Exploring the Spatial Relationship between Nighttime Light and Tourism Economy: Evidence from 31 Provinces in China

Abstract

Nighttime light (NTL) data have become increasingly practical and are now widely used in studies on urbanization, energy consumption, population estimation, socio-economic evaluation, etc. Based on NTL data and the basic tourism economy (TE) data from 31 provinces of China in 2019, this paper adopted a geographic concentration index, inconsistency index, spatial agglomeration coupling index, global and Local Moran’s index and geographical detector to explore the spatial relationship between NTL and TE. The results of the study were as follows. Firstly, there is a high spatial correlation between NTL and TE. Secondly, the concentration degree, as well as the concentrated distribution area of NTL and TE, are very similar, roughly showing a higher concentration in East and South-Central China. Thirdly, NTL and TE show a type of coordinated development in East and North China, and a TE surpassing NTL in Southwest and South-Central China. The spatial agglomeration coupling index is higher in North China, South-Central China and the coastal regions of East China, and relatively lower in Southwest and Northwest China. Furthermore, in the spatial agglomeration distribution of NTL and TE, there is an obvious high–high and low–low agglomeration. Finally, the geographical detector analysis showed that the driving factor of tourism economy level (TEL) also has a great influence on NTL. The spatial distribution of NTL and TE is integrated to reasonably allocate tourism resources for different areas and promote the sustainable development of NTL and TE among regions.

1. Introduction

In 2019, the global tourism industry created a total of 9.2 trillion in GDP, contributing 10.4% to the global economy. The value added to China’s tourism and related industries was CNY 449.89 billion, accounting for 4.56% of domestic GDP. Therefore, tourism plays an important role. With gradual improvements in science and technology and the broadening of tourism research horizons, scholars are no longer limited to the use of official statistical yearbook data when selecting data for research, but strive to conduct their research through a more objective perspective. Hence, NTL data, as more objective and realistic data, were gradually introduced into academic research. Internet Plus [1], remote sensing, geographic information systems [2] and global positioning [3] have been applied to tourism research. With the increasingly abundant remote sensing image data, coupled with the National Oceanic and Atmospheric Administration of the United States providing free and open global NTL data, the data acquisition process is more realistic and objective at present compared with artificial statistics; therefore, NTL data are widely used in various types of research. NTL data are mostly used for population, urbanization, economy, transportation, estimation of electric power consumption and CO₂ emission measurement. China is the world’s second most populated country, and its population is one of the key problematic factors that are restricting China’s economic and social development. Scholars have conducted a wealth of research on population issues, with studies mostly based on linear regression and geographically weighted regression analyses about population estimation [4] and population spatial pattern changes [5]. Furthermore, studies conclude that VIIRS NTL hold great potential and more widespread application of remote sensing information in studying dynamic demographic processes [6]. Urbanization plays a substantial role in anthropogenic environment changes, and the urbanization rate has increased steadily in China over the last decades, although the urbanization rate varies greatly in different parts of China. NTL offer a unique opportunity to understand the urbanization process through research into topics including urbanization [7,8] and urban built-up area extraction [9]. China being the second largest economy in the world, its economic development is highly regarded, with scholars striving to show the level of Chinese economic development to the world through a more objective perspective. Therefore, NTL were gradually introduced into research into GDP estimation [10] and simulation [11]. The results show that the NPP/VIIRS NTL data are better than the DMSP/OLS data for GDP estimation [12]. Transportation is one of the core elements of tourism development; convenient transportation can attract more tourists to the destination, and therefore the transportation research has become a pressing topic. Scholars made NTL as a proxy for freight traffic to predict traffic and freight volume [13]. NTL also can be used in research into the evaluation of highway traffic conditions [14] and the spatial distribution pattern of highways [15]. Electric power services are fundamental to prosperity and economic development, and increasing research effort has been made to relate electric energy consumption to NTL. These research areas include the estimation of electric power consumption [16,17] and monitoring simulation [18]. NTL data are most appropriate for estimating electric power consumption in developing countries; when estimating the power consumption of developed countries, more latent factors should be added into the model [19]. As is well known, carbon dioxide (CO₂) emissions are closely related to the scope and intensity of human activities. With the increasing trend of global warming, the world faces tremendous pressure to reduce CO₂ emissions. Nowadays, both DMSP/OLS and NPP/VIIRS image data are widely used in spatiotemporal variations of CO₂ emissions [20] and CO₂ emission measurements [21]. Besides, the results of simulations at different scales show that DMSP/OLS is superior to NPP/VIIRS in modeling CO₂ emissions at all spatial scales, and the bigger the scale, the more evident the advantages of DMSP/OLS [22].

At present, there are plenty of studies on the tourism economy (TE), such as research on the relationship of tourism with transportation, urbanization and culture. Tourism and transportation generally have an inseparable association, and transportation is an important topic in tourism research. Compared with high-speed rail, air travel or traditional rail remains central to the tourism economy. Meanwhile, a high level of tourist arrivals or a high level of tourism revenue are multiple paths that lead to regional tourism economic development [23]. Urbanization is an important factor for tourism development and has different effects according to the level of a city’s tourism development [24]. Urbanization has brought opportunities to develop the eco-tourism economy and promote eco-tourism development [25]. Culture has always played a crucial role in human life, and cultural tourism is one of the best ways to understand the cultural interdependence of nations. Scholars have explored the impact of culture on tourism and found that there is a negative relationship between cultural distance and tourism demand; tourism demand is less sensitive to change in cultural distance [26]. However, few studies have integrated NTL data with TE. The main reason for this is that some scholars believe that most tourism activities occur during the day. Nighttime tourism has developed gradually but not enough to represent the TE, thus NTL might be slightly inappropriate in tourism research. The author does not agree with such views. NTL data can estimate GDP more accurately [11]. At present, the proportion of tourism in the national GDP is gradually increasing; tourism has become an important force in promoting the development of the national economy, and studies have found a close relationship between NTL and socio-economic activities through correlation analysis and geographically weighted regression, etc. While studying night tourism, scholars have found that NTL has a strong impact on TE. An appropriate extension of the opening hours of tourist attractions and optimized city lighting projects can not only increase tourism revenue but also improve the competitiveness of urban tourism [27]. The latest research shows that a decrease in the number of tourists visiting destinations impacts the closure of most tourism facilities and furthermore reduces employee numbers and the brightness level of the nighttime light. There is therefore a close relationship between NTL and tourism activities [28].

Many cities around the world have taken the nighttime economy as an important strategy for urban spatial revitalization and economic revitalization. Night tourism has become a hot topic, and the seasonality and regularity of the night tourism market is becoming increasingly obvious, with an important role to play in the promotion of the regional economy and enhancement of urban construction [29]. Tourists’ satisfaction is considered an important factor in the sustainable development of the tourism industry. Scholars have studied the influence of tourists’ consumption experience and tourism image on satisfaction and willingness to revisit night markets in Taiwan, concluding that the night market tourism image has a significant positive influence on tourist satisfaction [30]. In addition, the improvement in tourist satisfaction also has a positive effect on the night TE. As night tourism has attracted extensive attention, studies have confirmed that the development of night tourism will promote the night economy and even the city’s GDP [31].

