Content uploaded by Chen Guo
Author content
All content in this area was uploaded by Chen Guo on Oct 14, 2021
Content may be subject to copyright.
Journal of Earth Science, Vol. 32, No. 5, p. 1079–1091, October 2021 ISSN 1674-487X
Printed in China
https://doi.org/10.1007/s12583-021-1467-2
Guo, C., Xu, Q., Dong, X. J., et al., 2021. Geohazard Recognition and Inventory Mapping Using Airborne LiDAR Data in Complex
Mountainous Areas. Journal of Earth Science, 32(5): 1079–1091. https://doi.org/10.1007/s12583-021-1467-2. http://en.earth-science.net
Geohazard Recognition and Inventory Mapping Using
Airborne LiDAR Data in Complex Mountainous Areas
Chen Guo , Qiang Xu *, Xiujun Dong, Weile Li, Kuanyao Zhao, Huiyan Lu, Yuanzhen Ju
State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology,
Chengdu 610059, China
Chen Guo: https://orcid.org/0000-0001-8028-0054; Qiang Xu: https://orcid.org/0000-0001-5894-4694
ABSTRACT: Geohazard recognition and inventory mapping are absolutely the keys to the establishment of
reliable susceptibility and hazard maps. However, it has been challenging to implement geohazards recogni-
tion and inventory mapping in mountainous areas with complex topography and vegetation cover. Progress
in the light detection and ranging (LiDAR) technology provides a new possibility for geohazard recognition
in such areas. Specifically, this study aims to evaluate the performances of the LiDAR technology in recog-
nizing geohazard in the mountainous areas of Southwest China through visually analyzing airborne LiDAR
DEM derivatives. Quasi-3D relief image maps are generated based on the sky-view factor (SVF), which
makes it feasible to interpret precisely the features of geohazard. A total of 146 geohazards are remotely
mapped in the entire 135 km
2
study area in Danba County, Southwest China, and classified as landslide, rock
fall, debris flow based on morphologic characteristics interpreted from SVF visualization maps. Field valida-
tion indicate the success rate of LiDAR-derived DEM in recognition and mapping geohazard with higher
precision and accuracy. These mapped geohazards lie along both sides of the river, and their spatial distribu-
tions are related highly to human engineering activities, such as road excavation and slope cutting. The min-
imum geohazard that can be recognized in the 0.5 m resolution DEM is about 900 m
2
. Meanwhile, the SVF
visualization method is demonstrated to be a great alternative to the classical hillshaded DEM method when
it comes to the determination of geomorphological properties of geohazard. Results of this study highlight the
importance of LiDAR data for creating complete and accurate geohazard inventories, which can then be used
for the production of reliable susceptibility and hazard maps and thus contribute to a better understanding
of the movement processes and reducing related losses.
KEY WORDS: geohazard, geohazard inventory, airborne LiDAR, sky view factor, remote sensing interpre-
tation, complex mountainous areas.
0 INTRODUCTION
Geohazard recognition and inventory mapping help to un-
derstand the correlation of distributions and patterns of geohaz-
ards at different scales with landforms and geological character-
istics (Peng et al., 2018; Trigila et al., 2010). Moreover, they also
help more clearly reveal deformation and failure of slopes as
well as the movement process and geomorphological evolution
characteristics of geohazards (Parker et al., 2011; Malamud et
al., 2004). Furthermore, geohazard inventory maps are the basis
for evaluating geohazard susceptibility, hazard and risk, and also
geohazard monitoring and early warning (Tang et al., 2019;
Guzzetti et al., 2012). Affected by the uplifting of the Qinghai-
Tibet Plateau, Southwest China is characterized by complex to-
pography, frequent tectonic movement, and dense vegetation,
which render itself highly prone to geohazards (Yin et al., 2009).
*Corresponding author: xq@cdut.edu.cn
© China University of Geosciences (Wuhan) and Springer-Verlag
GmbH Germany, Part of Springer Nature 2021
Manuscript received January 29, 2021.
Manuscript accepted April 6, 2021.
Over recent years, large catastrophic geohazards have succes-
sively occurred in mountainous areas of Southwest China, in-
cluding the 6·24 Xinmo Village landslide in the Mao County of
Sichuan (Fan et al., 2017), the 8·28 Nayong landslide in Guizhou
(Zhu et al., 2019; Zheng et al., 2018), and the 10·11 and 11·15
Baige landslides in Tibet (Zhang et al., 2020; Fan et al., 2019).
These landslides have caused huge casualties and severe finan-
cial loss, and they have aroused wide public concern. Subsequent
investigation shows that all these large catastrophic landslides
are not included in the identified hazard inventory, which means
that the conventional geohazard recognition and inventory map-
ping methods have failed to sufficiently meet the current de-
mands of hazard prevention and mitigation.
Before the emergence of remote sensing, geohazard recog-
nition is mainly dependent on manual field survey, which is seen
with not only low efficiency but also high difficulties in practice,
especially in areas isolated by high mountains and dangerous
terrains (Santangelo et al., 2010). With the continuous progress
in remote sensing, satellite remote sensing has now become the
most frequently used method for geohazard recognition, which
has benefited from the increasing quantity of remote sensing
satellites, refined ground resolutions, and accumulated remote
Chen Guo, Qiang Xu, Xiujun Dong, Weile Li, Kuanyao Zhao, Huiyan Lu and Yuanzhen Ju
1080
sensing data (Li et al., 2019; Gao and Maro, 2010; Nichol and
Wong, 2005). Besides, another two powerful methods have been
developed for geohazard recognition, namely the interferometric
synthetic aperture radar (InSAR) technology and the unmanned
aerial vehicle (UAV) photogrammetry. These two methods are
highly effective in recognizing and mapping geohazard induced
by rainfalls or earthquakes that have outstanding optical spectral
or textural features (Fan et al., 2018; Yamazaki et al., 2017; Fi-
orucci et al., 2011). However, geohazard recognition in moun-
tainous areas covered by dense vegetation is still tremendously
challenging (Tomás and Li, 2017).
Compared with optical imaging and other conventional re-
mote sensing methods, airborne LiDAR is favored by its certain
capacity to “penetrate” vegetation and thus obtain high-
resolution digital elevation models (DEMs) (Roering et al.,
2013; McKean and Roering, 2004). Experienced geologists can
recognize geohazard according to their various landform fea-
tures (Guzzetti et al., 2012). Therefore, airborne LiDAR is con-
sidered the most effective method for geohazard recognition
across dense-vegetation mountainous areas. This technology has
been widely applied to geohazard recognition and inventory
mapping, emergency investigation, and risk assessment by many
countries, including Italy, Austria, Japan, New Zealand, and so
on (Comert et al., 2019; Petschko et al., 2016; Chigira et al., 2004).
