Download as pdf or txt
Download as pdf or txt
You are on page 1of 14

Ecological Informatics 81 (2024) 102635

Contents lists available at ScienceDirect

Ecological Informatics
journal homepage: www.elsevier.com/locate/ecolinf

Construction of landscape eco-geological risk assessment framework in coal


mining area using multi-source remote sensing data
Xiaoya Zhu a, b, Peixian Li a, c, *, Bing Wang a, Sihai Zhao a, Tao Zhang a, Qingyue Yao a
a
College of Geoscience and Surveying Engineering, China University of Mining and Technology - Beijing, Beijing 100083, China
b
Key Laboratory of Mine Geological Hazards Mechanism and Control Ministry of Natural Resources, Xi’an 710054, China
c
Research Center of Eco-Geological Environment & Remote Sensing Big Data, Inner Mongolia Research Institute of China University of Mining and Technology-Beijing,
Ordos 010300, China

A R T I C L E I N F O A B S T R A C T

Keywords: High-intensity and large-scale mining activities have aggravated regional eco-geological risk. Therefore, it is
Landscape eco-geological risk significantly essential to conduct an assessment of the eco-geological risk of mining areas. Although some
Landscape loss index progress has been achieved in ecological risk assessment studies, existing approaches are not entirely suitable for
Multiscale geographically weighted regression
coal bases with high landscape fragmentation and dense coal mining activities. Here, we developed a novel
Shenfu mining area
landscape ecological and geological risk (LEGR) assessment framework based on theories that include landscape
ecological risk and eco-geological risk. The framework selected 10 indicators, including slope, fluctuation,
lithological hardness, soil type, FVC, RSEI, precipitation, biological abundance, distance to road and subsidence
rate, and calculated the weights of indicators by introducing the AHP-CRITIC coupled weighting model. Then,
the impact of landscape disturbances on eco-geological risk is quantified by measuring landscape losses. This
framework was applied to the Shenfu mining area (SFMA), a typical coal base in northwest China. The results
indicated the LEGR was moderate in the SFMA whose spatial distribution exhibited an increasing trend from
southwest to northeast. Besides, the high LEGR was mainly in the aggregated mining area with high subsidence.
For the eco-geological environment monitoring at the mine scale, a multiscale geographically weighted
regression (MGWR) model was utilized for analyzing the relationship between indicators and LEGR within the
disturbed range of coal mining. It provided valuable insights for the formulation of environmental protection
policies in the mining area.

1. Introduction Trofimov proposed the concept of ecological geology (Trofimov,


2008). Eco-geological risk (EGR) refers to the potential threats or harms
Coal is the main source of energy in China. Coal mining activities to ecosystems caused by geological processes resulting from human
have promoted the development of the local economy, but they have activities or natural factors. The EGR assessment can reflect the spatial
also inevitably caused a series of eco-geological environmental problems distribution of regional risk (Huang et al., 2022), and help relevant
(Zhu et al., 2022). These problems mainly include land subsidence, decision-makers to formulate regional eco-environmental protection
vegetation decline, and self-burning coal waste pile (Jin et al., 2021; measures or strategies for coordinating social development. The EGR
Zhang et al., 2023a). The eco-geological environment of mining areas in assessment needs to integrate the principles of multi-level and multi-
northwest China is affected by the dual pressures of inherent ecological perspective for designing a comprehensive assessment framework, and
vulnerability and the large-scale exploitation of coal resources (Li et al., the significant components include dividing assessment units, quanti­
2018b). Consequently, it is crucial for accurately obtaining detailed fying impact indicators, establishing an assessment model, and selecting
information about environmental disturbance to conduct an assessment a grading method. (Xu et al., 2023a) considered mineral development
on eco-geological risk of mining areas caused by mining activities. This activities as sources of ecological risk, characterized the ecological
will help decision-makers to implement new strategies for balancing pressure resulting from mining development through water conserva­
economic activities and ecological development. tion stress and ecosystem service functions, and established a

* Corresponding author at: College of Geoscience and Surveying Engineering, China University of Mining and Technology - Beijing, Beijing 100083, China.
E-mail address: lipx@cumtb.edu.cn (P. Li).

https://doi.org/10.1016/j.ecoinf.2024.102635
Received 15 January 2024; Received in revised form 6 May 2024; Accepted 6 May 2024
Available online 8 May 2024
1574-9541/© 2024 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
X. Zhu et al. Ecological Informatics 81 (2024) 102635

comprehensive evaluation framework for quantitative analysis of min­ development; (4) proposing an eco-geological risk monitoring plan for
ing area ecological risk. By analyzing the uncertainty of the mining mining areas based on the MGWR model.
technical parameters of 13,960 mines and the environmental detri­
mental extent caused by surrounding economic activities, (Dou et al., 2. Study area and data
2023) constructed an ecological risk framework suitable for the mine
scale, which helped to understand the degree of ecological threats from 2.1. Study area
mining activities. Despite the aforementioned studies have enriched the
theoretical system of EGR assessment in mining areas, its application The SFMA is located in the northern part of Yulin City, Shaanxi
only focused on weighting models and indicators selection, ignoring the Province, China (Fig. 1). It lies at the southern edge of the Mu Us Desert
impact of landscape spatial patterns on the eco-geological environment. and the northern boundary of the Loess Plateau, with an area of 6211.14
In the mining scenarios, the activities of underground mining and open- km2 (Liu and Li, 2019). The study area has a temperate semi-arid con­
pit mining destroy the original overburden structure, leading to the tinental monsoon climate, featuring an annual average precipitation
fragmentation of landscape type patches (Shang et al., 2022). In addi­ ranging from 396 mm to 650 mm and an average temperature between
tion, the self-burning coal waste pile can produce gas emissions and 6.6 and 8.5 ◦ C (Yang et al., 2019). The predominant soil types are
change the landscape structure, posing a serious threat to the sur­ aeolian sandy soil and loess soil, characterized by coarse texture and
rounding environment and human health (Teodoro et al., 2021). poor resistance to erosion. Multiple factors contribute to the vulnera­
Therefore, we incorporated the impact of landscape patch discretization bility of the eco-geological environment in this region, leading to serious
into the eco-geological risk framework. Landscape ecological risk is the land desertification and soil erosion. The confirmed coal reserves in
adverse result due to interaction between landscape pattern and Shenmu City and Fugu County are 56 billion tons and 18.3 billion tons,
ecological process, which can display the spatial heterogeneity of respectively. Their coal reserves account for 43.13% of the total coal
regional ecological risk degree (Cao et al., 2019; Kim et al., 2024). The reserves in Shaanxi Province. (http://www.sxsm.gov.cn/). In 2020, the
loss degree and risk probability of landscape types were used to coal production in Shenmu City and Fugu County was 2.67 billion tons
construct the LEGR index to reflect the disturbance and vulnerability of and 0.91 billion tons, respectively (http://www.fg.gov.cn/). The type of
external threats to landscape pattern (Xu et al., 2021a). coal mining in the region is dominated by underground mining, with 49
The weighting methods for ecological risk assessment indicators underground mines concentrated in the northeast of the SFMA.
include Analytic Hierarchy Process (AHP) (Kucuker and Cedano Gir­
aldo, 2022), Principal Component Analysis (PCA) (Chen et al., 2022), 2.2. Assessment indicators
Entropy Weight Method (EWM) (Wang et al., 2024). Compared with the
single weighting method, the coupled subjective and objective weight­ The internal characteristics of the natural environment and external
ing model has complementary advantages, and the indicators weight disturbances from human activities lead to changes in the eco-geological
can be determined more reliably. The typical subjective weighting risk (Li et al., 2022b). The research objective was the assessment of
method AHP transforms complex problems into a systematic hierarchy LEGR in arid and semi-arid mining areas, which was decomposed into
and its calculation process is simple. In addition, the CRITIC method the background characteristics of the eco-geological environment and
takes into account the degree of variability of the indicators and the the sources of risk pressure. Among these, background characteristics
correlation between different indicators. Therefore, this study inte­ reflect the environmental and geological condition of SFMA. The risk
grated AHP and CRITIC methods to determine the comprehensive pressure refers to the disturbance of human activities. Based on field
weights of eco-geological risk indicators, reducing the subjectivity of surveys, the background characteristics consisted of four indicators
artificial weighting and improving the reliability of indicators weight. related to topographical and geological conditions (i.e., slope, fluctua­
Traditional EGR assessments mostly rely on field surveys and utilize tion, lithological hardness, and soil type) and three indicators related to
discrete data, making regional monitoring challenging. Multi-source ecological conditions (i.e., fractional vegetation cover (FVC), remote
remote sensing data has great potential in spatiotemporal ecological sensing ecological index (RSEI), precipitation). Risk pressure includes
quality assessment. At present, optical satellites provide a wealth of three indicators influenced by human activities (i.e., biological abun­
images for the retrieval of eco-environment parameters. These param­ dance, distance to road, and subsidence rate). The sources and de­
eters include remote sensing ecological indices, vegetation cover, and scriptions for selected indicators are listed in Table 1.
net primary productivity (Li et al., 2020b; Xu et al., 2023b; Zhang et al.,
2023b). (Guo et al., 2023) proposed a new framework for overcoming 2.3. Data processing
the uncertainty of multi-source remote sensing inversion of surface
feature parameters, adopted three objective methods to calculate Firstly, we screened and processed the Landsat 8 SR and Sentinel-2
weights of indicators of four ecological pressure dimensions and MSI images by the Google Earth Engine platform (GEE). The spectral
analyzed the influence of the spatial scale effect of remote sensing im­ indices were then calculated using Sentinel-2 MSI image to construct the
ages on the assessment results. Time-series images from the Sentinel-1 feature space. A classifier was constructed using a random forest algo­
SAR satellite can derive the status of ground subsidence over a certain rithm. Finally, we split the real samples into 70% for model training and
period, providing an opportunity for regional geological environmental 30% for model validation. The overall accuracy and Kappa coefficient of
monitoring. We calculated the subsidence rate of the SFMA in 2022 the SFMA classification results were 77.9% and 0.72, meeting the re­
using the SBAS-InSAR technique to reflect the influence of mining ac­ quirements for calculating biological abundance and landscape vulner­
tivities and groundwater engineering on regional geological conditions. ability indices. Additionally, the Dimidiate Pixel Model and Principal
The eco-geological environmental problems caused by coal mining in Component Analysis based on Landsat 8 SR images were used to
the fragile areas of northwest China are attracting the attention of calculate FVC and RSEI. All Sentinel-1 SAR images in 2022 were
decision-makers. Here, we constructed a landscape eco-geological risk collected in batches, and then the subsidence rate of SFMA was calcu­
assessment framework by integrating multi-source remote sensing data lated by integrating the ISCE-2 and the StaMAP software with SBAS-
and a subjective-objective coupled weighting model and applied it to the InSAR technology, and the gap value was filled by the Kriging Interpo­
SFMA. The research purposes are as follows: (1) constructing a land­ lation method. This result was in good agreement with the mining po­
scape eco-geological risk assessment framework; (2) calculating the sition of the working face and the regional distribution of groundwater
landscape eco-geological risk index (LEGRI) for the SFMA and validating subsidence for many years (Fan et al., 2018). In addition, the rest of the
its reliability; (3) analyzing the spatial distribution characteristics of data products were quantified via thematic maps with different spatial
LEGRI to provide auxiliary suggestions for regional coordinated resolutions. Finally, all the data were down scaled to 30 m by the

