordable housing strategies that prioritize displaced service-sector workers, utilizing public-private partnerships, tax incentives, and density bonuses to promote subsidized housing near employment centers.
Our revised recommendations emphasize infrastructure modernization in growing eastern urban areas (e.g., Phuket Downtown, Kathu) to accommodate increasing population density.

Transportation Infrastructure to Reduce Congestion and Tourism Pressures

Our study highlights urban sprawl and traffic congestion, necessitating improved public transportation and sustainable mobility options.
We now propose high-capacity transit solutions, such as Bus Rapid Transit (BRT) or light rail, strategically linking key tourism hubs (airport, beaches, Phuket Town) to reduce car dependency.
Additionally, our integrated zoning and transit planning approach aims to minimize travel demand, reduce congestion, and enhance connectivity for both tourists and residents.

Environmental Resilience and Ecosystem Protection

Given our study’s findings on coastal and ecological degradation, we now emphasize stronger enforcement of conservation laws and the restoration of mangrove forests, which serve as natural storm barriers.
We also propose green infrastructure solutions, including permeable pavements, urban parks, and rooftop gardens, to mitigate flash flooding—a recurring challenge identified in our study.
Stricter environmental impact assessments (EIA) for both large-scale resorts and small tourism developments will be crucial in minimizing ecological damage.

Economic Diversification Beyond Tourism

Our findings indicate economic vulnerability due to Phuket’s overreliance on tourism, particularly during external shocks such as the COVID-19 pandemic.
To address this, we recommend strategies for economic diversification, including:

Innovation hubs and tax incentives to attract investment in technology, healthcare, and education sectors.
Sustainable tourism alternatives, such as agrotourism partnerships and eco-tourism initiatives, to protect agricultural land while generating conservation revenue.

Inclusive Governance and Climate Resilience

Given Phuket’s exposure to climate risks (e.g., sea level rise, resource competition), we now emphasize the need for resilience-focused policies that integrate climate adaptation strategies at all levels of urban planning.
Additionally, we stress the importance of inclusive governance, ensuring that local communities, businesses, and environmental NGOs are actively involved in decision-making to promote transparent and participatory development.

Through these revisions, our policy recommendations are now firmly grounded in our study’s findings, addressing the urban, environmental, economic, and governance challenges specific to Phuket. By integrating smart zoning, equitable housing policies, green infrastructure, and economic innovation, we provide a comprehensive roadmap for sustainable urban development. We believe these changes effectively respond to the reviewer's concerns and significantly enhance the practical relevance of our recommendations. Section 4.4 revision as follow,

Due to the rapid urban expansion, fueled by tourism industry in Phuket, to promote sustainable urban development. In this section we proposed the policy recommendations for adaptation in rapid urban expansion in Phuket. First the adoption of zoning regulations should geographically tailored. These regulations are designed to safeguard fragile coastal ecosystems and tourism sites, while simultaneously directing compact, mixed-use development to urban centers such as Phuket Downtown. By instituting definitive urban growth limits, the island can mitigate unrestricted expansion into agricultural area and vulnerable ecosystems, so assuring compatibility with long-term ecological and infrastructural capacities. To optimize land utilization, new buildings must comply with rigorous sustainability criteria, including energy-efficient designs and the obligatory incorporation of renewable energy sources. These strategies not only mitigate environmental impacts but also safeguard development against future climatic difficulties.

Next, to mitigate the severe housing crisis intensified by tourism-induced gentrification, focused affordable housing programs should concentrate service-sector workers displaced from expensive coastal regions. Public-private partnerships may motivate developers—via tax incentives or density bonuses—to incorporate subsidized units in buildings adjacent to employment centers. Concurrently, modernizing antiquated facilities (water, sewerage, electricity) in eastern regions will provide dependable services for expanding populations. Equitable planning must incorporate marginalized groups in decision-making processes, guaranteeing access to healthcare, education, and public places to promote inclusive and cohesive communities.

Plus, Phuket's transportation infrastructure necessitates immediate renovation to mitigate congestion and diminish dependence on private vehicles. Enhancing high-capacity transit systems, like as bus rapid transit (BRT) or light rail, to link essential locations like the airport, beaches, and Phuket Town, will optimize mobility. In addition, walkable urban planning and designated bike lanes in highly populated regions can facilitate sustainable commuting and improve quality of life. Integrating transit planning with zoning regulations will facilitate effortless access to employment, housing, and facilities, hence diminishing urban sprawl and travel demand.

Environmental resilience is also important. These include the destruction of essential ecosystems for tourism-driven development, for instance, the destruction of coastal mangroves that are natural storms barriers. Increasing the effectiveness of the enforcement of conservation laws and incorporating the restoration of the mangroves into the climate adaptation strategies could restore these vital systems. One of the natural disasters in Phuket is flash flood. Here, green infrastructure such as permeable pavements, urban parks and rooftop gardens should also be strengthened to help in fighting flooding while enhancing biodiversity and improving air quality. Further environmental impact assessment of all projects (from the high-end hotel to small resort) could also help to reduce the ecological effects. Economic diversification is therefore necessary to reduce over reliance on the tourism sector. Creating other sectors like technology, healthcare and education through innovation hubs or tax incentives makes the economy stronger against global shocks. However, sustainable tourism models for example, agrotourism partnerships and eco-tourism activities can be used to conserve agricultural land and earn money for conservation. Success also relies on democratic government. This ensures that citizens, businesses, and environmental NGOs are involved in the decision-making process and therefore, the process is transparent and participatory. Climate change threats such as sea level rise and resource competition must be addressed by policy makers and resilience must be addressed in every decision made on development. Therefore, through the integration of smart zoning, equitable housing, green infrastructure and economic innovation, Phuket can create a path for sustainable development to sustain its natural assets, improve the living standard of its citizens and build long term strength.

Comments 4. In addition, here are major concerns regarding methodology/findings. Mainly, these points are avenues that require further explanation.

Comments 4.1 I’m not entirely familiar with functions being offered in the Google Earth Engine (GEE). Hence, it seems unclear to me whether certain methodological choices were made by others or they are predetermined by the tool. Explicitly stating which part of analysis/data is from GEE would be helpful for readers to better understand the methodology.

See Response 4.2

Comments 4.2 The authors mention a +-6 month window around the target date for each data point (L103). But which date/month is this?

Response 4.2: We appreciate the reviewer’s request for further clarification regarding our methodological approach, particularly the use of Google Earth Engine (GEE) and its role in our analysis. To address these concerns, we have revised the methodology section to explicitly differentiate between automated processes performed within GEE and additional analyses conducted externally.

