Research Methodological Framework

In this study, we present a novel approach to establishing an indicator system for TV tourism from the perspective of new developmental concepts. Our empirical study employs the entropy-weighted TOPSIS method and nearest neighbor index analysis. We conduct research on the spatiotemporal distribution of TVTR in the YRDR, using relevant data from the National Bureau of Statistics of the People’s Republic of China. Data preprocessing is conducted, incorporating various dimensions of TVTR evaluation into the indicator system. We establish a framework comprising four dimensions, which include village resource endowment, ecological environment, socio-economics, and tourism vitality. The EM is subsequently utilized to ascertain the individual importance of each indicator. Thereafter, the TOPSIS is employed to evaluate the alignment between the TVTR assessment and the most favorable resolution. Employing comprehensive evaluation scores and analyzing various dimensions of TVTA development, we conduct an assessment and analysis of TVTA development in the YRDR. Furthermore, we explore the spatiotemporal evolution trends of six batches of TVs. Finally, the Geographic Detector method is employed to discern the driving factors among TVs. The procedural diagram of the mapping approach is depicted in Fig. 2.

Evaluation Methodology

The entropy method (EM) is employed to ensure the weights of indicators, and TOPSIS is utilized to compute the TVTR in the YRDR. ArcGIS 10.8 is employed to visually express the spatiotemporal differentiation of TVTR across various scales. Additionally, the Geographic Detector approach is employed to probe the determinants affecting the spatial variances of TR among TVs within the drainage basin [38].

Indicator Standardization Processing

Various indicators possess distinct units and dimensions, rendering direct comparison and calculation unfeasible. Therefore, before calculating the weights of each indicator, we tend to standardize the indicators.

For positive-oriented indicators, the formula for standardization is:

Fig. 1.

Study area introduction.

Fig. 2.

Methodology flow chart.

For negative-oriented indicators, the formula for standardization is:

The first step involves standardization, and the formula for standardization is:

yij represents the standardized indicator value after nondimensionalization for the jth indicator of the ith unit; xij stands for the original value of a specific sub-evaluation indicator; i denotes the ith (i=1,2,…,n) assessment region; j signifies the jth (j=1,2,… …,m) sub-evaluation indicator. Among the 14 indicators, all except the third indicator are positive-oriented.

Entropy Method for Weight Determination

The entropy weight method (EWM) is an objective assignment technique relying on the inherent discreteness of the data. The weight assigned to each evaluation indicator increases as the degree of discreteness of the indicator itself increases [39]. The core concept of this method is to standardize all indicators and sort them in order of their superiority or inferiority using the TOPSIS method. In this study, data processing is conducted using SPSSAU software, employing the EWM to determine the weights of sustainability evaluation indicators for water resources in China's western region.

The first step involves calculating the information entropy value for the jth indicator using the following formula:

The second step entails calculating the information utility value using the following formula:

The third step entails deriving the weight of the jth indicator using the formula:

After computing the entropy values, the weights for the ith evaluation indicator are then determined according to the formula (Table 1).

In this context: ej denotes the entropy value for the jth metric; dj symbolizes the superfluity of the entropy information for the same metric, commonly known as the coefficient of variation, where a larger figure suggests heightened significance; Wj signifies the weighting assigned to the jth metric, and wj represents the assigned weight to the ith metric.

Calculating Responsiveness Using the TOPSIS Method

The Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is a multi-criteria decision analysis method that ranks alternatives based on their closeness to an ideal solution. It is widely applied in systems engineering to evaluate and prioritize options in the presence of multiple, often conflicting, objectives [40]. It is an approximate ideal value ranking method suitable for assessing superiority and inferiority based on multiple indicators and various alternatives. The method utilizes the Euclidean distance between the alternatives and the positive and negative ideal resolutions to quantify the advantages and disadvantages of each alternative.

The TOPSIS method is applied to comprehensively evaluate the overall situation of the 41 cities (prefectures) within the YRDR urban agglomeration. Firstly, the positive and negative ideal solutions are computed [41]. Secondly, the distances between the indicator data for each city and the optimal and worst scenarios are calculated. Lastly, the proximity degree 'fi' is determined based on the computed ideal optimal and worst-case solutions. The ‘fi’ ranges between 0 and 1, with larger values indicating higher levels of TVTR in the respective region.

Firstly, the original matrix is constructed and normalized to obtain the canonical decision matrix:

Secondly, forming the weighted normalized matrix, where the weight vector is denoted as W=[W_1,W_2,⋯,W_n ]^T, as computed using the EWM:

Where cij represents the weighted normalized matrix, wj signifies the weight, with i representing the sample value and j representing the indicator value.

