Spatiotemporal Patterns and Driving Factors of the Ecological Environmental Quality along the Jakarta–Bandung High-Speed Railway in Indonesia

Abstract

Economic globalization and rapid urbanization have affected ecological environments in global regions to different degrees. Timely, objective and quantitative evaluations of the spatiotemporal variations in regional ecological environmental quality are the key to environmental protection and decision making. The spatial distributions of ecological environment quality levels along the Jakarta–Bandung high-speed railway from 2000 to 2020 were obtained based on Google Earth Engine and the Remote Sensing Ecological Index (RSEI). Then, the Theil–Sen median and Mann-Kendall methods were used to assess the temporal RSEI trend; the spatial autocorrelation evaluation index was used to evaluate RSEI clustering on a spatial scale. The results show that the overall ecological environmental quality from 2000 to 2020 was moderate, with temporally fluctuating changes and very significant spatial clustering. Approximately 20% of the area showed very strong changes (improvement or degradation). Areas with relatively better ecological quality were located mainly in relatively high-elevation and high-vegetation-coverage regions, while areas with poorer ecological quality were located mainly in the four major cities, including Jakarta, with concentrated populations and developed economies. The research results provide local governments with scientific suggestions regarding the synergistic development of high-speed railway construction and ecological environmental protection.

1. Introduction

Globally, climate change, dramatic increases in human activities and irrational natural resource use patterns have led to a rapid decline in the self-adjustment capacities of ecosystems and a proliferation of ecological and environmental problems [1,2,3,4]. For example, water resources have become polluted [5,6], extreme incidents occur frequently [7,8], vegetation destruction has led to species extinction [9] and air quality has degraded [10]. Ecological and environmental research has become a key concern in the fields of global environmental change and sustainable development. Timely, dynamic, objective and accurate monitoring of the ecological environment provides an important basis for promoting harmony between humans and nature.

The ecological environment is primarily monitored via ecological environmental quality (EEQ) assessments. EEQ assessment studies are mainly carried out based on ecological indices. For example, the normalized difference vegetation index (NDVI) [11,12] is used to evaluate vegetation coverage, density and vitality, while the land surface temperature (LST) [13,14] is used to assess heat stress and drought severity. These indices are simple and easy to use, but, due to the complexity of ecosystems, they often cannot comprehensively reflect various factors of the ecological environment [15,16]. Therefore, some scholars have developed comprehensive ecological indices to evaluate the EEQ. There are three main types: the ecological index (EI) [17], the analytic hierarchy process (AHP)-based ecological index [18] and the remote sensing ecological index (RSEI) [19,20]. Among them, the EI index is derived from the “Technical Specifications for the Evaluation of Ecological and Environmental Conditions” (HJ192-2015), which were published in China. It can effectively assess the ecological environment conditions at the county, provincial and ecological regional levels. However, the calculation of this index relies on statistical data. Detailed statistical data are difficult to obtain, especially in regions with relatively low levels of development, making long-term ecological environmental assessment impossible. In addition, evaluations based on statistical data often use the mean value of a certain indicator to represent the entire region, ignoring the spatial variability in geographic elements. The AHP method is also frequently used in EEQ research, but the calculation of indicator weights mainly relies on expert scoring, which has the drawback of subjectivity. The RSEI was proposed by Xu et al. [19,20]. This index is obtained by calculating four subindices of greenness, dryness, humidity and heat to obtain the RSEI, which can comprehensively reflect the overall ecological environment of a region that contains a variety of natural geographical elements, such as water bodies and soil. Its data, sourced entirely from remote sensing images, can be obtained easily and quickly. The index adopts the principal component analysis (PCA) method to determine the weight of the subindicators; thus, the weight calculations are completely dependent on the data themselves, improving the objectivity and reliability of the evaluation results. In addition, the RSEI has the characteristics of repeatability, comparability and visualization of the results and can be used for dynamic monitoring of the ecological environment over long time scales, making it the most widely used index in recent years [21,22,23,24,25]. It has greatly contributed to the development of the field of remote sensing ecology.

Through a review of domestic and international EEQ research, the following can be observed: (1) Currently, research on EEQ assessment mainly focuses on simple geographical elements, such as cities (e.g., Fuzhou [26], Meizhou [27]), urban clusters (e.g., the Changsha–Zhuzhou–Xiangtan Metropolitan Region [22,28], the Chengdu–Chongqing Economic Circle [29]), watersheds (e.g., Yangtze River [16,30], Yellow River [31,32], Chishui River [33], Ulan Moron River [34]) and lakes (e.g., Dongting Lake [23], Taihu Lake [24]). There is limited research on complex geographical regions, especially economic corridors of major engineering projects, which have received less attention from scholars. These regions are characterized by complex terrain, significant differences in topography, diverse landforms, large disparities in development levels and extremely imbalanced population distributions. However, these regions are often of significant importance for promoting economic growth, advancing regional integration and achieving sustainable regional development; (2) A comprehensive analysis of the spatiotemporal patterns in EEQ is significant for monitoring the ecological conditions of a region. However, previous studies have predominantly focused on one aspect (either temporal or spatial) for in-depth analysis. For example, Xiong et al. [25] extensively discussed the spatial distribution patterns in EEQ in the Erhai Lake Basin in Yunnan Province, primarily based on spatial autocorrelation indices. Zheng et al. [35], on the other hand, explored in depth the temporal variations in EEQ in five representative regions worldwide. In summary, current research on comprehensive spatiotemporal analysis in complex geographical regions is unable to meet the practical guidance requirements for the sustainable development of regional ecological environments.

Therefore, based on existing studies, for this paper, we selected the area along the Jakarta-Bandung High-Speed Railway (JBHSR) with complex geographical elements as the study case area and, based on the RSEI index, comprehensively explored and analyzed the spatiotemporal patterns in EEQ under these conditions, as well as its driving factors. First, by conducting field surveys in the research area and engaging in discussions with the high-speed rail construction authorities, we obtained vector data on the rail alignment and collected natural, economic and human geographic information for the study area. Then, the RSEI for the JBHSR corridor from 2000 to 2020 was constructed using the Google Earth Engine (GEE) platform, and the spatiotemporal evolution characteristics were analyzed in conjunction with the collected geographic information. Finally, the factors influencing the EEQ were explored using Geodetector. This research can provide scientific guidance and suggestions for the synergistic development of regional ecological and environmental protection under the conditions of complex geographical elements.

