Landscape Zoning Strategies for Small Mountainous Towns: Insights from Yuqian Town in China

Abstract

Small towns in mountainous regions face significant challenges in formulating effective landscape zoning strategies due to pronounced landscape fragmentation, which is driven by both the dominance of large-scale forest resources and the lack of coordination between administrative planning departments. To tackle this problem, this study focused on Yuqian, a quintessential small mountainous town in Hangzhou, Zhejiang Province. The town’s layout was divided into a grid network measuring 70 m × 70 m. A two-step cluster process was employed using ArcGIS and SPSS software to analyze five landscape variables: altitude, slope, land use, heritage density, and visual visibility. Further, eCognition software’s semi-automated segmentation technique, complemented by manual adjustments, helped delineate landscape character types and areas. The overlay analysis integrated these areas with administrative village units, identifying four landscape character types across 35 character areas, which were recategorized into four planning and management zones: urban comprehensive service areas, agricultural and cultural tourism development areas, industrial development growth areas, and mountain forest ecological conservation areas. This result optimizes the current zoning types. These zones closely match governmental sustainable development zoning requirements. Based on these findings, we propose integrated landscape management and conservation strategies, including the cautious expansion of urban areas, leveraging agricultural and cultural tourism, ensuring industrial activities do not impact the natural and village environment adversely, and prioritizing ecological conservation in sensitive areas. This approach integrates spatial and administrative dimensions to enhance landscape connectivity and resource sustainability, providing key guidance for small town development in mountainous regions with unique environmental and cultural contexts.

1. Introduction

Mountainous terrain constitutes two-thirds of China’s land area and harbors about half of its population in expansive towns and cities, highlighting the pivotal role of these regions in the country’s spatial dynamics [1]. As stipulated by the National New Urbanization Plan (2014–2020) and the rural revitalization strategies, urbanization in these mountainous regions is poised for significant growth over time. This shift of urban functionalities towards mountainous areas is increasingly becoming a discernible trend [2,3,4]. Small towns in these areas are characterized by their diverse terrain, complex spatial interfaces, fragmented landscape patterns, and constrained development land [5]. Together with urban expansion and population growth, these factors intensify the challenges of ecological conservation, landscape development, and planning management.

The urban expansion into mountainous areas necessitates a balanced approach to planning and managing small towns while preserving their landscapes. Numerous scholars have engaged in significant research to support these objectives. Some argue that the rational delineation of landscape characteristic units is a prerequisite for establishing sustainable planning strategies and management methods for landscape resources [2,6]. A case study in the Turkish Mediterranean Antalya examined the transitional zones between urban and rural areas, emphasizing the need for effective planning to maintain agricultural production, food safety, and traditional land use while preserving rural characteristics and the natural environment [2]. Similarly, studies in the Bekasi Regency in Indonesia [7], in Malaysia [8], and in Korea [9], which utilized quantitative zoning methods and a comprehensive urban–rural development framework, demonstrated how different areas can be strategically managed to balance industrial growth with the preservation of rural landscapes and agricultural productivity. At Lushan National Park in China, researchers have identified landscape character types and regions across three scales, facilitating the resolution of spatial and administrative management challenges within the park [10]. Additionally, an analysis in China’s southwestern mountainous regions has mapped the urban–rural–natural transition area pattern, identifying suitable habitats for different spatial groups and socio-economic activities through various developmental stages [11]. However, while these studies have provided valuable insights for large-scale mountainous regions, differences in regional scale, indicator selection, and data precision can make their findings less directly applicable to the landscape zoning of small mountainous towns [12].

Other scholars have explored ecological planning methods tailored to the landscape planning and construction systems of mountainous urban areas, focusing on aspects like ecological restoration [13], path intervention [14], and pattern conservation [15]. For instance, the impact of landscape pattern changes on ecosystem health in the Miaoling area of China has been analyzed using landscape ecology and terrain gradient theory, offering pathways to achieve synergistic effects in regional human-environmental systems [16]. Despite the focus on large-scale forest landscapes in ecological planning, the significance of other landscape resources often remains underemphasized, limiting the applicability of such approaches to small town planning in mountainous areas. Therefore, it is crucial to identify and prioritize the unique landscape characteristics of small towns, scattered villages, farmland, archaeological sites, and gently sloping areas surrounding the hills. In many regions, landscape management is complicated by the oversight of multiple administrative units, leading to a lack of synchronization in the identification, protection, and management of landscape resources [17]. The distribution of natural mountain resources located at the boundaries of rural areas frequently results in management conflicts, with different planning concepts and strategies applied to identical landscape character areas. For instance, while some mountainous resources are allocated for rural tourism projects, others are transformed into tea plantations. Such disparities hinder the ability to harmonize landscape identification and classification efforts across different administrative jurisdictions. This not only undermines the integrity of landscape characteristics but contravenes the principles of comprehensive landscape protection. In response to these challenges, China inaugurated the Ministry of Natural Resources in 2018. This new ministry aims to consolidate and streamline the formerly fragmented or overlapping responsibilities among various departments within the same region, thereby boosting the efficacy of natural resource management