2
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Fig. 1. Overview of the study area.

bilinear interpolation method to unify the spatial resolution of the data. is the weight of the j th index.
The spatial distribution of eco-geological risk assessment indicators in The testing formulas are as follows:
the SFMA is shown in Fig. 2.
CI
CR = (2)
RI
3. Methodology
λmax − n
CI = (3)
The overview of the methodology is illustrated in Fig. 3. Specifically, n− 1
(Step1) The thematic data, SAR images, and multi-source remote sensing
images were integrated to quantify risk assessment indicators. (Step2) A where, CI and RI are the consistency index and average random con­
comprehensive weighting method that coupled the AHP and the CRITIC sistency index of the judgment matrix; λmax is the maximum eigenvalue
methods was used to calculate the weights of assessment indicators of of the consistency matrix; n is the number of compared factors.
the eco-geological risk index (EGRI). (Step3) The landscape loss index The other detailed implementation method of AHP can be found in
was calculated by using the land classification data, and then combined Appendix A.
it with eco-geological risk index to construct LEGRI. (Step4) Frequency
ratio and frequency ratio accuracy were utilized to verify the consis­ 3.1.2. CRITIC
tency and reliability of the evaluation results for EGRI and LEGRI. The CRITIC is an objective weighting method put forward by Dia­
(Step5) The Moran’s I index was employed to analyze the spatial dis­ koulaki (Diakoulaki et al., 1995). This method determines the weight
tribution and clustering characteristics of LEGRI. Additionally, the based on the information of indicators and the correlation between in­
MGWR model parameters were determined to design an eco-geological dicators (Yang et al., 2023c). The objective weights are calculated using
risk monitoring plan at the mine scale. Eqs. (4)–(6).
Cj
w2j = ∑
3.1. Eco-geological environment risk assessment m
(4)
Ck
k=1
3.1.1. AHP √̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
The AHP is a subjective weighting method based on expert scoring √∑
√n ( )2
√ x − xj
and widely applied in risk assessment (Kokangül et al., 2017). The √i=1 i,j (5)
comparison matrix was constructed according to the importance of in­ σj =
n− 1
dicators and then tested whether it passed the random consistency rate
(CR). If CR > 0.1, adjust the judgment matrix; If CR ≤ 0.1, calculate the ∑
m
( )
subjective weight by the eigenvectors corresponding to the maximum Cj = σj 1 − rjk (6)
eigenvalue of the matrix. The subjective weights are calculated using Eq.
k=1

(1). where, σj and Cj represent the standard deviation and information en­
vj tropy of indicator j, respectively. rjk denotes the correlation coefficient
w1j =∑ (1)
m
vk between indicators j and k. n and m are the sample size and the number
k=1 of indicators, respectively. xj and xk indicate the sample means of the j-
th and k-th indicators, respectively.
where, vj is the eigenvector corresponding to the j largest eigenvalue, w1j

3
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Table 1 formula is as follows:


The sources and descriptions of indicators.
LEGRI = 0.5R + 0.5EGRI (8)
Factor Indicator Description and data source

Slope has a significant influence on soil 3.2.2. Landscape loss index


erosion, vegetation distribution and
Slope The landscape loss index represents the extent of ecological loss for
ecosystem stability (Löbmann et al., 2020).
https://www.gscloud.cn/
each landscape type when subjected to disturbance. Referring to rele­
Topographic fluctuation reflects the vant research results, the key parameter of landscape sample area should
topographic fluctuation of the ground surface reach 2–5 times of corresponding the landscape patch area, ensuring the
Fluctuation and affects the occurrence of geological sample can reflect the landscape pattern information around the sam­
disasters (Oh and Pradhan, 2011). https
Geological pling point (O’Neill et al., 1996). The average patch area of the study
://www.gscloud.cn/
condition
The physical and chemical characteristics of area was 0.013 km2. Considering the issues of calculation intensity and
different lithological types affect the stability. accuracy, the landscape of the SFMA was divided into a cell grid of 250
Lithological
hardness
Geological hazard potential is closely related m × 250 m, with a total of 100,589 sampling areas. The landscape loss
to lithological hardness (Wang et al., 2021b).
index can be calculated by Eq. 9 and Eq. 10.
https://www.resdc.cn/
Soil types are closely related to soil erosion ∑N Aki
Soil type and land desertification (An et al., 2019; R= i=1 A
Ri (9)
k
Sharma et al., 2012). https://www.resdc.cn/
FVC is the proportion of green vegetation per
unit area, which can measure the vegetation Ri = Ei × Fi (10)
FVC
condition (Chen et al., 2017). https://code.
earthengine.google.com/ where, N is the total number of landscape types. Ak is the area of the
The RSEI couples four ecological indicators assessment unit; Aki is the area of landscape type i in evaluation unit k; Ri
Ecological via the PCA method to comprehensively refers to the loss index of landscape type i; Ei and Fi are the landscape
condition RSEI portray the regional ecological quality (Zheng
et al., 2022). https://code.earthengine.goo
disturbance index and landscape vulnerability index of type i;
gle.com/
Ei = aCi + bNi + cDi (11)
Extreme rainfall events may lead to soil
Precipitation erosion, floods and other natural disasters (
Zhao et al., 2018). https://engine.piesat.cn/ where, according to its importance: a = 0.5, b = 0.3 and c = 0.2 (Li et al.,
Biological abundance refers to the 2018a).
measurement of biodiversity in a certain Landscape fragmentation Ci , landscape separation Ni and landscape
Biological period in a specific ecosystem, which is dominance Di are calculated by the following formulas:
abundance usually expressed as the diversity and richness
of biological species (Nicholson et al., 2021). ni
https://code.earthengine.google.com/
Ci = (12)
Ai
The construction of the road interferes with
the original topography and destroys the rock √̅̅̅̅
A ni
Human factor Distance to road
mass structure. The closer it is to the road, the Ni = (13)
easier it is to induce landslides and collapses ( 2Ai A
Shi et al., 2018; Tesfa and Woldearegay,
2021). https://www.dsac.cn/ 1 (ni mi ) Ai
Di = + + (14)
The subsidence rate refers to the rate of land 4 N M 2A
subsidence per unit time caused by mining
activities (Tang et al., 2022). Rapid where, ni is the number of patches for landscape type i; Ai is the total
Subsidence rate
subsidence rate can trigger ground cracking
area of landscape type i; A is the total landscape area; N is the total
and building damage (Zhao et al., 2023).
https://search.asf.alaska.edu number of landscape patches; mi is the number of sample plots for
landscape type i; M is the total number of sample plots in the landscape.
In addition to the degree of loss of landscape structure, landscape
3.1.3. Comprehensive weighting model types have different vulnerability in response to external environmental
The comprehensive weighting method that combines subjective with change pressures, namely landscape ecological vulnerability (Fi ) (Xie
objective weights can make the evaluation results more reliable (Deng et al., 2013). Based on the landscape disturbance resistance character­
et al., 2021; Yin et al., 2020). The comprehensive weight was obtained istics of the SFMA, the vulnerability weights of each landscape type were
by the method of minimum discriminant information, which minimized determined by the AHP method (Table 3). The implementation method
the bias between AHP and CRITIC (Yang et al., 2022). The formula is as of the AHP method is the same as section 3.1.1. The judgment matrix is
follows: as follows.
√̅̅̅̅̅̅̅̅̅̅̅ ⎡ ⎤
w1j w2j Y1 Y2 Y3 Y4 Y5 Y6
wj = ∑ m √̅̅̅̅̅̅̅̅̅̅̅ (j = 1, 2, ……, m) (7) ⎢ Y1 1 1/2 1/3 2 1/7 1/4 ⎥
⎢ ⎥
w1k w2k ⎢ Y2 2 1 1/2 3 1/6 1/3 ⎥
⎢ ⎥
k=1
F− Y = ⎢ ⎢ Y3 3 2 1 4 1/5 1/2 ⎥ ⎥
⎢ Y4 1/2 1/3 1/4 1 1/8 1/5 ⎥
where, w1j and w2j refer to the weights of AHP and CRITIC respec­ ⎢ ⎥
⎣ Y5 7 6 5 8 1 3 ⎦
tively. The results of comprehensive weights are shown in Table 2.
Y6 4 3 2 5 1/3 1