2. Study Area and Data Sources

2.1. Study Area

The JBHSR is located on the island of Java in Indonesia, connecting the capital city of Jakarta with the fourth-largest city, Bandung. The study area extends from 5°54${}^{\prime }$ to 7°19${}^{\prime }$ S and 106°42${}^{\prime }$ to 107°54${}^{\prime }$ E (Figure 1). The region has a tropical rainforest climate, with an average annual temperature ranging from 20 to 30 ℃. During the summer season, temperatures often exceed 30 °C, with the possibility of reaching 35 °C or higher. The relative humidity is relatively high, typically ranging from 70% to 90%. The area experiences seasonal climate variations. The rainy season is characterized by the northwest monsoon, with frequent rainfall, cloudy skies and thunderstorms. The dry season is dominated by the southeast monsoon, with more sunny days and less rainfall, creating a relatively dry climate. The region has diverse and complex landforms. The northern part is primarily composed of plains with relatively flat terrain, consisting of rivers, lakes and agricultural fields. The southern part is characterized by mountainous areas with higher elevations, steep peaks and winding valleys. The geological structure is also complex, with a significant presence of special types of soil and rock formations, such as soft soil, loose soil and volcanic sediment [36].

The JBHSR is Indonesia’s first high-speed rail project and an important achievement under the Belt and Road Initiative. It holds significant strategic importance [37]. In recent years, the region has faced various ecological threats due to human activities and rapid urban development. These threats include extreme drought [38], extreme high temperatures [39,40,41], environmental degradation [42], sea-level rise [43] and land subsidence [44,45], among others. Furthermore, activities such as land acquisition and construction may lead to negative impacts on land use changes, ecosystem degradation and loss of biodiversity in the areas along the railway line. However, there is currently a lack of relevant reports on the ecological environmental status in this region and a distinct lack of detailed ecological quality assessment results to guide the local government and people in making scientifically informed decisions for the harmonious development of high-speed rail construction and the protection of the ecological environment. Moreover, high-speed railway construction will inevitably cause significant population migration, industrial transfers and urban expansion activities, leading to spatiotemporal changes in the ecological environment. Therefore, an evaluation of the ecological environment in this region is of great practical significance.

Currently, there is a lack of evaluation studies related to the EEQ along the JBHSR. Our previous studies have analyzed extremely high temperatures in the region [39,41], but these reflect only regional temperature phenomena and do not allow for a comprehensive assessment of the ecological condition. Therefore, it is imperative to conduct more in-depth ecological environmental assessment research in this region to provide reference recommendations for regional sustainable development.

2.2. Data Collection and Preprocessing

The MODIS image data (MOD11A2, MOD13A1, MOD09A1) used in this study were obtained from the Google Earth Engine database. The MOD11A2 data product provides LST data, which are synthesized from MOD11A1 data and gridded at approximately 1 km spatial resolution [46]. These data were used to calculate the heat index. The MOD13A1 dataset utilizes the NDVI layer from MOD13A1 V6, which provides the best composite pixel values for a spatial resolution of 500 m over a 16-day period. This dataset was used to calculate the greenness index. The MOD09A1 dataset provides surface spectral reflectance estimates for the first to seventh bands of Terra MODIS. These data were atmospherically corrected and used to calculate the dryness and wetness indices. We obtained all images of the area along the JBHSR from January 2000 to December 2020 with the GEE platform and preprocessed them with cloud shadow removal, water masks and uniform spatial resolution (resampling all images to 1 km resolution). We then calculated four subindicators based on the GEE platform to determine the RSEI via the PCA module.

ASTER GDEM data were selected as the digital elevation model (DEM) data. Slope calculations were performed based on the DEM data using ArcGIS. Population data utilized population grid data, where each grid pixel value represented the population count within the unit; this dataset can be freely downloaded from WorldPop. Multiple studies have found a significant positive correlation between nighttime light data and regional economic levels [47,48,49]. Therefore, this study employed nighttime light remote sensing data as a representation of economic levels. The data were obtained from the globally unified nighttime light dataset developed by Chen et al. [50], covering the period from 2000 to 2020. OpenStreetMap (OSM) data were chosen, and relevant features of interest, including water bodies, roads, parks, industrial areas and community centers, were extracted using ArcGIS. The Euclidean distances to water bodies, roads, parks, industrial areas and community centers were calculated separately.

For the aforementioned 9 layers of data, the natural breaks method was employed to reclassify each layer into 10 classes and assign new values ranging from 1 to 10, achieving data discretization. Subsequently, a grid of fishing points measuring 1 km by 1 km was established, and the data values from the nine layers were extracted and assigned to their corresponding points. This preparation was conducted to serve as input data for the Geodetector. Detailed information on all the data is presented in

Table 1. Data sources.

Dataset	Resolution	Website
MOD13A1	500 m	Google Earth Engine (https://developers.google.cn/earth-engine (accessed on 1 December 2022))
MOD11A2	1000 m	Google Earth Engine (https://developers.google.cn/earth-engine (accessed on 1 December 2022))
MOD09A1	500 m	Google Earth Engine (https://developers.google.cn/earth-engine (accessed on 1 December 2022))
DEM	30 m	https://earthdata.nasa.gov/ (accessed on 20 December 2022)
Population data	100 m	WorldPop (https:125126//www.worldpop.org/ (accessed on 21 December 2022))
NTL	500 m	https://doi.org/10.7910/DVN/YGIVCD (accessed on 21 December 2022)
OSM	-	https://download.geofabrik.de/asia/indonesia.html (accessed on 22 December 2022)

.

3. Theory and Methodology

Figure 2 illustrates the comprehensive workflow employed in this study. First, the spatial distribution maps of the RSEI for 2000, 2005, 2010, 2015 and 2020 were computed using the GEE platform. Based on the obtained spatial distribution maps of RSEI, analyses were conducted from both temporal and spatial perspectives. Finally, Geodetector was employed to explore the factors influencing the EEQ.

3.1. Construction of RSEI

3.1.1. Construction of the RSEI Indicators

Construction of the RSEI requires four component indicators: greenness, wetness, heat and dryness, which are closely related to the ecological environment. The greenness index is represented by the NDVI, which can effectively detect vegetation growth status and coverage [51]. A higher NDVI value indicates a better ecological environment. The wetness indicator is derived from the tasseled cap transformation [52] and represents the overall moisture of water, soil and vegetation [21]. A higher wetness value indicates a better ecological environment. The dryness index, referred to as the NDBSI, is represented by the average value of the bare soil index (SI) and the index-based built-up index (IBI). A higher dryness index value indicates a poorer ecological environment. The heat index is described using the LST, which is obtained by converting the grayscale values of remote sensing data into degrees Celsius to determine the surface temperature of the region. A higher heat index indicates a poorer ecological environment. The formulas and parameter meanings for the four components are provided in

Table 2. Formulas and parameter meanings for the four components.

