Arabian Journal of Geosciences (2021) 14:1019

https://doi.org/10.1007/s12517-021-07396-6

ORIGINAL PAPER

Geotechnical investigation and landslide susceptibility assessment

along the Neelum road: a case study from Lesser Himalayas, Pakistan
Khawaja Shoaib Ahmed 1 & Muhammad Basharat 1 & Muhammad Tayyib Riaz 1,2 & Yasir Sarfraz 1 & Amir Shahzad 1,2

Received: 22 October 2020 / Accepted: 21 May 2021

# Saudi Society for Geosciences 2021

Abstract
Globally, and in Pakistan, among natural hazards, landslides are considered one of the most dangerous and frequently occurring
events having devastating impacts on society and economy. The present study deals with the susceptibility mapping and detailed
geological and geotechnical investigations on two large-scale landslides named Shahkot and Sandok, Lesser Himalayas, Pakistan.
Inventory of 74 landslides was developed by SPOT-5 images and further classified in the field. The goodness of developed landslide
susceptibility map was assessed by success rate curve (SRC) and prediction rate curve (PRC) which were 83.1% and 79.2%,
respectively. Geotechnical investigations of selected landslides were carried out to identify the causative factors and landslide
triggering mechanism. The Shahkot landslide is classified as complex (debris slide and slump failure), whereas the Sandok landslide
is classified as rockslide. Laboratory testing, i.e., sieve analysis, Atterberg limits, specific gravity, and X-Ray Diffraction (XRD)
analysis of the disturbed soil samples, reveal that the soils of Shahkot landslide are silty sand to clayey silty sand with plasticity
index (PI) ranging from 2.05 to 14.96%. Petrographic studies showed that the schist and granite of the Shahkot and Sandok
landslides contain quartz and feldspar with fair amounts of flaky minerals like chlorite, biotite, and muscovite. The XRD analysis
showed mineral composition of quartz, muscovite, kaolinite, goethite, aragonite, hematite, plagioclase, siderite, montmorillonite,
calcite, gypsum, orthoclase, dolomite, and illite. Intense jointing and fracturing in granite due to close proximity to faults imparts
low unconfined compressive strength (UCS) values which holds the major cause of Sandok landslide beside other geomorpholog-
ical and geological causes. The study depicts that drainage network, anthropogenic activities along steep slopes, fragile geology,
active faults, freeze, and thaw action are influential parameters which significantly contribute to the landslide events.

Keywords Landslide susceptibility . Geotechnical investigations . Petrography . XRD . UCS

Introduction areas (Petley et al. 2006; Jadoon et al. 2015; Owen et al. 2008;
Kamp et al. 2008; Basharat 2012). Particularly, northern part
Natural hazards especially landslides are major challenges of Azad Kashmir is characterized by steep slopes, active tec-
around the globe, and many countries are employing immense tonics, fragile rock units, and heavy rains that result the area
amount and resources to cope with them. In developing coun- more susceptible to landslides. Strong earthquakes are among
tries like Pakistan, landslide is among the major disasters re- the prime triggering factors of landslides. On October 8, 2005,
sponsible for fatalities, damages of the communication links, the Kashmir earthquake (Mw = 7.6, US Geological Survey)
loss of fertile soil, and economic losses in the mountainous occurred on the NW–SE trending Kashmir Boundary Thrust
(KBT), at the northwestern side of the Hazara Kashmir
Responsible Editor: Biswajeet Pradhan Syntaxis (HKS), northern Pakistan. The 2005 Kashmir earth-
quake triggered several thousands of landslides in the region.
* Muhammad Tayyib Riaz Various methods and techniques have been applied for
tayyibriaz@yahoo.com landslide susceptibility mapping, e.g., inventory-based
methods (Akgun 2012), logistic regression (LR) statistical in-
1
Institute of Geology, University of Azad Jammu and Kashmir
dex, artificial neural networks (ANN), frequency ratio (FR),
Muzaffarabad, Muzaffarabad, Pakistan cluster analysis, weight of evidence (WoE) (Pradhan and Lee
2
Department of Geology, University of Azad Jammu and Kashmir,
2010; Yılmaz 2009; Buša et al. 2019; Riaz et al. 2018; Sujatha
Neelum Campus, Athmuqam, Pakistan et al. 2012), analytic hierarchy process (AHP), fuzzy logic
1019 Page 2 of 19 Arab J Geosci (2021) 14:1019

(Basharat et al. 2016; Leonardi et al. 2016; Vojteková and more than 45 days as a result of massive landsliding activated
Vojtek 2020; Senouci et al. 2021), data mining, and machine by the earthquake (Basharat et al. 2012; Basharat et al. 2017).
learning (ML) methods (Ali et al. 2020; Lee et al. 2017; Resultantly, the communication link was totally disrupted be-
Kavzoglu et al. 2019; Al-Najjar and Pradhan 2020). tween Muzaffarabad and other parts of the Neelum Valley. In
ML and statistical models are mainly affected by the caus- response, the relief and rescue operation during the earthquake
ative factor selection (Feizizadeh et al. 2017; Can et al. 2019). was badly affected. These landslides caused the severe dam-
However, the integrated models have been effectively used age on landscape and road infrastructure. On the contrary,
during last few years to improve the susceptibility (Kalantar many other shallow landslides were also triggered during this
et al. 2018). Several studies of model integration include en- earthquake and resulted in number of fatalities and severe
semble models (Bragagnolo et al. 2020) and integration of damage to the main road.
knowledge-based and data-driven models (Zhang et al. There was a clear need and gap in the literature to investigate
2019). In addition, Yilmaz and Ercanoglu (2019) addressed the landslide susceptibility and to analyze the geotechnical and
the importance of data mining selection techniques and expe- mineralogical behavior of landslide material in the region.
rienced that polygon feature sampling methods are more real- Therefore, the current study was carried out along road section
istic in attaining the reliable maps than other techniques. from Nauseri to Athmuqam, district Neelum valley, Azad
After the Kashmir earthquake, various studies on co- Kashmir, Pakistan, to fill out this gap (Fig. 1). Firstly, the land-
seismic landslide identification, distribution analysis, evolu- slide inventory of the area was prepared using SPOT-5 satellite
tion, size, lithological control, and susceptibility mapping imageries and was classified in the field too. Landslide proba-
have been carried out in the area (e.g., Petley et al. 2006; bility map of the region by applying WoE method was devel-
Kumar et al. 2006; Sato et al. 2007; Dunning et al. 2007; oped to identify the probable hazardous zones. Two large-scale
Kamp et al. 2008; Owen et al. 2008; Khan et al. 2010; Saba catastrophic landslides (Sandok and Shahkot) were selected for
et al. 2010; Chini et al. 2011; Basharat et al. 2014; Basharat the detailed geotechnical and geomorphological investigation
et al. 2016; Shafique et al. 2016; Riaz et al. 2018; Riaz et al. to comprehend the causes and failure mechanism (Fig. 1). The
2019). However, few studies dealt with the geotechnical in- objectives of this study were to identify and characterize the
vestigations of landslides (e.g., Kiyota et al. 2011; Konagai landslide hazardous zones along Neelum Valley road that could
and Sattar 2012; Sattar et al. 2011; Riaz et al. 2019). They help the planners and decision makers for the safe continuity of
identified major landslides in terms of susceptibility and size traffic along the main Neelum road in future. To minimize the
in the region, i.e., Donga Kas and Hattian Bala landslides. landslide-associated risk in the future and to understand the
Riaz et al. (2019) investigated the Donga Kas landslide to landslide mechanism, this research focused on generating a
evaluate the possible initiation mechanism and movement. susceptibility map and geotechnical investigation. This study
However, the remaining landslides (i.e., Sandok and is the first attempt to understand the primary root cause of
Shahkot) needed to be investigated, so this study was an at- landslide mechanism by integrating geotechnical, mineralogi-
tempt to analyze the failure mechanisms of these catastrophic cal, and remote sensing–based approach in the region and will
landslides. Unfortunately, no landslide susceptibility map is contribute as primary database for future research in this
available along the studied road section which was the clear domain.
need to analyze the susceptibility in the region.
The study area lies in the Neelum valley situated towards
the northeast of the Muzaffarabad and is part of the lesser Geology and tectonics of the study area
Himalayan region. It is a steep, bow-shaped valley comprises
of an area about 3621 Km2 which hosts the total population of The study area is generally mountainous with narrow val-
about 200,000 inhabitants (Planning and Development leys. The altitude ranges between 923 and 1974 m asl
Department AJK 2015). The slope failures occur mainly along (Fig. 1). The steep slopes and escarpments are prominent
the main road built in the mountain environment. In addition, features of the area. The Neelum River and its tributaries
the deforestation has further enhanced the phenomenon of drain the entire region. Tectonically, the study area is part
mass wasting in the region (Rieux et al. 2007). The road along of the NW Himalayan fold and thrust belt (Kazmi and Jan
Neelum valley is the main transportation corridor which con- 1997). The area lies primarily in the eastern side of the
nects Muzaffarabad, the capital city of Pakistani HKS, Lesser Himalayas, across the Main Boundary
Administrated Kashmir (PAK), with the other localities of Thrust (MBT). The MBT, Panjal Thrust (PT) and local
the Neelum valley. The road has been badly affected due to faults, i.e., Barian Fault, Islampura Fault, and Bata
landslides during the rainy season and caused road blockage Fault, are present within the study area (Fig. 2). The
for many days to weeks. The blockage of this road often Panjal Formation is thrusted over the Murree Formation
causes isolation of the population and shortage of food, med- of Miocene age along the MBT (Khan 1994). The PT
icine, and other commodities. The road has been blocked marks the boundary between the Tanol Formation of
Arab J Geosci (2021) 14:1019 Page 3 of 19 1019