3.2. Improved eco-geological risk assessment where, Y1-Y6 represent forest, grassland, cropland, building, bare,
water, and F refers to the vulnerability index of the landscape type.
3.2.1. Landscape eco-geological risk index
Based on the previous research findings and considering the rela­ 3.3. Accuracy assessment
tionship between landscape pattern and eco-geological risk, the LEGRI
was constructed by integrating the landscape loss index (R) and the eco- Although there is no standardized method for assessing the reliability
geological risk index (EGRI) (Gong et al., 2021; Wang et al., 2021a). The of LEGRI results, existing studies indicate that cross-validation between

4
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Fig. 2. Spatial distribution of eco-geological risk assessment indicators.

different evaluation methods and regional geological disaster data ( )


Ni
provide effective indirect ways to evaluate the quality of the LEGRI
assessment result (Muço et al., 2012; Zou et al., 2023). The results of N
FRi = ( ) (15)
Si
LEGRI classification were overlapped with spatial distribution data of
geological disasters. The results of EGRI and LEGRI were validated by S
using geological disasters frequency ratio (FR) and frequency ratio ac­

n
curacy (FRA). The FR of geological disasters represents the ratio of the FRi
frequency of geological disasters in a certain partition, while the FRA FRA = i=m
∑n (16)
can directly compare the accuracy and rationality of different assess­ FRi
ment results. The formulas are as follows:
i=1

where, Ni and Si are the number and area of disasters occurring in


the i − th risk grade, respectively; the greater i, the higher the risk level;
N represents the total number of disasters, and S refers the total area of
the study area.

5
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Fig. 3. Overview of methodology.

3.4. Spatial correlation analysis n ∑


∑ n ( )
n wij (xi − x) xj − x
Spatial autocorrelation is a technique that tests whether an element ′
Global Moran s I =
i=1 j=1
(17)
∑n ∑ n ∑
n
is correlated with its neighbors and determines spatial heterogeneity wij (xi − x)2
(Cai et al., 2023; Chen et al., 2023). The Global Moran’s I was used to i=1 j=1 i=1

measure the overall correlation and variation of ecological phenomena,


and the Local Moran’s I effectively reflected the spatial clustering dis­
tribution of the study area (Yang et al., 2023a), which are expressed as:

6
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Table 2 tures of the MGWR model and mining disturbance characteristics. The
Calculation results of the comprehensive weight. conceptual model diagram is illustrated in Fig. 4. Specifically, existing
Indicator AHP CRITIC Comprehensive weight studies suggest an exponential change in the environmental disturbance
caused by mining activities as one moves away from the mining area,
Fluctuation 0.0442 0.0413 0.0430
Slope 0.0694 0.0449 0.0596 with an approximate disturbance range of 1 km (Wang et al., 2023).
Soil type 0.0264 0.1482 0.0752 These characteristics were utilized to determine the kernel function and
Lithological hardness 0.0235 0.1540 0.0758 different bandwidths of the MGWR model. We set up a bandwidth
Precipitation 0.0485 0.0498 0.0491 scheme from 500 m to 1500 m with 100 m intervals. Then, the optimal
FVC 0.0882 0.0821 0.0857
RSEI 0.1602 0.1756 0.1664
bandwidth for each parameter was determined by the minimum cor­
Biological abundance 0.1725 0.0715 0.1320 rected Akaike Information Criterion (AICC). In addition, the kernel
Distance to road 0.0658 0.1138 0.0850 function employed was the exponential function. The model is formu­
Subsidence rate 0.3013 0.1188 0.2282 lated as follows:

m
yi = β0 (ui , vi ) + βbwj (ui , vi )xij + εi (19)
Table 3 j=1

Calculation results of the landscape loss-related indices.


where, (ui , vi ) represents the coordinates of sampling point i, β0 (ui , vi )
Land Use Types Ci Ni Di Ei Fi Ri
represents the intercept at location i, βbwj refers the adjustment of the j th
Forest 7.3252 9.7688 0.1306 6.6194 0.0566 0.3747
bandwidth, m is the number of sampling points, xij represents the value
Grassland 0.1359 0.2073 0.6773 0.2656 0.0876 0.0233
Cropland 2.0021 2.3421 0.2555 1.7548 0.1346 0.2362 of independent variable xj at point i, εi is the error term of the model.
Building 2.1970 4.6182 0.0824 2.5005 0.0378 0.0945
Bare 4.1535 4.0325 0.2609 3.3387 0.4736 1.5812
4. Result
Water 0.7904 4.6046 0.0186 1.7803 0.2098 0.3735

4.1. Correlation and collinearity analysis of LEGRI indicators



n ( )
(xi − x) wij xj − x
To test the level of correlation and degree of multicollinearity be­
(18)
j=1
Local Moran’ s I = ∑
n
2 tween the LEGRI assessment indicators, we calculated the Pearson cor­
(xi − x)
i=1 relations (Fig. 5) and variance inflation factor (VIF). The results show
that slope and fluctuation (r = 0.45), FVC and RSEI (r = 0.53) have a
where, xi and xj are the attribute values of spatial units i and j, x is the certain degree of correlation, but the correlations between other in­
mean of all attribute values of spatial units, wij is the spatial weight dicators are weak. In addition. The VIF values of slope, fluctuation, FVC
matrix, and n is the total number of evaluation units. and RSEI are 5.07, 4.08, 8.91 and 7.29, respectively, and the VIF values
of other indicators are also <10, which proves that there is no multi­
collinearity relationship between indicators (Cheng et al., 2022).
3.5. Construction of MGWR model at the mine scale
Therefore, the selected indicators are relatively independent factors
suitable for LEGRI assessment in the SFMA.
The mining process leads to varying degrees of subsidence on the
corresponding surface and its surroundings, triggering eco-geological
environmental problems within a certain range. Thus, when analyzing 4.2. Accuracy verification
the relationship between assessment indicators and eco-geological risk,
it is necessary to consider the disturbance of mining activities to sur­ The results of LEGRI and EGRI on all assessment units were statis­
rounding environment. The geographically weighted regression (GWR) tically analyzed (Fig. 6a), indicating a good spatial consistency between
model is a local parameter estimation technique that considers spatial the two models. The overall EGRI showed higher degree of risk than the
non-stationarity based on ordinary least square method, utilizing LEGRI. Furthermore, the FR and FRA were used to quantitatively
geographic coordinates and kernel functions to perform local regression
estimates for neighboring entities in each research unit (Brunsdon et al.,
2010; Zhu et al., 2020). However, the bandwidth of GWR regression
parameters is the same. MGWR takes into account the heterogeneity
scale of different parameters, compensating for the shortcomings of a
single scale in GWR (Zhou et al., 2023).
In this work, we constructed the regression relationship between
influencing factors and LEGRI at the mine scale by combining the fea­

Fig. 4. Conceptual diagram of the MGWR model at the mine scale. Fig. 5. Heatmap of indicators correlation.

