Analysis of Ecological Environment Quality and Its Driving Factors in the Beijing-Tianjin-Hebei Region of China

Abstract

This study aims to explore the evolution of the pattern of ecological environment quality and its driving factors in the Beijing-Tianjin-Hebei (BTH) region, providing a basis for developing regional ecological protection policies. Based on remote sensing image data, the study developed a remote-sensing eco-environmental index (REI) from two dimensions, namely ecological quality and environmental quality, and evaluated the spatiotemporal changes of the eco-environment quality in the BTH region from 2000 to 2020. The main factors affecting the eco-environment quality and the changing trend of the eco-environment quality were subsequently analyzed using the geographic detector model and the GM1.1 model. The results show that the eco-environmental quality of the BTH region shows a fluctuating downward trend and distinct regional differences during the study period. The Yanshan Mountain in the north and the Taihang Mountain in the west have high ecological quality, while the Zhangjiakou area in the northwestern part of Hebei Province and the urban center in the southeastern part of Hebei Province suffer low ecological quality. Natural factors played a dominant role in influencing eco-environmental factors, but the proportion of economic and social factors increased over time. It is predicted that the number of counties in the region with poor eco-environmental factors will increase significantly. Therefore, the comprehensive assessment of regional ecological quality can be effectively achieved based on REI, thus providing a reference for the green and high-quality development of the regional social economy.

1. Introduction

Ecological environment quality refers to the suitability of the overall or partial combination of ecological environment factors in a specific time and space range for human survival and sustainable socio-economic development [1]. In 1969, the United States promulgated the National Environmental Policy Act, which first included the ecological environment in the system [2]. In 1974, the Canadian Department of Environmental Protection proposed a total environmental quality index covering elements, such as the atmosphere, water, and soil [3]. In the 1990s, the Organization for Economic Cooperation and Development (OECD) proposed the Pressure-State-Response (PSR) model for monitoring and evaluating regional ecological environment quality [4], which has been widely used by scholars in various countries. Improved models, such as the DSR model [5], PSIR model [6], and DPSIR model [7], have been developed based on the PSR model. Canadian scholars Rees WF and Wackernagel M proposed and developed the ecological footprint model [8,9], widely used in ecological health, ecological security, and ecological environment quality [10]. With the development and maturity of space science and geographic information systems, remote sensing (RS) and geographic information technology (GIS) have been increasingly used in ecological environment quality research. Currently, methods, such as the analytic hierarchy process, expert scoring, entropy weighting method, and fuzzy mathematics analysis method, are used to determine the weight of ecological environment quality evaluation indicators. In 2013, Xu Hanqiu proposed the Remote Sensing Ecological Index (RSEI), which includes greenness, wetness, warmth, and dryness, and is wholly based on remote sensing information [11]. The indicator weights are determined using principal component analysis, avoiding artificial interference factors. Subsequently, many scholars have conducted research on different unit scales, such as basins, mountainous areas, cities, counties, and specific regions, based on the RSEI index [12,13,14,15], and the results generally prove that RSEI can quickly, objectively, and comprehensively evaluate the ecological environment level of a region. However, the RSEI index places more emphasis on the ecological indicators of the region and cannot effectively reflect the environmental quality status of the region. Therefore, integrating ecological and environmental factors to accurately evaluate ecological environment quality is a significant problem to be solved at present [16].

The BTH region is one of the most developed regions in terms of the economy, science and technology, culture, and education in China. It has an important strategic position in China’s economic and social development. With the rapid increase of regional population and the rapid growth of industry, a series of ecological and environmental problems have occurred, such as accelerated shrinkage of plain depressions, degradation of coastal and estuarine ecosystems, land subsidence, seawater intrusion, frequent sandstorms, and heavily polluted weather in the recent years [17]. In order to improve the quality of ecological environment, local governments have implemented five kinds of comprehensive environmental control measures: industrial structure adjustment, energy structure adjustment, transportation structure adjustment, land use structure adjustment, and green transformation of agriculture and rural areas.

The scientific and accurate evaluation and prediction of ecological environment quality can provide a basis for local governments to adjust their ecological environment protection policies. Therefore, the BTH region was taken as the research object, considering the period 2000–2020 as the research range. The research combined with principal component analysis (PCA) and entropy value method, comprehensive analysis of a long sequence pattern in the spatial and temporal change of the ecological environment quality in BTH region, using a quantitative geography detector to reveal and discuss the key factors affecting the quality of river basin ecological environment and GM1.1 model in the prediction of the quality of the ecological environment in the future, thus laying the foundation for the high-quality development of the ecological environment in the BTH region.

2. Materials and Methods

The technical route of this research is depicted in Figure 1. Firstly, the ecological and environmental data from five periods spanning 2000 to 2020 were collected at five-year intervals. The Risk-Screening Environmental Index (RSEI) was obtained based on the principal component analysis method, the Environmental Quality Index (EQI)was acquired based on the entropy method, and the Remote-sensing Eco-environmental Index (REI)was obtained based on RSEI and EQI. Then, the geographical detector was used to analyze the factors influencing ecological environment quality in the BTH region from the two aspects, namely natural factors and economic and social factors, so as to understand the causes of its spatial differentiation and more accurately depict the mode, direction, path, and intensity of the influencing factors of eco-environmental quality in the BTH region. Finally, the GM1.1 model was used to predict the ecological environment trend of the BTH region in the future.