Fig. 1 Geographical location of

the study area

Precambrian age and the Panjal Formation of the Methodology

Carboniferous-Triassic age at the eastern limb of the
HKS. The Cambrian Neelum/Jura granite is intruded in The data related to landslide hazards along the Neelum road in
the Tanol Formation. Barian Fault lies between the general and about the landslides problem in particular was
Tanol schist and the Jura granite and dips to the northeast limited. However, all the available information was gathered
at about 30°. The fault lies towards south of the Sandok and used. Landslide inventory map of the area was prepared
landslide and effect of shearing is prominent along this by utilizing SPOT-5 satellite imageries and Google earth and
fault. The northern contact of granite exposed in Sandok further classified in the field in accordance to Varnes (1978)
landslide is also faulted which is exposed at the opposite classification. Landslide causative factors were identified, and
side on the left bank of the Neelum river near Islampura eight causative factor maps were prepared. Digital
village. The Bata Fault is exposed towards the southern elevation model (DEM) of 12.5m resolution was utilized for
end of the Shahkot landslide and lies between Athmuqam computation of slope, aspect, curvature, elevation, and drain-
biotite chlorite schist-phyllite and the Kundalshahi Nagdar age network causative factor rasters in Arc-GIS 10.8 environ-
garnet mica schist (Ghazanfar et al. 1983). ment. Landslide polygons were converted to point features
Rocks of Lesser Himalayas comprises the Tanol Formation and randomly subdivided into (70:30) training and validation
of Precambrian age, intruded by the Mansehra type granites of samples by utilizing subset features tool in Arc-GIS.
the Cambrian (Wadia 1931; Calkins et al. 1975; Fig. 2). Geological map and fault network were digitized from
Precambrian Tanol Formation generally comprises of meta- existing maps and then rasterized. A data-driven WoE method
sediments including quartzose schist, chlorite schist, graphitic was adopted to generate the susceptibility map of the area.
schist, garnet mica schist, and quartzite. The Tanol Formation Arc-SDM tool was used for computation of weights and other
is intruded by the Jura granite of the Cambrian age and further calculations. Developed landslide susceptibility map was val-
intruded by dolerite dykes. The general dip direction from idated by SRC and PRC curves. Based on developed suscep-
Sandok to Shahkot village is northeast (Ghazanfar et al. 1983). tibility map, the study region was classified to low susceptible,
1019 Page 4 of 19 Arab J Geosci (2021) 14:1019

Fig. 2 Geological map of the

study area

moderate susceptible, high susceptible, and very high suscep- strength parameters of the failure zones. The samples were
tible zones. Methodological steps for generating susceptibility taken systematically from main scarp, main body, and toe
maps are similar as Riaz et al. (2018). areas, three samples from each segment, i.e., right flank,
Detailed geotechnical and geochemical investigations were middle portion, and left flank. A total of 11 soil samples and
also carried out. Furthermore, base map of each landslide was 12 rock samples were collected according to the material ex-
prepared for detail geotechnical mapping and construction of posed along the sliding surfaces. The 5kg soil samples were
profiles. For this purpose, longitudinal and cross-sectional collected with the help of auger from 1 m depth and were
profiles have been constructed. The profiles basically show packed in air tight plastic bags to avoid loss of moisture
the initiation of landslide movement and related material ex- content.
posed on the landslide body. These profiles have been used to A series of laboratory tests such as grain size distribution,
understand the relationship between the intact mass and the Atterberg limits, specific gravity, UCS, petrography, and
material moved along the failure surfaces. The volume of XRD were performed on disturbed soil samples and rock core
landslides was roughly assessed by multiplying the landslide samples to determine the strength and physical characteristics
deposit area with the average thickness (Basharat et al. 2012). of the material for slope failure. These tests have been per-
The field investigations were carried out using Laser Distance formed according to the American Society for Testing and
Meter (RIEGL-F-21H), Clinometer, Global Positioning Materials (ASTM) standard test methods. Grain size analysis
System (GPS), Brunton, and tape measurements. Two major was performed according to ASTM D422 standard using stan-
landslides, i.e., Shahkot and Sandok, have been mapped on dard set of sieves. Atterberg limits were determined by ASTM
scale 1:6000 and 1:1000 respectively by using D4318-00 standard test method which include liquid limit by
ArcGIS software. The various lithological units have been Casagranda’s one-point method and plastic limit by using
observed and mapped. The rock and disturbed soil samples glass plate. ASTM standard test method D854-14 was used
have been taken for soil classification and to identify the shear to determine the specific gravity by Pycnometer. UCS test was
Arab J Geosci (2021) 14:1019 Page 5 of 19 1019

performed using ASTM D7012-14e1. Twelve core samples flow, rotational debris slide, translational slide, rock fall, com-
were cut to required dimensions, grinded to make the surfaces plex, and rockslides (Fig. 3). Majority of the landslides are
smooth, and then tested to determine the UCS of each sample. debris flow (26%), followed by rotational debris slide
For petrographic analysis, thin sections were prepared and (22%), translational debris slide (20%), translational
studied under petrographic microscope according to ASTM rockslides (18%), rock fall (13%), and complex slide (1%).
C295/C295M-19. XRD was performed on powdered samples
in the Geoscience Laboratory of Geological Survey of
Pakistan, Islamabad. Landslide causative factors

The landslide probability was evaluated by integrating

existing landslide locations with evidential rasters. Eight caus-
Results and discussions ative factors including topographic factors (slope aspect, slope
gradient, curvature, drainage network, and elevation), geolog-
Landslide inventory ical factors (lithology and fault lines), and road network were
derived (Fig. 4). These parameters were considered as causa-
The classification system used for this study is based on tive factors for the slope failure in the study area. These all 8
Varnes (1978). To classify the landslides along the Neelum causative factors were processed in raster in ArcGIS. Slope
Valley road, the terminology approved in the literature was gradient ascertains the divergence and convergence of rainwa-
used. Seventy-four landslides were observed and demarcated ter on a slope and thereby affects soil formation, soil water
via SPOT-5 and classified during field investigations. Based content, slope stability, and erosion potential (Vijith and
on field observation, these landslides were classified as debris Madhu 2008).