7
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Fig. 6. Accuracy verification of LEGRI and EGRI. (a). Comparison of pixel by pixel point density and histogram. (b). Distribution and frequency ratio of geological
hazards in two assessment results.

validate the assessment results. The distribution characteristics and the area with loose structure and susceptible to wind erosion. In addition,
FRs of geological disaster at different risk levels exhibited some degree due to shallow mining and relatively fragmented underground mining,
of variation (Fig. 6b). The FRA of EGRI and LEGRI are 73.67% and affecting the eco-geological environment of these townships. Specif­
74.87%, respectively. The result indicated that LEGRI has better ically, the LEGRI–III zone is distributed throughout the entire study area
assessment performance. without significant spatial clustering. The LEGRI–I-II zones are primarily
located in Erlintu, Zhongji, and Tianjiazhai towns, where there are fewer
mining activities and less landscape dispersion, resulting in a relatively
4.3. Spatial distribution characteristics of LEGRI
low eco-geological environmental risk. As a result, the LEGRI–II-III
zones account for 65.7%, while the EGERI–IV-V zones account for
The LEGRI in the SFMA was classified into five levels from low
19.3%, indicating that the overall risk level in the SFMA is moderate,
(LEGRI–I) to high (LEGRI–V) using the natural breaks method. The
high risk existed in the certain areas.
spatial distribution of LEGRI is presented in Fig. 7. The overall eco-
geological risk in the study area is low in the northwest and high in
the northeast and southwest. The LEGRI–IV-V risk zones are mainly 4.4. Spatial autocorrelation analysis of LEGRI
distributed in Laogaochuan, Sandaogou, Miaogoumen, and Dabaodang
towns. The areas of LEGRI–IV-V are 79.10 km2, 74.34 km2, 153.28 km2, From Fig. 8a, the global Moran’s I value is 0.798 (p < 0.01, z > 2.58),
and 201.97 km2, accounting for 34.28%, 49.73%, 62.11%, and 33.58% signifying a notably positive correlation in the spatial distribution of
of the respective town areas. Laogaochuan, Sandaogou, and Miaogou­ LEGRI in the SFMA. The LEGRI in SFMA has obvious spatial clustering
men towns are located on hilly loess areas with fragmented terrain and characteristics (Fig. 8b). On the whole, the clustering characteristics of
numerous gullies, while Dabaodang town is located on an aeolian sandy SFMA mainly manifest as high-high and low-low types. The high-high

Fig. 7. Evaluation results of landscape eco-geological risk index.

8
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Fig. 8. Spatial autocorrelation analysis of LEGRI. (a) Moran’s I scatter. (b) LISA clustering diagram.

clustering area mainly distributed in the northeast and southwest of the spatial distribution of mining intensity in the SFMA (Fan et al., 2016),
SFMA. The large number of mines and the high density of mining in we found that the LEGRI does not always higher in eco-geological risk
these areas have altered landscape pattern and aggravated its frag­ with increasing mining intensity. High-intensity mining often occurs in
mentation, which greatly affects the eco-geological conditions of these large and medium-sized mining areas that focus on implementing
regions. In contrast, the northwest and southeast parts of the study area disaster prevention, land management, and ecological restoration pro­
showed a low-low risk clustering phenomenon, indicating that these jects under policy driven initiatives. Therefore, the LEGRI levels in these
areas are less affected by coal mining activities and have relatively regions are at a medium to high level. On the contrary, although Lao­
better ecological conditions. The high-low and the low-high regions are gaochuan, Sandaogou, and Miaogoumen towns are medium mining in­
mainly distributed around both the high-high and low-low aggregation tensity, the high underground mining density and strong clustering
areas. This explained that the eco-geological risks in the study area are result in high LEGRI levels. Previous studies also support this viewpoint,
relatively stable. the standardization of mining activities in large and medium-sized
mining areas has gradually reduced the negative impact of mining on
the environment (Xiao et al., 2020). The mining intensity of a mining
4.5. MGWR-based monitoring plan for LEGRI in mining areas
area is determined by the length and depth of the mining face, rather
than its production capacity.
This study focuses on the application of eco-geological risk assess­
ment results at the mine scale. In previous studies, it is a conventional
idea to monitor the quality of eco-geological environments by over­ 5.2. Application of LEGR assessment framework
laying the multiplication of both indicator weights and time series of
indicator values on the evaluation unit (Li et al., 2023). However, the Based on the regional eco-geological risk situation differences, the
calculation of LEGRI across the region shares a common set of weights so scholars designed the assessment framework for different background
that above monitoring plan at the mine scale may be biased due to the conditions and risk factors (Ghosh and Maiti, 2021; Mantelli et al., 2011;
spatial heterogeneity of eco-geological conditions. In addition, the Weng et al., 2023). However, these frameworks do not fully consider the
relatively high subsidence rates may lead to decorrelation in SBAS- characteristics of geological risk. Only relying on indicators that reflect
InSAR derivation results, causing a limited number of effective subsi­ the inherent characteristics of landforms, such as faults, formation li­
dence points (Li et al., 2022a). To better address these problems, we thology and slope, lacking the risk sources that disturb the changes of
constructed a regression relationship between assessment indicators and geological conditions.(Bastos et al., 2023) also emphasized that the
LEGRI based on the MGWR model. Subsequently, we utilized these ecological risk assessment in the region with high vulnerability should
regression coefficients as indicator weights to devise an eco-geological be clear about the risk source. Therefore, the above framework is not
risk monitoring plan for mining areas. The regression coefficients of applicable to the eco-geological risk assessment of areas affected by
each underground mining were estimated (Table 4). The lithological groundwater extraction and mining activities. The framework of this
hardness and soil type are classified as a constant parameter of the study uses SAR satellite time series images to deduce the risk factors of
MGWR model due to not change significantly in individual mining areas. regional land subsidence, which provides several advantages over other
The LEGRI of a monitoring site or assessment unit within a mine site assessment frameworks. This framework offers a detailed quantification
varies as the value of one or more indicators changes. When the LEGRI is of the spatial distribution of surface disturbance during specific periods
above or below the critical threshold, the corresponding eco-geological and is suitable for eco-geological risk assessment of different scales (e.g.,
risk level is increased or decreased. single mining area, mining area group and mining city).
Fragmented landscape patterns lead to deterioration of ecological
5. Discussion quality compared to landscape types with a high degree of connectivity
(Biswas et al., 2023; Yang et al., 2023b). (Shi et al., 2023) utilized
5.1. Impact of mining intensity on LEGRI landscape pattern indices of the patch density, the landscape division
index, the Shannons diversity index, and the perimeter-area fractal
High-intensity coal mining exacerbates soil erosion and vegetation dimension to assess vulnerability in the ecological risk assessment
damage, leading to environmental degradation. Mining intensity is framework. The SFMA has a high degree of landscape pattern frag­
therefore considered to be a key factor contributing to the increase in mentation due to the long time-series coal mining activities. However,
LEGR (Xu et al., 2021b). However, through a comparison with the existing eco-geological risk assessments of the mine rarely focus on

9
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Table 4
MGWR parameters estimate for mining areas.
Mine β1(×103) β2 β3(×102) β4(×102) β5(×104) β6 β7(×102) β8 β0