2.1. Study Area

The BTH region includes two municipalities directly under the central government of Beijing and Tianjin, and 11 prefectural cities under the jurisdiction of Hebei Province, including Shijiazhuang, Chengde, Zhangjiakou, Qinhuangdao, Tangshan, Langfang, Baoding, Cangzhou, Hengshui, Xingtai, and Handan. It is located at 36°05′–42°40′ N, 113°27′–119°50′ E, with a total land area of 2.17 × 10⁵ km² and a population of 1.11 × 10⁸. The city cluster is located at the northern end of the North China Plain, belongs to a typical temperate semi-humid and semi-arid monsoon climate, with the overall terrain characteristics of high northwest and low southeast, and the landform types, including plain, mountain, and hill, etc. [18]. The BTH region is one of the regions with the fastest economic growth and the highest economic level in China, which regional development plays a pivotal role in the whole country. However, the regional ecosystem is under great pressure, and the environmental deterioration is increasingly serious due to the influence of human activities, destroying the sustainable development of ecological quality.

2.2. Data Source and Preprocessing

In this study, historical image data from 2000, 2005, 2010, 2015, and 2020 were selected to monitor and analyze the changes in ecological environment quality in the BTH region. This study used the GEE platform to collect and process Landsat-TM/OLI/TIRS images obtained from the United States Geological Survey (USGS). Moreover, Landsat5/8 products were collected and processed in the GEE platform without downloading [19,20]. Among them, the Normalized Digital Vegetation Index (NDVI), the Normalized Difference Bare Soil Index (NDBSI), Wet and the Land Surface Temperature (LST) data were derived from the GEE platform, and the LANDSAT image processing of the target year was adopted. The PM2.5 data were calibrated by combining the aerosol optical depth (AOD) inversion of NASA MODIS, MISR, and SeaWiFS instruments with the GEOS-Chem chemical transport model, obtaining long-time series ground observations by geographically weighted regression (GWR). Land use was generated by manual visual interpretation on the basis of Landsat-TM remote sensing images, with the types classified according to the degree of human interference on land. The value of unused land is 1, the value of woodland, grassland, and water is 2, the value of cultivated land is 3, and the value of construction land is 4. The biological abundance index was used to adjust the biodiversity. The biological abundance in the study area was obtained through the land use classification in the secondary class multiplied by the corresponding coefficient based on 100-m grid data of land use, and finally using the sum function corresponding to its 1000-m grid. NPP data were from NASA’s MODIS products website (https://ladsweb.Nascom.Nasa.gov, accessed on 1 January 2022), choosing the 2000 to 2020, 500 m resolution MOD17A3H synthetic data.

Nighttime light Data (DMSP-OLS&NPP-VIIRS) were obtained from NOAA’s National Geophysical Data Center, http://ngdc.noaa.gov/eog/download.html (accessed on 1 January 2022). DMSP-OLS data were only updated to 2013, while NPP-VIIRS data were updated from April 2012 to now. Therefore, the data processing method of Cao Ziyang [21] was used in this study to integrate two kinds of night light data to construct the long time series of night light data from 2000 to 2020, and the grid size was 1000 m. DEM data were obtained from the Cloud Platform of National Geographic Monitoring (http://www.dsac.cn/, accessed on 1 January 2022) with a raster resolution of 90 m. The average rainfall data were obtained from China Meteorological Data Network (http://data.cma.cn/, accessed on 1 January 2022), and the average rainfall data of each county were obtained by spatial interpolation of the annual average rainfall statistical data of national meteorological stations in this study. The average temperature data were from the U.S. National Oceanic and Atmospheric Administration (NOAA) consisting of the National Center for Environmental Information (NCEI), and the average annual temperature was calculated based on the latitude and longitude of the meteorological site and meteorological site’s daily average temperature value, by using inverse distance weighted interpolation to get, across the country, the daily average temperature in the form of a raster map, subsequently combining with the BTH region administrative boundaries to get daily average temperature. The population density and GDP data were obtained from the statistical yearbook of the BTH region, and the range method was used to normalize the original data to control the data in the range of [0, 1] to reduce the dimensional impact. The grid size of the temperature, population density, and GDP data was 1000 m.

Table 1. Data sources and preprocessing.

Data Type	Year	Spatial Resolution	Data Sources	Data Preprocessing
PM2.5	2000/2005/2010/2015/2020	0.01° × 0.01°	Aerosol optical depth (AOD) inversion of NASA MODIS, MISR and Sea-WiFS instruments	Calibrated by combining the datas with the GEOS-Chem chemical transport model, and obtained long-time series ground observations by GWR
Land use/land cover	2000/2005/2010/2015/2020	100 m	Resource and Enviroment Science and Data Center (https://www.resdc.cn/, accessed on 1 January 2022)	The biological abundance was obtained through the land use classification in the secondary class multiplied by the corresponding coefficient based on 100 m grid data of land use, and = using the sum function corresponding to its 1000 m grid
NPP	2000/2005/2010/2015/2020	500 m	NASA’s MODIS products website (https://ladsweb.Nascom.Nasa.gov, accessed on 1 January 2022)	Resample the grid to 1000 m
Nighttime light (DMSP-OLS & NPP-VIIRS)	2000/2005/2010/2015/2020	1000/500 m	NOAA’s National Geophysical Data Center (http://ngdc.noaa.gov/eog/download.html, accessed on 1 January 2022)	Integrate two kinds of night light data and resample the grid to 1000 m
DEM	-	90 m	Cloud Platform of National Geographic Monitoring (http://www.dsac.cn/, accessed on 1 January 2022)	Resample the grid to 1000 m
Average rainfall	2000–2020	1000 m	China Meteorological Data Network (http://data.cma.cn/, accessed on 1 January 2022)	-
Average temperature	2000–2020	1000 m	The national centre for environmental information (NCEI)	-
Population density	2000/2005/2010/2015/2020		The statistical yearbook of the Beijing-Tianjin-Hehei region	Added the statistical data to the attributes of countries, and converted to 1000 m grid
GDP	2000/2005/2010/2015/2020		The statistical yearbook of the Beijing-Tianjin-Hehei region	Added the statistical data to the attributes of countries, and converted to 1000 m grid

showed the data sources and preprocessing.