Third, the process involves identifying the positive ideal outcome C* and the negative ideal outcome C0. Presuming the jth attribute value for the positive ideal outcome C* is Cij, and for the negative ideal outcome C0 it is c0j, the equation used to calculate the positive ideal outcome is:

The formula for computing the negative ideal solution is:

Here, Y+ signifies the positive ideal outcome and Y- denotes the negative ideal outcome; i and j correspond to the sample and indicator values, respectively.

Fourth, the step entails computing the distances that separate each alternative from both the positive and negative ideal outcomes.

In this context, Cij signifies the weighted and normalized figure for the jth metric concerning the ith option; s+i and s-i stand for the positive and negative ideal values, respectively, for the jth metric across all available alternatives.

Fifthly, calculate the proximity degree for each alternative, i.e., the TR score for each city within the YRDR. The formula is:

Where a larger fi indicates that the level of the ith evaluation alternative is closer to the optimal level.

Nearest Neighbor Index

The Nearest Neighbor Index method, also known as Nearest Neighbor Distance Analysis, is a method used to assess the distribution pattern of point geographical features by comparing the value of the average nearest neighbor distance with the theoretical nearest neighbor distance [42-43]. It employs the Completely Spatially Random Pattern (CSR) as a reference for comparison. When the observed pattern’s nearest neighbor distance is greater than that of a random distribution, the pattern is dispersed; when it is less, the pattern is clustered [36]. The formula to calculate this is as follows:

Here, R designates the Nearest Neighbor Index value, rF symbolizes the empirically observed nearest neighbor distance, and rE represents the theoretical nearest neighbor distance. The variable n counts the number of villages, dmin indicates the minimum observed distance between two TVs, and A refers to the area of the region in question. An R value greater than 1 signals a random dispersion; less than 1 implies clustering and an R value of 1 indicates uniform distribution.

Kernel Density Analysis

Kernel Density Analysis computes the point feature density across each resulting raster, offering a visual depiction of TV's clustering and dispersal [44]. The corresponding equation is:

In this equation, k (x−xih) serves as the kernel density function, h signifies the bandwidth, n counts the points within the threshold radius, and (x-xi) indicates the distance separating the estimation point x from the occurrence xi.

Geographical Detector

The Geographical Detector, a statistical approach, is employed to investigate spatial discrepancies and the underlying causes [45-46]. Currently, it is mainly applied in research fields such as land use, public health, regional economics, regional planning, tourism, archaeology, etc. The central premise determines that if an explanatory variable substantially affects the spatial fluctuation of a dependent variable, then a notable resemblance should exist between the spatial distributions of both variables. In this study, this model is employed to detect the influencing factors of spatial variation in TVTR across 35 cities in the YRDR. By discerning the varying influences of these elements, the goal is to offer constructive guidelines for the judicious exploitation and preservation of TV heritage assets, as well as for the growth of tourism in TV regions across the YRDR.

Utilizing the Factor Detector, the 14 influencing factors from the indicator system are introduced into the geographical detector model [47]. This yields the influence value (q) and explanatory power value (p) of each influencing factor on TVTR. The q value ranges from 0 to 1. And if the value is closer to 1, it indicates greater influence [24]. A value closer to 0 implies a lesser influence of the factor on TVTR. The numerical expression formula is as follows:

Where q represents the magnitude of the detecting factor’s influence, with q ranging from 0 to 1. A larger q value indicates a higher degree of impact of the driving factor on TV tourism. K denotes the number of secondary samples. Ni and n stand for the number of regions in each sub-region and the research area, respectively. σ21 and σ2 refer to the variance of indicators for region i and the research area.

Overall Scores Differences on Traditional Village Tourism Responsiveness

In this study, the TOPSIS method is employed to quantitatively assess the tourism responsiveness of 41 cities in the YRDR. The specific overall scores (fi) can be obtained in Table 2. From the perspective of provinces, Jiangsu Province ranks first in terms of tourism responsiveness among all surveyed regions, while Anhui Province lags behind, occupying the last position. Delving into the city level, we observe that 15 cities have scores exceeding 0.4, including but not limited to Hangzhou, Shanghai, and Ningbo. These cities collectively account for 37% of the total of 41 cities in the YRDR. They demonstrate outstanding performance in various domains such as village endowment, socioeconomic factors, ecological environment, and tourism activity. In contrast, 10 cities have scores below 0.2, such as Huai’an, Suqian, and Lianyungang, accounting for 24% of the total. This suggests that these cities need to enhance their performance in the selected evaluation indicators, and the future development and planning of their tourism industry may require corresponding optimization and adjustments.

