Journal of Earth Science, Vol. 27, No. 6, p.

1008–1015, December 2016 ISSN 1674-487X

Printed in China
DOI: 10.1007/s12583-016-0684-6

Two Comparable Earthquakes Produced Greatly Different

Coseismic Landslides: The 2015 Gorkha, Nepal and
2008 Wenchuan, China Events
Chong Xu*1, Xiwei Xu*1, Yingying Tian1, Lingling Shen2, Qi Yao3, Xueqiang Huang4,
Junxue Ma4, Xi Chen4, Siyuan Ma1
1. Key Laboratory of Active Tectonics and Volcano, Institute of Geology, China Earthquake Administration, Beijing 100029, China
2. Beijing Meteorological Information Center, Beijing Meteorological Service, Beijing 100089, China
3. China Earthquake Networks Center, Beijing 100045, China
4. School of Engineering and Technology, China University of Geosciences, Beijing 100083, China

ABSTRACT: The 2015 Gorkha Earthquake in Nepal and the 2008 Wenchuan Earthquake in China
occurred at the south and southeast margins of the Tibetan Plateau, respectively. Both earthquakes had
similar magnitudes of Mw 7.8 and 7.9, caused catastrophic loss of life and damage to property, and
generated tens of thousands of landslides. Comparisons of pre- and post-quake satellite images sup-
ported by field investigations show that the Gorkha Earthquake triggered at least 2 064 large landslides
(defined as covering an area ≥10 000 m2) over a ~35 600 km2 region with a volume of (444–584)×106
(average 509×106) m3 and total area of 44.78×106 m2. In contrast, the Wenchuan Earthquake triggered
25 580 large landslides over a region of ~44 000 km2 with a volume of (7 128–9 479)×106 (average
8 219×106) m3 and a total area of about 670.65×106 m2. Several controlling factors including topograph-
ic relief, slope steepness, and regional peak ground acceleration (PGA) were investigated to try to ex-
plain the great differences between the number, volume and area of the coseismic landslides associated
with the two similar earthquakes. We found that the differences primarily arose from an unexpected
factor, the dip angle of the seismogenic fault. This discovery should aid understanding the failure me-
chanisms of quake-triggered landslides, and suggests that more factors should be taken into considera-
tion in estimating coseismic landslide volumes from earthquake magnitudes.
KEY WORDS: Gorkha Earthquake, Wenchuan Earthquake, landslide, dip angle, seismogenic fault.

0 INTRODUCTION geomorphic evolution of the affected area (Xu et al., 2014b;

The April 25, 2015 Gorkha, Nepal Mw 7.8 and May 12, Parker et al., 2011).
2008 Wenchuan, China Mw 7.9 earthquakes have received Although the Gorkha Earthquake occurred more recently,
much attention because both were humanitarian disasters that several studies (Kargel et al., 2016; British Geological Survey
generated large numbers of coseismic landslides and serious and Durham University, 2015; Chiaro et al., 2015; Collins and
damage (Yin and Liu, 2016). The ~200 000 landslides triggered Jibson, 2015; Hashash et al., 2015; Moss et al., 2015) have
by the Wenchuan Earthquake significantly changed the land- described its coseismic landslides using field reconnaissance
scape of the proximal area; their inventory (Xu et al., 2014a) is and satellite images, provided descriptions and analyses of
the largest ever recorded for landslides caused by an individual several large individual landslides, and analyzed their control-
earthquake. Relevant papers on the Wenchuan-event landslides ling factors. Two recently released inventories of coseismic
describe their database and spatial distribution patterns (Xu et landslides (Kargel et al., 2016; British Geological Survey and
al., 2014a; Gorum et al., 2011), discuss the characteristics and Durham University, 2015) record comparable numbers of
mechanisms of large individual landslides (Yin et al., 2015, landslides triggered by the Gorkha Earthquake but neither is a
2009), provide assessments of landslide susceptibility and ha- polygon-based inventory, so omissions of coseismic landslides
zards (Xu C et al., 2013), and describe the evolution of the are possible. One of the most interesting results recorded about
the Gorkha Earthquake is the unanticipated low number and
*Corresponding author: xuchong@ies.ac.cn; small scales of coseismic landslides. Kargel et al. (2016) attri-
xiweixu@vip.sina.com buted this to a lack of surface ruptures, the 10 to 15 km depth
© The Authors 2016. This article is published with open access of concentrated deformation, and the regional lithology but
at Springerlink.com these factors do not seem adequate to explain the actual situa-
tion. Therefore we undertook a comparative analysis of
Manuscript received May 10, 2016. landslides associated with the Wenchuan and Gorkha earth-
Manuscript accepted September 30, 2016. quakes to search for other relevant factors.

