Article

Monitoring and Mapping of Shallow Landslides in a Tropical

Environment Using Persistent Scatterer Interferometry: A Case
Study from the Western Ghats, India
Ambujendran Rajaneesh 1 , Natarajan Logesh 1,2 , Chakrapani Lekha Vishnu 1,3 , El Hachemi Bouali 4 ,
Thomas Oommen 3 , Vinayan Midhuna 5 and Kochappi Sathyan Sajinkumar 1,3, *

1 Department of Geology, University of Kerala, Thiruvananthapuram 695581, India;

rajaneesh@keralauniversity.ac.in (A.R.); logeshw77@gmail.com (N.L.); vchakrap@mtu.edu (C.L.V.)
2 Centre for Remote Sensing, Bharathidasan University, Tiruchirapalli 620023, India
3 Department of Geological & Mining Engineering & Sciences, Michigan Technological University, Houghton,
MI 49931, USA; toommen@mtu.edu
4 Department of Physical and Life Sciences, Nevada State College, Henderson, NV 89002, USA;
elhachemi.bouali@nsc.edu
5 Department of Applied Geology, The Gandhigram Rural Institute, Gandhigram 624302, India;
midhunavinayan123@gmail.com
* Correspondence: sajinks@keralauniversity.ac.in; Tel.: +91-9495-83-2220

Abstract: Persistent Scatterer Interferometry (PSI) techniques are now well established and accepted
for monitoring ground displacements. The presence of shallow-seated landslides, ubiquitous phe-
nomena in the tropics, offers an opportunity to monitor and map these hazards using PSI at the
regional scale. Thus, the Western Ghats of India, experiencing a tropical climate and in a topo-
graphically complex region of the world, provides an ideal study site to test the efficacy of landslide





 detection with PSI. The biggest challenge in using the PSI technique in tropical regions is the ad-
Citation: Rajaneesh, A.; Logesh, N.; ditional noise in data due to vegetation. In this study, we filtered these noises by utilizing the
Vishnu, C.L.; Bouali, E.H.; Oommen, 95-percentile of the highest coherence data, which also reduced the redundancy of the PSI points.
T.; Midhuna, V.; Sajinkumar, K.S. The study examined 12 landslides that occurred within one of the three temporal categories grouped
Monitoring and Mapping of Shallow as Group 1, Group 2, and Group 3, categorized in relation to PSI monitoring periods, which was
Landslides in a Tropical Environment also further classified into east- and west-facing landslides. The Synthetic Aperture Radar (SAR)
Using Persistent Scatterer data is in descending mode, and, therefore, the east-facing landslides are characterized by positive
Interferometry: A Case Study from deformation velocity values, whereas the west-facing landslides have negative deformation values.
the Western Ghats, India. Geomatics
Further, the landslide-prone areas, delineated using the conventional factor of safety (FS), were
2021, 1, 3–17. https://dx.doi.org/
refined and mapped using PSI velocity values. The combination of PSI with the conventional FS
10.3390/geomatics1010002
approach helped to identify exclusive zones prone to landslides. The main aim of such an attempt is
Received: 27 November 2020
to identify critical areas in the unstable category in the map prepared using FS and prioritizing the
Accepted: 20 December 2020 mitigation measures, and to develop a road map for any developmental activities. The approach also
Published: 29 December 2020 helps to increase confidence in the susceptibility mapping and reduce false alarms.

Publisher’s Note: MDPI stays neu- Keywords: persistent scatterer interferometry (PSI); landslides; Radar; Sentinel-1A; Western Ghats
tral with regard to jurisdictional claims
in published maps and institutional
affiliations.
1. Introduction
Landslides are natural hazards that pose severe social and economic consequences
Copyright: © 2020 by the authors. Li-
because they can cause widespread damage to life and property [1]. Landslide triggering
censee MDPI, Basel, Switzerland. This mechanisms fall into three critical categories [2]: geological (e.g., seismic activity from
article is an open access article distributed other hazards like earthquakes or volcanoes), geomorphological (e.g., changes in slope
under the terms and conditions of the features through surficial processes), and hydrological (e.g., precipitation, groundwater, or
Creative Commons Attribution (CC BY) freeze/thaw) [3]. The latter, hydrological triggering mechanisms, are the most prevalent
license (https://creativecommons.org/ cause of landslides around the world, especially in tropical regions and areas that receive
licenses/by/4.0/). heavy precipitation [4–7]. Oversaturated, non-cohesive soils and variegated debris may

Geomatics 2021, 1, 3–17. https://dx.doi.org/10.3390/geomatics1010002 https://www.mdpi.com/journal/geomatics