U1 8.194 − 0.323 2.581 0.012 1.018 − 0.080 − 1.089 − 4.002 1.571


U2 4.207 − 0.191 0.228 − 0.362 − 0.335 0.324 − 0.747 − 1.657 3.322
U3 0.310 − 0.782 1.338 1.934 − 1.853 − 0.142 1.581 − 3.553 − 4.150
U4 − 3.325 − 0.345 0.955 0.050 1.768 0.160 0.059 − 1.939 1.794
U5 2.202 − 0.097 1.772 0.071 4.553 − 0.131 − 2.339 − 1.180 0.495
U6 − 2.528 − 0.011 − 0.028 0.050 − 1.221 0.030 − 0.117 − 1.768 2.635
U7 − 1.244 − 0.645 0.662 0.743 − 0.777 − 0.144 1.335 − 4.097 0.396
U8 2.023 − 0.605 1.137 − 1.149 − 0.020 − 0.171 1.159 − 6.262 8.125
U9 − 0.435 − 0.599 1.048 0.627 0.201 − 0.144 1.156 − 4.016 0.561
U10 − 0.595 − 0.445 1.185 − 0.127 − 0.007 − 0.158 1.067 − 3.956 3.555
U11 − 4.493 0.038 − 0.614 0.115 − 0.531 − 0.047 0.447 4.418 0.685
U12 − 2.735 0.111 − 0.267 0.208 0.419 − 0.024 0.454 − 3.092 1.742
U13 − 5.276 − 0.543 − 0.135 0.105 − 0.491 − 0.073 0.482 − 5.494 3.201
U14 − 4.737 − 0.285 0.402 − 0.761 − 3.772 0.174 − 0.213 − 2.535 5.702
U15 − 1.141 − 0.041 0.010 0.168 0.259 0.015 − 0.397 − 3.904 2.239
U16 − 6.463 0.021 0.110 − 0.009 − 0.857 − 0.011 0.011 − 7.826 3.905
U17 − 4.539 − 0.166 0.022 0.051 − 0.902 0.002 0.210 − 6.048 3.574
U18 − 5.415 − 0.879 − 0.241 0.204 − 1.789 0.031 0.231 − 3.124 2.451
U19 − 3.348 − 0.290 0.753 − 5.517 0.024 − 0.212 0.979 − 8.451 28.019
U20 − 1.908 − 0.073 0.013 0.051 − 0.385 − 0.094 − 0.210 − 0.407 2.721
U21 − 20.776 − 0.514 0.813 0.564 0.452 − 0.197 1.040 − 3.304 0.556
U22 − 13.317 − 0.258 0.580 2.887 1.249 0.025 0.224 − 1.962 − 10.582
U23 − 9.083 − 0.039 0.272 0.116 − 0.311 − 0.097 − 0.109 − 5.861 3.095
U24 − 4.741 − 0.610 − 0.111 0.118 1.740 0.000 − 0.056 − 5.405 2.585
U25 − 23.401 − 0.410 0.626 0.920 − 2.397 − 0.204 0.824 − 0.516 − 1.252
U26 − 10.907 − 0.123 0.297 0.165 1.386 − 0.366 0.592 − 8.192 3.443
U27 − 20.707 − 0.765 1.224 1.567 0.245 − 0.119 1.016 − 4.288 − 3.396
U28 − 5.051 − 0.105 0.094 0.148 − 0.468 − 0.032 0.197 − 1.589 1.926
U29 − 4.839 − 0.592 0.717 1.074 − 0.082 − 0.144 1.270 − 6.919 − 0.478
U30 − 2.349 − 0.050 0.172 − 0.057 − 1.448 0.029 − 0.349 0.557 2.545
U31 0.074 − 0.462 0.727 − 0.138 0.050 − 0.174 1.183 − 3.496 3.578
U32 − 0.754 − 0.572 1.148 0.046 − 0.779 − 0.250 1.263 − 3.898 3.196
U33 1.913 0.091 1.313 − 0.001 − 0.695 − 0.037 1.514 1.866 1.917
U34 − 0.506 − 0.564 0.648 − 0.323 − 0.224 − 0.197 1.074 − 4.457 4.829
U35 − 2.907 − 0.873 0.181 − 72.709 0.371 0.261 0.885 − 3.066 292.796
U36 − 0.554 − 0.508 0.991 0.598 − 0.539 − 0.150 1.341 − 5.941 1.209
U37 − 3.812 − 0.466 0.148 0.068 − 0.144 0.007 1.383 − 8.055 3.600
U38 − 3.849 − 0.471 1.189 − 2.379 − 0.083 − 0.226 0.771 − 0.144 12.165
U39 − 0.326 − 0.887 − 0.986 0.152 − 3.397 − 0.012 − 0.569 − 2.771 2.793
U40 − 8.503 − 0.420 0.554 0.224 − 1.327 − 0.017 − 0.359 − 2.585 2.042
U41 − 17.971 − 0.498 1.264 0.851 − 0.138 − 0.177 0.979 − 3.453 − 0.424
U42 − 1.436 − 0.747 1.093 − 0.857 − 0.047 0.250 0.917 − 3.900 6.059
U43 0.305 − 0.414 0.775 0.274 0.068 − 0.189 1.198 − 3.138 1.869
U44 − 1.402 − 0.099 0.043 0.051 − 0.416 − 0.042 0.432 − 1.896 2.233
U45 0.703 − 0.462 0.817 − 13.261 − 0.787 − 0.164 0.941 − 3.835 58.535
U46 − 3.992 − 0.193 − 0.833 0.118 3.124 − 0.092 − 2.048 − 0.648 1.112
U47 − 0.032 − 0.115 0.164 0.952 1.722 − 0.130 0.000 − 4.102 − 1.833
U48 − 3.832 − 0.100 0.104 − 0.068 0.846 − 0.027 − 1.679 − 5.361 3.283
U49 0.650 0.062 0.229 0.007 0.174 − 0.011 0.312 − 3.262 2.848
O1 − 6.365 − 0.795 0.462 − 8.631 0.999 0.230 2.129 − 3.337 30.670
O2 − 5.590 0.040 − 0.865 0.319 − 0.539 − 0.010 0.574 − 4.987 2.265
O3 − 12.550 − 0.077 1.823 − 4.089 2.016 − 0.475 1.097 − 3.766 19.626
O4 − 9.518 − 0.416 2.550 0.047 − 1.771 − 0.370 1.039 − 6.381 3.957

Note: β1 ~ β8 represent the first-order coefficients of subsidence rate, FVC, fluctuation, precipitation, distance to road, RSEI, slope, and biological abundance in the
MGWR model. β0 is the constant term. “U” denotes underground mine, “O” denotes open-pit mine.

landscape pattern fragmentation. Therefore, the LEGR framework in­ suggestions for decision-makers to develop sustainable development
corporates a landscape ecological risk index to measure the impact of plans.
landscape patch variation on the regional eco-geological environment. High-intensity mining activities have led to surface subsidence and
In conclusion, the LEGR framework helps to understand the impact of ground fissures in the loess gully area of the SFMA, the surface water
human activities and climate change on eco-geological conditions. seeps into the deep layers of the surface along the fissures, causing
damage to vegetation and soil ecosystems (Li et al., 2020a; Xu et al.,
2021b), resulting in a relatively high for LEGRI level. Mining companies
5.3. Influence on decision-making
are advised to combine numerical modeling and SBAS-InSAR technology
to quantify subsidence above and around the working face (Woo et al.,
The LEGRI in the SFMA exhibited significant spatial variation. Un­
2012), and to backfill and remediate areas of high subsidence promptly
derstanding these differences is conducive to the more effective man­
after mining. Some towns, such as Miaogoumen, Sandaogou, and Lao­
agement of ecosystems and predicting their responses to natural and
gaochuan towns, have a high density of underground mining and
human interventions. This study emphasizes on both ecological and
increased landscape fragmentation, which has resulted in a reduction in
geological pressure factors. The results of the LEGR assessment help to
natural landscape connectivity and a significant spatial diffusion of
reduce the eco-geological risk of SFMA. Based on the township LEGRI
LEGRI-IV-V in the mining area. In addition, the LEGRI in different
characteristics and field survey, we provided the following ancillary

10
X. Zhu et al. Ecological Informatics 81 (2024) 102635

microtopographic units is different because of the complex topography mining areas aggravate degree of LEGRI. The areas of LEGRI–IV-V of
of the SFMA. Therefore, specific ecological restoration measures should mining-type towns such as Laogaochuan town and Sandaogou town
fully consider the local topographic and geomorphological characteris­ were 79.10km2 and 74.34km2, respectively, exceeding 34% of the
tics, optimizing the landscape pattern to reduce the overall LEGRI of the respective town areas. Therefore, ecological restoration projects should
mining area. The degree of desertification in the Mu Us Desert has been address the direct eco-geological environmental problems caused by
significantly reversed after decades of management and restoration (Liu mining activities. In addition, a monitoring plan for the eco-geological
et al., 2020). To prevent the rebound of desertification, it is recom­ risk in the mining area was developed by combining the LEGRI with
mended to integrate multi-source remote sensing to conduct large-scale the MGWR model. It provides scientific references for decision-makers
and refined desertification monitoring on the LEGRI-III-IV-V zones of the in formulating environmental protection policies.
sand-covered area. By identifying the rebound or expansion areas of
desertification, and then implementing targeted restoration measures to Funding
improve the stability of the ecological landscape.
This research was funded by the open funds from the State Key
5.4. Limitations Laboratory of Coal Mining and Clean Utilization (Granted No.2021-
CMCU-KF014), Fundamental Research Funds for the Key Laboratory of
The eco-geological risk assessment framework of this study used Mine Ecological Effects and Systematic Restoration, Ministry of Natural
multi-source remote sensing images to calculate ecological quality (e.g., Resources (Grant No. MEER2023–06), Natural Science Foundation of
RSEI), geological conditions (e.g., land subsidence rate), and landscape China (Grant No. 52174160), the Fundamental Research Funds for the
pattern fragmentation. This improves the reliability of the evaluation Central Universities (Grant Nos. 2023ZKPYDC04 and 2023YJSDC08),
system. Nevertheless, this study has certain limitations. Firstly, envi­ the Ecological-Smart Mines Joint Research Fund of the Natural Science
ronmental pollution resulting from the mining process is a factor Foundation of Hebei Province (Grant No. E2020402086).
contributing to the alteration in surface risk, but it is not included in the
LEGRI assessment index system due to the insufficiency of data such as CRediT authorship contribution statement
soil pollution and groundwater pollution. Secondly, different landscape
scales explain differences in external pressure. How to divide the Xiaoya Zhu: Writing – original draft, Visualization, Methodology,
optimal landscape scale and landscape pattern to respond to the impact Conceptualization. Peixian Li: Writing – review & editing, Investiga­
of external disturbances on LEGRI will be a key focus of future research. tion, Funding acquisition. Bing Wang: Visualization, Software. Sihai
Zhao: Supervision. Tao Zhang: Validation, Supervision. Qingyue Yao:
6. Conclusion Supervision.