Xu, C., Xu, X. W., Tian, Y. Y., et al., 2016. 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. http://en.earth-science.net
Two Comparable Earthquakes Produced Greatly Different Coseismic Landslides 1009

We compiled a polygon-based inventory of large quake of Wenchuan, China occurred respectively at the south-
landslides triggered by the Gorkha event, and compared the ern and southeastern boundaries of the Tibetan Plateau (Fig. 1a).
result to the Wenchuan inventory. We have analyzed and com- The collision between the Indian and Asian continents has
pared correlations between the large coseismic landslides (area created the Himalaya thrust belt (Najman et al., 2010). The
≥10 000 m2) for the two earthquakes with several other relevant present architecture of the belt is dominated by three approx-
factors, such as topographic relief, slope steepness, and PGA. imately parallel master thrusts with low dip angles (~10°): the
The results show that the Gorkha earthquake-triggered main frontal thrust (MFT), the main boundary thrust (MBT),
landslides were much fewer in the number, collectively covered and the main central thrust (MCT) (Fig. 1b), listed from south
a much smaller area, and had a much smaller total volume than to north in order of increasing age of thrust initiation. The three
those associated with the Wenchuan earthquake-triggered thrusts divide this region into four tectonic zones: the autoch-
landslides. Our comparisons show that slope angle and PGA thonous Indian Plate, the Sub-Himalayan zone (SHZ), the
were not the primary reasons for the abnormally low occur- Lesser Himalayan sequence (LHS), and the Greater Himalayan
rence of the Gorkha earthquake-triggered landslides. Instead, sequence (GHS) (Mukherjee, 2015; Bhakuni et al., 2013;
we suggest that the disparate dip angles of the seismogenic Searle et al., 2008; Upreti, 1999; Wesnousky et al., 1999; Le
faults and the resultant differences in the internal rock deforma- Fort, 1975). All the thrusts are believed to converge at depth
tion were the primary reason for the difference. Understanding into the Main Himalayan thrust (MHT). Results from various
this factor should improve understanding of coseismic landslide methods including field reconnaissance along the main faults
generation mechanisms and help mitigate their associated ha- and interferometric synthetic aperture radar (InSAR) imaging
zards. of the earthquake-affected area, confirm that the Gorkha
Earthquake (28.23°N, 84.731°E) did not produce surface rup-
1 DATA AND METHODS tures along faults (Angster et al., 2015; Galetzka et al., 2015;
1.1 Geologic Settings Kobayashi et al., 2015; Lindsey et al., 2015; Parameswaran et
The 2015 earthquakes of Gorkha, Nepal and 2008 earth- al., 2015).

Figure 1. Locations of earthquake epicenters (a) and geologic settings of the Gorkha (b) and Wenchuan (c) earthquakes. Red polygons in (b) and (c) represent
large coseismic landslides with individual areas ≥10 000 m2. MCT. main central thrust; MBT. main boundary thrust; MFT. main frontal thrust; MWF. Maowen
fault; YBF. Yingxiu-Beichuan fault; GJF. Guanxian-Jiangyou fault. The four black rectangles N-1, N-2, W-1, and W-2 were target areas for analysis of local
topographic relief and slope steepness.
1010 Chong Xu, Xiwei Xu, Yingying Tian, Lingling Shen, Qi Yao, Xueqiang Huang, Junxue Ma, Xi Chen and Siyuan Ma

The Wenchuan Earthquake occurred in the Longmenshan produce high-quality inventory maps, and are essential for the
belt that has resulted from extrusion of crustal material from recognition of small landslides. The accuracy of landslide
the high Tibetan Plateau in the west against strong crustal ma- boundary delineation and the ability to distinguish individual
terial of the Sichuan Basin in the east. The Longmenshan belt is landslides in areas of high landslide density are improved as
dominated by three main high-dip angle (~40°) fault systems, image quality improves. The pre- and post-quake satellite im-
from northwest to southeast; the Maowen fault (MWF), the ages used to map the Gorkha landslides were mainly taken from
Yingxiu-Beichuan fault (YBF), and the Guanxian-Jiangyou the Google Earth platform and Digital Globe Company, which
fault (GJF) (Fig. 1c). The Wenchuan Earthquake generated two mostly have resolutions of 0.5 to 2 m. For comparison, the im-
coseismic surface ruptures along the YBF and GJF with lengths ages used to map the landslides triggered by the Wenchuan
of 240 and 72 km, respectively (Xu et al., 2009). Earthquake mostly have somewhat lower resolutions of 2.5 to 5
m (Xu et al., 2014a). In order to avoid inventory map errors
1.2 Preparing Coseismic Landslide Inventories arising from the resolution difference for the two areas, we only
Visual interpretation of satellite images or aerial photo- include landslides with areas exceeding 10 000 m2, which are
graphs is the most useful popular method for compiling a cose- large enough to be clearly seen on either set of images.
ismic landslide inventory, especially for large earthquakes in
mountainous areas (Xu, 2015). Landslides triggered by such 2 RESULTS AND ANALYSIS
earthquakes often occur in remote areas blocking roads and 2.1 Coseismic Landslides Created by the Two Earthquakes
impeding the timely arrival of ground rescue crews. In addition, We have constructed an inventory of large landslides trig-
the large numbers and wide distribution of landslides triggered gered by the Gorkha Earthquake on the basis of visual inter-
by large earthquakes make detailed, comprehensive field inves- pretation of pre- and post-quake satellite images. Several days
tigations impossible. Even so, field renaissance is a very useful of field reconnaissance in vehicles and on foot along expedition
auxiliary for visual interpretation of remote sensing images and routes (see Fig. 2a) increase the credibility of our inventory.
improving the quality of inventories of landslides. Notable Our results show that this earthquake triggered at least 2 064
point- or polygon-based inventories of earthquake-triggered landslides exceeding an area of 10 000 m2. These landslides
landslides include those for the January 17, 1994 Northridge cover a total area of 44.78×106 m2. We selected the formula
Earthquake, USA (Harp and Jibson, 1995), September 21, 1999 V=1.314 7×A1.208 5 (±0.013 1) of Xu et al. (2016) to estimate the
Chi-Chi, Taiwan China Earthquake (Wang et al., 2003), May 12, total volume from a selection of “volume-area” power law rela-
2008 Wenchuan Earthquake, China (Xu et al., 2014a; Gorum et tionships for landslides, (e.g., Xu et al., 2016; Parker et al., 2011;
al., 2011), and the April 20, 2013 Lushan, China Earthquakes Larsen et al., 2010) because this scaling relation was derived
(Xu et al., 2015; Zhang et al., 2015). A landslide is represented from an original data- based nonlinear least square analysis and
by a point in point-based inventories and is delineated as a po- sampled 1 415 landslides triggered by the Wenchuan Earth-
lygon in polygon-based inventories. Point-based inventories can quake and is probably the most suitable for landslides triggered
only describe locations of landslides, whereas polygon-based by the two earthquakes. The relationship was applied to each
inventories can depict scales of landslides including landslide slide mass, yielding a total estimated volume of
area and derived volume. High-resolution images are needed to (444–584)×106 (average 509×106) m3.