In view of the sound development of night tourism, cultivating a new mode of nighttime cultural and touristic consumption has become a new trend in tourism development. In recent years, the number of night tourism participants has greatly increased in China, and night tourism has become an important way of meeting people’s desire for a better life and promote tourism consumption. At the same time, in response to the new trend of upgrading the quality and transformation of cultural and tourism consumption, the Chinese government is promoting the construction of a nighttime cultural and tourism agglomeration area by developing the nighttime culture and TE, improving cultural and tourism consumption facilities, optimizing the products and services of nighttime restaurants, shopping, performing arts and other venues, and continuously expanding the consumption scale of nighttime cultural and tourism. Therefore, there will be more tourism activities at night, and the impact of tourism activities on NTL will continue to increase. In tourism research, nighttime lighting has been used as an important evaluation indicator to reflect the intensity of human activities and has been used to evaluate the suitability of tourism destination development [32]. At the same time, it has been shown that tourism activities can promote local economic development and increase the number of tourists, which, to some extent, makes the management of protected areas more difficult and causes light pollution [33]. Taking astronomy tourism as an example, sustainable development strategies can effectively prevent light pollution [34]. Light pollution is generally produced by excessive outdoor lighting in cities, and since tourism activities can cause excessive lighting at night, the brightness of lights in different areas will be reduced when tourism activities are reduced [35].

Due to the increase in tourism activities, the brightness of lights at night has continued to improve. NTL has already been used to extract [36] and identify tourist attractions [37]. Ecological quality has been closely linked to tourism development; therefore, NTL data are used to study ecological changes [38]. As the accessibility of tourism POI data gradually improved, it was found that the POI of tourism factors had a significant influence on the difference in the spatial distribution of NTL [39]; the spatial coupling between the two is good. The area of light patches and light intensity became larger with the enhancement of various economic activities, while the density of urban leisure and tourism facilities was also gradually enlarged [40]. By studying the spatial distribution characteristics of NTL and tourism, POI can also provide useful suggestions for the spatial distribution characteristics of public tourism facilities and the balance of supply and demand [41]. Furthermore, the joint use of NTL and POI can identify urban nighttime leisure spaces more accurately and also help government departments better understand the local nightlife situation to rationally formulate planning and adjustment measures [42].

Eleni et al. studied the relationship between tourism activities and NTL in EU countries using remote-sensing image data from 2012 to 2013 together with remote-sensing techniques, GIS techniques, OLS and GWR, and finally concluded that NTL is highly correlated with tourism activities and the GWR model can better explain the relationship between NTL and tourism activities. In addition, Eleni et al. also pointed out that the accuracy of this model could be enhanced by adding other variables [43]. The development level of the TE directly reflects that of tourism. With the help of the spatial panel econometric model, scholars have confirmed that there is a high correlation between NTL and TE. Thus, NTL can be used as a proxy variable for the development of TE under certain conditions [44].

In summary, there is a close relationship between NTL and tourism activities, and these are influenced by each other. Studies on the application of NTL data are plenty, but there has been little research on its application in tourism. The existing research also has some limitations, as follows. First, the tourism data indicators selected in the exploration of the relationship between the two are limited and the analysis of TE activities is not comprehensive enough. Therefore, based on the existing research theory, this paper selects more varied TE indicators to analyze the spatial clustering relationship between different TE factors and NTL, and the influence intensity of TE factors on NTL. Second, in terms of methodology, most previous studies use linear regression and GWR regression to analyze the relationship between NTL and tourism. This paper uses the geographic concentration index, inconsistency index, spatial agglomeration coupling and Moran’s index to comprehensively analyze the spatial distribution pattern of NTL and TE, and then uses geographic detectors to further confirm the relationship between the two and the difference in influence intensity for different TE factors in NTL. This paper is one of only a few studies investigating the spatial relationship between NTL and TE, thus contributing significantly to the existing body of knowledge. The main objectives of this study are: firstly, to analyze the spatial relationship and distribution characteristics of NTL and TE, and secondly, to investigate the influence of TE on the spatial distribution of NTL.

The remainder of the article is structured as follows. Section 2 introduces the study area and data sources. Section 3 describes the methods of this research. Section 4 gives empirical results and describes the Spatial Agglomeration Pattern of NTL and TE, as well as the driving factors. Section 5 contains some discussion. Finally, Section 6 concludes the paper.

2. Study Area and Data

2.1. Study Area

As the second largest economy in the world, China has contributed greatly to the development of the world economy (Figure 1). With rapid socio-economic development, the level of urban modernization and the added value of the tourism industry has continued to increase, and China has now become a main driving force in the development of tourism worldwide, with a wealth of research being conducted on China’s tourism. At present, due to the lack of statistical data for Hong Kong, Macau and Taiwan, this paper takes 31 provincial-level administrative districts of China (including 22 provinces, 5 autonomous regions and 4 municipalities; in the following, province stands for provincial-level administrative district) as the study area and divides the study area into six major regions: Northeast, Northeast, South-Central, Northwest and Southwest China. Among these, East, North and South-Central China have better natural conditions, with higher population densities and higher levels of socio-economic development. Meanwhile, Southwest and Northwest China have fewer resources, smaller populations and relatively lower rates of development at the socio-economic level, and Northeast China is in between the two. On this basis, this paper analyzes the spatial distribution of the relationship between NTL and TE by using multiple data indicators combined with different research methods, which is expected to sustainably enrich the research results in related fields.

2.2. Study Data

2.2.1. Nighttime Light Data (NTL Data)

NTL data were divided into DMSP-OLS imagery data and NPP-VIIRS imagery data, which were published and made available to the public free of charge at (https://ngdc.noaa.gov/eog/download.html accessed on 4 April 2022) before October 2019 by the National Centers for Environmental Information (NCEI), a division of the National Oceanic and Atmospheric Administration (NOAA). On 15 October 2019, NCEI issued a statement discontinuing the NPP-VIIRS series of products, and the successor products were supplied to the public free of charge by the Payne Institute for Public Policy at the Colorado School of Mines, a public university in Golden, CO, USA (https://payneinstitute.mines.edu/eog/nightti-me-lights/ accessed on 4 April 2022). Three types of data are available in the NPP-VIIRS products at present. The first is monthly cloud-free coverage data. The second is the first version of the annual data in which the synthesis process is crude. The processing of outliers is performed on the scatter plot generated for each 15 arcsecond grid cell. During the process, the outliers are clipped from the high- and low-radiance sides of the scatter plot and the removal process continues until the standard deviation of the scatter plot is stable and constant. The third is the second version of the annual data and is derived from an improvement to the first version, with 12-month median radiance used to remove high- and low-radiance anomalies, filtering out most fires and removing noise. We obtained the NTL data from 31 provinces of China in 2019 based on the second version of the annual data, i.e., NTL, through the process of coordinate transformation, mask extraction, re-sampling, noise removal, outlier removal, etc.

2.2.2. Tourism Economy Data

When obtaining the basic TE data, we consulted the relevant statistical yearbooks of China’s tourism industry, which were The Yearbook of China Tourism Statistics, The Yearbook China Cultural Heritage and Tourism Statistics, and The Statistical Bulletin of National Economic and Social Development of some regions, and finally obtained the completed TE basic data for 2019 by unit transformation and by filling in missing values, etc., for the collected data (

Table 1. Essential Data of TE.