With the development of the LiDAR technology and continuous
reduction of its cost, the coverage and volume of the country-wide
airborne LiDAR data further expand and grow (Ardizzone et al.,
2007; Chen et al., 2006), which provides huge resources for geo-
hazard recognition in dense-vegetation areas and leads to new in-
sights into the evolution of various geomorphological landscapes
(Bell et al., 2012; Jaboyedoff et al., 2012, 2007).
Many countries have built or are building geohazard inven-
tories at different scales (national and/or regional). Nonetheless,
no geohazard inventories based on the airborne LiDAR have yet
been developed in mainland China, and there is little research
regarding applications of LiDAR data to geohazard survey and
recognition, except for attempts made by Chen et al. (2014) and
Li et al. (2015), who used object-oriented semi-automated land-
slide recognition in the Three-Gorge region (with an area of 21.6
km2 and a DEM resolution of 3 m). It should also be noted that
studies of LiDAR-based recognition of regional landslide around
the world are mainly focus on areas with low average altitudes
and small elevation differences, such as plains and hills. For ar-
eas with high mountains and steep gorges, like Southwest China
with an average altitude of about 3 500 m and a regional eleva-
tion difference above 2 000 m, relevant studies are rarely carried
out due to difficulties in data acquisition. Yet these areas are of-
ten highly prone to geohazard.
In this study, a geohazard inventory map was carried out
via sky view factor (SVF) terrain visualization, using high-
resolution DEM originating from the airborne LiDAR data in
Danba County, Southwest China, with the main objectives of:
(1) assessing the capability of the airborne LiDAR in mapping
geohazards in complex mountainous areas and (2) analyzing
morphometric characteristics and spatial distribution character-
istics and influencing factors of the geohazards in the study area.
This study provides references and guidance for the application
of airborne LiDAR technology in geohazard recognition and
mapping in mountainous areas with complex topography and
vegetation cover, and also provides data support for the geohaz-
ards prevention and risk assessment of regional geohazards.
1 STUDY AREA
The study area is located at the Danba County, the north-
eastern Ganzi Tibetan Autonomous Prefecture, Sichuan Prov-
ince, Southwest China, with a total area of about 135 km2. It is
about 400 km from Chengdu, the capital city of Sichuan Prov-
ince (Fig. 1a). The elevation difference in the study area is 2 720
m (Fig. 1c), with the minimum and maximum altitudes of about
1 840 and 4 560 m, respectively, which indicates a typical erosion-
denudation-dominated high mountain-gorge landform. Geologi-
cally, this area exposes silver-gray quartz-mica schist, the Fourth
Formation-complex (Smx4) of Maoxian Group, Silurian System
(Fig. 1b), and the overburden material consists of old landslide de-
bris (Q4del), rock fall deposits (Q4col+dl), and morainic deposits
(Q4fgl). In the study area flow the NE-SW Xiaojin River and the N-
S Dajinchuan River, which converge into the Dadu River after
flowing by the Danba County, and this area is characterized by
heavy erosion by the river. The precipitation in the study area is
found with uneven distribution throughout a year, and the monthly
precipitation peaks in June and September respectively (Fig. 1d).
The daily precipitation is large during the flood season, which very
likely to trigger geohazards (Li et al., 2008).
Physiographically, the study area is located in the eastern
part of Qinghai-Tibet Plateau, which has a prolonged geological
history, complex structure system, and fragile geological envi-
ronment, is characterized by neotectonic movements and seen
with frequent occurrences of geological hazards. Therefore, the
area has been graded as a high susceptibility zones of geohazards
in Sichuan Provence (Fig. 1a). The 6·26 and 7·11 debris flow
disasters in 2003 have caused severe casualties (50 missing and
1 dead, and 7 missing and 10 dead, respectively) and huge finan-
cial loss (Chen et al., 2005). In February 2005, the Jianshe Road
landslide at the county re-activated and imperiled the safety of
half of the county. This has aroused concerns of the state and the
Sichuan government, which actively dealt with the emergency
and spent tens of millions of RMB in emergency rescue and
comprehensive treatment (Huang, 2009; Fan et al., 2007). With
the continuously expanding human activity, the development of
geological hazards in the Danba County tends to grow, and mul-
tiple landslide deformations are re-activated, which brings about
major threats to the life and property security of local residents
(Dong et al., 2018). By March 20th 2019, a total of 844 geolog-
ical hazards have been found across the area of 5 649 km2 of
Danba County. Specifically, the area of 135 km2 covered by this
study alone includes 181 potential geological hazards, account-
ing for 21% of the total geological hazards of the county. Geo-
logical hazards have severely threatened the economic develop-
ment, social security, and local prosperity of Danba County.
2 DATA AND METHODS
In this work, the geohazard inventory map is obtained via
the visual interpretation of the DEM derivatives generated by the
airborne LiDAR data, with a spatial resolution of 0.5 m×0.5 m
and the processing steps followed in this study shown in Fig. 2.
The LiDAR data are acquired during the period from February
Geohazard Recognition and Inventory Mapping Using Airborne LiDAR Data in Complex Mountainous Areas
1081
13th 2019 to February 19th 2019. During the data acquisition, an
AS350 helicopter is equipped with a SKYEYE SE-J1200B air-
borne LiDAR system that applies to areas of high mountains and
hills, which is manufactured by the Mianyang Skyeye Laser
Technology Co. Ltd., China. The maximum field angle of the
laser generator is 50°, and the laser pulse frequency is 50–550
kHz. This LiDAR system integrates the laser scanner, the posi-
tion and orientation system (POS), the high-resolution digital
camera, and the embedded computer into one system. During the
operation of aerial photography, an 80-million-pixel optical
camera is loaded along the same axis of the laser scanner, which
enables simultaneous acquisition of laser-point clouds and aerial
photographs. Four georeferenced ground control points (GCPs)
are placed in the Danba County and the surrounding area to ensure
the data precision, and one GPS receiver is set up for simultaneous
observation, during the operation of aerial photography.
Figure 1. Geological map and precipitation data of the study area. (a) Geological hazard susceptibility map of Sichuan Province, modified from Geological Cloud
of China Geological Survey (http://geocloud.cgs.gov.cn/, last accessed on 2020-10-3); (b) tectonic and Geological map of the study area; (c) cross-section along
Line 1-1’ in Fig. 1b; (d) monthly precipitation data of the study area, data collected from the official website of China Meteorological Administration
(http://www.cma.gov.cn/, last accessed on 2020-11-3).