Landscape character assessment (LCA), as a tool for managing landscape changes and supporting decision-making, integrates natural and cultural landscapes with human perceptions, and provides methods for improving, restoring, and managing landscape characteristics to ensure future changes occur in a sustainable manner [2]. In practice, the identification and description of landscape characteristics are divided into three levels: national, regional, and local scales [18]. Key factors representing the distinctive features of the landscape are selected based on the actual conditions of the study area, and then the landscape is decomposed to analyze the constituent elements, structural composition, and their relationship with the overall landscape, aiming to understand how unique elements form local identity. Methodologically, based on the compositional structure and scale characteristics of landscape formation, dominant indicators are typically selected from three dimensions, including natural factors (e.g., climate, topography, vegetation types, land cover), cultural and social factors (e.g., land use, temporal dimensions), and landscape perception factors (e.g., sense of place, openness, aesthetic preferences) [19], with classification performed using overlay analysis or multivariate statistical analysis. Different indicators are suited to different research scales, and the results are influenced by indicator units, precision, and regional scale [12].

Despite advances in LCA, which often transcend administrative boundaries, emphasizing the integrity and coherence of landscape resources [10], gaps remain in the literature concerning the spatial integration of administrative regions at multiple scales [8]. While LCA is designed to preserve the diversity and authenticity of landscapes and to provide unified standards and terminologies for policy formulation [20,21], it faces challenges in addressing policy repetitiveness and inconsistencies. Moreover, although LCA aims to tailor spatial planning to specific landscape requirements [22], and the identification and classification of landscape characteristics serve as prerequisites for establishing effective planning and management systems [23], there is a lack of robust methods tailored specifically for small mountainous towns. Current methodologies may not adequately consider the unique characteristics of these areas or how to integrate landscape resources with the existing administrative framework effectively.

This study aims to fill these gaps by developing a novel method to assess the landscape character of small towns in mountainous areas, with Yuqian Town in Lin’an District, Hangzhou City, Zhejiang Province serving as a case study. This research employs ArcGIS, eCognition software, and a two-step clustering algorithm to analyze and classify the landscape character types within the town’s administrative boundaries. This study seeks to address two main questions: (1) How can mountainous landscape character management units be effectively divided at the scale of small towns? (2) How can landscape resource characteristics be integrated with the existing administrative management system? Our findings contribute to sustainable urban development and management strategies, providing a model that can be replicated in similar contexts.

2. Data Sources and Methods

2.1. Research Object

Yuqian Town, emblematic of the small mountainous towns scattered throughout western Zhejiang Province, spans 261.2 square km and encompasses 30 villages (Figure 1). Characterized by its predominantly hilly and mountainous terrain, the town boasts forest land comprising approximately 86% of its total area, leaving scant available land for construction. In recent years, the local government has articulated a strategic vision aimed at transforming Yuqian into a small, modern ecological city. This vision involves embracing the transfer of urban industries and extending urban functionalities, along with facilitating the spillover of economic and cultural factors from major cities. This strategic shift marks Yuqian Town’s gradual transition from a mere “township” to a burgeoning “small city”, highlighting the imperative to tackle the challenge of limited construction land in mountainous regions while fostering urban development in these areas.

Figure 1 illustrates the geographic setting of Yuqian Town within the broader context of Hangzhou City, Zhejiang Province, China. The main map details the administrative boundaries of Yuqian Town, delineating its internal divisions into village-level administrative units. The insets show the location of Hangzhou within Zhejiang Province and the positioning of Zhejiang within China and Asia, highlighting the regional context. Each administrative village within Yuqian Town is marked for precise spatial reference.

The selection of Yuqian Town as the focal point for this study is justified by several factors. Firstly, Yuqian Town exhibits a blend of mountainous and plain terrain typical of Zhejiang’s mountainous regions, thereby enhancing the generalizability of this study’s findings to similar contexts. Secondly, the town has a rich agricultural heritage dating back to the Song Dynasty, rendering it an exemplary site for exploring rural agricultural cultural landscapes. Thirdly, Yuqian Town has successfully leveraged its regional environment and cultural heritage to develop rural revitalization industries, including tourism, gastronomy, and research-based learning. These developments make it an ideal case for studying the interplay between cultural heritage and contemporary economic initiatives in rural settings.

2.2. Landscape Variables and Data Sources

Landscape research typically employs a range of indicators to identify, assess, or quantify the characteristics or conditions of a specified area [19]. In the case of Yuqian Town, the landscape attributes are assessed from natural physical, human social, and visual perception perspectives. To capture the natural physical attributes, altitude and slope are utilized, both of which are continuous variables. For the human social attributes, land use and heritage density are selected, reflecting socio-economic influences on the landscape. Visual visibility, representing the visual perception attributes, captures how visible and prominent landscape features are from various vantage points. These variables are divided into 23 distinct landscape indicators, each denoted by an uppercase English letter, as detailed in

Table 1. Variables used for general zoning in Yuqian Town.