	Index	Unit & Illustration
Essential Data of Tourism Economy	Domestic tourism revenue	100,000,000 USD
	Inbound tourism revenue	100,000,000 USD
	Domestic tourist arrivals	10,000 person–times
	Inbound tourist arrivals	10,000 person–times
	Abundance of tourism Resources	(Number of 5A ∗ 5 + number of 4A ∗ 2.5 + number of 3A ∗ 1.5 + number of 2A 3 0.75 + number of A ∗ 0.25)
	Number of star hotels	a
	Number of travel agencies	a
	Revenue of A-level scenic spots	10,000 USD
	Revenue of star hotels	10,000 USD
	Traffic convenience	(Highway mileage + railway mileage + river mileage)/administrative area
	Population	10,000 person
	Tertiary industry	100,000,000 USD

).

3. Methodologies

3.1. Geographic Concentration Index

Geographic concentration stands for the interrelationship between a factor and the size of the area in which it is located, reflecting the relative position and role of the factor in the whole region, as well as the spatial concentration characteristics [45,46]. Usually, a higher geographical concentration value indicates a higher degree of concentration, while a smaller value indicates a lower degree. At present, this indicator is widely used in population, economy and other social development aspects, and the calculation formula is as follows:

$$\mathrm{R}=\frac{\raisebox{1ex}{${\mathrm{K}}_{\mathrm{i}}$}\!\left/ \!\raisebox{-1ex}{${{\displaystyle \sum }}_i^{31}{\mathrm{K}}_{\mathrm{i}}$}\right.}{\raisebox{1ex}{${\mathrm{A}}_{\mathrm{i}}$}\!\left/ \!\raisebox{-1ex}{${{\displaystyle \sum }}_{\mathrm{i}}^{31}{\mathrm{A}}_{\mathrm{i}}$}\right.}$$

(1)

where R stands for the geographical concentration; K_i stands for the relevant factors of the ith region, and the factors involved in the study are the NTL, the number of A-level scenic spots, the number of star hotels, the number of travel agencies, the domestic tourism arrivals, the inbound tourism arrivals, the domestic tourism revenue, the inbound tourism revenue, the star hotel operating revenue, the operating revenue of the A-level scenic spot, the transportation mileage, etc. A_i stands for the administrative area of the ith province.

3.2. Inconsistency Index

The inconsistency index was first used in mathematical research to analyze the relationship between the overall variance and the sampling variance, on the basis of which this index can be widely used to analyze the relative coupling relationship between two variables. In this study, the relative agglomeration relationship between NTL and the TE factors will be analyzed in depth using the inconsistency index, and the calculation expression is the ratio of the geographic concentration of TE factors to the geographic concentration of NTL:

$${\mathrm{R}}_{\mathrm{AB}}=\frac{{\mathrm{R}}_{\mathrm{A}}}{{\mathrm{R}}_{\mathrm{B}}}$$

(2)

where R_AB denotes the inconsistency index. When R_A is less than 1, this indicates that the agglomeration of factor A lags behind that of factor B. When R_AB is equal to 1, this indicates that the two are developing in harmony. When R_AB is greater than 1, this indicates that the agglomeration of factor A is ahead of factor B [47,48,49].

3.3. Spatial Agglomeration Coupling Index

The spatial agglomeration coupling index is used to study the spatial agglomeration coupling of factors, and we will further elaborate on the spatial distribution characteristics of the two by calculating the spatial coupling relationship between NTL and TE factors. Recognition of such spatial interdependency in the standard procedure to define neighborhood relies on the distance between geometric means of territorial units which, so we use geometric means [50,51,52] to reflect the spatial agglomeration coupling and study the spatial agglomeration coupling of factors. The calculation formula is as follows:

$${\mathrm{I}}_{\mathrm{AB}}=\sqrt{{\mathrm{R}}_{\mathrm{A}} \ast {\mathrm{R}}_{\mathrm{B}}}$$

(3)

where I_AB indicates the spatial agglomeration coupling index between factor A and factor B. RA and RB indicate the geographical concentration of factor A and factor B. A larger IAB value indicates a stronger spatial agglomeration coupling ability between factors, while a smaller value indicates a weaker spatial coupling agglomeration ability.

3.4. Global Moran’s I and Local Moran’s I_i

Spatial autocorrelation analysis is used to study the degree of correlation among different provinces. According to the first law of geography, the closer the spatial location, the greater the degree of correlation among the things. Spatial autocorrelation analysis has been applied in many fields, and two types of research models, namely, global autocorrelation and local spatial autocorrelation, are usually used. Therefore, this section adopts Global Moran’s I [53] and Local Moran’s I_i [54], respectively, to observe the spatial relationships between 31 provinces, and the Moran’s index is calculated by using the following formula:

$$\mathrm{I}=\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\left({\mathrm{x}}_{\mathrm{i}}- \overline{\mathrm{x}}\right)}{{\mathrm{s}}^2{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}}\begin{array}{cc}& \end{array}{\mathrm{s}}^2=\frac{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}\left({\mathrm{x}}_{\mathrm{i}}- \overline{\mathrm{x}}\right)}{\mathrm{n}}$$

(4)

$${\mathrm{I}}_{\mathrm{i}}=\frac{\left({\mathrm{x}}_{\mathrm{i}}- \overline{\mathrm{x}}\right)}{{\mathrm{s}}^2}{\displaystyle \sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{ij}}\left({\mathrm{x}}_{\mathrm{j}}- \overline{\mathrm{x}}\right)$$

(5)

$${\mathrm{W}}_{\mathrm{ij}}=\{\begin{array}{c}1\\ 0\end{array}\begin{array}{ccc}& & \end{array}\begin{array}{c}\begin{array}{cc}\mathrm{i}\&\mathrm{j}& \mathrm{adjacent}\end{array}\\ \mathrm{i}\&\mathrm{j}\begin{array}{cc}\mathrm{not}& \mathrm{adjacent}\end{array}\end{array}$$

(6)

Equation (4) represents the Global Moran’s index model and Equation (5) represents the Local Moran’s index model. I stands for the Global Moran’s index. I_i stands for the Local Moran’s index. n is the total number of the study sample, i.e., 31 provinces. X_i and X_j represent the specific values of the neighboring regions i and j, respectively, which refer to the spatial agglomeration coupling index of NTL and TE. The adjacency weight matrix is calculated by setting the latitude and longitude coordinate bandwidth of each province to 12 through the “spatwmat” command in Stata, and is presented in Equation (6).