Geomatics 2021, 1 4

come down-slope during or after heavy rainfall [8]. These landslides are typically shallow-
seated, cover a small areal extent, and deposit thin layers of material [9,10].
The hilly terrain of the state of Kerala, a linear stretch of land with the Western Ghats
running throughout its length, is a landslide-prone area. Landslides here are shallow-seated
as the overburden thickness usually varies in the range of 1–5 m, with most of the region
having less than 2 m overburden [11], and they rest unconformably on the Precambrian
bedrock [12]. Thus, a detailed study of this hazard, using the latest technology, is inevitable.
The best possible method of studying these cataclysmic phenomena is through remote
sensing. Remote sensing allows spatial coverage at several orders of magnitudes, ranging
from detailed examinations of individual assets to exploration across hundreds of square
kilometers [13–17]. Being a region of intense rainfall and marred with clouds, microwave
remote sensing seems to be the best possible data source for studying landslides in tropics.
Thus, this study aims to monitor and map the shallow-seated landslides triggered during
monsoon season, using a satellite-based Synthetic Aperture Radar (SAR) technique, the
Persistent Scatterer Interferometric (PSI). PSI data modeling and procedures estimate
deformation time series data and average displacement rates [18]. PSI is used to measure
relatively small changes in ground deformation (mm-scale) between a reference radar
image and all other images [19,20]. PSI has been successfully used in a variety of unstable
slope cases [18,21–28]. This work intends to use successful applications of satellite-based
radar techniques to explicitly include mapping and monitoring of shallow landslides
triggered during the monsoon season in a tropical environment and combine the radar
monitoring to refine traditional slope stability maps.

2. Study Area
The Western Ghats, a mountain range, and prominent escarpment paralleling the
southwestern coast of India (Figure 1a) is a hotspot for hydrologically-triggered land-
slides [29]. About 8% (roughly 1400 km2 ) of the Western Ghats of Kerala is classified as a
critical zone for mass movements [11,30].
The present study is concentrated in the Ernakulam, Idukki, and Kottayam districts
of central Kerala (shown in Figure 1b) and lies between 9◦ 400 N to 10◦ 200 N latitude and
76◦ 400 E to 77◦ 100 E longitude covering an area of 2221 km2 . The average annual rainfall is
between 250 cm and 425 cm; the temperature varies between 21 ◦ C to 29 ◦ C. Landslides here
are shallow-seated and relatively small, typically less than 2 m in depth [11] (Figure 1c).
Rugged hills and deep valleys characterize this section of the Western Ghats. Holocene
sediments and soils rest unconformably atop Precambrian bedrock composed of gneiss
(e.g., biotite, hornblende–biotite, and more felsic granite gneiss), charnockite, granulite,
and quartzite [31]. The stratigraphic column of the state is shown in Table 1. The contact
between sediments and bedrock usually forms the landslide rupture surface.

Table 1. Stratigraphic column for the state of Kerala (Source: Nandakumar and Balakrishnan [32]).

Age Formation Lithology

Recent Alluvium Sand, clay, riverine alluvium etc.
Sub-recent Laterite Derived from crystalline and sedimentary rocks
Warkalli
Sandstone, clays with lignite
Quilon
Tertiary Limestone, marl and clay
Vaikom
Sandstone with pebbles, clay and lignite Carbonaceous Clay and fine sand
Alappuzha
Undated Intrusives Dolerite, Gabbro, Granites, Quartzo-feldspathic veins
Wayanad Group Granitic gneiss, Schists etc.
Archaean Charnockites Charnockites and associated rocks
Khondalites Khondalite suite of rocks and its associates
Geomatics 2021, 1 5

Geomatics 2021, 1, FOR PEER REVIEW 3

Figure 1. Location map of the study area: (a) with respect to South India (b) study area showing the location of landslides
used in
used in the
the present
present study
study (c)
(c) Field
Field reconnaissance
reconnaissance photo
photo showing
showing the
the ‘holotype’
‘holotype’ shallow
shallow landslide
landslide occurring
occurring in
in the
the study
study
area (Background image for (a,b): ArcGIS Earth).
area (Background image for (a,b): ArcGIS Earth).

Table 1. Stratigraphic column for the state of Kerala (Source: Nandakumar and Balakrishnan [32]).
3. Data and Methodology
Age Formation Twenty-two descending single lookLithology complex (SLC) synthetic aperture radar (SAR)
Recent Alluvium images acquired between March Sand,
2016clay,
andriverine
June 2017 alluvium
from theetc.
Sentinel-1A satellite (Table 2),
Sub-recent Lateritecourtesy of the European Spacefrom
Derived Agency (ESA), were
crystalline downloaded rocks
and sedimentary from the Alaska Satellite
WarkalliFacility at the University of Alaska Fairbanks [33]. Sentinel-1A acquires SAR images at an
approximate wavelength of 5.6 Sandstone,
cm (C-band) clays
over with
an lignite
area with dimensions 250 by 180 km
Quilon
Tertiary at a revisiting cycle of 12 days. Limestone,
The incidence marl andfor
angle clay
the Interferometric Wide Swath
Vaikom
(IWS) modeSandstone
rangeswith
frompebbles, clay and
29.1 degrees to lignite Carbonaceous
46 degrees, Clay and
with a central fine sand
incidence angle of
Alappuzha
approximately 37.55 degrees. Sentinel-1A images were processed using the PSI technique
Undated Intrusives Dolerite, Gabbro, Granites, Quartzo-feldspathic veins
to measure ground deformation rates with an accuracy up to 1 mm/yr [34] following the
Wayanad Group
default approach provided with the Granitic
ENVI gneiss,
SARscape Schists etc. [19,20,35]. The methodology
software
Archaean Charnockites Charnockites and associated
behind landslide mapping and monitoring is shown in Figure rocks2.
Khondalites Khondalite suite of rocks and its associates