Variables	Acronym	Variables	Acronym
Altitude		non-visible	V2
plain (<200 m)	H1	land use
hill (200~500 m)	H2	farmland	L1
low mountain (>500 m)	H3	garden land	L2
Slope		forest land	L3
flat slope (<6°)	S1	grassland	L4
gentle slope (6~15°)	S2	commercial and service land	L5
medium slope (15~25°)	S3	industrial and mining storage land	L6
steep slope (>25°)	S4	residential land	L7
Heritage density		public management and public service land	L8
existing	A1	special land	L9
non-existing	A2	transportation land	L10
visual visibility		waters and water conservancy facilities land	L11
visible	V1	other land	L12

and Figure 2.

The data sources are structured as follows: elevation and slope data originate from the 30 m-resolution DEM released by GEBCO in 2023; village boundaries and land use data derive from vector datasets of China’s Third National Land Survey (concluded 2019). These data enabled the classification of visual distance into visible and non-visible zones, facilitating a nuanced understanding of the visual landscape dynamics within the town. Heritage site density is calculated based on the 2023 List of Protected Cultural Heritage Sites published on the Hangzhou Municipal Government’s official portal (

Table 2. Heritage site density in the study area.

No.	Name	Period	Location
1	Tianmu Kiln Sites	Song-Yuan	Shaolu & Lingkou, Yuqian Town
2	Former Site of Minzu Daily	Republic Era	Hecun Hamlet, Houzhu Village, Yuqian Town
3	Qixiang Pagoda	Ming Dynasty	Guanyan Mountain, Yuqian Town
4	Tomb of Fang Keyou Couple	Qing Dynasty	Xujiawan, Genglou Hamlet, Fangyuan Village, Yuqian Town
5	Chang’an Bridge of Nanshan	Qing Dynasty	Nanshan Village, Yuqian Town

), with point features generated through ArcGIS (10.7 Esri, Redlands, CA, USA) spatial conversion. The 70 m × 70 m grid size was determined based on the scale of Yuqian Town and the precision of our data. This size balances detailed local analysis and computational feasibility, ensuring it is appropriate for a town-level study.

2.3. Research Methods

The methodology of this study is designed to analyze the landscape character of Yuqian Town through a systematic approach detailed in three sequential steps, as illustrated in Figure 3. The initial step involves clearly defining the geographic scope of this study using administrative boundaries, ensuring that the outcomes of the landscape character zoning are integrally linked to and supportive of spatial planning efforts [24]. This approach facilitates direct applicability to urban development strategies, incorporating not only the overarching town administrative boundary of Yuqian but extending to encompass the boundaries of all villages under its jurisdiction.

Step 1: Define the study scope.

(1)

Use administrative boundaries to clearly define the geographic scope of this study.

(2)

Employ ArcGIS 10.7 to delineate the boundary of Yuqian Town and all villages under its jurisdiction. This ensures that the landscape character zoning outcomes are integrally linked to and supportive of spatial planning efforts [25], facilitating direct applicability to urban development strategies.

Step 2: Zone the landscape character.

(1)

Classify natural, cultural, and visual variables into grid cells of 70 m × 70 m.

(2)

Conduct a two-step cluster analysis using SPSS.

(3)

Utilize eCognition software to integrate and delineate landscape character subareas.

(4)

Make manual adjustments after the analysis to refine the zoning accuracy.

Step 3: Allocate landscape resources.

(1)

Integrate landscape character areas with administrative village boundaries through overlay analysis.

(2)

Develop a comprehensive plan for landscape character management zones.

In the subsequent phase, the town is methodically divided into 70 m × 70 m grid cells, resulting in an extensive matrix of 51,816 grid cells. Each cell is filled with quantitative and qualitative data on five critical variables, including altitude, slope, land use, site density, and visibility. This granular approach establishes the basic landscape character units of Yuqian Town, laying a structured foundation for further analysis. The evaluation of these units employs the SPSS27 software’s two-step clustering algorithm, known for its effectiveness in managing large datasets and for autonomously determining the optimal number of clusters based on the Bayesian information criterion [25]. After clustering, the numbers are adjusted and integrated back into ArcGIS 10.7 software for visual representation, resulting in a detailed map that delineates various landscape character types across Yuqian Town.