This study proposed a novel landscape eco-geological risk assess­


ment framework by integrating regional intrinsic attributes, stress Declaration of competing interest
characteristics and landscape risk disturbances. The application of the
LEGRI evaluation framework not only demonstrates satisfactory per­ The authors declare no conflicts of interest.
formance in the SFMA (FRA = 74.87%), but also the framework can be
applied in other similar mining areas. The results indicated that the Data availability
overall eco-geological risk of the SFMA was moderate or below, the
LEGRI–I-II-III zones accounting for 80.7% of the study area. The spatial Data will be made available on request.
distribution of the LEGRI gradually increases from southwest to north­
east. The LEGRI global Moran’s I value of SFMA was 0.798, with high- Acknowledgments
high accumulations mainly influenced by human interference.
Notably, the surface subsidence above the mining area spreads to the The authors would like to acknowledge the open dataset of satellite
surrounding environment, resulting in a high degree of LEGRI in these images provided by the USGS and open survey data published by local
areas. Cumulative disturbance effects in long-term and high-intensity Chinese governments.

Appendix A

(1) Judgment matrix construction and quantitative grading of indicators


The eco-geological risk assessment model of SFMA can be divided into target layer (A), criterion layer (B) and indicator layer (C). The target layer is
the ecological and geological risk of SFMA; The criterion layer includes topographic and geological conditions, ecological conditions and anthro­
pogenic factors; The indicator layer includes 10 selected indicators. The discriminant matrix of the target layer (A-B) and the discriminant matrix of
the criterion layer (B1–C, B2–C, B3–C) are:
⎡ ⎤
B1 B2 B3
⎢ B1 1 1/2 1/3 ⎥
A− B=⎢ ⎣ B2 2

1 1/2 ⎦
B3 3 2 1
⎡ ⎤
C1 C2 C3 C4
⎢ C1 1 2 1/2 2 ⎥
⎢ ⎥
B1 − C⎢ ⎢ C2 1/2 1 1/2 1 ⎥ ⎥
⎣ C3 2 2 1 3 ⎦
C4 1/2 1 1/3 1

11
X. Zhu et al. Ecological Informatics 81 (2024) 102635

⎡ ⎤
C5 C6 C7
⎢ C5 1 1/2 1/3 ⎥
B2 − C = ⎢
⎣ C6

2 1 1/2 ⎦
C7 3 2 1
⎡ ⎤
C8 C9 C10
⎢ C8 1 3 1/2 ⎥
B3 − C = ⎢
⎣ C9 1/3

1 1/4 ⎦
C10 2 4 1

where, C1 to C11 represent the fluctuation, soil type, slope, lithological hardness, precipitation, FVC, RSEI, biological abundance, distance to roads,
and subsidence rate, respectively.
By referring to the literature and expert scoring, the data of the evaluation indicators were graded in a standard quantitative way to carry out the
comprehensive operation among the indicators (Table A.1). The evaluation indicators of EGR in the study area are divided into five levels from low to
high, assigned values 1, 3, 5, 7 and 9 respectively.

Table A.1
Quantitative grading standards of evaluation indicators.

score 1 3 5 7 9

Fluctuation (m) <5 5–10 10–15 15–20 >20


Slope (◦ ) <5 5–15 15–25 25–35 >35
Soil type – Semi hydrated soil Aqueous soil Primary soil Calcareous soil
Lithological hardness – – Medium hardness rock Soft rock Extremely soft rock
FVC 0.8–1.0 0.6–0.8 0.4–0.6 0.2–0.4 0–0.2
RSEI 0.8–1.0 0.6–0.8 0.4–0.6 0.2–0.4 0–0.2
Precipitation (mm) <370 370–390 390–410 410–430 >430
Biological abundance 0.8–1.0 0.6–0.8 0.4–0.6 0.2–0.4 0–0.2
Distance to roads (m) >500 300–500 200–300 100–200 <100
Sbusidence rate (mm/year) − 10–40 − 20 ~ − 10 − 30 ~ − 20 − 40 ~ − 30 <− 40 | >40

(2) Consistency Check

CI
CR = (A.1)
RI
λmax − n
CI = (A.2)
n− 1

where, CI and RI are the consistency index and average random consistency index of the judgment matrix; λmax is the maximum eigenvalue
of the consistency matrix; n is the number of compared factors.
Table A.2
Random consistency index.

Order 1 2 3 4 5 6 7 8 9 10

RI 0 0 0.52 0.89 1.12 1.26 1.36 1.41 1.46 1.49

References Chen, W., Xie, X., Wang, J., Pradhan, B., Hong, H., Bui, D.T., Duan, Z., Ma, J., 2017.
A comparative study of logistic model tree, random forest, and classification and
regression tree models for spatial prediction of landslide susceptibility. CATENA
An, H., Tang, Z., Keesstra, S., Shangguan, Z., 2019. Impact of desertification on soil and
151, 147–160. https://doi.org/10.1016/j.catena.2016.11.032.
plant nutrient stoichiometry in a desert grassland. Sci. Rep. 9, 9422. https://doi.org/
Chen, Z., Chen, J., Zhou, C., Li, Y., 2022. An ecological assessment process based on
10.1038/s41598-019-45927-0.
integrated remote sensing model: a case from Kaikukang-Walagan District, Greater
Bastos, M.I., Roebeling, P.C., Alves, F.L., Villasante, S., Magalhães Filho, L., 2023. High
Khingan Range, China. Ecol. Inform. 70, 101699 https://doi.org/10.1016/j.
risk water pollution hazards affecting Aveiro coastal lagoon (Portugal) – a habitat
ecoinf.2022.101699.
risk assessment using InVEST. Eco. Inform. 76, 102144 https://doi.org/10.1016/j.
Chen, L., Yang, H., Huang, H., Chang, M., Wang, X., Han, D., Liu, S., Xiao, Y., Yao, D.,
ecoinf.2023.102144.
Xiang, X., Yang, Q., 2023. How do natural and socio-economic factors influence the
Biswas, G., Sengupta, A., Alfaisal, F.M., Alam, S., Alharbi, R.S., Jeon, B.-H., 2023.
sustainable development of the ecological environment in the World Natural
Evaluating the effects of landscape fragmentation on ecosystem services: a three-
Heritage Sites? Evidence from the Jiuzhaigou, China. J. Clean. Prod. 428, 139238
decade perspective. Eco. Inform. 77, 102283 https://doi.org/10.1016/j.
https://doi.org/10.1016/j.jclepro.2023.139238.
ecoinf.2023.102283.
Cheng, J., Sun, J., Yao, K., Xu, M., Cao, Y., 2022. A variable selection method based on
Brunsdon, C., Fotheringham, A.S., Charlton, M.E., 2010. Geographically weighted
mutual information and variance inflation factor. Spectrochim. Acta A Mol. Biomol.
regression: a method for exploring spatial nonstationarity. Geogr. Anal. 28, 281–298.
Spectrosc. 268, 120652 https://doi.org/10.1016/j.saa.2021.120652.
https://doi.org/10.1111/j.1538-4632.1996.tb00936.x.
Deng, L., Yin, J., Tian, J., Li, Q., Guo, S., 2021. Comprehensive evaluation of water
Cai, Z., Zhang, Z., Zhao, F., Guo, X., Zhao, J., Xu, Y., Liu, X., 2023. Assessment of eco-
resources carrying capacity in the Han River basin. Water 13, 249. https://doi.org/
environmental quality changes and spatial heterogeneity in the Yellow River Delta
10.3390/w13030249.
based on the remote sensing ecological index and geo-detector model. Eco. Inform.
Diakoulaki, D., Mavrotas, G., Papayannakis, L., 1995. Determining objective weights in
77, 102203 https://doi.org/10.1016/j.ecoinf.2023.102203.
multiple criteria problems: the critic method. Comput. Oper. Res. 22, 763–770.
Cao, Q., Zhang, X., Lei, D., Guo, L., Sun, X., Kong, F., Wu, J., 2019. Multi-scenario
https://doi.org/10.1016/0305-0548(94)00059-H.
simulation of landscape ecological risk probability to facilitate different decision-
Dou, S., Xu, D., Keenan, R.J., 2023. Effect of income, industry structure and
making preferences. J. Clean. Prod. 227, 325–335. https://doi.org/10.1016/j.
environmental regulation on the ecological impacts of mining: An analysis for
jclepro.2019.03.125.
Guangxi Province in China. J. Clean. Prod. 400, 136654 https://doi.org/10.1016/j.
jclepro.2023.136654.