Figure 2. Distribution maps of landslides triggered by the Gorkha, Nepal (a) and Wenchuan, China (b) earthquakes.
Two Comparable Earthquakes Produced Greatly Different Coseismic Landslides 1011

In contrast, 25 580 large landslides (≥10 000 m2) were which is about 1.13 times that derived from Keefer’s equation
triggered by the 2008 Wenchuan Earthquake, according to the (Fig. 4). In other words, Keefer’s formula suggests that the total
detailed inventory by Xu et al., 2014a. These landslides collec- landslide volumes for the two events should be similar, when
tively cover 670.65×106 m2 and have a total volume (7 128– they differ by more than a factor of ten! However, neither of
9 479)×106 (average 8 219×106) m3. There is a startling contrast these two volumes appears to depart greatly from the trend
between large landslides caused by the Gorkha Earthquake derived from Keefer’s function (Fig. 4) due to the logarithmic
which are only 8.1% in number, 6.7% in area, and 6.2% in scale.
volume of those generated by the Wenchuan Earthquake. Nev- Many limitations apply to power law relationships be-
ertheless, the sizes of regions of the regions where there was tween volume and area, particularly when they are extended to
landsliding in these two earthquakes were similar, about 35 664 individual landslides. Typically, significant volume differences
and 44 031 km2, respectively. can occur for landslides with comparable areas if they have
The relation between landslide area and accumulated different depths. Actually, many landslides triggered by the
number is usefully depicted on a logarithmic graph. Fig. 3 Wenchuan Earthquake are bedrock landslides with large thick-
shows two nearly parallel curves that indicate similar power nesses, whereas landslides triggered by the Gorkha Earthquake
law relationships for the landslides of the two earthquakes, are almost all shallow landslides in regolith. For example, the
except for a few enormous landslides. This similarity spanning Langtang Landslide (28.215°N, 85.503°E), the largest triggered
all scales suggests that our inventories of large landslides are by the Gorkha Earthquake, covers 0.602×106 m2, yet the area of
quite complete and objective. seven Wenchuan-event landslides is greater than this, for ex-
Keefer (1994) obtained an empirical power law relation- amples the Daguangbao Landslide (31.643°N, 104.119°E),
ship between earthquake magnitude (M) and total volume (V) Wenjiagou (31.553°N, 104.148°E) and Donghekou Landslide
of seismic landslides, lg V=1.45M–2.50, by regression. Using (32.409°N, 105.109°E). The volume of the Langtang landslide
this relationship, the volumes of the Mw 7.8 Gorkha and Mw is estimated as (10.7–15.1)×106 (average 12.7×106) m3 from
7.9 Wenchuan earthquakes would be about 646×106 m3 and the relationship V=1.314 7×A1.208 5 (±0.013 1) (Xu et al., 2016), but
902×106 m3, respectively. For comparison, the total volume of its actual volume is only about 1.5×106 m3, given that its aver-
the landslides triggered by the Wenchuan Earthquake would be age depth is 2 to 3 m (Kargel et al., 2016; Collins and Jibson,
about 11 774×106 m3 (a value 13.05 times that suggested by 2015). Another useful comparison can be made with the Tang-
Keefer’s equations) using the inventory of 196 007 landslides jiashan Landslide (31.843°N, 104.432°E) triggered by the
triggered by the Wenchuan Earthquake (Xu et al., 2014a) and Wenchuan Earthquake, which covers an area 0.719×106 m2 that
the “volume-area” power law relationship V=1.314 7×A1.208 5 of is comparable to the Langtang slide. The volume of
Xu et al. (2016). Unfortunately, currently no inventory is (13.2–18.8)×106 (average 15.7×106) m3 estimated using the
available that spans all scales of coseismic landslides triggered power law (Xu et al., 2016) is smaller than its actual volume,
by the Gorkha Earthquake. Therefore, we estimated the total about 20×106 m3. Other Wenchuan-event examples are the
volume of Gorkha-event landslides by simple scaling of the Wangjiayan (31.826°N, 104.449°E) and Xinzhong (31.83°N,
Wenchuan-event results, using the respective volumes of large 104.459°E) landslides in Beichuan County that respectively
(≥10 000 m2) landslides, supported by the nearly parallel trends cover 0.129×106 and 0.121×106 m2. They would have power-
in Fig. 3. This indicated that the volume of all landslides trig- law volumes of about 2×106 and 1.8×106 m3 that are close to
gered by the Gorkha Earthquake can be estimated as their actual volumes of about 2×106 m3, calculated by their
(11 774×106 m3)×(509×106 m3/8 219×106 m3), i.e., 729×106 m3, average depths of about 15 m. These actual volumes of the
Wangjiayan and Xinzhong landslides are comparable with the
largest Langtang landslide triggered by the Gorkha Earthquake.
About 300 Wenchuan-event landslides covered larger areas

