Analyzing Spatiotemporal Variation Modes and Industry-Driving Force Research Using VIIRS Nighttime Light in China

Urbanization is a complex process closely involving the economy, society, and population. While monitoring urban development and exploring the industry-driving force in a real-time and effective way are the prerequisites for optimizing industry structure, narrowing the urban development gap, and achieving sustainable development. Nighttime light is an effective tool to monitor urban development from a macro perspective. However, the systematic research of nighttime light spatiotemporal variation modes and the industry-driving force of urban nighttime light are still unknown. Considering these issues, this paper analyzes the spatiotemporal variation modes of the average light index (ALI) and investigates the industry-driving force of ALI in 100 major prefecture-level cities across China mainland based on National Polar-Orbiting Partnership Satellite Visible Infrared Imaging Radiometer Suite (NPP VIIRS). The conclusions are as following three aspects. First, ALI is observed a funnel pattern among four regions in spatial dimension, with low in center and high in the surrounding, and it shows 5 variation modes (“W,” “√,” “Exponent,” “Logarithm,” and “N”) in temporal dimension, of which the “√” mode accounts for the highest proportion (60%). Second, the industry structure is closely related to ALI. Besides, the factor analysis result illustrates that the secondary and tertiary industry are the driving industries of ALI. Third, the classification result based on the industry contribution rate indicates that cities driven by different industries show significant spatial distribution differences. The three major industry-driving cities are mainly distributed in central and western regions, the secondary and tertiary industry-driving cities are evenly distributed, and the tertiary industry-driving cities are mainly distributed in provincial capitals. From 2013 to 2018, the fluctuation of city distribution driven by different industries changes obviously. The number of tertiary industry-driving cities increases steadily and the three major industry-driving cities are distributed wider spatially. Additionally, the impacts of location and raw coal on ALI are discussed. In general, these findings are essential to further research urban development mode and can be considered as the reference to narrow urban development gap.


Introduction
The world is in the stage of rapid urbanization [1,2], which is not only closely related to economic growth, expansion of the urban built-up area, improvement of infrastructure [3][4][5][6][7][8] but also causes changes in land use structure and transition of industry structure [7,9,10]. In the global context, Frick et al. (2018) [11] studied the relationship between the urban concentration and economic

Data
In 2013, NOAA/NGDC (https://ngdc.noaa.gov/eog/viirs/download_dnb_composites.html) released NPP VIIRS nighttime light dataset (unit: nanoWatts/cm 2 /sr) which is produced by the Earth Observation Group (EOG) using nighttime data from the Infrared Imaging Radiometer Suite Day/Night Band (DNB). The pixel values of DMSP-OLS ranges from 0 to 63 rather than absolute values in Watts per cm 2 per steradian, while the values of the NPP VIIRS are absolute values that greatly overcome pixel saturation in urban centers. Besides, the resolution of DMSP-OLS (30 arc-s) is far lower than NPP VIIRS (15 arc-s). Additionally, the DMSP-OLS sensors lack low light imaging spectral bands that cannot capture weak light. Despite the shortcomings that existed in DMSP-OLS, it plays a significant and irreplaceable role in early years. However, considering the advantages of the spatial resolution of 15 arc-s, wider radiation detection range and in-flight calibration, NPP VIIRS is selected to investigate the spatiotemporal variation of nighttime light. While there are also shortcomings in Remote Sens. 2020, 12, 2785 5 of 24 NPP VIIRS that the pixels of monthly composites are missing in high latitude in some months and NOAA/NGDC only provides the annual composites which remove temporal lights and background values in 2015 and 2016. Limited by the missing of monthly and annual composites nighttime light data, we select complete NPP VIIRS monthly composites data. Finally, referencing to   [6], we use the annual composites nighttime light data in 2015 and 2016 as the mask to remove background noise and use the maximum threshold method to remove outliers in monthly composites nighttime light data.
Statistical data used in this paper is provided by the National Bureau of Statistics (http://www. stats.gov.cn/tjgz/wzlj/dftjwz), including the GDP of 100 prefecture-level cities which is composed of urban and rural area, and three major industries data namely the primary (mainly in agriculture), secondary (mainly in industry) and tertiary (mainly in service) industry from 2013 to 2018. The GDP data is used to evaluate the ability of NPP VIIRS to characterize economic development and three major industries data is used to study the relationship between the rationality of the industry structure and NPP VIIRS. In China, the rationality of the industry structure refers to an industry structure based on the rational use of resources, coordinated development of various industries, high-tech industries as the leading, and advantageous industries as the base [18,49].
Raw coal production data from 2013 to 2018 are provided by the National Coal Industry Network (http://www.coalchina.org.cn/). Remote Sens. 2020, 12, x FOR PEER REVIEW 5 of 25 some months and NOAA/NGDC only provides the annual composites which remove temporal lights and background values in 2015 and 2016. Limited by the missing of monthly and annual composites nighttime light data, we select complete NPP VIIRS monthly composites data. Finally, referencing to   [6], we use the annual composites nighttime light data in 2015 and 2016 as the mask to remove background noise and use the maximum threshold method to remove outliers in monthly composites nighttime light data. Statistical data used in this paper is provided by the National Bureau of Statistics (http://www.stats.gov.cn/tjgz/wzlj/dftjwz), including the GDP of 100 prefecture-level cities which is composed of urban and rural area, and three major industries data namely the primary (mainly in agriculture), secondary (mainly in industry) and tertiary (mainly in service) industry from 2013 to 2018. The GDP data is used to evaluate the ability of NPP VIIRS to characterize economic development and three major industries data is used to study the relationship between the rationality of the industry structure and NPP VIIRS. In China, the rationality of the industry structure refers to an industry structure based on the rational use of resources, coordinated development of various industries, high-tech industries as the leading, and advantageous industries as the base [18,49].
Raw coal production data from 2013 to 2018 are provided by the National Coal Industry Network (http://www.coalchina.org.cn/).
RV ij (1) where ALI i represents the average light radiation of the ith city that includes urban and surrounding areas, RV ij represents the radiation of the jth pixel in the ith city, and N i represents the total number of pixels in the ith city.