2.3. Methodology

2.3.1. Construction of Remote-Sensing Eco-Environmental Index (REI)

As a comprehensive and multi-factor index, the remote-sensing eco-environmental index (REI) can assess and monitor ecological quality objectively, quickly, and efficiently. Considering the regional characteristics of the BTH region, the REI was constructed from two dimensions, namely ecological quality and environmental quality. Among them, the ecological quality dimension starts with four indicators, namely greenness, humidity, heat, and dryness [22], including LST, NDBSI, NDVI, and Wet, which are closely related to human life and can be intuitively perceived by people, to construct a remote sensing ecological index [23].

The humidity index is expressed by the wet component and monitored by using the tasselled cap transformation [22].

$$\mathrm{W}\mathrm{e}\mathrm{t}=0.2626{\mathsf{\rho }}_1+0.2141{\mathsf{\rho }}_2+0.0926{\mathsf{\rho }}_3+0.0656{\mathsf{\rho }}_4-0.7629{\mathsf{\rho }}_5-0.5388{\mathsf{\rho }}_7$$

(1)

where, ${\mathsf{\rho }}_{\mathrm{i}}$ (i = 1, …, 5, 7) is the influence corresponding to the reflectivity of each band respectively.

The greenness index is measured using NDVI, which is used most widely.

$$\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}=({\mathsf{\rho }}_4-{\mathsf{\rho }}_3)/({\mathsf{\rho }}_4+{\mathsf{\rho }}_3)$$

(2)

The heat index is calculated using the land surface temperature, and the land surface temperature is calculated using the Landsat user manual model and the newly revised calibration parameters such as Chandra:

$${\mathrm{L}}_6=\mathrm{g}\mathrm{a}\mathrm{i}\mathrm{n}\times \mathrm{D}\mathrm{N}+\mathrm{b}\mathrm{i}\mathrm{a}\mathrm{s}$$

(3)

$$\mathrm{T}={\mathrm{K}}_2/\mathrm{l}\mathrm{n}({\mathrm{K}}_1/{\mathrm{L}}_6+1)$$

(4)

where ${\mathrm{L}}_6$ is the radiation value of the pixel of thermal infrared band 6 at the sensor; DN is the grey value of the pixel, gain and bias are the gain value and bias value of the 6-band respectively, and T is the temperature value at the sensor, ${\mathrm{K}}_1 \mathrm{a}\mathrm{n}\mathrm{d} {\mathrm{K}}_2$ are calibration parameters respectively, ${\mathrm{K}}_1$ = 606.09 W/(m²·sr·μm), and ${\mathrm{K}}_2$ = 1282.71 K.

$$\mathrm{L}\mathrm{S}\mathrm{T}=\mathrm{T}/[1+(\mathsf{\lambda }\mathrm{T}/\mathsf{\rho }\left)\mathrm{l}\mathrm{n}\mathsf{\epsilon }\right]$$

(5)

where λ is the center wavelength of 6 bands ($\mathsf{\lambda }$ = 11.45 $\mathsf{\mu }$m); $\mathsf{\rho }$ = 1.438 × ${10}^{-2}$ mK; $\mathsf{\epsilon }$ is the surface emissivity, the value of which is shown in references [22].

The dryness index is measured by the IBI construction index and SI soil index.

$$\mathrm{N}\mathrm{D}\mathrm{B}\mathrm{S}\mathrm{I}=(\mathrm{I}\mathrm{B}\mathrm{I}+\mathrm{S}\mathrm{I})/2$$

(6)

where

$$\mathrm{I}\mathrm{B}\mathrm{I}=\{2{\mathsf{\rho }}_5/{(\mathsf{\rho }}_4+{\mathsf{\rho }}_5)-[{\mathsf{\rho }}_4/({\mathsf{\rho }}_4+{\mathsf{\rho }}_3)+{\mathsf{\rho }}_2/({\mathsf{\rho }}_2+{\mathsf{\rho }}_5\left)\right\}/\{2{\mathsf{\rho }}_5/{(\mathsf{\rho }}_4+{\mathsf{\rho }}_5)+[{\mathsf{\rho }}_4/({\mathsf{\rho }}_4+{\mathsf{\rho }}_3)+{\mathsf{\rho }}_2/({\mathsf{\rho }}_2+{\mathsf{\rho }}_5\left)\right\}$$

(7)

$$\mathrm{S}\mathrm{I}=\left[\right({\mathsf{\rho }}_5+{\mathsf{\rho }}_3)-({\mathsf{\rho }}_4+{\mathsf{\rho }}_1\left)\right]/\left[\right({{\mathsf{\rho }}_{5+}\mathsf{\rho }}_3)+({\mathsf{\rho }}_4+{\mathsf{\rho }}_1\left)\right]$$

(8)

The construction of RSEI could not only be shown by a single indicator, but also by synthesizing the information of the above indicators. Therefore, the principal component analysis (PCA) method was used in this study to compress the features of multiple variables through linear transformation [24]. The PCA concentrated the multi-dimensional information of vertical rotation coordinates in turn into a few feature components. Each feature component often represents certain feature information, with the weight automatically and objectively determined according to the contribution degree of each index to each principal component. In this study, the spatial coordinate axis of the characteristic spectrum is rotated to remove the correlation between the indicators, and then used to focus the information on one principal component, so as to avoid the result deviation caused by subjective weight setting.

$${\mathrm{R}\mathrm{S}\mathrm{E}\mathrm{I}}_0=1-\left\{\mathrm{P}\mathrm{C}1\right[\mathrm{f}(\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I},\mathrm{W}\mathrm{e}\mathrm{t},\mathrm{L}\mathrm{S}\mathrm{T},\mathrm{N}\mathrm{D}\mathrm{B}\mathrm{S}\mathrm{I})\left]\right\}$$