12
X. Zhu et al. Ecological Informatics 81 (2024) 102635

Fan, L., Ma, X., Li, Y., Li, C., Yao, C., Xiang, M., WU, B., Peng, J., 2016. Geological Shang, Y., Ye, X., Dong, L., Liu, S., Du, T., Wang, G., 2022. Landscape pattern evolution
disasters and control technology in high intensity mining area of western China[J]. in a Mining City: an urban life cycle perspective. Sustainability 14, 8492. https://doi.
J. China Coal Soc. 42 (2), 276–285. https://doi.org/10.13225/j.cnki.jccs.2016.6002, org/10.3390/su14148492.
2017. Sharma, L.P., Patel, N., Debnath, P., Ghose, M.K., 2012. Assessing landslide vulnerability
Fan, L., Li, T., Xiang, M., He, W., Wu, B., Peng, J., Li, Y., Li, C., Zheng, M., Chen, J., from soil characteristics—a GIS-based analysis. Arab. J. Geosci. 5, 789–796. https://
Gao, S., Du, J., Ji, Y., 2018. Effect of coal mining on springs in the Yushenfu Mining doi.org/10.1007/s12517-010-0272-5.
Area of China. Geofluids 2018, 1–16. https://doi.org/10.1155/2018/3564360. Shi, H., Shi, T., Yang, Z., Wang, Z., Han, F., Wang, C., 2018. Effect of roads on ecological
Ghosh, A., Maiti, R., 2021. Development of new Ecological Susceptibility Index (ESI) for corridors used for wildlife movement in a natural heritage site. Sustainability 10,
monitoring ecological risk of river corridor using F-AHP and AHP and its application 2725. https://doi.org/10.3390/su10082725.
on the Mayurakshi river of Eastern India. Eco. Inform. 63, 101318 https://doi.org/ Shi, Z., Xiang, F., Guo, Y., 2023. Ecological risk of geohazards and its combination
10.1016/j.ecoinf.2021.101318. patterns: a case study of an ecologically fragile region, NW ChinaEcol. Inform. 77,
Gong, J., Cao, E., Xie, Y., Xu, C., Li, H., Yan, L., 2021. Integrating ecosystem services and 102153. doi:https://doi.org/10.1016/j.ecoinf.2023.102153.
landscape ecological risk into adaptive management: insights from a western Tang, W., Zhao, X., Motagh, M., Bi, G., Li, J., Chen, M., Chen, H., Liao, M., 2022. Land
mountain-basin area, China. J. Environ. Manag. 281, 111817 https://doi.org/ subsidence and rebound in the Taiyuan basin, northern China, in the context of inter-
10.1016/j.jenvman.2020.111817. basin water transfer and groundwater management. Remote Sens. Environ. 269,
Guo, Z., Xie, Y., Guo, H., Zhang, X., Wang, H., Bie, Q., Xi, G., Ma, C., 2023. Do the 112792 https://doi.org/10.1016/j.rse.2021.112792.
ecosystems of Gansu Province in Western China’s crucial ecological security barrier Teodoro, A., Santos, P., Espinha Marques, J., Ribeiro, J., Mansilha, C., Melo, A.,
remain vulnerable? Evidence from remote sensing based on geospatial analysis. Duarte, L., Rodrigues De Almeida, C., Flores, D., 2021. An integrated multi-approach
J. Clean. Prod. 402, 136740 https://doi.org/10.1016/j.jclepro.2023.136740. to environmental monitoring of a self-burning coal waste pile: the São Pedro da Cova
Huang, X., Wang, X., Zhang, X., Zhou, C., Ma, J., Feng, X., 2022. Ecological risk mine (Porto, Portugal) study case. Environments 8, 48. https://doi.org/10.3390/
assessment and identification of risk control priority areas based on degradation of environments8060048.
ecosystem services: a case study in the Tibetan Plateau. Ecol. Indic. 141, 109078 Tesfa, C., Woldearegay, K., 2021. Characteristics and susceptibility zonation of landslides
https://doi.org/10.1016/j.ecolind.2022.109078. in Wabe Shebelle Gorge, south eastern Ethiopia. J. Afr. Earth Sci. 182, 104275
Jin, J., Yan, C., Tang, Y., Yin, Y., 2021. Mine geological environment monitoring and risk https://doi.org/10.1016/j.jafrearsci.2021.104275.
assessment in arid and semiarid areas. Complexity 2021, 1–10. https://doi.org/ Trofimov, V.T., 2008. Ecological geology, environmental geology, geoecology: contents
10.1155/2021/3896130. and relations. Mosc. Univ. Geol. Bull. 63, 59–69. https://doi.org/10.3103/
Kim, E.S., Lee, D.K., Choi, J., Kim, J.H., Mo, Y., Jeon, Y., Kim, J.Y., 2024. Metapopulation S0145875208020014.
models using landscape connectivity can better reflect landscape heterogeneity. Eco. Wang, H., Liu, X., Zhao, C., Chang, Y., Liu, Y., Zang, F., 2021a. Spatial-temporal pattern
Inform. 80, 102464 https://doi.org/10.1016/j.ecoinf.2024.102464. analysis of landscape ecological risk assessment based on land use/land cover
Kokangül, A., Polat, U., Dağsuyu, C., 2017. A new approximation for risk assessment change in Baishuijiang national nature reserve in Gansu Province, China. Ecol. Indic.
using the AHP and Fine Kinney methodologies. Saf. Sci. 91, 24–32. https://doi.org/ 124, 107454 https://doi.org/10.1016/j.ecolind.2021.107454.
10.1016/j.ssci.2016.07.015. Wang, X., Zhang, C., Wang, C., Liu, G., Wang, H., 2021b. GIS-based for prediction and
Kucuker, D.M., Cedano Giraldo, D., 2022. Assessment of soil erosion risk using an prevention of environmental geological disaster susceptibility: from a perspective of
integrated approach of GIS and Analytic Hierarchy Process (AHP) in Erzurum, sustainable development. Ecotoxicol. Environ. Saf. 226, 112881 https://doi.org/
Turkiye. Ecol. Inform. 71, 101788 https://doi.org/10.1016/j.ecoinf.2022.101788. 10.1016/j.ecoenv.2021.112881.
Li, C., Chen, J., Liao, M., Chen, G., Zhou, Q., 2018a. Ecological risk assessment of Shan Wang, B., Li, P., Zhu, X., 2023. Quantification of vegetation phenological disturbance
Xin Mining area based on remote sensing and geography information system characteristics in open-pit coal mines of arid and semi-arid regions using harmonized
technology. JGIS 10, 234–246. https://doi.org/10.4236/jgis.2018.102012. Landsat 8 and Sentinel-2. Remote Sens. 15, 5257. https://doi.org/10.3390/
Li, W., Liu, S., Pei, Y., He, J., Wang, Q., 2018b. Zoning for eco-geological environment rs15215257.
before mining in Yushenfu mining area, northern Shaanxi, China. Environ. Monit. Wang, X., Xiong, J., Wang, J., Liu, M., Zhang, J., 2024. Spatiotemporal evolution and
Assess. 190, 619. https://doi.org/10.1007/s10661-018-6996-5. driving factors of tourism ecological adaptation in the Dongting Lake Area, China.
Li, Jianwei, Li, X., Liu, C., Wu, X., 2020a. Dynamic changes in surface damage induced by Ecol. Inform. 80, 102459 https://doi.org/10.1016/j.ecoinf.2024.102459.
high-intensity mining of shallow, thick coal seams in gully areas. Adv. Civil Eng. Weng, C., Bai, Y., Chen, B., Hu, Y., Shu, J., Chen, Q., Wang, P., 2023. Assessing the
2020, 1–16. https://doi.org/10.1155/2020/5151246. vulnerability to climate change of a semi-arid pastoral social–ecological system: a
Li, Jun, Pei, Y., Zhao, S., Xiao, R., Sang, X., Zhang, C., 2020b. A review of remote sensing case study in Hulunbuir, China. Ecol. Inform. 76, 102139 https://doi.org/10.1016/j.
for environmental monitoring in China. Remote Sens. 12, 1130. https://doi.org/ ecoinf.2023.102139.
10.3390/rs12071130. Woo, K.-S., Eberhardt, E., Rabus, B., Stead, D., Vyazmensky, A., 2012. Integration of field
Li, S., Xu, W., Li, Z., 2022a. Review of the SBAS InSAR time-series algorithms, characterisation, mine production and InSAR monitoring data to constrain and
applications, and challenges. Geodesy Geodynam. 13, 114–126. https://doi.org/ calibrate 3-D numerical modelling of block caving-induced subsidence. Int. J. Rock
10.1016/j.geog.2021.09.007. Mech. Min. Sci. 53, 166–178. https://doi.org/10.1016/j.ijrmms.2012.05.008.
Li, L., Zhou, X., Yang, L., Duan, J., Zeng, Z., 2022b. Spatio-temporal characteristics and Xiao, W., Zhang, W., Lyu, X., Wang, X., 2020. Spatio-temporal patterns of ecological
influencing factors of ecological Risk in China’s North–South Transition Zone. capital under different mining intensities in an ecologically fragile mining area in
Sustainability 14, 5464. https://doi.org/10.3390/su14095464. Western China: a case study of Shenfu mining area. J. Nat. Resourc. 35 (1), 68–81.
Li, P., Wang, B., Chen, P., Zhang, Y., Zhao, S., 2023. Vulnerability assessment of the eco- https://doi.org/10.31497/zrzyxb.20200107, 2020.
geo-environment of mining cities in arid and semi-arid areas: a case study from Xie, H., Wang, P., Huang, H., 2013. Ecological risk assessment of land use change in the
Zhungeer, China. Ecol. Indic. 152, 110364 https://doi.org/10.1016/j. Poyang Lake Eco-economic Zone, China. IJERPH 10, 328–346. https://doi.org/
ecolind.2023.110364. 10.3390/ijerph10010328.
Liu, S., Li, W., 2019. Zoning and management of phreatic water resource conservation Xu, W., Wang, J., Zhang, M., Li, S., 2021a. Construction of landscape ecological network
impacted by underground coal mining: a case study in arid and semiarid areas. based on landscape ecological risk assessment in a large-scale opencast coal mine
J. Clean. Prod. 224, 677–685. https://doi.org/10.1016/j.jclepro.2019.03.282. area. J. Clean. Prod. 286, 125523 https://doi.org/10.1016/j.jclepro.2020.125523.
Liu, Q., Zhang, Q., Yan, Y., Zhang, X., Niu, J., Svenning, J.-C., 2020. Ecological Xu, J., Zhu, W., Xu, Jialin, Wu, J., Li, Y., 2021b. High-intensity longwall mining-induced
restoration is the dominant driver of the recent reversal of desertification in the Mu ground subsidence in Shendong coalfield, China. Int. J. Rock Mech. Min. Sci. 141,
Us Desert (China). J. Clean. Prod. 268, 122241 https://doi.org/10.1016/j. 104730 https://doi.org/10.1016/j.ijrmms.2021.104730.
jclepro.2020.122241. Xu, W., Yao, W., Bai, Z., Yang, J., Li, L., 2023a. Ecological risk evaluation and ecological
Löbmann, M.T., Geitner, C., Wellstein, C., Zerbe, S., 2020. The influence of herbaceous restoration model of mining in the source area of the Yellow River Basin. Land 12,
vegetation on slope stability – a review. Earth Sci. Rev. 209, 103328 https://doi.org/ 933. https://doi.org/10.3390/land12040933.
10.1016/j.earscirev.2020.103328. Xu, Q., Zhao, B., Dai, K., Dong, X., Li, W., Zhu, X., Yang, Y., Xiao, X., Wang, X., Huang, J.,
Mantelli, L.R., Barbosa, J.M., Bitencourt, M.D., 2011. Assessing ecological risk through Lu, H., Deng, B., Ge, D., 2023b. Remote sensing for landslide investigations: a
automated drainage extraction and watershed delineation. Eco. Inform. 6, 325–331. progress report from China. Eng. Geol. 321, 107156 https://doi.org/10.1016/j.
https://doi.org/10.1016/j.ecoinf.2011.04.003. enggeo.2023.107156.
Muço, B., Alexiev, G., Aliaj, S., Elezi, Z., Grecu, B., Mandrescu, N., Milutinovic, Z., Yang, Z., Li, W., Li, X., Wang, Q., He, J., 2019. Assessment of eco-geo-environment
Radulian, M., Ranguelov, B., Shkupi, D., 2012. Geohazards assessment and mapping quality using multivariate data: a case study in a coal mining area of Western China.
of some Balkan countries. Nat. Hazards 64, 943–981. https://doi.org/10.1007/ Ecol. Indic. 107, 105651 https://doi.org/10.1016/j.ecolind.2019.105651.
s11069-012-0185-6. Yang, Y., Wang, H., Wang, C., Yang, B., 2022. Model-based temporal evolution and
Nicholson, E., Watermeyer, K.E., Rowland, J.A., Sato, C.F., Stevenson, S.L., Andrade, A., spatial equilibrium analysis of green development in China’s Yangtze River
Brooks, T.M., Burgess, N.D., Cheng, S.-T., Grantham, H.S., Hill, S.L., Keith, D.A., Economic Belt from 2009 to 2018. Ecol. Indic. 141, 109071 https://doi.org/
Maron, M., Metzke, D., Murray, N.J., Nelson, C.R., Obura, D., Plumptre, A., 10.1016/j.ecolind.2022.109071.
Skowno, A.L., Watson, J.E.M., 2021. Scientific foundations for an ecosystem goal, Yang, L., Pan, S., Chen, W., Zeng, J., Xu, H., Gu, T., 2023a. Spatially non-stationary
milestones and indicators for the post-2020 global biodiversity framework. Nat Ecol response of habitat quality to land use activities in World’s protected areas over 20
Evol 5, 1338–1349. https://doi.org/10.1038/s41559-021-01538-5. years. J. Clean. Prod. 419, 138245 https://doi.org/10.1016/j.jclepro.2023.138245.
Oh, H.-J., Pradhan, B., 2011. Application of a neuro-fuzzy model to landslide- Yang, H., Xu, W., Yu, J., Xie, X., Xie, Z., Lei, X., Wu, Z., Ding, Z., 2023b. Exploring the
susceptibility mapping for shallow landslides in a tropical hilly area. Comput. impact of changing landscape patterns on ecological quality in different cities: a
Geosci. 37, 1264–1276. https://doi.org/10.1016/j.cageo.2010.10.012. comparative study among three megacities in eastern and western China. Eco.
O’Neill, R.V., Hunsaker, C.T., Timmins, S.P., Jackson, B.L., Jones, K.B., Riitters, K.H., Inform. 77, 102255 https://doi.org/10.1016/j.ecoinf.2023.102255.
Wickham, J.D., 1996. Scale problems in reporting landscape pattern at the regional Yang, X., Zheng, X., Zhou, Z., Miao, H., Liu, H., Wang, Y., Zhang, H., You, S., Wei, S.,
scale. Landsc. Ecol. 11, 169–180. https://doi.org/10.1007/BF02447515. 2023c. A novel multilevel decision-making evaluation approach for the renewable