Figure 2. Overview of the processing steps followed in this study.
Chen Guo, Qiang Xu, Xiujun Dong, Weile Li, Kuanyao Zhao, Huiyan Lu and Yuanzhen Ju
1082
The first aim of the classification was to determine the topo-
graphical and non-topographical points (Fig. 3a). With the help
of the TerraScan module in the Terrasolid software, the macro
command was created and the automatic classification of point
clouds was realized. The point clouds obtained from the LiDAR
data were classified with respect to the standard parameter cate-
gories, including ground, low vegetation, medium vegetation,
high vegetation, building, noisy low points, and aerial points.
Arc GIS 10.3 was used to convert point cloud data into a digital
surface model (DSM) and a digital elevation model (DEM). In
order to help interpret geohazards, Pix4d Mapper 4.3.2 was used
to generate a digital orthophoto map (DOM) of the study area,
with a resolution of 0.2 m. Accuracy check result shown that the
elevation errors of the DSM and DEM are smaller than 0.6 m,
while the error of the DOM data are smaller than 0.5 m. The
system for automated geoscientific analyses (SAGA GIS) (Con-
rad et al., 2015) was also used to further investigate and produce
DEM derivatives, such as a hillshade map and a sky-view factor
(SVF) map (Fig. 3b).
As a derivative of the DEM visualization, hillshade maps
are used to assist geohazard identification. However, due to the
effects of the solar azimuth and altitude, the generation of hill-
shade maps are often greatly varied, which brings about difficul-
ties in geohazard recognition and may result in missed or false
recognition of geohazards. The SVF represents the ratio of the
visible sky in the entire hemisphere at a point in space and is a
quantitative measure of the openness of the landform and thus
the blocking degree of sky within sight. It effectively solves the
shadow problem in the case of a single light source using diffuse
reflection. For instance, steep ridges or peaks are brighter than
valleys, because they receive more light from the sky.
The simplest way to calculate SVF is to measure the solid
angle (Fig. 4). In the case of an observation point in the DEM,
the solid angle is defined as: a unit sphere centered at the obser-
vation point, and the area of any object projected onto this unit
sphere as defined by the solid angle of this object relative to the
observation point of interest. The solid angle is an area of the
unit sphere, representing the maximum area that can be covered
in the observation at one point (the projected area of the hemi-
sphere above the observer)
cos ∙
∙2 (1)
where and are the latitude and longitude within the hemi-
sphere. Since visible sky is limited by surface features, the solid
angle can be calculated as shown below, provided that the azi-
muths along the horizontal direction above the horizontal plane
share the same angles of elevation
cos ∙
∙2∙1sin (2)
where represents the elevation angle of the relief horizon.
Since actual observation points have varied surface heights along
different azimuths, they also have varied angles of elevation.
Therefore, the solid angle can be equivalently calculated by se-
lecting a certain quantity of azimuths along the horizontal plane
and computing the angle of elevation
corresponding to each
azimuth (Fig. 4)
∑cos ∙
2∙1
∑
(3)
where n is the quantity of search directions;
is the angle of
elevation at different azimuths.
Hence, the SVF can be calculated as below
SVF1
∑
(4)
As shown in Fig. 5, the SVF map generated from the air-
borne LiDAR-DEM can clearly reveal topological features in-
duced by geohazards, such as fractures, hummocks, pressure
ridges, and depressions, which in turn can be used to effectively
identify geohazards (Chen et al., 2015), even though most geo-
hazards are covered by vegetation and artificially modified.
3 RESULTS
3.1 Geohazards Inventory and Typology
The study area covers 135 km
2
with a total of 146 geohaz-
ards identified and interpreted (Fig. 6). They are then classified
according to the standard of classification for geological hazards
(T/CAGHP 001-2018), which results in landslides (n=86), rock-
falls (n=45), and debris flows (n=15) (Fig. 5), with a total geo-
hazard area of about 46.418 km
2
(accounting for 33.4% of the total
study area) and a surface density of about 1.08 geohazards/km
2
. In
the study area, the identified maximum and minimum geohaz-
ards are considered to be landslides, which presents a huge dif-
ference (937 m
2
and 9.36 km
2
, respectively).
Figure 3. LiDAR data processing and LiDAR data products and topographic derivatives.
Geohazard Recognition and Inventory Mapping Using Airborne LiDAR Data in Complex Mountainous Areas
1083
Figure 4. Schematic diagram of calculating sky-view factors (Zakšek et al., 2011).
Figure 5. An example of LiDAR data products and interpreted geohazards for a sample area. (a) Geohazard morphological characteristics used in the interpreta-
tions; (b) interpreted geohazards from SVF map.
Figure 6. LiDAR-based geohazard inventory map of the study area. Geohazard types are classified based on the standard of classification for geological hazards.
The pie chart (left upper panel) shows the percentage of each type.
Chen Guo, Qiang Xu, Xiujun Dong, Weile Li, Kuanyao Zhao, Huiyan Lu and Yuanzhen Ju
1084
Landslides are the most commonly-seen type of geohazard
in the study area (Fig. 6), they account for 58.90% of the total
geohazard. The identified landslides are mostly at the large to
medium scale. An educated guess made by this research is that
relatively large slides are generally formed by tectonic or glacier
movement and thus relatively old, such as the two landslides in-
terpreted at the Jiaju and Zhonglu towns in the study area, both
with an area of about 900×10
4
m
2
. As for some other medium-
small landslides, they are induced by rainfalls, river undercut-
ting, and artificial engineering activities (e.g., cutting slopes to
build houses and road excavation). Such landslides are often de-
bris slide developed on the deposition of older landslides, char-
acterized by generally low altitude, distribution along rivers and
human settlements next to highways, and multi-stage sliding.
Rockfalls are also one of the frequently-identified geohaz-
ard types (Fig. 7), accounting for 30.82% of the total geohazards.
Rockfalls often occur at relatively steep slopes on both sides of
highways and rivers, with average slope gradients of 40°–50°.
Such geohazards, which are relatively small, are found with de-
finitive source area and deposit zones. The source area is seen
with relatively high slope gradients, generally developed struc-
tural planes, and coarse textures. The deposit zone typically pre-
sents itself as an inverted cone shape, with small slope gradients
and smooth textures.