3.5. Entropy Method

The entropy method is an objective weighting method which refers to the determination of indicator weights based on the magnitude of the information provided by the observations of each indicator. In information theory, entropy is a measurement of uncertain information. The higher the amount of information, the lower the uncertainty and entropy, and vice versa. Compared with other methods, the entropy method can effectively avoid the subjectivity of index weights and can be objectively evaluated [55]. The TEL value was obtained by weighting entropy value to the selected factors of TE, and the calculation formulas are as follows:

$${\mathrm{Z}}_{\mathrm{ij}}={\frac{{\mathrm{X}}_{\mathrm{ij}}-\mathrm{MIN}\left({\mathrm{X}}_{\mathrm{j}}\right)}{\mathrm{MAX}\left({\mathrm{X}}_{\mathrm{j}}\right)-\mathrm{MIN}\left({\mathrm{X}}_{\mathrm{j}}\right)}}_{\begin{array}{cc}\mathrm{i}=1,2,3\dots \mathrm{n}& \mathrm{j}=1,2,3\dots \mathrm{m}\end{array}}$$

(7)

$${\mathrm{P}}_{\mathrm{ij}}={\frac{{\mathrm{Z}}_{\mathrm{ij}}}{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{Z}}_{\mathrm{ij}}}}_{\begin{array}{cc}\mathrm{i}=1,2,3\dots \mathrm{n}& \mathrm{j}=1,2,3\dots \mathrm{m}\end{array}}$$

(8)

$${\mathrm{e}}_{\mathrm{j}}=-\mathrm{k}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{P}}_{\mathrm{ij}}\mathrm{Ln}\left({\mathrm{P}}_{\mathrm{ij}}\right) {\mathrm{d}}_{\mathrm{j}}=1-{\mathrm{e} \mathrm{w}}_{\mathrm{j}}=\frac{{\mathrm{d}}_{\mathrm{j}}}{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{d}}_{\mathrm{j}}}$$

(9)

$${\mathrm{T}={\displaystyle \sum }_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{w}}_{\mathrm{j}}{\mathrm{P}}_{\mathrm{ij}}}_{\begin{array}{c}\left(\mathrm{i}=1,2,3\dots \mathrm{n}\right)\end{array}}$$

(10)

where Z_ij is the standardized value of the jth indicator of sample i. P_ij is the proportion of the jth indicator in the ith sample of that indicator. e_j is the entropy value of the jth indicator, in which k is the reciprocal of the sample Ln(n). d_j is the effective information entropy of the jth indicator. w_j is the weight of the jth indicator and T is the comprehensive score of the sample, which, in this paper, refers to the TEL value.

3.6. Geographical Detector

3.6.1. Differentiation and Factor Analysis

This is used to detect the spatial differentiation of the studied factors [56] and the strength of its influence factor X in explaining the studied factor Y. The spatial differentiation is indicated as q and calculated as follows:

$$\begin{gathered} \begin{array}{l}\mathrm{q}=1-\frac{{{\displaystyle \sum }}_{\mathrm{h}=1}^{\mathrm{L}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2}{{\mathrm{N}\mathsf{\sigma }}^2}=1-\frac{\mathrm{SSW}}{\mathrm{SST}}\\ \\ \mathrm{SSW}={\displaystyle \sum _{\mathrm{h}=1}^{\mathrm{L}}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2\\ \\ \mathrm{SST}={\mathrm{N}\mathsf{\sigma }}^2\end{array} \end{gathered}$$

(11)

where h = 1, …, L is the stratification (Strata) of variable Y or factor X; N_h and N are the number of units in stratum h and the whole area, respectively; σ2 h and σ2 are the variance of Y in stratum h and the whole area, respectively. SSW and SST are the sum of within Sum of Squares (SSW) and Total Sum of Squares (SST), respectively. The range of q is [0, 1]. When q values 0, this indicates that X has no effect on Y. A higher q value indicates that X explains Y more strongly. When q values are 1, this indicates that factor X completely controls Y. q indicates that X can explain 100 * q% of Y. The factors involved in this study are shown in

Table 2. Variables and Factors of Geographical Detector.

Variables		Factor
TEL NTL	Tourism Economy Level Nighttime Lighting	DTA	Domestic tourist arrivals
		ITA	Inbound tourist arrivals
		ATR	Abundance of tourism resources
		NSH	Number of star hotels
		NTA	Number of travel agencies
		RA-LSS	Revenue of A-level scenic spots
		RSRH	Revenue of star hotels
		TC	Traffic convenience
		POP	Population
		TI	Tertiary industry

.

A simple change in the q value satisfies the noncentral F distribution:

$$\begin{gathered} \begin{array}{l}\mathrm{F}=\frac{\mathrm{N}-\mathrm{L}}{\mathrm{L}-1}\frac{\mathrm{q}}{1-\mathrm{q}}~\mathrm{F}\left(\mathrm{L}-1,\mathrm{N}-\mathrm{L};\mathsf{\lambda }\right)\\ \\ \mathsf{\lambda }=\frac{1}{{\mathsf{\sigma }}^2}\left[{\displaystyle \sum _{\mathrm{h}=1}^{\mathrm{L}}}{\overline{{\mathrm{Y}}_{\mathrm{h}}}}^2-\frac{1}{\mathrm{N}}{{\displaystyle (\sum _{\mathrm{h}=1}^{\mathrm{L}}}\sqrt{{\mathrm{N}}_{\mathrm{h}}}\overline{{\mathrm{Y}}_{\mathrm{h}}})}^2\right]\end{array} \end{gathered}$$

(12)

where $\mathsf{\lambda }$ stands for the noncentral parameter and $\overline{{\mathrm{Y}}_{\mathrm{h}}}$ stands for the mean value of layer h. Equation (12) can be used to determine whether the q value is significant [57].

3.6.2. Interaction Detection

This is used to detect and assess the degree of influence on the studied factors when different factors act simultaneously, as well as the direction of the effect. The five interaction types are shown as follows in Figure 2 [57].

3.6.3. Ecological Detection

This is used to assess whether the effect on the dependent variable is significant when the two factors interact, and is expressed by the statistic F as follows:

$$\begin{gathered} \begin{array}{l}\mathrm{F}=\frac{{\mathrm{N}}_{\mathrm{X}1}\left({\mathrm{N}}_{\mathrm{X}2}-1\right){\mathrm{SSW}}_{\mathrm{X}1}}{{\mathrm{N}}_{\mathrm{X}2}\left({\mathrm{N}}_{\mathrm{X}1}-1\right){\mathrm{SSW}}_{\mathrm{X}2}}\\ \\ {\mathrm{SSW}}_{\mathrm{X}1}={\displaystyle \sum _{\mathrm{h}=1}^{\mathrm{L}1}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2\\ \\ {\mathrm{SSW}}_{\mathrm{X}2}={\displaystyle \sum _{\mathrm{h}=1}^{\mathrm{L}2}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2\end{array} \end{gathered}$$

(13)

where N_X1 and N_X2 denote the sample sizes of the two factors X1 and X2, respectively; SSW_x1 and SSW_x2 denote the sum of within-layer variances of the strata formed by X1 and X2, respectively; L1 and L2 denote the number of strata of the variables X1 and X2, respectively. It is almost impossible for SSWx1 is equal to SSWx2, so there is a significant difference in the effect of the two factors of X1 and X2 on the spatial distribution of attribute Y.

4. Results

4.1. The Spatial Agglomeration Pattern of NTL and TE

In this paper, the spatial distribution, the spatial imbalance and the coordination degree of NTL and TE were analyzed through the geographical concentration index and inconsistency index to better reflect the spatial clustering characteristics of both and the clustering differences in the same region. Then, the spatial clustering coupling index was used to measure the joint development relationship between NTL and TE to more objectively compare the differences between provinces. Finally, in order to more comprehensively derive the inner spatial connection between provinces, a spatial autocorrelation analysis was conducted on the spatial agglomeration coupling between NTL and TE with the help of Stata software in order to show the spatial agglomeration coupling association between NTL and TE at a deeper level, and the spatial agglomeration pattern of the two was obtained by combining the above three analyses.