Components	Formulas	Parameter Meanings
Greenness	$NDVI=\frac{{\rho }_{NIR}-{\rho }_{\mathrm{Re}d}}{{\rho }_{NIR}+{\rho }_{\mathrm{Re}d}}$	${\rho }_{\mathrm{Red}}$ and ${\rho }_{NIR}$ refer to the spectral reflectances in the red and near-infrared bands, respectively.
Heat	$LST=0.02D{N}_S-273.15$	$D{N}_S$ denotes the grayscale value of the surface temperature image.
Humidity	$WET=0.1147b_1+0.2489b_2+0.2408b_3+0.3132b_4$ $-0.3122b_5-0.6416b_6-0.5087b_7$	$b_1$ to $b_7$ denote the surface reflectance of MODIS images in bands 1–7, respectively.
Dryness	$NDBSI=(SI+IBI)/2$ $SI=[(b_6+b_1)-(b_2+b_3)]/[(b_6+b_1)+(b_2+b_3)]$ $IBI=\frac{2b_6/(b_6+b_2)-\left[b_2/(b_2+b_1)+b_4/(b_4+b_6)\right]}{2b_6/(b_6+b_2)+\left[b_2/(b_2+b_1)+b_4/(b_4+b_6)\right]}$	$b_1$ to $b_7$ denote the surface reflectance of MODIS images in bands 1–7, respectively.

.

3.1.2. Calculation of the RSEI

(1) Due to the inconsistency of the four indicators, normalization was performed as a preliminary step:

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

(1)

where $N{I}_i$ is the pixel value of the image and $I_i$ is the pixel value at raster i. $I_{max}$ and $I_{min}$ are the maximum and minimum values of the image, respectively;

(2) We conducted PCA. The calculation formula is as follows:

$$RSE{I}_0=PC1\left[f(WET,NDVI,LST,NDBSI)\right]$$

(2)

Here, f refers to the PCA function;

(3) If $RSE{I}_0$ showed low values under good ecological conditions and high values under poor ecological conditions, then the $RSE{I}_0$ value was subtracted from 1, such that high values represented better ecological conditions. The RSEI is expressed as follows:

$$RSEI=1-RSE{I}_0=1-PC1\left[f(WET,NDVI,LST,NDBSI)\right];$$

(3)

(4) Finally, the RSEI was normalized to a range of 0 to 1. Using ArcGIS, an equal interval method was employed to divide the RSEI values into five categories [20,53]: poor, fair, moderate, good and excellent.

3.2. Pixel-Level Dynamic Detection

The Theil–Sen (T–S) median method is a robust nonparametric statistical approach for trend estimation. It is often used in conjunction with the Mann–Kendall (M–K) nonparametric test. The combination of these two methods can be used to observe significant changes in the locations of individual pixels in annual data distributions and has been widely used to monitor trends in vegetation, precipitation, temperature and other long-term time series data [54,55,56]. Therefore, in this study, we combined the T–S and M–K methods to characterize the long-term spatiotemporal changes in EEQ along the JBHSR at the pixel scale. The T–S calculation formula is expressed as follows:

$$\beta =median{\frac{RSE{I}_j-RSE{I}_i}{j-i}}^{}{}^{}\left(j>i\right)$$

(4)

where $RSE{I}_j$ and $RSE{I}_i$ denote the RSEI values in the corresponding years j and i, respectively. When $\beta$ is greater than 0, the ecological environment is improving; otherwise, it is deteriorating. The formula used to calculate M–K is expressed as follows:

$$Z=\left\{\begin{array}{cc}\frac{S-1}{\sqrt{Var\left(S\right)}}& (S>0)\\ 0& (S=0)\\ \frac{S+1}{\sqrt{Var\left(S\right)}}& (S<0)\end{array}\right.$$

(5)

$$S=\sum _{i=1}^{n-1}\sum _{j=i+1}^{n}sgn(RSE{I}_j-RSE{I}_i)$$

(6)

$$sgn(RSE{I}_j-RSE{I}_i)=\left\{\begin{array}{cc}1& (RSE{I}_j-RSE{I}_i>0)\\ 0& (RSE{I}_j-RSE{I}_i=0)\\ -1& (RSE{I}_j-RSE{I}_i<0)\end{array}\right.$$

(7)

$$Var\left(S\right)=\frac{n(n-1)(2n+5)}{18}$$

(8)

Here, n represents the number of years, sgn denotes the sign function and Z is the significance statistic of the RSEI trend. The significance level of RSEI is generally set at 0.05. When the absolute value of Z exceeds 1.65, 1.96 and 2.58, the trend passes the significance test with confidence levels of 90%, 95% and 99%, respectively. The method for determining trend significance is presented in

Table A1. Mann–Kendall test trend categories (adapted from [16,63]).

$\mathit{\beta }$	Z	Trend Type	Trend Features
$\beta >0$	$2.58<Z$	4	Very strong improvement
	$1.96<Z\le 2.58$	3	Strong improvement
	$1.65<Z\le 1.96$	2	Moderate improvement
	$Z\le 1.65$	1	Light improvement
$\beta =0$	Z	0	No change
$\beta <0$	$Z\le 1.65$	−1	Light degradation
	$1.65<Z\le 1.96$	−2	Moderate degradation
	$1.96<Z\le 2.58$	−3	Serious degradation
	$2.58<Z$	−4	Very serious degradation

.

Dynamic monitoring at the pixel scale was implemented with MATLAB code via the following steps: (1) A georeferenced raster file (GeoTIFF format) was imported with projection information as the input data to obtain the projection information; (2) The code read the raster data for each year; (3) The Theil–Sen median was calculated for each pixel, and the test statistic S was calculated for the Mann-Kendall test; (4) For each pixel, different trend levels were assigned based on the $\beta$ value and the significance level Z (Table A1). The trend levels were used to indicate the trend of changes in EEQ, where positive values indicated improvement, negative values indicated degradation and 0 indicated no change.

3.3. Spatial Autocorrelation Analysis

Spatial autocorrelation is a collective term for a set of statistical measures that describe the spatial dependence or clustering of geographic variables [57]. The basic idea is to determine the spatial correlation of a geographic phenomenon by calculating its similarity across space.

Global Moran’s I is a statistical measure that can be used to describe the spatial correlation of geographic phenomena at a global scale. It evaluates the spatial clustering of EEQ by comparing the RSEI values of a particular spatial unit with those of its neighboring units [22,58]. It ranges from −1 to 1.

$$GlobalMora{n}^{\prime }{s}_{}I=\frac{N{\sum }_i{\sum }_j{w}_{ij}(x_i-\mu )(x_j-\mu )}{({\sum }_i{\sum }_j{w}_{ij}){\sum }_i{(x_i-\mu )}^2}$$

(9)

where $x_i$ and $x_j$ represent the RSEI values of grid i and grid j, respectively. $\mu$ represents the average RSEI value, and $w_{ij}$ represents the weight.