Another mapped geohazard type are debris flows (Fig. 7),
which occurrence constitutes 10.27% of the total geohazard
events. Such geohazards are generally elongated, with a large
watershed area, an identifiable source area, a narrow flow zone,
and a fan-shaped accumulation zone. They often develop at the
mid-upper part of slopes with relatively high altitudes.
3.2 Validation of LiDAR Results with Field Surveys
In order to validate the LiDAR recognition results, field inves-
tigation on the mapped geohazards in the study area was conducted.
It was found that the field evidences and topographic features were
consistent with the mapped geohazards using the LiDAR data. Be-
low we introduce two typical landslides that were investigated.
Landslide L01 is located in the Xiaoniega Village, Jiaju
Town (Fig. 6), with a front elevation of 1 920 m, a rear elevation
of 2 470 m, and thus an overall elevation difference of about 550
m. The slip direction is about 100°, with an average slope gradi-
ent of about 24°. The length of the landslide is about 1 200 m,
and the width, about 1 100 m, leading to a coverage of 1.01 km
2
.
The orthophotograph shows that a lot of vegetation developed
around the landslide (Fig. 8a), and the SVF map generated from
the LiDAR-DEM filtering vegetation presents clear signs of the
previous sliding. Scarps indicating multiple stages of sliding (at
least three stages) exist on the slope. The boundary of secondary
sliding is highly identifiable, and slope materials were relatively
complex and mostly crushed rocks (Figs. 8b, 8c).
Figure 7. Two- and three-dimensional examples and block diagram of landslide, rock fall, and debris flow.
Geohazard Recognition and Inventory Mapping Using Airborne LiDAR Data in Complex Mountainous Areas
1085
Field survey identifies this landslide as a large debris slide.
The landslide rear is associated with an apparent downward dislo-
cation. According to local residents, the downward dislocation of
the rear surface of the slope has been present for over ten years,
and the rear surface dives downward by 3–5 cm annually. Inten-
sive deformation is observed during the rainy season. The field
survey discovered apparently newly-born fractures (Fig. 8e),
nearly N-S striking, with a width of about 10 cm and an exten-
sion of 0.5–3 m. The surface material at the mid-rear slope of the
landslide is relatively crushed and accumulated relatively irreg-
ularly, and the field photo taken by UAV clearly shows that a
debris slide occurred (Fig. 8d). The accumulated materials are
mostly rock blocks, with apparent hanging phenomena. The
front of the landslide is found with good free surfaces due to the
slope that is cut by highways and rivers. Highway excavation
leads to the high steep side slope, which is seen with local rock-
falls and now reinforced by active nets (Fig. 8f). This landslide
imperils the scattered residential areas, highways, and the Dajin-
chuan River at the front.
Landslide L02 occurs at the right bank of the Dajinchuan
River, the Wulipai New District, Zhanggu Town (Fig. 6). The
plan view of the landslide is generally shaped like a long tongue.
Compared with the hillshade map (Fig. 9c), the SVF map cap-
tures with greater clarity the boundary of the landslide (Fig. 9b).
The front elevation of the landslide is 1 877 m, and the rear ele-
vation, 2 210 m, which leads to an elevation difference of 330
m. The landslide front is found with gentle relief, while the rear
is relatively steep. The slope gradients are between 25°–30° and
50°–60° respectively, with an overall slope gradient of about
31°. The length of the landslide is about 475 m; the width, about
260 m; the covered area, about 12.4×10
4
m
2
. Moreover, the
average thickness is 20 m or so, and thus the estimated landslide
volume reaches about 240×10
4
m
3
, indicating a large scale debris
slide. The slip direction of the landslide is about 40°. The surface
of the landslide is mainly composed of blocky rock-soil, and
rock blocks are varied in size. The thickness of the landslide is
relatively large. Rock blocks at the rear part of the landslide
present themselves like bedrocks, while the blocky rock-soil at
the mid-front part is loose-slightly tight.
Construction of cities and towns at Danba is seen with rapid
progress over recent years, construction of the Wulipai New Dis-
trict was initiated in 2013. Excavation of the slope toe to build
houses reduced the stability of the slope and subjected to local
sliding at the toe of the old landslide (Fig. 9g), and this ultimately
leads to the apparent overall sliding of the landslide. To prevent
further sliding of the landslide, a barrier wall was constructed at
the landslide front. However, due to intensive rainfall during the
rainy seasons over the past years, the front barrier wall is ob-
served with apparent bulging, cracking, and dislocation (Fig.
9h). The UAV photo shows the presence of crushed rock mass
that has not completely disintegrated at the landslide rear (Figs.
9d, 9e), which may roll along the slope in the case of earthquakes
or extreme weather conditions. The landslide endangers several
residential communities, apartments, hotels, and highways below
it and therefore lives and properties of several thousand people.
3.3 Geohazard Morphometric Characteristics and Spatial
Distribution
As shown in Fig. 10, morphometric parameters of the three
types of geohazards are differentiated from each other. Land-
slides show great variations in size. Rockfalls are found with
limited variations in various parameters such as areas, lengths,
and widths. The average elevation difference of slides is 283.4
m. As for debris flows, they also present a large average eleva-
tion difference (813.2 m). Concerning the average slope gradi-
ent, the slides (only 30°) is smaller than those of the other types
Figure 8. LiDAR and field investigation result of landslide L01. (a) Aerial orthophoto; (b) sky view factor map; (c) the landslide feature based on SVF map; (d)
the rock slide deposit in the middle of the landslide; (e) scarp in the back of the landslide; (f) active protection net in front of the landslide.
Chen Guo, Qiang Xu, Xiujun Dong, Weile Li, Kuanyao Zhao, Huiyan Lu and Yuanzhen Ju
1086
of geohazards (44.5° for rockfalls). The elevation distributions
of the three geohazard types have characteristics similar to slope
gradient distributions. Furthermore, the geohazard recognition
results show that the smallest landslide identified in the study
area with a resolution of 0.5 m covers an area of about 933 m2,
with a corresponding length of about 43 m, a width of about 24
m, an elevation difference of about 16 m, and an average slope
gradient of about 9°. As the mountains are high and steep, the
movement of geohazards is less limited by the terrain, the geo-
hazards showing the characteristics of a relatively great length-
width ratio in the study area, especially debris flow is the most
significant. In addition, the results indicate that the relationship
between length and width of all geohazards can be met with the
formula L=a+b×W0.95, and a=62.5, b=2.15.
