Next Article in Journal
Anti-Vibration Method for the Near-Bit Measurement While Drilling of Pneumatic Down-the-Hole Hammer Drilling
Previous Article in Journal
Substituting Sugar in Pastry and Bakery Products with Functional Ingredients
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Applied Sciences
Volume 14
Issue 18
10.3390/app14188562
applsci-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Prediction of Landslide Susceptibility in the Karakorum under the Context of Climate Change

by
Yanqian Pei
1,*,
Haijun Qiu
2,* and
Yaru Zhu
2
1
School of Geographical Sciences, Qinghai Normal University, Xining 810016, China
2
College of Urban and Environmental Sciences, Northwest University, Xi’an 710127, China
*
Authors to whom correspondence should be addressed.
Appl. Sci. 2024, 14(18), 8562; https://doi.org/10.3390/app14188562
Submission received: 23 August 2024 / Revised: 16 September 2024 / Accepted: 19 September 2024 / Published: 23 September 2024
Browse Figures
Versions Notes

Abstract

:
Climate change has recently increased the frequency of landslides in alpine areas. Susceptibility mapping is crucial for anticipating and assessing landslide risk. However, traditional methods focus on static environmental variables to emphasize the spatial distribution of landslides, ignoring temporal dynamics in landslide development in the context of climate change. In this work, we focused on static and dynamic environment factors and utilized the certainty factor-logistic regression (CF-LR) model to assess and predict landslide susceptibility in Taxkorgan County, located in the Karakorum. The assessment and prediction were based on a catalog of climate change-related landslides over the past 20 years, the causative factors, and predicted climatic variables for the Shared Socioeconomic Pathways (SSP1-2.6) scenario. The results indicated that elevation, slope, groundwater, slope length gradient (LS) factor, Topographic Wetness Index (TWI), valley depth, and maximum precipitation were the key causes of slides below the snow line. The key factors causing debris flow above the snow line were elevation, slope, topographic relief, aspect, LS factor, distance to the river, and maximum temperature. The accuracy of slide and debris flow susceptibility was 0.92 and 0.89, respectively. The area of slides with medium, high, and very high susceptibility is 25.5% of the Taxkorgan. In addition, 82.6% of the slides happened in this region, and 49.5% of the entire area is covered by debris flows with medium, high, and very high susceptibility. Moreover, this area accounts for 91.8% of all debris flows. Until 2060, the region’s climate is anticipated to become warmer and wetter. Slides below the snow line will gradually decrease and shift eastward, and debris flows above the snow line will expand. Our findings will contribute to the management of landslide risks at the regional scale.

1. Introduction

Landslides are natural hazards in alpine regions around the world [1,2,3]. Climate change has increased the frequency and amplitude of landslides [4,5,6], endangering mountain people’s property and lives and impeding social development [7,8,9]. The task of landslide susceptibility mapping (LSM) is typically assigned to regional landslide risk management and land use planning [10,11]. Accurate evaluation and prediction of landslide susceptibility under dynamic conditions, such as climate change, can support areas in better planning and managing their natural resources and informing efforts to reduce and control landslide risk [12,13].
Landslides are the result of a combination of dynamic and static variables [14]. Among them, static factors include soil composition, geology, topography, geomorphology, etc., whereas dynamic factors include climatic variables (temperature, heavy rainfall), land use/cover, and earthquakes [15]. Most previous landslide susceptibility results were based on static factors, which were assessed using both qualitative and quantitative methods. Qualitative methods mainly consist of heuristic models such as expert empirical methods [16,17]. However, qualitative methods rely on subjective knowledge and the experience of experts, which can result in uncertainty in the results [18]. The quantitative approach is the calculation of landslide probability by physical models [19], which require a detailed, highly accurate catalog of landslides [20]. Statistical models cannot deal with complex nonlinear problems; hence, they cannot reveal the relationship between landslides and environmental conditions [21]. Machine learning methods, e.g., the logistic regression (LR) model, have the advantage of capturing complex relationships in data in landslide spatiotemporal susceptibility assessments, which is becoming more popular in landslide susceptibility assessments [22,23]. Nevertheless, landslide susceptibility assessment methods still vary in their performance [24], suggesting that selecting the most appropriate method is crucial for accurately assessing and predicting regional landslide susceptibility.
Recently, global warming has exacerbated landslide activities in alpine areas [25,26]. The previous regional landslide susceptibility based on static factor assessments can no longer meet the requirements for regional disaster prevention and mitigation, and there is an urgent need for LSM that takes dynamic factors into consideration. A significant number of researchers have conducted extensive studies on picking one or more dynamic elements and integrating them with climate model prediction and In-SAR technologies to detect landslides [15,27]. The superimposed In-SAR method is a fast and effective way to identify probable landslides in large mountainous areas [9,28,29]. This method has modest technical requirements and computing effort, making it suitable for monitoring landslide susceptibility in remote and challenging terrain. However, the In-SAR method may overlook some active slopes with tiny spatial scales, low displacement levels, and air effects [30]. Furthermore, the superimposed In-SAR technique is insufficient for predicting future landslide susceptibility. According to climate change projections, the severity and frequency of landslides would fluctuate due to changed precipitation and temperature regimes [31]. Additionally, dynamic landslide susceptibility mapping usually employs dynamic variables (e.g., cumulative rainfall and snowmelt) and static parameters (e.g., lithology and topographic features) as model inputs, and thus the use of a revised machine learning model, which is capable of evaluating the spatiotemporal probability of landslides, then realizes the assessment and prediction of dynamic landslide susceptibility [32,33,34,35]. The model successfully overcomes the limitations of traditional static machine learning applications, and the results have the potential to be applied to early warning systems for disasters [34,35,36]. A CA-ANN model was used in conjunction with the temporal dynamics of land use/cover to forecast the size and geographical distribution of landslides [31]. Landslide forecasts of key triggers (temperature, precipitation) were used to show that climate projections can provide useful information about the possible impact of future climatic conditions on landslide susceptibility [2]. Despite breakthroughs in merging dynamic landslide susceptibility forecasts with climate change scenarios, knowledge gaps and challenges in predicting future landslide susceptibility were reported in some studies [37]. Future studies should concentrate on merging climatic projections with LULC data and using locally relevant modeling methodologies to increase the precision of LSM.
In this work, we employed the most recent climate change trends, predicted climate variables, and inventory of climate change-related landslides over the last 20 years based on the optimized machine learning model CF-LR (i) to identify key causative factors for landslides; (ii) to access the LSM for 2020; (iii) to simulate climate variables (2030, 2060); (iv) to project the LSM for 2030 and 2060 under the SSP1-2.6 scenario.