Economic Density Index (ED)
GDP, a reflection of the comprehensive economic level, is related to the urban economic development level and the urban administrative area [17], which means two regions with the same economic development level, the GDP is higher in the larger urban administrative. While the GDP only measures the general economic status and development level of a country or region [48]. In order to represent the average development level of the city, we select economic density (ED) which can characterize the efficiency of economic activities per unit area of a city as the economic indicator: where ED i represents the economic density index of the ith city, GDP i represents the GDP of the ith city, and N i represents the total number of pixels in the ith city.

Regression Analysis
Regression models help to find the linear or non-linear relationship between independent and dependent variable from data mixed with noise and predict the inversion variables quantitatively. Because of the superiority of quantifiable nature and reliability in regression models, there are a large number of studies that quantitatively analyze the relationship between independent variables and dependent variables based on regression models [22,53]. For example, Hu et al. (2019) [25] modeled the global electric power consumption using linear, logarithmic, exponential, and second-order polynomial regression models. Besides, Zhu et al. (2017) [54] applied linear, quadratic polynomial, and power function model to predict GDP in China. Among the regression models, the linear regression model is a commonly used correlation analysis model. For example,   [40] used the linear regression model to predict GDP and electric power consumption.   [53] established a linear regression model to study the correlation between electric power consumption and nighttime light. Because of the advantages of simplicity, intuitiveness, and high accuracy, we select the linear regression model to explore the correlation between independent and dependent variables in this paper.
where Y ij represents the standardized value of the jth industry in the ith city, X ij represents the original value of the jth industry in the ith city, and minX ij and maxX ij represent the minimum and maximum value of the jth industry in the ith city in each year, respectively. Using the standardized value, the industry structure score is calculated based on the entropy weight method: where IS i represents the industry structure score in the ith city, W j represents the weight of the jth industry, and K equals to 3, representing the three major industries.

Factor Analysis
Nighttime light is related to economy. For example, Levin et al. (2017) [44] illustrated that nighttime light positively correlated with GDP per capita. Besides, Han et al. (2018) [49] have pointed out that nighttime light intensity is related to industry structural changes that are composed of the primary, secondary and tertiary industry. Therefore, we conduct factor analysis to further investigate the factor contribution of the three major industries to ALI.
Random forest (RF) is an ensemble of prediction trees based on the regression and classification trees. It was integrated into a scikit-learn python library. RF can predict results by averaging values calculated by each tree.   [52] estimated poverty in Bangladesh using RF. Liang et al. (2020) [56] analyzed spatial distribution employing RFR.   [57] used RF to research the factor contribution of six potential factors to surface urban heat island intensity difference. Therefore, RF is utilized to conduct factor analysis because of the advantages of robust and high precision.

Characterization of ALI to Economic Development
The regression analysis results (R 2 ) of ALI and ED are greater than 0.8 (Figure 2), which represent that ALI can explain economic changes at a level higher than 80%, illustrating that ALI has the potentiality to characterize the economy. In Figure 2, a good correlation between ALI and ED is observed among most cities that belong to medium cities, large cities, and megacities. However, a few super-cities and megacities are far from the regression line. Among these cities, the economic development of Shenzhen reaches a mature stage, the spatial and economic urbanization is high, and the speed of economic development is faster than ALI. On the contrary, the economic development of Soochow, Dongguan, and Shantou is in the stage of rapid development and the ability of ALI

Spatial Distribution Pattern of ALI
To investigate the spatial distribution of nighttime light, ALI of the 100 cities calculated by corrected NPP VIIRS is shown in Figure 3 and the mean value of ALI in four regions are shown in Table 1. In Figure 3, ALI is observed a funnel pattern in four regions, with low in center region and high in the surrounding. According to Table 1, we can know that the growth rates of ALI from 2013 to 2018 in the eastern, western, central, and northeastern regions are 0.038, 0.047, 0.078, and 0.31, respectively. The level of economic development in the central and western regions is relatively low, while the in-depth implementation in the great western development strategy and the Rise of Central China has accelerated urbanization [16,58] which results in the high growth rate of ALI. However, compared with the eastern region, since the initial endowment of production factors is relatively low [16], socioeconomic activities are weak and the net increase of ALI is low. As an old industrial base, ALI in the northeastern region is relatively low as the population is massively outflowing and urban development lacks vitality [43]. Because of the high economic development level and a strong economic foundation [19], the net increase of ALI in the eastern region is high. On the whole, ALI shows a noteworthy difference in space because of economic and natural conditions.