3. Data and Methodology

Twenty-two descending single look complex (SLC) synthetic aperture radar (SAR)
images acquired between March 2016 and June 2017 from the Sentinel-1A satellite (Table
2), courtesy of the European Space Agency (ESA), were downloaded from the Alaska Sat-
ellite Facility at the University of Alaska Fairbanks [33]. Sentinel-1A acquires SAR images
at an approximate wavelength of 5.6 cm (C-band) over an area with dimensions 250 by
180 km at a revisiting cycle of 12 days. The incidence angle for the Interferometric Wide
Swath (IWS) mode ranges from 29.1 degrees to 46 degrees, with a central incidence angle
 Geomatics 2021, 1, FOR PEER REVIEW

of approximately 37.55 degrees. Sentinel-1A images were processed using the PS

Geomatics 2021, 1 6 [34] foll
nique to measure ground deformation rates with an accuracy up to 1 mm/yr
the default approach provided with the ENVI SARscape software [19,20,35]. The m
ology behind landslide mapping and monitoring is shown in Figure 2.
Table 2. Dates of image acquisition of single look complex (SLC) synthetic aperture radar (SAR) images.
Table 2. Dates of image acquisition of single look complex (SLC) synthetic aperture radar (S
images. Date of Image Date of Image
S.No S.No
Acquisition Acquisition
S.No Date of Image Acquisition S.No Date of Image Acquisition
1 1 4 March 20162016
4 March 12
12 05 December
05 December 2016 2016
2 2 15 May 20162016
15 May 13
13 29 December
29 December 2016 2016
3 08 June 2016 14 10 January 2017
3 08 June2016 14 10 January 2017
4 14 July 2016 15 11 March 2017
5 4 14 July
19 August 2016
2016 15
16 11 March 2017 2017
23 March
6 5 19 August
12 September 2016 2016 16
17 23 March 2017 2017
04 April
7 6 12 September
24 September 20162016 17
18 04 April 2017 2017
16 April
8 7 24 September
06 October 2016 2016 18
19 16 April 2017 2017
28 April
9 8 30 October 20162016
06 October 20
19 10 May
28 April 2017 2017
10 9 11 November 2016
30 October 2016 21
20 10 May 2017 2017
22 May
11 10 23 November 2016
11 November 2016 22
21 22 May03 2017
June 2017
11 23 November 2016 22 03 June 2017

Figure
Figure 2. Flow 2. describing
chart Flow chart the
describing the methodology
methodology adopted
adopted in the studyinarea.
the study area.

3.1. Selection of 3.1.

Landslide Location
Selection of Landslide Location
Twelve different landslide
Twelve locations
different were collated
landslide fromwere
locations various media from
collated sources, mostlymedia so
various
newspaper articles,
mostly and verified through
newspaper articles,field
and studies.
verified The details
through of these
field landslides
studies. are of these
The details
shown in Tableslides
3. Allare
these landslides were confined to roadcuts. These landslides
shown in Table 3. All these landslides were confined to roadcuts. These were
categorized intoslides
threewere
groups based on activity
categorized periods
into three groupsthatbased
align with PSI data
on activity acquisition:
periods that align wi
Landslides that dataoccurred beforeLandslides
acquisition: March 2016, thati.e., representing
occurred landslide
before March 2016, that
i.e.,occurred
representing lan
before the SARthatdataoccurred
collection, are referred
before the SARtodataas ‘Group
collection,1’; between March
are referred 2016
to as and June
‘Group 1’; between
2017, i.e., landslides that occurred
2016 and June 2017, during the SAR data
i.e., landslides thatcollection
occurredtime-period,
during the are SARreferred
data collection
to as ‘Group 2’,period,
and after June
are 2017, i.e.,
referred to aslandslides
‘Group 2’,after
andthe SAR
after Junedata collection
2017, time-period,
i.e., landslides after the SA
are referred to collection
as ‘Group time-period,
3’ (Figure 3). areIndividual landslides were studied against
referred to as ‘Group 3’ (Figure 3). Individual the PSIlandslide
deformation values.
studied The dates the
against of data were selectedvalues.
PSI deformation prior toThe thedates
2018 anomalous
of data were rainfall
selected prior
that occurred in Kerala which resulted in one of the biggest floods and 4726
2018 anomalous rainfall that occurred in Kerala which resulted in one of the biggest landslides,
so that more significant movementssoassociated
and 4726 landslides, that morewith this anomalous
significant movements rainfall should
associated notthis anom
with
manifest in ourrainfall
results,should
thereby, any anomaly can be avoided.
not manifest in our results, thereby, any anomaly can be avoided.
Geomatics 2021, 1 7