2. Study Area

After reviewing the literature, five bellwether sites were selected around the world where landslide activities are most sensitive to climate change and long-term, systematic observations could be obtained: Fairweather Range, North America; Northern Patagonia Icefield, South America; European Alps, Europe; Karakoram, Asia; and Aoraki/Mount Cook, Oceania [38]. In this study, we used Taxkorgan County, which is located in the eastern Pamir Plateau and the Karakoram, China, as the study area (Figure 1a) to study landslide susceptibility. This area is characterized by sparse vegetation [39], and extensive glaciers and permafrost (Figure 1c). These provide favorable conditions for landslides. In Taxkorgan County, from 2000 to 2019, there were a total of 258 landslides (138 slides and 120 debris flows) with areas greater than or equal to 0.5 km2 (Figure 1b), which were identified using historical archives, visual interpretations of Google Earth satellite images, optical image contrast, and field investigations. Overall, 90% of the large-sized debris flows induced by snow melting, glacier retreating, and freeze-thawing resulting from rising temperatures were concentrated in the area above the snow line, which is covered by glaciers and permafrost all year round (Figure 1c,d). In contrast, the small-sized slides (73.2%) were widespread in the region below the snow line and were triggered by rainfall [6].
In the past, the climate in Taxkorgan County has exhibited a warming and wetting trend, which has manifested as a result of increases in the temperature and maximum precipitation (Figure 2a) [40]. Spatially, the temperature and precipitation have exhibited elevation-dependent warming and wetting trends (Figure 1d). That is, the temperature and precipitation trends are amplified at higher elevations, especially in the region above the snow line (elevation of 4076.3 m). According to Figure 1d, the torrential rainfall has been primarily distributed in areas with elevations of less than 4000 m.
In climate scenario SSP1-2.6, the climate in the region continues to become warmer and wetter before 2060, exhibiting a positive warming rate and increasing precipitation trend (Figure 2b). Notably, the future rate of change in the temperature above the snow line (0.29 °C/decade) is slightly higher than the past rate of change (0.28 °C/decade), while the future warming rate below the snow line (0.25 °C/decade) is lower than the past warming rate in this region. The precipitation above the snow line in the region is greater than that below the snow line. However, compared to the trend of the maximum precipitation in this region in the past (7.63 mm/decade), the precipitation in the region, especially below the snow line (1.08 mm/decade), will decrease significantly in the future.

3. Materials and Methods

Identifying the primary factors that trigger landslides is crucial for assessing landslide susceptibility. However, there is currently a lack of standardized methodologies to determine these vital factors [41,42]. Our assessment framework for landslide susceptibility in Taxkorgan takes into account terrain, geomorphology, hydrology, climate change, and land use/cover, particularly sparse vegetation and low population density. We delineate fourteen potential landslide factors that are both static and dynamic: elevation, slope, topographic relief, aspect, Terrain Roughness Index (TRI), landform type, lithology, hydrogeology, LS factor, TWI, valley depth, distance to river, temperature, and precipitation (Figure 3). Initially, both static and dynamic factors were optimized; to achieve this, we applied the CF model and conducted susceptibility using the LR model to determine the key factors contributing to landslides. Subsequently, we calculated the spatial probability by generating a landslide map for 2020 with the LR model, which was validated using the area under the receiver operating characteristic (ROC) curve methodology. Finally, we employed a statistical downscaling model to project future climate variables (temperature, rainfall, and snowfall) under the SSP1-2.6 scenario based on the Coupled Model Intercomparison Project Phase 6 (CMIP6) dataset [43]. The LR model was also utilized to forecast the landslide susceptibility of the years 2030 and 2060. Figure 3 illustrates the methodological approach used in this study in a flowchart.
In this study, we conducted downscaling for the GLDAS dataset based on data from 10 observation stations to obtain climate variables for 1948–2020. Then, we simulated the temperature and precipitation for Taxkorgan from 2020 to 2100. This simulation was conducted using the SSP1-2.6 scenario and 27 climate models from the IPCC6 [43,44]. The detailed calculation processes for the climate variables and the projected climate variables can be found in the literature [5,6].
The SSPs represent a crucial enhancement within the CMIP6 framework. The SSP1-2.6 scenario aims to limit the temperature trend below 2.0 °C per decade by 2060, driven by national policy interventions; the SSP2-4.5 scenario predicts a temperature trend of approximately 2.7 °C per decade by 2030; and the SSP5-8.5 scenario illustrates an extreme case of no policy intervention [45]. Recently, countries worldwide have actively implemented energy-saving and emission-reduction policies in an effort to control the temperature trend below 2.0 °C per decade by 2060. Therefore, this paper focuses on predicting landslide susceptibility for the years 2030 and 2060 under the SSP1-2.6 scenario.

3.1. Data Sources

3.1.1. Static Data

Digital elevation models (DEMs) and optical images were acquired from the USGS network. Reference [46] and the disaster management department’s reports provided updated lithology data. We acquired the hydrogeology and geomorphology types from the Disaster Management Department of Taxkorgan. The 2016 China–Pakistan Economic Corridor Permafrost Dataset and the sixth edition of the Glacier Catalogue provided the glacier and permafrost data [47,48,49]. With SAGA-3.0 software (https://saga-gis.sourceforge.io/en/index.html, accessed on 18 September 2024), the topographic information (slope, relative relief, aspect, TRI, LS factor, TWI, valley depth, and distance to rivers) was extracted from the DEM. The landslide inventory was derived from the visual interpretation of optical images, the provision of the Disaster Management Department of Taxkorgan County, and field investigation. Well-known information on landslides and their surrounding environments extracted from factor data with a higher resolution reflects landslide development better [50]. However, the spatial resolutions of most variables used in LSM are inconsistent [51,52,53], which causes uncertainty in the assessment results [54]. To avoid the uncertainty linked to using datasets with different spatial resolutions, we resampled all of the causes and triggers of slides and debris flows and standardized their resolution to 30 m.

3.1.2. Dynamic Data

The dynamic data in this paper refer to meteorological data, including climate data (temperature and precipitation) for 1948–2019 and simulated climate scenario data (temperature, rainfall, and snowfall) for the years 2030 and 2060. The detailed calculation processes for the climate variables and the projected climate variables can be found in the literature [5,6].

3.2. Methodology

3.2.1. The Certainty of Landslide Key Causal Factors

The CF model can express the contribution of each classification of potential influencing factors to the landslide and simplify the mixing process of the debris flow [55,56]. The CF value is a function of the probability, and it can be expressed as follows:
C F = P a − P S P a 1 − P s   P a ≥ P S P a − P S P S 1 − P A   P a < P S
where Pa is the ratio of the area of the landslide in a classification of causative elements to the total area of the classification. Ps represents the likelihood of a landslide occurring in the research region. The CF value ranges between −1 and 1. When the CF value is more than 0, the likelihood of a landslide increases. Landslides are less likely to occur when CF is less than 0. When the CF value approaches 0, it is impossible to anticipate whether a landslide will occur [57,58].
Because the CF value represents the contribution of each grade of influencing factors for landslides, it is impossible to determine a factor’s contribution to the landslide. As a result, the CF values from each classification must be combined to generate a ZCF value that demonstrates the contribution of each factor to landslides. The calculation formula for the ZCF is as follows:
Z C F = x + y − x y ,   x ,   y ≥ 0 x + y 1 − m i n x , y , x y < 0 x + y + x y , x , y < 0
where x and y denote the two CF values being combined. A ZCF value greater than zero indicates that the factor is a key triggering factor for landslides, while a value less than zero indicates that it is not.

3.2.2. Landslide Susceptibility Assessment and Prediction

The LR model is a complex machine learning method used to evaluate dependent and independent variables [59,60]. In this method, individual susceptibility maps are constructed considering a dependent variable (e.g., flash floods, landslides, and mudflows), with 1 indicating landslides and 0 indicating non-landslides. The positive and negative LR coefficients indicate their impact on landslides and their role in landslide formation [61,62,63]. Specifically, a positive coefficient implies that the variable promotes the occurrence of landslides, while a negative coefficient indicates that the variable hinders the occurrence of landslides. The LR method is mathematically represented by the following equations [64]:
Z = A 0 + A 1 I 1 + A 2 I 2 + ⋯ + A i I i
P = e z 1 + e z
where A0 is a constant and is the intercept of the logistic regression equation, A is the binary regression coefficient, I is the index of the landslide sensitivity analysis, P is the probability of landslides, and Z is a linear combination.