4.1.1. Analysis of Spatial Agglomeration Characteristics

The geographical concentration index of TE and NTL in each province in 2019 was calculated by Formula (1) (

Table 3. Geographical Concentration Index of NTL and TE.

Region	Province	NTL	NA-LSS	NSH	NTA	DTA	ITA	DTR	ITR	RSRH	RA-LSS	TM
North China	Beijing	1.2058	1.1719	1.3632	1.4381	1.2258	1.3519	1.2485	1.3824	1.3829	1.2532	1.0574
	Tianjin	1.2348	1.0087	0.9878	1.1524	1.2359	1.2366	1.1096	1.2692	1.1470	0.8215	1.0600
	Hebei	1.0599	1.0330	1.0321	1.0477	1.0636	0.9530	1.0494	0.9607	1.0198	0.9721	1.0251
	Shanxi	1.0535	0.9336	0.9534	0.9927	1.0865	0.8010	1.0488	0.9049	0.9665	1.1171	1.0148
	Inner Mongolia	0.8832	0.8808	0.8276	0.8759	0.8097	0.8346	0.8412	0.8606	0.8368	0.6416	0.8964
Northeast China	Liaoning	1.0478	1.0888	1.0439	1.0702	1.0662	1.0558	1.0222	1.0331	1.0166	0.9124	1.0084
	Jilin	0.9781	0.9285	0.8110	0.9353	0.9527	0.8934	0.9736	0.9228	0.9025	0.8736	0.9730
	Heilongjiang	0.9599	0.9567	0.8387	0.8950	0.8775	0.7963	0.8400	0.8617	0.8601	0.9170	0.9430
East China	Shanghai	1.3084	1.0922	1.2683	1.4443	1.3387	1.6719	1.3179	1.5457	1.4843	1.2431	1.0929
	Jiangsu	1.1351	1.1561	1.1322	1.2031	0.9531	0.8519	0.9566	1.0115	1.1399	1.2450	1.0715
	Zhejiang	1.0991	1.1994	1.1772	1.1903	1.1085	1.1741	1.1193	1.1001	1.1440	1.4401	1.0388
	Anhui	1.0368	1.1225	1.0251	1.0710	1.0946	1.2097	1.0625	1.0946	1.0360	1.2161	1.0599
	Fujian	1.0431	1.0486	1.0494	1.0471	1.0619	1.2926	1.0590	1.2011	1.0804	0.8906	1.0115
	Jiangxi	0.9712	1.0421	1.0350	0.9829	1.0738	0.9698	1.0625	0.9617	0.9967	1.4115	1.0396
	Shandong	1.1017	1.2362	1.1306	1.1439	0.8915	1.1570	1.3362	1.0860	1.0859	1.2457	1.0706
South-Central China	Henan	1.0639	1.0807	1.0650	1.0200	1.0885	1.0785	1.0640	0.9987	1.0344	1.0236	1.0619
	Hubei	0.9923	1.0340	1.0382	1.0228	1.0396	1.1126	1.0137	1.0483	1.0130	1.1482	1.0584
	Hunan	0.9654	1.0447	1.0103	0.9962	1.0571	1.1061	1.0420	1.0223	1.0030	1.2354	1.0326
	Guangdong	1.0932	1.0314	1.1413	1.1594	1.0213	1.4983	1.0958	1.2205	1.1042	1.2050	1.0383
	Guangxi	0.9717	1.0586	1.0428	0.9440	1.0504	1.1458	1.0352	1.0478	0.9633	0.9703	0.9707
	Hainan	1.0608	0.8140	0.9522	1.0197	0.9808	1.0425	0.9143	1.1111	1.1355	0.9426	1.0230
Southwest China	Chongqing	0.9936	1.0067	0.9726	0.9992	1.1226	1.1750	1.0628	1.1191	1.0319	1.0493	1.0882
	Sichuan	0.9255	1.0336	0.9807	0.9448	0.9824	1.0169	0.9954	0.9500	1.0144	1.3272	0.9924
	Guizhou	0.9425	1.0367	0.9725	0.9171	1.1028	0.9285	1.0860	0.8797	0.9595	1.1682	1.0318
	Yunnan	0.9483	0.9405	1.0160	0.9445	1.0041	1.2521	1.0041	1.0394	0.9361	0.8934	0.9876
	Tibet	0.6937	0.7182	0.7920	0.7116	0.6780	0.6300	0.6269	0.7445	0.7203	0.2500	0.8399
Northwest China	Shaanxi	1.0059	1.0411	1.0054	0.9603	1.0462	1.1104	1.0099	1.0602	0.9986	1.0636	1.0101
	Gansu	0.9024	0.7603	0.9644	0.8826	0.9316	0.5094	0.8499	0.6247	0.8880	0.8638	0.9397
	Qinghai	0.7820	0.8866	0.8613	0.8064	0.7271	0.3266	0.6562	0.6143	0.8102	0.4168	0.8604
	Ningxia	1.0156	0.8736	0.8988	0.8167	0.8764	0.5156	0.7480	0.8240	0.8782	0.5874	0.9899
	Xinjiang	0.8596	0.9004	0.8642	0.7636	0.7977	0.8007	1.0225	0.9451	0.8153	0.9567	0.8683

) and the geographical concentration of these places was spatially visualized through ArcGIS (Figure 3).

Table 2 shows the geographical concentration indices of NTL and TE factors. The highest concentration of NTL, number of travel agencies, domestic tourism arrivals, inbound tourism arrivals, inbound tourism revenue, revenue of star hotels and traffic mileage are in Shanghai, which is a fashionable port city with a developed economy and well-developed infrastructure. With its superior location and rich history and culture, Shanghai attracts a large number of domestic and foreign tourists. The highest concentration of A-level scenic spots and domestic tourism revenue is in Shandong, as it has the highest number of A-level scenic spots in the country and is rich in Confucian culture as well as its own various tourism products. Thus, domestic tourists’ tourism consumption in Shandong is higher and more diverse. The highest concentration of star hotels is in Beijing, because Beijing, as the capital of China, is the center of domestic and international exchange and communication, and hotels undertake important hospitality tasks. Regarding the revenue of A-level scenic spots, the highest concentration is in Zhejiang. Through further analysis of the standard deviation of the geographic concentration of all factors, we found that although Shanghai has a higher degree of concentration in all aspects, its coefficient of variation is also the largest, which indicates that the concentration situation among TE factors in Shanghai differs widely. The smallest coefficient of variation is in Henan Province, so it can be concluded that the concentration of each TE factors in Henan Province is more even. Figure 3 shows the following trends: there is a higher concentration of NTL located mainly in North and East China (a); the number of star hotels (c), domestic tourism revenue (g), inbound tourism revenue (h), revenue of star hotels (j) and revenue of A-level scenic spots (i) have a higher geographical concentration of distribution in East and South-Central China; the number of travel agencies has a higher concentration of distribution in East China (d); and the number of A-level scenic spots (b), domestic tourism arrivals (e) and traffic mileage (k) are more concentrated in North, East and South-Central China.