Local Moran’s I (LISA) is a statistical measure that can be used to describe the spatial correlation of geographic phenomena at a regional scale. It further reveals the local spatial clustering patterns within a study area by measuring the difference in RSEI values between a specific unit and its neighboring units [59]. It helps identify regions with significant spatial clustering characteristics. The values of LISA also range from −1 to 1.

$$LocalMora{n}^{\prime }{s}_{}I=\frac{x_i-\mu }{{\sum }_i{(x_i-\mu )}^2}\sum _j{w}_{ij}(x_j-\mu )$$

(10)

where $x_i$, $x_j$, $\mu$ and $w_{ij}$ are the same as those defined for the Global Moran’s I formula.

The LISA cluster map exhibits five local spatial clustering types, namely, high–high (H–H), low–low (L–L), low–high (L–H), high–low (H–L) and nonsignificant. H–H indicates the presence of significant high-value clusters, where both the selected region and its neighboring areas exhibit high EEQ. L–L suggests the presence of significant low-value clusters, where both the selected region and its neighboring areas have low EEQs. L–H indicates a scenario where the selected region has a low EEQ while its neighboring areas have a high EEQ, indicating the existence of dispersed regions. H–L indicates a scenario where the selected region has a high EEQ while its neighboring areas have a low EEQ, also representing the presence of dispersed regions. Nonsignificance indicates the absence of a significant spatial clustering pattern, as the local Moran’s index values are not statistically significant.

3.4. Geodetector

Geodetector is a statistical tool used to analyze the spatial correlation and factors influencing geographic phenomena [60]. It is based on the concept of spatial heterogeneity, which recognizes that different locations are influenced by distinct natural and human factors. The software is available free of charge (https://www.geodetector.org/ (accessed on 25 December 2022)). We used factor detection and cross-detection methods to explore the impacts of each factor and the pairwise cross-factors on the EEQ. The Geodetector calculation formula is as follows:

$$q=1-\frac{1}{N{\sigma }^2}\sum _{h=1}^L{N}_h{\sigma }_h^2$$

(11)

In this context, the “q-value” is a metric used to explain the strength of the interaction between EEQ and influencing factors. A higher q-value indicates that the EEQ is more sensitive to a particular factor. By analyzing and comparing the q-values of different influencing factors, we can identify which factors have greater impacts on the variation in EEQ and further investigate their interrelationships. The significance of the remaining parameters can be seen in

Table A2. Meaning of the parameters.

Section	Parameters	Meaning
3.2. Pixel-Level Dynamic Detection	$\beta$	Theil–Sen median
	$RSE{I}_j$ and $RSE{I}_i$	Corresponding RSEI values for years j and i
	Z	Significance statistics for RSEI trends
	S	The test statistic for the Mann–Kendall test
	$Var$	Variance
	$sgn$	Symbolic function
	n	Number of time series
3.3. Spatial Autocorrelation Analysis	$x_i$ and $x_j$	RSEI values for raster i and raster j
	$\mu$	Average value of RSEI
	$w_{ij}$	Spatial weights between raster i and raster j
	N	Total number of grids
3.4. Geodetector	q	Explanatory power
	L	Hierarchy of variables
	$N_h$ and N	Number of cells in layer h and the whole area, respectively
	${{\sigma }_h}^2$ and ${\sigma }^2$	Variance in the variables in layer h and the whole area, respectively

.

The specific methodology was as follows. First, the q-values were calculated for each factor, and then the q-values were calculated for pairwise cross-factors. Finally, the q-values of individual factors and those of cross-factors were compared based on the cross-type table (

Table A3. Types of interactions (adapted from [24,60]).

Types of the Interactions	Basis for Judgment
Nonlinear weakening	$q(X_1\cap X_2)<min[q\left(X_1\right),q\left(X_2\right)]$
Single linear weakening	$min[q\left(X_1\right),q\left(X_2\right)]<q(X_1\cap X_2)<max[q\left(X_1\right),q\left(X_2\right)]$
Two-factor enhancement	$q(X_1\cap X_2)>max[q\left(X_1\right),q\left(X_2\right)]$
Mutual independence	$q(X_1\cap X_2)=q\left(X_1\right)+q\left(X_2\right)$
Nonlinear enhancement	$q(X1\cap X2)>q\left(X1\right)+q\left(X2\right)$

).

The specific implementation process was as follows: (1) The factor data were collected and preprocessed; (2) The data were discretized; (3) Fishnet points were created, and all factor indicator data values from the corresponding points were extracted; (4) This information was imported into Geodetector, and the program was run.

4. Results

4.1. PCA Results

In this study, the PCA module of the GEE platform was used to calculate the RSEI values for the region along the JBHSR from 2000 to 2020. The PCA results are presented in

Table 3. Results of PCA of four components.

Year		NDVI	LST	WET	NDBSI	ECR
2000	PC1	0.637	−0.595	0.014	−0.491	71.32%
	PC2	0.523	0.238	−0.731	0.368	21.64%
	PC3	0.295	0.766	0.189	−0.539	5.12%
	PC4	0.484	0.058	0.655	0.577	1.92%
2005	PC1	0.532	−0.569	0.096	−0.619	71.85%
	PC2	0.597	0.051	−0.718	0.355	21.35%
	PC3	0.292	0.818	0.058	−0.492	5.04%
	PC4	0.524	0.064	0.687	0.499	1.75%
2010	PC1	0.532	−0.563	0.134	−0.619	71.39%
	PC2	0.684	0.114	−0.632	0.347	21.36%
	PC3	0.213	0.812	0.082	−0.537	5.03%
	PC4	0.452	0.106	0.759	0.457	2.21%
2015	PC1	0.488	−0.598	0.172	−0.612	82.01%
	PC2	0.678	0.057	−0.670	0.297	12.70%
	PC3	0.248	0.791	0.073	−0.555	3.98%
	PC4	0.490	0.116	0.719	0.480	1.31%
2020	PC1	0.468	−0.573	0.206	−0.641	78.91%
	PC2	0.608	−0.084	−0.737	0.283	14.86%
	PC3	0.334	0.810	−0.002	−0.481	4.99%
	PC4	0.547	0.089	0.644	0.527	1.23%

(Note: ECR indicates the contribution rate of each principal component to the original data.)

and indicate the following: (1) The contribution rate of PC1 for this region during each of the five periods was consistently above 70%, with PC1 even reaching a contribution rate of over 80% in 2015. The average contribution rate of PC1 for the five periods was 75.10%. Thus, PC1 contains the majority of the information, making it suitable for reflecting the overall EEQ of the region; (2) The values of the PCA results show that, in PC1, both NDVI and WET have positive values, indicating their positive effects on the ecological environment. Conversely, the NDBSI and LST have negative values, indicating their negative impact on the ecological environment. This observation is consistent with reality.