4. Results

4.1. Landslide Potential Controlling Factors

4.1.1. Static Factors

Topography is a causative factor of landslides and plays a controlling role in their occurrence. Elevation is a fundamental aspect of analyzing landslides. In Taxkorgan, 98% of slides occurred below the snow line (H < 3400 m), while 95.7% of debris flows occurred above the snow line (4200–5400 m) (Figure 4a). The slope affects the sedimentation and accumulation of loose debris. Steep slopes are more prone to landslides. Slides and debris flow mainly occurred on slopes ranging from 10 to 60° (Figure 4b). The potential energy of slides and debris flows is determined by their relative relief, which ranges from 0 to 695 m/km2 (Figure 4c). However, the slides and debris flows were primarily concentrated in locations with a relative relief of less than 100 m/km2. The aspect affects hydrological processes and meteorological events, which in turn result in landslides (Figure 4d). The TRI shows the degree of regional ground relief. Landslides in the study area tended to occur in high/mid-rough terrain, where the TRI ranged from 7.1 to 21.1 (Figure 4e). Taxkorgan, which crosses the Pamirs and the Karakorum, has a complex landform type, with landslides primarily occurring in glacial erosion and denudation landscapes, valleys, and moraines in the alpine and subalpine zones (Figure 4f).
Lithology can reflect rock weathering and erosion while also influencing landforms. Figure 5a displays the distribution of rock hardness, revealing that slides were more distributed in softer rock regions and debris flows in hard rock. Figure 5b depicts the water yield property of Taxkorgan, which has a scarcity of groundwater, particularly fissure and pore water. Landslides were statistically more common in fissure water (water production of 1 L/s). The LS factor is an important element in estimating regional erosion. The TWI might reflect the soil’s moisture content. Larger valley depths may expand the slope area, providing materials and pathways for landslides. River erosion can destabilize hillsides, resulting in landslides. Landslides in the region are statistically dispersed in places with an LS factor of less than 19.8, a TWI of less than 14.9, valley depths ranging from 159.1 to 641.4 m, and river distances of less than 69.1 m (Figure 5c–f).

4.1.2. Dynamic Factors

This study focused on landslides caused by climate change. According to prior research [5,6], the triggering factors for slides and debris flows were heavy rainfall and maximum temperature, respectively. Increasing maximum temperatures cause glaciers to retreat, snow to melt, and freeze–thaw, resulting in debris flows. Figure 6 depicts the region’s highest temperature and precipitation over the last 30 years. Figure 6a shows that the bulk of debris flows occurred above the snow line, where the temperature was lower and the snowfall was higher. The increased rainfall below the snow line has exacerbated the occurrence of slides, with a maximum rainfall of 30 mm at the snow line over the last 30 years (Figure 6b).
In climate scenario SSP1-2.6, the climate variables were simulated for the years between 2030 and 2060. Temperature, rainfall, and snowfall have all shown an increasing trend in future predictions (Figure 7). The average annual maximum temperature, maximum rainfall, and maximum snowfall in 2060 increased to 25.7 °C, 65.40 mm, and 52.46 mm, respectively.

4.2. Optimized Key Factors for Landslides

Initially, the potential factors were reclassified in ArcGIS utilizing the natural breakpoint method, which was then overlaid with 80% of the slides and debris flows, respectively, to ascertain the key factors based on the ZCF value (Table 1). The key factors for slides included elevation, slope, groundwater, LS factor, TWI, valley depth, and maximum precipitation, as evidenced by their ZCF values exceeding 0. The regions with positive CF values had elevations below 3412 m, slopes between 10 and 40°, groundwater yields >1 L/S, TWI values ranging from 7.3 to 8.7, 10.4 to 12.3, and 14.9 to 18.5, LS factors from 3.7 to 12.4 and 16.1 to 19.8, and maximum precipitation > 24 mm. Elevation, slope, relative relief, aspect, LS factor, distance to the river, and maximum temperature were the key factors for debris flows. Positive CF values were found in places with elevations between 4598 and 5523 m, slopes ranging from 30 to 60°, relative reliefs from 34 to 79 m/km2, north and northeast aspects, LS factor > 16.1, distance to rivers < 177.6 m, and maximum temperatures between 9 and 15 °C.
Furthermore, the created random point tool was used to generate 200 non-slides and non-debris flows in the study area, which were then mixed with 80% of the slides and debris flows to produce slide and debris flow samples for the LR model. The key factors of the slides and debris flows were analyzed using a binary LR model of the SPSS-20.0 software (https://www.ibm.com/spss, accessed on 18 September 2024). The Sig values of the key factors were all less than 0.05, indicating that the binary LR model results met the 95% significance level test and were statistically significant (Table 1).

4.3. Landslide Susceptibility Analysis

4.3.1. Slide Susceptibility

Figure 8 depicts the slide susceptibility. The study area was divided into four categories based on normalized values: low susceptibility (<0.36), medium susceptibility (0.36–0.54), high susceptibility (0.54–0.71), and extremely high susceptibility (>0.71) (Figure 8a). In total, 75.4% of the area has low slide susceptibility. Only 19.5% of the region was very susceptible to slides, which were all located below the snow line (Figure 8b). Statistically, 82.6% of the slides were distributed in areas with medium, high, and very high susceptibilities. The accuracy for the slide susceptibility was 0.92 (Figure 8c), which was estimated using the remaining 20% of slides.

4.3.2. Susceptibility of Debris Flows

Figure 9 depicts the susceptibility of debris flows. According to the normalized susceptibility value, the study area was divided into four categories: low susceptibility (<0.28), medium susceptibility (0.28–0.47), high susceptibility (0.47–0.68), and very high susceptibility (>0.68) (Figure 9a). The low-susceptibility area of debris flows accounted for 79.1% of the study area. The high-susceptibility area accounted for 20.9%, which was all located above the snow line. Statistically, 84.8% of the debris flows were distributed in areas with high and very high susceptibility (Figure 9b). The accuracy for the debris flow susceptibility reached 0.92 (Figure 9c), verified by the remaining 20% of debris flows.

4.4. Landslide Susceptibility Prediction

4.4.1. Predicting Slide Susceptibility

Without considering the changes in the static factors of slides, we predicted the slide susceptibility in this region in 2030 and 2060 (Figure 10) based on the predicted maximum rainfall. In 2030, the area with very high susceptibility will account for 4.5% (Figure 10a), and by 2060, the area with high susceptibility will only account for 2.9% (Figure 10b). High and very high susceptibility areas will gradually decrease and expand eastward with climate change.

4.4.2. Predicting Susceptibility of Debris Flow

Similarly, we projected the susceptibility of debris flows in 2030 and 2060 (Figure 11) using the predicted maximum temperature and snowfall. In 2030, the area with very high susceptibility to debris flows will account for 12% (Figure 11a), whereas by 2060, it will account for 12.4% (Figure 11b). According to the statistics, the high and very high debris flow susceptibility zones will progressively increase, whereas the high susceptibility area will expand more quickly in the future.