1×10 11

10 Wenchuan
1×10 13.05 times
Gorkha
Landslide volume (m 3)

9
1×10 1.13 times

1×10 8

7
1×10
1.45× M -2.50
V=10
6
1×10

1×10 5

1×10 4
5 6 7 8 9
Mw

Figure 3. Relationships between coseismic landslides area (≥10 000 m2) Figure 4. Empirical power law relationship between earthquake magnitude
and accumulated number of the Gorkha and Wenchuan earthquakes. and total volume of triggered landslides.
1012 Chong Xu, Xiwei Xu, Yingying Tian, Lingling Shen, Qi Yao, Xueqiang Huang, Junxue Ma, Xi Chen and Siyuan Ma

than the Wangjiayan and Xinzhong landslides, so several hun- quakes are plotted against distance from the epicenter in Fig. 5
dred Wenchuan-event landslides must be larger than the Lang- and landslide density for each slope angle bin in Fig. 6a. Areas
tang landslide. Therefore, the total volume ratio between with slope angles of 40°–50° register the largest number of
landslides triggered by the two earthquakes should be greater landslides, and the landslide density increases with increasing
than the aforementioned value by about 13.05/1.13=11.55 times, slope angle, especially for the Wenchuan event. The primary
normalized by earthquake magnitude. However, the uncertainty difference between the curves for the two earthquakes is that
caused by various landslide depths is difficult to quantify be- the landslide numbers and densities related to the Wenchuan
cause the actual depths and volumes remain unknown. Never- Earthquake are much higher than the values of the Gorkha
theless, it is certain that the Gorkha-event landslides are much Earthquake in every class of slope angle. Slope steepness alone
fewer and smaller than the Wenchuan-event landslides. cannot explain the low number of Gorkha landslides.
Similarly, we compared the correlations between PGA
2.2 Topography and PGA Related to the Two Earthquakes (US Geological Survey, 2015, 2008) and landslides for the two
Coseismic landslides are generally considered to be pri- earthquakes, dividing the data into 0.04 interval bins for PGA.
marily controlled by topographic, geologic, and seismic factors. The results show that PGA had a greater influence on Wen-
To evaluate these factors, we constructed profiles along four 10 chuan landslides than Gorkha landslides. Landslide numbers
km×100 km rectangular areas G-1, G-2, W-1, and W-2, (Fig. 1) and density related to the Wenchuan Earthquake are much
to show the topographic relief and slope steepness. The average higher than those of the Gorkha Earthquake at the same PGA
elevation and slope angle for each 1 km×10 km segment were values in every PGA class (Fig. 6b). We conclude that the low
determined along these four rectangles. The results show that capability of triggering landslides for the Gorkha Earthquake is
the landslide affected areas for the two earthquakes have simi- not primarily due to strong ground shaking.
lar topographic relief and slope steepness (Fig. 5). The topo- Both the areas affected by the two earthquakes are charac-
graphic relief of the two profiles G-1, G-2 in the Gorkha area is terized by steep terrain with widespread high-angle slopes. The
mostly a little greater than that of the Wenchuan profiles, but
12 000 (a) 6
topographic profile W-1 shows a peak relief between distances
Number-G

Landslide density (km -1)