4.2. Spatiotemporal EEQ Variability Analysis

4.2.1. Spatiotemporal Variations in Regional EEQ

Figure 3 shows the spatiotemporal variations in the EEQ along the JBHSR from 2000 to 2020. This indicates that the overall EEQ was at a moderate level. Additionally, there were significant spatial differences in the EEQ within the region, with the Jakarta area exhibiting the lowest quality.

Combined with

Table 4. Percentage changes in EEQ at different levels from 2000–2020.

Year	Poor	Fair	Moderate	Good	Excellent
2000	4.15%	20.10%	38.76%	28.61%	8.37%
2005	5.58%	12.51%	41.20%	29.88%	10.80%
2010	6.21%	10.01%	34.54%	35.94%	13.30%
2015	11.01%	18.97%	37.64%	24.24%	8.14%
2020	9.94%	13.76%	32.65%	31.96%	11.67%

, it can be seen that, from 2000 to 2020, the EEQ level showed an “N”-shaped trend of improving, followed by worsening and then improving again. The overall EEQ level was the lowest in 2015.

The most significant changes in the EEQ level percentages corresponded to the fair and moderate levels, at 6.34% and 6.12%, respectively. The proportions of the ’good’ and ’excellent’ levels both increased between 2000 and 2020, with the summed increases of these two levels being 6.65%.

Spatial analysis reveals the following findings: (1) Areas with poor EEQ were primarily concentrated in the four major metropolitan areas of Jakarta, Bandung, Bekasi and Cimahi, with Jakarta being the most affected. In 2000, the poorly performing areas were predominantly in the central regions of these cities. For Jakarta specifically, nearly half of the area was classified as having poor ecological quality. From 2000 to 2020, the distribution of areas with poor quality gradually expanded toward the outskirts, with approximately 90% of Jakarta’s area falling into the poor category by 2020. Similar trends in the evolution of urban EEQ were observed in the other three cities. Notably, this expansion trend gradually affected neighboring counties as well. For instance, in adjacent counties such as Bekasi County, scattered areas with poor quality started appearing in the central regions in approximately 2005, and, by 2015, clearly visible clusters of poor quality had formed.

Analyzing the reasons for this variation, these areas were mainly influenced by LST and NDBSI. Previous studies [39,41] on the region have shown that areas with poorer EEQ face higher risks of heat stress, as evidenced by the increasing surface temperature. The impact of surface temperature has also expanded over time, with the highest recorded surface temperature reaching 47 °C in 2019. This indicates that LST has a significant influence on the EEQ. Further analysis reveals that population plays a role in these changes. Over the past 20 years, the rapid economic growth in these areas has attracted a large influx of population, leading to increased concentration of people, which in turn contributes to rising regional temperatures. The increased population also leads to higher demand for vehicles and housing, resulting in traffic congestion, increased impervious surfaces and elevated carbon dioxide levels, further exacerbating regional temperatures. Additionally, rising temperatures enhance vegetation evaporation and increase aridity in urban areas, exacerbating the urban heat island effect and deteriorating the urban EEQ.

(2) Regions with relatively good EEQ are mainly distributed in the southern part of Bandung County and the northeastern region of West Bandung County. These areas are characterized by high vegetation coverage and higher elevations, primarily consisting of forests and grasslands, with forests being the dominant type—specifically, evergreen broad-leaved forests—and grasslands being multitree grasslands. These areas have minimal human habitation and are subjected to fewer human activities, resulting in better EEQ.

4.2.2. Spatiotemporal EEQ Variation Analysis Based on Subdistricts

The EEQ raster data were spatially linked with subdistrict vector data to calculate the average EEQ of 208 subdistricts, as shown in Figure 4.

The calculation results (

Table A4. Number of subdistricts with various EEQ levels.

Year	Poor	Fair	Moderate	Good	Excellent
2000	36	62	76	32	2
2005	45	49	78	32	4
2010	44	45	65	48	6
2015	65	52	63	26	2
2020	59	46	66	34	3

and

Table A5. Number of subdistricts at each EEQ level based on the city/county.

City/County	Year	Poor	Fair	Moderate	Good	Excellent
Jakarta	2000	28	17	0	0	0
	2005	32	13	0	0	0
	2010	31	14	0	0	0
	2015	39	6	0	0	0
	2020	34	11	0	0	0
Bandung	2000	8	19	2	1	0
	2005	11	16	2	1	0
	2010	11	15	3	1	0
	2015	17	10	2	1	0
	2020	17	10	2	1	0
Bekasi	2000	0	8	4	0	0
	2005	2	6	4	0	0
	2010	2	8	2	0	0
	2015	8	4	0	0	0
	2020	8	4	0	0	0
Cimahi	2000	0	2	1	0	0
	2005	0	2	1	0	0
	2010	0	2	1	0	0
	2015	0	3	0	0	0
	2020	0	3	0	0	0
Bekasi county	2000	0	6	17	0	0
	2005	0	4	17	2	0
	2010	0	4	17	2	0
	2015	1	15	7	0	0
	2020	0	10	12	1	0
Karawang	2000	0	7	22	1	0
	2005	0	3	26	1	0
	2010	0	0	27	3	0
	2015	0	8	21	1	0
	2020	0	2	27	1	0
Purwakarta	2000	0	0	10	8	0
	2005	0	1	9	8	0
	2010	0	0	4	14	0
	2015	0	2	10	6	0
	2020	0	1	7	10	0
Bandung county	2000	0	3	15	11	2
	2005	0	2	14	11	4
	2010	0	2	9	16	4
	2015	0	4	17	8	2
	2020	0	5	13	10	3
West Bandung	2000	0	0	5	11	0
	2005	0	0	5	11	0
	2010	0	0	2	12	2
	2015	0	0	6	10	0
	2020	0	0	6	10	0

) show that, over the past 20 years, the majority of subdistricts have a moderate level of EEQ, while the number of subdistricts with excellent levels is the lowest. Notably, the number of subdistricts that have a poor level has been increasing; they are primarily concentrated in the four cities, including Jakarta (Figure 4). By 2015, most subdistricts in these four cities were at a poor level. These regions have experienced high levels of development, attracting a large population, which has resulted in urban traffic congestion, environmental pollution and other issues. Although some measures have been taken to improve the EEQ, such as tree planting along roadsides, the overall efforts have been insufficient to mitigate the ecological damage caused by the large population. Of particular concern is Jakarta, where all regions have been below the moderate level over the past 20 years, and the number of administrative units classified as poor has shown an overall increasing trend. This elucidates the urgent need for Jakarta to receive more attention and take effective measures to improve the EEQ of the city.