5. Discussions

Climate change-induced slides, debris flows, floods, and avalanches have grown more common in alpine places around the world [45,65], which pose serious threats to people’s lives and property. As a result, it is vital to assess and predict the landslide susceptibility.
In this work, the CF-LR model was utilized to assess and forecast landslide susceptibility. The predicted climatic variables (temperature, rainfall, and snowfall) were used as driving factors for predicting future landslide maps. Landslide (slide and debris flow) susceptibility mapping for 2020, based on the last 20 years’ landslide inventory, was utilized to train and validate the model. This inventory demonstrates that slides below the snowline region were produced by heavy rainfall, and debris flows above the snowline were caused by glacier retreat, snow melt, and other temperature-induced processes. As a result, in the training and validation models, the maximum precipitation (heavy rainfall) and maximum temperature were employed as the drivers of slide and debris flow susceptibility, respectively. The climate in the region showed a warm and wet trend before 2060, and the wet phenomenon was mainly reflected by an increase in precipitation. However, precipitation refers to both solid and liquid precipitation, with liquid being rainfall below the snow line and solid being snowfall above the snow line. Rainfall can reflect the regional heavy rainfall, whereas snowfall can indicate the state of glacial snow accumulation above the snow line, which responds to the dynamic environment of debris flow formation. Therefore, this paper predicts debris flow using predicted snowfall and temperature, and rainfall is utilized to forecast slides. Additionally, we have considered terrain, geomorphology, hydrology, and lithology; the CF value was utilized to identify the key causes of landslides. The key causes of slides were elevation, slope, groundwater, LS factor, TWI, and valley depth. Elevation, slope, relative relief, aspect, LS factor, and distance to rivers were the key causes of debris flows. Elevation, slope, and LS factor were discovered to have roles in both slides and debris flows. The landslides were centered in regions with slopes ranging from 20 to 50°. On steep slopes, the gravity coefficient is higher, resulting in increased shear stress, making the slope’s surface material more prone to damage and creating a favorable environment for landslides [66]. Furthermore, groundwater and TWI can reflect soil dryness and wetness, influencing landslide susceptibilities [54]. Landslides tended to occur in areas with shaded slopes (north and northeast views), lower relative relief, and shorter distances from the river. This result is consistent with the previous findings [67,68,69,70].
Previous research has found that assessing landslide susceptibility using traditional models remains a significant difficulty [71,72,73]. Traditional and single-susceptibility models fail to identify the key causes of landslides, resulting in inaccurate landslide susceptibility [74,75,76]. Furthermore, most previous susceptibility maps solely considered static catastrophic factors [77]. However, in the case of climate change, the susceptibility map was assessed using static causative factors, which cannot effectively guide regional catastrophe prevention and mitigation nor enhance disaster contingency planning [65]. We assessed and projected landslide susceptibility in Taxkorgan County using climate variables. The future (2020–2060) will see a gradual increase in the susceptibility to debris flow in areas above the snow line. This is linked to an increasing temperature trend with elevation, which will exacerbate the region’s permafrost degradation, glacier retreat, and snowmelt [78]. As a result, there will be more debris flows in areas above the snowline [79]. Below the snow line, slide susceptibility gradually decreases. This is despite the fact that an increase in the westerly wind index (WI) enhances the warm and wet airflow from the North Atlantic and the Caspian Sea, exacerbating the frequency of rainfall-induced slides in the region below the snow line [80,81,82]. However, slide susceptibility is more related to its size; therefore, a high number of slides does not always translate into a high slide susceptibility in the area [6,83]. In the future, we should pay more attention to large-scale landslides above the snow line caused by snow melting, glacier retreat, and ice collapse, which may be more destructive. Typical examples are the glacial debris flow that occurred in the Bulunkou Township [84], the snowmelt-induced landslides in Taheman Township on the Pamir Plateau [85,86], and the catastrophic landslides that occurred in Chamoli, India, at the southern foothills of the Himalayas on 7 February 2021 [87], which all caused significant damage to human life and property.

6. Conclusions

Taxkorgan County is a bridge for the China–Pakistan Economic Corridor (CPEC) project. Landslides are numerous and widespread. The frequent landslides hinder CPEC construction and threaten the locals’ safety and livelihoods. Recently, climate change-related landslides have increased in alpine regions, particularly during the warm season. The scientific and accurate assessment and prediction of landslide susceptibility can provide a scientific reference for disaster prevention, land-use planning, and CPEC construction. In this paper, we focused on the static and dynamic environment factors (temperature and precipitation) and used the CF-LR model to evaluate and predict landslide susceptibility in Taxkorgan County. The assessment and prediction were based on a catalog of climate change-related landslides over the past 20 years, the causative factors, and predicted climatic variables for the SSP1-2.6 scenario. The results indicated that elevation, slope, groundwater, LS factor, TWI, valley depth, and maximum precipitation were the key causes of slides below the snow line. The key causing factors of debris flow above the snow line were elevation, slope, topographic relief, aspect, LS factor, distance to the river, and maximum temperature. The accuracy of slide and debris flow susceptibility was 0.92 and 0.89, respectively. The area of slides with medium, high, and very high susceptibility is 25.5% of the total area. In addition, 82.6% of the slides occurred in this region, and 49.5% of the entire area is covered by debris flows with medium, high, and very high susceptibility. Moreover, this area accounts for 91.8% of all debris flows. Until 2060, the region’s climate is anticipated to become warmer and wetter. Slides below the snow line will gradually decrease and shift eastward, and debris flows above the snow line will expand. Our findings will contribute to the management of landslide risks at the regional scale.

Author Contributions

Conceptualization, H.Q.; methodology, Y.P.; software, Y.P. and Y.Z.; investigation, Y.P. and Y.Z.; writing—original draft preparation, Y.P.; writing—review and editing, H.Q.; funding acquisition, H.Q. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the International Science and Technology Cooperation Program of China (Grant No. 2018YFE0100100).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original contributions presented in the study are included in the article, and further inquiries can be directed to the corresponding authors.

Acknowledgments

We acknowledge the reviewers and academic editors for their positive and constructive comments and suggestions. The authors also thank Engineer Zhiyuang Feng from the Administration of Natural Resources of Kashgar for his assistance in the fieldwork.

Conflicts of Interest

The authors declare no conflicts of interest.