Key Revisions and Clarifications

Explicit Differentiation of GEE Functions vs. External Analysis

We now clearly delineate which processes were performed directly within GEE and which were conducted externally:

Within GEE:

Acquisition of Landsat 4-5 TM, 7 ETM+, and 8-9 OLI satellite imagery from GEE’s public repository.
Automated cloud masking algorithms tailored to each sensor to filter cloud-covered pixels and atmospheric noise.
Radiometric calibration to ensure sensor harmonization across datasets.
Median compositing to create representative cloud-free images for each time point.

Post-GEE Analysis:

Land Use/Land Cover (LULC) classification based on the processed Landsat images.
Calculation of Urban Expansion Intensity Index (UEII) and Shannon Entropy for quantitative assessment of urban growth patterns.
Supplementary manual validation to ensure robust analysis and reduce errors in classification.

Justification of Temporal Data Selection and Image Collection Process

We have provided a more detailed explanation of our temporal data selection methodology, ensuring stable observing conditions by selecting target dates that coincide with the dry season (January to July of each year).
To account for seasonal variations, we incorporate a ±6-month window around each target date, ensuring consistent comparisons over time.

Reproducibility and Robustness of the Approach

We now emphasize the computational advantages of GEE, particularly its efficient cloud-based data processing, which allowed us to create a high-quality time series of cloud-free Landsat imagery spanning four decades.
The integration of automated tools with manual oversight ensures a robust and reproducible methodology.

Supporting Figures for Methodological Transparency

Figure 3 now explicitly illustrates the number of Landsat image scenes used in the study, reinforcing the comprehensiveness of our dataset.
This addition improves the visual clarity of our methodology and provides transparency regarding data coverage. Now 2.2 Data Collection was revised as follow

Here, we use Google Earth Engine (GEE), a cloud-based platform that is that is a key component for data acquisition, preprocessing, and initial analysis [33]. We directly accessed life-long-spanning satellite Landsat imagery from the Landsat 4-5 TM, 7 ETM+, and 8-9 OLI sensors from GEE's extensive public repository. However, due to the high cloud cover during the monsoon season, to guarantee low levels of cloud cover and stable observing conditions, target dates for each time point were chosen to coincide with the dry seasons (between January to July of each year). Hence, the process began with the creation of image collections for each five-year interval, incorporating a ±6-month window around the target date to account for seasonal variations. Next, automated cloud masking algorithms, tailored to each sensor, were applied to filter out cloud-covered pixels and atmospheric noise. Then, median composites were made to get clear, representative images for each time point, and GEE's radiometric calibration tools made sure that all the sensors worked together [34]. Besides these automated steps, we also export the processed Landsat images from GEE for LULC classification and more calculations, like the Urban Expansion Intensity Index (UEII) and Shannon Entropy, which are important for the study's objective. This comprehensive data preparation methodology, supported by the computational framework of GEE, resulted in a high-quality time series of cloud-free Landsat imagery spanning nearly four decades. Figure 2 exhibited number of scenes of Landsat images use in this study. This integration of automated tools with manual oversight ensured a robust and reproducible analysis framework. These datasets form the foundation for analyzing urban growth and land cover changes in Phuket, providing insights into the impacts of economic policies and tourism-driven development on the region's landscape.

We hope that these revisions ensure that our methodology is clear, well-documented, and reproducible, addressing the reviewer's concern about whether certain analytical choices were predetermined by GEE or made independently by the authors. By explicitly outlining each step of our methodology, we provide a more transparent and structured framework for readers unfamiliar with GEE.

Comments 4.3: Satellite imagery from LANDSAT is being classified into six categories using a Random Forest model (L139). However, a Random Forest is a supervised algorithm, meaning that data used for training and validation must have been pre-classified. Where did the authors obtain this from? What data years had such “pre-classification” and what years were “predicted” from trained models?

Response 4.3: The reviewer’s concerns regarding the training and validation data used for the Random Forest (RF) classification. We recognize the importance of clarifying the source and temporal coverage of our training data, as well as the distinction between training years and predicted years.

Key Clarifications and Revisions:

Supervised Learning and Training Data Approach

The reviewer correctly notes that RF is a supervised classification algorithm, requiring labeled training data. However, pre-classified data is not required for every year in the study period.
As per standard remote sensing classification methodology, we only require labeled data for selected training years, from which the trained model is then used to classify subsequent years.

Source of Labeled Training Data

We explicitly clarify that training data were manually created by categorizing land cover classes using the following high-resolution reference datasets:

Google Earth Pro high-resolution imagery
Open Street View for ground truth validation
Historical maps for tracking land cover transitions
Field survey data collected to refine classification accuracy
Expert validation to ensure the reliability of labeled data

Temporal Coverage of Training and Predicted Years

Training Data Years: Manually labeled data were created for each classification period, ensuring that key land cover transitions were captured.
Predicted Years: Once trained, the RF model classified land cover for other years without labeled data, ensuring a consistent and scalable classification process.

Enhancements to Manuscript for Transparency

We have revised the methodology section to clearly differentiate between training and predicted years and explicitly state our training data sources.
This ensures methodological transparency and helps readers unfamiliar with RF classification techniques understand how land cover classification was conducted. Revision is presented in 2.3 LULC Classification L136-139:

"We created labeled training data for each classification period by manually categorizing land cover classes using high-resolution imagery from Google Earth pro and supplementary datasets, including Open Street view, historical maps, field survey data, and expert validation"

Comments 4.4 : I could not follow the authors’ adaptation of the Shannon entropy index, and I do not see this use from Dou & Han (2022) that the authors are referencing. If you’re determining p_i by “dividing each year's urban area by the total urban area over all years” (L163), and summing them up for all years to get H_n, shouldn’t there be a single index over 37 years? Or does n expand each year to include previous years? Further explanations are needed.

Response 4.4: We appreciate the reviewer’s feedback regarding our use of the Shannon Entropy Index and the need for a clearer explanation of its calculation methodology. We acknowledge the confusion regarding the reference to Dou & Han (2022) and have corrected it to cite Das and Angadi (2021) instead.

Key Clarifications and Revisions:

Correction of References

The incorrect reference to Dou & Han (2022) has been removed and replaced with the appropriate citation of Das and Angadi (2021).