4.3. EEQ Trend Analysis at a Pixel-Scale

Using MATLAB, trend analysis results were obtained (Figure 5). Obvious spatial differences can be observed in the RSEI change trends within different regions. Most of the regions experienced slight EEQ changes from 2000 to 2020, with approximately 23.3% of the regions experiencing mild degradation and 34.01% of the regions experiencing mild improvement (

Table A6. Statistics on the percentages of different trend variations.

Trend Feature	Percentage
Very serious degeneration	11.80%
Serious degeneration	4.84%
Moderate degeneration	2.50%
Light degeneration	23.30%
No change	3.57%
Light improvement	34.01%
Moderate improvement	4.95%
Strong improvement	8.33%
Very strong improvement	6.70%

).

Figure 5 indicates that the spatial distribution of degraded areas mainly radiates from the central regions of each county, forming concentric circles. The degree of degradation is most severe in the central regions, gradually weakening as the distance from the center increases. The reason for this pattern is that the central regions of each city and county have experienced rapid urbanization over the past 20 years, leading to the conversion of agricultural and grassland areas into built-up land. Consequently, surface temperature has increased, vegetation coverage has decreased, aridity has intensified, humidity has decreased and the EEQ has continuously degraded. The regions with severe EEQ degradation are the most critical areas that require attention. These areas are mainly located in the suburbs of Jakarta and the suburbs of Bandung, Bekasi and Cimahi. To analyze the degradation types in these regions in detail, two typical cities (Jakarta and Bandung) within the study area were selected. The change trend maps of EEQ over the past 20 years at a 30 m resolution were plotted for these cities (Figure A1). In the EEQ trend maps at 30 m resolution, four severely degraded small areas were selected. Combined with the remote sensing images of these areas in 2000 and 2020 (Figure A2), it was observed that these regions experienced an occupation of green spaces by construction land over the past 20 years. This observation is consistent with previous studies [61].

The areas of ecological improvement are located mainly in areas away from the centers of the counties. The types of land in these areas are mainly arable land, grasslands and woodlands, with an overall high level of vegetation cover. Notably, in the majority of northern Karawang County, the ecological improvement was more significant. In addition to the influence of vegetation cover, farmers in this area have exploited the unique geographical location to open up many fish ponds and salt pans, which have also contributed to the ecological improvement in this region. For Jakarta and Bandung, the overall EEQ (Figure 3) has remained poor over the past 20 years. However, the EEQ in the city center areas is gradually improving. This improvement is primarily due to these cities having relatively high economic levels and implementing measures to protect the urban ecological environment, such as increasing greening in public areas (e.g., parks) and regular watering activities throughout the city.

4.4. Comparative Analysis of the EEQ Results

Since no studies have been conducted to assess the EEQ specifically around the JBHSR, the results were compared with those of previous large-scale studies in terms of the general spatial distribution pattern: (1) A comparison with the regional distribution in EEQ in Southeast Asian countries conducted by Liao et al. [62] indicated that the EEQ in the Jakarta region was generally at a moderate level or below, while the surrounding areas exhibited significantly better EEQ; (2) Hong et al. [61] depicted the spatiotemporal distribution in the EEQ in the Pan-Third Pole region for 2000 and 2020, uncovering changes over the past 20 years. The EEQ in Jakarta was relatively moderate compared to the overall Pan-Third Pole region but lower than other regions in Indonesia.

In summary, previous studies indicated that the EEQ in the Jakarta region is significantly lower than that in its neighbors in terms of the overall spatial distribution of its dynamics. The research results within this paper (Figure 3) also show that the EEQ in the Jakarta region is the worst. Therefore, although there may be some differences in the evaluation results due to differences in data sources and assessment methods, the overall spatial dynamics depicted in this study align well with previous research findings. Hence, the evaluation results presented in this study are considered reliable.

4.5. Spatial Autocorrelation Analysis of the EEQ Results

Using ArcGIS, grid cells of 1 km × 1 km were used as sampling units, and a total of 9439 grid points were extracted for each year. Subsequently, using GeoDa, Moran scatter plots (Figure 6) and LISA cluster maps (Figure 7) were generated.

Based on the global spatial autocorrelation in the EEQ, as shown in the Moran scatter plots, the Moran’s I values for the five years were 0.947, 0.946, 0.944, 0.948 and 0.942. These indicate a significant positive spatial correlation and clustering pattern in the EEQ in the JBHSR corridor. The Moran’s I values for all five years were relatively high (above 0.94), indicating a consistently strong spatial autocorrelation in the region. Additionally, the values exhibited some fluctuation, which is consistent with the dynamic nature of the EEQ.

To further understand the spatial characteristics of the EEQ in the JBHSR corridor, a LISA cluster map was used to analyze the local spatial patterns. The LISA cluster map reveals that the overall spatial clustering pattern in the study area was characterized by H–H and L–L clusters (positive correlation pattern), consistent with the results from the Moran scatter plot. The H–H clusters were primarily in small areas in the southern and central parts of the study area, characterized by higher elevations, mountainous terrain and extensive forest coverage. These areas experience less human activity and exhibit a stronger capacity for self-regulation and protection. The L–L regions were mainly in the major urban areas, such as Jakarta, Bandung, Bekasi and Cimahi, as well as the main urban areas of Bekasi County and Karawang. These regions are characterized by significant commercial development and population concentration. However, they suffer from severe traffic congestion and high vehicle volume, leading to increased carbon dioxide emissions and significant ecological degradation. Additionally, the presence of extensive impervious surfaces within cities further exacerbates the urban heat island effect, negatively impacting the ecological environment. Notably, over time, areas with poor EEQ showed an expansion trend toward the outskirts, which aligns with the direction of urban expansion in the region, indirectly indicating the influence of urban development on the regional ecological environment. Therefore, in future urban construction and development processes, attention should also be given to the protection of the ecological environment. Nonsignificant regions were mainly distributed in rural areas of each county, characterized by lower levels of development and sparse populations. Farmers in these areas cultivate different crops, fruit orchards, tea gardens and salt fields based on the characteristics of different terrains and landscapes, resulting in strong spatial heterogeneity and no significant clustering effects.