References

  1. Petley, D. Global patterns of loss of life from landslides. Geology 2012, 40, 927–930. [Google Scholar] [CrossRef]
  2. Gariano, S.L.; Guzzetti, F. Landslides in a changing climate. Earth Sci. Rev. 2016, 162, 227–252. [Google Scholar] [CrossRef]
  3. An, H.; Viet, T.T.; Lee, G.; Kim, Y.; Kim, M.; Noh, S.; Noh, J. Development of time-variant landslide-prediction software considering three-dimensional subsurface unsaturated flow. Environ. Model. Softw. 2016, 85, 172–183. [Google Scholar] [CrossRef]
  4. Huggel, C.; Clague, J.J.; Korup, O. Is climate change responsible for changing landslide activity in high mountains? Earth Surf. Process. Landf. 2012, 37, 77–91. [Google Scholar] [CrossRef]
  5. Pei, Y.; Qiu, H.; Yang, D.; Liu, Z.; Ma, S.; Li, J.; Cao, M.; Wufuer, W. Increasing landslide activity in the Taxkorgan River Basin (eastern Pamirs Plateau, China) driven by climate change. Catena 2023, 223, 106911. [Google Scholar] [CrossRef]
  6. Pei, Y.Q.; Qiu, H.J.; Zhu, Y.R.; Wang, J.D.; Yang, D.D.; Tang, B.Z.; Wang, F.; Cao, M.M. Elevation dependence of landslide activity induced by climate change in the eastern Pamirs. Landslides 2023, 20, 1115–1133. [Google Scholar] [CrossRef]
  7. Froude, M.J.; Petley, D.N. Global fatal landslide occurrence from 2004 to 2016. Nat. Hazards Earth Syst. Sci. 2018, 18, 2161–2181. [Google Scholar] [CrossRef]
  8. Zhang, Y.; Gu, J.; Liu, S.Y.; Wang, X.; Jiang, Z.L.; Wei, J.F.; Zheng, Y.J. Spatial pattern of the debris-cover effect and its role in the Hindu Kush-Pamir-Karakoram-Himalaya glaciers. J. Hydrol. 2022, 615, 128613. [Google Scholar] [CrossRef]
  9. Qiu, H.J.; Su, L.L.; Tang, B.Z.; Yang, D.D.; Ullah, M.; Zhu, Y.R.; Kamp, U. The effect of location and geometric properties of landslides caused by rainstorms and earthquakes. Earth Surf. Process. Landf. 2024, 49, 2067–2079. [Google Scholar] [CrossRef]
  10. Fell, R.; Corominas, J.; Bonnard, C.; Cascini, L.; Leroi, E.; Savage, W.Z. Guidelines for landslide susceptibility, hazard and risk zoning for land use planning. Eng. Geol. 2008, 102, 85–98. [Google Scholar] [CrossRef]
  11. Pawluszek, K.; Borkowski, A. Impact of DEM-derived factors and analytical hierarchy process on landslide susceptibility mapping in the region of Rożnów Lake, Poland. Nat. Hazards 2017, 86, 919–952. [Google Scholar] [CrossRef]
  12. El Jazouli, A.; Barakat, A.; Khellouk, R. GIS-multicriteria evaluation using AHP for landslide susceptibility mapping in Oum Er Rbia high basin (Morocco). Geoenvironmental Disasters 2019, 6, 3. [Google Scholar] [CrossRef]
  13. Li, D.F.; Lu, X.X.; Walling, D.; Zhang, T.; Steiner, J.; Wasson, R.; Harrison, S.; Nepal, S.; Nie, Y.; Immerzeel, W.; et al. High Mountain Asia hydropower systems threatened by climate-driven landscape instability. Nat. Geosci. 2022, 15, 520–530. [Google Scholar] [CrossRef]
  14. Dai, F.C.; Lee, C.F. Landslide characteristics and slope instability modeling using GIS, Lantau Island, Hong Kong. Geomorphology 2002, 42, 213–228. [Google Scholar] [CrossRef]
  15. Tyagi, A.; Tiwari, R.K.; James, N. Prediction of the future landslide susceptibility scenario based on LULC and climate projections. Landslides 2023, 20, 1837–1852. [Google Scholar] [CrossRef]
  16. Devara, M.; Tiwari, A.; Dwivedi, R. Landslide susceptibility mapping using MT-InSAR and AHP enabled GIS-based multi-criteria decision analysis. Geomat. Nat. Hazards Risk 2021, 12, 675–693. [Google Scholar] [CrossRef]
  17. Gantimurova, S.; Parshin, A.; Erofeev, V. GIS-Based Landslide Susceptibility Mapping of the Circum-Baikal Railway in Russia Using UAV Data. Remote Sens. 2021, 13, 3629. [Google Scholar] [CrossRef]
  18. Hong, H. Assessing landslide susceptibility based on hybrid Best-first decision tree with ensemble learning model. Ecol. Ind. 2023, 147, 109968. [Google Scholar] [CrossRef]
  19. Medina, V.; Hürlimann, M.; Guo, Z.; Lloret, A.; Vaunat, J. Fast physically-based model for rainfall-induced landslide susceptibility assessment at regional scale. Catena 2021, 201, 105213. [Google Scholar] [CrossRef]
  20. Huang, F.; Cao, Z.; Guo, J.; Jiang, S.H.; Li, S.; Guo, Z. Comparisons of heuristic, general statistical and machine learning models for landslide susceptibility prediction and mapping. Catena 2020, 191, 104580. [Google Scholar] [CrossRef]
  21. Steger, S.; Mair, V.; Kofler, C.; Pittore, M.; Zebisch, M.; Schneiderbauer, S. Correlation does not imply geomorphic causation in data-driven landslide susceptibility modelling-Benefits of exploring landslide data collection effects. Sci. Total Environ. 2021, 776, 145935. [Google Scholar] [CrossRef] [PubMed]
  22. Gayen, A.; Pourghasemi, H.; Saha, S.; Keesstra, S.; Bai, S. Gully erosion susceptibility assessment and management of hazard-prone areas in India using different machine learning algorithms. Sci. Total Environ. 2019, 668, 124–138. [Google Scholar] [CrossRef] [PubMed]
  23. Islam, A.; Talukdar, S.; Mahato, S.; Kundu, S.; Eibek, K.; Pham, Q.; Kuriqi, A.; Linh, N.T. Flood susceptibility modelling using advanced ensemble machine learning models. Geosci. Front. 2021, 12, 101075. [Google Scholar] [CrossRef]
  24. Bui, D.; Tsangaratos, P.; Nguyen, V.; Liem, N.; Trinh, P.T. Comparing the prediction performance of a Deep Learning Neural Network model with conventional machine learning models in landslide susceptibility assessment. Catena 2020, 188, 104426. [Google Scholar] [CrossRef]
  25. Hewitt, K. Rock avalanches that travel onto glaciers and related developments, Karakoram Himalaya. Geomorphology 2009, 103, 66–79. [Google Scholar] [CrossRef]
  26. Chiarle, M.; Geertsema, M.; Mortara, G.; Clague, J.J. Relations between climate change and mass movement: Perspectives from the Canadian Vordillera and the European Alps. Glob. Planet. Change 2021, 25, 103499. [Google Scholar] [CrossRef]
  27. Wei, Y.D.; Qiu, H.J.; Liu, Z.J.; Huangfu, W.C.; Zhu, Y.R.; Liu, Y.; Yang, D.D.; Kamp, U. Refined and dynamic susceptibility assessment of landslides using InSAR and machine learning models. Geosci. Front. 2024, 15, 101890. [Google Scholar] [CrossRef]
  28. Casagli, N.; Frodella, W.; Morelli, S.; Tofani, V.; Ciampalini, A.; Intrieri, E.; Raspini, F.; Rossi, G.; Tanteri, L.; Lu, P. Spaceborne, UAV and ground-based remote sensing techniques for landslide mapping, monitoring and early warning. Geoenvironmental Disasters 2017, 4, 9. [Google Scholar] [CrossRef]
  29. Zhu, Y.R.; Qiu, H.J.; Liu, Z.J.; Ye, B.F.; Tang, B.Z.; Li, Y.J.; Kamp, U. Rainfall and water level fluctuations dominated the landslide deformation at Baihetan Reservoir, China. J. Hydrol. 2024, 642, 131871. [Google Scholar] [CrossRef]