50 and 70 km (Fig. 5a). In addition, the slope steepness of pro- 10 000
Number-W 5
Landslide number

file W-1 between distances 60 and 80 km is greater than the 8 000 Density-G
4
Density-W
values in other profiles. It is noted that this high, steep area
6 000 3
roughly corresponds to the hanging wall of the southwest sec-
4 000 2
tion of the seismogenic fault (YBF) of the Wenchuan Earth-
quake, and the greatest landslide density occurs there. 2 000 1
To test whether topography or PGA can explain the small 0 0
number of Gorkha-event large landslides, we evaluated the 0-10 10-20 20-30 30-40 40-50 50-60 >60
Slope angle ( ° )
correlations between the numbers of coseismic landslides and
these factors. To test the influence of topography, data were 10 000 (b) 4

Landslide density (km -1)

divided into bins for each 10° interval of slope steepness and 8 000
Landslide number

3
the numbers of landslides and the landslide. Correlations be-
tween topographic relief and slope steepness for the two earth- 6 000
2
4 000
6 000 (a)
G-1 1
5 000 2 000
Elevation (m a.s.l.)

G-2
W-1
4 000 W-2 0 0
0.28

0.48

0.68

0.8
0.2

0.32

0.52

0.72
0.36

0.56
0.6

0.76
0.24

0.4
0.44

0.64

0.84

3 000
PGA (g)
2 000
15 000 (c) 25
Landslide density (km -1)

1 000
Landslide number

20
0 10 20 30 40 50 60 70 80 90 100 10 000
Distance (km) 15
50 (b)
10
40 5 000
Slope angle ( ° )

5
30 0 0
4

4
.6

.6

.6

.6

.6

.6

20
-0.

-0.

-0.
4-0

>0

4-0

>0

4-0

>0
0.2

0.2

0.2

0&

&

0&
0.4

0.4

0.4

10
-40
0&

&

0&

0-2

>4
0&

&

0&
-40

20
0-2

>4

-40
0-2

>4
20

20

0 10 20 30 40 50 60 70 80 90 100
Distance (km) Slope angle ( ° ) & PGA (g)

Figure 5. Curves of topographic relief (a) and slope steepness (b) along the Figure 6. Relationships between coseismic landslides and slope angle (a),
four rectangles perpendicular to the seismogenic faults in Fig. 1. PGA (b), and the combination of them (c).
Two Comparable Earthquakes Produced Greatly Different Coseismic Landslides 1013

distribution of PGA data related to the earthquakes shows ap- small, concentrating the surface energy and resulting in many
proximate elliptical rings around the meizoseismal areas. Both large, intense landslides. In contrast, the Gorkha Earthquake
the two cases show a high inconsistency between slope angle occurred along a low-angle fault, so the overlying surface area
and PGA data and therefore univariate relationships between was wide, the surface energy less concentrated, and fewer
landslides, slope angle and PGA appear unlikely. In order to smaller landslides were triggered. In short, we suggest that that
validate this inference, we correlated the coseismic landslides these two effects determine the capability of an earthquake to
with nine combinations of slope angle and PGA (Fig. 6c). This trigger landslides because the dip angle of the seismogenic fault
indicates a similar result to the correlations between landslide controls the manner of earthquake energy release, and thus
densities, slope angle and PGA, i.e., the landslide numbers and influences the deformation of the hanging wall bedrock.
density related to the Wenchuan Earthquake are much higher
than those of the Gorkha Earthquake in any combinations of σ1

slope angle and PGA. Both slope steepness and PGA cannot
explain the low number of Gorkha earthquake-triggered
landslides.
σ1
2.3 Behavior of Seismogenic Faults
For earthquakes generated by reverse (or thrust) faults, the tal
iz on
distribution of coseismic landslides could depend on whether or

ane
H
surface ruptures are generated. Xu (2014) compared three in-

lt pl

L
L×
dices of coseismic landslides, including landslide distribution

Fau
tan
α
area, landslide coverage area, and landslide number, for eight
reverse-fault earthquakes, of which five had surface ruptures.
The results suggest that surface-rupture earthquakes trigger α
fewer but larger landslides in a smaller area than buried-rupture
earthquakes of comparable magnitudes. In most cases, buried-
rupture earthquakes trigger relatively more landslides than
those caused by surface-rupture events, as exemplified by the
1994 Northridge, USA earthquake (Harp and Jibson, 1995; Figure 7. 3-D view of the stress components along and perpendicular to the

Hauksson et al., 1995), 2010 Haiti (Xu et al., 2014c; Hayes et rupture surface, and slip components on the fault surface.