Explicit Explanation of Shannon Entropy Calculation

To clarify our methodology, we have revised L160-162 to explicitly describe the calculation approach:

We calculate Shannon Entropy (H_n) dynamically for each year rather than as a single index over 37 years.
The entropy value expands year by year, capturing cumulative urban expansion patterns rather than a static, one-time calculation.
Instead of computing entropy over a fixed 37-year period, we analyze its progressive trends, reflecting ongoing urban growth patterns in successive years.

Clarification of Urban Expansion Intensity Index (UEII)

Alongside Shannon Entropy, we analyze the Urban Expansion Intensity Index (UEII) to assess spatial urbanization trends.
The integration of both indices provides a comprehensive understanding of urban expansion dynamics in Phuket.

Enhancements to Manuscript for Transparency

We have explicitly defined Shannon Entropy’s calculation formula in the revised methodology section.
We now provide a step-by-step explanation of how urban area proportions are calculated for each time step to ensure clarity. A revision is presented in L160-162:

Here, we are calculating Urban Expansion Intensity Index (UEII) [42,43] and Shannon Entropy [44,45] and analyzing their changes and trends, offered a comprehensive understanding

Comments 4.5 Table 1 mentions Low/Moderate/High levels of UEII and Entropy. How were these levels determined? In addition, the authors mention certain UEII/Entropy values as “moderate” or high” in section 3.3. But they don’t seem to match the defined Low/Moderate/High levels in Table 1. For example, shouldn’t “moderate UEII of 9.15” (L282) be “High UEII (>2%)”? Further explanations regarding how these cut-offs were determined are needed, and the interpretations of the values also need to match the levels being defined here.

Response 4.5: We appreciate the reviewer’s comments regarding the determination of UEII (Urban Expansion Intensity Index) and Shannon Entropy thresholds and their consistency with interpretations in Table 1 and Section 3.3. We acknowledge the need for a clearer explanation of these classification levels and have revised our manuscript accordingly.

Key Revisions and Clarifications:

Explicit Explanation of UEII Thresholds

We reviewed prior studies and determined that the thresholds for Low (<0.5%), Moderate (0.5%–2%), and High (>2%) UEII are consistent with established methodologies used in similar urban expansion studies.
We have included specific references to relevant literature, including studies by Al-Sharif & Pradhan (2014), Norouzi (2023), Das & Angadi (2021), and Manesha et al. (2021), which support the validity of our cutoff values.

Clarification of UEII Interpretation in the Text

We identified inconsistencies in Section 3.3, where some UEII values were described as “moderate” or “high” without matching the cutoff definitions in Table 1.
These inconsistencies have been corrected to align with the standardized UEII classification, ensuring terminological consistency throughout the manuscript.

Enhancements to Table 1 Explanation

We have updated Table 1’s description to explicitly define how these UEII categories were determined and clarify their relevance in assessing urban expansion trends.
We have now detailed the step-by-step methodology used to derive these thresholds, ensuring transparency in our classification framework.

Shannon Entropy Threshold Explanation

Similar to UEII, we have ensured that Shannon Entropy values are consistently classified in both Table 1 and Section 3.3, eliminating any discrepancies.

Comments 4.6 Throughout the paper, the authors mainly rely on a 5-year timeframe in their findings and results. Although, I’ve noticed several instances where the authors use a shorter 1-year timeframe (e.g., L255) or a longer 10-year timeframe (e.g., section 3.4). Hopping to a shorter 1-year timeframe would be particularly problematic since readers cannot simply “interpolate” 1-year data that’s not being presented in the paper. For example, if “The city area surged from 85 sq km to around 160 sq km during the period of 2005-2006” (L255), how is this land use comparable to 2003 or 2004 levels? I suggest either providing these more granular data as yearly line graphs or unifying the temporal unit of analysis being used throughout the paper.

Response 4.6: We corrected as suggested. We have unified the temporal unit of analysis throughout the paper as highlighted in pink. This adjustment ensures consistency in presenting and interpreting the data, addressing the variations in timeframes that were previously noted. By aligning the analysis within a consistent temporal framework, the findings are now more cohesive and easier for readers to follow.

L109-113: It should be noted that due to the data availability, we have made a modest adjustment to the analysis timeframe for the periods 1987–1990 and 2021–2024. The analysis for 1987–1990 was initiated from this year due to the insufficient satellite imagery available prior to 1987. In the same vein, the most recent data available for the period 2021–2024 only encompassed 2024

L244-248: 1996-2000. The early 2000s witnessed growth with the urban area reaching 223.5 square kilometers during 2006 - 2010 period. This expansion trend. Throughout 2016-2020, Phuket urban area had grown to encompass 312.8 square kilometers reflecting ongoing development pressures and urban sprawl. By the period of 2021- 2024,

L263-264: In the period of 1996-2000

L269-271: In the 2001-2005, there was a resurgence in growth marked by significant changes following the devastating impact of the 2004 Indian Ocean Tsunami. The city area surged from 85 sq km to around 160 sq km during the 2006-2010 period due to intensive recons

L275: In the 2021- 2024

L320-321: The period around the year 2000-2005 marked a phase of urban growth for Phuket. Between 2000-2005 and 2006-2010 there was development within

L325-328: Moving into the decade from 2011-2015 to 2016- 2020 there was a shift towards consolidating and densifying spaces. High density projects became more prevalent in central city areas. Suburbs also experienced population densities supported by enhanced infrastructure and transportation systems. The recent period spanning from 2016-2020 and 2021-2024 demonstrates a more balanced approach, to urban expansion focusing on sustainable

- Some minor points that need attention

- In Figure 4, the colors for Agriculture and Mangrove seem to be identical and hence cannot be distinguished in the panel (a).

We corrected as suggest. Now, Figure 5

- In Figure 6, the authors used discrete colors to distinguish the urban expansion areas. I would instead recommend graduated/continuous colors to easily distinguish urban expansion over time.

We used continuous colors as suggested but for some reason the map didn't came out well, we designed to use discrete colors.

- I suggest adding row-wise percentages in Table 2 for each land use category. Section 3.1 is describing these percentages and would be helpful for readers to see them side by side in the table.

We added row-wise percentages in Table 2 as suggested.

We rewrited L290-291

The Shannon Entropy rose to 2.29 reflecting increased diversity in urban area distribution patterns [44]

and L301-302:

Despite this, the Shannon Entropy remained high at 2.31, indicating stable diversity in the spatial distribution of urban areas.

Comments on the Quality of English Language

We correct as the reviewer suggested.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Based on tourism-induced urbanization in Phuket, Thailand, between 1987 and 2024, this study evaluated the impact of tourism and economic policies on land use and land cover change by integrating Landsat satellite imagery and advanced analytical methods.