Spatial Distribution of TV Tourism Responsiveness in Different Cities Within the YRDR Metropolitan Region

In this study, the natural breakpoint method is employed to categorize the tourism responsiveness of each city (prefecture) into five levels, ensuring maximum similarity within each group and maximum differences between groups. The specific classification is as follows: high responsiveness area, moderately high responsiveness area, moderate responsiveness area, moderately low responsiveness area, and low responsiveness area. The visualization of the results is presented in Fig. 3. In the YRDR, there are 15 cities (prefectures) at Level 2 or above, accounting for 37% of the total, primarily distributed in Zhejiang Province and Shanghai. Moreover, the tourism responsiveness of two-thirds of the provinces and cities in the YRDR falls below Level 3, mainly concentrated in Jiangsu Province and Anhui Province. Considering the difference in the number of cities within each province, we calculated the average score to determine the TVTR of each province (region). The results demonstrate that Shanghai ranks first with a score of 0.613, followed by Zhejiang with a score of 0.5499. Jiangsu scores 0.2814, while Anhui has the lowest score of 0.2188. This is positively correlated with the number of TVs in the provinces (regions), where Shanghai has the highest number of villages while Anhui has the least.

In general, TVTR in Shanghai and Zhejiang ranks at the forefront, followed by Jiangsu. The responsiveness of Anhui, which is located inland, is generally lower, highlighting the spatial imbalance across different areas in the Yangtze River Delta. The TVTR in the YRDR exhibits a notable spatial imbalance, with coastal areas significantly outperforming inland regions. This disparity may be attributed to geographical location, transportation convenience, and other regional factors.

Spatiotemporal Evolution of Traditional Village Tourism Responsiveness in the YRD Urban Agglomeration

In this study, the TOPSIS method is employed to quantitatively assess the tourism responsiveness of 41 cities in the YRDR. The specific overall scores (fi) can be obtained in Table 2. From the perspective of provinces, Jiangsu Province ranks first in terms of tourism responsiveness among all surveyed regions, while Anhui Province lags behind, occupying the last position. Delving into the city level, we observe that 15 cities have scores exceeding 0.4, including but not limited to Hangzhou, Shanghai, and Ningbo. These cities collectively account for 37% of the total of 41 cities in the YRDR. They demonstrate outstanding performance in various domains such as village endowment, socioeconomic factors, ecological environment, and tourism activity. In contrast, 10 cities have scores below 0.2, such as Huai’an, Suqian, and Lianyungang, accounting for 24% of the total. This suggests that these cities need to enhance their performance in the selected evaluation indicators, and the future development and planning of their tourism industry may require corresponding optimization and adjustments.

Fig. 3.

Methodology flow chart

Spatial Distribution of TV Tourism Responsiveness in Different Cities Within the YRDR Metropolitan Region

In this study, the natural breakpoint method is employed to categorize the tourism responsiveness of each city (prefecture) into five levels, ensuring maximum similarity within each group and maximum differences between groups. The specific classification is as follows: high responsiveness area, moderately high responsiveness area, moderate responsiveness area, moderately low responsiveness area, and low responsiveness area. The visualization of the results is presented in Fig. 3. In the YRDR, there are 15 cities (prefectures) at Level 2 or above, accounting for 37% of the total, primarily distributed in Zhejiang Province and Shanghai. Moreover, the tourism responsiveness of two-thirds of the provinces and cities in the YRDR falls below Level 3, mainly concentrated in Jiangsu Province and Anhui Province. Considering the difference in the number of cities within each province, we calculated the average score to determine the TVTR of each province (region). The results demonstrate that Shanghai ranks first with a score of 0.613, followed by Zhejiang with a score of 0.5499. Jiangsu scores 0.2814, while Anhui has the lowest score of 0.2188. This is positively correlated with the number of TVs in the provinces (regions), where Shanghai has the highest number of villages while Anhui has the least.

In general, TVTR in Shanghai and Zhejiang ranks at the forefront, followed by Jiangsu. The responsiveness of Anhui, which is located inland, is generally lower, highlighting the spatial imbalance across different areas in the Yangtze River Delta. The TVTR in the YRDR exhibits a notable spatial imbalance, with coastal areas significantly outperforming inland regions. This disparity may be attributed to geographical location, transportation convenience, and other regional factors.