  30. Chang, Z.L.; Du, Z.; Zhang, F.; Huang, F.M.; Chen, J.W.; Li, W.B.; Guo, Z.Z. Landslide Susceptibility Prediction Based on Remote Sensing Images and GIS: Comparisons of Supervised and Unsupervised Machine Learning Models. Remote Sens. 2020, 12, 502. [Google Scholar] [CrossRef]
  31. Rahman, M.; Tabassum, F.; Rasheduzzaman, M.; Saba, H.; Sarkar, L.; Ferdous, J.; Uddin, S.Z.; Islam, A.Z.M.Z. Temporal dynamics of land use/land cover change and its prediction using CA-ANN model for southwestern coastal Bangladesh. Environ. Monit. Assess. 2017, 189, 565. [Google Scholar] [CrossRef]
  32. Stanley, T.A.; Kirschbaum, D.B.; Benz, G.; Emberson, R.A.; Amatya, P.M.; Medwedeff, W.; Clark, M.K. Data-Driven Landslide Nowcasting at the Global Scale. Front. Earth Sci. 2021, 9, 640043. [Google Scholar] [CrossRef]
  33. Li, B.H.; Liu, K.; Wang, M.; He, Q.J.; Jiang, Z.Y.; Zhu, W.H.; Qiao, N.N. Global Dynamic Rainfall-Induced Landslide Susceptibility Mapping Using Machine Learning. Remote Sens. 2022, 14, 5795. [Google Scholar] [CrossRef]
  34. Fang, Z.C.; Wang, Y.; Westen, C.J.V.; Lombardo, L. Landslide hazard spatiotemporal prediction based on data-driven models: Estimating where, when and how large landslide may be. Int. J. Appl. Earth Obs. Geoinf. 2024, 126, 103631. [Google Scholar] [CrossRef]
  35. Nocentini, N.; Rosi, A.; Piciullo, L.; Liu, Z.Q.; Segoni, S.; Fanti, R. Regional-scale spatiotemporal landslide probability assessment through machine learning and potential applications for operational warning systems: A case study in Kvam (Norway). Landslides 2024, 21, 2369–2387. [Google Scholar] [CrossRef]
  36. Cui, Y.F.; Li, Y.; Tang, H.; Turowski, J.M.; Yan, Y.; Bazai, N.A.; Wei, R.L.; Li, L. A digital-twin platform for cryospheric disaster warning. Natl. Sci. Rev. 2024, nwae300. [Google Scholar] [CrossRef]
  37. Maynard, J.; Hooidonk, R.; Eakin, C.M.; Puotinen, M.; Garren, M.; Williams, G.J.; Heron, S.; Lamb, J.B.; Weil, E.; Willis, B.; et al. Projections of climate conditions that increase coral disease susceptibility and pathogen abundance and virulence. Nat. Clim. Change 2015, 5, 688–694. [Google Scholar] [CrossRef]
  38. Coe, J.A. Bellwether sites for evaluating changes in landslide frequency and magnitude in cryospheric mountainous terrain: A call for systematic, long-term observations to decipher the impact of climate change. Landslides 2020, 17, 2483–2501. [Google Scholar] [CrossRef]
  39. Chen, F.H.; Yu, Z.C.; Yang, M.L.; Wang, S.M.; Madsen, D.B.; Huang, X.Z.; Zhao, Y.; Sato, T.; Birks, H.J.B.; Boomer, I.; et al. Holocene moisture evolution in arid central Asia and its out-of-phase relationship with Asian monsoon history. Quat. Sci. Rev. 2008, 27, 351–364. [Google Scholar] [CrossRef]
  40. Yao, J.; Chen, Y.; Guan, X.; Zhao, Y.; Chen, J.; Mao, W. Recent climate and hydrological changes in a mountain-basin system in Xinjiang, China. Earth Sci. Rev. 2022, 226, 103957. [Google Scholar] [CrossRef]
  41. Ayalew, L.; Yamagishi, H. The application of GIS-based logistic regression for landslide susceptibility mapping in the Kakuda-Yahiko Mountains Central Japan. Geomorphology 2005, 65, 15–31. [Google Scholar] [CrossRef]
  42. Reichenbach, P.; Rossi, M.; Malamud, B.D.; Mihir, M.; Guzzetti, F. A review of statistically-based landslide susceptibility models. Earth Sci. Rev. 2018, 180, 60–91. [Google Scholar] [CrossRef]
  43. Grose, M.R.; Narsey, S.; Delage, F.P.; Dowdy, A.J.; Bador, M.; Boschat, G.; Chung, C.; Kajtar, J.B.; Sauniyar, S.; Freund, M.B.; et al. Insights from CMIP6 for Australia’s future climate. Earth’s Future 2020, 8, 480–504. [Google Scholar] [CrossRef]
  44. Kirschbaum, D.; Kapnick, S.B.; Stanley, T.; Pascale, S. Changes in extreme precipitation and landslides over high mountain Asia. Geophys. Res. Lett. 2020, 47, e2019GL085347. [Google Scholar] [CrossRef]
  45. Allan, R.P.; Hawkins, E.; Bellouin, N.; Collins, B. Summary for Policymakers; IPCC: Geneva, Switzerland, 2021. [Google Scholar]
  46. Robinson, A.C.; Yin, A.; Manning, C.E.; Harrison, T.M.; Zhang, S.H.; Wang, X.F. Cenozoic evolution of the eastern Pamir: Implications for strain-accommodation mechanisms at the western end of the Himalayan-Tibetan orogen. Geol. Soc. Am. Bull. 2007, 119, 882–896. [Google Scholar] [CrossRef]
  47. Zhang, G.Q.; Yao, T.D.; Xie, H.J.; Wang, W.C.; Yang, W. An inventory of glacial lakes in the Third Pole region and their changes in response to global warming. Glob. Planet. Change 2015, 131, 148–157. [Google Scholar] [CrossRef]
  48. Ai, M.H.; Zhang, Y.N.; Kang, J.F.; Feng, K.T.; Tian, D.Y. Dataset of China-Pakistan Economic Corridor Permafrost Distribution in 2016; Science Data Bank: Beijing, China, 2018. [Google Scholar]
  49. Rgi, C.; Nosenko, G. Randolph Glacier Inventory (RGI)-A Dataset of Global Glacier Outlines: Version6.0. Technical Report; In Global Land Ice Measurements from Space; NASA: Washington, DC, USA, 2017. [Google Scholar]
  50. Ma, L.; Liu, Y.; Zhang, X.; Ye, Y.; Yin, G.; Johnson, B.A. Deep learning in remote sensing applications: A meta-analysis and review. ISPRS J. Photogramm. Remote Sens. 2019, 152, 166–177. [Google Scholar] [CrossRef]
  51. Catani, F.; Lagomarsino, D.; Segoni, S.; Tofani, V. Landslide susceptibility estimation by random forests technique: Sensitivity and scaling issues. Nat. Hazards Earth Syst. Sci. 2013, 13, 2815–2831. [Google Scholar] [CrossRef]
  52. Costache, R.; Popa, M.C.; Bui, D.T.; Diaconu, D.C.; Ciubotaru, N.; Minea, G.; Pham, Q.B. Spatial predicting of flood potential areas using novel hybridizations of fuzzy decision-making, bivariate statistics, and machine learning. J. Hydrol. 2020, 585, 124808. [Google Scholar] [CrossRef]
  53. Rahman, M.; Chen, N.; Elbeltagi, A.; Islam, M.M.; Alam, M.; Pourghasemi, H.R.; Tao, W.; Zhang, J.; Shufeng, T.; Faiz, H. Application of stacking hybrid machine learning algorithms in delineating multi-type flooding in Bangladesh. J. Environ. Manag. 2021, 295, 113086. [Google Scholar] [CrossRef]
  54. Fang, Z.; Wang, Y.; Peng, L.; Hong, H. Integration of convolutional neural network and conventional machine learning classifiers for landslide susceptibility mapping. Comput. Geosci. 2020, 139, 104470. [Google Scholar] [CrossRef]
  55. Pourghasemi, H.R.; Pradhan, B.; Gokceoglu, C.; Mohammadi, M.; Moradi, H.R. Application of weights-of evidence and certainty factor models and their comparison in landslide susceptibility mapping at haraz watershed, Iran. Arab. J. Geosci. 2012, 6, 2351–2365. [Google Scholar] [CrossRef]
  56. Angillieri, M.Y.E. Debris flow susceptibility mapping using frequency ratio and seed cells, in a portion of a mountain international route, Dry Central Andes of Argentina. Catena 2020, 189, 104504. [Google Scholar] [CrossRef]