al., 2010), and the 2013 Lushan, China events (Xu X et al.,
2015, 2013). The relation between the surface-rupture Wen- 3 CONCLUSIONS
chuan Earthquake and the buried-rupture Gorkha Earthquake is The 2008 Wenchuan and 2015 Gorkha earthquakes have
the opposite. The lack of surface ruptures for the Gorkha event comparable magnitudes, both occurred on the margins of the
cannot account for its relatively small number of associated Tibetan Plateau, and both are characterized by high topography
landslides. and steep slopes. These similarities might suggest that the
A significant difference between the Gorkha and Wen- number of landslides triggered by these two events would be
chuan earthquakes is that the dip of the seismogenic fault of the comparable but however, the Gorkha Earthquake triggered only
Wenchuan Earthquake is much steeper than that of the Gorkha 2 064 large landslides (≥10 000 m2), with a total area of
Earthquake. The Gorkha Earthquake occurred at the boundary 44.78×106 m2 and volume of (444–584)×106 (average 509×106)
between the Indian and Eurasian plates, along a shallow fault m3, whereas the Wenchuan Earthquake generated 25 580 large
plane (dip angle ~10°), whereas the Wenchuan Earthquake landslides that cover 670.65×106 m2 and have a total volume of
occurred in the plate interior, along a steep (~40°) seismogenic (7 128–9 479)×106 (average 8 219×106) m3. There is only one
thrust fault. We consider that this difference of dip angle can similarity, namely that the regions of landslide generation asso-
explain the difference in the number and scale of landsides of ciated with these two earthquakes have comparable areas. Our
the Wenchuan and Gorkha events. The relative magnitudes of research indicates that local topography and PGA cannot ex-
the shear and normal components of maximum stress depend plain the large difference of landsliding between the events, and
on the dip angle of the fault plane. For a reverse fault with dip the much different dip angles of their seismogenic faults is the
angle α and slip displacement L, the maximum principal stress primary reason. We suggest that high dip reverse faults cause
(σ1) is horizontal and perpendicular to the strike of the rupture more intense deformation of interior blocks and create larger
plane (Fig. 7). Therefore, the normal stress is proportional to normal stresses perpendicular to rupture surfaces, inducing
L×tanα, and this component becomes greater as the dip angle α joints and cracks in brittle rocks conducive to landslide genera-
increases. The deformation of the hanging wall block is closely tion. In contrast, fewer smaller landslides will be associated
related to this component, and the occurrence of coseismic with earthquakes on reverse faults with low dip angles, for
landslides is therefore triggered by strong ground shaking. In equivalent earthquake magnitude, topography, and other condi-
addition, the surface region directly above the rupture plane tions. This study should be a valuable supplement to traditional
receives most of the earthquake energy. Because the Wenchuan views that coseismic landslides are mainly controlled by topo-
Earthquake occurred along a high-angle reverse fault, the area graphy, lithology, PGA , etc.. We suggest that recognition of the
of the surface directly overlying the fault surface was relatively important effect of fault dip angle will contribute to a better
1014 Chong Xu, Xiwei Xu, Yingying Tian, Lingling Shen, Qi Yao, Xueqiang Huang, Junxue Ma, Xi Chen and Siyuan Ma