In the introduction section, this study points out the transformation of Phuket from tin mining industry to global tourism hub, which has had a significant impact on land use and urban expansion. In addition, through the review of relevant researches on urbanization, it is pointed out that urbanization has a profound impact on land use and land cover patterns, and urbanization patterns are used as key indicators of regional economic development and spatial growth characteristics. However, this manuscript fails to clearly point out the shortcomings of the existing research, that is, the research gap, and proposes to further emphasize the uniqueness of Phuket as a tourism-driven urbanization and its differences compared with other urbanization models, so as to put forward the research gap more clearly.

In the section of materials and methods, Phuket is chosen as a typical case first, mainly because of its unique geographical characteristics and economic transformation background, which is in line with the research goal of demonstrating the impact of tourism on urbanization. Secondly, in terms of data sources, the data sources are mainly Landsat 4-5 TM, Landsat 7 ETM+ and Landsat 8-9 OLI datasets. As for the research methods, Google Earth Engine (GEE) is mainly used for cloud data processing, Random Forest (RF) model is used for classification. As well as the Urban Expansion Intensity Index (UEII) and Shannon Entropy indicators to measure the intensity and diversity of urban expansion, the overall selection is reasonable, but it is recommended to further supplement the comparison of different classification algorithms. And why the UEII and Shannon Entropy metrics are best suited to measure Phuket's urban sprawl characteristics.

In the results and discussion section, in the presentation of the results, the research revealed the significant reduction of forest and mangrove cover, and detailed the changes of land use and land cover in different time periods, as well as the analysis of the intensity and diversity of urban expansion. On this basis, manuscript discusses the drivers of urban growth, including population growth, economic development and urban planning initiatives, noting the concern of urban sprawl over the loss of forest cover. However, the policy recommendations are more general, and lack of specific measures and suggestions. The suggestions can be combined with the results of this study and the discussion part to put forward more specific and operational policy suggestions, provide specific implementation steps and possible challenges and solutions, so as to enhance the practical application value of the suggestions.

Author Response

Comments and Suggestions for Authors

Comments 1: In the introduction section, this study points out the transformation of Phuket from tin mining industry to global tourism hub, which has had a significant impact on land use and urban expansion. In addition, through the review of relevant researches on urbanization, it is pointed out that urbanization has a profound impact on land use and land cover patterns, and urbanization patterns are used as key indicators of regional economic development and spatial growth characteristics. However, this manuscript fails to clearly point out the shortcomings of the existing research, that is, the research gap, and proposes to further emphasize the uniqueness of Phuket as a tourism-driven urbanization and its differences compared with other urbanization models, so as to put forward the research gap more clearly.

Response 1: We value the reviewer’s insightful feedback concerning the necessity to explicitly delineate the research gap and underscore the distinctiveness of Phuket’s tourism-driven development relative to other urbanization paradigms. In reaction to your remark, we have amended the Introduction section (L73-96) to clearly:

Current urbanization research predominantly emphasizes economic zones, transportation systems, and commercial centers, neglecting the influence of seasonal tourism fluctuations, policy-induced economic shifts, and topographical limitations on urban growth in tourism-reliant islands.

This study addresses key research gaps, including:

In what ways does tourism-driven urbanization contrast with traditional urbanization models?
How can economic policies and external crises (such as the 1997 Asian Financial Crisis, the 2004 Tsunami, and COVID-19) influence urban growth patterns?
In what ways do land alterations driven by tourism affect long-term sustainability and resilience in Phuket?

This work seeks to address the research gap by examining spatiotemporal urbanization patterns in Phuket from 1987 to 2024, utilizing remote sensing methodologies, machine learning categorization (RF models), and urban growth indices.

We employ Google Earth Engine (GEE) for cloud-based processing of Landsat images, facilitating long-term observation of land use alterations. Utilize the Urban Expansion Intensity Index (UEII) and Shannon Entropy to measure urbanization intensity and spatial complexity. Subsequently, we analyze the impact of economic policies, tourism surges, and external disturbances on land cover changes in Phuket.

This research establishes a fresh analytical paradigm that connects urbanization studies with tourism-driven spatial growth, relevant to other coastal economies reliant on tourism. The results will be essential for urban planners, legislators, and sustainability experts aiming to reconcile economic growth with environmental preservation in swiftly changing coastal areas.

Comments 2. In the section of materials and methods, Phuket is chosen as a typical case first, mainly because of its unique geographical characteristics and economic transformation background, which is in line with the research goal of demonstrating the impact of tourism on urbanization. Secondly, in terms of data sources, the data sources are mainly Landsat 4-5 TM, Landsat 7 ETM+ and Landsat 8-9 OLI datasets. As for the research methods, Google Earth Engine (GEE) is mainly used for cloud data processing, Random Forest (RF) model is used for classification. As well as the Urban Expansion Intensity Index (UEII) and Shannon Entropy indicators to measure the intensity and diversity of urban expansion, the overall selection is reasonable, but it is recommended to further supplement the comparison of different classification algorithms. And why the UEII and Shannon Entropy metrics are best suited to measure Phuket's urban sprawl characteristics.

Response 2: We value the reviewer's astute observations concerning the methodological decisions in our research. We offer a comprehensive explanation to substantiate the use of the Random Forest (RF) model for classification, as well as the Urban Expansion Intensity Index (UEII) and Shannon Entropy for quantifying urban sprawl. We assert that the RF model is the most appropriate selection for our study. Our literature evaluation indicates that we do not include comparisons with alternative algorithms, as previous research has consistently shown RF's higher efficacy in remote sensing applications. In light of these advantages, we assert that RF is the optimal selection and refrain from making superfluous comparisons with alternative methods.

The Urban Expansion Intensity Index (UEII) and Shannon Entropy are employed to assess urban sprawl in Phuket, as they effectively encapsulate both the intensity and spatial complexity of urbanization, particularly in tourism-centric economies. The UEII is optimal for assessing land transformation generated by tourism in Phuket, as it quantifies urbanization. UEII differentiates between high-growth tourism areas, such as coastal hotel and resort complexes, and slower-growing inland regions by assessing the percentage of land transitioned to urban use during a certain timeframe. This distinction elucidates how tourism-related economic activities affect land utilization over time. Shannon Entropy quantifies urban sprawl and spatial complexity, which is significant in the irregular, tourism-driven expansion of Phuket. Phuket's development has been fragmented and spread since the construction of tourism infrastructure, including hotels, commercial centers, and recreational facilities. Shannon Entropy assesses the compactness or dispersion of urban growth, elucidating the spatial structure of tourism-related urban expansion. Prior research on urban expansion in tourism-dependent economies has validated both indicators, enhancing their significance in our analysis. They provide a comprehensive examination of the dynamics of urbanization in Phuket, encompassing the intensity and spatial distribution of land cover changes generated by tourists. We amended section 2.3 regarding LULC Classification as illustrated in lines 152-158.