After the landscape character units within Yuqian Town were clustered, the resulting map revealed areas where the landscape character units appeared notably fragmented. Managing these fragmented patches effectively presents a significant challenge in landscape management, particularly from the standpoint of maintaining continuity and coherence across the landscape. To address this issue, it was deemed necessary to integrate these disparate and isolated landscape units while adhering to the principle of boundary integrity. This integration process was facilitated by employing eCognition software, which utilizes a raster-object-based image processing technique used to segment and merge images into various-sized multi-pixel units [26]. The software’s multi-scale segmentation algorithm was particularly useful, allowing for the adjustment of parameters such as scale, shape, and compactness to optimize the merging of landscape units. For this specific analysis, the settings chosen for the scale, shape, and compactness parameters were 10, 0.4, and 0.5, respectively. These values were selected to provide a balance that reflects the actual landscape characteristics while minimizing artificial distortions. However, the initial outcomes from the eCognition software tended to show over-segmentation, indicating an excess of small, disconnected units that did not accurately represent the physical continuity of the landscape. To correct this, manual adjustments were necessary. These adjustments were based on detailed satellite imagery and aimed to correct deviations from the semi-automated delineation results, ensuring that the final landscape character map more accurately reflected real-world conditions [27].

The third step in our study’s methodology focuses on the zoning of landscape character planning and management areas. This step involves integrating the various landscape character types identified earlier with the existing administrative village-level management boundaries within Yuqian Town. This integration is achieved through an overlay analysis that merges areas sharing similar landscape characteristics. The process assigns these unified landscape zones to geographically adjacent administrative villages. This strategic grouping ensures that the resulting landscape character planning and management areas are spatially linked and coordinated, enhancing the efficacy of landscape governance. Each of these designated regions comprises multiple landscape character types, encompassing several administrative villages.

3. Results

3.1. Landscape Character Types

Yuqian Town is categorized into four distinct landscape character types, as illustrated in Figure 4. These types are delineated using a coding system based on the prevalence of landscape indicators: a dominant indicator (over 60% prevalence) is denoted by “X”; a substantial presence (30% to 60%) by “{X}”; and a moderate presence (10% and 30%) by “(X)”. Indicators contributing less than 10% are excluded from this classification.

Figure 4 displays the spatial distribution of four distinct landscape character types within Yuqian Town, classified according to a combination of natural, cultural, and visual variables. The types are coded as follows: type I (H2.S3.A2.V2.L3), type II (H2.S3.A2.V2.L2), type III (H1.S3.A2.A1.V2.V1.L3), and type IV (H1.S2.A2.A1.V2.V1.L3). Each color on the map represents a different landscape character type, illustrating the varied environmental and aesthetic qualities across the town. The map is oriented to north at the top and includes a scale for distance measurement, providing a comprehensive overview of the landscape’s diversity at the municipal scale.

The distribution of these landscape indicators across the different character types is shown in Figure 5, with specific numerical values available in

Table A1. Percentage composition of landscape character types.

	I	II	III	IV
H1	0.002	0.002	0.183	0.198
H2	0.199	0.145	0.015	0.003
H3	0.000	0.050	0.000	0.000
S1	0.000	0.000	0.000	0.037
S2	0.002	0.000	0.000	0.165
S3	0.197	0.029	0.193	0.000
S4	0.001	0.167	0.005	0.000
A1	0.000	0.002	0.044	0.064
A2	0.199	0.195	0.154	0.139
V1	0.000	0.075	0.059	0.071
V2	0.199	0.135	0.139	0.132
L1	0.003	0.001	0.015	0.014
L2	0.003	0.001	0.006	0.012
L3	0.183	0.192	0.172	0.124
L4	0.000	0.000	0.000	0.000
L5	0.000	0.000	0.000	0.001
L6	0.000	0.000	0.000	0.006
L7	0.002	0.001	0.005	0.015
L8	0.000	0.000	0.000	0.002
L9	0.000	0.000	0.000	0.000
L10	0.001	0.001	0.003	0.008
L11	0.008	0.004	0.005	0.009
L12	0.000	0.000	0.000	0.001

. This framework facilitates a nuanced analysis of Yuqian Town’s land use and village landscape, providing a comprehensive overview of each character type’s unique attributes.

The Figure 5 chart illustrates the percentage distribution of various landscape indicators within the four distinct landscape character clusters (I, II, III, and IV) identified in Yuqian Town. Each color in the chart corresponds to a different indicator, as delineated in the legend, ranging from natural features (H1, H2, H3), slope categories (S1 to S4), area types (A1, A2), visual aspects (V1, V2), to a series of twelve landscape layers (L1 to L12). The chart provides a clear visual representation of how these indicators are apportioned among the different clusters, highlighting the unique composition of each cluster.