understanding of mechanisms giving rise to coseismic GEER Association Report No. GEER-040: 250
landslides, and facilitate more reliable assessments of landslide Hauksson, E., Jones, L. M., Hutton, K., 1995. The 1994 North-
susceptibility and associated hazards. ridge Earthquake Sequence in California: Seismological
and Tectonic Aspects. Journal of Geophysical Research:
ACKNOWLEDGMENTS Solid Earth (1978–2012), 100(B7): 12335–12355.
This research was supported by the National Natural doi:10.1029/95JB00865
Science Foundation of China (Nos. 41472202). Comments and Hayes, G. P., Briggs, R. W., Sladen, A., et al., 2010. Complex
revisions by Robert E. Criss of Washington University greatly Rupture during the 12 January 2010 Haiti Earthquake.
improved the manuscript. The final publication is available at Nature Geoscience, 3(11): 800–805. doi:10.1038/ngeo977
Springer via http://dx.doi.org/10.1007/s12583-016-0684-6. Kargel, J. S., Leonard, G. J., Shugar, D. H., et al., 2016. Geo-
morphic and Geologic Controls of Geohazards Induced by
Open Access This article is distributed under the terms of the Nepal’s 2015 Gorkha Earthquake. Science, 351(6269):
Creative Commons Attribution 4.0 International License 140. doi:10.1126/science.aac8353
(http://creativecommons.org/licenses/by/4.0/), which permits Keefer, D. K., 1994. The Importance of Earthquake-Induced
unrestricted use, distribution, and reproduction in any medium, Landslides to Long-Term Slope Erosion and Slope-Failure
provided you give appropriate credit to the original authors and Hazards in Seismically Active Regions. Geomorphology,
the source, provide a link to the Creative Commons license, 10(1–4): 265–284. doi:10.1016/0169-555X(94)90021-3
and indicate if changes were made. Kobayashi, T., Morishita, Y., Yarai, H., 2015. Detailed Crustal
Deformation and Fault Rupture of the 2015 Gorkha
REFERENCES CITED Earthquake, Nepal, Revealed from ScanSAR-Based Inter-
Angster, S., Fielding, E. J., Wesnousky, S., et al., 2015. Field ferograms of ALOS-2. Earth, Planets and Space, 67(1):
Reconnaissance after the 25 April 2015 M 7.8 Gorkha 1–13. doi:10.1186/s40623-015-0359-z
Earthquake. Seismological Research Letters, 86(6): Larsen, I. J., Montgomery, D. R., Korup, O., 2010. Landslide
1506–1513. doi:10.1785/0220150135 Erosion Controlled by Hillslope Material. Nature Geos-
Bhakuni, S., Luirei, K., Kothyari, G. C., 2013. Neotectonic cience, 3(4): 247–251. doi:10.1038/ngeo776
Fault in the Middle Part of Lesser Himalaya, Arunachal Le Fort, P., 1975. Himalayas: the Collided Range. Present
Pradesh: A Study Based on Structural and Morphotectonic Knowledge of the Continental Arc. American Journal of
Analyses. Himalayan Geology, 34(1): 57–64 Science, 275-A: 1–44
British Geological Survey, Durham University, 2015. Nepal Lindsey, E. O., Natsuaki, R., Xu, X., et al., 2015. Line-of-Sight
Earthquake Response 2015. [2016-10-01]. Displacement from ALOS-2 Interferometry: Mw 7.8
http://www.bgs.ac.uk/research/earthHazards/epom/nepalE Gorkha Earthquake and Mw 7.3 Aftershock. Geophysical
arthquakeResponse.html Research Letters, 42(16): 6655–6661.
Chiaro, G., Kiyota, T., Pokhrel, R. M., et al., 2015. Reconnais- doi:10.1002/2015GL065385
sance Report on Geotechnical and Structural Damage Moss, R. E. S., Thompson, E. M., Kieffer, D. S., et al., 2015.
Caused by the 2015 Gorkha Earthquake, Nepal. Soils and Geotechnical Effects of the 2015 Magnitude 7.8 Gorkha,
Foundations, 55(5): 1030–1043. Nepal, Earthquake and Aftershocks. Seismological Re-
doi:10.1016/j.sandf.2015.09.006 search Letters, 86(6): 1514–1523.
Collins, B. D., Jibson, R. W., 2015. Assessment of Existing and doi:10.1785/0220150158
Potential Landslide Hazards Resulting from the April 25, Mukherjee, S., 2015. A Review on Out-of-Sequence Deforma-
2015 Gorkha, Nepal Earthquake Sequence. US Geological tion in the Himalaya. Geological Society, London, Special
Survey. doi:10.3133/ofr20151142 Publications, 412: 412. doi:10.1144/SP412.13
Galetzka, J., Melgar, D., Genrich, J. F., et al., 2015. Slip Pulse Najman, Y., Appel, E., Boudagher-Fadel, M., et al., 2010. Tim-
and Resonance of the Kathmandu Basin during the 2015 ing of India-Asia Collision: Geological, Biostratigraphic,
Gorkha Earthquake, Nepal. Science, 349(6252): and Palaeomagnetic Constraints. Journal of Geophysical
1091–1095. doi:10.1126/science.aac6383 Research: Solid Earth (1978–2012), 115(B12).
Gorum, T., Fan, X. M., van Westen, C. J., et al., 2011. Distribu- doi:10.1029/2010JB007673
tion Pattern of Earthquake-Induced Landslides Triggered Parameswaran, R. M., Natarajan, T., Rajendran, K., et al., 2015.
by the 12 May 2008 Wenchuan Earthquake. Geomor- Seismotectonics of the April–May 2015 Nepal Earth-
phology, 133(3–4): 152–167. quakes: An Assessment Based on the Aftershock Patterns,
doi:10.1016/j.geomorph.2010.12.030 Surface Effects and Deformational Characteristics. Jour-
Harp, E. L., Jibson, R. W., 1995. Inventory of Landslides Trig- nal of Asian Earth Sciences, 111: 161–174.
gered by the 1994 Northridge, California Earthquake. doi:10.1016/j.jseaes.2015.07.030
[2016-10-01]. http://geo-nsdi.er.usgs.gov/metadata/open- Parker, R. N., Densmore, A. L., Rosser, N. J., et al., 2011. Mass
file/95-213/ Wasting Triggered by the 2008 Wenchuan Earthquake is
Hashash, Y. M. A., Tiwari, B., Moss, R. E. S., et al., 2015. Greater than Orogenic Growth. Nature Geoscience, 4(7):
Geotechnical Field Reconnaissance: Gorkha (Nepal) 449–452. doi:10.1038/ngeo1154
Earthquake of April 25, 2015 and Related Shaking Se- Searle, M. P., Law, R. D., Godin, L., et al., 2008. Defining the
quence. Geotechnical Extreme Event Reconnaisance Himalayan Main Central Thrust in Nepal. Journal of the
Two Comparable Earthquakes Produced Greatly Different Coseismic Landslides 1015