Fig. 3 Landslide inventory of the

area
1019 Page 6 of 19 Arab J Geosci (2021) 14:1019

Fig. 4 Landslide susceptibility causative factors: a slope gradient; b slope aspect; c curvature; d terrain elevation; e distance to streams; f distance to road;
g distance to faults; h lithological units

Slope aspect is another important contributing factor, e.g., methods for landslide probability mapping depend on the data
slope directly faces the hitting of the sun light resulting in the accessibility and local ground settings.
melting of snow and water infiltration, which increases the Different quantitative and qualitative methods are devel-
freezing and thawing action, due to which slope failure occurs. oped for landslide probability analysis (e.g., Aleotti and
Slope failure is also affected by curvature (Nefeslioglu et al. Chowdhury 1999; Guzzetti et al. 2005; Soeters and van
2008). Elevation is a dominant contributory factor considered Westen 1996). In this research, we adopted data-driven WoE
for landslide probability mapping (Dai and Lee 2001). method for landslide probability mapping (Fig. 5). The WoE
Lithological units are one of the more important contributing works by allocating weights to each class of the thematic
parameters for landslide distribution (Yalcin 2008). Loose and parameter, based on spatial association to that parameter and
fragile lithological units are more prone to landslides (Kamp identified landslide locations. Bayesian approach was
et al. 2008). Fault rupture and geometry have direct impact on employed in WoE to ascertain the impact of the spatial asso-
the occurrence of landslides (Mahmood et al. 2015). Road ciation between landslide locations and classes of evidential
construction activities, e.g., blasting, excavation, undercut- factors to compute weights.
ting, and heavy traffic vibrations are also responsible for land- To evaluate and investigate the association of each eviden-
slides (Mittal et al. 2008). According to Dahal et al. (2008), tial factor with landslides, a set of tables were generated by the
undercutting and toe erosion due to stream network are con- Calculate Weights tool in Arc-SDM (Table 1). The analysis of
sidered as significant contributing parameters for landslides. slope gradient evidential theme showed the slope class (61–
70°) has highest weight (2.336) and has positive correlation
Landslide susceptibility mapping with landslides, while the least weight value is generated by
slope class (0–10°) which is (− 2.861). These values indicate
The probability of occurring a landslide in the area based on the strong association between landslide points and slope clas-
local terrain environment fell under the definition of landslide ses. The analysis of weight tables for slope aspect indicates
susceptibility (Brabb 1984; Fell et al. 2008). Probability as- that higher weight value obtains by east-facing slope which is
sessment starts with valid landslide inventory, evidential pa- 0.705 followed by south-facing slope which is 0.522. The
rameters selection, and computation of appropriate approach minimum weight value is yielded by southwest-facing slope
to analyze the contribution of controlling factors on landslide (− 0.898). The weights for curvature causative factor revealed
distribution (Fell et al. 2008; Van Westen et al. 1997). Various that the maximum weight value was assigned to concave
Arab J Geosci (2021) 14:1019 Page 7 of 19 1019

Fig. 5 Developed landslide

susceptibility map of the area

curvature which is 0.220. While the lowest weight value is for which is − 3.258. We concluded from these results that drain-
convex curvature which is − 0.168. age is also another important causative factor for landsliding
The analysis of weights for elevation shows that maximum phenomenon. Distance to road is our last evidential theme and
weight value is 1.013 which is at elevation class 1074–1223 m classified into five classes using Euclidean distances. The re-
covered an area of 11.094 km2 having 13 landslide points, sults of weights table show that highest weight value is of
while the lowest weight value is − 2.538 at elevation ranges 0.705 which is the class of 0–50 m while the lowest weight
from 1524 to 1673 m. This analysis of geological units reveals value is − 0.302 which is for class 200–300 m indicating no
that highest weight value is of Panjal volcanics which is 1.145. positive association between training points and this class of
This reveals that this class is most susceptible to landsliding. evidential theme.
In contrast, lowest contrast value is of alluvial deposit which is In this study, all the causative factors along with their cal-
− 0.076. These results indicate that there is a strong associa- culated weights were combined in Arc-SDM tool to develop
tion between landslides and lithological units. The weight the susceptibility map. The developed probability map was
analysis of distance to fault causative factor indicate that max- then classified into low, moderate, high, and very high sus-
imum weight value was yielded by the class 0–50 m which is ceptible zone using cumulative area posterior probability
1.139, while lowest contrast value was obtained by the class curve (CAPP). The validation of the susceptibility map was
301–400 m which is − 1.299 having zero landslide points. carried out by SRC and PRC curves. The susceptibility map
These outcomes indicate that faults have great effect in trig- has 83% (SRC) and 79% (PRC) which is good enough for
gering landslides. 76.6 km2 of the area (Fig. 6). Overall, 44% of the area fall
The weight analysis of distance to drainage discloses that under low susceptible zone, 22% in moderate susceptible
class 0–50 m computed the highest weight value which is zone, 18% in high susceptible zone, while 16% of the area
2.866 while the lowest weight value is of the class > 300 m fall in very high susceptible zone.
1019 Page 8 of 19 Arab J Geosci (2021) 14:1019

Table 1 Weights of each factor estimated by weights-of-evidence considered in this study

Causative factors Classes Landslide points W+ W− Contrast (C) Weight Weight STD

Slope (degree) 0–10 1 − 2.861 0.052 − 2.914 − 2.861 1.004

11–20 2 − 1.590 0.149 − 1.739 − 1.590 0.306
21–30 7 − 0.443 0.186 − 0.538 − 0.443 0.435
31–40 9 − 0.841 0.204 − 1.219 − 0.841 0.349
41–50 10 − 0.897 0.175 − 0.933 − 0.875 0.779
51–60 11 − 0.443 0.186 − 0.538 − 0.443 0.435
61–70 12 2.336 0.184 2.530 2.336 0.794
Slope aspect North 5 0.132 0.180 − 0.340 − 0.302 0.525
Northeast 5 0.127 0.179 − 1.424 − 0.127 0.538
East 14 0.709 0.194 0.876 0.705 0.381
Southeast 10 0.474 0.187 0.560 0.474 0.427
South 6 0.522 0.179 0.574 0.522 0.557
Southwest 3 0.892 0.178 − 0.991 − 0.898 0.634
West 5 0.577 0.182 − 0.664 − 0.577 0.507
Northwest 4 0.498 0.179 − 0.553 − 0.498 0.572
Curvature Concave 25 0.220 − 0.168 0.389 0.220 0.256
Convex 27 − 0.168 0.220 − 0.389 − 0.168 0.230
Elevation (m) 923–1073 10 − 0.897 0.175 − 0.933 − 0.875 0.779
1074–1223 13 1.013 −.0.193 1.205 1.013 0.430
1224–1373 19 0.676 0.249 0.926 0.676 0.325
1374–1523 9 − 0.388 0.143 − 0.493 − 0.388 0.387
1524–1673 1 − 2.538 0.259 − 2.798 − 2.538 1.020
1674–1823 0 0 0 0 − 0.215 0.379
1824–1973 0 0 0 0 − 0.215 0.379
Above 1974 0 0 0 0 − 0.215 0.379
Lithology Tonal Formation 31 0.167 − 0.201 0.365 0.136 0.227
Neelum granite 7 − 0.767 0.196 − 0.958 − 0.762 0.420
Kundulshahi group 2 0.040 − 0.001 0.045 − 0.064 0.498
Glacial deposits 1 − 0.217 0.004 − 0.221 − 0.064 0.496
Alluvial deposit 1 − 0.258 0.005 − 0.258 − 0.076 0.489
Nauseri granite 5 0.408 − 0.035 0.443 0.488 0.595
Panjal volcanic 3 1.043 − 0.040 1.186 1.145 0.932
Murree Formation 2 0.032 − 0.009 0.024 − 0.046 0.496
Distance to faults (m) 0–50 4 1.139 − 0.026 1.153 1.139 1.141
51–100 3 0.558 − 0.258 0.813 0.558 0.794
101–200 2 0.040 − 0.001 0.045 − 0.064 0.498
201–300 1 − 0.851 0.026 1.117 − 0.851 1.103
301–400 0 0 0 0 − 1.299 1.067
401–500 1 0.600 0.016 1.144 − 0.600 1.131
Above 500 41 0.051 − 0.241 0.458 0.051 0.187
Distance to drainage (m) 0–50 21 2.866 − 0.157 2.380 2.866 1.792
50–100 12 1.319 − 0.198 1.581 1.319 0.492
100–200 10 − 0.206 0.056 − 0.262 − 0.206 0.376
200–300 8 − 0.256 0.054 − 0.310 − 2.256 0.414
Above 300 1 − 3.258 0.673 − 3.932 − 3.258 1.009
Distance to road (m) 50–100 14 0.705 − 0.170 0.876 0.705 0.538
100–200 5 − 0.127 0.014 − 0.142 − 0.127 0.525
200–300 5 − 0.302 0.383 0.340 − 0.302 0.381
300–400 10 0.476 − 0.086 0.560 0.474 0.427
Above 300 6 0.525 − 0.051 0.574 0.525 0.557