The inconsistency index of NTL and elements of TE is calculated by Formula (2). Using the ArcGIS software, the relative agglomeration type distribution of the two was plotted as shown in Figure 4, which can be divided into three types. First, the type of TE factors that surpass NTL. This type is mainly distributed in Southwest and South-Central China. Among these, the more obvious TE factors include the number of A-level scenic spots (a), the number of star hotels (b), domestic tourism arrivals (d), domestic tourism revenue (f), and inbound tourism revenue (h), etc., indicating that the Southwest and South-Central regions have invested more in tourism infrastructure and the construction of A-level scenic spot in recent years and have thus attracted a large number of domestic and foreign tourists. Second, the coordinated development type. Except for the scenic spot income, other TE factors and NTL in North China are shown as the coordinated development type. Meanwhile, the number of star hotels (a), the number of travel agencies (c), inbound tourism revenue (g), revenue of A-level scenic spots (i) and traffic mileage (j) also show a coordinated development type in Northeast China. Third, the type of NTLs that surpass TE. This type of distribution is more scattered and the performance is more obvious regionally. Gansu’s NTL is surpassing in terms of the number of A-level scenic spots (a), inbound tourism trips (e), inbound tourism revenue (g) and star hotel revenue (h); Qinghai’s NTL is surpassing in terms of inbound tourism arrivals (e), domestic tourism revenue (f), inbound tourism revenue (g) and revenue of A-level scenic spots (h); Jiangsu’s NTL is surpassing in terms of domestic tourism trips (d), inbound tourism trips (e) and domestic tourism revenue (f); Jiangsu’s NTL is surpassing in terms of domestic tourism arrivals (d), inbound tourism arrivals (e) and domestic tourism revenue (f). In terms of inbound tourism arrivals, NTL in Gansu, Qinghai and Jiangsu all show the type of NTL-surpassing TE, suggesting that these three provinces have fewer inbound tourists.

4.1.2. Spatial Clustering Coupling Characteristics of NTL and TE

The spatial agglomeration coupling index of NTL and TE factors was calculated by Equation (3) (

Table 4. Spatial Agglomeration Coupling Index of NTL and TE.

Region	Province	NA-LSS	NSH	NTA	DTA	ITA	DTR	ITR	RSRH	RA-LSS	TM
North China	Beijing	1.1887	1.2821	1.3168	1.2157	1.2768	1.2270	1.2911	1.2913	1.2292	1.1291
	Tianjin	1.1160	1.1044	1.1929	1.2353	1.2357	1.1705	1.2519	1.1901	1.0072	1.1441
	Hebei	1.0464	1.0459	1.0537	1.0617	1.0050	1.0546	1.0091	1.0396	1.0151	1.0423
	Shanxi	0.9917	1.0022	1.0226	1.0699	0.9186	1.0511	0.9764	1.0091	1.0848	1.0340
	Inner Mongolia	0.8820	0.8549	0.8795	0.8456	0.8586	0.8620	0.8718	0.8597	0.7528	0.8898
Northeast China	Liaoning	1.0681	1.0458	1.0589	1.0569	1.0518	1.0349	1.0404	1.0321	0.9777	1.0279
	Jilin	0.9530	0.8906	0.9564	0.9653	0.9347	0.9758	0.9501	0.9395	0.9244	0.9755
	Heilongjiang	0.9583	0.8973	0.9269	0.9178	0.8743	0.8980	0.9095	0.9086	0.9382	0.9514
East China	Shanghai	1.1954	1.2882	1.3747	1.3235	1.4790	1.3131	1.4221	1.3936	1.2753	1.1958
	Jiangsu	1.1456	1.1337	1.1686	1.0401	0.9834	1.0421	1.0715	1.1375	1.1888	1.1028
	Zhejiang	1.1482	1.1375	1.1438	1.1038	1.1360	1.1091	1.0996	1.1213	1.2581	1.0685
	Anhui	1.0788	1.0309	1.0537	1.0653	1.1199	1.0496	1.0653	1.0364	1.1229	1.0483
	Fujian	1.0458	1.0463	1.0451	1.0525	1.1612	1.0510	1.1193	1.0616	0.9639	1.0272
	Jiangxi	1.0060	1.0026	0.9771	1.0212	0.9705	1.0158	0.9665	0.9839	1.1709	1.0048
	Shandong	1.1670	1.1160	1.1226	0.9910	1.1290	1.2133	1.0938	1.0938	1.1715	1.0861
South-Central China	Henan	1.0723	1.0644	1.0417	1.0761	1.0711	1.0640	1.0308	1.0490	1.0436	1.0629
	Hubei	1.0129	1.0150	1.0074	1.0157	1.0507	1.0029	1.0199	1.0026	1.0674	1.0248
	Hunan	1.0043	0.9876	0.9807	1.0102	1.0334	1.0030	0.9935	0.9841	1.0921	0.9985
	Guangdong	1.0619	1.1170	1.1258	1.0566	1.2798	1.0945	1.1551	1.0986	1.1477	1.0654
	Guangxi	1.0142	1.0066	0.9578	1.0103	1.0552	1.0030	1.0090	0.9675	0.9710	0.9712
	Hainan	0.9292	1.0051	1.0401	1.0200	1.0516	0.9848	1.0857	1.0975	1.0000	1.0417
Southwest China	Chongqing	1.0001	0.9830	0.9964	1.0561	1.0805	1.0276	1.0545	1.0125	1.0210	1.0398
	Sichuan	0.9781	0.9527	0.9351	0.9536	0.9701	0.9599	0.9377	0.9690	1.1083	0.9584
	Guizhou	0.9885	0.9574	0.9297	1.0195	0.9355	1.0117	0.9106	0.9510	1.0493	0.9861
	Yunnan	0.9444	0.9816	0.9464	0.9758	1.0897	0.9758	0.9928	0.9422	0.9205	0.9678
	Tibet	0.7059	0.7412	0.7026	0.6858	0.6611	0.6594	0.7187	0.7069	0.4165	0.7633
Northwest China	Shaanxi	1.0234	1.0057	0.9828	1.0259	1.0569	1.0079	1.0327	1.0022	1.0343	1.0080
	Gansu	0.8283	0.9329	0.8925	0.9169	0.6780	0.8758	0.7509	0.8952	0.8829	0.9209
	Qinghai	0.8326	0.8207	0.7941	0.7540	0.5054	0.7163	0.6931	0.7960	0.5709	0.8202
	Ningxia	0.9419	0.9554	0.9107	0.9434	0.7237	0.8716	0.9148	0.9444	0.7723	1.0026
	Xinjiang	0.8798	0.8619	0.8102	0.8281	0.8297	0.9375	0.9014	0.8371	0.9069	0.8639

), and the spatial agglomeration coupling index of the two was visualized with the help of ArcGIS (Figure 5).