(9)

$$\mathrm{R}\mathrm{S}\mathrm{E}\mathrm{I}=({\mathrm{R}\mathrm{S}\mathrm{E}\mathrm{I}}_0-{\mathrm{R}\mathrm{S}\mathrm{E}\mathrm{I}}_{0\_\mathrm{m}\mathrm{i}\mathrm{n}})/({\mathrm{R}\mathrm{S}\mathrm{E}\mathrm{I}}_{0\_\mathrm{m}\mathrm{a}\mathrm{x}}-{\mathrm{R}\mathrm{S}\mathrm{E}\mathrm{I}}_{0\_\mathrm{m}\mathrm{i}\mathrm{n}})$$

(10)

In addition, taking into account the dimensional difference between each index, in order to avoid weight imbalance, the index is normalized before principal component transformation, and the dimension is unified as [0, 1]. The formula is:

$${NI}_i=(I_i-I_{min})/(I_{max}-I_i)$$

(11)

where NI_i is the index value after normalization, I_i is the index value in pixel I, I_max is the maximum value of the index, and I_min is the minimum value of the index.

Table 2. Results of RSEI principal component analysis.

Components	2000	2005	2010	2015	2020
PC1	94.6234	95.6959	93.3761	94.7164	95.7468
PC2	98.4907	98.5908	97.9041	98.5448	98.4990
PC3	99.9982	99.8824	99.4882	99.5559	99.5543
PC4	100.0000	100.0000	100.0000	100.0000	100.0000

showed the principal component analysis of RSEI in the BTH region in five years. In the five years, PC1 and the feature values contained accounted for more than 90%, indicating that PC1 contains most of the features of ecological indicators, so it can be used to construct remote sensing ecological indexes.

From the four dimensions of atmosphere, biology, plants, and land, the Ecological Quality Index (EQI) was constructed by using the entropy method, including PM2.5, land use (LULC), biomass abundance (BA), and net initial productivity of vegetation (NPP). The formula was as follows:

Calculate the proportion of index $x_{\theta ij}^{″}$ as $R_{\theta ij}$:

$$R_{\theta ij}=x_{\theta ij}^{″}/\sum _{\theta =1}^r\sum _{i=1}^m{x}_{\theta ij}^{″}$$

(12)

Calculate the entropy value of the jth index as $e_j$:

$$e_j=-(\frac{1}{lnrm})\sum _{\theta =1}^r\sum _{i=1}^m{R}_{\theta ij}ln{R}_{\theta ij}$$

(13)

Calculate the difference coefficient of the jth index was calculated as:

$$g_j=1-e_j$$

(14)

Calculate the weight of item j as $w_j$:

$$w_j=g_j/\sum _{j=1}^n{g}_j$$

(15)

Calculate the corresponding index Z:

$$Z=\sum _{j=1}^q{w}_j{x}_{\theta ij}^{″}$$

(16)

Among them, PM2.5 and land use are negative indicators. The higher the values, the worse the environmental quality. Therefore, negative normalization should be carried out during data normalization.

Finally, RSEI is obtained based on RSEI and EQI as follows:

REI = 0.5 × RSEI + 0.5 × EQI

(17)

2.3.2. Geographic Detector and Driving Factors

The geographic detector is the detection of spatial differentiation in the study area [25]. The principle is that the higher the influence weight of an independent variable on the dependent variable, the higher the spatial distribution correlation between them [26,27]. Geographic detectors can analyze the differences and similarities of geographical phenomena in a certain region, which can be responsible for forming the driving force of spatial differentiation [28].

There are four modules of geographic detectors, among which the factor detection module can count the explanatory power of independent variables on dependent variables, and the interaction detection module can judge whether there is an interaction between independent variables, as well as its direction and type [23,29]. In this paper, we used the factor detection module and the interaction detection module to study the influencing factors of the ecological environment in the BTH region.

The factor detection module is used to study the explanation degree of x of a factor to the dependent variable y, which is measured by q value and expressed as follows:

$$\mathrm{q}=1-\frac{{\sum }_{\mathrm{h}=1}^{\mathrm{L}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2}{\mathrm{N}{\mathsf{\sigma }}^2}=1-\frac{\mathrm{S}\mathrm{S}\mathrm{W}}{\mathrm{S}\mathrm{S}\mathrm{T}}$$

(18)

$$\mathrm{S}\mathrm{S}\mathrm{W}={\sum }_{\mathrm{h}=1}^{\mathrm{L}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2$$

(19)

$$\mathrm{SST}=\mathrm{N}{\mathsf{\sigma }}^2$$

(20)

where SST is the total variance of the whole area, SSW is the sum of intra-layer variance, q is the explanation degree of factor x to the dependent variable y, the value range is 0~1. The larger the q value, the more obvious the spatial differentiation, and vice versa.

The interaction detection module is used to make a pairwise comparison before each factor. Firstly, the q value of the two factors on the dependent variable’s y is calculated respectively, and then the q value of the factor after the two factors are superimposed is calculated, and the three values are compared to judge the type of interaction between the two independent variables on the dependent variable. The judgment basis is shown in

Table 3. Judgment methods of geographic detectors interaction detection module.

Interaction Types	Judgment
Attenuation of nonlinearity	q(x1∩x2) < min(q(x1), q(x2))
Double factor enhancement	q(x1∩x2) > max(q(x1), q(x2))
The single-factor nonlinearity is weakened	min(q(x1), q(x2)) < q(x1∩x2) < max(q(x1), q(x2))
Nonlinear enhancement	q(x1∩x2) > q(x1) + q(x2)
Independent	q(x1∩x2) = q(x1) + q(x2)

.