Geomatics 2021, 1, FOR PEER REVIEW 5

Table 3. Landslide inventory of the study area.
Table 3. Landslide inventory of the study area.
Lat Long Date of Dimension (m) Average PSI
S.No Landslide Location Direction Dimension (m) Average
Lat
N LongE Occurrence Velocity (mm/yr)
PSI Velocity
S.No Landslide Location Direction Date of Occurrence L W D
N E L W D (mm/yr)
Before Monitoring
Before Monitoring
1 1 NH-49
NH-49Salaf
Salafmasjid
masjid West
West 10.030 76.879
10.030 76.879 05 August
05 August
20132013 5 5 5 5 0.5 0.5 −4.6721
−4.6721
2 NH-49
NH-49Machiplavu
Machiplavu village
vil- West 10.023 76.910 05 August 2013 10 30 0.3 −2.0795
3 2 Painavu–Kattapana
lage road West
East 10.023
09.893 76.910
76.972 05 August 2013
05 August 2013 10 12 30 8 0.3 0.5 −2.0795
−0.94653
4 Adimali–Kirittodu road East 10.007 76.959 05 August 2013 10 10 1 −1.6030
3 Painavu–Kattapana road East 09.893 76.972 05 August 2013 12 8 0.5 −0.94653
5 Thadiyam padu East 09.882 76.970 04 August 2013 10 5 1 −0.0448
4 Adimali–Kirittodu road East 10.007 76.959 05 August 2013 10 10 1 −1.6030
6 Munnar Tea County West 10.088 77.065 24 June2013 7.5 8.5 1 −4.9353
7 5 Thadiyam
Munnar padu
MG colony East
West 09.882
10.088 76.970
77.070 04 August
24 June2013
2013 10 16 5 15 1 2 −0.0448
−0.0091
8 6 Munnar new
Munnar Tea County
colony West
West 10.088
10.089 77.065
77.075 24 June2013
24 June2013 7.5 4 8.5 5 1 0.4 −4.9353
−0.5933
7 Munnar MG colony West 10.088 77.070 24 June 2013 16 15 2 −0.0091
During Monitoring
8 Munnar new colony West 10.089 77.075 24 June2013 4 5 0.4 −0.5933
St. Mary’s orthodox
9 East 10.083During Monitoring
77.070 28 June 2016 9 15 2 −0.1944
church
St. Mary’s orthodox
10 9 Munnar–Gudaral road East
East 10.083
10.085 77.070
77.070 28 June 2016
28 June 2016 9 6 15 15 2 1.5 −1.9065
−0.1944
church
10 Munnar–Gudaral road East 10.085 After Monitoring
77.070 28 June 2016 6 15 1.5 −1.9065
11 Adimali–Rajakad road East 09.985 After Monitoring
76.998 19 September 2017 15 8 2 −1.3890
1211 Kamakshi village
Adimali–Rajakad road West
East 09.835 76.998
09.985 77.053 19 September
09 August2017
2017 15 75 8 30 2 4 −6.0173
−1.3890
12 Kamakshi village West 09.835 77.053 09 August 2017 75 30 4 −6.0173

Figure3.3.Categorization
Figure Categorizationofoflandslides
landslidesininthe
thestudy
studyarea
areabased
basedon
onPersistent
PersistentScatterer
ScattererInterferometry
Interferome-
try (PSI) monitoring.
(PSI) monitoring.

3.2.Monitoring
3.2. MonitoringofofIndividual
IndividualLandslides
Landslides
Thesethree
These threecategories
categoriesofoflandslides
landslides were
were correlated
correlated with
with thethe temporal
temporal deformation
deformation of
of PSI
PSI points.
points. These
These threethree groups
groups werewere further
further subdivided
subdivided intobased
into two two based
on the on the east-
east-facing
facing
and and west-facing
west-facing slope.
slope. This This classification
classification is performedis performed
to account to
foraccount
the Linefor the Line
of Sight of
(LoS)
Sight
of (LoS) ofsatellite
Sentinel-1A Sentinel-1A satellite
that plays that
a vital plays
role a vital
in the role in the
deformation deformation
direction measured direction
using
PSI. These Sentinel-1A
measured images
using PSI. These were acquired
Sentinel-1A images inwere
descending
acquired mode. Thus, landslides
in descending can
mode. Thus,
show up in can
landslides PSI show
data inup
two ways:
in PSI data in two ways:
1.1. Negative
Negative velocity
velocity values
values on
onwest-facing
west-facingslopes,
slopes,indicating
indicatingdownward
downwardor orwestward
westward
(downslope) movement,
(downslope) movement, and and
2.2. Positive
Positive velocity
velocity values
values on
on east-facing
east-facing slopes,
slopes, indicating
indicating downward
downwardor oreastward
eastward
(downslope)
(downslope)movement.
movement.
This
Thismethod
methodwill show
will show good results
good for for
results landslides withwith
landslides E–WE–W
oriented (±45◦(±45°)
slopesslopes
oriented ) [36].
The
[36].east
Theand west-facing
east slopes were
and west-facing slopesdetermined using the
were determined 30 mthe
using spatial
30 mresolution Shuttle
spatial resolution
Radar
Shuttle Topographic Mission (SRTM)
Radar Topographic Missiondigital
(SRTM) elevation model (DEM)
digital elevation downloaded
model from the
(DEM) downloaded
United
from the States Geological
United Survey (USGS)
States Geological Survey website
(USGS) [37].
website [37].