Kernel density analysis has been regarded as a standard
technique for the spatial evaluation of geohazards (Görüm, 2019;
Pedrazzini et al., 2016). Accordingly, we created a kernel density
map in ArcGIS for the entire geohazards dataset. Figure 11 il-
lustrates the spatial distribution characteristics of geohazards
across the study area. Geohazards are mainly linearly distributed
along lateral sides of rivers and highways, which is related to the
extremely steep terrain of the study area. Human working and
living activities are concentrated in the low-altitude gentle-relief
area at the banks of river valleys. Besides linear distribution
along rivers, geohazards in the study area are found to be con-
centrated in three zones (Fig. 11), namely the Danba County-
Wulipai New District zone (Part 1), the Jiaju Tibetan Village
tourist center zone (Part 2), and the Jiaju Town zone (Part 3).
Based on the field survey, the following reasons are suggested
to explain concentrations of geohazards in these three zones.
First, for the rightmost Danba County-Wulipai New Dis-
trict (Part 1 in Fig. 11a), engineering construction and ur-
ban/township expansion have to excavate the slope toe for site
construction, due to terrain limitation. This, on the one hand,
leads to stress relaxation at the slope toe. On the other hand, it
creates extra effective free faces at the slope front. Therefore,
local instability and failure of the slope are induced, which reac-
tivates the old landslide. For instance, slope toe excavation for
new housing in 2005 reactivated the Danba County landslide
(Huang, 2009), which endangered more than half of the county.
Afterward, the county continuously expands northward to the
Wulipai New District. Numerous landslides (e.g., the Wulipai
New District landslide) were induced, due to further slope exca-
vation for building infrastructures such as hotels and hospitals.
Figure 9. LiDAR and field investigation result of landslide L02. (a) Aerial orthophoto; (b) sky view factor map; (c) hillshade map of 315° azimuths; (d) SVF
map and; (e) orthophoto of the broken rock in the back of the landslide; (f) the landslide feature based on the SVF map; (g) collapse in front of a landslide caused
by slope cutting; (h) crack in the retaining wall in front of the landslide.
Geohazard Recognition and Inventory Mapping Using Airborne LiDAR Data in Complex Mountainous Areas
1087
Figure 10. Morphometric parameters of the LiDAR-based geohazards.
Second, in zone of the Jiaju Tibetan Village Tourist Center
(Part 2 in Fig. 11a), the slope materials, mainly composed of silty
clays and crushed rock-soils, are loose mixed deposits of soils
and rocks, with irregular structures (Bai et al., 2020). The depos-
its have loose structures, with low mechanical strength and poor
stability, and thus are prone to deformation and failure induced
by exogenic load such as groundwater. This combined with the
construction of the tourist center and highways inevitably leads
to occurrences of numerous landslides.
Third, as for the leftmost Niega zone (Part 3 in Fig. 11a),
the slope is composed of the large old landslide deposits. The
Dajinchuan River flows through the toe of the slope, where the
narrow watercourse is seen with the rapid water level rising and
falling during the flood season, with a water level variation of 5–
12 m. The slope is intensively eroded by the river and thus has a
wide free face. Due to years of river undercutting and partially
slope cutting by highways, the slope adjacent to the river creeps
and slides towards the free face, which leads to numerous small
landslides.
4 DISCUSSION
4.1 The Differences of DEM Visualization Methods
When applying the high resolution DEM to geohazard
recognition and mapping, it is necessary to first visualize the
DEM. The conventional and also most prevailing method is to
generate hillshade map from DEM, which mimic the lightness
variation induced by the terrain under the sunlight to render the
topographic map and enable the three-dimensional representation
of the two-dimensional topological data. However, hillshade
maps are subjected to the influence of the solar azimuth and al-
titude, and thus are often greatly varied, which leads to high dif-
ficulties in effectively identifying small features of potential
landslides (Fig. 12) and thus reduce the accuracy and reliability
of geohazard recognition. In order to mitigate the effects of the
single light source upon visualization, Chiba et al. (2008) pro-
pose the red relief image map (RRIM) based on the slope gradi-
ent as well as positive and negative terrain openness to identity
and map individual landslides (Chiba et al., 2008).
Several conventional DEM visualization methods are com-
pared in this study, and the visual interpretation is used to eval-
uate the differentiation of landslide features (Fig. 12). Figure 12a
shows the orthophotograph of the same zone. A light source az-
imuth at 45°, perpendicular to the slope, leaves no shadow on the
slope (Fig. 12b); in contrast, a light source azimuth at 225°, par-
allel to the slope, results in excessive shadow on the slope (Fig.
12d). These two cases both fail to truly reflect the overall mor-
phological characteristics of landslides. Moreover, the landslide
feature can be most clearly captured with a light source azimuth
of 315° (Fig. 12e), followed by the case of 135° (Fig. 12c). The
positive openness, slpoe map and RRIM are independent of the
light source, and can relatively well illustrate the boundary char-
acteristics of landslides (Figs. 12f, 12g and 12h). Nevertheless,
the SVF map, in comparison, is able to clearly demonstrate not
only the boundary of the overall landslide but also features such
as rear scarps and front deposition in a way more clear than all
of the above methods (Fig. 12i).
Chen Guo, Qiang Xu, Xiujun Dong, Weile Li, Kuanyao Zhao, Huiyan Lu and Yuanzhen Ju
1088
Figure 11. The spatial distribution characteristics and influencing factors of the geohazards in the study area. (a) Kernel density analysis of all geohazards
(searching radius 1.2 km); (b) an example of the common V-shaped valley in the study area; (c) excavation for building at the toe of the slope; (d) loose deposits
widely distributed on the slope surface; (e) rock-soil aggregate of the deposits; (f) road construction, including the provincial highway and the township road; (g)
road excavation; (h) flood-control dam destroyed by river erosion.
4.2 Geohazard Scale Comparison between Ground-Based
and LiDAR-Based Inventories
We managed to obtain a copy of the geohazard inventory
based on a ground survey from the Sichuan Research Institute of
National Land Space Ecological Remediation and Geological
Hazard Prevention and Treatment. This geohazard inventory in-
cludes geohazard information such as the position coordinate
and coarsely estimated volume, whereas there is no information
on the boundary and area of every geohazards. According to the
empirical geohazard volume-area conversion equation (Guzzetti
et al., 2009), namely VL=0.074×AL1.45, the area of the geohazard
based on the ground survey was calculated, with the smallest
geohazard area of about 1 m2 and the largest of about 49×104 m2.
The area of LiDAR-based geohazards is larger than that of the
ground-based geohazards by one order of magnitude (Fig. 13a).