Spatial Distribution Pattern of ALI
To investigate the spatial distribution of nighttime light, ALI of the 100 cities calculated by corrected NPP VIIRS is shown in Figure 3 and the mean value of ALI in four regions are shown in Table 1. In Figure 3, ALI is observed a funnel pattern in four regions, with low in center region and high in the surrounding. According to Table 1, we can know that the growth rates of ALI from 2013 to 2018 in the eastern, western, central, and northeastern regions are 0.038, 0.047, 0.078, and 0.31, respectively. The level of economic development in the central and western regions is relatively low, while the in-depth implementation in the great western development strategy and the Rise of Central China has accelerated urbanization [16,58] which results in the high growth rate of ALI. However, compared with the eastern region, since the initial endowment of production factors is relatively low [16], socioeconomic activities are weak and the net increase of ALI is low. As an old industrial base, ALI in the northeastern region is relatively low as the population is massively outflowing and urban development lacks vitality [43]. Because of the high economic development level and a strong economic foundation [19], the net increase of ALI in the eastern region is high. On the whole, ALI shows a noteworthy difference in space because of economic and natural conditions.  In order to clearly investigate the distribution difference among the four regions, the distribution characteristics of ALI from 2013 to 2018 are represented by box-plots ( Figure 4). According to Table  1 and Figure 4, the ALI shows significant difference characteristics in four regions. Take 2018 as an example, in the northeastern region, ALI in Shenyang is the highest (2.66 nanoWatts/cm 2 /sr), and ALI in Suihua is the lowest (0.18 nanoWatts/cm 2 /sr); in the eastern region, ALI in Shenzhen is the highest (18.19 nanoWatts/cm 2 /sr), and ALI in Maoming is the lowest (0.70 nanoWatts/cm 2 /sr); in the central region, ALI in Zhengzhou is the highest (4.6 nanoWatts/cm 2 /sr) and ALI in Yongzhou is the lowest (0.16 nanoWatts/cm 2 /sr); in the western region, ALI in Chengdu is the highest (3.71 nanoWatts/cm 2 /sr), and ALI in Hulunbuir is the lowest (0.03 nanoWatts/cm 2 /sr). Among the four regions, ALI in the eastern region is the highest overall and the average ALI are 3.39-, 3.32-, 3.37-, 3.40-, 3.79-, and 4.08 nanoWatts/cm 2 /sr from 2013 to 2018, respectively. Although the 38 cities are all in the eastern region, while limited by location, resource, and transportation conditions, the ALI of cities outside the Yangtze River Delta, the Pearl River Delta, and the Bohai Rim Economic Circle are  In order to clearly investigate the distribution difference among the four regions, the distribution characteristics of ALI from 2013 to 2018 are represented by box-plots ( Figure 4). According to Table 1 and Figure 4, the ALI shows significant difference characteristics in four regions. Take 2018 as an example, in the northeastern region, ALI in Shenyang is the highest (2.66 nanoWatts/cm 2 /sr), and ALI in Suihua is the lowest (0.18 nanoWatts/cm 2 /sr); in the eastern region, ALI in Shenzhen is the highest (18.19 nanoWatts/cm 2 /sr), and ALI in Maoming is the lowest (0.70 nanoWatts/cm 2 /sr); in the central region, ALI in Zhengzhou is the highest (4.6 nanoWatts/cm 2 /sr) and ALI in Yongzhou is the lowest (0.16 nanoWatts/cm 2 /sr); in the western region, ALI in Chengdu is the highest (3.71 nanoWatts/cm 2 /sr), and ALI in Hulunbuir is the lowest (0.03 nanoWatts/cm 2 /sr). Among the four regions, ALI in the eastern region is the highest overall and the average ALI are 3.39-, 3.32-, 3.37-, 3.40-, 3.79-, and 4.08 nanoWatts/cm 2 /sr from 2013 to 2018, respectively. Although the 38 cities are all in the eastern region, while limited by location, resource, and transportation conditions, the ALI of cities outside the Yangtze River Delta, the Pearl River Delta, and the Bohai Rim Economic Circle are much lower than those within the economic circle, resulting in a great difference of ALI in the eastern region. Since half of the selected cities in the western region are provincial capital cities whose main functions are to enhance economic radiation capacity, rationally allocate resources, and improve regional industry structure and economic development quality, the ALI of provincial capital cities is much higher than that of non-provincial cities, leading to large ALI difference in the western region. Besides, ALI in the northeastern region is higher than that in the central region and the difference of ALI in the northeastern region is relatively small with the benefit of a strong industrial foundation. While compared with the other three regions, the cities in the central region have a relatively dense population (429 population/km 2 in the central region) which is much higher than that in the northeastern (215 population/km 2 ) and western region (202 population/km 2 ), and lower than that in the eastern region (869 population/km 2 ). Besides, cities in the central region have relatively developed agriculture and stable economic development, with the lowest ALI and smallest difference.
Remote Sens. 2020, 12, x FOR PEER REVIEW 10 of 25 much lower than those within the economic circle, resulting in a great difference of ALI in the eastern region. Since half of the selected cities in the western region are provincial capital cities whose main functions are to enhance economic radiation capacity, rationally allocate resources, and improve regional industry structure and economic development quality, the ALI of provincial capital cities is much higher than that of non-provincial cities, leading to large ALI difference in the western region. Besides, ALI in the northeastern region is higher than that in the central region and the difference of ALI in the northeastern region is relatively small with the benefit of a strong industrial foundation. While compared with the other three regions, the cities in the central region have a relatively dense population (429 population/ km 2 in the central region) which is much higher than that in the northeastern (215 population/km 2 ) and western region (202 population/km 2 ), and lower than that in the eastern region (869 population/km 2 ). Besides, cities in the central region have relatively developed agriculture and stable economic development, with the lowest ALI and smallest difference.