3.3. Filtering of PSI Points

The PSI technique can produce a large number of points depending on the choice of
a coherence threshold. Although all the points convey the same amount of information,
Geomatics 2021, 1 8

3.3. Filtering of PSI Points

The PSI technique can produce a large number of points depending on the choice of a
coherence threshold. Although all the points convey the same amount of information, they
are affected differently by random noise and the strength of the signal being measured. To
identify and isolate only the points that convey the maximum amount of information and
avoid potential noise and artifacts (especially the influence of vegetation in the tropical
region), we undertook a selection process looking for points with a high signal to noise
ratio and signal quality utilizing the coherence value. For the filtering, we select only
Persistent Scatterer (PS) points that have higher coherence than the 95-percentile.

3.4. Creation of Slope Stability Map

Geologic, hydrologic, and geomorphologic data, including the normalized difference
vegetation index (NDVI), slope angle, groundwater level, and soil strength parameters,
were used to create a landslide susceptibility map. NDVI was generated using the Near-
Infrared (NIR) and Red bands of Landsat 8 [36]. The formula for NDVI generation is:

(NIR − Red)
NDVI = (1)
(NIR + Red)

and was reclassified based on Holben [38]. The soil map was taken from the National
Bureau of Soil Survey data [39], which used the Unified Soil Classification System (USCS).
The slope map was prepared using SRTM data [40]. The slope map was classified based on
the Bureau of Indian Standard [41]. All thematic maps used for the creation of landslide
susceptibility map are shown in Figure 4. The spatial extent of the landslide susceptibility
map was then reduced by a factor of safety (FS) analysis to identify and target landslide-
prone slopes. An FS map was prepared based on input factors used in the Geographic
Information System Tool for Infinite Slope Stability Analysis (GIS-TISSA) [42] and Prob-
abilistic Infinite Slope Analysis model (PISA-m) [43]. The equation for calculating the
FS is:
cr + cs + [qt + γm D + (γsat − γw − γm )Hw D] cos 2β tanφ
FS = (2)
[qt + γm D + (γsat − γm )Hw D]sin β cos β
where cr is root cohesive strength (pressure); cs is soil cohesive strength (pressure); qt
is tree surcharge (pressure); γm is saturated unit weight of the soil (force/volume); γsat
is moist unit weight of the soil (force/volume); D is soil thickness; Hw is pore pressure
coefficient (0 ≤ Hw ≤ 1); β is the angle of topographical slope; ϕ is the angle of internal
friction; In this study, we assumed a partially saturated soil condition (Hw = 0.5) for the
slope stability analysis.

3.5. Refinement of Slope Stability Map

A slope aspect map of the study area was prepared, which was further edited as an
east and west aspect map, by taking all pixels with values 0–180 as east and 180–360 as
west. PSI points with positive values were extracted from eastern slopes, and PSI points
with negative values were extracted from west-facing slopes. All PSI points belonging to a
slope of 5 degrees or less were avoided as this area was deemed flat. A refinement matrix
was created to refine the conventional FS map prepared using GIS-TISSA by taking into
consideration the PS LoS velocities (VLoS ).
Overlaying of these two derivatives, i.e., one prepared using GIS-TISSA and other
using VLoS resulted in the combinations of different slope conditions. This final landslide-
prone zone map includes three hazard classifications: unstable, critical, and stable.
Geomatics 2021, 1 9

Geomatics 2021, 1, FOR PEER REVIEW 7

Figure 4. Data used for deriving landslide susceptibility map (a) normalized difference vegetation
Figure 4. Data used for deriving landslide susceptibility map (a) normalized difference vegetation
index (NDVI) map; (b) Soil map; (c) Elevation map draped over hillshade; (d) slope map.
index (NDVI) map; (b) Soil map; (c) Elevation map draped over hillshade; (d) slope map.