Moreover, the gap between the areas of debris flows mapped by
the two methods reaches even two orders of magnitude. This
indicates that LiDAR fails to present satisfactory performance in
identifying small or shallow landslides within a relatively large
area, because small landslides show no apparent topographic
features (Lin et al., 2013). However, it performs tremendously
well in recognizing large landslides (van den Eeckhaut et al.,
2007). Hence, the airborne LiDAR data seem to be able to pro-
vide satisfactory results in identifying medium and large geohaz-
ards in the study area.
In addition, the probability density of the LiDAR-based
geohazards in the study area also follows the inverse gamma
function (Malamud et al., 2004), which is consistent with most
earthquake-induced geohazards, such as Wenchuan earthquake
in May 12, 2008 (Ms8.0) and Jiuzhaigou earthquake in August 8,
2017 (Ms7.0)(Fan et al., 2018). But its probability density is
much lower than earthquake-induced geohazards. Nonetheless,
it is seen that on the left of the rollover point exists a limited
quantity of small geohazards, while on the right of the rollover
point lie relatively larger geohazards (Fig. 13b), and the large
geohazards (on the right of the rollover point) follow the power-
law distribution, with the expanded of area, the probability den-
sity continuously declines. LiDAR-based inventory and ground-
based inventory are compared in this study, the result shows that
ground survey can recognize small-scale geohazards better than
LiDAR. Moreover, the probability density of LiDAR-based in-
ventory in this study is consistent with the findings of Tolga
Görüm in western Black Sea, Turkey, which contains 167 deep-
seated landslides (depth of the landslide failure surface >5 m)
(Görüm, 2019). And compared with the results based on API
(aerial photo interpretation), the API-based landslide inventory
recognized comparatively large landslides in western Black Sea,
Turkey.
In conclusion, LiDAR provides satisfactory results in iden-
tifying medium and large geohazards but fails to present satis-
factory performance in identifying small or shallow geohazards.
These can be explained by resolution limitations of the airborne
LiDAR-derived DEM. For small geohazards with no apparent
features, it is difficult to observe their topographical features
(such as the scarp, lateral edges, deposit, toe of landslide). Con-
sequently, they may be missed in the recognition process.
Geohazard Recognition and Inventory Mapping Using Airborne LiDAR Data in Complex Mountainous Areas
1089
Figure 12. The differences of DEM visualization methods. (a) Aerial orthophoto image of the landslide area; (b) hillshade map of 45° azimuth; (c) 135° azimuth;
(d) 225° azimuth; (e) 315° azimuth; (f) positive openness map; (g) slope map; (h) red relief image map; (i) sky view factor map.
Figure 13. Geohazard scale comparison of the two inventories. (a) Box plots showing areas of LiDAR-based and ground-based landslides; (b) probability density
functions (inverse gamma (Malamud et al., 2004)) for the landslide area distribution of the two inventories (API represents aerial photo interpretation).
Chen Guo, Qiang Xu, Xiujun Dong, Weile Li, Kuanyao Zhao, Huiyan Lu and Yuanzhen Ju
1090
5 CONCLUSIONS
This study presents a new geohazard inventory based on in-
terpretations of the airborne LiDAR data acquired in the Danba
county, southwest China, with the aims to better understand and
quantify the capacity of the LiDAR method in recognizing geo-
hazards in the mountainous areas with high altitudes and great
elevation differences. This is the first geohazard inventory ever
produced from airborne LiDAR data in China. Moreover, a new
method based on SVF visualization maps is proposed to recog-
nize the topographic signatures related to geohazards, which pro-
vides better topographic information than the classical hillshade
maps. A total of 146 geohazards are remotely identified and
mapped in the entire 135 km2 study area using the airborne Li-
DAR data, and nearly one-third of the study area is dominated
by geohazards. Results of this study further demonstrate the sig-
nificant contribution of high-resolution topographical data such
as LiDAR to improve recognition of old and/or previously un-
known geohazards. The minimum landslide area that can be rec-
ognized by the 0.5 m resolution DEM is about 900 m2. Spatial
distributions of mapped geohazards are found to be highly re-
lated to human engineering activities, such as road excavation
and slope cutting. Moreover, unreasonable human engineering
in the process of urban expansion in mountainous areas has in-
duced a large number of landslides and caused serious economic
losses and casualties. In this context, a more precise assessment
of these landslides is critical. In this respect, airborne LiDAR
data provide a unique opportunity for understanding the land-
slide processes and reducing related losses, especially for the
Southwest China mountainous areas, where altitude, elevation
difference, and forest cover density are very high and the hazards
caused by landslides are very significant.
ACKNOWLEDGMENTS
The research was supported by the National Innovation Re-
search Group Science Fund (No. 41521002) and the National
Key Research and Development Program of China (No.
2018YFC1505202). The authors thank the Sichuan Provincial
Surveying and Mapping Geographic Information Bureau for
providing airborne LiDAR data and the Sichuan Research Insti-
tute of National Land Space Ecological Remediation and Geo-
logical Hazard Prevention and Treatment for providing ground-
based landslide inventory. The final publication is available at
Springer via https://doi.org/10.1007/s12583-021-1467-2.
REFERENCES CITED
Ardizzone, F., Cardinali, M., Galli, M., et al., 2007. Identification and Map-
ping of Recent Rainfall-Induced Landslides Using Elevation Data Col-
lected by Airborne Lidar. Natural Hazards and Earth System Sciences,
7(6): 637–650. https://doi.org/10.5194/nhess-7-637-2007
Bai, Y. J., Wang, Y. S., Ge, H., et al., 2020. Slope Structures and Formation
of Rock-Soil Aggregate Landslides in Deeply Incised Valleys. Journal
of Mountain Science, 17(2): 316–328. https://doi.org/10.1007/s11629-
019-5623-4
Bell, R., Petschko, H., Röhrs, M., et al., 2012. Assessment of Landslide Age,
Landslide Persistence and Human Impact Using Airborne Laser Scanning
Digital Terrain Models. Geografiska Annaler: Series A, Physical Geogra-
phy, 94(1): 135–156. https://doi.org/10.1111/j.1468-0459.2012.00454.x
Chen, N. S., Li, T. C., Gao, Y. C., 2005. A Great Disastrous Debris Flow on
11 July 2003 in Shuikazi Valley, Danba County, Western Sichuan,
China. Landslides, 2(1): 71–74. https://doi.org/10.1007/s10346-004-
0041-1
Chen, R. F., Chang, K. J., Angelier, J., et al., 2006. Topographical Changes
Revealed by High-Resolution Airborne LiDAR Data: The 1999 Tsaol-
ing Landslide Induced by the Chi-Chi Earthquake. Engineering Geol-
ogy, 88(3/4): 160–172. https://doi.org/10.1016/j.enggeo.2006.09.008
Chen, R. F., Lin, C. W., Chen, Y. H., et al., 2015. Detecting and Characteriz-
ing Active Thrust Fault and Deep-Seated Landslides in Dense Forest Ar-
eas of Southern Taiwan Using Airborne LiDAR DEM. Remote Sensing,
7(11): 15443–15466. https://doi.org/10.3390/rs71115443
Chen, W. T., Li, X. J., Wang, Y. X., et al., 2014. Forested Landslide Detection
Using LiDAR Data and the Random Forest Algorithm: A Case Study of
the Three Gorges, China. Remote Sensing of Environment, 152: 291–
301. https://doi.org/10.1016/j.rse.2014.07.004
Chiba, T., Kaneta, S., Suzuki, Y., 2008. Red Relief Image Map: New Visual-
ization Method for Three Dimensional Data. The International Archives
of the Photogrammetry, Remote Sensing and Spatial Information Sci-
ences, 37(B2), 1071–1076. http://www.isprs.org/proceed-
ings/xxxvii/congress/2_pdf/11_ths-6/08.pdf.