Landscape character type I in Yuqian Town is distinguished by its hilly slopes, predominantly spanning rural and extensive forest areas. This type features a scattered layout with limited visibility, primarily due to the sparse distribution of villages, thereby reducing the number of available points for visibility analysis. Despite its secluded nature, this area is rich in natural resource characteristics and offers stunning scenic views. Type II is concentrated in the northwest of Yuqian Town, characterized by low mountains and steep slopes, posing a potential risk for landslides. This region is extensively covered by forests, with limited village distribution, resulting in low visibility. A notable feature of this area is the Yingong Reservoir, which plays a crucial role in the ecological conservation of the surrounding mountain forests. Type III exhibits a relatively dispersed distribution, mainly situated on the sloping plains at the town’s outskirts. Although visibility remains low due to the spread of villages, this area is culturally significant, housing historical sites and relics, including the nationally recognized Tianmu Kiln Site and the provincial-level heritage site of the former Ethnic Daily Newspaper. Lastly, type IV encompasses the main urban built-up areas of Yuqian Town, characterized by gentle slopes on the plains. This type is the most visible and socially active, encapsulating nearly all urban functionalities and serving as a central gathering point for the local population.

3.2. Landscape Character Areas

Building on the initial identification of four landscape character types in Yuqian Town, a detailed subdivision resulted in 105 distinct landscape character areas (Figure 6A). Further refinement through manual adjustments led to the delineation of 35 specific landscape character areas, as shown in Figure 6B. Descriptions for each of these areas, based on field surveys (Figure A1) and their respective landscape character types, are systematically catalogued in

Table 3. Descriptions of landscape character areas.

Areas	Types	Descriptions
1, 29, 33, 34	II	Low mountain and hill, steep slope, forest land, non-existing, low visibility.
2, 12, 15, 16	II	Hill, steep slope, forest land, non-existing, non-visible.
4	I	Hill, medium slope, waters and water conservancy facilities land, non-existing, low visibility.
3, 5, 13, 25	I, III	Hill, medium slope, forest land, non-existing, non-visible.
6, 11, 17	IV	Plain, gentle slope, residential land, farmland, high heritage density, High visibility.
7, 8	III	Hill, gentle slope, forest land, high heritage density, low visibility.
9, 10, 28, 23, 32, 35	I, III	Plain, medium slope, forest land, non-existing, non-visible.
14, 26	III	Plain, medium slope, forest land, low heritage density, low visibility.
18	III	Plain, medium slope, forest land, high heritage density, high visibility.
19, 30	IV	Plain, medium slope residential land, industrial and mining storage land, low heritage density, high visibility.
20	III	Plain, medium slope residential land, non-existing, high visibility.
21, 22	III	Plain, steep slope, forest land, low heritage density, high visibility.
24, 31	II	Hill, medium slope, forest land, non-existing, low visibility.

. Four types were classified as follows: I included hill, medium slope, waters and water conservancy facilities land, non-existing, low visibility; II included low mountain and hill, steep slope, forest land, non-existing, low visibility; III included hill, gentle slope, forest land, high heritage density, low visibility; IV included plain, gentle slope, residential land, farmland, high heritage density, high visibility.

Figure 6A shows the initial identification of landscape character subareas using eCognition software, showcasing a detailed partitioning into numerous smaller subareas based on automated analysis with refined delineation following manual adjustment. Figure 6B shows that the subareas have been consolidated into clearly defined units, each numbered to facilitate further analysis and discussion. This figure highlights the transition from complex, automated segmentation to a more streamlined and practical subdivision suitable for management and planning purposes.

A holistic analysis reveals that forest land constitutes the majority of the area within these character types. However, when describing landscape characteristics, it is essential to focus on key features rather than just the proportion of the areas covered. For example, in landscape character area 4, the presence of the Yingong Reservoir, despite the predominance of forest land, is deemed more significant, and is thus described primarily as an area dedicated to water resources and aquatic facilities. This approach highlights the importance of selecting distinctive or defining features that contribute to the identity of the area. Visibility also emerges as a crucial indicator for delineating mountainous landscapes, with the extent of visual observation points indicating the primary activity zones of local residents. From a management perspective, remote hilly or low mountain regions, characterized by steep slopes and minimal human activity, are identified for specialized management to preserve their ecological and aesthetic values. In regions where plains are scarce, many villages adapt by situating themselves on gentler slopes and sloping areas within the hilly regions. These locales often serve as transitional zones between urban centers and the natural mountainous terrain, scattered with numerous historical sites and relics. Notably, landscape character areas 7, 8, and 18 include significant cultural heritage sites, such as the nationally protected Tianmu Kiln Site and provincial-level heritage sites like the former site of the Ethnic Daily Newspaper, highlighting their cultural and historical importance.

3.3. Management Zoning Strategies for Small Towns Based on LCA

In Yuqian Town, areas exhibiting relatively uniform landscape characteristics offer the potential to be developed into distinctive landscape character areas by leveraging their inherent attributes. By contrast, regions marked by a complex and varied landscape require a focused approach that integrates these diverse landscape characteristics, developing them into zones with diverse and distinctive features while concurrently prioritizing their preservation. Adhering to this planning principle, regions sharing similar landscape characteristics are integrated and assigned to administrative villages in close geographical proximity, forming four categories of landscape areas (Figure 7A). A subsequent overlay analysis results in the creation of the landscape character planning and management area map for Yuqian Town, as shown in Figure 7B. This map outlines several strategic areas, including the urban comprehensive service areas, agricultural tourism development areas, industrial development growth areas, and mountain forest ecological conservation areas. Each area is tailored to leverage its unique landscape features to fulfil specific developmental and conservation goals, thereby enhancing the town’s overall spatial planning and sustainability.