Field investigation, geotechnical mapping, and Shahkot landslide

construction of profiles
The Shahkot landslide is situated about 72 Km North of
The discrete geological units of the investigated landslides Muzaffarabad alongside Neelum road, district Neelum
have been recognized and mapped. The lithological units Valley, Azad Kashmir. According to the native people, the
and geological formations have also been presented on the Shahkot landslide was activated first time due to heavy flood
maps with bedding attitude where available. in river Neelum in 1992. The landslide area was composed of
Arab J Geosci (2021) 14:1019 Page 9 of 19 1019

Fig. 6 Landslide susceptibility

validation curves: a success rate
curve (SRC); b prediction rate
curve (PRC)

cultivated terraces with irrigation channels. These terraces and schistose conglomerate. The scarp on the right flank consists
irrigation channels were destroyed due to the landslide. of chlorite schist, graphitic schist, and phyllite of the Tanol
Consequently, the cultivation in adjacent area downstream Formation, while towards the middle and left flank, the de-
was also abandoned. About 490m road section has been dam- tachment surface consists of clays, sub-phyllite, and schist
aged by the landslide and caused disruption to the continuity debris (Fig. 8d). Thick vegetation cover is present above the
of the traffic along this road (Fig. 7a, b). The landslide occurs crown and along the flanks as well as within the slumped
within the meta-sediments of the Tanol Formation which material in the middle portion (Fig. 7a, b). Several seepages
comprises mainly of quartzose schist, garnetiferrous chlorite were also observed near the main scarp on the right flank and
mica schist, graphitic schist, quartzite bands, phyllite, and near the bottom of the main body on the left flank along the
1019 Page 10 of 19 Arab J Geosci (2021) 14:1019

Fig. 7 Field photographs: a

Shahkot landslide; b disturbed
road through landslide material; c
exposed scarp and landslide
material of the Sandok landslide;
d huge displaced material along
the Sandok landslide

road. The landside activity here relates to steepness of slope, The length of this zone is 100–180 m and the width is 350–
presence of swelling clays and non-cohesive soils, river under 365 m. It consists mainly of mixed debris material with some
cutting, and deforestation. boulders and slabs. The toe area on the right flank of the
The zone of depletion of landslide on the right flank shows landslide is categorized by debris flow with well-developed
rock and debris slide and toppling, while along the middle fan adjacent to the river. The length and width of the landslide
portion and left flank, it shows slump movement with slightly are 650 m and 350 m, respectively. The thickness of the ma-
curved surface. The measurement of the depletion zone of the terial is about 90 m and the estimated volume of the landslide
landslide is variable along the circumference of the main is 1.35 × 107 m3.
scarp. On the right flank, it is 15–190 m wide and 85–150 m
long, while on the left flank, it is about 10–190 m wide and
about 90 m long (Fig. 8c). The exposed area of the zone is Sandok landslide
32,300 m2. The sandy, silty, and clayey material is deposited
in the depletion zone with coarse, platy fragments of schist, The Sandok landslide is located approximately 65 Km north
and sub-phyllite. of the Muzaffarabad, alongside Neelum valley road, Azad
The transitional zone extends in length from 35–230 m and Kashmir. The elevation varies from 1291 m at bottom to
is 340–350 m in width (Fig. 8c). Area covered by the transi- 1391 m at the top. This landslide was activated in year 2013
tion zone of landslide is 32,880 m2. The slope angle varies due to heavy and consistent rain fall for many days and initi-
from 30° to 70° (Fig. 9a, c). The zone is considered as sandy ates with huge mass towards the Neelum river. Resultantly,
zone with abundant gravel and pebble fraction. The main the river was completely blocked for a while, and the water
body of the landslide consists of clays, sand, gravel, and cob- level rose to approach the nearby village of Islampura. The
ble fractions. The thick deposit of loose material along the blockage was then cleared through blasting. About 115m road
traveling path enlarged the volume of the landslide. In the section has been damaged by the landslide which disrupted
lower portion of the slide, the displaced material measures the continuity of traffic along this road (Fig. 7c, d).
about 10–12 m above the road. The thickness of debris mate- Considering the aforementioned problems, the landslide was
rial is more below the road due to the dumping of material studied and mapped on a scale 1:1000 to know the reasons and
while clearance of the road from landslide material. The veg- the failure mechanisms (Fig. 8a, b). The Sandok landslide
etation cover is also present within the transition zone towards occurs in the granite intruded into the Tanol Formation. The
the left flank where the landslide shows slump failure down- rocks exposed along landslide and adjacent areas are highly
slope. The accumulation zone covers an area of 46830 m2. jointed and cracked. The landslide material is displaced from
Arab J Geosci (2021) 14:1019 Page 11 of 19 1019

Fig. 8 a Morphological map of the Sandok landslide; b geotechnical map of the Sandok landslide; c morphological map of the Shahkot landslide; d
geotechnical map of the Shahkot landslide

the crown to toe. However, huge material is remained along 9b, d). The deposition zone of the Sandok landslide is 75–100
the sliding surface. m long and 100–115 m wide. The calculated surface area of
The depletion zone of the Sandok landslide is 50–80 m deposition zone is 9440 m2. The zone of accumulation of the
wide and 10–15 m long (Fig. 8a). The main scarp is nearly slide consists of large blocks and boulders of granite.
vertical and extends about 15 m from the top of the slide. The Evidences of blasting were observed which was done to clear
total area of depletion zone is about 1250 m2. The scarp seems the Neelum road which may have further accelerated the
to have been formed due to detachment of blocks of granite rockslide. The diameter of the boulders ranges up to 5 m in
along the nearly vertical jointed surface. The boulders of gran- diameter. Total volume is calculated about 0.11 million m3
ite are stacked along the slope which may slide or topple as the estimated by multiplying the average thickness with the cov-
landslide may reactivate. The rock mass in source area is ered area. The Neelum road runs through deposition zone of
highly jointed and fractured granite body. The vertical eleva- the landslide.
tion from crown to toe is about 125 m. The absolute horizontal
distance is calculated bout 150 m. The accumulation of large Laboratory tests
boulders (several meters in diameter) with some unconsolidat-
ed material along down slope has increased the volume of Grain size distribution
landslide (Fig. 7c). The transition zone of the Sandok land-
slide consists mainly of overhanging boulders of granite. The The representative soil sample was obtained from the bulk soil
transitional zone extends from 15–45 m in length and 70–95 sample by reducing the sample through quartering. The soil
m in width (Fig. 8a). Area covered by the transition zone of sample was then placed in the sieve shaker in ascending order.
landslide is 3060 m2. Average thickness is measured about 6– The mass of the material retained on each sieve was recorded.
8 m. The slope angle in this zone ranges from 40 to 70° (Fig. The quantity of the soil is based on the maximum particle size
1019 Page 12 of 19 Arab J Geosci (2021) 14:1019