In order to further understand the status of ecological environment quality in the BTH region, the influencing factors of ecological environment quality in the BTH region in the past two decades were detected by the geographic detector. The influencing factors are divided into natural factors and economic and social factors. The natural factors include elevation (X1), average rainfall (X2), and average temperature (X3), and the economic and social factors include GDP(X4), population density (X5), and night light data (X6).

2.3.3. Grey Model (GM) 1.1

The GM1.1 can accumulate and generate the original irregular data and turn them into regular generative series, which can be used to establish the corresponding differential equation model, so as to predict the future development trend and state [30]. Therefore, this paper used GM1.1 to predict the future trend of ecological environmental quality in the BTH region. The specific modelling process is as follows:

Set the original sequence $X_t$ (t = 1, 2, … n), and then add them once to generate a sequence $Y_t$:

$$Y_t={\sum }_{t=1}^n{X}_t, (t=1,2,\dots \mathrm{n})$$

(21)

Take the average of the sequence $Y_t$:

$$Z_t=\frac{1}{2}(Y_t+Y_{t-1}) (t=1,2,\dots \mathrm{n})$$

(22)

Let the estimate of $Y_t$ be $Y_0$. The first-order linear differential equation is:

$$\frac{{dY}_t}{dt}+\alpha Y_t=\mu$$

(23)

where $\alpha$ and $\mu$ represent the undetermined parameter, $\alpha$ is the endogenous grey number, and $\mu$ is the grey development number. The data matrix B is established according to the least square method:

$$\mathrm{B}=\left(\begin{array}{c}\begin{array}{c}\begin{array}{cc}-{\mathrm{Z}}_2& 1\end{array}\\ \begin{array}{cc}-{\mathrm{Z}}_3& 1\end{array}\end{array}\\ \begin{array}{c}\begin{array}{cc}^& ^\end{array}\\ \begin{array}{cc}-{\mathrm{Z}}_{\mathrm{t}}& 1\end{array}\end{array}\end{array}\right)$$

(24)

and,

$$\mathrm{a}=\left(\begin{array}{c}\alpha \\ \mu \end{array}\right)=\frac{{\mathrm{B}}^{\mathrm{T}\mathrm{Y}}}{{\mathrm{B}}^{\mathrm{T}}\mathrm{B}}$$

(25)

Through matrix operation, the undetermined parameters $\alpha ,\mu$ in the matrix a can be obtained, and the following can be obtained by solving and differentiating:

$${\mathrm{Y}}_0=({\mathrm{X}}_1-\frac{\alpha }{\mu }){\mathrm{e}}^{-\mathrm{a}(t-1)}+\frac{\mu }{\alpha }$$

(26)

Hence, the estimated value $X_0$ is obtained:

$$X_0=X_0-Y_{t-1}$$

(27)

Finally, the prediction accuracy is verified according to the fitting accuracy grade table, shown in

Table 4. Fitting accuracy grade table.

Precision Grade	p Value	C Value
Good	>0.95	<0.35
Qualified	>0.80	<0.45
Barely qualified	>0.70	<0.50
Unqualified	≤0.70	≥0.65

.

3. Analysis

3.1. Dynamic Analysis of the Ecological Environment in the BTH Region

3.1.1. RSEI Analysis

The results from statistically analyzing the four ecological indicators and RSEI in the BTH region each year are shown in

Table 5. Statistical Table of ecological indicators in the BTH Region from 2000 to 2020.

Index	Year	Mean Value	Index	Year	Mean Value	Index	Year	Mean Value
LST	2000	0.011	NDBSI	2000	0.343	RSEI	2000	0.846
	2005	0.433		2005	0.342		2005	0.784
	2010	0.307		2010	0.337		2010	0.755
	2015	0.313		2015	0.289		2015	0.802
	2020	0.294		2020	0.197		2020	0.755
Index	Year	Mean Value	Index	Year	Mean Value
NDVI	2000	0.391	Wet	2000	0.282
	2005	0.361		2005	0.271
	2010	0.399		2010	0.303
	2015	0.443		2015	0.293
	2020	0.453		2020	0.282

. The statistical results showed that the RSEI of the BTH region decreased first and then increased during the past 20 years. From 2000 to 2010, the average value of RSEI in the study area decreased from 0.846 in 2000 to 0.755 in 2010, with an overall decrease of approximately 10.7%. Then, the average value of RSEI slightly increased in 2015 and decreased again in 2020, indicating that with the improvement of economic conditions in the Beijing-Tianjin-Hebei region and the strong support of central government policies, its ecological environment protection measures have been further enhanced, leading to a rebound in RSEI.

In terms of spatial distribution, based on the ecological distribution area, the BTH region is divided into a plateau ecological zone (located in the north of Hebei Province), a mountain ecological zone (north and west of Hebei Province, with Shanxi and Henan provinces in the west and south, Liaoning Province in the northeast, and plateau ecological zone in the north, Hebei Plain ecological zone in the east and south), plain ecological zone (including all plain areas south of Yanshan Mountain and east of Taihang Mountain, including Beijing), and ecological sea zone (including Tianjin). It was found that from 2000 to 2020, the RSEI of the mountain ecological zone was generally higher than that of other regions, because woodland and grassland were concentrated in mountainous areas. Hence the NDVI value is higher than in other regions. The plateau ecological zone had a relatively high RSEI in the early 21st century, but decreased with time, while the plain ecological zone was demonstrated the opposite, and the ecological sea zone was relatively stable. Analysis of the reasons that the early twentieth century biome in the mountainous regions and plateau biota demonstrated good ecological conditions produced a high RSEI value, but with economic development, the ecology was destroyed, leading to a weakening trend. However, the RSEI of the plain ecological zone displayed a rising trend in recent years, possibly because its geographical location is located in the economic and political center of the BTH region, the ecological protection measures of which are better than other regions, with the attention of the government focused on ecological protection, meaning that it displays an upward trend.