Spatiotemporal Evolution of Traditional Village Tourism Responsiveness in the YRD Urban Agglomeration

Based on the spatial positioning of TVTA in the YRDR, the geographical coordinate distribution areas of TVs in the region for the years 2012, 2013, 2014, 2016, 2019, and 2022 are selected, as shown in Fig. 4. At the provincial level, TVs can be considered a point feature, with their geographic locations represented by coordinate points. Geographical coordinates for Traditional Villages (TVs) within the YRDR are incorporated into the ArcGIS 10.8 platform. Utilizing the Average Nearest Neighbor instrument from the Spatial Statistics suite, the nearest neighbor indices across six distinct time frames in the YRDR are computed. The findings are presented in Table 3. The Nearest Neighbor Index (R) for the TVTA in the YRDR for the six time periods is all less than 1, indicating that the TVTA is in an aggregated state during these periods. From the temporal changes in R, the value of R for 2022 is relatively high at 0.8967, approaching 1, with a confidence level (P) of 0.0079. The value of R for 2013 and 2016 decreases, and both of these periods have a P-value close to zero, indicating a confidence level of over 99%. This suggests that the spatial clustering trend of TVTA in the YRDR has been gradually strengthening during these periods. Furthermore, the values of R are between 0.6 and 0.9, all greater than 0.5, indicating that there is still some room for development in the maturity of TVTA in the YRDR.

To further analyze the changing characteristics of spatial clustering of TVTA in the YRDR, the Density Analysis tool in ArcGIS 10.8 is employed to analyze the spatial distribution density of TVTA in the six time periods. The findings are depicted in Fig. 5. The overall evolution trend of the kernel density of TVTA in the YRDR over the six time periods shifts from a dispersed structure to a ‘clustered structure and then to a combination of clustered and ribbon-like structures’, generally forming a C-shaped high-density distribution area. In 2012, a C-shaped kernel density structure emerged with the Huangshan-Anqing high-density belt as the vertex, the Jinhua-Lishui-Quzhou high-density belt, and the Huzhou-Shanghai high-density belt as the wings. In 2013, the C-shaped kernel density structure became clearer with the addition of Zhenjiang and Taizhou to the existing high-density belt. In 2014, the high-density areas decreased significantly, leaving only the Huangshan-Lishui high-density belt. By 2016, the high-density distribution areas of TVTA in the YRDR experienced rapid growth, with the Lishui-Taizhou high-density belt as the vertex and areas including Huangshan, Xuancheng, Anqing, Lishui-Jinhua, Hangzhou, and Wenzhou as the high-density belts. In 2019, the high-density distribution areas of TVTA in the YRDR were relatively sparse. Only sporadic high-density distribution points were observed, mainly in Huangshan, Quzhou, Jinhua, and Lishui, primarily within Anhui Province. By the year 2022, with the number of TVTAs increasing in the YRDR, the overall kernel density intensified. In addition to the stable growth of kernel density in Huangshan, Quzhou, Jinhua, and Lishui, new high-density distribution points emerged, primarily within Jiangsu, including Suzhou, Huzhou, Yixing, Changzhou, Zhenjiang, and Yangzhou. This gradual emergence contributed to the formation of a high-density distribution belt, indicating a strengthening of kernel density due to the increased number of TVTAs in the region.

Fig 4.

Schematic Diagram of Spatial Distribution of TV in Different Time Periods in the YRDR

Fig 5.

Analysis Diagram of TV Kernel Density in different period of the YRDR.

Detection Result of Drivers

Within the 41 cities in the YRDR, we conducted a deep investigation of the influential factors on TVTR through geographical detector analysis. The results indicate that the P-values of 9 factors are less than 0.1, as depicted in Table 4. This confirms the diversified nature of factors affecting TVTR in the YRDR area. From the classification of each influential factor, their impacts on the spatial distribution of TV tourism vary, with the specific ranking as follows: X6 > X2 > X5 > X1 > X3 > X4 > X8 > X7 > X9. Notably, factors with q-values exceeding 0.3 are primarily associated with ecological and village endowment, involving X6, X2, and X5. These three factors exhibit significant explanatory power and influence over the spatial distribution of TV tourism. Conversely, the explanatory power and influence of X9 are comparatively limited, revealing a lower sensitivity of TV tourism responses in the YRDR and a stronger dependence on regional tourism development and the economic environment.

Initially, ecological elements had the most substantial impact on the spatial dispersion of TVTA. X5 and X6 emerge as key determinants in the layout of TVTR. This could be attributed to the fact that a healthy and vibrant ecological environment can provide a more appealing atmosphere, attracting tourists to experience a harmonious rural life coexisting with nature. In particular, forest coverage can impact climate regulation, biodiversity, and landscape aesthetics, while total water resources are directly linked to regional ecological health and sustainability. These ecological factors also align with the principles of sustainable development and ecotourism for TVs. From the data of the YRDR, these ecological indicators exhibit q-values greater than 0.3, indicating a strong correlation with TVTR. For decision-makers and planners, placing emphasis on the preservation of these critical ecological resources can not only enhance the region’s tourism appeal but also play a vital role in ensuring long-term ecological health and the sustainable development of the tourism industry in the area.