4.6. Geodetector Results Analysis

Combining previous research and data availability, nine indicator factors were selected for the exploratory analysis of drivers: the population count (X1), economic level (X2), distance from roads (X3), distance from water bodies (X4), distance from parks (X5), elevation (X6), slope (X7), distance from industrial areas (X8) and distance from community centers (X9). A fishnet point with a size of 1 km × 1 km was created in ArcGIS, and the values of the nine indicators were extracted at the corresponding points. These data were then imported into Geodetector. Here, due to data limitations, only one year of geodetection results was produced, and the results corresponding to 2020 were obtained.

The results from the Geodetector analysis indicated that the factor with the highest q-value was elevation (X6), with a value of 0.339, indicating that elevation is the most significant influencing factor for the spatial distribution in EEQ (

Table 5. Detection results for each factor.

Factors	X1	X2	X3	X4	X5	X6	X7	X8	X9
q	0.0591	0.1949	0.0605	0.0849	0.1462	0.3390	0.1111	0.2063	0.2126
p	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000	0.0000

). The distance to the community center and the distance to the industrial area were the next most influential factors, with mean q-values of 0.2126 and 0.2063, respectively, both exceeding 0.2. The former suggests that buildings and human activities also have a significant impact on EEQ. The latter indicates that industrial pollution has a noticeable effect on the EEQ and measures should be taken by the government to reduce its impact, such as treating industrial waste gases and wastewater before discharge. In comparison, the factors of population count, distance to roads and distance to water bodies had relatively minor impacts, with q-values below 0.1.

The factor interaction results (

Table 6. Interaction detection influence degrees of the analysed factors.

Factor	X1	X2	X3	X4	X5	X6	X7	X8	X9
X1	0.0591
X2	$0.2410\#$	0.1949
X3	$0.1116\#$	$0.2347\#$	0.0605
X4	$0.1512*$	$0.2855*$	$0.1469*$	0.0849
X5	$0.1992\#$	$0.3305\#$	$0.1971\#$	$0.2251\#$	0.1462
X6	$0.3545\#$	$0.4002\#$	$0.3913\#$	$0.4488*$	$0.5115*$	0.3390
X7	$0.1582\#$	$0.2513\#$	$0.1665\#$	$0.1987*$	$0.2512\#$	$0.3527\#$	0.1111
X8	$0.2485\#$	$0.3245\#$	$0.2470\#$	$0.2586\#$	$0.3168\#$	$0.4855\#$	$0.2806\#$	0.2063
X9	$0.2636\#$	$0.3672\#$	$0.2466\#$	$0.2994*$	$0.2800\#$	$0.5160\#$	$0.2908\#$	$0.3465\#$	0.2126

(The symbols # and * denote two-factor enhancement and nonlinear enhancement, respectively.)

) illustrate that the interactions between the EEQ drivers all showed enhancing effects. Among them, 80.56% showed two-factor enhancement effects, 19.44% showed nonlinear enhancement effects and no independent factors were obtained, indicating that the interaction between any two drivers had a greater influence on the spatial distribution in the EEQ than any single factor alone.

The most significant interactions were X2 ∩ X6, X4 ∩ X6, X5 ∩ X6, X6 ∩ X8 and X6 ∩ X9, with q-values of 0.4002, 0.4488, 0.5115, 0.4855 and 0.5160, respectively; all of these resulted from interactions between elevation and other detection factors, further indicating that elevation influences the regional EEQ to a large extent. Notably, even the factors with weak explanatory power alone (X1, X3 and X4) showed significant increases in their cross-explanatory power when analyzed together with other factors, indicating that these factors indirectly influence the spatial differentiation pattern in the EEQ results.

5. Discussion

In this study, we utilized the powerful capabilities of the GEE cloud platform to construct an RSEI for the JBHSR corridor. Based on this index system, a comprehensive evaluation and analysis of the EEQ status along the JBHSR was conducted, revealing the trends in and characteristics of EEQ changes over the past 20 years. This is the first study to evaluate the EEQ for the JBHSR corridor and the first long-term EEQ-monitoring study in this region. The research findings not only provide scientific data support and methodological guidance for ecological protection, but also offer important references and a basis for the optimal allocation of ecological resources in this region. This study has significant theoretical and practical implications for promoting sustainable development in the JBHSR corridor.

Previous studies have mainly focused on administrative units (cities) or specific features (such as watersheds and wetlands) as research units. However, few studies have been conducted on complex terrain types, particularly in economic belts along major engineering projects. These types of projects often play a significant role in promoting local urban development and economic growth. However, the areas they traverse often exhibit significant disparities in urban development, uneven population distribution, variations in topography and complex geological structures. We selected the JBHSR project in Indonesia and conducted long-term, dynamic monitoring of the EEQ in its economic belt. The spatial distribution and spatiotemporal variation characteristics of the EEQ results in the four major cities, including Jakarta and Bandung, were consistent with the research results of Zhang et al. [22], Geng et al. [26] and Chen et al. [27]. The results show that the EEQ is worse in the town centers and shows a trend of gradually improving as the distance from the town centers increases toward the periphery. In addition, the distribution trend in areas with relatively poor EEQ was consistent with the trend in urban expansion. The distribution in areas with better ecological quality agreed well with the findings of Yuan et al. [23] and Zhou et al. [24], showing that the areas with relatively good EEQ are located mainly in areas covered by woodlands and grasslands, as these areas are less disturbed by human activities.

Land use has been proven in several previous studies to be an important factor influencing the EEQ. For example, Zhou et al. [24] considered land use as the first driver of the EEQ (obtaining a corresponding q-value of 0.594) and conducted an exploration of the relationship between EEQ changes and land use changes. Yuan et al. [23] further explored the relationships between different EEQ levels and 14 different land use types while considering the special importance of land use factors. Most of these past studies proved that the EEQ of built-up land-type areas in relatively populated urban areas is relatively poor, while the EEQ of woodlands–grasslands—land use types that are less impacted by human activities—is better [22,23,24,26,27]; this is an accepted general fact. Therefore, in terms of drivers, the impact of land use type on ecosystem quality is not reiterated in this paper. Furthermore, more detailed indicators, such as the distance to parks, distance to community centers, distance to roads, distance to water bodies and distance to industrial areas, were analyzed in this work. These indicators were also found to significantly impact the EEQ [26] but were not explored in previous studies.

