Mechanistic Investigation of Typical Loess Landslide Disasters in Ili Basin, Xinjiang, China

sustainability

Article Mechanistic Investigation of Typical Loess Landslide Disasters in Ili Basin, Xinjiang, China

Maoguo Zhuang 1,2,3,4, Wenwei Gao 5,* , Tianjie Zhao 6, Ruilin Hu 1,2,3, Yunjie Wei 4, Hai Shao 4 and Sainan Zhu 4

1 Key Laboratory of Shale Gas and Geoengineering, Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing 100029, China; [email protected] (M.Z.); [email protected] (R.H.) 2 Institutions of Earth Science, Chinese Academy of Sciences, Beijing 100029, China 3 College of Earth and Planetary Sciences, University of Chinese Academy of Sciences, Beijing 100049, China 4 China Geological Environment Monitoring Institute, Beijing 100081, China; [email protected] (Y.W.); [email protected] (H.S.); [email protected] (S.Z.) 5 Institute of Architectural Engineering, Yan’an University, Yan’an 716000, China 6 Aerospace Information Research Institute, Chinese Academy of Sciences, Beijing 100101, China; [email protected] * Correspondence: [email protected]

Abstract: In the period from 2010 to 2018, a total of 302 geological disasters occurred in the Xinjiang Autonomous Region, China, of which 136 occurred in the Ili Basin. Compared with those in other regions, the loess landslides in the Ili Basin are strongly influenced by the seasonal freeze–thaw effect. Taking the No. 2 Piliqinghe landslide as an example and based on the field geological investigation, it was found in the present study that the main triggering factors of this landslide were the snowmelt erosion of the slope toe and meltwater infiltration into the trailing edge of the slope. The mechanism of loess landslide instability was studied using numerical simulation. The results showed that (1) the Piliqinghe landslide disaster was formed through a process composed of the local sliding of the leading edge → the creep sliding and tension cracking of the slope surface → the overall sliding stage; (2) the infiltration of snowmelt was the direct cause of the landslide formation; (3) the fluvial Citation: Zhuang, M.; Gao, W.; Zhao, erosion and softening caused the soil of the slope toe to slide. The results can be used as a reference T.; Hu, R.; Wei, Y.; Shao, H.; Zhu, S. for the analysis of the disaster mechanism and movement characteristics of similar loess landslides. Mechanistic Investigation of Typical Loess Landslide Disasters in Ili Basin, Keywords: loess landslide; freeze–thaw effect; disaster mechanism; numerical simulation Xinjiang, China. Sustainability 2021, 13, 635. https://doi.org/10.3390/su 13020635

Received: 17 December 2020 1. Introduction Accepted: 8 January 2021 Loess is a special Quaternary loose sediment deposited and consolidated in a special Published: 11 January 2021 natural environment [1]. Due to the special structural and physico-mechanical properties of loess, loess landslides have occurred frequently in the world, such as the loess landslides in Publisher’s Note: MDPI stays neu- the Kyrgyz and TajikTien Shan (e.g., the Kainama earthflow, the Shrara flowslide, the Firma tral with regard to jurisdictional clai- flowslide, the 1 May flowslide, etc.) [2,3]. In China, loess is distributed over an area of ms in published maps and institutio- 64 × 104 km2, accounting for 6.3% of the total area of the country [4]. Statistically, excluding nal affiliations. earthquake-induced landslides, approximately one-third of all landslides in China occurred in loess areas [5,6]. In recent years, the intensification of human economic and engineering activities have been accompanied by the increasing occurrence of loess landslide disasters,

Copyright: © 2021 by the authors. Li- posing a serious threat to people’s lives and property and even causing mass injuries censee MDPI, Basel, Switzerland. and casualties. Research on the triggering factors and formation mechanisms of loess This article is an open access article landslides is the basis for effective mitigation of the risk of these landslides [7–9]. Since the distributed under the terms and con- 1980s, Chinese researchers have been investigating loess landslides and proposed a number ditions of the Creative Commons At- of mechanisms, including those for the movement and liquefaction of saturated sand– tribution (CC BY) license (https:// gravel layers of the sliding bed in high-speed long-runout landslides, earthquake-induced creativecommons.org/licenses/by/ liquefaction and pulverization effects, and static liquefaction and ﬂow liquefaction [10–12]. 4.0/).

Sustainability 2021, 13, 635. https://doi.org/10.3390/su13020635 https://www.mdpi.com/journal/sustainability Sustainability 2021, 13, 635 2 of 14

Rainfall, irrigation, earthquakes, and human activities are considered the main factors triggering loess landslides [13–18]. For example, in August 1955, a loess landslide with a volume of 3 × 107 m3 occurred in Wolongsi of Baoji, Shaanxi Province, destroy- ing a railroad, and a landslide occurred again at this location in 1971, killing 28 people [19]. On 7 March 1983, a landslide with a volume of approximately 4.1 × 107 m3 occurred in Dongxiang, Gansu Province, causing the collapse of more than 500 houses and the death of 237 people [20]. In 1991, a massive landslide occurred in Tianshui, Gansu Province, killing 200 people [21]. On 15 March 2019, a loess landslide occurred in Zaoling, Shanxi Province, causing the collapse of multiple houses and killing 20 people [4]. These catastrophic events have greatly raised the public awareness of loess geological disasters and promoted special research on loess landslide-triggering factors. The role of seasonal freeze–thaw cycles is often overlooked among the many landslide- triggering factors such as rainfall, climate, earthquakes and human activities [22]. However, in the arid and semiarid regions of Northwest China, freeze–thaw-induced landslides occur frequently, causing significant human injuries and property damage [23]. For example, on 10 March 2010, the Zizhou landslide in Yulin, Shaanxi Province, killed 27 people; and on 7 February 2012, the Jiaojia landslide in Yongjing, Gansu Province, pushed two vehicles into the Yellow River, killing one and leaving three people missing [24]. In recent years, the continuous occurrence of geological disasters during the freeze–thaw period has attracted increasing attention from the public and researchers. Studies on the mechanisms of freeze–thaw-induced landslides can be divided into two categories: one focuses on the destruction of the soil structure caused by freeze–thaw cycles, leading to a decrease in the physico-mechanical parameters of the soil and thereby triggering landslides, and the other pays attention to the water table rise due to freezing, which results in an increase in the pore water pressure and thus triggers landslides [25,26]. Although the Xinjiang region of China is not a high-risk area for geological disasters, a relatively large number of loess landslides occur in the Ili region of Xinjiang. A study of the vertical zone spectrum of natural disasters in the Tianshan Mountains region revealed that snowmelt and rainfall in spring are usually the main factors triggering landslides [27]. The frequent occurrence of landslides in the Ili region during the winter–spring transition period makes the investigation of the seasonal freeze–thaw effect the first choice. To further improve the accuracy of disaster assessment for this region, researchers investigated the formation mechanism and dynamics of the 31 July 2012, catastrophic landslide in Ili and found that the loess landslides in this region are characterized by structural complexity and instability due to unsaturated rainfall seepage. In particular, the freeze–thaw cycle leads to the enrichment and dissipation of groundwater in the slope. Enrichment results in the softening of the soil in the loess slope, while dissipation causes abrupt changes in the hydrostatic and hydrodynamic pressures, which simultaneously changes the pore characteristics and permeability, hence affecting the stability of the loess slope [28,29]. Sud- den heavy rainfall is an important factor triggering loess landslides in the Ili region [30,31]; therefore, the study of loess landslide in the Ili region for the seasonal freeze–thaw effect loess landslide is typical and representative. In this paper, taking the No. 2 Piliqinghe landslide as an example and based on a field geological survey, the characteristics and disaster mechanism of this landslide are analyzed in depth, and the disaster mode of this type of landslide is summarized to provide scientific experience and a theoretical basis for the rational disaster prevention and mitigation of and emergency responses to landslides in the loess region of Xinjiang, China.