Second, natural variables occupy the secondary rank concerning their sway on the spatial allocation of TVTA. Elements X1 and X2, primarily natural in essence, play a supportive but crucial role in the geographical patterning of TV tourism. They form the foundation of attraction for tourism destinations, influencing the seasonality of tourism, feasibility, and special activities of the areas. In the YRDR, these natural factors, along with ecological indicators X5 and X6, collectively contribute to providing TVs with a unique tourism backdrop and allure. While ecological factors play a significant role in tourism responsiveness, the contributions of the natural factors represented by X1 and X2 are not to be underestimated. Therefore, decision-makers must pay full attention to them when formulating tourism strategies.

Thirdly, socio-economic factors exhibit an influence on the spatial distribution of TV tourism destinations. Variables X3 and X4 serve as indicators of an area’s economic health and residents’ consumption capacity. The economic prosperity of a region is closely linked to the maturity of its tourism industry, the level of tourism facilities, and the effectiveness of external publicity. Particularly within the context of rural tourism, variable X4 might be associated with the willingness and capacity of local residents to engage in tourism-related activities such as providing accommodations, agritainment, and other services. From the data in the YRDR, although the influence of socio-economic factors in determining the responsiveness of TV tourism is not as prominent as ecological and natural factors, they still hold significant importance in comprehending the economic disparities and strengths within the region. Therefore, further analysis and research on these indicators will contribute to providing policymakers with a more comprehensive basis for decision-making.

Fourth, the factor of tourism vitality exhibits a relatively minor impact on the spatial distribution of TVTA. Both X8 and X9 demonstrate limited influence on spatial distribution, particularly X9, which shows a relatively modest explanatory power. Possible explanations for this observation include regional characteristics specific to the YRDR, which might possess distinct geographical and socio-economic attributes that result in ecological and natural factors as primary driving forces. In addition, the type of tourism prevalent in this area, primarily centered around ecological and natural attractions, may render factors related to tourism vitality less pivotal in consideration.

Detection Results of Temporal-Spatial Evolution Driving Factors

Table 5 comprehensively demonstrates the changing influence of core factors on TVTR in the YRD during six different time periods: 2012, 2013, 2014, 2016, 2019, and 2022. Among the nine factors, the explanatory power of X3 and X7 has gradually increased since 2012, while the impact of other factors has diminished year by year. Notably, the influence of X2 has decreased from 0.484 in 2012 to 0.269 in 2022, which might reflect the gradual expansion of development from core cities to all cities in the YRDR. Coinciding with this trend, the influence of X1 has noticeably decreased since 2016, validating the spatial-temporal evolution from the core cities toward all cities within the region, as highlighted in this study. Furthermore, the influence of X4, X5, X6, X8, and X9 has slightly decreased since 2012, though the overall changes are relatively modest. This may suggest that these factors have exhibited a relatively stable role in the spatial-temporal evolution of TVs in the YRDR. By analyzing the changes in the above factors across six different time points in the region, this study has unveiled the primary driving factors behind the spatial-temporal evolution of TVs. The variation of these factors reflects the interplay among cities in the region and the overall development trends. Future research could delve into the specific mechanisms underlying these factors and provide more targeted recommendations for regional tourism development policies.

Results of Interaction Exploration

Interaction Exploration is an analytical approach used to uncover the combined impact of two or more influencing factors on changes in a target variable. This section focuses on analyzing whether these interactions enhance or diminish the explanatory power of TVTA spatial distribution in the YRDR. As presented in Table 6, all pairwise combinations of influencing factors exhibit nonlinearity and augmentation effects. This suggests that the spatial distribution of TVTA is not solely governed by a single factor but by multiple factors. This finding is consistent with the previous analysis and emphasizes the importance of understanding the interactive mechanisms behind the complex geographic and spatial distribution. The strongest interaction combinations are X2 and X7, X6 and X7, and X7 and X9, all of which exhibit explanatory power exceeding 0.9. These strong interaction effects further validate the necessity of considering the interactions among multiple influencing factors when evaluating the spatial distribution of TVTA in the YRDR. The results of interaction detection underscore the complexity of spatial distribution in TVTA, and the analysis of multifactor interactions provides profound insights into understanding the spatiotemporal variations in tourism responsiveness in the region. By identifying and analyzing critical interacting factors, targeted guidance can be provided for the formulation of tourism development strategies in the YRDR. This analytical approach may also offer valuable references for studies in other regions or fields.