Fig. 9 Landslide profiles: a longitudinal profile of the Shahkot landslide; b longitudinal profile of the Sandok landslide; c cross profile of the Shahkot
landslide; d cross profile of the Sandok landslide

present in the soil. Retained percentage of soil samples of the which holds the major failure along the slope due to the water
Shahkot landslide from the depletion zone on sieve no. 4 is seepages within the landslide body.
38.2–54.6%, on sieve no. 50 is 89.9–90.6%, on sieve no. 100 Retained percentage of soil sample of the Sandok landslide
is 95.9–96.2%, and on sieve no. 200 is 97.8–98.7%. Based on from accumulation zone at right flank on sieve no. 4 is 12.2%,
these results, the soil from the accumulation zone at right flank on sieve no. 50 is 72.4%, on sieve no. 100 is 85.4%, and on
is classified as silty sand (non-cohesive). Retained percentage sieve no. 200 is 93.6%. Based on these results, the soil from
of soil from the accumulation zone at right flank and left flank accumulation zone at right flank is classified as clayey silty
on sieve no. 4 is 33–36%, on sieve no. 50 is 79.6–89%, on sand (non-cohesive). Retained percentage of soil of accumu-
sieve no. 100 is 85–93.4%, and on sieve no. 200 is 93–97%. lation zone at left flank on sieve no. 4 is 38.3%, on sieve no.
Based on these results, the soil from accumulation zone at 50 is 78.5%, on sieve no. 100 is 86.9%, and on sieve no. 200 is
right and left flank is classified as clayey silty sand (non-co- 91.4%. On the basis of these results, the soil from accumula-
hesive). Retained percentage of soil from accumulation zone tion zone at left flank is classified as silty clayey sand (non-
at middle portion on sieve no. 4 is 44.3%, on sieve no. 50 is cohesive). Based on grain size distribution analysis, it is clas-
88.3%, on sieve no. 100 is 96.2%, and on sieve no. 200 is sified that the soils of transitional and accumulation zones are
97.8%. Based on these results, the soil from accumulation silty clayey sands.
zone at middle is classified as sandy silt (non-plastic).
Retained percentage of soil of the toe area of accumulation Atterberg limits and plasticity index of soils
zone on sieve no. 4 is 46.7–48.5%, on sieve no. 50 is 85.7–
88.8%, on sieve no. 100 is 91.3–94.4%, and on sieve no. 200 The liquid limit (LL), plastic limit (PL), and plasticity index
is 96–97.6%. Based on these results, the soil from depletion (PI) of soils are also used extensively, either individually or
zone is classified as clayey sandy silt (non-cohesive). Based together, with other soil properties to correlate with engineer-
on grain size distribution analysis, it is classified that the soils ing behavior such as compressibility, hydraulic conductivity
of transitional and accumulation zones are silty clayey sands (permeability), compactibility, shrink-swell, and shear
Arab J Geosci (2021) 14:1019 Page 13 of 19 1019

strength. Fine soils (i.e., finer than 425 μm) are tested accord- relates to the fact that the rock is highly fractured as observed
ing to the ASTM standard D 4318–00. LL of Shahkot land- in field as well as in petrography. The intensive fracturing of
slide soil samples ranges from 37.05 to 23.9%, PL ranges the rock unit is attributed to the presence of localized thrust
from 28.57 to 21.75%, and PI ranges from 14.96 to 2.15% faults in the vicinity of the landslide. The shearing strength of
that indicates the landslide material has high swelling potential granite was overcome by the overburden pressure which
(Table 2), whereas the LL of Sandok landslide soil samples causes the major failure along the slope. The very low UCS
ranges from 38.09 to 27.9%, PL ranges from 33.06 to 24.78%, values of the Tanol schist in the Shahkot landslide correspond
and PI ranges 6.48 to 2.74 which indicates that the finer land- to the presence of flaky minerals like chlorite, micas, and other
slide material has slightly plastic behavior (Table 2). clay minerals observed in XRD and petrography which reduce
the strength of the rock unit. Moreover, UCS was performed
on the core samples oriented parallel to the shistosity planes
Specific gravity
due the fact that the direction of landslide failure plane also
corresponds to the shistosity planes (Fig. 10).
To understand the physical characteristics of the landslide
material, samples for specific gravity test were collected from
top, middle, and base which were tested according to ASTM Petrography of exposed rocks in landslides
standard test method D854-14 and test results are presented in
(Table 2). The specific gravity of the soil samples taken from Tanol Formation at Shahkot landslide
the Shahkot landslide ranges from 2.72 to 2.33. The specific
gravity of the soil samples taken from the Sandok landslide Three samples from the Shahkot landslide were collected for the
ranges from 2.71 to 2.53. These results indicate that the land- petrographic study (SH-1, 2, and 3) from the Tanol formation
slide body contains coarse grain materials. exposed along the right flank of the landslide. In hand specimen,
the fresh color of schist is greenish grey to light grey except few
dark grey graphitic bands at places. The rocks are soft, fine
Unconfined compressive strength (UCS) of rock cores
grained, easily scratched with knife, and have low shistosity.
Petrographic studies show that it contains subhedral to euhedral
UCS test was performed using ASTM D7012-14e1. A total of
quartz with fair amount of flaky minerals like chlorite, biotite,
12 samples were collected from the top, middle, and base of
and muscovite. Minor amount of garnet and other opaque min-
both the landslides. Three samples were collected from
erals is also present. Mineralogically, the schist is composed of
garnetiferrous chlorite mica schist of the Tanol Formation in
quartz 40–50%, muscovite 15–20%, chlorite 10–15%, opaque
Shahkot landslide and 9 samples were collected from the Jura
minerals 5–10%, and clay minerals 5–10% (Fig. 11a).
granite in Sandok landslide given in (Table 3). Shahkot land-
slide sample UCS test values range from 16.05 to 26.45 MPa
which indicate very low compressive strength. The Sandok Granite at Sandok landslide
landslide sample UCS tests values range from 36.22 to
50.80 MPa which also indicate low compressive strength. Three samples were collected from the Sandok landslide for
The low values of the UCS in granite of the Sandok landslide the petrographic study (SD-1, 2, and 3) from the Jura granite.

Table 2 Summary of
geotechnical characteristics of No. Sample ID. Liquid limit Plastic limit Plasticity index Specific gravity
soil samples from Sandok and
Shahkot landslides Shahkot landslide
1 SHKT-MS-(RF) - - - 2.33
2 SHKT-MS-(M) - - - 2.33
3 SHKT-MS-(LF) - - - 2.59
4 SHKT-MB-(RF) - - - 2.44
5 SHKT-MB-(M) 37.05 22.09 14.96 2.66
6 SHKT MB-(LF) - - - 2.45
7 SHKT-TOE-(RF) 31.3 28.57 2.73 2.67
8 SHKT-TOE-(M) 26 23.95 2.05 2.72
9 SHKT-TOE-(LF) 23.9 21.75 2.15 2.45
Sandok landslide
10 SDK-MB-(RF) - - - 2.71
11 SDK-MB-(M) 38 31.52 6.48 2.53
1019 Page 14 of 19 Arab J Geosci (2021) 14:1019

Table 3 Uniaxial compressive

strength test results on rock cores Sample Dia of sample Length of sample Area Failure load Stress rate (kN/ UCS
from Shahkot and Sandok ID (mm) (Mm) (mm2) (kN) min) (MPa)
landslides
STS-1 47 94 1734.065 38.41 4.0 22.15
STS-2 47 94 1734.065 45.87 4.0 26.45
STS-3 47 94 1734.065 27.83 4.0 16.05
SJG-1 47 94 1734.065 71.50 4.0 41.23
SJG-2 47 94 1734.065 80.37 4.0 46.35
SJG-3 47 94 1734.065 78.40 4.0 45.21
SJG-4 47 94 1734.065 88.09 4.0 50.8
SJG-5 47 94 1734.065 80.98 4.0 46.7
SJG-6 47 94 1734.065 77.46 4.0 44.67
SJG-7 47 94 1734.065 87.62 4.0 50.53
SJG-8 47 94 1734.065 62.81 4.0 36.22
SJG-9 47 94 1734.065 67.98 4.0 39.2