The RF algorithm, recognized for its methodology, was particularly appropriate for our research because of its capacity to handle extensive datasets and its resilience to disruptions such as noise and outliers typically present in remote sensing data [5,36]. We created labeled training data for each classification period by manually categorizing land cover classes using high-resolution imagery from Google Earth pro and supplementary datasets, including Open Street view, historical maps, field survey data, and expert validation.

Comments 3. In the results and discussion section, in the presentation of the results, the research revealed the significant reduction of forest and mangrove cover, and detailed the changes of land use and land cover in different time periods, as well as the analysis of the intensity and diversity of urban expansion. On this basis, manuscript discusses the drivers of urban growth, including population growth, economic development and urban planning initiatives, noting the concern of urban sprawl over the loss of forest cover. However, the policy recommendations are more general, and lack of specific measures and suggestions. The suggestions can be combined with the results of this study and the discussion part to put forward more specific and operational policy suggestions, provide specific implementation steps and possible challenges and solutions, so as to enhance the practical application value of the suggestions.

Response 3: We appreciate the reviewer’s insightful comments and constructive feedback regarding the policy recommendations. Based on the suggestion to provide more specific and operational policy measures, we have thoroughly revised Section 4.4, “Policy Recommendations for Urban Development in Phuket,” to align closely with the findings of our study and enhance the practical application value of the recommendations.

4.4 Policy Recommendations for Urban Development in Phuket

Environmental resilience is also important. These include the destruction of essential ecosystems for tourism-driven development, for instance, the destruction of coastal mangroves that are natural storms barriers. Increasing the effectiveness of the enforcement of conservation laws and incorporating the restoration of the mangroves into the climate adaptation strategies could restore these vital systems. One of the frequent natural disasters in Phuket is a flash flood. Here, green infrastructure such as permeable pavements, urban parks and rooftop gardens might also be strengthened to help in fighting flooding while enhancing biodiversity and improving air quality. Further environmental impact assessment of all projects (from the high-end hotel to small resort) could also help to reduce the ecological effects. Economic diversification is therefore necessary to reduce over reliance on the tourism sector. Creating other sectors like technology, healthcare and education through innovation hubs or tax incentives makes the economy stronger against global shocks. However, sustainable tourism models for example, agrotourism partnerships and eco-tourism activities can be used to conserve agricultural land and earn money for conservation. Success also relies on democratic government. This ensures that citizens, businesses, and environmental NGOs are involved in the decision-making process and therefore, the process is transparent and participatory. Climate change threats such as sea level rise and resource competition must be addressed by policy makers and resilience must be addressed in every decision made on development. Therefore, through the integration of smart zoning, equitable housing, green infrastructure and economic innovation, Phuket can create a path for sustainable development to sustain its natural assets, improve the living standard of its citizens and build long term strength.

Reviewer 3 Report

Comments and Suggestions for Authors

The paper titled "Tourism-Induced Urbanization in Phuket Island, Thailand (1987–2024): A Spatiotemporal Analysis” uses GEE to analyze satellite images and introduces the urbanization process in Phuket Island, Thailand. The authors detect land use land cover change combined with economic and environmental occurrences to understand land use dynamics and urban expansion from a historical point of view. Here are some concerns about the paper:

1. The LULC categories contain Urban areas, Water bodies, Agricultural zones, Forested areas, Barren lands and Mangrove forests. What's the difference between the Agriculture and Barren lands, and also between Forest and Mangrove? Is that possible to add a confusion matrix of classification?

2. In Figure 4 (also Figure 3), the color of agriculture and mangrove lands are both in orange?

3. In Figure 5, the urban area looks very intensive in 2005. Did the southeast area have denser development than the later years? Or, is that the image data quality issue?

4. Color in Figure 6 is a bit of a mess. Combined with Figure 5, it's hard to observe the urban growth pattern as described in Sec 3.4. Maybe authors could modify the Figure 1, study area map, with adding major transportation routes/city cores, also landscape features like elevation as the natural terrain is the major constraint of urban development.

5. Beyond the land cover percentage change in Figure 4, land use changes are not well established among different types of land. Is the urban sprawl along with the deforestation or converting from the agricultural or barren land? Flow chart of land-use and land cover change directions is recommended.

6. The timeline will be more clear if authors could add a line plot with time of occurrences and urban dynamics. For example, the x-axis is the year, and the y-axis is the total/new urban area, then using the vertical line to refer to occurrences such as the Financial Crisis in 1997 and 2008. This would be a more clear visual of impacts.

7. The title uses “Tourism-Induced Urbanization”, but from the satellite images, it’s hard to claim the urbanization driven factors. If authors want to discuss the socio-economic impacts on the urbanization process, more data are needed to support the study such as the economic investment of tourism sectors, employees in the tourism industry, or number of hotel/resort/attractions.

Author Response

We sincerely thank the reviewer for their valuable feedback and constructive suggestions, which have helped us improve the clarity and robustness of our study, “Tourism-Induced Urbanization in Phuket Island, Thailand (1987–2024): A Spatiotemporal Analysis.” Below, we provide detailed responses to each comment and outline how we have addressed these concerns in the revised manuscript.

Comment 1: The LULC categories contain Urban areas, Water bodies, Agricultural zones, Forested areas, Barren lands and Mangrove forests. What's the difference between the Agriculture and Barren lands, and also between Forest and Mangrove? Is that possible to add a confusion matrix of classification?

Response 1: We have clarified these distinctions in the manuscript. Agricultural zones are characterized by cultivated or actively farmed land with significant vegetation, whereas barren lands represent areas with minimal vegetation or exposed soil due to erosion, mining, or construction activities. Similarly, forests and mangroves are differentiated based on ecosystem type; mangroves are salt-tolerant forests found in coastal zones. To address classification accuracy concerns, we have added a confusion matrix in the supplementary material to validate the performance of our LULC classification using the Random Forest (RF) model. This matrix demonstrates high accuracy for all categories, ensuring the reliability of the classification results.

Comment 2: In Figure 4 (also Figure 3), the color of agriculture and mangrove lands are both in orange?