13
X. Zhu et al. Ecological Informatics 81 (2024) 102635

energy heating systems: a case study in China. J. Clean. Prod. 390, 135934 https:// Zheng, Z., Wu, Z., Chen, Y., Guo, C., Marinello, F., 2022. Instability of remote sensing
doi.org/10.1016/j.jclepro.2023.135934. based ecological index (RSEI) and its improvement for time series analysis. Sci. Total
Yin, S., Li, J., Liang, J., Jia, K., Yang, Z., Wang, Y., 2020. Optimization of the weighted Environ. 814, 152595 https://doi.org/10.1016/j.scitotenv.2021.152595.
linear combination method for agricultural land suitability evaluation considering Zhou, Y., Xu, K., Feng, Z., Wu, K., 2023. Quantification and driving mechanism of
current land use and regional differences. Sustainability 12, 10134. https://doi.org/ cultivated land fragmentation under scale differences. Eco. Inform. 78, 102336
10.3390/su122310134. https://doi.org/10.1016/j.ecoinf.2023.102336.
Zhang, P., Chen, X., Ren, Y., Lu, S., Song, D., Wang, Y., 2023a. A novel mine-specific eco- Zhu, C., Zhang, X., Zhou, M., He, S., Gan, M., Yang, L., Wang, K., 2020. Impacts of
environment index (MSEEI) for mine ecological environment monitoring using urbanization and landscape pattern on habitat quality using OLS and GWR models in
Landsat imagery. Remote Sens. 15, 933. https://doi.org/10.3390/rs15040933. Hangzhou, China. Ecol. Indic. 117, 106654 https://doi.org/10.1016/j.
Zhang, Z., Lu, L., Zhao, Y., Wang, Y., Wei, D., Wu, X., Ma, X., 2023b. Recent advances in ecolind.2020.106654.
using Chinese earth observation satellites for remote sensing of vegetation. ISPRS J. Zhu, X., Li, J., Cheng, H., Zheng, L., Huang, W., Yan, Y., Liu, H., Yang, X., 2022. Assessing
Photogramm. Remote Sens. 195, 393–407. https://doi.org/10.1016/j. the impacts of ecological governance on carbon storage in an urban coal mining
isprsjprs.2022.12.006. subsidence area. Eco. Inform. 72, 101901 https://doi.org/10.1016/j.
Zhao, J., Ji, G., Tian, Y., Chen, Y., Wang, Z., 2018. Environmental vulnerability ecoinf.2022.101901.
assessment for mainland China based on entropy method. Ecol. Indic. 91, 410–422. Zou, F., Che, E., Long, M., 2023. Quantitative assessment of geological hazard risk with
https://doi.org/10.1016/j.ecolind.2018.04.016. different hazard indexes in mountainous areas. J. Clean. Prod. 413, 137467 https://
Zhao, S., Li, P., Li, H., Zhang, T., Wang, B., 2023. Monitoring and comparative analysis of doi.org/10.1016/j.jclepro.2023.137467.
Hohhot Subway subsidence using StaMPS-PS based on two DEMS. Remote Sens. 15,
4011. https://doi.org/10.3390/rs15164011.

14

You might also like