  57. Gorum, T.; Gonencgil, B.; Gokceoglu, C.; Nefeslioglu, H.A. Implementation of reconstructed geomorphologic units in landslide susceptibility mapping: The Melen Gorge (NW Turkey). Nat. Hazards 2008, 46, 323–351. [Google Scholar] [CrossRef]
  58. Qiu, C.C.; Su, L.J.; Zou, Q.; Geng, X.Y. A hybrid machine-learning model to map glacier-related debris flow susceptibility along Gyirong Zangbo watershed under the changing climate. Sci. Total Environ. 2022, 818, 151752. [Google Scholar] [CrossRef]
  59. Ghosh, A.; Maiti, R. Soil erosion susceptibility assessment using logistic regression, decision tree and random forest: Study on the Mayurakshi river basin of Eastern India. Environ. Earth Sci. 2021, 80, 328. [Google Scholar] [CrossRef]
  60. Goyes-Peñafiel, P.; Hernandez-Rojas, A. Landslide susceptibility index based on the integration of logistic regression and weights of evidence: A case study in Popayan, Colombia. Eng. Geol. 2021, 280, 105958. [Google Scholar] [CrossRef]
  61. Yilmaz, I. Landslide susceptibility mapping using frequency ration, logistic regression, artificial neural networks and their comparison: A case study from Kat landslides (Tokat—Turkey). Comput. Geosci. 2009, 35, 1125–1138. [Google Scholar] [CrossRef]
  62. Yalcin, A.; Reis, S.; Aydinoglu, A.C.; Yomralioglu, T. A GIS-based comparative study of frequency ratio, analytical hierarchy process, bivariate statistics and logistics regression methods for landslide susceptibility mapping in Trabzon, NE Turkey. Catena 2011, 85, 274–287. [Google Scholar] [CrossRef]
  63. Niu, Q.F.; Dang, X.H.; Li, Y.F.; Zhang, Y.X.; Lu, X.L.; Gao, W.X. Suitability analysis for topographic factors in loess landslide research: A case study of Gangu County, China. Environ. Earth Sci. 2018, 77, 294–306. [Google Scholar] [CrossRef]
  64. Ahmad, H.; Ningsheng, C.; Rahman, M.; Islam, M.M.; Pourghasemi, H.R.; Hussain, S.F.; Habumugisha, J.M.; Liu, E.; Zheng, H.; Ni, H.; et al. Geohazards susceptibility assessment along the upper Indus basin using four machine learning and statistical models. ISPRS Int. J. Geo-Inf. 2021, 10, 315. [Google Scholar] [CrossRef]
  65. Harmeling, S. Global Climate Risk Index 2021. Climate Change and Law Collection; Germanwatch: Bonn, Germany, 2022; p. 12. [Google Scholar]
  66. Kritikos, T.; Robinson, T.R.; Davies, T.R.H. Regional coseismic landslide hazard assessment without historical landslide inventories: A new approach. J. Geophys. Res. Earth Surf. 2015, 120, 711–729. [Google Scholar] [CrossRef]
  67. Marra, F.; Destro, E.; Nikolopoulos, E.I.; Zoccatelli, D.; Creutin, J.D.; Guzzetti, F.; Borga, M. Impact of rainfall spatial aggregation on the identification of debris flow occurrence thresholds. Hydrol. Earth Syst. Sci. 2017, 21, 4525–4532. [Google Scholar] [CrossRef]
  68. Wang, Y.; Fang, Z.; Hong, H. Comparison of convolutional neural networks for landslide susceptibility mapping in Yanshan County, China. Sci. Total Environ. 2019, 666, 975–993. [Google Scholar] [CrossRef]
  69. Wu, S.; Chen, J.; Zhou, W.; Iqbal, J.; Yao, L. A modified Logit model for assessment and validation of debris-flow susceptibility. Bull. Eng. Geol. Environ. 2019, 78, 4421–4438. [Google Scholar] [CrossRef]
  70. Luo, J.; Niu, F.; Lin, Z.; Liu, M.; Yin, G.; Gao, Z. Inventory and frequency of retrogressive thaw slumps in permafrost region of the Qinghai–Tibet Plateau. Geophys. Res. Lett. 2022, 49, e2022GL099829. [Google Scholar] [CrossRef]
  71. Takahashi, T. Debris Flow: Mechanics, Prediction and Countermeasures; Taylor & Francis: Abingdon, UK, 2007. [Google Scholar]
  72. Kappes, M.S.; Keiler, M.; Glade, T. From Single-to Multi-Hazard Risk Analyses: A Concept Addressing Emerging Challenges; CERG, Ed.; Mountain Risks: Firenze, Italy, 2010; pp. 351–356. [Google Scholar]
  73. Erokhin, S.A.; Zaginaev, V.V.; Meleshko, A.A.; Ruiz-Villanueva, V.; Petrakov, D.A.; Chernomorets, S.S.; Viskhadzhieva, K.S.; Tutubalina, O.V.; Stoffel, M. Debris flows triggered from nonstationary glacier lake outbursts: The case of the Teztor Lake complex (Northern Tian Shan, Kyrgyzstan). Landslides 2018, 15, 83–98. [Google Scholar] [CrossRef]
  74. Yilmaz, I. Comparison of landslide susceptibility mapping methodologies for Koyulhisar, Turkey: Conditional probability, logistic regression, artificial neural networks, and support vector machine. Environ. Earth Sci. 2010, 61, 821–836. [Google Scholar] [CrossRef]
  75. Chen, X.C.; Chen, H.; You, Y.; Liu, J. Susceptibility assessment of debris flows using the analytic hierarchy process method-a case study in Subao river valley, China. J. Rock Mech. Geotech. Eng. 2015, 7, 404–410. [Google Scholar] [CrossRef]
  76. Xue, R.; Wang, C.; Liu, M.L.; Zhang, D.; Li, K.L.; Li, N. A new method for soil health assessment based on analytic hierarchy process and meta-analysis. Sci. Total Environ. 2019, 650, 2771–2777. [Google Scholar] [CrossRef]
  77. Ullah, K.; Wang, Y.; Fang, Z.; Wang, L.; Rahman, M. Multi-hazard susceptibility mapping based on convolutional neural networks. Geosci. Front. 2022, 13, 101425. [Google Scholar] [CrossRef]
  78. Yan, L.; Liu, X. Has climatic warming over the Tibetan Plateau paused or continued in recent years? Earth Ocean Atmos. Sci. 2014, 1, 13–28. [Google Scholar]
  79. Seneviratne, S.I. Changes in Climate Extremes and Their Impacts on the Natural Physical Environment; Cambridge University Press: Cambridge, UK, 2012; pp. 109–230. [Google Scholar]
  80. Schär, C.; Frei, C.; Lüthi, D.; Davies, H.C. Surrogate climate change scenarios for regional climate models. Geophys. Res. Lett. 1996, 23, 669–672. [Google Scholar] [CrossRef]
  81. Dai, X.G.; Wang, P. Zonal mean mode of global warming over the past 50 years. Atmos. Ocean. Sci. Lett. 2010, 3, 45–50. [Google Scholar]
  82. Yao, J.Q.; Chen, Y.N.; Yang, Q. Spatial and temporal variability of water vapor pressure in the arid region of northwest China, during 1961–2011. Theor. Appl. Climatol. 2015, 123, 683–691. [Google Scholar] [CrossRef]
  83. Li, Y.; Cui, Y.; Hu, X.; Lu, Z.; Guo, J.; Wang, Y.; Wang, H.; Wang, S.F.; Zhou, X.Z. Glacier retreat in Eastern Himalaya drives catastrophic glacier hazard chain. Geophys. Res. Lett. 2024, 51, e2024GL108202. [Google Scholar] [CrossRef]
  84. Jiang, N.; Su, F.; Li, Y.; Guo, X.; Liu, X. Debris flow assessment in the Gaizi-Bulunkou section of Karakoram Highway. Front. Earth Sci. 2012, 9, 660579. [Google Scholar]
  85. Seong, Y.B.; Owen, L.A.; Yi, C.; Finkel, R.C.; Schoenbohm, L. Geomorphology of anomalously high glaciated mountains at the northwestern end of Tibet: Muztag Ata and Kongur Shan. Geomorphology 2009, 103, 227–250. [Google Scholar] [CrossRef]
  86. Yuan, Z.D.; Chen, J.; Owen, L.A.; Hedrick, K.A.; Caffee, M.W.; Li, W.Q.; Schoenbohm, L.M.; Robinson, A.C. Nature and timing of large landslides within an active orogen, eastern Pamir, China. Geomorphology 2013, 182, 49–65. [Google Scholar] [CrossRef]