STS stands for Shahkot Tanol schist

SJG stands for Sandok Jura granite

In hand specimen, the fresh surface of granite is light grey or indicate that the main constituted minerals present are quartz,
whitish grey, while on the weathered surface, it appears yel- muscovite, kaolinite, montmorillonite, illite, goethite, hema-
lowish grey or brownish black. Apart from feldspar, the main tite, aragonite, siderite, calcite, dolomite, orthoclase, plagio-
minerals seen in the hand specimen include muscovite, biotite, clase, and gypsum. The results showed the presence of con-
and quartz with minor amount of tourmaline and hornblende. siderable amount of clay minerals montmorillonite, illite, and
Mineralogically, the granite is composed of quartz 40–55%, kaolinite. These minerals, specially montmorillonite, have
plagioclase 10–20%, potash feldspar 10–20%, muscovite 3– greater swelling potential and increase the pore water pressure
6%, biotite 2–4%, sericite 2–4%, opaque minerals 1–3%, and which is one of the causes of slope failure. The minerals like
tourmaline 0–1% (Fig. 11b–d). orthoclase and plagioclase are high-temperature minerals
which are being chemically unstable and are more prone to
Clay mineralogy of sliding surface material chemical alteration. The other minerals like aragonite, calcite,
and hematite also have greater affinity to weathering.
XRD technique was utilized in this study to establish bulk
mineralogy of shale and clay of the Tanol Formation in the
Shahkot landslide. The observed crystal lattice distance (d- Discussion
spacing) in clays is correlated with joint committee on powder
diffraction standard (JCPDS) for identification of different The presence of different types of landslides in the study area
minerals. The results of XRD pattern presented in Fig. 12 is because of the seismicity, intense rainfall, fragile nature of

Fig. 10 a Failure mode observed

on Tanol schist, failure along
foliation planes; b Jura granite
specimen under UCS; axial
splitting and multiple fractured
observed
Arab J Geosci (2021) 14:1019 Page 15 of 19 1019

Fig. 11 a Photomicrograph showing schist of Tanol Formation exposed showing Jura granite exposed in Sandok landslide showing fractured
in Shahkot landslide; Q quartz, MUC muscovite, BIO biotite, CL feldspar crystal; FR fracture, SRT siricite (10x:xl); d photomicrograph
chloride, OP opaque mineral (4x:xl); b photomicrograph showing Jura showing Jura granite exposed in Sandok landslide showing multiple
granite exposed in Sandok landslide. Q quartz, Pl plagioclase, AP altered fractures in quartz crystals; FR fracture (10x:xl)
plagioclase, MU muscovite, BIO biotite (4x:xl); c photomicrograph

outcropping rocks, steep slopes, and narrow valley (Riaz et al. A detailed field survey was carried out for two large-scale
2019). Multiple field visits found that about 74 landslides landslides. Detailed landslide mapping, profile drawings, and
having different types occurred in the study area. Based on laboratory analysis such as geotechnical analysis, sieve anal-
landslide susceptibility results, a problematic zone with a high ysis, Atterberg limits, specific gravity, UCS, and XRD analy-
frequency of landslide was selected for a more detailed inves- sis concluded that the Shahkot landslide is a rock and debris
tigation. Data-driven WoE method for landslide susceptibility slide with slump failure towards the left flank. The sieve anal-
analysis has been applied in landslide prone area, Lesser ysis, Atterberg limits, and specific gravity of disturbed soil
Himalayas, Pakistan. Contrary to the other studies, e.g., Lee sample suggest that the soils range from sandy, silty, and clay
and Choi (2004); Dahal et al. 2008) and Pradhan et al. (2010), to gravel fraction which show cohesive, non-plastic to non-
weights were computed via Arc-SDM tool. Landslide suscep- cohesive behavior. The PI indicates the occurrence of organic
tibility map integrating eight causative factors reveals that matter in the soil which has less impact on slope failure during
drainage network, faults, road network, and steep riverbed rain fall or seepages. The UCS test values indicate high com-
slopes are more prone for landslides in the studied area. pressive strength of the material. The XRD analysis shows
There are various factors that donate to the landslide suscep- mineral composition of quartz, muscovite, kaolinite, goethite,
tibility, e.g., lithological units, slope gradient, slope aspect, terrain aragonite, hematite, plagioclase, siderite, montmorillonite,
elevation, landcover, discontinuity, and the location of roads and calcite, gypsum, orthoclase, dolomite, and illite. Mechanical
rivers. However, these causative factors vary from area to area, deformation is enhanced with richness of kaolinite, smectite,
their effect is, to some extent, known, and have not revealed a and illite having strong water absorption power. This study
great difference apart from lithology (El Jazouli et al. 2020). revealed that shear strength decreased with increasing
1019 Page 16 of 19 Arab J Geosci (2021) 14:1019

Fig. 12 XRD analysis of Tanol

Formation exposed in Shahkot
landslide

moisture content, and hence, number of landslides increased abundant moisture provided by drainage network and rainwa-
relative to the particle size distribution of the clay. The Sandok ter percolations. Furthermore, undercutting by the river as
landslide is a rockslide resulted by the failure of highly jointed well as anthropogenic activities along the steep slopes also
and fractured granite. The sieve analysis, Atterberg limits, and increases the probability of slope failures.
specific gravity of disturbed soil sample suggest that the soil is
non-cohesive with mainly clayey silty sand and gravels which
is also supported by the XRD analysis showing higher con- Conclusion
centration of clay mineral such as illite, which has higher
swelling potential. The UCS test results show low compres- Landslide inventory mapping and classification were carried
sive strength of granite, possibly reduced by previous seismic out, and subsequently, landslide susceptibility map was devel-
activities, presence of quartz filled veins, weathering effects, oped to document the most probable triggering factors along
fracturing, and jointing in granite body experiencing freeze the studied road section. The Shahkot and the Sandok land-
and thaw repeated episodes. Combining all analyses reveals slides in Athmuqam area along Neelum valley road were iden-
that the Sandok granite is highly jointed and sheared which tified, first time documented, and classified as a debris slide-
leads to the conclusion that this landslide occurred mainly due slump failure and rockslide, respectively. The possible causes
to fracture/joint failure in bed rock. Besides that, after the of these landsides include steepness of slope, drainage net-
initial landslide episode, blasting method was applied to clear work, existence of swelling clays and non-cohesive soils, in-
the road section which contributed in further fracturing of the tensive jointing and fracturing, repeated freeze and thaw epi-
granite body. The weight results of susceptibility mapping sodes, seismic activities, river under cutting during flooding,
revealed that steep slopes, stream networks, lithological vari- and deforestation.
ations, and fault networks are the more influential factors for The sieve analysis, Atterberg limits, specific gravity, and
slope failure in the region. Geotechnical and geochemical XRD analysis of the disturbed soil sample and petrographic
analysis also support these results as weathered and crushed analysis show higher concentration of clay mineral such as
rocks having close proximity to faults revealed by low UCS, montmorillonite, illite, and kaolinite which possess high
presence of clay minerals having swelling potential, and swelling potential, show non-plastic to non-cohesive behavior
Arab J Geosci (2021) 14:1019 Page 17 of 19 1019