4.3.5. Refinement of Slope Stability Map

Results
A slope aspect
4.1. Monitoring Shallowmap of the
Debris study area was prepared, which was further edited as an
Flows
eastPSI
andresults
west aspect map, by taking
using the Sentinel-1A SLC all pixels with values
SAR image 0–180
are shown inas east and
Figure 180–360
5a. The smallas
west. PSI points with positive values were extracted from eastern slopes,
triangular cut-over portion is due to that fact that one Sentinel-1 image is composed and PSI points
of
27 smaller bursts. When such data is used for delineating the area of interest (AOI), AOIto
with negative values were extracted from west-facing slopes. All PSI points belonging
a slope
may of 5 degrees
overlap or less were
with multiple avoided
bursts. Thus, as this
it is area was deemed
a pre-processing flat. A Figure
artifact. refinement matrix
5a shows
was
the created
VLoS (bothto refine the
positive andconventional FS map
negative values). The prepared using
PSI analysis GIS-TISSA
provided by PS
153,876 taking into
points,
consideration
and the PS average
each point yields LoS velocities (Vvalues
velocity LoS). (mm/yr), incremental displacement values
 4.1. Monitoring Shallow Debris Flows
PSI results using the Sentinel-1A SLC SAR image are shown in Figure 5a. The small
triangular cut-over portion is due to that fact that one Sentinel-1 image is composed of 27
Geomatics 2021, 1
smaller bursts. When such data is used for delineating the area of interest (AOI), AOI may 10
overlap with multiple bursts. Thus, it is a pre-processing artifact. Figure 5a shows the VLoS
(both positive and negative values). The PSI analysis provided 153,876 PS points, and each
point yields average velocity values (mm/yr), incremental displacement values (mm) and
(mm)
total and total displacement
displacement values The
values (mm). (mm). The velocity
velocity varies varies −24.38−24.38
between
between mm/yr mm/yr to
to +28.22
+28.22
mm/yr. mm/yr. 95-percentile
95-percentile coherence
coherence PS points
PS points were were only selected
only selected for further
for further analysis
analysis to
to avoid
avoid noise. The resulting PSI points were 9206 (Figure 5b). And these are the
noise. The resulting PSI points were 9206 (Figure 5b). And these are the PSI points that PSI points
that
werewere
usedused in this
in this study.
study. The The velocity
velocity of these
of these points
points variesvaries −13.56−mm/yr
between
between 13.56 mm/yr
to +11.28
tomm/yr.
+11.28 The
mm/yr. The 12 landslide locations, collected from different sources and field
12 landslide locations, collected from different sources and field verification
verification
(shown in (shown
Figure 1) inwere
Figure 1) were
placed intoplaced
three into three temporal
temporal groups:
groups: eight eight landslides
landslides in “Group
in1”,
“Group 1”, two landslides
two landslides in “Groupin2”,
“Group
and two 2”, and two landslides
landslides in “Groupin 3”
“Group 3” andclassified
and further further
classified into east and west-facing
into east and west-facing (Table 3). (Table 3).

Figure 5. PSI results using the Sentinel-1A SLC SAR image (Background image is Shuttle Radar
Figure 5. PSI results using the Sentinel-1A SLC SAR image (Background image is Shuttle Radar
Topographic Mission (SRTM) elevation map draped over hillshade) (a) Both positive and negative
Topographic Mission (SRTM) elevation map draped over hillshade) (a) Both positive and negative
VLoS values; (b) Persistent Scatter (PS) points that have greater than 95-percentile coherence.
VLoS values; (b) Persistent Scatter (PS) points that have greater than 95-percentile coherence.
PSImonitoring
PSI monitoringof oflandslides
landslides was
was done
done for
for these
these12
12landslides.
landslides.Sentinel-1A
Sentinel-1A onon
itsits
de-
scending mode characterizes the west-facing slope with negative velocity
descending mode characterizes the west-facing slope with negative velocity values, in- values, indicat-
ing down-slope
dicating down-slope or lateral movement
or lateral movementto the west,
to the andand
west, positive velocity
positive values
velocity on east-
values on
facing slopes, indicating down-slope or lateral movement to the east [44–46].
east-facing slopes, indicating down-slope or lateral movement to the east [44–46]. Buffering Buffering
wascarried
was carriedoutoutasasthe
thelandslides
landslideswere
weredemarcated
demarcatedbasedbasedononlocal
localreports
reportsand
andhence
hencemay
may
notrepresent
not representlandslide
landslideinitiation
initiationpoints.
points.Buffering
Bufferingwas
wasperformed
performedbased basedononthe
therun-out
run-out
lengthofoflandslides,
length landslides,which
whichisis101
101m.m.The
Theeast-facing
east-facinglandslide
landslidehaving
havingwest-face
west-faceslope
slopeinin
the buffer was removed and vice-versa, for getting more accurate
the buffer was removed and vice-versa, for getting more accurate results. results.
Out of the three different temporal periods, the Group 2 category shows the maximum
deformation value (St Mary’s Orthodox Church with −35.67 mm) followed by Group
3 (Kamakshi village landslide with a deformation value of −28.03 mm) and Group 1
categories (NH-49 Salaf masjid landslide with −21.07 mm deformation). These three
typical cases are shown in Figure 6. The rest of the temporal deformation graphs for other
landslides are shown in Supplementary Material. All these landslides are west-facing.
The maximum deformation recorded for the east-facing landslides is +10.11 mm/yr for
Munnar–Gudaral Road landslide that belongs to Group 2. This was followed by +8.66 mm
deformation for the Adimali–Rajakkad Road landslide that belongs to Group 3 category
landslide. Group 1 has a maximum deformation of +8.53 mm, both for Painavu–Kattapana
landslide and Thadiyampadu landslide. These deformations indicate that the maximum
 cases are shown in Figure 6. The rest of the temporal deformation graphs for other land-
slides are shown in Supplementary Material. All these landslides are west-facing. The
maximum deformation recorded for the east-facing landslides is +10.11 mm/yr for Mun-
nar–Gudaral Road landslide that belongs to Group 2. This was followed by +8.66 mm
Geomatics 2021, 1 deformation for the Adimali–Rajakkad Road landslide that belongs to Group 3 category 11
landslide. Group 1 has a maximum deformation of +8.53 mm, both for Painavu–Kattapana
landslide and Thadiyampadu landslide. These deformations indicate that the maximum
deformation
deformation isisfor
forthe
thelandslides thatoccurred
landslides that occurred during
during the the
timetime
of PSIofmonitoring
PSI monitoring (Group-
(Group-2),
2), whereas the landslides that are going to happen in the future have smaller
whereas the landslides that are going to happen in the future have smaller deformation deformation
compared
compared to to
Group-2.
Group-2.ItItisisalso
also noted thatsites
noted that sitesthat
that have
have experienced
experienced landslides
landslides in thein the
past (Group-1)
past (Group-1) are
arealso
alsoshowing signsof
showing signs ofdeformation,
deformation, which
which is indicative
is indicative of reactivation
of reactivation
and/or
and/or consolidation/settlementofoflandslides.
consolidation/settlement landslides.