2. Geological Conditions of the Study Area and Research Methods 2.1. Geological Conditions The Ili Basin is located in the western part of the Tianshan Mountains in Xinjiang, with longitude in the range of 80◦100 to 84◦580 E and latitude in the range of 42◦200 to 44◦490 N. Geological disasters occur frequently in this region due to natural factors such as a complex geological structure, rock fragmentation, and unique hydrometeorology as Sustainability 2021, 13, 635 3 of 14

well as intense human activities. According to the survey data of the China Geological Environment Monitoring Institute, a total of 302 geological disasters occurred in Xinjiang

Sustainability 2021from, 13, x 2010FOR PEER to 2018,REVIEW of which 136 occurred in the Ili Basin (Figure1), accounting for 45.03%,4 of 16 with landslides as the main type of geological disasters but fewer debris ﬂows and collapse.

Figure 1. Distribution of landslide disasters in the Ili River Valley region of Xinjiang. Figure 1. Distribution of landslide disasters in the Ili River Valley region of Xinjiang.

The Ili Basin has a complex topography and geomorphology, presenting an overall pattern of “three mountains sandwiching two valleys and one basin”. The geomorphological types are divided into three primary geomorphological units, namely, fault-block mountains with eroded folds, block uplift mountains by denudation and accumulation, and accumulation plains. The geological disasters in the region mostly develop in piedmont loess hills, which are distributed in the piedmont area and belong to block uplift mountains by erosion accumulation, with elevations of 900–1600 m and relative heights of 50–100 m. The hills are banded and were formed by uplift due to neotectonic movement. The Meso-Cenozoic strata are often covered with gravel and loess as well as lush vegetation, and the local lowlands are prone to landslides due to spring outcrops. The loess strata are mostly distributed in the foothills on two sides of the valleys or cover the slopes of the bedrock in the mountainous area, with the upper boundary reaching the forest belt and the lower boundary meeting the river valley plains. Their thickness is generally approximately 5–30 m and can reach a maximum of 80 m at some individual locations. The Ili BasinFigure is 2. located Monthly in temperature a high seismicity in Ili between zone 2010 withand 2018. intense neotectonic movements that mainly manifest as the inherited elevation and subsidence and the accompanying faults and folds. This basin is located in the North Tianshan seismic zone and immediately adjacent to the South Tianshan seismic zone in the south. Both seismic zones compose the main high seismicity regions in Xinjiang. Since 1812, there have been 14 strong earthquakes with magnitudes greater than six in the Ili Basin and adjacent regions. The Ili Basin is situated in the center of the Eurasian continent and in western Xinjiang, which is far from the ocean, thereby forming a typical continental climate with an average annual temperature of 7.8 ◦C and an average annual precipitation of 349.3 mm. In a year, the temperature rises rapidly in spring, the summer is mild and rainy, the temperature drops quickly in the autumn, and the winter is long and warm (Figure2). The supply of surface water is clearly controlled by seasonal precipitation and inﬂuenced by both alpine snowmelt and precipitation recharge (Figure3). Geological disasters in the region are often distributed along both sides of the valleys, and perennial ﬂowing water has a direct impact on geological disasters. Sustainability 2021, 13, x FOR PEER REVIEW 4 of 16

Sustainability 2021, 13, 635 4 of 14 Figure 1. Distribution of landslide disasters in the Ili River Valley region of Xinjiang.

Sustainability 2021, 13, x FOR PEER REVIEW 5 of 16

FigureFigure 2. 2. MonthlyMonthly temperature temperature in in Ili Ili between between 2010 and 2018.

FigureFigure 3. 3.MonthlyMonthly precipitation precipitation in inIli Ili between between 2010 2010 and and 2018. 2018.

2.2. MethodsThe mountainous area in the Ili Basin accounts for more than 70% of the total area of the region, with a mild climate and abundant precipitation. Human engineering and economic Mechanistic analysis of landslide instability is the core issue in landslide disaster re- activities are mainly concentrated in the valley plain area and in the ﬁelds of, for example, search. Due to the complexity of geological conditions and the limited understanding of agricultural and hydraulic engineering, animal husbandry, and the tourism industry. special landslides, the field investigation and analysis of engineering geological condi- tions2.2. are Methods of great importance in landslide research. The qualitative analysis of engineering geology is especially valuable for research on landslide disasters with complex geological Mechanistic analysis of landslide instability is the core issue in landslide disaster conditions and diverse triggering factors. research. Due to the complexity of geological conditions and the limited understanding of Stability under special conditions is another core issue in landslide disaster research. special landslides, the ﬁeld investigation and analysis of engineering geological conditions Atare present, of great the importance numerical si inmulation landslide method research. is commonly The qualitative employed. analysis When of heterogene- engineering ousgeology and nonlinear is especially slopes valuable with complex for research boundaries on landslide are disastersstudied with with this complex method, geological computer-basedconditions andnumerical diverse software triggering is used factors. to obtain the stress–strain relations of rock and soil andStability to quantitatively under special analyze conditions the slop ise another deformation/damage core issue in landslide and groundwater disaster research. seep- ageAt to present, investigate the numerical the interaction simulation between method the is rock commonly and soil. employed. The complex When heterogeneousstratigraphic conditionsand nonlinear of the slopes slope can with be complex simulated boundaries in a cost-effective are studied and with efficient this method,manner to computer- obtain thebased stability numerical state of software a landslide is used under to obtaindifferent the working stress–strain conditions. relations of rock and soil and to quantitativelyThe integration analyze of field theinvestigation, slope deformation/damage qualitative engineering and groundwatergeological analysis, seepage and to quantitativeinvestigate thenumerical interaction simulation between is the th rocke most and common soil. The complexand most stratigraphic comprehensive conditions approach in the study of landslide mechanisms. The integrated approach makes it possible to reveal the fundamental causes and formation modes of landslides and provide theoretical support for disaster prevention and mitigation. A schematic diagram of the landslides research methods of this study is detailed in Figure 4. First, we research the landslides in the slope region of the Ili Basin in Xinjiang using remote sensing image analysis and geological investigation. Then, we build a geological structural model of the landslide and simulate its deformation and failure process using the numerical software FLAC 3D 5.00. Finally, we discuss the loess landslide’s stability, deformation and failure mechanism. Sustainability 2021, 13, 635 5 of 14