in case of Shahkot landslide. Very low UCS values of schist in Basharat M, Rohn J, Baig MS, Ehret D (2012) Lithological and structural
control of Hattian Bala rock avalanche triggered by the Kashmir
Shahkot landslide indicate presence of flaky minerals and
earthquake 2005, NW Himalaya, Pakistan. J Earth Sci 23(2):213–
schistosity in relation to the failure planes. The UCS test re- 224
sults of the granite of Sandok landslide suggests the possible Basharat M, Rohn J, Baig MS, Khan MR, Schleier M (2014) Large scale
strength reduction by the presence of megascopic and micro- landslides triggered by the Kashmir Earthquake 2005. Pakistan J
Mountain Science 11(1):19–30
scopic fractures and joints produced as result of intense shear-
Basharat M, Shah HR, Hameed N (2016) Landslide susceptibility map-
ing along localized thrust faults and folds. Presence of calcite ping using GIS and weighted overlay method: a case study from
and quartz veins and repeated episodes of freeze and thaw NW Himalayas, Pakistan. Arab J Geosci 9(4):1–19
action contributed to weathering and erosion. Gentle joint Basharat M, Sarfraz Y, Ahmed KS et al (2017) A Preliminary investiga-
tion of reactivated mass movement near the epicenter of 2005
dip and open fractures are found in the scarp of the Sandok
Kashmir earthquake, NW Himalayas, Pakistan. Journal of
landslide and dip gently parallel to the escarpment slope. The Himalayan Earth Sciences 50(1A):57–65
landslide is probably advanced by stress release, freeze thaw, Brabb EE (1984) Innovative approaches to landslide hazard mapping. In
variations in moisture content, and creep movement. The dis- Proc., 4th Int. Symp. On Landslides, 307–324. Toronoto
Bragagnolo L, Silva RVD, Grzybowski JMV (2020) Artificial neural
integration affects the surface rocks to some extent, so that
network ensembles applied to the mapping of landslide susceptibil-
permeability and strength influenced badly at the weathered ity. Catena 184:104240. https://doi.org/10.1016/j.catena.2019.
zone. The weaker zone shows the development of fracturing 104240
along the slope and finally slides due to heavy rainfall when it Buša J, Tornyai R, Bednarik M et al (2019) Hodnotenie zosuvného
is immediately saturated or due to seismicity. Alternatively, hazardu pomocou multivariačnej a bivariačnej štatistickej analýzy
v Košickej kotline (Západné Karpaty). Geografický časopis 71:383–
these findings highlight the worth potential of landslide sus- 405 (in Slovakm with English abstract). https://www.sav.sk/index.
ceptibility analysis to identify the hazardous zones and geo- php?lang=en&doc=journallist&part=article_response_
technical investigations and geochemical analysis to gain in- age&journal_article_no=17766
sight into the mineral compositions, rock strength, and mois- Calkins JA, Offield TW, Abdullah SKM et al (1975) Geology of the
southern Himalaya in Hazara, Pakistan, and adjacent areas. US
ture conditions which assumed the root cause of the landslid- Geol Surv Prof Pap:716–c 29
ing phenomenon. Can A, Dagdelenler G, Ercanoglu M, Sonmez H (2019) Landslide sus-
ceptibility mapping at Ovacık-Karabük (Turkey) using different ar-
tificial neural network models: comparison of training algorithms.
Acknowledgements The authors are grateful to the Director Institute of Bull Eng Geol Environ 78(89):102–102. https://doi.org/10.1007/
Geology for providing transportation facilities during field work and Mr. s10064-017-1034-3
Mudassir Shakeel for assisting in lab. work. Authors also express their Chini M, Cinti FR, Stramondo S (2011) Co-seismic surface effects from
sincere acknowledgements to Biswajeet Pradhan,Editor of the Arabian very high-resolution panchromatic images: the case of the 2005
Journal of Geosciences andthe two anonymous reviewers. Kashmir (Pakistan) earthquake. Nat Hazards Earth Syst Sci 11:
931–943
Dahal RK, Hasegawa S, Nonomura A, Yamanaka M, Masuda T, Nishino
Declarations K (2008) GIS based weights-of-evidence modelling of rainfall-
induced landslides in small catchments for landslide susceptibility
Conflict of interest The authors declare that they have no competing mapping. Environ Geol 54(2):311–324. https://doi.org/10.1007/
interests. s00254-007-0818-3
Dai FC, Lee CF (2001) Terrain-based mapping of landslide susceptibility
using a geographical information system: a case study. Can Geotech
J 38(5):911–923. https://doi.org/10.1139/t01-021
References Dunning SA, Mitchell WA, Rosser NJ, Petley DN (2007) The Hattian
Bala rock avalanche and associated landslides triggered by the
Kashmir earthquake of 8 October 2005. Eng Geol 93(3–4):130–
Akgun A (2012) A comparison of landslide susceptibility maps produced 144. https://doi.org/10.1016/j.enggeo.2007.07.003
by logistic regression, multi-criteria decision, and likelihood ratio El Jazouli A, Barakat A, Khellouk R (2020) Geotechnical studies for
methods: a case study at İzmir, Turkey. Landslides 9(1):93–106 landslide susceptibility in the high basin of the Oum Er Rbia river
Aleotti P, Chowdhury R (1999) Landslide hazard assessment: summary (Morocco). Geology, Ecology, and Landscapes:1–8
review and new perspectives. Bull Eng Geol Environ 58(1):21–44. Feizizadeh B, Roodposhti MS, Blaschke T, Aryal J (2017) Comparing
https://doi.org/10.1007/s100640050066 GIS-based support vector machine kernel functions for landslide
Ali SA, Parvin F, Vojteková J et al (2020) GIS-based landslide suscepti- susceptibility mapping. Arab J Geosci 10:122. https://doi.org/10.
bility modeling: a comparison between fuzzy multi-criteria and ma- 1007/s12517-017-2918-z
chine learning algorithms. Geoscience Frontiers Fell R, Corominas J, Bonnard C, Cascini L, Leroi E, Savage WZ (2008)
Al-Najjar HH, Pradhan B (2020) Spatial landslide susceptibility assess- Guidelines for landslide susceptibility, hazard and risk zoning for
ment using machine learning techniques assisted by additional data land use planning. Eng Geol 102(34):85–98. https://doi.org/10.
created with generative adversarial networks. Geosci Front 12(2): 1016/j.enggeo.2008.03.022
625–637 Ghazanfar M, Baig MS, Chaudhry MN (1983) Geology of Tithwal Kel
Basharat M (2012) The distribution, characteristics and behavior of land- area Neelum Valley, Azad Jammu and Kashmir. Kashmir J Geol 1:
slides triggered by the Kashmir Earthquake 2005, NW Himalaya, 1–10
Pakistan, PhD thesis, University of Erlangen-Nuremberg, Germany, Guzzetti F, Reichenbach P, Cardinali M, Galli M, Ardizzone F (2005)
186 Probabilistic landslide hazard assessment at the basin scale.
1019 Page 18 of 19 Arab J Geosci (2021) 14:1019