Response 2: We acknowledge the oversight in color differentiation. In the revised manuscript, the color schemes for Figures 3 and 4 (now revised to Figures 4 and 5) have been adjusted to ensure distinct visual representation for all LULC categories, particularly between agriculture and mangroves. These modifications enhance the interpretability of the figures.

Comment 3: In Figure 5, the urban area looks very intensive in 2005. Did the southeast area have denser development than the later years? Or, is that the image data quality issue?

Response 3: The dense urbanization observed in 2005 reflects actual development trends in the southeast, driven by rapid tourism and commercial expansion during that period. We have re-evaluated the imagery and confirmed its accuracy. To avoid potential misinterpretations, additional explanations have been included in the results section to contextualize the density patterns observed.

Comment 4: Color in Figure 6 is a bit of a mess. Combined with Figure 5, it's hard to observe the urban growth pattern as described in Sec 3.4. Maybe authors could modify the Figure 1, study area map, with adding major transportation routes/city cores, also landscape features like elevation as the natural terrain is the major constraint of urban development.

Response 4 : In the revised manuscript, we have updated Figure 1(now Figure 2) to include major transportation routes, city cores, and key topographical features, such as elevation. These additions provide context for the constraints and drivers of urban expansion. Furthermore, the color scheme in Figure 6 (now Figure7) has been improved for clarity, allowing readers to better observe urban growth patterns.

Comment 5: Beyond the land cover percentage change in Figure 4, land use changes are not well established among different types of land. Is the urban sprawl along with the deforestation or converting from the agricultural or barren land? Flow chart of land-use and land cover change directions is recommended.

Response 5: We appreciate the reviewer’s suggestion to visualize land-use and land-cover (LULC) change directions using a flowchart. While we acknowledge the potential benefits of such a visualization, we believe that the detailed LULC percentage changes provided in Table 2 already address the dynamics of land-cover transitions effectively. Table 2 comprehensively illustrates the shifts in land-cover categories over time, including urban areas, forest, agricultural zones, barren lands, water bodies, and mangroves.

Additionally, the percentage changes presented in Table 2, coupled with our narrative in Section 3.1, clearly describe the patterns of urban sprawl and its relationship with deforestation, agricultural conversion, and barren land reclamation. For example, the reduction in forest and agricultural areas is directly linked to urban expansion, which is further elaborated in the discussion section. These data-driven insights provide a nuanced understanding of land-use dynamics without necessitating a separate flowchart.

To improve clarity, we have enhanced the discussion in Section 3.1 by explicitly highlighting key transitions (e.g., deforestation contributing to urban expansion) and their driving forces, ensuring a better connection between the LULC changes in Table 2 and the urbanization patterns discussed in the text.

Comment 6: The timeline will be more clear if authors could add a line plot with time of occurrences and urban dynamics. For example, the x-axis is the year, and the y-axis is the total/new urban area, then using the vertical line to refer to occurrences such as the Financial Crisis in 1997 and 2008. This would be a more clear visual of impacts.

Response 6:We have added a timeline plot (Figure 1) in the revised manuscript, illustrating the total and newly developed urban areas over time. Key events, such as the 1997 Asian Financial Crisis, the 2004 Indian Ocean Tsunami, and the COVID-19 pandemic, are marked on the timeline. This figure visually connects urbanization dynamics with these occurrences, enhancing the discussion of their impacts in Section 3.4.

Comment 7: The title uses “Tourism-Induced Urbanization”, but from the satellite images, it’s hard to claim the urbanization driven factors. If authors want to discuss the socio-economic impacts on the urbanization process, more data are needed to support the study such as the economic investment of tourism sectors, employees in the tourism industry, or number of hotel/resort/attractions.

Response 7: We appreciate the reviewer’s comment and agree that satellite imagery alone may not fully establish the socio-economic drivers of urbanization. However, our study primarily focuses on the spatiotemporal analysis of land-use and land-cover (LULC) changes over nearly four decades. The phrase "Tourism-Induced Urbanization" in the title reflects the well-documented historical and contextual evidence of tourism's role in shaping Phuket’s urban landscape, which is extensively discussed in the manuscript.

In Section 1 (Introduction) and Section 4.2 (Phuket Changing Urban Landscape), we highlight key economic and tourism-driven developments, including the decline of tin mining, the rise of tourism as Phuket’s dominant industry, and major infrastructure projects such as the expansion of Phuket International Airport. These contextual elements underscore tourism's significant role in driving urban expansion.

While we acknowledge the merit of integrating additional socio-economic data, such as tourism revenue, employment trends, and the distribution of hotels and resorts, these data are beyond the scope of our current remote-sensing-focused methodology. Instead, we have supported our claims by correlating urban growth patterns with significant tourism-related milestones (e.g., "Amazing Thailand" campaigns, tsunami recovery efforts, and the Phuket Sandbox program). These correlations provide a narrative link between tourism and urbanization trends. Additionally, we have strengthened the discussion in Section 4.2 to include more detailed connections between urban expansion and tourism-related factors, using the existing evidence from historical and policy narratives.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for taking the time to revise the manuscript based on the reviewer’s comments.

I believe that the consistency and clarity of the manuscript have been considerably improved after the revision. Please see the following comments, which focus on points that have not been addressed in this revision.

* Addition of Figure 1 and revision of Figure 2 is certainly helpful in understanding how a) Thailand’s tourism industry grew over time, and b) where major tourism attractions are located in Thailand.

* Although the present study still does not directly examine the relationship between such growth in Thailand tourism and changes in land use. To put it differently, the authors have presented circumstantial evidence that the growth of the Thailand tourism industry could be one reason that impacted the observed changes in land use, but have yet to present substantive and direct evidence to support the urbanization is tourism-induced.

* Since this argument is the very core of the present research, I'll have to repeat the previous comment that the authors need to present the evidence to support whether LULC is the result of growth in tourism.

* Here are some examples/ideas on how the authors could approach this issue by looking at the relationship temporally or spatially:

* Can the time series of tourist visitation predict changes in the LULC categories? Analysis techniques such as Vector Autoregression could be used to answer such questions; or

* Are the urban expansion areas (shown in Figures 6 and 7) spatially correlated with areas that have tourist attractions? Spatial models such as GWR would be helpful in directly examining these questions.

* The authors mention that Figure 5 has been changed. Although it still has categories with duplicate colors, where Agriculture and Mangrove categories are using the same color. Could you recheck whether the change has been made in the manuscript?

* The rest of revisions appear to adequately address the previously raised concerns.

Best luck to the authors.

Author Response

reviewer 1

Thank you for taking the time to revise the manuscript based on the reviewer’s comments.