of the slope can be simulated in a cost-effective and efﬁcient manner to obtain the stability state of a landslide under different working conditions. The integration of ﬁeld investigation, qualitative engineering geological analysis, and quantitative numerical simulation is the most common and most comprehensive approach in the study of landslide mechanisms. The integrated approach makes it possible to reveal the fundamental causes and formation modes of landslides and provide theoretical support for disaster prevention and mitigation. A schematic diagram of the landslides research methods of this study is detailed in Figure4. First, we research the landslides in the slope region of the Ili Basin in Xinjiang using remote sensing image analysis and geological investigation. Then, we build a geological structural model of the landslide and simulate its deformation and failure 3D Sustainability 2021, 13, x FOR PEER REVIEWprocess using the numerical software FLAC 5.00. Finally, we discuss6 of 16 the loess landslide’s stability, deformation and failure mechanism.

Figure 4. Schematic 4. Schematic diagram diagram of the landslides of the research landslides methods. research methods. 3. Case Study of Typical Geological Disasters in the Study Area and their Disaster Modes3. Case Study of Typical Geological Disasters in the Study Area and their Disaster Modes Geological disasters in Ili between 2010 and 2018 are shown in Table 1. Geological disastersGeological occurred mainly disasters in spring in Iliand between summer in 2010 the andIli Basin, 2018 with are 92 shown (67.65%) in in Table 1. Geological dis- spring,asters 40 occurred (29.41%) in mainly summer, in 4 (2.94%) spring in and autumn, summer and none in in the winter. Ili Basin, The Ili withValley 92is (67.65%) in spring, 40 a region with extensive development of freeze–thaw landslides. Past records show that 152(29.41%) landslides in summer,occurred during 4 (2.94%) the freeze–thaw in autumn, periods, and accounting none in for winter. 40% of Thethe 380 Ili Valley is a region with landslidesextensive recorded development in the Ili Valley of freeze–thaw region [31,32]. According landslides. to the Past spatiotemporal records dis- show that 152 landslides tributionoccurred characteristics during the of freeze–thawloess landslides in periods, the Ili Basin, accounting the main triggering for 40% factors of the for 380 landslides recorded loess landslides in this region are precipitation, fluvial erosion, and freeze–thaw cycles [33,34].in the Ili Valley region [31,32]. According to the spatiotemporal distribution characteristics of loess landslides in the Ili Basin, the main triggering factors for loess landslides in this regionTable 1. are Geological precipitation, disasters in Ili ﬂuvial between erosion,2010 and 2018. and freeze–thaw cycles [33,34]. Total Num- Ili Year ber in Xin- Total SpringTable 1. GeologicalSummer disasters inAutumn Ili between 2010 and 2018.Winter jiang Number (March to May) (June to August) (September to November) (December to next February) 2010 68 31 31 Ili 2011 11 Total 6Number 3 1 2 Year Summer Autumn Winter 2012 40 in Xinjiang 6 4 1 Spring 1 Total Number (June to (September to (December to 2013 29 6 1 5 (March to May) 2014 15 2 2 August) November) next February) 20102015 13 4 68 2 31 2 31 20112016 65 35 11 17 6 18 3 1 2 2017 61 40 29 11 2012 40 6 4 1 1 2018 28 6 3 2 1 2013 29 6 1 5 Sum 330 136 92 40 4 2014 15 2 2 2015 133.1. Piliqinghe Landslide 4 2 2 2016 65Located in 35the west of the Ili Valley, 17 the Piliqinghe River 18 is a typical area in the Ili 2017 61region that experiences 40 loess landslide 29 disasters. As shown in 11 the remote sensing image 2018 28of the Kezilesai Gully 6 (Figure 5), the survey 3 found that there 2were 14 loess landslides 1 in Sum 330the Piliqinghe River 136 Basin, all of which 92 developed in the three 40 first-level tributaries in 4 the upper reaches of the Piliqinghe River. Among them, the third branch gully (the Kezilesai Sustainability 2021, 13, 635 6 of 14

3.1. Piliqinghe Landslide SustainabilitySustainability 2021 2021, ,13 13, ,x x FOR FOR PEER PEER REVIEW REVIEW 77 of of 16 16

Located in the west of the Ili Valley, the Piliqinghe River is a typical area in the Ili region that experiences loess landslide disasters. As shown in the remote sensing image GullyGullyof thein in Figure Figure Kezilesai 6, 6, the the remote remote Gully sensing sensing (Figure image image5 was), was the taken taken survey by by UAV(Unmanned UAV(Unmanned found that Aerial Aerial there Vehicle) Vehicle) were 14 loess landslides in withwith a a resolution resolution of of 50 50 cm), cm), which which is is relatively relatively small, small, developed developed nine nine landslides landslides (No. (No. 1 1 to to 9),9),the with with Piliqinghe the the leading leading edges Riveredges of of Basin,two two landslides landslides all of located whichlocated on on developed floodplains; floodplains; the inthe thebranch branch three gully gully first-level on on tributaries in the thetheupper right right side side reaches of of the the two two of main main the branches Piliqinghebranches is is th thee River.Piliqinghe Piliqinghe Among River, River, whic whic them,hh developed developed the third five five land- land- branch gully (the Kezilesai slidesslidesGully (No. (No. in 10 10 Figureto to 14), 14), with with6, the the leading leading remote edges edges sensing of of four four landslides imagelandslides was located located taken on on first-level first-level by UAV(Unmanned terraces terraces Aerial Vehicle) oror floodplains; floodplains; and and the the branch branch gully gully on on the the left left side side is is the the Axi Axi Gully, Gully, which which developed developed fourfourwith landslides landslides a resolution (nos. (nos. 15 15 to to of 18), 18), 50 with with cm), the the which le leadingading isedges edges relatively of of two two landslides landslides small, developedlocated located on on first- first- nine landslides (No. 1 to orderorder9), withterraces terraces the or or river leadingriver floodplains. floodplains. edges of two landslides located on floodplains; the branch gully on the rightTheThe sideKezilesai Kezilesai of theGully Gully two landslides landslides main in branchesin the the Piliqinghe Piliqinghe is the Basin Basin Piliqinghe are are shown shown in River,in Figure Figure which7; 7; the the Pili- Pili- developed five landslides qingheqinghe landslide landslide is is a a homogeneous homogeneous loess loess landslide landslide of of a a single single lithology. lithology. The The Ili Ili loess loess is is QuaternaryQuaternary(No. 10 aeolian toaeolian 14), loess, loess, with which which the has has leading a a highly highly edgesuniform uniform of grain grain four size size landslides composition composition and locatedand mainly mainly on first-level terraces or consistsconsistsfloodplains; of of silt. silt. Comparison Comparison and the of of branch the the Ili Ili loe loe gullyssss with with on the the the loess loess left from from side the the isLoess Loess the Plateau Plateau Axi Gully, found found which developed four thatthatlandslides the the content content of (nos.of the the grain grain 15 tosize size 18), of of 0.05–0.005 0.05–0.005 with the mm mm leading in in the the Ili Ili loess edges loess is is markedly markedly of two higher landslides higher than than located on first-order those in the Malan loess from eastern China and the loess from the northern piedmont of thoseterraces in the Malan or river loess floodplains. from eastern China and the loess from the northern piedmont of thethe Tianshan Tianshan Mountains. Mountains.