Temporal Variation Mode of ALI
ALI distribution difference and the raw coal production change from 2013 to 2018 are shown in Figure 5. The median and mean value of ALI are in the upward trend from 2013 to 2018, and the upward trend is slight in 2014 and 2016 and the upward extend is obvious in 2017 and 2018. The trend of raw coal production is consistent with the trend of ALI, which is in the downward during 2015 and 2016 in most cities. Studies illustrated that nighttime light significantly positively correlates with electricity consumption [22,53,59], as well as about 70 percent of our country's electricity is produced from coal through thermal power generation [60]. Thus, we inferred that the raw coal production indirectly affects nighttime light through electricity produced by raw coal. Besides, ALI in 100 selected Chinese cities shows an increasing disparity from 2013 to 2018. The eastern region is the center of the economy that focuses on the rapid development of the high-tech industry [61]. Besides, under the in-depth implementation of the east-west region industry transfer, the rise of the central region and other national strategies, the central and western regions accelerate the speed of economic development [62]. However, compared with the eastern region, the development level in the western region is still low as the western region is limited by natural and traffic conditions [63], leading to an increasing disparity in ALI from 2013 to 2018.

Temporal Variation Mode of ALI
ALI distribution difference and the raw coal production change from 2013 to 2018 are shown in Figure 5. The median and mean value of ALI are in the upward trend from 2013 to 2018, and the upward trend is slight in 2014 and 2016 and the upward extend is obvious in 2017 and 2018. The trend of raw coal production is consistent with the trend of ALI, which is in the downward during 2015 and 2016 in most cities. Studies illustrated that nighttime light significantly positively correlates with electricity consumption [22,53,59], as well as about 70 percent of our country's electricity is produced from coal through thermal power generation [60]. Thus, we inferred that the raw coal production indirectly affects nighttime light through electricity produced by raw coal. Besides, ALI in 100 selected Chinese cities shows an increasing disparity from 2013 to 2018. The eastern region is the center of the economy that focuses on the rapid development of the high-tech industry [61]. Besides, under the in-depth implementation of the east-west region industry transfer, the rise of the central region and other national strategies, the central and western regions accelerate the speed of economic development [62]. However, compared with the eastern region, the development level in the western region is still low as the western region is limited by natural and traffic conditions [63], leading to an increasing disparity in ALI from 2013 to 2018.
By defining different ALI fluctuation states, the ALI of the 100 cities from 2013 to 2018 shows five variation modes ( Figure 6) namely "W" mode (ALI experiences decrease twice), "√" mode (ALI experiences decrease once), "Logarithm" mode (ALI increases quickly in the first period and slows down in the second period), "N" mode (ALI changes greatly), and "Exponent" mode (ALI increases steadily). In Figure 6, the "√" mode accounts for 60% among the five modes, which is consistent with the Chinese economic development situation [16], while Liupanshui, Chenzhou, and Fushun, accounting for only 3% among 100 cities, are coal mining cities where coal resource depletion leads to increased urban vulnerability [64], changes in land use structure [12] and lacking vitality in urban development, resulting in the decreased light radiation intensity. According to Figures 5 and 6, 60% of the selected cities experience a decline in ALI with the sharp decline of raw coal production in 2015 and 2016, especially coal mining cities like Handan, Tangshan, and Fushun. The phenomenon of ALI rising and falling together with raw coal production indicates that the raw coal production not only seriously affect the ALI variation mode in coal mining cities, but also affect the ALI variation mode in other cities.
Counting the number of 5 ALI variation modes in four regions (Table 2), the "√" mode concentrates in the central and eastern regions, the "Exponent" mode concentrates in the eastern and central regions, the "Logarithm" mode evenly concentrates in four regions, and the "N" and "W" mode mainly concentrates in the western region. On the whole, the "W" mode and "N" mode whose fluctuation is large mainly concentrates in the western region, while the mode of "√," "Exponent," and "Logarithm" whose fluctuation is small mainly concentrates in the eastern and central regions. This phenomenon is associated with the economic development situation and initial endowment of production factors in different regions.  √   1  27  10  22  60  Exponent  1  7  1  8  17  Logarithm  1  1  1  -3  N  1  1  3  1  6  W  2  2  9  1  14  Total  6  38  24 32 100 By defining different ALI fluctuation states, the ALI of the 100 cities from 2013 to 2018 shows five variation modes ( Figure 6) namely "W" mode (ALI experiences decrease twice), " √ " mode (ALI experiences decrease once), "Logarithm" mode (ALI increases quickly in the first period and slows down in the second period), "N" mode (ALI changes greatly), and "Exponent" mode (ALI increases steadily). In Figure 6, the " √ " mode accounts for 60% among the five modes, which is consistent with the Chinese economic development situation [16], while Liupanshui, Chenzhou, and Fushun, accounting for only 3% among 100 cities, are coal mining cities where coal resource depletion leads to increased urban vulnerability [64], changes in land use structure [12] and lacking vitality in urban development, resulting in the decreased light radiation intensity.
According to Figures 5 and 6, 60% of the selected cities experience a decline in ALI with the sharp decline of raw coal production in 2015 and 2016, especially coal mining cities like Handan, Tangshan, and Fushun. The phenomenon of ALI rising and falling together with raw coal production indicates that the raw coal production not only seriously affect the ALI variation mode in coal mining cities, but also affect the ALI variation mode in other cities.
Counting the number of 5 ALI variation modes in four regions (Table 2), the " √ " mode concentrates in the central and eastern regions, the "Exponent" mode concentrates in the eastern and central regions, the "Logarithm" mode evenly concentrates in four regions, and the "N" and "W" mode mainly concentrates in the western region. On the whole, the "W" mode and "N" mode whose fluctuation is large mainly concentrates in the western region, while the mode of " √ ," "Exponent," and "Logarithm" whose fluctuation is small mainly concentrates in the eastern and central regions. This phenomenon is associated with the economic development situation and initial endowment of production factors in different regions.  1  27  10  22  60  Exponent  1  7  1  8  17  Logarithm  1  1  1  -3  N  1  1  3  1  6  W  2  2  9  1  14  Total  6  38  24  32  100 Remote Sens. 2020, 12, 2785 12 of 24 Remote Sens. 2020, 12, x FOR PEER REVIEW 12 of 25

Industry-Driving Force
The nighttime light radiation comes from the three major industries. It is conducive to in-depth research if we explore the effect of the industry structure layout on ALI and investigate the industrydriving force of ALI.