From Table 4, it can be seen that the average spatial agglomeration coupling index of tourism economy factors and NTL in 2019 in each province is 1.0084. The regions below the average level are Inner Mongolia, Jilin Province, Heilongjiang, Guangxi, Sichuan, Guizhou, Yunnan, Tibet, Gansu Province, Qinghai, Ningxia and Xinjiang. The spatial agglomeration coupling degree of all TE factors and NTL in Shanghai is higher than that in other regions. The spatial agglomeration coupling degree of all TE factors and NTL in Tibet is the lowest in China, except for inbound tourism revenue and inbound tourism arrivals. The lowest spatial agglomeration coupling degree of inbound tourism revenue and inbound tourism trips and NTL is in Qinghai Province. As can be seen from Figure 5, the spatial agglomeration coupling index of each factor with NTL is generally higher in North, East and South-Central China. The more obvious differences in Northwest China are in Xinjiang and Gansu Province. In Xinjiang, the spatial agglomeration coupling indexes of the number of A-level scenic spots (c), domestic tourism revenue (g), revenue of A-level scenic spots (j) and NTL are slightly higher than those of other regions, while in Gansu Province, the spatial agglomeration coupling indexes of the number of star hotels (a), traffic mileage (d), domestic tourism arrivals (e), inbound tourism arrivals (f) and NTL are slightly higher than those of other regions.

4.1.3. Spatial Clustering Association between NTL and TE

Table 5. Global Moran’s I of Spatial Agglomeration Coupling Between NTL and TE factors.

Variables	I	E(I)	sd(I)	z	p-Value *
NA-LSS	0.451	−0.033	0.098	4.958	0.000
NSH	0.368	−0.033	0.097	4.135	0.000
NTA	0.396	−0.033	0.097	4.436	0.000
DTA	0.240	−0.033	0.096	2.841	0.002
ITA	0.290	−0.033	0.097	3.338	0.000
DTR	0.323	−0.033	0.096	3.702	0.000
ITR	0.325	−0.033	0.096	3.712	0.000
RSRH	0.351	−0.033	0.096	4.003	0.000
RA-LSS	0.336	−0.033	0.095	3.911	0.000
TM	0.310	−0.033	0.097	3.543	0.000

* 1-tail test.

shows the results of the Global Moran’s I analysis of the coupling index of spatial agglomeration of NTL and TE factors. Firstly, the p-values are all significant at 0.01 significant level, indicating that their spatial relationships are significant. Secondly, the Moran’s I is greater than 0, which indicates that there is a positive spatial relationship. The number of A-level scenic spots and the spatial agglomeration coupling of NTL show a strong spatial autocorrelation, and the number of domestic tourists and the spatial agglomeration coupling of NTL have a weak spatial autocorrelation. Figure 6 shows the Local Moran’s I_i scatter plot of the spatial agglomeration coupling index of different TE factors and NTL. From the figure, it can be noted that the clusters of spatial agglomeration coupling index of TE and NTL can be divided into four types. The first quadrant stands for provinces with high–high (high coupling surrounded by high coupling), the third quadrant stands for provinces with low–low (low coupling surrounded by low coupling), and the first and the third quadrants stand for unidirectional agglomeration regions. The second quadrant is for provinces with high–low (high coupling surrounded by low coupling), the fourth quadrant is for provinces with low–high (low coupling surrounded by high coupling), and the second and fourth quadrants are reverse agglomeration regions. In terms of the spatial coupling index between NTL and TE factors, the provinces in East China and South-Central China, except Shanghai and Hubei Province, all appeared in the first quadrant, while Southwest, Northwest and Northeast China all appeared in the third quadrant. More than half of the provinces in East and South China appeared in the third quadrant, while fewer appeared in the fourth quadrant; these are Fujian Province, Guangxi, Hainan Province, Henan Province, Hunan Province, Jiangsu Province, Jiangxi Province, Qinghai Province, Shandong Province and Yunnan Province.

4.2. Analysis of Driving Factors

4.2.1. Analysis of Driving Factors of Tourism Economy Level

In order to further explore the relationship between NTL and TE, firstly, the comprehensive score of each factor of TE was found using the entropy value method and defined as TEL. Secondly, the influence strength of each factor on TEL and NTL was analyzed using a geographic detector, and the influence difference for the same factor on TEL and NTL was comparatively analyzed.

(1)

Factor Detector

The purpose of differentiation and factor detection is to analyze the explanatory strength of each factor to the explained factor. The value q of the explanatory strength of each factor compared to the spatial differentiation of TEL was obtained by differentiation and factor detection (

Table 6. Factor Detector of TEL.

	DTA	ITA	ATR	NSH	NTA	RA-LSS	RSRH	TC	POP	TI
q statistic	0.893	0.708	0.807	0.919	0.721	0.780	0.671	0.721	0.957	0.963
p value	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000

). From Table 6, it can be found that the tertiary industry has the strongest explanatory strength for the TEL, followed by local population and then the number of star hotels. Thus, the main factors affecting the spatial differentiation of TEL are the above three factors.

(2)

Interaction Detector

Interaction detection analysis was used to assess the influence degree and the direction of the effect on TEL when different factors act simultaneously. The results are shown in

Table 7. Interaction Detector of TEL.

	DTA	ITA	ATR	NSH	NTA	RA-LSS	RSRH	TC	POP	TI
DTA	0.893
ITA	0.97	0.708
ATR	0.987	0.973	0.807
NSH	0.986	0.975	0.98	0.919
NTA	0.990	0.982	0.987	0.98	0.721
RA-LSS	0.979	0.98	0.967	0.977	0.942	0.780
RSRH	0.962	0.908	0.893	0.977	0.964	0.963	0.671
TC	0.998	0.976	0.992	0.995	0.94	0.950	0.984	0.721
POP	0.981	0.992	0.995	0.996	0.995	0.991	0.973	1.00	0.957
TI	0.995	0.977	0.982	0.978	0.98	0.970	0.979	0.998	0.991	0.963

: the local population together with inbound tourism arrivals, the abundance of tourism resources, the number of star hotels, the number of travel agencies, the revenue of A-level scenic spots and the convenience of transportation act simultaneously, which has a great impact on the TEL. The interaction value with the convenience of transportation is 1, which is the strongest interaction group, followed by the interaction value of 0.998 when the number of domestic tourist arrivals and traffic convenience act at the same time. The next interaction is local population and tertiary industry, valuing 0.991. On this basis, a superposition calculation of each factor shows that the interactions are all dual-factor enhancement types.

(3)

Ecological Detector

Whether there is a significant difference in the effect of the two factors on the spatial distribution of TEL is analyzed by ecological detection (Equation (13)); the detection results are shown in

Table 8. Ecological Detector TEL.

	DTA	ITA	ATR	NSH	NTA	RA-LSS	RSRH	TC	POP	TI
DTA
ITA	Y
ATR	Y	Y
NSH	Y	Y	Y
NTA	Y	Y	Y	Y
RA-LSS	Y	Y	Y	Y	Y
RSRH	Y	Y	Y	Y	Y	Y
TC	Y	Y	Y	Y	N	Y	Y
POP	Y	Y	Y	Y	Y	Y	Y	Y
TI	Y	Y	Y	Y	Y	Y	Y	Y	Y

. The influence differences of the number of travel agencies and traffic convenience on the TEL are not significant, while the effects of the other factors on the TEL are significantly different.

4.2.2. Analysis of Nighttime Lighting Driving Factors

(1)

Factor Detector

In order to further study the influence of TE on NTL, the same factors as the TE driving factors were adopted to detect the spatial differentiation factors of NTL. The detection results are shown in

Table 9. Factor Detector of NTL.

	DTA	ITA	ATR	NSH	NTA	RA-LSS	RSRH	TC	POP	TI
q statistic	0.828	0.575	0.720	0.578	0.402	0.650	0.652	0.215	0.758	0.823
p value	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000

. The factors with a greater influence on the TEL, such as domestic tourist arrivals, an abundance of tourism resources, the revenue of A-level scenic spots and revenue of star hotels, also have a greater influence on NTL. In addition, the population and the tertiary industry also have a greater explanatory strength for the spatial differentiation of NTL.