From the changes in each index, NDVI and Wet, which represent better ecological conditions, showed a fluctuating trend in the past 20 years. Wet showed a small low peak and a small peak in 2005 and 2010, respectively, and began to show an upward trend in 2010. However, NDVI showed a downward trend from 2000 to 2005 and then showed an overall upward trend. LST and NDBSI, which represent the poor ecological environment, showed a downward trend in the past 20 years, indicating that the overall ecological quality has improved in the past 20 years.

3.1.2. EQI Analysis

Statistically analyzing the four environmental indicators and EQI of the BTH region in each year, the results are shown in

Table 6. Statistical Table of environmental indicators in the BTH region from 2000 to 2020.

Index	Year	Mean Value	Index	Year	Mean Value	Index	Year	Mean Value
LULC	2000	0.583	PM2.5	2000	0.841	EQI	2000	0.199
	2005	0.584		2005	0.820		2005	0.242
	2010	0.574		2010	0.842		2010	0.209
	2015	0.579		2015	0.808		2015	0.212
	2020	0.573		2020	0.883		2020	0.223
Index	Year	Mean Value	Index	Year	Mean Value
NPP	2000	0.137	BA	2000	0.235
	2005	0.166		2005	0.293
	2010	0.151		2010	0.237
	2015	0.162		2015	0.230
	2020	0.163		2020	0.235

. The statistical results show that EQI has fluctuated and increased in the past two decades, and its mean value showed an upward trend from 2000 to 2005, indicating that the rapid economic development in the early 20th century did not have an excessive impact on the environment of the BTH region. Moreover, from 2005 to 2010, EQI showed a downward trend, indicating that the environment of the BTH region has been damaged due to the influence of economic development.

In terms of spatial distribution, based on the ecological distribution area, the BTH region is divided into a plateau ecological zone, mountain ecological zone, plain ecological zone, and ecological sea zone. Moreover, it was found that the spatial distribution in 2000–2020 displayed no obvious change. Among them, the EQI of the mountain ecological zone has been at a higher level, and the cities represented by Beijing have a higher environmental level. The specific reason for this is that the mountain ecological zone is the concentrated distribution area of mountains, hills, and basins, as well as the concentrated distribution area of forests in Hebei; its zonal vegetation is deciduous broad-leaved forest, and the vertical distribution of vegetation is obvious. In addition, it is not only the most important forest and animal husbandry production development base in Hebei Province, but also a relatively poor area in Hebei Province, so it is less affected by social and economic development than other areas, consistently placing its EQI in a high position in the 20 years considered.

From the changes of each index, BA and NPP, which represent the better environment condition, both experienced a fluctuation and tended to stabilize. BA peaked in 2005 and returned to the level of 2000 in 2010, and then tended to be balanced in the following ten years, while the NPP index improved from 2000 to 2005, and remained relatively stable for the next 15 years. After the standardized operation of LULC and PM2.5, which represent poor environmental conditions, it is found that the fluctuation of land use is relatively small, and the overall value has shown a downward trend in the past 20 years, indicating that the overall quality of the environment has deteriorated. However, PM2.5 shows a fluctuation state, where the value decreased from 2000 to 2005, and then rose in 2010. The fluctuation state is consistent with reality, indicating that the BTH region was greatly affected by haze in 2015, and the negative impact of haze on the environment was alleviated in 2020 with the strengthening of environmental management.

3.1.3. REI Analysis

Analyzing the ecological environment quality in the BTH region, it was found that it fluctuated and declined in the past 20 years. Specifically, it showed a downward trend from 2000 to 2010, partially rebounded in 2015, and decreased again in 2020. Shown in Figure 2, in terms of spatial distribution, the overall ecological environment quality of Beijing is relatively good, the overall ecological environment quality of Tianjin is relatively poor, and the overall ecological environment quality of northern and western Hebei is relatively good compared with southern and eastern Hebei. With respect to ecological environment quality changes in the BTH area, the area of REI for level 1 increased by 6.5%, the area of REI for level 2 remained unchanged, accounting for approximately 52%, the area of REI for level 3 was reduced by 3%, the area of REI for level 4 was reduced by 2.5%, and the area of REI for level 5 was reduced by 0.5% from 2000 to 2020. This means that in the past 20 years, overall ecological environment quality has shown a downward trend.

Shown in Figure 3, in terms of grades, the area of REI for level 1 increased by 7% from 2000 to 2010, then decreased again in 2015, before increasing again by 6% in 2020. The area of level 2 also increased by 6% between 2000 and 2010, and then fell back to the 2000 level again. However, the area of REI for level 3, level 4, and level 5 plummeted from 2005 to 2010, recovering by 2015, and falling again by 2020. This indicates that the ecological environment quality was poor in 2010, picked up in 2015, and declined again in 2020 on the whole. Therefore, the ecological environment quality still needs great attention at this stage.

Shown in Figure 4, from the perspective of the changes in ecological environment quality in the BTH region, 85% of the areas experienced a decline in REI from 2000 to 2020, while 15% of the areas experienced an increase in REI, which was mainly concentrated in Beijing and the southern part of Hebei region. In terms of five years, only from 2010 to 2015 were there more areas of REI improvement than areas of decline, and in other years, there were more areas of decline than areas of increase. This indicated that the improvement effect of ecological environment quality in 2015 was good, while the overall decline in other years was consistent with the above research results.