The Quality of Industry Structure
Using the three major industries data to obtain IS based on entropy weight method and conducting regression analysis between IS and ALI, the goodness of fit ( 2 ) is greater than 0.8 (Figure

Industry-Driving Force
The nighttime light radiation comes from the three major industries. It is conducive to in-depth research if we explore the effect of the industry structure layout on ALI and investigate the industry-driving force of ALI.

The Quality of Industry Structure
Using the three major industries data to obtain IS based on entropy weight method and conducting regression analysis between IS and ALI, the goodness of fit (R 2 ) is greater than 0.8 (Figure 7), and cities Remote Sens. 2020, 12, 2785 13 of 24 far from the fit line are basically consistent with those in Figure 2. According to the regression results between IS and ALI, IS affects ALI significantly. Besides, research indicated that GDP is positively correlated with NPP VIIRS [56], and the essence of the GDP is the three major industries in China. Additionally, Han et al. (2018) have demonstrated that the industry structure change in China is related to nighttime light based on DMSP-OLS. Therefore, we inferred that the industry structure layout indirectly affects ALI and it may be the fundamental cause of ALI variation.
Remote Sens. 2020, 12, x FOR PEER REVIEW 13 of 25 7), and cities far from the fit line are basically consistent with those in Figure 2. According to the regression results between IS and ALI, IS affects ALI significantly. Besides, research indicated that GDP is positively correlated with NPP VIIRS [56], and the essence of the GDP is the three major industries in China. Additionally, Han et al. (2018) have demonstrated that the industry structure change in China is related to nighttime light based on DMSP-OLS. Therefore, we inferred that the industry structure layout indirectly affects ALI and it may be the fundamental cause of ALI variation.

Industry-Driving Force
Using the three major industries data and GDP from 2013 to 2018 to calculate the contribution of the three major industries of 100 cities, the position of dots presents the industry structure layout, and the color of dots presents ALI ( Figure 8). In Figure 8, since super-cities and megacities like Beijing, Shanghai, and Shenzhen with sound economic development are mainly dominated by the tertiary industry, ALI is high. However, since large and medium cities like Hulunbuir and Qiqihar with relatively depressed economic development are mainly dominated by the secondary or primary industry, ALI is relatively low. Besides, compared with cities in the eastern region, cities in the western region like the Ordos dominated by the secondary industry and Lhasa dominated by the tertiary industry have an expansive district but a sparse population with relatively weaker socioeconomic activities, leading to relatively lower ALI. Additionally, ALI decreases with the rise of the proportion of the primary industry, while it increases with the rise of the proportion of the secondary and tertiary industry, indicating that the secondary and tertiary industry are the industrydriving force of ALI. From 2013 to 2018, the primary industry is gradually transformed into the secondary and tertiary industry in most cities, and the industry structure layout of some cities is mainly in the tertiary industry. Therefore, ALI experiences a significant rise. (a-f) represent the regression results of IS and ALI from 2013 to 2018, respectively (1 represents super-cities, 2 represents megacities, 3 represents large cities, and 4 represents medium cities).