In this study, both the single-factor detection results and the cross-factor detection results showed that elevation is the most important factor affecting the EEQ of the region. However, in previous studies, the explanatory power of the elevation factor on EEQ was relatively weak. For example, in the Taihu Lake region [24], elevation ranked fourth in terms of explanatory power among seven factors; in the Chishui River region [33], elevation also ranked fourth. Furthermore, population density has consistently been the most significant influencing factor in previous studies. For example, in the same studies as those mentioned above, population density was the first influencing factor for EEQ in the Chishui Basin [33] and the second influencing factor for EEQ in the Taihu Basin [24]. In addition, studies on the EEQ in the Three Gorges Ecological Corridor region have noted that population density is always one of the top three influencing factors in different years [21]. However, in the research findings of this study, the influence of population density on EEQ is relatively weak. For different study areas, the main influencing factors on the EEQ vary greatly due to the great differences in their topography and population distribution. In addition to population and elevation, wind speed, solar radiation and distance from industrial areas are also likely to be the main factors influencing the EEQ in different regions. Therefore, clarifying the different driving factors of EEQ in different regions is an important guide to the rational formulation of regional ecological environmental improvement and protection policies.

This study used the GEE platform to construct and process the RESI over a long temporal scale and large spatial extent. Direct access to the remote sensing data repository of GEE, in which cloud computing architecture can be leveraged and open algorithmic tools can be utilized, enables efficient, rapid and large-scale processing and analysis of remote sensing data. Compared to traditional remote sensing software, this approach saves a significant amount of time and cost. It provides a novel approach for processing long-term and wide-area remote sensing data.

However, there are some limitations in this study. (1) Regarding the selection of indicators, this study considered only four subindicators to calculate the RSEI value. Multiple studies have indicated that these four indicators can effectively reflect the regional EEQ to some extent; however, importantly, the EEQ is not only related to these four factors, but is also closely associated with factors such as air pollution and carbon emissions. Therefore, in future research, additional data related to these factors will be incorporated into the indicator construction to better reflect and describe the regional EEQ. (2) Regarding the comparison and validation of research results, due to the lack of previous studies in this specific area, this study provides only a qualitative comparison of the overall EEQ distribution trends within the study area with broader research conducted by previous researchers. In future research, more diverse data from multiple sources will be collected to further validate the research findings in this region. (3) The exploration of driving factors in this study is also limited. The OpenStreetMap platform cannot provide historical data for very long periods. Therefore, this study explored only the driving factors of EEQ for 2020. In the future, we will consider using high-resolution remote sensing images and deep learning algorithms to extract historical road data and water body data, enabling exploration of changes in dominant driving factors over long-term time series. Additionally, once data on other indicators such as urbanization rate and the status of primary, secondary and tertiary industries become available (or data on other potential factors affecting the ecological environment), we will incorporate them to update the research and further explore the more detailed and diversified influencing factors and mechanisms of the ecological environment.

The current research was based on MODIS data with a resolution of 1 km. The construction of the high-speed rail will undoubtedly bring significant economic and population growth to the areas along the railway, especially near high-speed rail stations. Meanwhile, studies have shown that human activities have a significant negative impact on EEQ. Therefore, in the future, we will conduct long-term monitoring of the EEQ within a 5 km radius around the stations. However, the current 1 km resolution data will not be sufficient; more detailed data will be needed. Landsat data can meet the requirements for fine-scale analysis. However, the Landsat images in the research area are heavily affected by clouds, which affect the quality of the inversion results. In the future, we will consider using algorithms such as downscaling to construct more refined indicator data. This will not only improve the accuracy of EEQ assessment but also enable the monitoring of dynamic changes in the EEQ around the stations over longer time series. After the completion of the high-speed rail project, we will continue to monitor and evaluate the EEQ in the region. This will not only provide a scientific basis for the sustainable development strategies of the JBHSR economic belt, but also serve as an important research reference for other countries and regions, particularly in Southeast Asia, with similar geographical environments, regarding the changes in the surrounding ecological environment caused by high-speed rail development. This holds significant importance for sustainable development.

Human beings are an important component of the Earth’s ecosystem. However, in the context of global climate warming and continuous deterioration of the ecological environment, human health is facing increasing threats. Most existing studies have focused on the impact of air pollutants on human health, the influence of soil heavy metal ions on human health and the effects of water pollution on human health. A clear analysis regarding the relationship between EEQ and human health is still lacking. Therefore, in future research, it will be necessary to integrate multiple influencing factors and explore the quantitative relationship between the ecological environment and overall human health impact. This will provide scientific data support for the coordinated development and protection of resources and the environment, as well as human sustainable development.

6. Conclusions and Suggestions

Based on GEE and the RSEI, this study monitored the spatiotemporal changes in EEQ along the JBHSR for the past 20 years and explored its spatial differentiation characteristics and influencing factors. The results show the following:

(1)

The overall EEQ along the JBHSR is moderate, but there are significant spatial variations. Temporally, the EEQ shows an ”N”-shaped fluctuation, reaching its lowest point in 2015 and then gradually recovering. Spatially, areas with better EEQ are primarily concentrated in mountainous and forested regions with higher elevations and vegetation coverage, while areas with poorer EEQ are found mainly in the four major cities, such as Jakarta (which has a highly concentrated population and a developed economy), and in the urban areas of the districts;

(2)

The EEQ was slightly altered in most areas from 2000–2020, with approximately 23.3% of the areas experiencing mild degradation and approximately 34.01% of the areas seeing mild improvement. Approximately 20% of the areas showed a very strong EEQ alteration;

(3)

Altitude, distance from a community center and distance from an industrial area were the top-three most important influencing factors. Altitude had a positive effect on EEQ. The concentration of human groups and pollution from industry had a significant negative effect on the EEQ.

Based on the multiangle analysis, the following suggestions are proposed to achieve sustainable development of the ecological environment along the JBHSR corridor: (1) Public awareness and consciousness about ecological environmental protection should be raised. The development, utilization and conservation of resources in the JBHSR region should be regulated; (2) Businesses should be encouraged to adopt sustainable production methods and strengthen the governance of pollution sources, including the treatment of industrial wastewater and emissions. Strict standards should be enforced to ensure safe discharge; (3) Ecological restoration and conservation efforts along the railway route should be enhanced. Measures such as afforestation, wetland restoration and biodiversity conservation should be implemented to mitigate the environmental impact and promote ecological balance; (4) It is suggested that the counties, in their future development, should first optimize the structure of their towns and cities, which could be designed as multiple central regional towns, to mitigate the threat to the EEQ from human aggregation; (5) The area around high-speed railway stations will support high levels of economic development in the future. Therefore, reasonable planning should be carried out in advance for the areas around the four high-speed railway stations, to build a blueprint for the synergistic and sustainable development of regional construction, economic development and ecological protection.