3.2. Analysis of Influencing Factors of Ecological Environment Quality in the BTH Region

In order to integrate the spatial resolution of different data sources and collect special diagnosis, a 5 km × 5 km grid was constructed to divide the spatial sample units of the whole watershed, and the corresponding values of REI and each independent variable were obtained by sampling the center point of the grid. Among them, the dependent variable REI adopts the continuous value, and the other independent variables are discretized into ten categories by the natural segment point method or the equal spacing method to obtain the detection results. Shown in Figure 5, among the natural factors, the q value of dem ranked first from 2000 to 2005, and then it ranked second from 2005 to 2020, indicating that topographic conditions have always been an important criterion affecting the quality of the ecological environment. The q value of average temperature ranked third in 2000, second in 2005, and then always ranked first, indicating that with the environmental trend of global warming, the influence of temperature condition degree on ecological environment conditions is becoming more and more serious. The q value of population density ranked second in 2000, then became worse, and ranked fifth in 2020, indicating that in the modern society with more and more frequent population migration, the impact of population on ecological environment quality is gradually decreasing. The q value of night light data ranked fifth in 2000 and increased to third in 2020, with a rising trend, which means that, with the development of urban and rural construction of the BTH region, the population density and social impact on the quality of the ecological environment will strengthen gradually, so it is necessary to regulate the behavior of society, so as to achieve the protection of the ecological environment. In addition, the q values of average rainfall and GDP rank lower and have not changed significantly in the past 20 years, indicating that the two indexes have little impact on the ecological environment quality of the BTH region.

On the whole, compared with economic and social factors, natural factors have a more obvious driving effect on ecological environment quality in the BTH region. Among them, the average temperature and elevation demonstrated strong explanatory power on the spatial heterogeneity of ecological environment quality changes in the BTH region, while the average rainfall was relatively low. However, with the passage of time, the impact of social and economic factors on the quality of the ecological environment is becoming higher and higher. Among them, population density has a more significant impact on the ecological environment quality of the BTH region than GDP, and night light data has a higher influence on the ecological environment quality of the BTH region than GDP and population density, indicating that urbanization is an important factor affecting the ecological environment quality of the BTH region at present and in the future.

Shown in

Table 7. Interactive probe statistics.

	2000	2005	2010	2015	2020
dem ∩ GDP	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement	Nonlinear enhancement
dem ∩ pre	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement	Nonlinear enhancement
dem ∩ rk	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement
dem ∩ tmp	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement
dem ∩ yj	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement
GDP ∩ rk	Nonlinear enhancement	Nonlinear enhancement	Nonlinear enhancement	Nonlinear enhancement	Nonlinear enhancement
GDP ∩ yj	Nonlinear enhancement	Nonlinear enhancement	Nonlinear enhancement	Nonlinear enhancement	Nonlinear enhancement
pre ∩ GDP	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement	Nonlinear enhancement
pre ∩ rk	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement	Nonlinear enhancement
pre ∩ tmp	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement	Nonlinear enhancement
pre ∩ yj	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement	Nonlinear enhancement
rk ∩ yj	Nonlinear enhancement	Nonlinear enhancement	Nonlinear enhancement	Double factor enhancement	Double factor enhancement
tmp ∩ GDP	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement	Nonlinear enhancement
tmp ∩ rk	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement	Nonlinear enhancement
tmp ∩ yj	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement	Double factor enhancement

, by interaction detection result, the interactions between factors showed double factor enhancement and nonlinear enhancement. There is no independence or weakening, which shows that the interaction factors related to the impact of the change of the BTH region ecological environment quality are higher than the original factors separately related to the impact of the BTH region ecological environment quality changes. From the interaction results of factors in different years, in 2000, 2005, and 2010, most of the interaction results between factors displayed double factor enhancement, while in 2015 and 2020, most of the interaction results between factors displayed nonlinear enhancement, further indicating that the interaction between factors became more and more complex with the passage of time.

Shown in Figure 6, in terms of factors, from the interaction between factors, the interaction between average rainfall and the average temperature had the most significant influence on the ecological environment quality in the BTH region from 2000 to 2020, followed by average rainfall and elevation. On the one hand, this verifies that the interaction between factors in the BTH region has a stronger influence on the quality of the ecological environment than a single factor. On the other hand, it also indicates that the internal interaction between environmental factors is still the critical factor affecting the ecological environment quality in the BTH region. In addition, it is not difficult to find that the interaction detection value of GDP and average rainfall with other factors is significantly higher than the single factor explanatory value, indicating that the single factor without the good explanatory ability for ecological environment quality will enhance the effect more obviously under the joint action of other factors. With the interaction of the values of the average temperature and the density of the population, GDP also increased after 2010, indicating that economic activities in different environments, especially influenced by temperature, will have a more significant effect on ecological environment quality in the BTH region. This also means that the temperature will not only have a certain impact on the economic production activities of agriculture, forestry, fishery, and animal husbandry, but also that the temperature environment will have an impact on various economic activities. This further shows that economic and social factors have an increasingly greater influence on the change of the BTH ecological environment quality, which is consistent with the interaction results of factors.

3.3. Prediction and Analysis of Ecological Environment Quality in the BTH Region

The change in the ecological environment is closely related to human civilization, which means that predicting the future trend of ecological environment quality can provide good measures and suggestions for the current social behavior. In this paper, spatial raster data are used to analyze the ecological environment quality, which can directly observe the spatial dynamic changes of the ecological environment quality in the BTH region. However, there are certain limitations in judging the changes in the ecological environment quality in the BTH region. Therefore, in order to better understand the administrative divisions of the BTH region under the current situation and future trend of the ecological environment change while considering the availability of data, this paper adopts the GM1.1 model, for five years as the unit of time, and with area, city, and county as division units, to undertake the dynamic monitoring of ecological environment quality in the study period, 2000–2020, in the BTH region, as well as to predict the future development trend of ecological environment quality in the BTH region in 2025, 2030, and 2035.