Industry-Driving Force
Using the three major industries data and GDP from 2013 to 2018 to calculate the contribution of the three major industries of 100 cities, the position of dots presents the industry structure layout, and the color of dots presents ALI ( Figure 8). In Figure 8, since super-cities and megacities like Beijing, Shanghai, and Shenzhen with sound economic development are mainly dominated by the tertiary industry, ALI is high. However, since large and medium cities like Hulunbuir and Qiqihar with relatively depressed economic development are mainly dominated by the secondary or primary industry, ALI is relatively low. Besides, compared with cities in the eastern region, cities in the western region like the Ordos dominated by the secondary industry and Lhasa dominated by the tertiary industry have an expansive district but a sparse population with relatively weaker socioeconomic activities, leading to relatively lower ALI. Additionally, ALI decreases with the rise of the proportion of the primary industry, while it increases with the rise of the proportion of the secondary and tertiary industry, indicating that the secondary and tertiary industry are the industry-driving force of ALI. From 2013 to 2018, the primary industry is gradually transformed into the secondary and tertiary industry in most cities, and the industry structure layout of some cities is mainly in the tertiary industry. Therefore, ALI experiences a significant rise. (a-f) represent the industry structure layout from 2013 to 2018, respectively. The location of dots presents industry structure layout in 100 cities, and the color of dots presents ALI (red means the high ALI, blue means the low ALI).
By defining different morphological characteristics of the temporal moving trajectories of 100 cities from 2013 to 2018 on the ternary diagrams, we find six types of industry structure layout changing trajectories (Figure 9). Represented by Lhasa, Bozhou, Xuzhou, Lu'an, Fushun, and Zunyi, respectively, the meaning of the six types of industry structure layout changing characteristics of cities are as follows: (a) Represents that the proportion of secondary industry increases, while the proportion of tertiary industry decreases; (b) represents that the proportion of tertiary industry (a-f) represent the industry structure layout from 2013 to 2018, respectively. The location of dots presents industry structure layout in 100 cities, and the color of dots presents ALI (red means the high ALI, blue means the low ALI).
By defining different morphological characteristics of the temporal moving trajectories of 100 cities from 2013 to 2018 on the ternary diagrams, we find six types of industry structure layout changing trajectories ( Figure 9). Represented by Lhasa, Bozhou, Xuzhou, Lu'an, Fushun, and Zunyi, respectively, the meaning of the six types of industry structure layout changing characteristics of cities are as follows: (a) Represents that the proportion of secondary industry increases, while the proportion of tertiary industry decreases; (b) represents that the proportion of tertiary industry increases, while the proportion of primary industry decreases; (c) represents that the proportion of tertiary industry increases, while the proportion of secondary industry decreases; (d) represents that the proportion of tertiary industry increases, while the proportion of primary and secondary industry decreases; (e) represents that the proportion of the three major industries reciprocates; (f) represents that the proportion of the tertiary industry changes slightly. There are spatial differences among the four regions in the industry structure layout changing types ( Figure 10). The eastern region concentrates on the development of the high-tech industry in recent years [18], with a significant increase in the proportion of the tertiary industry and the industry structure layout changing types mainly in the type of c and d. Besides, under the radiation of the eastern region, the tertiary industry in the central region obtains relatively speedy development and the main type of industry structure layout change is d. All the six types exist in the western region, of which c is the main type of industry structure layout changing types. Additionally, the industry structure layout changing types in the northeastern region are c, d, and e. Generally, in recent years, China focuses on developing high-tech industries to optimize industry structure [18], resulting in b, c, and d are the main types in 100 cities, accounting for 81%. Remote Sens. 2020, 12, x FOR PEER REVIEW 16 of 25 Figure 10. Distribution of the type of industry structure layout change in four regions.

Factor Analysis
The importance of the primary industry to ALI is significantly lower than that of the secondary and tertiary industry ( Figure 11). The factor analysis result suggests that about 10% of nighttime light comes from the primary industry, and the other 90% nighttime light comes from the secondary and tertiary industry. Thus, the ALI variation is more controlled by the second and tertiary industry. Transportation, infrastructure, business, industry, and manufacturing that produce light at night all belong to the secondary and tertiary industry. Thus, the importance of the secondary and tertiary industry to ALI is much higher than that of the primary industry, which illustrates that the secondary and tertiary industry are the driving industries of ALI. In Figure 7, the regression results illustrate that the higher the IS is, the higher the ALI is, and the cities with relatively high ALI are super-cities and megacities. According to Figure 8, the cities where the proportion of the secondary and tertiary industry is high, ALI is relatively high. Combined with Figure 7, Figure 8, and Figure 11, we can infer that the higher the IS and the proportion of the secondary and tertiary industry are, the higher ALI is. But this relationship between industry structure and ALI will show deviation in several western cities like Lhasa and Ordos because of poor initial endowment of production factors.
Under the fast economic development and the policy of transformation and upgrading of the industry structure in China [49], the primary industry will be transferred into the secondary and tertiary industry, which will consistently strengthen the nighttime light radiation intensity. While when nighttime light reaches saturation referring nighttime light intensity will not be enhanced with the economic development in the urban and rural areas, the ALI will keep stable. Therefore, the "Exponent" ALI variation mode is expected in a short period, and the "Logarithm" ALI variation mode is expected in a long period under ideal circumstances. On the whole, the contribution of the secondary and tertiary industry to ALI is significantly greater than the primary industry.

City Classification Based on Urban Industry-Driving Force
Based on the proportion of the three major industries from 2013 to 2018, the clustering analysis of 100 cities is performed ( Figure 12). The three major industry-driving cities are mainly distributed in the central and western regions with relatively developed agriculture. Studies indicated that the central region possesses the highest percentage of the agricultural area to the total area, while compared with other three regions, the western region has the smallest agricultural area loss [7]. Besides, the secondary and tertiary industry-driving cities are evenly distributed in China as the secondary and tertiary industry are the leading industries in Chinese economic development [10]. Additionally, as the economic center of the province, some provincial capitals are the tertiary industry-driving cities that focus on education, scientific research, service industry, and high-tech industry.

Factor Analysis
The importance of the primary industry to ALI is significantly lower than that of the secondary and tertiary industry ( Figure 11). The factor analysis result suggests that about 10% of nighttime light comes from the primary industry, and the other 90% nighttime light comes from the secondary and tertiary industry. Thus, the ALI variation is more controlled by the second and tertiary industry. Transportation, infrastructure, business, industry, and manufacturing that produce light at night all belong to the secondary and tertiary industry. Thus, the importance of the secondary and tertiary industry to ALI is much higher than that of the primary industry, which illustrates that the secondary and tertiary industry are the driving industries of ALI. In Figure 7, the regression results illustrate that the higher the IS is, the higher the ALI is, and the cities with relatively high ALI are super-cities and megacities. According to Figure 8, the cities where the proportion of the secondary and tertiary industry is high, ALI is relatively high. Combined with Figure 7, Figure 8, and Figure 11, we can infer that the higher the IS and the proportion of the secondary and tertiary industry are, the higher ALI is. But this relationship between industry structure and ALI will show deviation in several western cities like Lhasa and Ordos because of poor initial endowment of production factors.
Remote Sens. 2020, 12, x FOR PEER REVIEW 17 of 25 Figure 11. Feature importance of the industry (PI represents the primary industry, SI represents the secondary industry, TI represents the tertiary industry). Under the fast economic development and the policy of transformation and upgrading of the industry structure in China [49], the primary industry will be transferred into the secondary and tertiary industry, which will consistently strengthen the nighttime light radiation intensity. While when nighttime light reaches saturation referring nighttime light intensity will not be enhanced with the economic development in the urban and rural areas, the ALI will keep stable. Therefore, the "Exponent" ALI variation mode is expected in a short period, and the "Logarithm" ALI variation mode is expected in a long period under ideal circumstances. On the whole, the contribution of the secondary and tertiary industry to ALI is significantly greater than the primary industry.