(2)

Interaction Detector

The analysis of factor interaction detection revealed (

Table 10. Interaction Detector of NTL.

	DTA	ITA	ATR	NSH	NTA	RA-LSS	RSRH	TC	POP	TI
DTA	0.828
ITA	0.949	0.575
ATR	0.960	0.883	0.720
NSH	0.904	0.903	0.928	0.578
NTA	0.981	0.899	0.950	0.926	0.402
RA-LSS	0.944	0.946	0.935	0.895	0.990	0.650
RSRH	0.949	0.899	0.877	0.904	0.978	0.900	0.652
TC	0.990	0.952	0.967	0.971	0.789	0.759	1.000	0.215
POP	0.937	0.978	0.960	0.973	0.965	0.989	0.914	0.994	0.758
TI	0.939	0.918	0.933	0.869	0.915	0.915	0.922	0.983	0.969	0.823

) that the strongest interactions were found between transportation convenience and domestic tourism arrivals, as well as the abundance of tourism resources and revenue of star hotels. The strongest interactions were found between the population and inbound tourism arrivals, number of star hotels, revenue of A-level scenic spots and transportation convenience. The interaction type among all factors is dual-factor enhancement types.

(3)

Ecological Detector

Whether there is a significant difference between the influence of the two factors on NTL is analyzed by ecological detection, and the detection results are shown in

Table 11. Ecological Detection of NTL.

	DTA	ITA	ATR	NSH	NTA	RA-LSS	RSRH	TC	POP	TI
DTA
ITA	Y
ATR	Y	Y
NSH	Y	N	Y
NTA	Y	Y	Y	Y
RA-LSS	Y	Y	Y	Y	Y
RSRH	Y	Y	Y	Y	Y	N
TC	Y	Y	Y	Y	Y	Y	Y
POP	Y	Y	Y	Y	Y	Y	Y	Y
TI	N	Y	Y	Y	Y	Y	Y	Y	Y

. The difference between the influence of domestic tourist arrivals and the output value of tertiary industry on the spatial distribution of NTL is not significant. The difference between the influence of inbound tourist arrivals and the number of star hotels on the spatial distribution of NTL is not significant. The difference between the influence of revenue of A-level scenic spots and the revenue of star hotels on the spatial distribution of NTL is not significant, but the influence of other factors on the spatial distribution of NTL is significantly different.

In summary, the influence of each factor on the TEL and NTL was analyzed through a geographical detector. The study results showed that there is a close relationship between NTL and TE, and the driving factors composed of TE elements make a significant contribution to NTL.

5. Discussion

As an important industry promoting China’s social and economic development, tourism is gradually becoming an indispensable part of people’s daily life, and the lighting of cities at night represents their vitality to a certain extent. In this study, spatial characteristics were studied in terms of the spatial agglomeration pattern and driving factors of both using NTL and the basic data of TE in 31 provinces in China.

After NTL data became widely used in various fields, scholars of tourism research initially used the NTL intensity index to characterize the intensity of human activities in the evaluation of tourism destination suitability [32]. Then, the application of night light data in domestic tourism research began. Based on a spatial panel econometric model, scholars conducted a regression analysis with TE, using the NTL intensity as the core explanatory variable. The model fit was high and could be used as a proxy variable for the TE [44]. Then, another scholar confirmed that NTLs are highly correlated with TE activities through OLS and GWR regression models [43] and that the decrease in the number of visiting tourists to destinations impacts the closure of most tourism facilities and reduces the number of employees and the brightness level of the nighttime light [28]. This is consistent with the conclusion that we reached. However, there is insufficient research on the spatial relationship between NTL and TE, and this paper selected more abundant TE variables based on existing research and analyzed the spatial clustering characteristics of NTL and several TE factors from a spatial perspective. Finally, the influence intensity of the factors comprised of TE factors on NTL and TEL was compared using geographic detector analysis, and it was found that most influencing factors have a significant impact on NTL. In summary, the research content of this paper is more detailed than previous scholars’ research and the conclusion about the relationship between NTL and TE is clearer, which not only fully proves the high correlation between NTL and TE and enriches the research results in related fields, but also shows the factors affecting the spatial differentiation of NTL and TE; furthermore, it can be combined with the spatial clustering pattern research on NTL and TE factors in the previous paper. It can reasonably allocate tourism resources for different regions and promote the sustainable development of NTL and TE among regions.

Since the research involved in this paper is relatively new in the field of tourism, there is still much room for improvement in terms of the research process and findings. Many studies indicate that NTL brightness can be a sound proxy of socioeconomic factors on large scales, such as the population size, gross domestic product, electric power consumption, etc. However, few studies have been dedicated to these topics on fine scales [58]. Likewise, we took 31 provinces as the research area, which is a large scale, and the data accuracy was limited after integrating the NTL data with TE, which may lead to unclear relationships in some areas. Follow-up studies could select a smaller scale on the basis of the available tourism data. For example, if we used counties as the object of study, we could try to resample them into grids of about the same size as the counties used when dealing with the NTL, reduce the overall size and enlarge the local size. In addition, the NTL data can be used as a social environment indicator to assess the safety of the nighttime environment, which can be introduced into the study of tourism behaviors and intentions. Finally, since NTL is updated quickly, monthly or quarterly studies can be attempted, such as the link between the TE and NTL during the peak and low seasons of a region, or the link between tourism activities and NTL during a certain event.

6. Conclusions

We analyzed and verified the close connection between NTL and TE through many research methods. The main conclusions are as follows. First, from the perspective of spatial agglomeration characteristics, both the concentration degree and the concentrated distribution area of NTL and TE are very similar, roughly showing a higher concentration in East China and South-Central China, indicating that the brightness of the lights in these areas is generally higher and TE is more developed. Compared with the agglomeration of the two, the development of NTL and TE can be divided into three types: NTL surpassing TE, coordinated development of NTL and TE, and TE surpassing NTL. In East China and North China the two are mostly shown as coordinated development, while in Southwest and South-Central China they are mostly shown as the type of TE surpassing NTL. Second, the analysis of the coupling characteristics of the spatial agglomeration of NTL and TE reveals that there is a high coupling relationship between the two, with higher spatial agglomeration coupling in North, South-Central and the coastal regions of East China, and relatively lower coupling in Southwest and Northwest China. Third, the spatial agglomeration correlation analysis of NTL and TE shows that the coupling degree of NTL and TE shows a positive spatial autocorrelation property, and there are obvious high–high coupling agglomeration and low–low coupling agglomeration between NTL and TE in spatial agglomeration distribution. Fourth, through geographical detector analysis, the driving factors affecting NTL and TEL can be concluded and the development of TE is proved to promote the development of NTL. In conclusion, it can be seen that the concentration and spatial agglomeration coupling degree of NTL and TE is relatively high in areas with better natural conditions, denser population and a higher level of socio-economic development, such as that in East China, North China and South-Central China. Meanwhile, the coupling degree of the two is relatively low in Southwest and Northwest China, with scarce resources, a smaller population and a slower socio-economic development level.