Four types of landscape areas, each categorized by distinct landscape character areas, are numbered and shaded in various colors, indicating their unique environmental and cultural characteristics: a, b, c, and d represent types I, II, III, and IV landscape character types, respectively (Figure 7A). For landscape character planning and management zoning of Yuqian Town, different zones are color-coded as follows: green represents the mountain forest ecological conservation zones; yellow highlights the agricultural and cultural tourism development zones; red designates the urban comprehensive service zones; brown dentifies the industrial development growth zones (Figure 7B).

The urban comprehensive service area in Yuqian Town spans approximately 10.368 square km, covering 4.1% of the town’s total area. It incorporates four administrative villages and is primarily situated in the regions defined by landscape character type IV, the central urban built-up land. This area also integrates select cultural and historical sites from landscape character type III, rendering it possible to create a dense hub of commercial, social, cultural, and entertainment facilities and services. Plain and gentle slope terrain (low altitude, gentle slope) shapes the foundation of urban construction; high visibility facilitates the use of facilities and cultural displays; heritage resources are integrated into urban services; and through the layout of residential, commercial, and service land, a convenient and culturally vibrant comprehensive service area is constructed. The focus is on enhancing the quality of life for urban residents and fostering vibrant social and cultural exchanges within a concentrated urban environment.

Meanwhile, the agricultural and cultural tourism development area includes eleven administrative villages, encompassing about 98.878 square km or 38.1% of the town’s total area. This area merges elements from landscape character types I, III, and IV, orchestrating a blend of environments that enrich its cultural and agricultural appeal. The spatial heterogeneity of terrains generates diverse land-use patterns and visual landscapes, thereby enhancing agricultural productivity and livability. High-density heritage resources function as tangible carriers of traditional farming culture. Furthermore, the strategic integration of regions with distinct landscape characteristics establishes a rural development zone characterized by a synergistic coupling of ecology, culture, and livelihoods. It features cultural resources predominantly from type III across five administrative villages, and the natural and semi-urban landscapes of type I across six villages, with additional urban features from type IV. This strategic integration forms a cohesive development zone for agricultural and cultural tourism, centered on celebrating and preserving the local agricultural heritage. The area is designed to highlight the town’s history, traditional agricultural practices, handicrafts, and other cultural facets, thereby promoting economic diversification and sustainable development in rural areas.

The industrial development growth area in Yuqian Town spans approximately 60.108 square km, which makes up about 23.3% of the town’s total area. This region, which covers seven administrative villages, is primarily situated in areas classified under landscape character types IV and III. It features relatively flat terrain and predominantly hosts industrial, mining, and warehousing facilities. The flat terrain meets the needs of industrial construction; high visibility facilitates the diffusion of the industrial growth effects; low site density reduces conflicts between industrial development and historical preservation; and, through the coordination of industrial, mining, and residential land, an industry-led development zone spreading to rural areas is constructed. This area is designated as a crucial zone for the town’s industrial expansion and functional diversification. The strategic zoning of this area is conducive to the adjustment and upgrading of the town’s internal industrial structure, exerting a scattering effect on the surrounding villages and promoting their economic, social, and cultural development.

The mountain and forest ecological conservation area, encompassing nine administrative villages and about 90.501 square km, or 34.5% of the town’s total area, is characterized by its remoteness from main residential activities. The visibility here is generally low, and the area is mainly classified under landscape character type II, with certain higher-altitude regions falling under landscape character type III. High-altitude, steep slope terrain (S4) and low visibility (V2) jointly create a closed ecological environment with little disturbance; low site density and forest-dominated land use ensure ecosystem integrity, constructing a mountain forest ecological conservation area to maintain ecological value. The primary aim of zoning this area is to safeguard its critical ecological functions, including water conservation, soil retention, and biodiversity preservation. This dedicated focus ensures that the mountain and forest ecosystems within this region are effectively conserved and managed, supporting the overall environmental health and sustainability of Yuqian Town.