Figure 5. Distribution of the Piliqinghe landslides. FigureFigure 5. Distribution 5. Distribution of the Piliqinghe of the landslides. Piliqinghe landslides.

Figure 6. Remote sensing image of the Kezilesai Gully landslides in the Piliqinghe basin. FigureFigure 6. Remote 6. Remote sensing image sensing of the image Kezilesai of Gully the Kezilesailandslides in Gully the Piliqinghe landslides basin. in the Piliqinghe basin.

The Kezilesai Gully landslides in the Piliqinghe Basin are shown in Figure7; the Pili qinghe landslide is a homogeneous loess landslide of a single lithology. The Ili loess is Quaternary aeolian loess, which has a highly uniform grain size composition and mainly consists of silt. Comparison of the Ili loess with the loess from the Loess Plateau found that the content of the grain size of 0.05–0.005 mm in the Ili loess is markedly higher than those in the Malan loess from eastern China and the loess from the northern piedmont of the Tianshan Mountains. Sustainability 2021, 13, 635 7 of 14 Sustainability 2021, 13, x FOR PEER REVIEW 8 of 16

Figure 7. The Kezilesai Gully landslide in the Piliqinghe basin. Figure 7. The Kezilesai Gully landslide in the Piliqinghe basin. The landslide that occurred on 24 March 2017 in the Kalayagaqi Village of Kalayagaqi The landslideTownship was that typical occurred of thoseon developed 24 March in the 2017Piliqinghe in thebasin. Kalayagaqi This landslide, Villagereferred of Kalayagaqi to as the No. 2 Piliqinghe landslide, blocked the Piliqinghe River, forming a barrier lake Township was(Figure typical 8). This oflandslide those buried developed a farmer’s infishpond the Piliqinghe (with an area basin.of 2000 m This2), a farm- landslide, referred to as the No.house, 2 Piliqinghe a small tractor, landslide, an excavator, blocked and dozens the of fruit Piliqinghe trees on the River, north bank forming of the a barrier lake Piliqinghe River, with a direct economic loss of more than 200,000 yuan. The Piliqinghe 2 (Figure8). Thislandslide landslide had a clear buried perimeter a farmer’s and a tongue-shaped ﬁshpond pattern (with in plane, an area with ofa gradient 2000 m of ), a farmhouse, a small tractor,37° and an a main excavator, sliding direction and dozensof 32° (Figures of fruit 9 and trees10). The on landslide the north had a width bank of 96 of the Piliqinghe River, withm, a directa smallest economic width of 60 m loss at the of leading more edge, than a maximum 200,000 longitudinal yuan. The length Piliqinghe of 175 m, landslide had a thickness of 3–6 m, and a total volume of approximately 6.08 × 104 m3. The landslide was ◦ a clear perimetercovered with and snow a tongue-shaped approximately 20 cm pattern thick (Figure inplane, 11). with a gradient of 37 and a main sliding direction of 32◦ (Figures9 and 10). The landslide had a width of 96 m, a smallest width of 60 m at the leading edge, a maximum longitudinal length of 175 m, a thickness of 3–6 m, and a total volume of approximately 6.08 × 104 m3. The landslide was covered Sustainability 2021, 13, x FOR PEER REVIEW 9 of 16 with snow approximately 20 cm thick (Figure 11).

Figure 8. Remote sensing image of the No. 2 Piliqinghe landslide. Figure 8. Remote sensing image of the No. 2 Piliqinghe landslide.

Figure 9. Panoramic view of the No. 2 Piliqinghe landslide. Sustainability 2021, 13, x FOR PEER REVIEW 9 of 16

Sustainability 2021, 13, 635 8 of 14

Figure 8. Remote sensing image of the No. 2 Piliqinghe landslide.

Sustainability 2021, 13, x FOR PEER REVIEW 10 of 16 Sustainability 2021, 13, x FOR PEER REVIEWFigure 9. Panoramic view of the No. 2 Piliqinghe landslide. 10 of 16 Figure 9. Panoramic view of the No. 2 Piliqinghe landslide.

Figure 10. Scheme of the engineering geology of the No. 2 Piliqinghe landslide. Figure 10. Scheme of the engineering geology of the No. 2 Piliqinghe landslide.

Figure 11. Longitudinal profile of the No. 2 Piliqinghe landslide. FigureFigure 11. 11.Longitudinal Longitudinal proﬁle profile of of the the No. No. 2 2 Piliqinghe Piliqinghe landslide. landslide. As shown in Figure 12, according to the data of the Yining County Meteorological As shown in Figure 12, according to the data of the Yining County Meteorological Station, from 1 March to 22 March 2017, the daily minimum and maximum temperatures Station, from 1 March to 22 March 2017, the daily minimum and maximum temperatures in the Piliqinghe Basin were below and above 0 °C, respectively, and there was an intense in the Piliqinghe Basin were below and above 0 °C, respectively, and there was an intense freeze–thaw cycle before the disaster. The rainfall data showed that the total rainfall in the freeze–thaw cycle before the disaster. The rainfall data showed that the total rainfall in the region before the occurrence of the No. 2 Piliqinghe landslide from 10 to 24 March was region before the occurrence of the No. 2 Piliqinghe landslide from 10 to 24 March was approximately 20 mm, and there was no rainfall from 19 to 24 March; hence, it was un- approximately 20 mm, and there was no rainfall from 19 to 24 March; hence, it was unlikely that rainfall induced the landslide, which occurred on 24 March 2017. likely that rainfall induced the landslide, which occurred on 24 March 2017. Sustainability 2021, 13, 635 9 of 14

As shown in Figure 12, according to the data of the Yining County Meteorological Station, from 1 March to 22 March 2017, the daily minimum and maximum temperatures in the Piliqinghe Basin were below and above 0 ◦C, respectively, and there was an intense freeze–thaw cycle before the disaster. The rainfall data showed that the total rainfall in the region before the occurrence of the No. 2 Piliqinghe landslide from 10 to 24 March Sustainability 2021, 13, x FOR PEER REVIEWwas approximately 20 mm, and there was no rainfall from 19 to 24 March; hence,11 it wasof 16

unlikely that rainfall induced the landslide, which occurred on 24 March 2017.