Figure 6. Deformation
Figure values
6. Deformation ofof
values three
threetypical
typical landslides fromGroup
landslides from Group1, 1, Group
Group 2 and
2 and Group
Group 3. 3.

4.2.4.2.
Mapping and
Mapping Refining
and Refiningthe
the Conventional FSMap
Conventional FS Map
The conventional
The conventionalFSFSmap
map prepared usingGIS-TISSA
prepared using GIS-TISSA (Figure
(Figure 7a) 7a)
waswas refined
refined usingusing
PSIPSI
velocity
velocityvalues
values(Figure
(Figure 7b,c).
7b,c). Since PSImeasures
Since PSI measuresthethe slant
slant range
range displacement
displacement and and
velocities of ground points, landslide movements appear differently in the eastern
velocities of ground points, landslide movements appear differently in the eastern and and
western slopes over any satellite orbital geometry (ascending or descending). Since all
the images downloaded were of the descending mode, the landslides are observed as
negative velocity points in the western slope and positive velocity points in the eastern
slope. Thus, the PS points were divided on the basis of slope aspect as east and west. All
points falling in the flat aspect were avoided as this area is physiographically plain and
free from landslides. The rest of the points, after the 95-percentile filtering, were extracted.
 Geomatics 2021, 1, FOR PEER REVIEW 10

western slopes over any satellite orbital geometry (ascending or descending). Since all the
images downloaded were of the descending mode, the landslides are observed as nega-
Geomatics 2021, 1 tive velocity points in the western slope and positive velocity points in the eastern slope. 12
Thus, the PS points were divided on the basis of slope aspect as east and west. All points
falling in the flat aspect were avoided as this area is physiographically plain and free from
landslides. The rest of the points, after the 95-percentile filtering, were extracted. Upon
Upon extracting, the positive PS points in the eastern slope comprise 3852 points, and
extracting, the positive PS points in the eastern slope comprise 3852 points, and the nega-
the negative PS points in the western slope amounted to 5023. It is to be noted that these
tive PS points in the western slope amounted to 5023. It is to be noted that these numbers
numbers were generated after reducing the points on flat ground. The PS points in the
were generated after reducing the points on flat ground. The PS points in the east slope
east slope range from 0 to +11.28 mm/yr, whereas in the west slope, it spans from 0 to
range from 0 to +11.28 mm/yr, whereas in the west slope, it spans from 0 to −23.03 mm/yr.
−23.03 mm/yr. This was divided into the equal class interval. The values between 0 and
This was divided into the equal class interval. The values between 0 and ±8 mm/yr are
±8 mm/yr are considered as low VLOS ; velocity values between ±8 and ±16 mm/yr as
considered as low VLOS; velocity values between ±8 and ±16 mm/yr as medium VLOS and
medium VLOS and the rest as high VLOS . Thus, the eastern slopes do not have a high
the rest as high VLOS. Thus, the eastern slopes do not have a high VLOS class. These points
VLOS class. These points were used to refine the corresponding GIS-TISSA pixels in the
were used to refine the corresponding GIS-TISSA pixels in the eastern and western slopes
eastern and western slopes using a conversion matrix (Figure 8a). Such a matrix was used
using a conversion matrix (Figure 8a). Such a matrix was used by Ciampalini et al. [47].
by Ciampalini et al. [47]. Pixels in the unstable class were not changed. In the critical
Pixels in the unstable class were not changed. In the critical class, pixels coinciding with
class, pixels coinciding with high-velocity PS points were moved to unstable class. Pixels
high-velocity PS points were moved to unstable class. Pixels coinciding with medium ve-
coinciding with medium velocity were unchanged, while those that coincided with low
locity were unchanged, while those that coincided with low velocity were moved to stable
velocity were moved to stable class. In the stable class, pixels coinciding with high-velocity
class. In the stable class, pixels coinciding with high-velocity PS points were moved to the
PS points were moved to the critical class. At the same time, those coinciding with medium
critical class. At the same time, those coinciding with medium and low velocity was re-
and low velocity was retained in the same class. After refining, it was found that there was
tained in the same class. After refining, it was found that there was a marked decrease in
a marked decrease in the percentage of pixels in the critical class (Figure 8b). Most pixels in
the percentage of pixels in the critical class (Figure 8b). Most pixels in the critical classes
the critical classes were deemed as stable, thereby reducing the number of false positives.
were deemed as stable, thereby reducing the number of false positives.