Addition of Figure 1 and revision of Figure 2 is certainly helpful in understanding how a) Thailand’s tourism industry grew over time, and b) where major tourism attractions are located in Thailand.

Comment #1:

Although the present study still does not directly examine the relationship between such growth in Thailand tourism and changes in land use. To put it differently, the authors have presented circumstantial evidence that the growth of the Thailand tourism industry could be one reason that impacted the observed changes in land use, but have yet to present substantive and direct evidence to support the urbanization is tourism-induced. Since this argument is the very core of the present research, I'll have to repeat the previous comment that the authors need to present the evidence to support whether LULC is the result of growth in tourism. Here are some examples/ideas on how the authors could approach this issue by looking at the relationship temporally or spatially:

Response #1

We appreciate the reviewer’s insightful comments regarding the need for more direct evidence on whether urbanization is tourism-induced. To address this, we have expanded our analysis by incorporating annual urban area data derived from Landsat imagery. This allows us to examine its relationship with the number of visitors and tourism revenue over time. Furthermore, to enhance our understanding of the potential influence of tourism on urban expansion, we have included additional economic and demographic variables in our analysis. Specifically, we have collected data on the Gross Provincial Product (GPP) for the service sector in Phuket, as well as population density, which includes both registered residents and migrant workers, who are likely employed in the tourism sector. Additionally, we have considered housing density, as an indicator of residential expansion. By integrating these factors, we aim to provide a more comprehensive assessment of whether and to what extent urban expansion is influenced by tourism-related activities. (Appendix A)

Comment #2:

Can the time series of tourist visitation predict changes in the LULC categories? Analysis techniques such as Vector Autoregression could be used to answer such questions; or

Response #2

We try to perform Vector Autoregression but for some reason it turned out the dataset is too small, we then try Laggaed Regression Analysis which is provides direct causal relationships between an independent variable (e.g., visitor numbers) and a dependent variable (e.g., accumulated urban area). Below is the revisdion for a whole new 3.5 setion

3.5 Analysis of Urban expansion and tourism

To understand the effect of tourism on urban expansion, we extent our analysis by extracting urban area data from Landsat imagery using GEE. This allows us to examine its relationship with visitor number over time. Additionally, we incorporated supplementary economic and demographic variables into our analysis. We gathered information on the Gross Provincial Product (GPP) of the service sector in Phuket and the population density, which encompasses both registered residents and migrant laborers who are likely to be employed in the tourism sector. Furthermore, we investigated the density of housing, which serves as a metric for the expansion of residential areas (See Appendix A)

Then, we apply the Lagged Regression Analysis on Urban Expansion and Tourism to investigate the correlation between accumulated urban area expansion and tourism growth (visitor numbers), while integrating economic and demographic variables, including Gross Provincial Product (GPP) in the service sector and housing density. The investigation employed a lagged regression model to forecast urban expansion based on historical data of tourist numbers, GPP, and housing density (Figure 8).

(a)

(b)

(c)

Figure 8 presents a correlation heatmap, revealing strong links between urban expansion, housing density, and economic indicators. (a), the relationship between urban growth, visitor numbers, and economic activity, highlighting long-term urbanization trends. (b) a strong alignment between actual and predicted urban expansion, confirming the model’s accuracy (c)

The regression model (appendix A1.2) indicates that visitor numbers are not statistically significant in forecasting urban expansion at 1-year, 2-year, or 3-year intervals. The one-year lag coefficient (17.14, p = 0.113) indicates a minor although statistically insignificant effect. This outcome suggests that urbanization is not primarily influenced by transient variations in tourism, but instead by a confluence of enduring economic and demographic elements.

The lagged housing density variable (1-year lag) exerted the most significant impact on urban expansion (coef = 6.52, p < 0.0001). This indicates that urbanization is mostly propelled by residential development rather than direct increase related to tourism. The 1-year lagged GPP variable was statistically significant (p = 0.021) and had a negative coefficient (-0.0017). This may suggest that heightened economic activity could impede urban expansion, potentially due to rising land prices, legislative changes, or congestion impacts. The regression model accounted for 99.3% of the variance in urban expansion (R² = 0.993, Adjusted R² = 0.991). The F-statistic (489.0, p < 0.0001) indicates that the model is very significant and appropriately fitted (appendix A1.3). The map comparing anticipated and actual accumulated urban area demonstrates that the model accurately reflects real-world urbanization trends.

Urban expansion adheres to a protracted trajectory, predominantly shaped by housing development and economic conditions rather than transient tourism swings. Although tourism fosters economic growth, its direct influence on urban expansion is neither quick nor substantial. Policymakers must prioritize the equilibrium between tourism expansion and housing and infrastructure policy to facilitate sustainable urban development.

Comment # 3:

Are the urban expansion areas (shown in Figures 6 and 7) spatially correlated with areas that have tourist attractions? Spatial models such as GWR would be helpful in directly examining these questions.

Response #3

We perform Kernel Density analysis for tourist attractions.Consequently, the pattern of urban expansion does not consistently correspond with the distribution of tourist attractions. Although tourism may play a role in urban growth, the relationship is intricate and shaped by a multitude of factors, including economic development, infrastructure expansion, and land use policies. These findings imply that their spatial distribution does not provide meaningful insights into the patterns of urban expansion. The results merely confirm that tourist attractions located within urban areas remain in urban areas, without offering any explanatory power regarding the relationship between tourism and urbanization. Consequently, the spatial distribution of attractions does not serve as a significant factor in explaining urban expansion patterns in Phuket.

Comment#4:

The authors mention that Figure 5 has been changed. Although it still has categories with duplicate colors, where Agriculture and Mangrove categories are using the same color. Could you recheck whether the change has been made in the manuscript?

Response # 4

We appreciate the reviewer’s observation regarding the duplicate colors in Figure 5. We have rechecked and updated the figure to ensure that each category is distinctly represented. The revised figure is now included in the updated manuscript.

The rest of revisions appear to adequately address the previously raised concerns.

Best luck to the authors.

Thank you once again for your time and thoughtful suggestions. We hope that these revisions satisfactorily address your concerns and look forward to your feedback.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

In this version of the manuscript, the authors have made substantial improvements, and basically responds well to the comments. In this regard, I would like to give minor revision suggestions, except for the following suggestions:

1. In the discussion section, the impact of specific economic policies and tourism initiatives on urbanization of Phuket can be analyzed in more depth. For example, a detailed analysis of how the "Wonderful Thailand" campaign or the Phuket Sandbox initiative directly affects land use and urban expansion would make the research more convincing.