Figure 12. Daily minimum and maximum temperatures and rainfall beforebefore and after the occurrence ofof the No. 2 Piliqinghe landslide on 24 March 2017.2017.

Combined with the field field investigation result resultss and Zhu Sainan’s investigation in this area [32], [32], the the results results showed showed that that since since 1 Marc 1 Marchh 2017, 2017, the temperature the temperature in this in region this region grad- uallygradually increased, increased, leading leading to an to increase an increase in snowmelt. in snowmelt. As a Asresult, a result, the volume the volume of the of Pili- the qinghePiliqinghe River River increased increased dramatically, dramatically, the the water water level level rose, rose, and and the the flow flow velocity velocity increased. increased. Consequently, the lateral erosion of the leading edge of the landslide on the concave bank was intensified, intensified, the the strength strength of of the the soil soil at atthe the leading leading edge edge decreased, decreased, and and collapsed collapsed occurred,occurred, and and there there was was serious serious scouring scouring of the of therock rock mass mass at the at theslope slope toe, toe,which which was was the mainthe main factor factor causing causing the thelandslide. landslide. In additi In addition,on, as the as theweather weather warmed, warmed, the thesnowmelt snowmelt in- filtratedinfiltrated along along the the cracks cracks in inthe the soil, soil, increasing increasing the thesliding sliding force force and and deadweight deadweight of the of slidingthe sliding mass, mass, which which was another was another important important factor factorin the inlandslide. the landslide. Therefore, Therefore, the mecha- the nismmechanism of the No. of the 2 Piliqinghe No. 2 Piliqinghe landslide landslide can be cansummarized be summarized by the following by the following three stages: three localstages: collapse local collapseat the leading at the edge leading → creep edge sliding→ creep and sliding tension and cracking tension of cracking the surface of the → overallsurface sliding→ overall (Figure sliding 13). (Figure 13). (1) ErosionErosion and and collapse collapse of of the the leading edge. In In the spring, the snowmelt increased the flowflow rate rate of of the the Piliqinghe Piliqinghe River. River. As As a a result, result, the lateral erosion was intensified,intensified, and thethe bank bank slope slope was was undercut, undercut, forming forming a a free free face. Therefore, Therefore, the leading edge of the landslidelandslide was was partially partially unbalanced, unbalanced, and and multiple multiple sliding sliding collapses collapses occurred, as can bebe observed observed in in the the multi-period multi-period remote remote sensing sensing images. images. (2) CreepCreep sliding and tensiontension crackingcracking of of the the surface. surface. As As the the snowmelt snowmelt infiltrated infiltrated into into the ◦ theslope, slope, the the variation variation in the in diurnalthe diurnal temperature temperature around around 0 C 0 led °C to led intense to intense freeze–thaw freeze– thawcycles, cycles, which which reduced reduced the mechanical the mechanical strength strength of the soil, of the resulting soil, resulting in the formation in the for- of mationcracks onof cracks the landslide on the surface.landslide Meanwhile, surface. Meanwhile, as the water as contentthe water of content the soil of increased, the soil increased,shear creep shear occurred creep occurred on the slope on the surface, slope surface, causing causing creep tension creep tension cracking cracking under undergravity. gravity. Hence, Hence, the cracks the cracks in the slopein the wereslope widened, were widened, and continuous and continuous tensile tensile cracks cracksoccurred occurred at the trailingat the trailing edge. edge. (3) Overall deformation and sliding. As the freeze–thaw cycles continued, the frost heave effect further increased the cracks at the trailing edge of the landslide. With the infiltration of a large amount of snowmelt along the cracks, the water content rose sharply, and the soil became nearly saturated. As a result, the friction decreased dramatically, disrupting the original force balance of the loess structure. Consequently, the landslide slid down at high speed, rushing into and burying the fishpond and blocking the Piliqinghe River to form a barrier lake. Sustainability 2021, 13, 635 10 of 14

(3) Overall deformation and sliding. As the freeze–thaw cycles continued, the frost heave effect further increased the cracks at the trailing edge of the landslide. With the inﬁltration of a large amount of snowmelt along the cracks, the water content rose sharply, and the soil became nearly saturated. As a result, the friction decreased dramatically, disrupting the original force balance of the loess structure. Consequently, the landslide slid down at high speed, rushing into and burying the ﬁshpond and Sustainability 2021, 13, x FOR PEER REVIEW 12 of 16 blocking the Piliqinghe River to form a barrier lake.

Figure 13. Sketch map showing the gestation model of the No. 2 Piliqinghe landslide. Figure 13. Sketch map showing the gestation model of the No. 2 Piliqinghe landslide. 3.2. Numerical Simulation of the Landslide 3.2. NumericalLandslide Simulation deformation of the and Landslide failure is a typical large deformation problem, for which FastLandslide Lagrangian deformation Analysis of andContinua failure in 3D is a(FLAC typical 3D) largesoftware deformation has been widely problem, used as for which Fast Lagrangiana numerical simulation Analysis tool. of Continua This software in 3D is based (FLAC on 3Dthe) fi softwarenite difference has beenmethod widely with a used as a fast Lagrangian analysis algorithm for a continuum and uses an explicit time approxima- numerical simulation tool. This software is based on the ﬁnite difference method with a fast tion to obtain the full time-step solution of all equations of motion, thereby allowing track- Lagrangianing of the analysis progressive algorithm failure of for the a continuumslope. For this and reason, uses FLAC an explicit 3D 5.00was time used approximation in the to obtainpresent the full study time-step to simulate solution the No. 2 of Piliqinghe all equations landslide. of motion, thereby allowing tracking of the progressiveDuring the failure computing of the procedure, slope. For the this stabil reason,ity factor FLAC of slope3D 5.00waswas computed used by in the present studyshear to simulate strength reduction the No. 2method. Piliqinghe This landslide.method has been widely used in slope stability analysis.During Using the computing a series of trial procedure, factors of thesafety stability Fs, the soil factor strength of slope parameters was computed c and φ by the shearare strength reduced as reduction follows: method. This method has been widely used in slope stability 𝑐 =𝑐/𝐹 (1) Sustainability 2021, 13, 635 11 of 14