To vertically contrast the district ecological environment quality of cities or counties in the BTH region under long time series, this paper divided REI into three categories using a spacing method based on standardizing data and defined it according to the three categories of good, general, and poor. The statistical results are shown in

Table 8. County risk statistics of BTH region.

Year	Number of Counties in Good Condition	Number of Counties in Poor Condition
2000	22	0
2005	22	2
2010	14	3
2015	23	1
2020	12	5
2025	11	7
2030	14	14
2035	14	25

. On the whole, the ecological environment changes in the BTH region showed an overall trend of improvement in the short term, and the overall REI increased stably.

Shown in Figure 7, in terms of spatial distribution, the future spatial distribution of ecological environment quality in the BTH region has no significant change from the current one, and is in a good state in Beijing, a small part of southwest Hebei province and northeast Hebei province, while it is poor in northwest Hebei province and Southeast Hebei province.

In terms of quantity distribution, the number of counties with good ecological environment quality was small in 2010, rose in 2015, and declined again in 2020, which led to the prediction that that the number of counties with good ecological environment quality would be relatively stable in the short term in the future. However, there were no counties with poor ecological environment quality in 2000, in the future forecast period from 2025 to 2035, shown in Figure 8, the number of counties with poor ecological environment quality in 2025 does not change much compared with that in 2020. However, the number of counties with poor ecological environment quality in the northwest and southeast of Hebei will increase significantly when extrapolated to 2030 and 2035, which indicates that in the future, more attention should be paid to the ecological environment quality of these regions, and the ecological quality of these regions should be improved through policy guidance, people’s cooperation, and other ways.

4. Discussion and Conclusions

4.1. Discussion

(1) This paper not only starts from the ecological dimension, but also considers the environmental dimension in the construction of ecological environment quality, which more comprehensively considers the ecological environment quality of the BTH region. At the same time, the combination of principal component analysis and the entropy method reduces the influence of human factors on the results, which makes it possible to completely evaluate the ecological environment quality more scientifically, and further provides a new idea in the direction of multi-factor detection of ecological environment quality. In addition, the data selected in this paper are mostly comprised of remote sensing data, whose characteristics, such as fast data update and convenient acquisition, improve the practicability of dynamic monitoring of the ecological environmental quality in the BTH region.

(2) This paper selected natural, economic, and social factors only when analyzing the influencing factors of ecological environment quality in the BTH region, presenting certain limitations in the selection of indicators. The nature of ecological environment quality makes it not only affected by multiple factors, but also influenced by the interaction between different factors. Therefore, it is necessary to further increase the dynamic relationship between the internal connection of factors and the quality of the ecological environment.

(3) In this paper, the future ecological environment quality of the BTH region is predicted by the GM1.1 model, which showed the future trend of the ecological environment quality of the BTH region to a certain extent. However, the prediction model cannot determine the stability of the trend, so it is necessary to analyze the stability of the trend in further study, so as to evaluate the ecological environment development of the BTH region in the future.

4.2. Conclusions

(1) This paper, based on remote sensing data, combined with principal component analysis (PCA) and entropy value method, analyzed the ecological environment quality from the ecological dimension and the environmental dimension of 2000 to 2020 in the BTH region, and found that the ecological quality recovered better after destruction, environmental quality continued at low-quality level after destruction, and the ecological environment quality was decreased on the whole. Moreover, long time series studies found that the change in the ecological environment is closely related to the implementation of the policies and measures. The rapid development of the social economy in the BTH region destroyed the ecological environment quality to a certain extent, but it is obvious that when the state issues response policies and actively responds to certain measures to protect the ecological environment, the RSEI, REI, and EQI are improved, which indicates that the ecological environment quality of BTH region is greatly affected by the policy, and the ecological environment protection measures can be strengthened to improve the regional ecological environment in the future.

(2) This paper analyzed the factors influencing ecological environment quality in the BTH region based on the geographic detector, finding that the ecological environment quality in the BTH region is more affected by natural factors. However, with the development of the urban economy, the impact of economic and social factors is increasing. Therefore, when formulating relevant regulations, it is necessary for the government to consider the effect on the ecological environment in advance in urban planning and urban and rural construction, so as to reduce the damage to the ecological environment and lay the foundation for building a harmonious and beautiful ecological environment.

(3) This paper predicted future ecological environment quality in the BTH region based on the GM1.1 model. The results showed that in terms of spatial distribution, namely comparing the ecological environment quality of the future spatial distribution and the present, there is no apparent change on the whole. The ecological environment quality is in a good state in Beijing, a small part of southwest Hebei province, and northeast Hebei province, while it is poor in northwest Hebei province and Southeast Hebei province. In terms of quantity distribution, 2025 will not change much compared with 2020, but from the prediction results in 2030 and 2035, the number of counties with poor ecological environment quality in the northwest and southeast Hebei will increase significantly, indicating that more attention should be paid to the ecological environment quality of these regions in the future, and regional ecological quality should be improved through policy guidance, people’s cooperation, and other ways.

(4) In order to improve the ecological environment quality of the BTH region, the local governments should carry out economic transformation and structural reform and make the BTH region enter a virtuous circle of green and low-carbon development. The local governments should strengthen regional linkage and departmental cooperation, ensure multi-factor coordination and cross-regional coordination of the ecological environment, and realize the overall protection, system restoration, and comprehensive management of the BTH region’s ecological environment. The local governments should strengthen the pertinence and innovation of the BTH region’s ecological environment factor management and continue to further promote the coordinated management of air pollution. Moreover, full use should be made of the South-to-North Water Diversion Project to improve the regional water ecology and environment. Local governments should pay attention to research on the protection and governance of the ecological environment in urban agglomerations, strengthen basic research and development in relation to key technologies, explore the scientific laws and governance methods of urban agglomerations’ ecological environment changes, and increase the application, demonstration, and promotion of existing, high-quality, scientific, and technological achievements.