City Classification Based on Urban Industry-Driving Force
Based on the proportion of the three major industries from 2013 to 2018, the clustering analysis of 100 cities is performed ( Figure 12). The three major industry-driving cities are mainly distributed in the central and western regions with relatively developed agriculture. Studies indicated that the central region possesses the highest percentage of the agricultural area to the total area, while compared with other three regions, the western region has the smallest agricultural area loss [7]. Besides, the secondary and tertiary industry-driving cities are evenly distributed in China as the secondary and tertiary industry are the leading industries in Chinese economic development [10]. Additionally, as the economic center of the province, some provincial capitals are the tertiary industry-driving cities that focus on education, scientific research, service industry, and high-tech industry.
Remote Sens. 2020, 12, x FOR PEER REVIEW 17 of 25 Figure 11. Feature importance of the industry (PI represents the primary industry, SI represents the secondary industry, TI represents the tertiary industry). Counting the number of cities driven by different industries from 2013 to 2018, the result is shown in Figure 13. In 2013, only Beijing is the tertiary industry-driving city and the three major industry-driving cities are mainly distributed in the central region as shown in Figure 12. According to Figures 12 and 13, with the implementation of industry transformation and industry optimization Counting the number of cities driven by different industries from 2013 to 2018, the result is shown in Figure 13. In 2013, only Beijing is the tertiary industry-driving city and the three major industry-driving cities are mainly distributed in the central region as shown in Figure 12. According to Figures 12 and 13, with the implementation of industry transformation and industry optimization policies in recent years, the number of tertiary industry-driving cities is increasing, all of which are provincial capital cities. Besides, agriculture is the foundation of a country, and the Chinese government has formulated agricultural development policies and the red line for arable land to promote modern agricultural development [65]. Therefore, the proportion of the primary industry in some cities is significantly increasing, resulting in that the number of cities driven by the three major industries increases and become more widely distributed.
Remote Sens. 2020, 12, x FOR PEER REVIEW 18 of 25 policies in recent years, the number of tertiary industry-driving cities is increasing, all of which are provincial capital cities. Besides, agriculture is the foundation of a country, and the Chinese government has formulated agricultural development policies and the red line for arable land to promote modern agricultural development [65]. Therefore, the proportion of the primary industry in some cities is significantly increasing, resulting in that the number of cities driven by the three major industries increases and become more widely distributed.   [19] proposed that ALI represents the average level of economic development in a city, and cities with the same industry structure layout and development stage have little difference in ALI. From these studies, we can know that the indicators related to the specific socioeconomic indicators were predicated by TNL, while the indicators related to the whole socioeconomic development like poverty and inequality were predicated by ALI. In this paper, ED represents the average development level that is correlated with every inch of land in cities. Thus, ALI is selected to study the economic spatiotemporal distribution and analyze the relationship between ALI and the three major industries. In order to further demonstrate the reliability of ALI to characterize ED, we conducted the regression analysis between TNL and ED ( Figure 14). According to Figures 2 and 14, we can know that ALI is indeed superior to TNL in characterizing ED.   [19] proposed that ALI represents the average level of economic development in a city, and cities with the same industry structure layout and development stage have little difference in ALI. From these studies, we can know that the indicators related to the specific socioeconomic indicators were predicated by TNL, while the indicators related to the whole socioeconomic development like poverty and inequality were predicated by ALI. In this paper, ED represents the average development level that is correlated with every inch of land in cities. Thus, ALI is selected to study the economic spatiotemporal distribution and analyze the relationship between ALI and the three major industries. In order to further demonstrate the reliability of ALI to characterize ED, we conducted the regression analysis between TNL and ED ( Figure 14). According to Figures 2 and 14, we can know that ALI is indeed superior to TNL in characterizing ED.

Potential Influencing Factors of ALI Changes
NPP VIIRS is not only affected by festivals, seasons, infrastructure, GDP [43][44][45]66] but also affected by location and electricity power consumption [21,23]. We will discuss the effect of location, and raw coal which is the main resource of electricity power consumption on ALI in this paper. In Figures 5 and 6, the result shows that raw coal production indirectly affects NPP VIIRS. In Figure 3, the cities in the Yangtze River Delta, the Pearl River Delta, and the Bohai Rim Economic Circle that to some extent represent the highest level of Chinese economic development have the highest ALI values. The selected cities include 24 cities in the economic circle and 17 coal mining cities. To better understand the influence of raw coal and location on ALI, we analyze the ALI in different cities under the two influencing factors.
The impacts of location and raw coal on ALI are shown in Tables 3 and 4. In Table 3, the p-value (<0.05) illustrates that ALI shows the great difference between cities within and outside the economic circle, which demonstrates that the economic circles produce a significant influence on economic development in China. In Table 4, the p-value (>0.05) suggests that there is no significant difference between coal mining cities and other cities, which demonstrate that raw coal production affects the ALI of coal mining cities and other cities. These discussion further demonstrate the aforementioned result (Figures 3 and 6).