Sustainability 2021, 13, x FOR PEER REVIEW 13 of 16 analysis. Using a series of trial factors of safety Fs, the soil strength parameters c and ϕ are reduced as follows: 0 c = c/Fs (1) tan 0 𝜑 =arctan( ϕ ). ϕ = arctan . (2)(2) Fs DuringDuring the the numerical numerical computing computing process, process,c 0 𝑐and andϕ0 are𝜑 areused used as the as inputthe input parameters. parameters. The trial factor of safety 𝐹 incrementally increases until the convergent criterion is The trial factor of safety Fs incrementally increases until the convergent criterion is not not satisfied. In this case, the slope is in the limit equilibrium state, and the corresponding satisﬁed. In this case, the slope is in the limit equilibrium state, and the corresponding Fs is considered𝐹 is considered as the as real the factor real factor of safety of safetyF. F. AA numerical numerical model model (Figure (Figure 14 14)) was was established established based based on on the the engineering engineering geological geological modelmodel (Figure (Figure 10 10).). TheThe whole whole model model was was meshed meshed using using 5078 5078 quadrilateral quadrilateral elements elements and and 68,44868,448 nodes. nodes. The The default default static static mode mode of FLAC of FLAC was adoptedwas adopted in the simulation.in the simulation. In the model, In the themodel, displacements the displacements in the x-direction in the x-direction on the two on sides, the two all displacementssides, all displacements in the y-direction, in the y- anddirection, the displacement and the displacement in the z-direction in the at z-di therection bottom wereat the constrained. bottom were The constrained. surfaces of theThe landslidesurfaces of model the landslide were set asmodel free were boundaries. set as free The boundaries. ideal elastic–plastic The ideal constitutive elastic–plastic model con- andstitutive the Mohr–Coulomb model and the strength Mohr–Coulomb criterion werestrength adopted, criterion and were the slope adopted, stability and factor the slope was calculatedstability factor using was the calculated strength reduction using the method. strength reduction method.

FigureFigure 14. 14.A A numerical numerical simulation simulation model model of of the the landslide landslide proﬁle. profile.

AccordingAccording to to the the ﬁeld field investigation investigation data data and and the the model model generalization, generalization, three three different differ- typesent types of rock–soil of rock–soil material material were were considered considered in the in calculation, the calculation, namely, namely, loess, loess, alluvium alluvium and diluvium,and diluvium, and bedrock. and bedrock. The physico-mechanical The physico-mechan parametersical parameters of these of types these of types rock–soil of rock– and watersoil and table water level table refer level to the refer results to the of results Zhu et of al. Zhu [32], et with al. [32], the speciﬁcwith the parameters specific parameters listed in Tablelisted2 .in Table 2.

3D Table 2. Physical andTable mechanical 2. Physical parameters and mechanical of the paramete materialsrs for of FLAC the materials. for FLAC 3D. Elastic Tensile Internal Material Name of Density Poisson’s Tensile Cohesion Internal Fric- Name of Ma- Density 𝜸/𝒈∗ElasticModulus Modu- Poisson’s Ra- Strength Cohesion Friction MaterialType type Material γ/g∗m−3 Ratio v Strength c/KPa tion Angle terial 𝒎 𝟑 lusE/GPa E/GPa tio 𝝂 t/KPa c/KPa Angle ϕ/◦ t/KPa 𝝋/° A Loess 1670 (1930) 0.043 (0.021) 0.30 (0.32) 15.72 (13.44) 17.78 (14.32) 23.30 (20.08) A Loess 1670 (1930) 0.043 (0.021) 0.30 (0.32) 15.72 (13.44) 17.78 (14.32) 23.30 (20.08) Alluvium & B Alluvium & 2100 0.042 0.32 13.00 13.38 21.47 B diluvium 2100 0.042 0.32 13.00 13.38 21.47 C Clasticdiluvium rock 2542 30 0.28 113.65 120 45 C Clastic rock 2542 30 0.28 113.65 120 45 AlthoughAlthough the the dynamic dynamic processes processes of of the the slope slope toe toe erosion erosion and and rainfall rainfall inﬁltration infiltration were were notnot simulatedsimulated inin thethe numericalnumerical analysis,analysis, thethe effecteffect ofof snowmeltsnowmelt onon thethe softening softening ofof the the slopeslope toe toe and and the the inﬂuence influence of of rainfall rainfall on on thethe soilsoilproperties propertieswere were considered. considered. ToTo this this end, end, thethe groundwatergroundwater levellevel was determined determined according according to to the the results results of of Zhu Zhu et etal. al.[32]; [32 then,]; then, the thesaturated saturated parameters parameters were were used used for forthe therock–s rock–soiloil of the of slope the slope below below the groundwater the groundwater level levelin the in rainy the rainy season, season, and the and saturated the saturated parameters parameters were were used used for all for the all soil the below soil below the river the riverlevel. level. The numerical simulation results for dry and wet season are shown in Figure 15. In the dry season, the plastic shear zone of the slope was distributed at the front of the slope (Figure 15a), and the overall slope stability factor was calculated to be 1.243, indicating Sustainability 2021, 13, x FOR PEER REVIEW 14 of 16

that the slope was stable. The shear strain was large at the front of the slope, with a maximum value of 9.61 × 10−4, which could be a potential failure surface of the slope toe. Figure 15c shows the contours of the displacement in the landslide, with a maximum value of 0.544 m in the dry season. In the rainy season, the slope was in the limit equilibrium state, with a calculated stability factor of 0.981. The plastic shear strain, with a maximum value of 2.56 × 10−3, was distributed along the sliding zone (Figure 15b), and the trailing edge of the landslide was located at the crack position in the upper part of the slope. In addition, the displacement contours indicate that the maximum displacement in the slope was 0.864 m (Figure 15d). In fact, the cracks at the trailing edge of the Piliqinghe landslide formed a channel for the infiltration of snowmelt and rainfall, resulting in the softening of the soil in the trailing Sustainability 2021, 13, 635 edge of the slope and thus extending the sliding from the back to the middle of the slope.12 of 14 The fluvial erosion and softening of the slope toe also caused the soil of the slope toe to slide. The two factors had an important influence on the progressive failure of the Pili- qinghe landslide. TheAt numerical present, the simulation numerical results simulation for dry soft andware wet cannot season simulate are shown the in influence Figure 15 of. In the dryfreeze–thaw season, the cycle plastic on landslide shear zone stability. of the How slope to realize was distributedthe simulation at of the the front freeze–thaw of the slope (Figureprocess 15 a),can and be studied the overall in the slope future. stability In this paper, factor through was calculated the adjustment to be 1.243, of geotechnical indicating that theparameters, slope was stable.the softening The shear of slope strain toe wascaused large by snowmelt at the front infiltration of the slope, and the with influence a maximum of rainfall on soil properties are considered, which is a good method for the simulation value of 9.61 × 10−4, which could be a potential failure surface of the slope toe. Figure 15c study of loess landslide under the influence of snowmelt infiltration. For the loess land- shows the contours of the displacement in the landslide, with a maximum value of 0.544 m slide, water is the most important factor. Therefore, under the influence of snowmelt inin thefiltration, dry season. the occurrence process of landslide disaster is very complex.