  87. Jiang, R.; Zhang, L.; Peng, D.; He, X.; He, J. The landslide hazard chain in the Tapovan of the Himalayas on 7 February 2021. Geophys. Res. Lett. 2021, 48, e2021GL093723. [Google Scholar] [CrossRef]
Applsci 14 08562 g001
Figure 1. Overview of the study area. (a) The location of Taxkorgan County; (b) temporal distribution of landslides over the past 20 years; (c) spatial distribution of landslides. The photos were taken during field investigations; (d) landslide size, days of torrential rainfall, and trends in climate within different elevation intervals (revised from the literature [5,6]).
Figure 1. Overview of the study area. (a) The location of Taxkorgan County; (b) temporal distribution of landslides over the past 20 years; (c) spatial distribution of landslides. The photos were taken during field investigations; (d) landslide size, days of torrential rainfall, and trends in climate within different elevation intervals (revised from the literature [5,6]).
Applsci 14 08562 g001
Applsci 14 08562 g002
Figure 2. Climate change in the study area. (a) Previous climate change; (b) anticipated changes in climate (revised from the literature [5,6]).
Figure 2. Climate change in the study area. (a) Previous climate change; (b) anticipated changes in climate (revised from the literature [5,6]).
Applsci 14 08562 g002
Applsci 14 08562 g003
Figure 3. The proposed workflow chart of this manuscript.
Figure 3. The proposed workflow chart of this manuscript.
Applsci 14 08562 g003
Applsci 14 08562 g004
Figure 4. Distribution of landslides in the topography and landform.
Figure 4. Distribution of landslides in the topography and landform.
Applsci 14 08562 g004
Applsci 14 08562 g005
Figure 5. Distribution of landslides in geology and hydrology conditions.
Figure 5. Distribution of landslides in geology and hydrology conditions.
Applsci 14 08562 g005
Applsci 14 08562 g006
Figure 6. Maximum temperature, maximum precipitation, and landslide distribution.
Figure 6. Maximum temperature, maximum precipitation, and landslide distribution.
Applsci 14 08562 g006
Applsci 14 08562 g007
Figure 7. Predicted temperature, rainfall, and snowfall under the climate scenario of SSP1-2.6.
Figure 7. Predicted temperature, rainfall, and snowfall under the climate scenario of SSP1-2.6.
Applsci 14 08562 g007
Applsci 14 08562 g008
Figure 8. Susceptibility of slides. (a) The distribution result of the susceptibility in each grade by the LR model. (b) The statistical proportion of slide area and each susceptibility grade area. (c) The evaluated accuracy of slide susceptibility. ROC curve: receiver operating characteristic curve; AUC: area under curve.
Figure 8. Susceptibility of slides. (a) The distribution result of the susceptibility in each grade by the LR model. (b) The statistical proportion of slide area and each susceptibility grade area. (c) The evaluated accuracy of slide susceptibility. ROC curve: receiver operating characteristic curve; AUC: area under curve.
Applsci 14 08562 g008
Applsci 14 08562 g009
Figure 9. Susceptibility of debris flows. (a) The distribution result of the susceptibility in each grade by the LR model. (b) The statistical proportion of debris flow area and each susceptibility grade area. (c) The evaluated accuracy of debris flow susceptibility.
Figure 9. Susceptibility of debris flows. (a) The distribution result of the susceptibility in each grade by the LR model. (b) The statistical proportion of debris flow area and each susceptibility grade area. (c) The evaluated accuracy of debris flow susceptibility.
Applsci 14 08562 g009
Applsci 14 08562 g010
Figure 10. Predicted slide susceptibility. (a) Slide susceptibility in 2030. (b) The area proportion of each susceptibility grade in 2030. (c) Slide susceptibility in 2060. (d) The area proportion of each susceptibility grade in 2060.
Figure 10. Predicted slide susceptibility. (a) Slide susceptibility in 2030. (b) The area proportion of each susceptibility grade in 2030. (c) Slide susceptibility in 2060. (d) The area proportion of each susceptibility grade in 2060.
Applsci 14 08562 g010
Applsci 14 08562 g011
Figure 11. Predicted debris flow susceptibility. (a) Debris flow susceptibility in 2030. (b) The area proportion of each susceptibility grade in 2030. (c) Debris flow susceptibility in 2060. (d) The area proportion of each susceptibility grade in 2060.
Figure 11. Predicted debris flow susceptibility. (a) Debris flow susceptibility in 2030. (b) The area proportion of each susceptibility grade in 2030. (c) Debris flow susceptibility in 2060. (d) The area proportion of each susceptibility grade in 2060.
Applsci 14 08562 g011
Table 1. ZCF and LR regression coefficients of key triggering factors for slide and debris flow. β denotes the regression coefficient, S.Emeans standard error, and CI indicates confidence interval.
Table 1. ZCF and LR regression coefficients of key triggering factors for slide and debris flow. β denotes the regression coefficient, S.Emeans standard error, and CI indicates confidence interval.
SlidesDebris Flows
FactorZCFβS.E95%CI. of Exp(B)FactorZCFβS.E95%CI. of Exp(B)
LowerUpperLowerUpper
Elevation (m)0.74−0.780.19−1.37−0.42Elevation (m)0.870.550.220.131.06
Slope (°)0.780.210.25−0.280.87Slope (°)0.670.140.31−0.500.82
Hydrogeology0.14−0.090.27−0.700.46Relative relief0.43−0.010.25−0.560.54
LS factor0.52−0.110.23−0.700.39Aspect0.59−0.090.07−0.250.06
TWI0.820.070.13−0.230.34LS factor0.320.050.18−0.35−0.45
Valley depth0.47−0.320.110.150.59Distance to river0.79−0.430.14−0.76−0.21
Pmax (mm)0.800.250.22−0.160.79Tmax0.910.320.21−0.080.78
Constant−−0.941.09−3.131.42Constant−−3.692.08−8.180.11
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Pei, Y.; Qiu, H.; Zhu, Y. Prediction of Landslide Susceptibility in the Karakorum under the Context of Climate Change. Appl. Sci. 2024, 14, 8562. https://doi.org/10.3390/app14188562

AMA Style

Pei Y, Qiu H, Zhu Y. Prediction of Landslide Susceptibility in the Karakorum under the Context of Climate Change. Applied Sciences. 2024; 14(18):8562. https://doi.org/10.3390/app14188562

Chicago/Turabian Style

Pei, Yanqian, Haijun Qiu, and Yaru Zhu. 2024. "Prediction of Landslide Susceptibility in the Karakorum under the Context of Climate Change" Applied Sciences 14, no. 18: 8562. https://doi.org/10.3390/app14188562

APA Style

Pei, Y., Qiu, H., & Zhu, Y. (2024). Prediction of Landslide Susceptibility in the Karakorum under the Context of Climate Change. Applied Sciences, 14(18), 8562. https://doi.org/10.3390/app14188562

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop