Open Access

Landslide damage along Araniko highway and Pasang Lhamu highway and regional assessment of landslide hazard related to the Gorkha, Nepal earthquake of 25 April 2015

Geoenvironmental Disasters20174:14

https://doi.org/10.1186/s40677-017-0078-9

Received: 9 August 2016

Accepted: 30 March 2017

Published: 5 April 2017

Abstract

Background

The Gorkha, Nepal Mw 7.8 earthquake of 25 April 2015 triggered a large number of coseismic landslides in a broad area. Two highways, Araniko Highway and Pasang Lhamu Highway, that connect Tibet of China and Nepal, were affected seriously by these landslides. The purpose of this study was to investigate the landslide damage along the two highways, construct a detailed and complete inventory of coseismic landslides in the 5-km buffer area of the Araniko Highway, and perform a regional assessment of landslide hazard in the affected area.

Findings

Based on visual interpretation of high-resolution satellite images, field investigations, and GIS technology, we investigated the coseismic landslides along the Araniko Highway and Pasang Lhamu Highway. A detailed point-based inventory of coseismic landslides was constructed and spatial distributions of the landslides were analyzed. Correlations between the landslides and five controlling factors, i.e. elevation, slope angle, slope aspect, lithology, and seismic intensity, were illustrated statistically which permitted to assess landslides hazard in a larger rectangle area.

Conclusions

We examined the coseismic landslides of the 2015 Gorkha earthquake that blocked or damaged the Araniko Highway (117.3 km) and Pasang Lhamu Highway (139.3 km) in Nepal. Results show 35 coseismic landslides damaged the Araniko Highway along a total length 1,415 m. The total volume of them was estimated to be 0.37 million m3. We delineated 89 coseismic landslides that damaged the Pasang Lhamu Highway, where the total length of the damaged or buried roads is about 2,842 m and the total volume of the 89 landslides is about 1.47 million m3. In the 5-km buffer area along the Araniko Highway, we mapped 3,005 landslides caused by the Gorkha earthquake. The landslide number density of the study area is 2.925 km-2. The places with elevations 2,000-2,500 m have the highest landslide concentration. Landslide number density values increase with the slope angle. The slope aspects E and SE correspond to the highest concentrations of coseismic landslides. The underlying bedrock of Precambrian rocks-1 (Pc1) registered the largest landslide number density. The area of seismic intensity IX has a much higher LND value than that of the intensity VIII. We used the weigh index method to perform landslide hazard assessment in the 5-km buffer area on either side of the highway, which shows a success ratio of 85.9%. This method has been applied to a larger area mainly encompassing Rasuwa and Sindhupalchok counties of Nepal.

Keywords

Gorkha earthquakeCoseismic landslidesField investigationVisual interpretationLandslide hazard assessment

Introduction

The 25 April 2015 Gorkha, Nepal Mw 7.8 earthquake caused more than 8,800 fatalities and enormous economic losses. It also triggered a large number of coseismic landslides, mainly shallow and disrupted landslides and a few deep-seated landslides, some of which buried villages, roads, and valleys (Hashash et al. 2015; Moss et al. 2015; Dahal 2016; Gnyawali et al. 2016; Martha et al. 2016; Wang et al. 2016; Xu et al. 2016a). The affected areas include Central Nepal and Gyirong and Nielamu counties of southern Tibet, China. The coseismic landslides seriously damaged two highways, Pasang Lhamu Highway and Araniko Highway, connecting China and Nepal. After the event, several research teams carried out field investigations of seismic damages and earthquake-triggered landslides (Collins and Jibson 2015; Hashash et al. 2015; Sun and Yan 2015; Kargel et al. 2016; Lacroix 2016; Sharma et al. 2016). Until now, however, little work focuses on the landslides that damaged these two highways. Although the materials of the coseismic landslides blocking the two main roads have been cleaned up in time, some new landslides were triggered by aftershocks or strong rainfalls, resulting in further damage. Therefore, identifying the landslides destroying the roads and assessment of landslide hazard is very important for prevention and mitigation of future geologic hazard around these two roads. In this work, we firstly identified the coseismic landslides that destroyed the Lhamu Highway and Araniko Highway using field investigation and visual interpretation of satellite images. Then we constructed a detailed inventory map containing 3,005 individual coseismic landslides in the buffer area of 5 km to the Araniko Highway. Next, correlations between the 3,005 landslides and five landslide controlling factors were analyzed. Finally, we performed landslide hazard assessment for a larger area affected by the Gorkha earthquake using the weigh index (WI) method.

Data and methods

The study area

Despite its large magnitude, the Gorkha earthquake did not produce visible ruptures on the surface, which was confined to the subsurface at depths 10–15 km (Angster et al. 2015; Avouac et al. 2015; Hashash et al. 2015; Parameswaran et al. 2015; Duputel et al. 2016; Elliott et al. 2016). The earthquake-affected area is mainly in the east to the epicenter (28.23°N, 84.731°E), likely associated with the eastward rupturing directivity (Wang and Fialko 2015; Koketsu et al. 2016), from which we selected is a rectangular area as the study area, which has a length of 113 km in east-west direction and width of 92 km in north-south direction (Fig. 1), covering 10,396 km2. From north to south, the elevation of the study area generally declines from 7,975 m to 387 m, i.e. more than 7,500 m elevation drop in an about 100 km-wide zone. The area encompasses the Rasuwa and Sindhupalchok counties of Nepal (Fig. 1). The Araniko Highway passes through Sindhupalchok county and the Pasang Lhamu Highway passes through Rasuwa county, respectively. Based on the seismic intensity map released by the China Earthquake Administration (www.cea.gov.cn), most of the study area lies in the IX intensity zone, and part in VIII and VII intensity zones (Fig. 1).
Fig. 1

Shaded topographic relief map showing the study area (big black box) and two highways

Data

The satellite images for landslide interpretation are from the Google Earth (GE) platform. After the earthquake occurred, several organizations have implemented specialized tasks to obtain post-earthquake satellite images. Some of the images with very high resolution (1 m or better) are available on the Google Earth platform. In addition, pre-earthquake images with high quality and resolution in the area are also available on the GE platform. These images allow researchers to map co-seismic landslides conveniently and accurately. The regional DEM for analyzing correlations between topography and coseismic landslides were derived from SRTM DEM in 3-arc-second resolution (Fig. 2a). The slope angle map (Fig. 2b) and aspect map (Fig. 2c) were derived from the regional DEM on the GIS platform. The geologic map (Fig. 2d) of the study area was clipped and revised from “World Geologic Maps” on the USGS Website (www.usgs.gov).
Fig. 2

Maps showing controlling factors of coseismic landslides of the study area. a Elevation. b Slope angle. c Aspect. d Lithology (sources are mentioned in the text)

Methods

Landslide identification

In this study, we used two methods to identify landslides, i.e. visual interpretation of pre- and post-earthquake satellite images and field investigation. Computer screen-based visual interpretation of satellite images is the most widely used method for earthquake-triggered landslide mapping which permits to prepare high-quality landslide inventories (Xu 2015). As a supplement and verification of results from visual interpretation, we carried out several days of field investigation mainly along the Pasang Lhamu Highway and Araniko Highway.

Spatial distribution and hazard assessment of landslides

The Gorkha, Nepal earthquake affected a very large area about tens of thousands of square kilometers. Immediately after the quake, it was difficult to construct a detailed and complete landslide inventory throughout the affected area. Fortunately, spatial distribution of the partial affected area can represent the overall spatial patterns of landslides under some conditions (Lee et al. 2008; Xu et al. 2013a). Therefore, we selected a 5-km buffer area on either side of the Araniko Highway to construct a detailed landslide inventory. Although we prepared a polygon-based inventory of landslides that directly damaged Araniko Highway and Pasang Lhamu Highway, we chose point-based inventory of coseismic landslides and landslide number density (LND, defined as the number of landslides per square kilometers (Xu et al. 2013b) to conduct analysis of the spatial distribution and hazard assessment of landslides. The reasons include: (1) The precise source area of a landslide is very difficult to be distinguished from the whole landslide area because the boundaries of the source area, movement area, and accumulation area of the landslide are usually in the subsurface, thus cannot be exactly delineated, which perhaps reduce the objectiveness of landslides hazard assessment. (2) Preparation of a point-based landslide inventory is relatively time-saving, permitting to carry out a quick regional assessment of earthquake-triggered landslides. Five controlling factors, including elevation, slope angle, slope aspect, lithology, and seismic intensity were taken into account for a statistical analysis. Currently, many statistical methods are available for landslide hazard assessment (Xu et al. 2012; Feng et al. 2016; Pathak 2016; Tsangaratos and Ilia 2016), among which the bivariate statistical analysis method has been widely used in various areas because it is time-saving and does not need complex calculations (Xu et al. 2013b). In this study, a weight index (WI) model was employed to perform landslide susceptibility mapping in the 5-km buffer area aforementioned. This WI method is based on a bivariate statistical analysis based on calculating landslide number density (LND). In this method, the weigh value of each factor class is defined as the natural logarithm of the LND in the class divided by the LND of the whole area (Sarkar et al. 2008; Yalcin 2008; Xu et al. 2013b):
$$ \begin{array}{l}\mathrm{W}{\mathrm{I}}_i= \ln \left(\mathrm{L}\mathrm{N}{\mathrm{D}}_{\mathrm{i}}/\mathrm{LND}\right)\\ {}\kern3.12em = \ln \left(\left(\mathrm{L}{\mathrm{N}}_{\mathrm{i}}/\mathrm{Are}{\mathrm{a}}_{\mathrm{i}}\right)/\left(\mathrm{L}\mathrm{N}/\mathrm{Are}\mathrm{a}\right)\right)\end{array} $$
(1)
where WI i is the weight of the factor-class i, LND i is the landslide number intensity within the area of the ith factor class, and LND is the landslide number intensity in the whole area. In this study, the value of LND is 3,005/1,027.4 km2 = 2.925 km-2.

Findings, results and analysis

Landslides on satellite images

In this section, we present several groups of comparisons of satellite images and field photos of coseismic landslides to illustrate the excellent capacity of detecting coseismic landslides on high-resolution satellite images. The satellite images used in this study are from the GE platform collected in early May, 2015. The red solid arrow on Fig. 3 shows a coherent landslide (27.87°N, 85.911°E) with clear exposed bedrocks in the landslide source area and partly damaged vegetation stayed at its deposit area. The red dotted line defines several shallow, disrupted landslides along the Araniko Highway road. Due to the high resolution and quality of the satellite image, the locations and boundaries of the landslides can be mapped correctly and conveniently on the ortho images.
Fig. 3

Coseismic landslides at the Araniko Highway. a Satellite image of 4 May 2015. b Field photo of 14 June 2015 (by Chong Xu, view to south). The solid and dotted arrows in (a) and (b) show the same places, respectively

Figure 4 shows two rockfalls (27.927°N, 85.932°E) occurred on the upper slope at an inspection station of Nepal, which originated from nearby the ridge of the reverse slope and accumulated into two conical heaps with two narrow runout paths. The broken accumulate materials are dangerous for the structures on the toe of the slope. Despite different expressions of the rockfalls on the image and field photos due to the image stretching caused by steep topography, the rockfalls can be easily identified on the satellite image with the aid of field investigations. They have short runout distances on the image, whereas the actual runout distances of them are likely longer. This is because the slope of the rockfalls occurrence is almost vertical. Small rockfalls or falling stones are more susceptible than large deep-seated landslides on such a reverse slope.
Fig. 4

Rockfalls on the upper slope of an inspection station of Nepal. a Satellite image of 4 May 2015. b Field photo of 14 June 2015 (by Chong Xu, view to northwest). The solid and dotted arrows in (a) and (b) indicate the same places, respectively

After the main shock, a series of aftershocks and rainfalls struck the affected area and caused more landslides. For example, the satellite image of 3 May 2015 (Fig. 5a) shows quite a few shallow, disrupted landslides (located at 28.064°N, 85.225°W) that occurred in weathering layers and blocked the Pasang Lhamu Highway. Fig. 5b shows the road was blocked by a secondary landslide caused by a heavy rainfall in the area. Information from local residents suggests that the landslide accumulation material that blocked the road was not triggered by the main shock, but by a strong rainfall a few days before. All the landslide materials related to the main shock and subsequent triggers blocking the roads have been cleared up or were being cleared away in time to keep the traffic flowing.
Fig. 5

A series of shallow, disrupted landslides blocking the Pasang Lhamu Highway. a Satellite image of 3 May 2015. b Field photo of 15 June 2015 (by Chong Xu, view to southeast) showing the road was blocked by a secondary landslide caused by a heavy rainfall. The red solid arrow in (a) shows the location of field photo in (b)

The satellite image (Fig. 6a) shows an area with high density of coseismic landslides, dominated by shallow, disrupted landslides. The red solid arrows wherein indicate a large rock slide located at 28.079°N, 85.194°W. It occurred at the lower part of the slope and blocked the valley. However, they did not create a lake because of the small area of the catchment upstream. The red dotted arrows show a shallow, disrupted landslide originated from a ridge (located at 28.08°N, 85.208°W). Most of the landslides shown in Fig. 6a are distributed along the rivers, likely associated with river incision or loose deluvium with high landslide susceptibility.
Fig. 6

An area characterized by high density of coseismic landslides. a Satellite image taken on 3 May 2015. b Field photo taken on 15 June 2015 (by Chong Xu, view to west). The solid and dotted arrows denote the same places, respectively

Landslide damage to the two roads

The Pasang Lhamu Highway and Araniko Highway are two most important roads connecting Nepal and China. The Araniko Highway links Kathmandu, Nepal and Nielamu County, China. In this study, the section of the Araniko Highway between the place (27.987262°N, 85.982552°E) nearby Zhangmu Port and the location (27.678835°N, 85.349647°E) southeast to Kathmandu was selected as the target to investigate the damage of coseismic landslides on the road. This section of Araniko Highway is about 117.3 km long. Based on visual interpretation of high-resolution satellite images and field investigations, we delineated 35 coseismic landslides damaging the Araniko Highway at 36 places. The longest section of the road damaged is about 156 m long which was buried by a landslide at 27.91744°N, 85.93064°E. Considering the previous correlations between area and volume of individual landslides (Larsen et al. 2010; Xu et al. 2016b) and field investigations, the total volume of the 35 coseismic landslides was estimated to be about 0.37 million m3. Figure 7 shows the Araniko Highway damaged by coseismic landslides on satellite images at four places. Table 1 lists the detailed information on the 35 coseismic landslides and hazards on the road they caused.
Fig. 7

Damaged sites along Araniko Highway by coseismic landslides have been seen on satellite images. The detailed information of the landslides and associated damages on the road is listed in Table 1. a the No. 1, 2, and 3 landslides, (b) the No. 21, 22, and 23 landslides, (c) the No. 24, 25, 26, and 27 landslides, and (d) the No. 31, 32, 33, and 34 landslides along Araniko Highway. All the satellite images were acquired on 4 May, 2015

Table 1

Information of damage along the Araniko Highway and associated coseismic landslides

No.

Longitude (°)

Latitude (°)

Length of road damaged (m)

Area (m2)

Estimated volume (m3)

1

85.98343

27.9868

45

29131

50000

2

85.98196

27.98562

52

6180

12000

3

85.9791

27.98369

35

22953

60000

4

85.96235

27.97215

16

3648

6000

5

85.96186

27.97166

27

3756

7000

6

85.96482

27.96926

29

8383

20000

7

85.96511

27.96912

30

ditto

ditto

8

85.95799

27.96591

10

435

400

9

85.93064

27.91744

156

31816

20000

10

85.92627

27.91352

19

4233

8000

11

85.9227

27.90877

5

972

1000

12

85.92241

27.90828

27

2271

3000

13

85.9222

27.9079

33

2630

4000

14

85.91496

27.8927

10

15890

30000

15

85.91307

27.88253

30

1739

2000

16

85.90599

27.87934

46

12025

30000

17

85.90493

27.87892

12

268

200

18

85.90185

27.87782

115

6224

10000

19

85.89957

27.87686

49

4245

6000

20

85.89595

27.87602

15

303

200

21

85.88781

27.87385

95

9237

20000

22

85.88341

27.87226

25

608

500

23

85.88309

27.87113

20

3583

5000

24

85.88062

27.85115

43

3805

5000

25

85.88092

27.85034

94

7461

15000

26

85.88045

27.84828

54

9367

20000

27

85.88063

27.84779

42

7061

15000

28

85.87291

27.82729

57

3399

5000

29

85.87352

27.82707

43

3094

5000

30

85.89329

27.8036

15

1250

1500

31

85.89425

27.80022

30

949

1000

32

85.89475

27.80006

13

522

300

33

85.89492

27.79994

7

937

1000

34

85.89613

27.7986

52

1971

2500

35

85.88417

27.77117

1

628

500

36

85.77906

27.73006

63

3778

5000

Total

  

1,415

214,751

372,100

No. 6 and No. 7 landslide-damaged sections were caused by one landslide

Table 2

Information of damaged along the Pasang Lhamu Highway and associated coseismic landslides

No.

Longitude (°)

Latitude (°)

Length of road damaged (m)

Area (m2)

Estimated volume (m3)

1

85.24384

27.82563

12

166

100

2

85.23569

27.82943

14

613

500

3

85.20104

27.83575

14

275

200

4

85.13849

27.8637

17

628

500

5

85.18591

27.98131

18

311

200

6

85.1884

27.98278

14

197

100

7

85.18866

27.98288

16

209

100

8

85.18777

27.9828

50

1144

1200

9

85.21029

28.00465

16

336

300

10

85.2181

28.01873

9

177

100

11

85.22025

28.02089

21

553

400

12

85.22087

28.02175

28

2766

4000

13

85.22311

28.02457

40

32568

80000

14

85.22349

28.02528

48

17673

50000

15

85.22334

28.02617

39

10247

20000

16

85.22309

28.02716

45

9266

20000

17

85.22299

28.02766

18

10071

20000

18

85.21933

28.0394

14

199

100

19

85.22441

28.04657

14

227

100

20

85.22547

28.04732

7

171

100

21

85.22722

28.04826

24

805

800

22

85.22892

28.05021

22

279

200

23

85.22869

28.05054

14

103

100

24

85.22923

28.05373

5

271

200

25

85.22867

28.05858

19

11564

5000

26

85.22811

28.06002

53

3790

5000

27

85.22798

28.061

67

7609

15000

28

85.22674

28.0633

57

5222

10000

29

85.22643

28.06371

19

1815

2000

30

85.22562

28.0644

28

9168

20000

31

85.22542

28.06454

15

3018

4000

32

85.22515

28.06481

14

525

300

33

85.22494

28.06501

21

2679

3000

34

85.22447

28.06545

31

5012

5000

35

85.22374

28.06637

36

3662

5000

36

85.22535

28.06689

25

2876

4000

37

85.22884

28.06806

59

11267

20000

38

85.23007

28.06827

99

20358

60000

39

85.23904

28.07238

209

54607

200000

40

85.25552

28.0772

16

920

800

41

85.2558

28.07737

11

152

100

42

85.25023

28.07813

13

500

300

43

85.277

28.09335

18

241

100

44

85.27853

28.09454

15

169

100

45

85.2867

28.10178

9

95

100

46

85.28784

28.104

7

109

40000

47

85.2877

28.10459

104

15961

3000

48

85.28754

28.1055

30

2082

3000

49

85.28715

28.10675

14

577

500

50

85.31217

28.10857

19

1711

2000

51

85.31202

28.10873

22

1306

1500

52

85.31183

28.11008

60

10024

250000

53

85.31135

28.11061

49

4482

8000

54

85.31117

28.11098

17

1550

2000

55

85.29239

28.11159

24

555

500

56

85.31072

28.11175

41

3888

6000

57

85.31043

28.11217

55

7463

15000

58

85.29454

28.11298

6

168

100

59

85.30909

28.11914

11

117

100

60

85.30815

28.12234

14

5918

12000

61

85.30685

28.12406

4

357

300

62

85.34242

28.17233

25

957

1000

63

85.34248

28.17269

31

4223

8000

64

85.34254

28.17314

14

1996

2000

65

85.3426

28.17358

63

21402

60000

66

85.34237

28.178

18

1374

2000

67

85.34225

28.17896

11

695

800

68

85.3439

28.18229

14

197

100

69

85.34455

28.18387

17

4005

6000

70

85.34617

28.18621

30

33718

100000

71

85.34634

28.18678

57

32800

100000

72

85.34741

28.18903

16

565

500

73

85.34761

28.18932

24

8214

10000

74

85.34883

28.19127

59

18536

30000

75

85.34972

28.19332

14

445

300

76

85.35092

28.19643

19

398

300

77

85.35147

28.19703

41

4677

80000

78

85.35182

28.19742

27

3538

5000

79

85.35224

28.19803

13

1716

1000

80

85.35285

28.19931

19

3032

3000

81

85.35292

28.19967

12

1805

1000

82

85.35304

28.20057

43

2852

5000

83

85.35425

28.20403

41

1693

2500

84

85.35507

28.20837

38

4976

10000

85

85.35804

28.2197

102

13969

40000

86

85.36041

28.22164

19

3995

5000

87

85.36106

28.22261

107

14279

30000

88

85.36698

28.25649

44

2549

4000

89

85.37802

28.27514

92

21169

60000

Total

  

2,842

500,552

1,470,600

The Pasang Lhamu Highway connects Kathmandu, Nepal and Gyirong County. The length of the section between the point (27.735268°N, 85.305939°E) northwest to Kathmandu and the point (28.278972°N, 85.377904°E) China-Nepal border is about 139.3 km. Visual interpretation of satellite images and field investigations allowed us to delineate 89 coseismic landslides that damaged the Pasang Lhamu Highway. The total length of damaged or buried roads is about 2,842 m, of which the longest section is about 209 m long, caused by a landslide located at 27.91744°N, 85.93064°E. The total volume of the 89 landslides was estimated to be 1.47 million m3. Figure 8 shows satellite images of coseismic landslides along the Pasang Lhamu Highway. Table 2 shows the detailed information on the 89 coseismic landslides and their hazards on the road.
Fig. 8

Damaged sites along Pasang Lhamu Highway by coseismic landslides on satellite images. The detailed information of the landslides and associated damages on the road is listed in Table 2. a the No. 11 and 12 landslides, (b) the No. 28~36 landslides, (c) the No. 59, 60, and 61 landslides, (d) the No. 62, 63, 64, and 65 landslides along Pasang Lhamu Highway. All the satellite images were acquired on 3 May, 2015

Landslide inventory along the Araniko highway

On either side of the 117.3 km-long Araniko Highway, we constructed a 5-km buffer region to construct a detailed and complete point-based coseismic landslide inventory. The area of this buffer region is 1,027.4 km2. Individual coseismic landslides were mapped as points at the central of the landslide. Consequently, we mapped 3,005 coseismic landslides in the area (Fig. 9), and calculated the landslide number density to be 3,005/1,027.4 km2 = 2.925 km-2. The spatial distribution of the coseismic landslides along the Araniko Highway is quite uneven. Most of the landslides occurred in the mountainous areas to the north, where the landslide inventory is complete and detailed, i.e. small landslides are included. The buffer area only accounts for less 10% than the primary affected area of the main shock. The buffer area is approximately normal to the strike of the seismogenic structure (EW trending). Usually the seismic landslide density along the causative fault is uniform. Therefore, we estimated the Gorkha quake triggered at least 30,000 landslides. Several other teams have released coseismic landslides related to the Gorkha quake. For example, Kargel (Kargel et al. 2016) only mapped 4,312 coseismic and postseismic landslides. A team from British Geological Survey et al. (British Geological Survey et al. 2015) identified about 5,600 coseismic landslides as polylines marking the location and movement path from head to toe of a landslide. Therefore, there might be false negative errors (omissions) in these released inventories of landslides triggered by the Gorkha quake.
Fig. 9

Distribution map of coseismic landslides in the 5-km buffer area on either side the Araniko Highway

Spatial distribution of landslides along the Araniko highway

As a common index to reflect landslide abundance, landslide number density was employed as the index to measure spatial distribution of the 3,005 landslides in the 5-km buffer area of the Araniko Highway. In this study, five controlling factors including elevation, slope angle, slope aspect, lithology, and seismic intensity were selected to analyze their correlations with the landslides (Figs. 10, 11, 12, 13 and 14). The DEM of the area was derived from SRTM in 3 arc-second, which permitted to determine the elevations of the buffer area vary from 610 m to 4,750 m. The study area was divided into six classes based on 500 m of elevation intervals, i.e. 610–1000 m, 1000–1500 m, 1500–2000 m, 2000–2500 m, 2500–3000 m, and 3000–4750 m (Fig. 10). The elevations of most of the area (814.97 km2, 79.3% of the total) are lower than 2,000 m. The class 1000–1500 m occupies the largest area, which is 433 km2, accounting for 42.1% of the total. The class 2,000-2,500 m registered the largest LND value, which is 10.3 km-2. The landslide number density values gradually decrease at the elevations higher than 2,500 m and lower than 2,000 m.
Fig. 10

Relationships between elevation class (horizontal axis), its area (vertical axis on left) and landslide number density (LND, vertical axis on right)

Fig. 11

Same as Fig. 10 but for slope angle (horizontal axis, unit: degree)

Fig. 12

Same as Fig. 10 but for slope aspects (horizontal axis)

Fig. 13

Same as Fig. 10 but for lithology (horizontal axis)

Fig. 14

Same as Fig. 10 but for seismic intensity (horizontal axis)

Slope angle is an important controlling factor of coseismic landslides. In this study, the slope angle of the buffer area ranges from 0° to 74°, which was divided into 9 classes with an interval of 5°. Majority of the area (780.7 km2, 76% of the total) has slope angles lower than 30°. As shown in Fig. 11, the landslide number density increases with the growing slope angle. The class >40° corresponds to the largest LND value, which is 12.92 km-2. In addition, the LND curve shows a concave form, implying the LND increases with the slope angle gradually. This suggests a very strong control of the slope angle on the coseismic landslides. Such a situation is also common in other earthquake events (Gorum et al. 2014; Xu et al. 2014; Xu et al. 2015; Tian et al. 2016).

Slope aspects (or facing directions) can affect the pattern of coseismic landslides because slopes with different aspects have varied responses to the movement directions of blocks or the propagating direction of seismic waves. The study area has nine classes of slope aspects, i.e. flat, north (N), northeast (NE), east (E), southeast (SE), south (S), southwest (SW), west (W), and northwest (NW). Fig. 12 shows the correlations between the slope aspect, area of its classes and landslide number density. It is clear that the slope aspects E and SE correspond to the two largest LND values, which are 4.87 km-2 and 4.58 km-2, respectively. This is perhaps related to the movement direction of the hanging wall of the seismogenic fault or the propagation direction of seismic wave (Shen et al. 2016). The study area is located east of the epicenter of the Gorkha main shock, and thus the propagating direction of seismic waves is eastward. During the Gorkha earthquake, the hanging wall of the fault, where the buffer area is located, moved toward south and probably generated inertia effect to the south. In addition, the slopes of southward aspect in the area are easily exposed to sunlight and rainfall, thus leading to widespread weathering layers and high susceptibility to seismic landslides.

The Gorkha earthquake affected area can be divided into a series of east-west trending major tectonic regions by three major active fault zones, including MFT, MBT, and MCT (Le Fort 1975; Nakata 1989; Upreti 1999; Wesnousky et al. 1999; Mukherjee 2015). Based on the geologic map of South Asia, the study area has five classes of lithology (rock types) generally from north to south, i.e. Tertiary igneous rocks (Ti), Precambrian rocks-1 (Pc1), Precambrian rocks-2 (Pc2), Precambrian rocks-3 (Pc3), and Mesozoic intrusive rocks (Mi). Figure 13 shows the correlations between lithology, its class area and landslide number density. The three groups of Precambrian rocks cover most of the area, which is 942.8 km-2, occupying 91.8% of the total. The lithology class Pc1 registered the largest landslide number density, which is 8.32 km-2.

Seismic intensity and peak ground accumulation (PGA) are two common proxies representing the degree of seismic energy and often used to explore the effect of earthquakes on landslides. The PGA distribution map released by USGS (www.usgs.gov) is rather irregular in the 5-km buffer area of the Araniko Highway because the buffer area is relatively small and there are perhaps significant errors generated by numerical simulation and limited stations. Therefore, we preferred to analyze the correlation between coseismic landslides and seismic intensity in this study. The seismic intensity map of the Gorkha earthquake was produced by China Earthquake Administration (CEA) (Fig. 1). Only VIII and IX intensity zones appear in the study area, which have the landslide number density values 0.32 km-2 and 3.5 km-2, respectively (Fig. 14). Despite merely two data points, these data show a positive correlation with the coseismic landslides, i.e. the place with larger seismic intensity has a higher landslide number density.

Landslide hazard assessment

In the aforementioned 5-km buffer area of the Araniko Highway, the WI vales were calculated to each class of all the five controlling factors, respectively. Then, the weighted thematic maps of the five factors were produced and were superposed to derive a landslide hazard index (LHI) map:
$$ \mathrm{L}\mathrm{H}\mathrm{I}=\mathrm{W}{\mathrm{I}}_{\mathrm{Elevation}}+\mathrm{W}{\mathrm{I}}_{\mathrm{Slope}\ \mathrm{angle}}+\mathrm{W}{\mathrm{I}}_{\mathrm{Slope}\ \mathrm{aspect}}+\mathrm{W}{\mathrm{I}}_{\mathrm{Lithology}}+\mathrm{W}{\mathrm{I}}_{\mathrm{Seismic}\ \mathrm{intensity}} $$
(2)
The WI values indicate the relative importance of each factor to landslide hazard. Positive WI values mean the factor-class area is prone to coseismic landslides, whereas negative WI values represent the opposite. WI values close to zero mean moderate probabilities of occurrence of coseismic landslides. Results show the LHI of the area is in the range from -23.546 to 4.48. In order to examine the validity of the model, the 3,005 coseismic landslides aforementioned were employed to compare the known landslides with the landslide hazard index map. By referring to a common method, the regional area was categorized into 100 classes with a same area by the LHI value and the percentages of landslide number in each class were calculated. Then, a correlation curve between cumulative area percentages and cumulative percentage of landslide number from high to low LHI in a descending order was drawn (Fig. 15). It shows the area under the curve (AUC) is as much as 85.9%, i.e. a quite satisfactory success ratio. The curve also reveals that 10% of the area with the highest landslide hazard index could cover 1,514 landslides, about 50.4% of the total. Likewise, 20 and 30% of the area with the highest landslide hazard index can account for 2,207 and 2,645 landslides, about 73.4 and 88% of the total, respectively.
Fig. 15

The area under the curve representing the success ratio of the landslide hazard assessment. Area % means the percentage of area to the study area for each factor class. Landslide number % means the percentage of landslide number in a factor class to the total landslide number

We applied the WI values in Table 3 to a larger area, i.e. the rectangle in Fig. 1, to construct a landslide hazards map. For the areas with factor-attribute values beyond the ranges of the 5-km buffer area, they are classified into the classes that are closest to them. The elevation ranges from 291 m to 7968 m in the area. The area with elevation less than 610 m was classified into the class 610–1000 m and the area with elevation higher than 4,750 m was classified into the class 3000–4750 m. The range of slope angle of the rectangle area is 0–81.7°, therefore, the range of 74°–81.7° was merged into the class 40–74°. The northern part of the rectangle area outcrop several other lithology types, such as Quaternary perennial ice and snow, Quaternary sediments, Neogene granite, Triassic metamorphic and sedimentary rocks, and Cretaceous sedimentary rocks. The northern rock group located at the northern part of the study area was not subdivided further because of limited geologic information there. The WI values of these lithology types were assigned the values same as the type Tertiary igneous rocks (Ti). The rectangle area includes three seismic intensity zones, i.e. VII, VIII, and IX (Fig. 1). The seismic intensity VII is out of the range of the 5-km buffer area and its WI value was calculated by linear extrapolation, which is -4.58. Subsequently, we constructed the LHI map of the rectangle area. We divided the map into four classes, i.e. very low, low, high, and very high, based on three breakpoints of the index values, i.e. -3, -1, 1, and 3. Figure 16 shows the derived landslide hazard map of the study area. The high zone and very high zone show a NWW-SEE directed distribution, which is coincident with the seismogenic fault and earthquake damage area. Figure 17 shows a three-dimensional view on the landslide hazard map. We overlaid the hazard map with 50% transparency on satellite images of the Google Earth platform. However, several limitations of this result should be noted, including (1) Only five common factors were considered, while there should more factors can affect the occurrence of the coseismic landslides, such as rivers and seismogenic fault. (2) The weight value method is a bivariate method, interactions among factors cannot be considered; and (3) WI values were calculated based on the 3,005 landslides in the 5-km buffer area along the Araniko Highway. Of course, it is inferior to use of a complete inventory of landslides throughout the earthquake-affected area. These limitations are expected to be improved in future research.
Table 3

Weight index values of various classes of five controlling factors

Factor

Class area

Landslide number

LND

WI

Elevation

 610–1000 m

155.94

159

1.02

-1.05

 1000–1500 m

433

438

1.01

-1.06

 1500–2000 m

226.03

1000

4.42

0.41

 2000–2500 m

95.24

981

10.3

1.26

 2500–3000 m

58.83

327

5.56

0.64

 3000–4750 m

58.39

100

1.71

-0.54

Slope angle

 0–5°

139.07

15

0.11

-3.3

 5–10°

86.75

42

0.48

-1.8

 10–15°

96.61

71

0.73

-1.38

 15–20°

141.86

144

1.02

-1.06

 20–25°

163.49

290

1.77

-0.5

 25–30°

152.92

458

3.00

0.02

 30–35°

115.59

641

5.55

0.64

 35–40°

76.65

640

8.35

1.05

 40–74°

54.5

704

12.92

1.49

Slope aspect

 Flat

16.63

6

0.36

-2.09

 N

120.07

300

2.50

-0.16

 NE

131.27

459

3.50

0.18

 E

112.63

549

4.87

0.51

 SE

119.02

545

4.58

0.45

 S

121.36

305

2.51

-0.15

 SW

158.51

295

1.86

-0.45

 W

127.22

312

2.45

-0.18

 NW

120.72

234

1.94

-0.41

Lithology

 Ti

29.34

101

3.44

0.16

 Pc1

308.10

2564

8.32

1.05

 Pc2

288.82

324

1.12

-0.96

 Pc3

345.91

16

0.05

-4.15

 Mi

55.26

0

0

-14.89

Seismic intensity

 VIII

185.60

60

0.32

-2.2

 IX

841.84

2945

3.5

0.18

Lithology type Mi registered no landslide. In order to avoid ln(0) in calculating WI value, we assigned the LND value of the class Mi with 0.000001, i.e. a small enough value, and the WI value of the class is -14.89

Fig. 16

Landslide hazard map for a part of the Gorkha earthquake region

Fig. 17

A printing screen showing a three-dimensional perspective on the landslide hazard map. View to north

Conclusions

Based on high-resolution satellite images, field investigation, and GIS technology, we examined the coseismic landslides of the 2015 Gorhka, Nelpal earthquake that blocked or damaged the Araniko Highway (117.3 km) and Pasang Lhamu Highway (139.3 km) in Nepal. Results show 35 coseismic landslides damaged the Araniko Highway with a total length of the sections of the damaged road about 1,415 m. The total volume of these 35 coseismic landslides was estimated to be about 0.37 million m3. We delineated 89 coseismic landslides that damaged the Pasang Lhamu Highway. The total length of the damaged or buried roads is about 2,842 m. The total volume of these 89 landslides was estimated to be 1.47 million m3. In the 5-km buffer area on either side of the Araniko Highway, we mapped 3,005 landslides caused by the Gorkha earthquake. The landslide number density of the buffer area is 2.925 km-2. Correlations between the landslides and five controlling factors were analyzed based on the bivariate method. The results show the elevation class 2,000–2,500 m has the highest landslide concentration. The landslide number density value increases with the slope angle. The slope aspects E and SE correspond to the highest concentrations of coseismic landslides. The underlying bedrock of Precambrian rocks-1 (Pc1) registered the largest landslide number density. The area with seismic intensity IX has a much higher LND value than the area of the intensity VIII. We used the weigh index method to perform landslide hazard assessment in the 5-km buffer area. Result shows the success ratio as high as 85.9%. In addition, we prepared a landslide hazard assessment map for a larger area encompassing Rasuwa and Sindhupalchok counties of Nepal. It indicates the areas most likely to be prone to coseismic landslides, which would be helpful for constructing a more detailed and complete coseismic landslide inventory map throughout the earthquake-affected area subsequently. The result is also helpful predict the locations of landslides triggered by subsequent events, e.g. strong aftershocks and rainfalls, which would be a scientific reference for restoration, reconstruction, and landslide reduction and mitigation in the Gorkha earthquake-affected area.

Declarations

Acknowledgments

This research was supported by the National Natural Science Foundation of China (41661144037, 41472202, 41472264) and the Special Project for China Earthquake Research (201508024). We are very grateful to the comments of two anonymous reviewers and the help of Ranjan Kumar Dahal (Editor).

Competing interests

The authors declare that they have no competing interests.

Authors’ contributions

CX and YT mapped landslides on GE images. CX and GL participated the field investigations. CX, BZ, and HR provided the regional geologic and tectonic data. CX drafted the manuscript. All the authors reviewed and approved the manuscript.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.

Authors’ Affiliations

(1)
Key Laboratory of Active Tectonics and Volcano, Institute of Geology, China Earthquake Administration
(2)
CCCC Highway Consultants Co., Ltd.

References

  1. Angster, S., E.J. Fielding, S. Wesnousky, I. Pierce, D. Chamlagain, D. Gautam, B.N. Upreti, Y. Kumahara, and T. Nakata. 2015. Field reconnaissance after the 25 April 2015 M 7.8 Gorkha earthquake. Seismological Research Letters 86(6): 1506–1513. doi:10.1785/0220150135.View ArticleGoogle Scholar
  2. Avouac, J.-P., L. Meng, S. Wei, T. Wang, and J.-P. Ampuero. 2015. Lower edge of locked main Himalayan Thrust unzipped by the 2015 Gorkha earthquake. Nature Geoscience 8: 701–711. doi:10.1038/ngeo2518.View ArticleGoogle Scholar
  3. British Geological Survey, Earthquakes without Frontiers, Durham University. 2015. 2015 Nepal Earthquakes mapped landslide intensity (Revision 4.0 - 19 June 2015). https://data.hdx.rwlabs.org/group/nepal-earthquake.Google Scholar
  4. Collins, B.D., and R.W. Jibson. 2015. Assessment of existing and potential landslide hazards resulting from the April 25, 2015 Gorkha, Nepal earthquake sequence. US Geological Survey.. https://pubs.er.usgs.gov/publication/ofr20151142. doi:10.3133/ofr20151142.Google Scholar
  5. Dahal, R.K. 2016. Initiatives for rockfall hazard mitigation in Nepal. Bulletin of Nepal Geological Society 33: 51–56.Google Scholar
  6. Duputel, Z., J. Vergne, L. Rivera, G. Wittlinger, V. Farra, and G. Hetényi. 2016. The 2015 Gorkha earthquake: a large event illuminating the main Himalayan thrust fault. Geophysical Research Letters 43(6): 2517–2525. doi:10.1002/2016GL068083.View ArticleGoogle Scholar
  7. Elliott, J.R., R. Jolivet, P.J. González, J.-P. Avouac, J. Hollingsworth, M.P. Searle, and V.L. Stevens. 2016. Himalayan megathrust geometry and relation to topography revealed by the Gorkha earthquake. Nature Geoscience 9: 174–180. doi:10.1038/ngeo2623.View ArticleGoogle Scholar
  8. Feng, H., A. Zhou, J. Yu, X. Tang, J. Zheng, X. Chen, and S. You. 2016. A comparative study on plum-rain-triggered landslide susceptibility assessment models in West Zhejiang Province. Earth Science 41(3): 403–415.Google Scholar
  9. Gnyawali, K.R., S. Maka, B.R. Adhikari, D. Chamlagain, S. Duwal, and A.R. Dhungana. 2016. Spatial implications of earthquake induced landslides triggered by the April 25 Gorkha earthquake Mw 7.8: preliminary analysis and findings. In International conference on earthquake engineering and post disastor reconstruction planning 24 – 26 April, 2016, Bhaktapur, Nepal, 50–58.Google Scholar
  10. Gorum, T., O. Korup, C.J. van Westen, M. van der Meijde, C. Xu, and F.D. van der Meer. 2014. Why so few? Landslides triggered by the 2002 Denali earthquake, Alaska. Quaternary Science Reviews 95: 80–94. doi:10.1016/j.quascirev.2014.04.032.View ArticleGoogle Scholar
  11. Hashash, Y.M.A., B. Tiwari, R.E.S. Moss, D. Asimaki, K.B. Clahan, D.S. Kieffer, D.S. Dreger, A. Macdonald, C.M. Madugo, H.B. Mason, M. Pehlivan, D. Rayamajhi, I. Acharya, and B. Adhikari. 2015. Geotechnical field reconnaissance: Gorkha (Nepal) earthquake of April 25, 2015 and related shaking sequence. In Geotechnical extreme event reconnaisance GEER association report No. GEER-040.. https://works.bepress.com/rmoss/47: 250 pages.Google Scholar
  12. Kargel JS, Leonard GJ, Shugar DH, Haritashya UK, Bevington A, Fielding EJ, Fujita K, Geertsema M, Miles ES, Steiner J, Anderson E, Bajracharya S, Bawden GW, Breashears DF, Byers A, Collins B, Dhital MR, Donnellan A, Evans TL, Geai ML, Glasscoe MT, Green D, Gurung DR, Heijenk R, Hilborn A, Hudnut K, Huyck C, Immerzeel WW, Jiang L, Jibson R, Kääb A, Khanal NR, Kirschbaum D, Kraaijenbrink PDA, Lamsal D, Liu S, Lv M, McKinney D, Nahirnick NK, Nan Z, Ojha S, Olsenholler J, Painter TH, Pleasants M, Pratima KC, Qi Y, Raup BH, Regmi D, Rounce DR, Sakai A, Shangguan D, Shea JM, Shrestha AB, Shukla A, Stumm D, van der Kooij M, Voss K, Wang X, Weihs B, Wolfe D, Wu L, Yao X, Yoder MR, Young N. 2016. Geomorphic and geologic controls of geohazards induced by Nepal’s 2015 Gorkha earthquake. Science 351(6269): aac8353. doi:10.1126/science.aac8353
  13. Koketsu, K., H. Miyake, Y. Guo, H. Kobayashi, T. Masuda, S. Davuluri, M. Bhattarai, L.B. Adhikari, and S.N. Sapkota. 2016. Widespread ground motion distribution caused by rupture directivity during the 2015 Gorkha, Nepal earthquake. Scientific Reports 6: 28536. doi:10.1038/srep28536.View ArticleGoogle Scholar
  14. Lacroix, P. 2016. Landslides triggered by the Gorkha earthquake in the Langtang valley, volumes and initiation processes. Earth, Planets and Space 68(1): 46. doi:10.1186/s40623-016-0423-3.View ArticleGoogle Scholar
  15. Larsen, I.J., D.R. Montgomery, and O. Korup. 2010. Landslide erosion controlled by hillslope material. Nature Geoscience 3(4): 247–251. doi:10.1038/ngeo776.View ArticleGoogle Scholar
  16. Le Fort, P. 1975. Himalayas: the collided range. Present knowledge of the continental arc. American Journal of Science 275-A: 1–44.Google Scholar
  17. Lee, C.T., C.C. Huang, J.F. Lee, K.L. Pan, M.L. Lin, and J.J. Dong. 2008. Statistical approach to earthquake-induced landslide susceptibility. Engineering Geology 100(1-2): 43–58. doi:10.1016/j.enggeo.2008.03.004.View ArticleGoogle Scholar
  18. Martha TR, Roy P, Mazumdar R, Govindharaj KB, Kumar KV. 2016. Spatial characteristics of landslides triggered by the 2015 Mw 7.8 (Gorkha) and Mw 7.3 (Dolakha) earthquakes in Nepal. Landslides. doi:10.1007/s10346-016-0763-x
  19. Moss, R.E.S., E.M. Thompson, D.S. Kieffer, B. Tiwari, Y.M.A. Hashash, I. Acharya, B.R. Adhikari, D. Asimaki, K.B. Clahan, and B.D. Collins. 2015. Geotechnical effects of the 2015 magnitude 7.8 Gorkha, Nepal, earthquake and aftershocks. Seismological Research Letters 86(6): 1514–1523. doi:10.1785/0220150158.View ArticleGoogle Scholar
  20. Mukherjee, S. 2015. A review on out-of-sequence deformation in the Himalaya. Geological Society, London, Special Publications 412(S): 67–109. doi:10.1144/SP412.13.View ArticleGoogle Scholar