Figure 15. Modeling results for dry and wet season: (a) contours of the shear strain for the dry season; (b) contours of the Figure 15. Modeling results for dry and wet season: (a) contours of the shear strain for the dry season; (b) contours of the shear strain for wet season; (c) contours of the displacement for the dry season; (d) contours of the displacement for the shear strainwet season. for wet season; (c) contours of the displacement for the dry season; (d) contours of the displacement for the wet season. 4. Conclusions InBased the rainy on geological season, theinvestigation slope was and in research the limit on equilibriumthe landslides state, in the withslope aregion calculated −3 stabilityof the Ili factor Basin of in 0.981. Xinjiang, The the plastic present shear stud strain,y examined with the a maximummode of geological value of disasters 2.56 × 10 , wasin distributed the region and along carried the slidingout the zoneengineering (Figure geological 15b), and zoning the trailing of the edgeNo. 2 ofPiliqinghe the landslide waslandslide. located atIn the addition, crack position a geological in the structural upper part model of the of slope.the landslide In addition, was built, the displacement and its contours indicate that the maximum displacement in the slope was 0.864 m (Figure 15d). In fact, the cracks at the trailing edge of the Piliqinghe landslide formed a channel for the infiltration of snowmelt and rainfall, resulting in the softening of the soil in the trailing edge of the slope and thus extending the sliding from the back to the middle of the slope. The fluvial erosion and softening of the slope toe also caused the soil of the slope toe to slide. The two factors had an important influence on the progressive failure of the Piliqinghe landslide. At present, the numerical simulation software cannot simulate the influence of freeze– thaw cycle on landslide stability. How to realize the simulation of the freeze–thaw process can be studied in the future. In this paper, through the adjustment of geotechnical parameters, the softening of slope toe caused by snowmelt infiltration and the influence of rainfall on soil properties are considered, which is a good method for the simulation study of loess landslide under the influence of snowmelt infiltration. For the loess landslide, water is the most important factor. Therefore, under the influence of snowmelt infiltration, the occurrence process of landslide disaster is very complex. Sustainability 2021, 13, 635 13 of 14

4. Conclusions Based on geological investigation and research on the landslides in the slope region of the Ili Basin in Xinjiang, the present study examined the mode of geological disasters in the region and carried out the engineering geological zoning of the No. 2 Piliqinghe landslide. In addition, a geological structural model of the landslide was built, and its deformation and failure process was simulated using the numerical software FLAC 3D. The following conclusions were drawn: (1) The disaster process of the No. 2 Piliqinghe landslide was as follows: fluvial scouring of the slope toe leading to the local sliding of the leading edge → snowmelt infiltration resulting in the creep sliding and tension cracking of the slope surface → overall sliding of the slope. (2) The infiltration of snowmelt due to temperature increase was the direct cause of landslide formation. The infiltrated snowmelt was transformed into groundwater, which saturated the loess and increased the pore water pressure, thereby decreasing the slope stability factor and leading to an overall sliding under gravity. (3) The stability and the underlying deformation and failure mechanism of the No. 2 Piliqinghe landslide were investigated using the finite difference strength reduction method. The results showed the following. In the dry season, the slope had a stability factor of 1.254, indicating that it was stable. The shear strain at the front of the slope was large, with a maximum value of 3.59 × 10−4, and the maximum displacement in the slope was 0.304 m. In the rainy season, the slope was in the limit equilibrium state, and the stability factor was calculated to be 0.985. The plastic shear strain, which had a maximum value of 9.29 × 10−4, was distributed along the sliding zone, and the trailing edge of the landslide was consistent with the crack position in the upper part of the slope. The maximum displacement in the slope was 0.636 m. Therefore, regarding the prevention of loess landslide disasters in the Ili region, the results suggest that cracks in the trailing edge of the slope have an extremely important influence on the landslide and that additionally, instability in the middle and back parts of the landslide has a strong impact on the stability of the whole slope and should be avoided. These findings can be used as a reference for the analysis of the disaster mechanism and movement characteristics of similar loess landslides and provide further theoretical support for disaster prevention and mitigation.

Author Contributions: Methodology, M.Z. and W.G.; Field Investigation, M.Z., H.S., S.Z. and Y.W.; Software, M.Z., T.Z. and W.G.; Data curation, M.Z.; Writing—Original Draft Preparation, M.Z. and W.G.; Writing—Review and Editing, M.Z., R.H. and T.Z. All authors have read and agreed to the published version of the manuscript. Funding: We appreciate financial support by the National Natural Science Foundation of China (NSFC) (No. 41671355); Geological Survey Project (No. DD20179609); The Chinese Academy of Sciences (CAS) “Light of West China” Program; Yan’an University Project (YDBK2019-37). Grateful appreciation is expressed for the support. Data Availability Statement: Data sharing not applicable. Acknowledgments: This authors are grateful to editors and reviewers for their enthusiastic help and valuable comments. Conflicts of Interest: The authors declare no conflict of interest.

References 1. Liu, D. Loess and Environment; Science Press: Beijing, China, 1985. 2. Ishihara, K.; Okusa, S.; Oyagi, N.; Ischuk, A. Liquefaction-induced ﬂow slide in the collapsible loess deposit in Soviet Tajik. Soils Found. 1990, 30, 73–89. [CrossRef] 3. Havenith, H.-B.; Torgoev, A.; Schlögel, R.; Braun, A.; Torgoev, I.; Ischuk, A. Tien Shan geohazards database: Landslide susceptibility analysis. Geomorphology 2015, 249, 32–43. [CrossRef] Sustainability 2021, 13, 635 14 of 14