Geological Society, 165(2): 523–534. doi:10.1144/qjegh2012-006

doi:10.1144/0016-76492007-081 Xu, X., Wen, X., Han, Z., et al., 2013. Lushan Ms 7.0 Earth-
Sun, B., Yan, P., 2015. Damage Characteristics and Seismic quake: A Blind Reserve-Fault Event. Chinese Science
Capacity of Buildings During Nepal Ms 8.1 Earthquake. Bulletin, 58(28–29): 3437–3443.
Earthquake Engineering and Engineering Vibration, 14(3): doi:10.1007/s11434-013-5999-4
571–578. doi:10.1007/s11803-015-0046-x Xu, C., Xu, X., Yao, X., et al., 2014a. Three (Nearly) Complete
Upreti, B., 1999. An Overview of the Stratigraphy and Tecton- Inventories of Landslides Triggered by the May 12, 2008
ics of the Nepal Himalaya. Journal of Asian Earth Wenchuan Mw 7.9 Earthquake of China and Their Spatial
Sciences, 17(5): 577–606. Distribution Statistical Analysis. Landslides, 11(3):
doi:10.1016/S1367-9120(99)00047-4 441–461. doi:10.1007/s10346-013-0404-6
US Geological Survey, 2008. Shakemap US2008ryan. Xu, C., Xu, X., Gorum, T., et al., 2014b. Did the 2008 Wen-
[2016-10-01]. http://earthquake.usgs.gov/earthquakes/ chuan Earthquake Lead to a Net Volume Loss? Landslide
shakemap/ global/ shake/2008ryan/ Science for a Safer Geoenvironment, Proceedings of
US Geological Survey, 2015. Shakemap US20002926. World Landslide Forum 3, 2–6 June 2014, Beijing. 3:
[2016-10-01]. http://earthquake.usgs.gov/earthquakes/ 191–196. doi:10.1007/978-3-319-04996-0_30
shakemap/global/shake/20002926/ Xu, C., Shyu, J. B. H., Xu, X., 2014c. Landslides Triggered by
Wang, W. N., Wu, H. L., Nakamura, H., et al., 2003. Mass the 12 January 2010 Port-au-Prince, Haiti, Mw=7.0
Movements Caused by Recent Tectonic Activity: The Earthquake: Visual Interpretation, Inventory Compiling,
1999 Chi-Chi Earthquake in Central Taiwan. The Island and Spatial Distribution Statistical Analysis. Natural Ha-
Arc, 12(4): 325–334. zards and Earth System Sciences, 14(7): 1789–1818.
doi:10.1046/j.1440-1738.2003.00400.x doi:10.5194/nhess-14-1789-2014
Wesnousky, S. G., Kumar, S., Mohindra, R., et al., 1999. Uplift Xu, C., Xu, X., Shyu, J. B. H., 2015. Database and Spatial Dis-
and Convergence along the Himalayan Frontal Thrust of tribution of Landslides Triggered by the Lushan, China
India. Tectonics, 18(6): 967–976. Mw 6.6 Earthquake of 20 April 2013. Geomorphology,
doi:10.1029/1999TC900026 248: 77–92. doi:10.1016/j.geomorph.2015.07.002
Xu, C., 2014. Do Buried-Rupture Earthquakes Trigger Less Xu, C., Xu, X., Shen, L., et al., 2016. Optimized Volume Mod-
Landslides than Surface-Rupture Earthquakes for Reverse els of Earthquake-Triggered Landslides. Scientific Reports,
Faults? Geomorphology, 216: 53–57. 6: 29797. doi:10.1038/srep29797
doi:10.1016/j.geomorph.2014.03.029 Yin, Y., Wang, F., Sun, P., 2009. Landslide Hazards Triggered
Xu, C., 2015. Preparation of Earthquake-Triggered Landslide by the 2008 Wenchuan Earthquake, Sichuan, China.
Inventory Maps Using Remote Sensing and GIS Technol- Landslides, 6(2): 139–152.
ogies: Principles and Case Studies. Geoscience Frontiers, doi:10.1007/s10346-009-0148-5
6(6): 825–836. doi:10.1016/j.gsf.2014.03.004 Yin, Y., Li, B., Wang, W., 2015. Dynamic Analysis of the Stabi-
Xu, X., Wen, X., Yu, G., et al., 2009. Coseismic Reverse- and lized Wangjiayan Landslide in the Wenchuan Ms 8.0
Oblique-Slip Surface Faulting Generated by the 2008 Mw Earthquake and Aftershocks. Landslides, 12(3): 537–547.
7.9 Wenchuan Earthquake, China. Geology, 37(6): doi:10.1007/s10346-014-0497-6
515–518. doi:10.1130/G25462A.1 Zhang, Z. W., Zhou, L. Q., Cheng, W. Z., et al., 2015. Focal
Xu, C., Xu, X., Yao, Q., et al., 2013. GIS-Based Bivariate Sta- Mechanism Solutions of Lushan Mw 6.6 Earthquake Se-
tistical Modelling for Earthquake-Triggered Landslides quence and Stress Field for Aftershock Zone. Earth
Susceptibility Mapping Related to the 2008 Wenchuan Science––Journal of China University of Geosciences,
Earthquake, China. Quarterly Journal of Engineering Ge- 40(10): 1781–1793
ology and Hydrogeology, 46(2): 221–236.