4. Discussion

4.1. Selection of Clustering Methods

Landscape classification is a prerequisite for conducting landscape character assessments, serving as a foundational step in understanding the spatial organization and qualitative features of a landscape [28]. Cluster analysis facilitates understanding the relationships between landscape characters, types, and regions [29]. In practical applications, landscape variables used in clustering can be either continuous or categorical, with common methods including k-means clustering, the k-modes algorithm, the k-prototypes clustering algorithm, and the affinity propagation (AP) algorithm. The k-means clustering and k-modes algorithms allow researchers flexibility in selecting the size of clustering grids and the number of clusters [30]. The k-modes algorithm offers advantages in reducing sensitivity to noise and cluster shapes compared to k-means, although it demands higher requirements for initializing cluster centroids [31]. However, these methods are only suitable for analyzing continuous variables [30]. To address this, some scholars have attempted to integrate a principal component analysis (PCA) with these algorithms to identify dominant character types [10,32]. Yet, this approach tends to diminish the influence of non-forest areas in clustering results, particularly in mountainous small towns. The k-prototypes algorithm integrates the features of k-means and k-modes, facilitating the mixed clustering of both continuous and categorical data [33]. Despite its robustness, its implementation in Python (2.7.10, Python Software Foundation, Wilmington, DE, USA) poses challenges due to inherent limitations in clustering ordering and the need for manual construction of a clustering evaluation metric, which increases the difficulty and complexity of the operation [34]. Meanwhile, the AP algorithm, which does not require predefined initial cluster centers, mitigates the necessity for prior knowledge, which enhances its applicability [35]. However, it still requires the manual tuning of key parameters that influence the number of clusters and algorithm convergence. Given these limitations, this study opts for a two-step clustering algorithm, which adeptly manages both continuous and categorical variables and efficiently computes the optimal number of clusters. The method not only tackles the challenges of quantifying diverse cultural landscape variables but amplifies the impact of categorical variables in clustering results. It thus holds significant promise for accurately delineating landscape characteristics in mountainous small towns, ensuring a comprehensive and nuanced classification.

The choice of grid cell size is a critical factor that significantly influences the outcomes of landscape clustering [36]. Utilization of the original resolution of 30 m for grid partitioning would result in a total of 285,329 grid cells, which poses practical challenges for processing with ArcGIS software due to the vast volume of data. Extensive testing revealed that a cell size of 70 m represents a crucial threshold, enabling ArcGIS 10.7 software to manage and perform statistical operations effectively. While this adjustment in cell size may slightly compromise the precision of certain results, it generally maintains the integrity of statistical outcomes, especially relevant in the context of mountainous small towns. Moreover, from a landscape management perspective, controlling highly granular, heterogeneous spaces is often impractical. Thus, the selection of a 70-m cell size is not only a pragmatic decision but aligns well with the operational capabilities of the software and the management needs of the landscape, ensuring an efficient balance between data management and analytical accuracy. The commonly used clustering methods and their application scenarios are presented in

Table 4. Advantages, Disadvantages, and Application Environments of Clustering Methods.

Clustering Method	Core Characteristics	Advantages	Limitations	Applicable Scenarios
K-means Clustering	Targets continuous variables, optimizes cluster centers through iteration, requires preset number of clusters.	Simple operation, high computational efficiency, flexible adjustment of grid size and cluster number.	Sensitive to noise and outliers; relies on initial cluster centers; only suitable for continuous variables.	Preliminary clustering of large-scale continuous data.
K-modes Algorithm	Focuses on categorical variables, uses mode instead of mean to calculate cluster centers, reduces sensitivity to noise.	Better performance for categorical data than K-means; less dependent on cluster shapes.	High requirements for initializing cluster centers; still limited to categorical variables.	Clustering of pure categorical landscape features.
K-prototypes Algorithm	Integrates K-means and K-modes, supports mixed clustering of continuous and categorical data.	Handles multi-type variables; strong robustness.	Complex implementation in Python; requires manual construction of evaluation metrics; inherent limitations in clustering ordering.	Comprehensive landscape classification with mixed variables.
Affinity Propagation (AP) Algorithm	Eliminates the need for predefined initial cluster centers, determines cluster centers via message passing.	Reduces dependency on prior knowledge; wide applicability.	Requires manual tuning of key parameters (e.g., damping factor), affecting cluster number and convergence.	Exploratory clustering without prior information.
Two-step Clustering Algorithm (This Study)	First performs K-means clustering on continuous variables, then hierarchical clustering on categorical variables, automatically determines optimal cluster number.	Simultaneously processes continuous and categorical variables; quantifies cultural landscape variables; enhances weight of categorical variables in results; no manual parameter tuning needed.	Slight loss of precision for extremely fine-grained data (balanced by 70 m grid in this study).	Accurate delineation of multi-dimensional landscape characteristics in mountainous small towns

.

4.2. Classification, Zoning, and Landscape Planning Strategies for Mountain Towns

The traditional planning and organization of individual villages are proving insufficient to address the evolving development needs of mountainous small towns [34]. This shortfall emphasizes the necessity not only to maximize the limited available land for development but to consider the strategic integration and development of bordering areas that share similar landscape resource characteristics. This approach aims to harmonize development while leveraging the unique landscape attributes of these regions.

To address these challenges, this study employs the LCA method to systematically classify the research area across five dimensions: elevation, slope gradient, heritage density, visual accessibility, and land use patterns. Based on these criteria, mountainous urban development was categorized into four types: comprehensive urban service zones, agricultural tourism zones, industrial growth zones, and forest ecological conservation zones. The findings reveal that natural factors like terrain play a dominant role in spatial functional differentiation and pattern evolution within mountainous small towns. Relatively gentle slopes and flat terrains facilitate urban function concentration and development, while steep or high-altitude areas primarily serve as ecological barriers. Furthermore, visual accessibility, land use configurations, and cultural heritage distribution collectively create distinctive landscape identities across different regions.

Notably, compared with existing studies focusing on mountainous rural areas, the previous research predominantly emphasizes endogenous ecosystem indicators (such as plant diversity) [10,30,37,38,39,40], highlighting the ecological value of forest resources. By contrast, this study highlights the intrinsic needs and multidimensional characteristics of urban “core” development in mountainous regions. Particularly, some existing classification frameworks directly apply urban life cycle assessment theories, narrowly positioning mountainous areas as ecological buffer zones for cities, failing to fully reflect regional diversity and functional differences. This approach has somewhat limited the relevance and practicality of planning measures. Therefore, the empirical findings presented in this paper based on a novel zoning system can provide reference and valuable insights for specialized planning, management, and conservation efforts in related regions across China.

Specifically, this study further recommends formulating differentiated management strategies for four core sectors: urban integrated service zones, agricultural tourism development zones, industrial growth zones, and mountain-forest ecological conservation zones. During urban expansion, priority should be given to preserving urban–rural fringe areas and key mountain landscapes, avoiding unnecessary interventions on complex scenery. By integrating cultural heritage resources within these zones, distinctive cultural landmarks can be shaped. Agricultural tourism development zones should emphasize topographic variations at landscape unit boundaries and agricultural landscape protection, promoting deep integration of multidimensional landscapes with tourism to achieve coordinated growth in local agritourism economies. Industrial growth zones require the enhanced comprehensive planning of composite spaces and industrial clustering, standardized land use practices, and risk prevention measures for ecological environments. For mountain-forest conservation zones, monitoring and management of ecologically fragile and high-risk areas remain crucial for ensuring regional sustainable development.

However, this study also has certain limitations. First, in the quantitative analysis framework, it primarily focuses on natural physical elements (such as topography and land use) while inadequately incorporating “soft” factors, like historical development trajectories [41], architectural settlement characteristics [42], and socio-cultural elements [43]. These elements play a crucial role in shaping local landscape identity and enhancing spatial perception experiences. Future research should broaden its analytical dimensions to explore the interactive mechanisms between natural and socio-cultural elements, thereby achieving both scientific rigor and human-centered approaches in landscape planning for mountainous small towns.

5. Conclusions

This study utilized ArcGIS, eCognition software, and a two-step clustering algorithm to identify landscape character types and areas within Yuqian Town, Zhejiang Province. The landscape character assessment (LCA) method screened five critical indicators—altitude, slope, heritage density, visual visibility, and land use—to classify the mountainous town development into four distinct types: urban comprehensive service areas, agricultural and cultural tourism development areas, industrial development growth areas, and mountain-forest ecological conservation areas. By integrating landscape resource characteristics with administrative boundaries, the research effectively addressed the challenges of landscape character zoning and planning management in mountainous small towns.

This study reveals that natural elements, particularly elevation and slope gradients, play a decisive role in shaping landscape zoning patterns. Flat areas naturally develop into urban functional clusters, while steep slopes serve as ecological barriers. The interplay of visual corridors, land use patterns, and cultural heritage creates differentiated landscape expressions, with scientifically planned visual corridors significantly enhancing the landscape perception value of cultural tourism integration zones.

Methodologically, the utilization of the two-step cluster analysis was particularly beneficial for handling categorical variables, helping to reduce the disproportionate influence of forested areas on the clustering results, which is a common issue in mountainous regions. The combination of semi-automatic segmentation and manual delineation techniques enabled precise identification of key characteristics, fostering a systematic integration of Yuqian Town in spatial and administrative dimensions. This approach not only supports the effective zoning of landscape characters but establishes a model for coordinated development and conservation efforts in similar environments.

Nonetheless, there are areas for further refinement. This study utilized a select number of indicators, which, given the dynamic nature of landscapes, suggests that classification results might benefit from a broader indicator set. Additionally, while forming landscape character management units, factors like public opinion and environmental carrying capacity assessments were not included, which could enrich the management zoning process. Future research could enhance the depth and applicability of landscape classification by expanding the range of landscape indicators to incorporate both quantitative measures of built environments and qualitative assessments of cultural landscapes. Incorporating stakeholder engagement, such as community consultations or participatory mapping, will also help ensure that management zoning aligns more effectively with community needs and ecological sustainability, thereby achieving a more balanced approach that considers both natural and cultural landscape dimensions, leading to a more holistic approach to landscape management.