Geomorphology 72(1–4):272–299. https://doi.org/10.1016/j. earthquake. Disaster Mitigation of Rock Flows, Slope failures and
geomorph.2005.06.002 Landslides by Universal Academy Press:1–9
Jadoon IAK, Hinderer M, Kausar AB, Qureshi AA, Baig MS, Basharat Planning and Development Department AJK (2015) Geological mapping
M, Frisch W (2015) Structural interpretation and geo-hazard assess- of Azad Jammu and Kashmir. No paging.
ment of a locking line: 2005 Kashmir Earthquake, western Pradhan B, Lee S (2010) Delineation of landslide hazard areas on Penang
Himalayas. Environ Earth Sci 73:7587–7602. https://doi.org/10. Island, Malaysia, by using frequency ratio, logistic regression, and
1007/s12665-0143929-7 artificial neural network models. Environ Earth Sci 60:1037–1054
Kalantar B, Pradhan B, Amir Naghibi S et al (2018) Assessment of the Pradhan B, Oh JH, Buchroithner M (2010) Use of remote sensing data
effects of training data selection on the landslide susceptibility map- and GIS to produce a landslide susceptibility map of a landslide
ping: a comparison between support vector machine (SVM), logistic prone area using a weight of evidence model. Remote Sensing
regression (LR) and artificial neural networks (ANN). Geomatics, Science Center for Cultural Heritage:395–402
Natural Hazards and Risk 9:49–69. https://doi.org/10.1080/ Riaz MT, Basharat M, Hameed N, Shafique M, Luo J (2018) A data-
19475705.2017.1407368 driven approach to landslide-susceptibility mapping in mountainous
Kamp UB, Growley J, Khattak GA, Owen LA (2008) GIS-based land- terrain: case study from the Northwest Himalayas, Pakistan. Natural
slide susceptibility mapping for the 2005 Kashmir earthquake re- Hazards Review 19(4):05018007
gion. Geomorphol. 101(4):631–642 Riaz S, Wang G, Basharat M, Takara K (2019) Experimental investiga-
Kavzoglu T, Colkese I, Sahin EK (2019) Machine learning techniques in tion of a catastrophic landslide in northern Pakistan. Landslides
landslide susceptibility mapping: a survey and a case study. In: 16(10):2017–2032
Pradhan S, Vishal V, Singh T (eds) Landslides: theory, practice Rieux SK, Qureshi RA, Peduzzi P et al (2007) An interdisciplinary ap-
and modelling. Advances in Natural and Technological Hazards proach to understanding landslides and risk management: a case
Research, vol 50. Springer, Cham, pp 283–301. https://doi.org/10. study from earthquake affected Kashmir. In Proc., Mountain
1007/978-3-319-77377-3_13 Forum, Mountain GIS Conf
Kazmi AH, Jan MQ (1997) Geology and tectonics of Pakistan, 557. Saba SB, Meijde VDM, Werff HVD (2010) Spatiotemporal landslide
Graphic Publishers, Karachi, Pakistan detection for the 2005 Kashmir earthquake region.
Khan MS (1994) Petrology and geochemistry of the Panjal volcanic in the Geomorphology 124(1–2):17–25. https://doi.org/10.1016/j.
Azad Kashmir and Kaghan areas of the NW Himalaya. Ph.D. dis- geomorph.2010.07.026
sertation, Univ. of Punjab
Sato PH, Hasegawa H, Fujiwara S et al (2007) Interpretation of landslide
Khan KP, Mohanty S, Mohanty M (2010) Geodynamics response of 8
distribution triggered by the 2005 northern Pakistan earthquake
October 2005 North Pakistan earthquake. Geophys Surv 31(1):85–
using SPOT 5 imagery. Landslides 4(2):113–122. https://doi.org/
106. https://doi.org/10.1007/s10712-009-9083-1
10.1007/s10346-0060069-5
Kiyota T, Sattar A, Konagai K, Kazmi ZA, Okuno D, Ikeda T (2011)
Sattar A, Konagai K, Kiyota T, Ikeda T, Johansson J (2011)
Breaching failure of a huge landslide dam formed by the 2005
Measurement of debris mass changes and assessment of the dam-
Kashmir earthquake. Soils Found 51(6):1179–1190
break flood potential of earthquake-triggered Hattian landslide dam.
Konagai K, Sattar A (2012) Partial breaching of Hattian Bala landslide
Landslides 8(2):171–182
dam formed in the 8th October 2005 Kashmir earthquake, Pakistan.
Landslides 9(1):1–11 Senouci R, Taibi NE, Teodoro AC, Duarte L, Mansour H, Yahia Meddah
Kumar KV, Martha TR, Roy PS (2006) Mapping damage in the Jammu R (2021) GIS-based expert knowledge for landslide susceptibility
and Kashmir caused by 8 October 2005 Mw 7.3 earthquake from the mapping (LSM): case of Mostaganem Coast District, West of
Cartosat-1 and Resourcesat-1 imagery. Int J Remote Sens 27(20): Algeria. Sustainability 13(2):630
4449–4459. https://doi.org/10.1080/01431160600702376 Shafique M, Meijde MVD, Khan MA (2016) A review of the 2005
Lee S, Choi J (2004) Landslide susceptibility mapping using GIS and the Kashmir earthquake-induced landslides; from remote sensing per-
weight of evidence model. Int J Geogr Inf Sci 18(8):789–814. spective. J Asian Earth Sci 118(Mar):68–80. https://doi.org/10.
https://doi.org/10.1080/13658810410001702003 1016/j.jseaes.2016.01.002
Lee S, Lee MJ, Jung HS (2017) Data mining approaches for landslide Soeters R, Van Westen CJ (1996) In: Turner AK, Schuster RL (eds)
susceptibility mapping in Umyeonsan, Seoul, South Korea. Appl Slope instability recognition, analysis, and zonation. In
Sci 7(7):683. https://doi.org/10.3390/app7070683 Landslides-investigation and mitigation. National Research
Leonardi G, Palamara R, Cirianni F (2016) Landslide susceptibility map- Council, Washington, DC, pp 129–177
ping using a fuzzy approach. Procedia Eng 161:380–387. https:// Sujatha ER, Kumarave P, Rajamanickam V (2012) Landslide suscepti-
doi.org/10.1016/j.proeng.2016.08.578 bility mapping using remotely sensed data through conditional prob-
Mahmood I, Qureshi SN, Tariq S, Atique L, Iqbal MF (2015) Analysis of ability analysis using seed cell and point sampling techniques. J Ind
landslides triggered by October 2005, Kashmir Earthquake. PLoS Soc Remote Sens 40(4):669–678
Curr Disasters:1–7. https://doi.org/10.1371/currents.dis. Van Westen CJ, Rengers N, Terlien MT, Soeters R (1997) Prediction of
0bc3ebc5b8adf5c7fe9fd3d702d44a99 the occurrence of slope instability phenomena through GIS-based
Mittal S, Singh M, Kapur P et al (2008) Design and development of hazard zonation. Geol Rundsch 86(2):404–414. https://doi.org/10.
instrumentation network for landslide monitoring and issue an early 1007/s005310050149
warning. J Sci Ind Res 67(5):361 Varnes DJ (1978) Slope movement type and processes. In: Schuster RL,
Nefeslioglu HA, Gokceoglu C, Sonmez H (2008) An assessment on the Krizek RJ (eds) Landslides: analysis and control. National Academy
use of logistic regression and artificial neural networks with different of Sciences, vol 76. Transportation Research Board Special report,
sampling strategies for the preparation of landslide susceptibility pp 12–33
maps. Eng Geol 97(3):171–191. https://doi.org/10.1016/j.enggeo. Vijith H, Madhu G (2008) Estimating potential landslide sites of an up-
2008.01.004 land sub-watershed in Western Ghat’s of Kerala (India) through
Owen LA, Kamp U, Khattak GA, Harp EL, Keefer DK, Bauer MA frequency ratio and GIS. Environ Geol 55(7):1397–1405. https://
(2008) Landslides triggered by the October 8, 2005, Kashmir earth- doi.org/10.1007/s00254-007-1090-2
quake. Geomorphology 94:1–9 Vojteková J, Vojtek M (2020) Assessment of landslide susceptibility at a
Petley D, Dunning S, Rosser N, Kausar AB (2006) Incipient landslides in local spatial scale applying the multi-criteria analysis and GIS: a
the Jhelum Valley, Pakistan following the 8th October 2005 case study from Slovakia. Geom Nat Hazards Risk 11(1):131–148
Arab J Geosci (2021) 14:1019 Page 19 of 19 1019

Wadia DN (1931) The syntaxis of the north–west Himalaya-its rocks, Yilmaz I, Ercanoglu M (2019) Landslide inventory, sampling and effect
tectonics, and orogeny. Geol., Surv. India 65:189–220 of sampling strategies on landslide susceptibility/hazard modelling
Yalcin A (2008) GIS-based landslide susceptibility mapping using ana- at a glance. In: Pourghasemi, H., Rossi, M. (Eds.), Natural hazards
lytical hierarchy process and bivariate statistics in Ardesen (Turkey): GIS-based spatial modeling using data mining techniques.
comparison of results and confirmations. Catena 72(1):1–12. https:// Advances in Natural and Technological Hazards Research, vol 48.
doi.org/10.1016/j.catena.2007.01.003 Springer, Cham. https://doi.org/10.1007/978-3-319-73383-8_9
Yılmaz I (2009) Landslide susceptibility mapping using frequency ratio, Zhang TY, Han L, Zhang H, Zhao YH, Li XA, Zhao L (2019) GIS-based
logistic regression, artificial neural networks and their comparison: a landslide susceptibility mapping using hybrid integration ap-
case study from Kat landslides (Tokat-Turkey). Comput Geosci proaches of fractal dimension with index of entropy and support
35(6):1125–1138 vector machine. J Mt Sci 16:1275–1288. https://doi.org/10.1007/
s11629-018-5337-z