4. Cui, Y.; Xu, C.; Xu, S.; Chai, S.; Fu, G.; Bao, P. Small-scale catastrophic landslides in loess areas of China: An example of the March 15, 2019, Zaoling landslide in Shanxi Province. Landslides 2020, 17, 669–676. [CrossRef] 5. Xu, C.; Xu, X.; Shyu, J.B.H.; Zheng, W.; Min, W. Landslides triggered by the 22 July 2013 Minxian–Zhangxian, China, Mw 5.9 earthquake: Inventory compiling and spatial distribution analysis. J. Asian Earth Sci. 2014, 92, 125–142. [CrossRef] 6. Zhou, J.-X.; Zhu, C.-Y.; Zheng, J.-M.; Wang, X.-H.; Liu, Z.-H. Landslide disaster in the loess area of China. J. For. Res. 2002, 13, 157–161. 7. Derbyshire, E.; Jingtai, W.; Zexian, J. In the Gansu Loess of China. Catena Suppl. 1991, 20, 119–145. 8. Tu, X.; Kwong, A.; Dai, F.; Tham, L.; Min, H. Field monitoring of rainfall infiltration in a loess slope and analysis of failure mechanism of rainfall-induced landslides. Eng. Geol. 2009, 105, 134–150. [CrossRef] 9. Peng, J.; Wang, G.; Wang, Q.; Zhang, F. Shear wave velocity imaging of landslide debris deposited on an erodible bed and possible movement mechanism for a loess landslide in Jingyang, Xi’an, China. Landslides 2017, 14, 1503–1512. [CrossRef] 10. Fan, X.; Xu, Q.; Scaringi, G.; Li, S.; Peng, D. A chemo-mechanical insight into the failure mechanism of frequently occurred landslides in the Loess Plateau, Gansu Province, China. Eng. Geol. 2017, 228, 337–345. [CrossRef] 11. Jin, Y.-L.; Dai, F.-C. The mechanism of irrigation-induced landslides of loess. Yantu Gongcheng Xuebao (Chin. J. Geotech. Eng.) 2007, 29, 1493–1499. 12. Wang, Z.R.; Wu, W.; Zhou, Z.Q. Landslide induced by over-irrigation in loess platform areas in Gansu Province. Chin. J. Geol. Hazard Control 2004, 15, 43–46. 13. Close, U. Where the nountains walked. Natl. Geogr. Mag. 1922, 41, 445–464. 14. Derbyshire, E.; Dijkstra, T.; Smalley, I.; Li, Y. Failure mechanisms in loess and the effects of moisture content changes on remoulded strength. Quat. Int. 1994, 24, 5–15. [CrossRef] 15. Wang, J.; Liu, W.; Chen, W.; Liu, P.; Jia, B.; Xu, H.; Wen, L. Study on the mechanism of loess landslide induced by chlorine salt in Heifangtai terran. Jpn. Geotech. Soc. Spec. Publ. 2019, 7, 159–167. [CrossRef] 16. Xu, L.; Coop, M.R.; Zhang, M.; Wang, G. The mechanics of a saturated silty loess and implications for landslides. Eng. Geol. 2018, 236, 29–42. [CrossRef] 17. Zhang, F.; Wang, G. Effect of irrigation-induced densification on the post-failure behavior of loess flowslides occurring on the Heifangtai area, Gansu, China. Eng. Geol. 2018, 236, 111–118. [CrossRef] 18. Lei, X. Geo-Hazards in Loess Plateau and Human Activity; Science Press: Beijing, China, 2001. 19. Guangtao, H. Landslide Dynamics; Geological Publishing House: Beijing, China, 1995. 20. Guorui, G. Formation and development of the structure of collapsing loess in China. Eng. Geol. 1988, 25, 235–245. [CrossRef] 21. Zhou, Y.; Tham, L.; Yan, W.; Dai, F.; Xu, L. Laboratory study on soil behavior in loess slope subjected to infiltration. Eng. Geol. 2014, 183, 31–38. [CrossRef] 22. Wu, W.; Wang, N. Lanslides in Gansu Province; Lanzhou University Press: Lanzhou, China, 2006. 23. Zhang, M.-S.; Liu, J. Controlling factors of loess landslides in western China. Environ. Earth Sci. 2010, 59, 1671–1680. [CrossRef] 24. Maosheng, Z.; Tonglu, L. Triggering factors and forming mechanism of loess landslides. J. Eng. Geol. 2011, 19, 530–540. 25. Ge, S.; McKenzie, J.; Voss, C.; Wu, Q. Exchange of groundwater and surface-water mediated by permafrost response to seasonal and long term air temperature variation. Geophys. Res. Lett. 2011, 38, 14. [CrossRef] 26. McKenzie, J.M.; Voss, C.I.; Siegel, D.I. Groundwater flow with energy transport and water–ice phase change: Numerical simulations, benchmarks, and application to freezing in peat bogs. Adv. Water Resour. 2007, 30, 966–983. [CrossRef] 27. Xiong, H.; Liu, G.; Cui, Z. The characteristics and vertical zone spectrum of natural disasters in the Tianshan Mountains, Xinjiang. Chin. Geogr. Sci. 1999, 9, 126–133. [CrossRef] 28. Li’nan, L.; Shouding, L.; Yue, J.; Yaheng, B.; Yu, L. Failure mechanism of loess landslides due to saturated-unsaturated seepage-case study of Gallente landslide in Ili, Xinjiang. J. Eng. Geol. 2017, 25, 1230–1237. 29. Yang, L.; Wei, Y.; Zhu, S. Formation mechanism of the Kezilesai Landslide in Yining County, Xinjiang and its dynamic characteristics. Chin. J. Geol. Hazard Control 2018, 29, 18–24. 30. Wang, W.; Yin, Y.; Zhu, S.; Wei, Y.; Zhang, N.; Yan, J. Dynamic analysis of a long-runout, flow-like landslide at Areletuobie, Yili River valley, northwestern China. Bull. Eng. Geol. Environ. 2019, 78, 3143–3157. [CrossRef] 31. Zhuang, M.; Wei, Y.; Shao, H.; Zhu, S. Type and characteristics of loess landslides in Piliqing River. Yili of Xinjiang Uygur autonomous region. Chin. J. Geol. Hazard Control 2018, 29, 54–59. 32. Zhu, S.; Yin, Y.; Wang, W.; Wei, Y. Mechanism of Freeze-thaw Loess Landslide in Yili River Valley, Xinjiang. Acta Geosci. Sin. 2019, 40, 339–349. 33. An, H.; Liu, P. Genesis and influencing factors of loess landslides in Yili region in Xinjiang. J. Geol. Hazards Environ. Preserv. 2010, 21, 22–25. 34. Chen, P. Causal Analysis and Stability Evaluation of Loess Landslide in Yili Region of Xinjiang–A Case Study of Alar Village Landslide, Recent Research on Engineering Geology and Geological Engineering. In Proceedings of the 2nd GeoMEast Inter- national Congress and Exhibition on Sustainable Civil Infrastructures, Egypt 2018—The Official International Congress of the Soil-Structure Interaction Group in Egypt (SSIGE), Giza, Egypt, 24–28 November 2018; Springer: Berlin/Heidelberg, Germany, 2018; p. 94.