  21. Nakata, T. 1989. Active faults of the Himalaya of India and Nepal. Geological Society of America Special Papers 232: 243–264. doi:10.1130/SPE232-p243.View ArticleGoogle Scholar
  22. Parameswaran, R.M., T. Natarajan, K. Rajendran, C.P. Rajendran, R. Mallick, M. Wood, and H.C. Lekhak. 2015. Seismotectonics of the April-May 2015 Nepal earthquakes: An assessment based on the aftershock patterns, surface effects and deformational characteristics. Journal of Asian Earth Sciences 111: 161–174. doi:10.1016/j.jseaes.2015.07.030.View ArticleGoogle Scholar
  23. Pathak, D. 2016. Knowledge based landslide susceptibility mapping in the Himalayas. Geoenvironmental Disasters 3(1): 8. doi:10.1186/s40677-016-0042-0.View ArticleGoogle Scholar
  24. Sarkar, S., D.P. Kanungo, A.K. Patra, and P. Kumar. 2008. GIS based spatial data analysis for landslide susceptibility mapping. Journal of Mountain Science 5(1): 52–62. doi:10.1007/s11629-008-0052-9.View ArticleGoogle Scholar
  25. Sharma, K., L. Deng, and C.C. Noguez. 2016. Field investigation on the performance of building structures during the April 25, 2015, Gorkha earthquake in Nepal. Engineering Structures 121: 61–74. doi:10.1016/j.engstruct.2016.04.043.View ArticleGoogle Scholar
  26. Shen, L., C. Xu, and L. Liu. 2016. Interaction among controlling factors for landslides triggered by the 2008 Wenchuan, China Mw 7.9 earthquake. Frontiers of Earth Science 10(2): 264–273. doi:10.1007/s11707-015-0517-4.View ArticleGoogle Scholar
  27. Sun, B., and P. Yan. 2015. Damage characteristics and seismic capacity of buildings during Nepal Ms 8.1 earthquake. Earthquake Engineering and Engineering Vibration 14(3): 571–578. doi:10.1007/s11803-015-0046-x.View ArticleGoogle Scholar
  28. Tian, Y., C. Xu, X. Xu, and J. Chen. 2016. Detailed inventory mapping and spatial analyses to landslides induced by the 2013 Ms 6.6 Minxian earthquake of China. Journal of Earth Science 27(6): 1016–1026. doi:10.1007/s12583-016-0905-z.View ArticleGoogle Scholar
  29. Tsangaratos, P., and I. Ilia. 2016. Comparison of a logistic regression and Naïve Bayes classifier in landslide susceptibility assessments: The influence of models complexity and training dataset size. Catena 145: 164–179. doi:10.1016/j.catena.2016.06.004.View ArticleGoogle Scholar
  30. Upreti, B.N. 1999. An overview of the stratigraphy and tectonics of the Nepal Himalaya. Journal of Asian Earth Sciences 17(5-6): 577–606. doi:10.1016/S1367-9120(99)00047-4.View ArticleGoogle Scholar
  31. Wang, K., and Y. Fialko. 2015. Slip model of the 2015 Mw 7.8 Gorkha (Nepal) earthquake from inversions of ALOS‐2 and GPS data. Geophysical Research Letters 42(18): 7452–7458. doi:10.1002/2015GL065201.View ArticleGoogle Scholar
  32. Wang, F., M. Miyajima, R. Dahal, M. Timilsina, T. Li, M. Fujiu, Y. Kuwada, and Q. Zhao. 2016. Effects of topographic and geological features on building damage caused by 2015.4.25 Mw7.8 Gorkha earthquake in Nepal: a preliminary investigation report. Geoenvironmental Disasters 3(1): 7. doi:10.1186/s40677-016-0040-2.View ArticleGoogle Scholar
  33. Wesnousky, S.G., S. Kumar, R. Mohindra, and V. Thakur. 1999. Uplift and convergence along the Himalayan Frontal Thrust of India. Tectonics 18(6): 967–976. doi:10.1029/1999TC900026.View ArticleGoogle Scholar
  34. Xu, C. 2015. Preparation of earthquake-triggered landslide inventory maps using remote sensing and GIS technologies: principles and case studies. Geoscience Frontiers 6(6): 825–836. doi:10.1016/j.gsf.2014.03.004.View ArticleGoogle Scholar
  35. Xu, C., X. Xu, F. Dai, and A.K. Saraf. 2012. Comparison of different models for susceptibility mapping of earthquake triggered landslides related with the 2008 Wenchuan earthquake in China. Computers & Geosciences 46: 317–329. doi:10.1016/j.cageo.2012.01.002.View ArticleGoogle Scholar
  36. Xu C, Xu X, Dai F, Wu Z, He H, Wu X, Xu S, Shi F. 2013a. Application of an incomplete landslide inventory, logistic regression model and its validation for landslide susceptibility mapping related to the May 12, 2008 Wenchuan earthquake of China. Natural Hazards 68(2): 883-900. doi:10.1007/s11069-013-0661-7.Google Scholar
  37. Xu C, Xu X, Yao Q, Wang Y. 2013b. GIS-based bivariate statistical modelling for earthquake-triggered landslides susceptibility mapping related to the 2008 Wenchuan earthquake, China. Quarterly Journal of Engineering Geology and Hydrogeology 46(2): 221-236. doi:10.1144/qjegh2012-006.Google Scholar
  38. Xu, C., X. Xu, X. Yao, and F. Dai. 2014. Three (nearly) complete inventories of landslides triggered by the May 12, 2008 Wenchuan Mw 7.9 earthquake of China and their spatial distribution statistical analysis. Landslides 11(3): 441–461. doi:10.1007/s10346-013-0404-6.View ArticleGoogle Scholar
  39. Xu, C., X. Xu, and J.B.H. Shyu. 2015. Database and spatial distribution of landslides triggered by the Lushan, China Mw 6.6 earthquake of 20 April 2013. Geomorphology 248: 77–92. doi:10.1016/j.geomorph.2015.07.002.View ArticleGoogle Scholar
  40. Xu C, Xu X, Tian Y, Shen L, Yao Q, Huang X, Ma J, Chen X, Ma S. 2016a. Two comparable earthquakes produced greatly different coseismic landslides: The 2015 Gorkha, Nepal and 2008 Wenchuan, China events. Journal of Earth Science 27(6): 1008-1015. doi:10.1007/s12583-016-0684-6.Google Scholar
  41. Xu C, Xu X, Shen L, Yao Q, Tan X, Kang W, Ma S, Wu X, Cai J, Gao M, Li K. 2016b. Optimized volume models of earthquake-triggered landslides. Scientific Reports 6: 29797. doi:10.1038/srep29797.Google Scholar
  42. Yalcin, A. 2008. GIS-based landslide susceptibility mapping using analytical hierarchy process and bivariate statistics in Ardesen (Turkey): Comparisons of results and confirmations. Catena 72(1): 1–12. doi:10.1016/j.catena.2007.01.003.View ArticleGoogle Scholar

Copyright

© The Author(s). 2017