Chigira, M., Duan, F. J., Yagi, H., et al., 2004. Using an Airborne Laser Scan-
ner for the Identification of Shallow Landslides and Susceptibility As-
sessment in an Area of Ignimbrite Overlain by Permeable Pyroclastics.
Landslides, 1(3): 203–209. https://doi.org/10.1007/s10346-004-0029-x
Comert, R., Avdan, U., Gorum, T., et al., 2019. Mapping of Shallow Land-
slides with Object-Based Image Analysis from Unmanned Aerial Vehi-
cle Data. Engineering Geology, 260(1): 105264.
https://doi.org/10.1016/j.enggeo.2019.105264
Conrad, O., Bechtel, B., Bock, M., et al., 2015. System for Automated Geo-
scientific Analyses (SAGA) V.2.1.4. Geoscientific Model Development,
8(7): 1991–2007. https://doi.org/10.5194/gmd-8-1991-2015
Dong, J., Zhang, L., Tang, M. G., et al., 2018. Mapping Landslide Surface
Displacements with Time Series SAR Interferometry by Combining Per-
sistent and Distributed Scatterers: A Case Study of Jiaju Landslide in
Danba, China. Remote Sensing of Environment, 205: 180–198.
https://doi.org/10.1016/j.rse.2017.11.022
Eeckhaut, M. V. D., Poesen, J., Verstraeten, G., et al., 2007. Use of LIDAR-
Derived Images for Mapping Old Landslides under Forest. Earth Surface
Processes and Landforms, 32(5): 754–769. https://doi.org/10.1002/esp.1417
Fan, X. M., Scaringi, G., Xu, Q., et al., 2018. Coseismic Landslides Triggered
by the 8th August 2017 Ms7.0 Jiuzhaigou Earthquake (Sichuan, China):
Factors Controlling Their Spatial Distribution and Implications for the
Seismogenic Blind Fault Identification. Landslides, 15(5): 967–983.
https://doi.org/10.1007/s10346-018-0960-x
Fan, X. M., Xu, Q., Alonso-Rodriguez, A., et al., 2019. Successive Landslid-
ing and Damming of the Jinsha River in Eastern Tibet, China: Prime
Investigation, Early Warning, and Emergency Response. Landslides,
16(5): 1003–1020. https://doi.org/10.1007/s10346-019-01159-x
Fan, X. M., Xu, Q., Scaringi, G., et al., 2017. Failure Mechanism and Kinematics
of the Deadly June 24th 2017 Xinmo Landslide, Maoxian, Sichuan, China.
Landslides, 14(6): 2129–2146. https://doi.org/10.1007/s10346-017-0907-7
Fan, X., Xu, Q., Huang, R., et al., 2007. Dynamical Optimal Anchoring De-
sign and Information Construction of Danba Landslide. Chinese Journal
of Rock Mechanics and Engineering, 26 (S2): 4139–4146 (in Chinese
with English Abstract)
Fiorucci, F., Cardinali, M., Carlà, R., et al., 2011. Seasonal Landslide Map-
ping and Estimation of Landslide Mobilization Rates Using Aerial and
Satellite Images. Geomorphology, 129(1/2): 59–70.
Geohazard Recognition and Inventory Mapping Using Airborne LiDAR Data in Complex Mountainous Areas
1091
https://doi.org/10.1016/j.geomorph.2011.01.013
Gao, J., Maro, J., 2010. Topographic Controls on Evolution of Shallow Land-
slides in Pastoral Wairarapa, New Zealand, 1979–2003. Geomorphol-
ogy, 114(3): 373–381. https://doi.org/10.1016/j.geomorph.2009.08.002
Görüm, T., 2019. Landslide Recognition and Mapping in a Mixed Forest En-
vironment from Airborne LiDAR Data. Engineering Geology, 258:
105155. https://doi.org/10.1016/j.enggeo.2019.105155
Guzzetti, F., Ardizzone, F., Cardinali, M., et al., 2009. Landslide Volumes and Land-
slide Mobilization Rates in Umbria, Centr al Italy. Earth and Planetary Science
Letters, 279(3/4): 222–229. https://doi.org/10.1016/j.epsl.2009.01.005
Guzzetti, F., Mondini, A. C., Cardinali, M., et al., 2012. Landslide Inventory
Maps: New Tools for an Old Problem. Earth-Science Reviews, 112(1/2):
42–66. https://doi.org/10.1016/j.earscirev.2012.02.001
Huang, R. Q., 2009. Some Catastrophic Landslides since the Twentieth Cen-
tury in the Southwest of China. Landslides, 6(1): 69–81.
https://doi.org/10.1007/s10346-009-0142-y
Jaboyedoff, M., Metzger, R., Oppikofer, T., et al., 2007. New Insight Techniques
to Analyze Rock-Slope Relief Using DEM and 3D-Imaging Cloud Points.