2. While the loss of forest and mangrove cover is mentioned, the specific environmental consequences of these changes could be discussed in more detail, such as impacts on biodiversity, ecosystem services and climate change mitigation efforts. When discussing environmental resilience, it can be further expanded to specific strategies to strengthen islands against future shocks, such as sea level rise and extreme weather events.

3. It is suggested to provide some other tourism-driven urbanization models of coastal areas for comparison. This will provide a broader context for Phuket's urbanisation and highlight the unique challenges and opportunities it faces.

4. In the presentation of results, is it possible to add some additional maps or visualizations to show the distribution of urban expansion intensity and Shannon entropy over time, which will help to understand urbanization dynamics more intuitively?

Author Response

Reviewer_2

We sincerely appreciate your constructive feedback and the time you have taken to review our manuscript. We are pleased that you found our revisions to be substantial and that they adequately addressed most of the previous comments. Below, we provide a detailed response to your remaining suggestions, along with explanations of the modifications made to the manuscript.

Comment 1. In the discussion section, the impact of specific economic policies and tourism initiatives on urbanization of Phuket can be analyzed in more depth. For example, a detailed analysis of how the "Wonderful Thailand" campaign or the Phuket Sandbox initiative directly affects land use and urban expansion would make the research more convincing.

Response 1

We appreciate your suggestion to provide a more detailed discussion of the impact of economic policies and tourism initiatives on Phuket's urbanization. In response, we have expanded our discussion section to include an analysis of key initiatives such as Amazing Thailand (1998), Visit Thailand Year (2015), and the Phuket Sandbox (2021).

Specifically, we discuss how these initiatives accelerated real estate and infrastructure development, leading to increased land-use changes in Phuket. We highlight the impact of financial incentives for tourism investment, the easing of foreign ownership laws, and the diversification of the economy through the Smart City Initiative (2015), which introduced digital innovation and sustainable urban planning.

These developments have influenced land-use objectives by shifting focus toward urban densification and mixed-use developments in inland areas such as Kathu and Chalong. Meanwhile, rapid tourism growth associated with the Phuket Sandbox initiative has driven infrastructure upgrades, though challenges related to resource management, environmental conservation, and sustainable tourism remain significant.

These additions provide a clearer link between economic policies and urban expansion, making the analysis more robust. The revised discussion can be found in Section 4.2 L468-488 of the manuscript.

The island had a substantial rise in the development of resorts and condominiums due to initiatives like Amazing Thailand (1998) and Visit Thailand Year (2015). The campaigns provided financial incentives for tourism investment. Consequently, they boosted tourism and economic diversification, leading to increased investment in tourism infrastructure and real estate expansion. The expansion of luxury home projects in sought-after locations like Laguna and Kata Beach was greatly impacted by these policies. The alteration of Phuket's land-use patterns was expedited by the easing of foreign ownership laws for real estate, drawing international purchasers and developers [60,61]. The Smart City Initiative, however, launched in 2015, marked a decisive shift in Phuket's developmental path by emphasizing sustainable urban planning and digital innovation. The island's economy has diversified with investments in technological parks, transportation facilities, and upscale residential complexes, thus reducing its reliance on conventional tourism. These shifts have also impacted land-use objectives, placing increased focus on urban densification and mixed-use developments in inland areas such as Kathu and Chalong. While Phuket Sandbox initiative, allowing vaccinated international tourists to visit without quarantine, sparked a surge in tourism-driven businesses and real estate investment [62,63]. The government expedited infrastructure upgrades, but faced challenges related to resource management, environmental conservation, and sustainable tourism. Both campaigns highlight the complex relationship between economic growth, urban planning, and environmental sustainability[64].

Comment 2. While the loss of forest and mangrove cover is mentioned, the specific environmental consequences of these changes could be discussed in more detail, such as impacts on biodiversity, ecosystem services and climate change mitigation efforts. When discussing environmental resilience, it can be further expanded to specific strategies to strengthen islands against future shocks, such as sea level rise and extreme weather events.

Response 2

We acknowledge the importance of discussing the broader environmental consequences of land-use changes, particularly in relation to biodiversity loss, ecosystem services, and climate change mitigation. However, the primary focus of our study is on land-use change analysis and spatial patterns, and a comprehensive evaluation of these environmental factors would require additional data and methodological approaches beyond the scope of this research.

While we recognize that these issues are critical, we have briefly acknowledged their importance in the discussion section and have suggested that future research could explore these aspects in greater detail. This ensures that the manuscript remains focused on its core objectives while recognizing avenues for further investigation.

Comment 3. It is suggested to provide some other tourism-driven urbanization models of coastal areas for comparison. This will provide a broader context for Phuket's urbanisation and highlight the unique challenges and opportunities it faces.

Response 3

We appreciate the suggestion to compare Phuket with other tourism-driven coastal urbanization models. However, to the best of our knowledge, Phuket represents a unique case due to its distinct combination of geographic, economic, and regulatory factors.

Unlike many other tourism-driven coastal cities, Phuket's rapid urbanization is occurring within a constrained island environment, with specific regulatory, infrastructural, and environmental challenges that make direct comparisons difficult. Differences in governance structures, land use policies, and economic dependencies further complicate drawing direct parallels with other destinations.

For these reasons, our study remains focused on analyzing Phuket's urbanization within its own context, ensuring that the findings are directly applicable and relevant. Nonetheless, we acknowledge that future research could explore comparative case studies in greater depth.

Comment 4. In the presentation of results, is it possible to add some additional maps or visualizations to show the distribution of urban expansion intensity and Shannon entropy over time, which will help to understand urbanization dynamics more intuitively?

Response 4

We appreciate this recommendation and have carefully examined the feasibility of adding additional maps and visualizations. However, upon further analysis, we found that the generated maps did not reveal significant spatial patterns that would meaningfully enhance the understanding of urbanization dynamics.

To avoid potential misinterpretation or confusion, we have chosen not to include these additional maps in the manuscript. Instead, we have focused on presenting the results through quantitative analyses that provide clearer and more interpretable insights. This decision ensures that our findings are conveyed with the highest possible clarity and precision.

We thank you again for your valuable time and thoughtful suggestions, and we hope that our revisions satisfactorily address your concerns.

Author Response File: Author Response.pdf

Article Menu

Tourism-Induced Urbanization in Phuket Island, Thailand (1987–2024): A Spatiotemporal Analysis

Further Information

Guidelines

MDPI Initiatives

Follow MDPI