Potential Influencing Factors of ALI Changes
NPP VIIRS is not only affected by festivals, seasons, infrastructure, GDP [43][44][45]66] but also affected by location and electricity power consumption [21,23]. We will discuss the effect of location, and raw coal which is the main resource of electricity power consumption on ALI in this paper. In Figures 5 and 6, the result shows that raw coal production indirectly affects NPP VIIRS. In Figure 3, the cities in the Yangtze River Delta, the Pearl River Delta, and the Bohai Rim Economic Circle that to some extent represent the highest level of Chinese economic development have the highest ALI values. The selected cities include 24 cities in the economic circle and 17 coal mining cities. To better understand the influence of raw coal and location on ALI, we analyze the ALI in different cities under the two influencing factors.
The impacts of location and raw coal on ALI are shown in Tables 3 and 4. In Table 3, the p-value (<0.05) illustrates that ALI shows the great difference between cities within and outside the economic circle, which demonstrates that the economic circles produce a significant influence on economic development in China. In Table 4, the p-value (>0.05) suggests that there is no significant difference between coal mining cities and other cities, which demonstrate that raw coal production affects the ALI of coal mining cities and other cities. These discussion further demonstrate the aforementioned result (Figures 3 and 6).
We have demonstrated that the secondary and tertiary industry are the driving industries of ALI. Besides, we mastered the contribution of the three major industries to ALI. However, there is still a lot of work to be done. In this paper, because of the different data statistic standards of the 100 cities, it is difficult to find consistent research indicators. Thus, we only discuss the impact of different industry structures on ALI from the perspective of the three major industries. In fact, the three major industries have many different structural components. For example, the tertiary industry includes service, education and scientific research, tourism, and others. Therefore, we need to further study the specific driving forces of ALI in our future work from a more micro perspective.

Conclusions
This paper analyzes ALI variation mode and industry-driving force of ALI in China. It is the first attempt to synthetically analyze the influence of the three major industries to nighttime light. This study proves that nighttime light is closely related to industry structure, and helps us further understand the source of nighttime light from the industry structure perspective. The results indicate that there are obvious differences in ALI among the 100 cities, and the importance of the primary, secondary, and tertiary to ALI presents great disparities.
The detailed conclusions are summarized as three following aspects. First, ALI is observed a concave funnel pattern in four regions, with low in center and high in the surrounding. Meanwhile, ALI shows five temporal variation modes ("W," " √ ," "Exponent," "Logarithm," and "N"), of which the " √ " mode accounts for the highest proportion (60%). Second, the regression results of IS and ALI (R 2 > 0.8) suggest that the industry structure is significantly correlated with ALI. The factor analysis result illustrates that the secondary and tertiary industry are the driving industries of ALI. Third, the classification result cities based on the industry contribution rate indicates that cities driven by different industries show significant spatial distribution differences. The three major industry-driving cities are mainly distributed in the central and western regions, the secondary and tertiary industry-driving cities are evenly distributed in four regions, and the tertiary industry-driving cities are mainly distributed in provincial capitals. From 2013 to 2018, the fluctuation of cities driven by different industries changes obviously. The number of the tertiary industry-driving cities increases steadily and the three major industry-driving cities are distributed widely spatially.
In the section of the discussion, the results of the regression analysis further demonstrate that ALI is the better indicator to characterize economic development. Additionally, the influence of raw coal and the location on ALI are conducted. The discussion illustrates that the location of cities has a significant effect on ALI spatial distribution, while the raw coal does not. preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Handan Remote Sens. 2020, 12, x FOR PEER REVIEW preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Beijing Baoding Harbin Handan Jining Nanjing Tangshan Hefei Hengyang Xinyang acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Beijing Baoding Harbin Handan Jining Nanjing Tangshan Hefei Hengyang Remote Sens. 2020, 12, x FOR PEER REVIEW preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Handan Huanggang preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Beijing Harbin Handan Jining Nanjing Tangshan Hefei Hengyang Remote Sens. 2020, 12, x FOR PEER REVIEW preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A   Table A1. The selected cities across China mainland.
Xingtai Ⅱ Y 89 Guiyang Ⅲ N Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Foshan Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A  Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Acknowledgments: The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; supervision, X.S.; funding acquisition, Y.C. All authors have read and agreed to the published version of the manuscript.
Funding: This work was supported by National Key R&D Program on monitoring, early warning and prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laboratory of Water Resources Security.

Acknowledgments:
The authors would like to thank the editor and the anonymous reviews for their constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A preparation, X.S.; writing-review and editing, Y.C., K.L., and X.S.; visualization, X.S.; sup acquisition, Y.C. All authors have read and agreed to the published version of the manus Funding: This work was supported by National Key R&D Program on monitoring prevention of major natural disasters (2017YFC1502406), and in part by Beijing Laborato Security.

Acknowledgments:
The authors would like to thank the editor and the anonymo constructive and significant comments on the earlier version of this paper.

Conflicts of Interest:
The authors declare no conflict of interest.
Appendix A Note: CI represent city ID, CN represent the name of the city, R represent the region (green, orange, red, and blue represent cities in the Central, Northeast, East, and West regions, respectively.), CT represents the city size (I, II, III, and IV represent the super-cities, megacities, large cities, and medium cities, respectively), CI represent coal mining cities (Y represent Yes, N represent No).