Rock Mechanics: Meeting Society’s Challenges and Demands. In: Eber-
hardt, E., Stead, D., Morrison, T., eds., Rock mechanics: Meeting Society’s
challenges and demands. Proceedings of the 1st Canada––U.S. Rock Me-
chanics Symposium, Vancouver, Canada, 27–31 May 2007. Taylor & Fran-
cis, 61–68. https://doi.org/10.1201/noe0415444019-c8
Jaboyedoff, M., Oppikofer, T., Abellán, A., et al., 2012. Use of LIDAR in
Landslide Investigations: A Review. Natural Hazards, 61(1): 5–28.
https://doi.org/10.1007/s11069-010-9634-2
Li, M. H., Zheng, W. M., Shi, S. W., et al., 2008. The Revival Mechanism
and Stability Analysis to Jiaju Landslide of Danba County in Sichuan
Province. Journal of Mountain Science, 26(5): 577–582 (in Chinese with
English Abstract)
Li, W., Xu, Q., Lu, H., et al., 2019. Tracking the Deformation History of
Large-Scale Rocky Landslides and Its Enlightenment. Geomatics and
Information Science of Wuhan University, 44 (7): 1043–1053 (in Chi-
nese with English Abstract)
Li, X. J., Cheng, X. W., Chen, W. T., et al., 2015. Identification of Forested
Landslides Using LiDar Data, Object-Based Image Analysis, and Ma-
chine Learning Algorithms. Remote Sensing, 7(8): 9705–9726.
https://doi.org/10.3390/rs70809705
Lin, M. L., Chen, T. W., Lin, C. W., et al., 2013. Detecting Large-Scale Landslides
Using Lidar Data and Aerial Photos in the Namasha- Liuoguey Area, Taiwan.
Remote Sensing, 6(1): 42–63. https://doi.org/10.3390/rs6010042
Malamud, B. D., Turcotte, D. L., Guzzetti, F., et al., 2004. Landslides, Earth-
quakes, and Erosion. Earth and Planetary Science Letters, 229(1/2): 45–
59. https://doi.org/10.1016/j.epsl.2004.10.018
McKean, J., Roering, J., 2004. Objective Landslide Detection and Surface
Morphology Mapping Using High-Resolution Airborne Laser Altime-
try. Geomorphology, 57(3/4): 331–351. https://doi.org/10.1016/s0169-
555x(03)00164-8
Nichol, J., Wong, M. S., 2005. Detection and Interpretation of Landslides Us-
ing Satellite Images. Land Degradation & Development, 16(3): 243–
255. https://doi.org/10.1002/ldr.648
Parker, R. N., Densmore, A. L., Rosser, N. J., et al., 2011. Mass Wasting Trig-
gered by the 2008 Wenchuan Earthquake is Greater than Orogenic Growth.
Nature Geoscience, 4(7): 449–452. https://doi.org/10.1038/ngeo1154
Pedrazzini, A., Humair, F., Jaboyedoff, M., et al., 2016. Characterisation and
Spatial Distribution of Gravitational Slope Deformation in the Upper
Rhone Catchment (Western Swiss Alps). Landslides, 13(2): 259–277.
https://doi.org/10.1007/s10346-015-0562-9
Peng, D. L., Xu, Q., Liu, F. Z., et al., 2018. Distribution and Failure Modes
of the Landslides in Heitai Terrace, China. Engineering Geology, 236:
97–110. https://doi.org/10.1016/j.enggeo.2017.09.016
Petschko, H., Bell, R., Glade, T., 2016. Effectiveness of Visually Analyzing
LiDAR DTM Derivatives for Earth and Debris Slide Inventory Mapping
for Statistical Susceptibility Modeling. Landslides, 13(5): 857–872.
https://doi.org/10.1007/s10346-015-0622-1
Roering, J. J., MacKey, B. H., Marshall, J. A., et al., 2013. ‘You are HERE’: Con-
necting the Dots with Airborne Lidar for Geomorphic Fieldwork. Geomor-
phology, 200: 172–183. https://doi.org/10.1016/j.geomorph.2013.04.009
Santangelo, M., Cardinali, M., Rossi, M., et al., 2010. Remote Landslide
Mapping Using a Laser Rangefinder Binocular and GPS. Natural Haz-
ards and Earth System Sciences, 10(12): 2539–2546.
https://doi.org/10.5194/nhess-10-2539-2010
Tang, C. X., Tanyas, H., van Westen, C. J., et al., 2019. Analysing Post-Earthquake
Mass Movement Volume Dynamics with Multi-Source DEMs. Engineering
Geology, 248: 89–101. https://doi.org/10.1016/j.enggeo.2018.11.010
Tomás, R., Li, Z. H., 2017. Earth Observations for Geohazards: Present and Future
Challenges. Remote Sensing, 9(3): 194. https://doi.org/10.3390/rs9030194
Trigila, A., Iadanza, C., Spizzichino, D., 2010. Quality Assessment of the
Italian Landslide Inventory Using GIS Processing. Landslides, 7(4):
455–470. https://doi.org/10.1007/s10346-010-0213-0
Yamazaki, F., Kubo, K., Tanabe, R., et al., 2017. Damage Assessment and 3d
Modeling by UAV Flights after the 2016 Kumamoto, Japan Earthquake.
2017 IEEE International Geoscience and Remote Sensing Symposium
(IGARSS). July 23–28, 2017, Fort Worth, TX, USA. IEEE, 3182–3185.
https://doi.org/10.1109/igarss.2017.8127673
Yin, Y. P., Wang, F. W., Sun, P., 2009. Landslide Hazards Triggered by the
2008 Wenchuan Earthquake, Sichuan, China. Landslides, 6(2): 139–
152. https://doi.org/10.1007/s10346-009-0148-5
Zakšek, K., Oštir, K., Kokalj, Ž., 2011. Sky-View Factor as a Relief Visualization
Technique. Remote Sensing, 3(2): 398–415. https://doi.org/10.3390/rs3020398
Zhang, S. L., Yin, Y. P., Hu, X. W., et al., 2020. Initiation Mechanism of the
Baige Landslide on the Upper Reaches of the Jinsha River, China. Land-
slides, 17(12): 2865–2877. https://doi.org/10.1007/s10346-020-01495-3
Zheng, G., Xu, Q., Ju, Y. Z., et al., 2018. The Pusacun Rockavalanche on
August 28, 2017 in Zhangjiawan Nayongxian, Guizhou: Characteristics
and Failure Mechanism. Journal of Engineering Geology, 26(1): 223–
240 (in Chinese with English Abstract)
Zhu, Y. Q., Xu, S. M., Zhuang, Y., et al., 2019. Characteristics and Runout
Behaviour of the Disastrous 28 August 2017 Rock Avalanche in
Nayong, Guizhou, China. Engineering Geology, 259: 105154.
https://doi.org/10.1016/j.enggeo.2019.105154