Figure7.7.(a)
Figure (a)Landslide
Landslidesusceptibility
susceptibility map
map prepared
prepared using
using Geographic
Geographic Information
InformationSystem
SystemTool
Toolfor
for
Infinite Slope Stability Analysis (GIS TISSA) (b) Refined landslide susceptibility map (c) Zoomed
Infinite Slope Stability Analysis (GIS TISSA) (b) Refined landslide susceptibility map (c) Zoomed
part of the refined landslide susceptibility map.
part of the refined landslide susceptibility map.
Geomatics 2021, 1 13

Geomatics 2021, 1, FOR PEER REVIEW 11

Figure
Figure 8. 8.
(a)(a) Conversion
Conversion matrix
matrix used
used for
for refining
refining theconventional
the conventionalsusceptibility
susceptibilitymap
map(b)
(b)Pie
Piechart
chart
showing the changes in the stability classes prior and after refinement.
showing the changes in the stability classes prior and after refinement.

5. 5. Discussion
Discussion
TheThe purpose
purpose ofof this
this study
study is is
toto show
show thethe efficiency
efficiency ofof PSI
PSI inin monitoring
monitoring andand mapping
mapping
shallow-seated
shallow-seated landslides
landslides inin tropical
tropical areas.
areas. Deformation
Deformation caused
caused byby landslides,
landslides, both
both local
local
andandregional
regionalscale,scale,as as well as as both
bothslow-
slow-oror extremely
extremely slow-moving
slow-moving andandfast-moving
fast-moving land-
landslides
slides cancan be be studied
studied by by
radarradarandandcancan be evaluated
be evaluated usingusing
PSI PSI
[48].[48]. In this
In this study, study, 12
12 land-
landslides, separated
slides, separated intothree
into threetemporal
temporalcategories
categories based
based on PSI PSI monitoring
monitoringasasGroup Group1, 1,
Group
Group 2, 2,
and andGroup
Group 3, 3,
which
which werewere further
further divided
divided into east
into and
east and west-facing
west-facing landslides,
landslides,
were
were studied
studied inin detail
detail totounderstand
understand howhow PSIPSIdata
datacan
canbebeutilized
utilized forfor
studying
studying shallow-
shallow-
seated landslides. The analyzed PSI data spanned from March
seated landslides. The analyzed PSI data spanned from March 2016 through July 2017. 2016 through July 2017.
Landslides
Landslides ofofthese
these three
threeperiods
periods also
alsoshow
show characteristic
characteristic deformation
deformation rates.
rates.This
Thisis is
because
because the
the monitoring
monitoring period
period covers
covers allall
thethesequence
sequence ofof
development
development of of
shallow-seated
shallow-seated
landslide
landslide viz. thethe
viz. monsoon
monsoon season,
season,the the
subsequent
subsequent increase in pore-water
increase in pore-water pressure, and the
pressure, and
final
thefailure. The west-facing
final failure. slopes showed
The west-facing the largest
slopes showed thedeformation values (e.g.,
largest deformation valuesSt. Mary’s
(e.g., St.
Orthodox Church landslide
Mary’s Orthodox Churchwith a valuewith
landslide of −a35.67
valuemm), whereas
of −35.67 mm),the maximum
whereas the formaximum
the east-
facing landslides is +10.11 mm/yr for the Munnar–Gudaral
for the east-facing landslides is +10.11 mm/yr for the Munnar–Gudaral Road landslide.Road landslide. Both belong
toBoth
the Group 2 landslides. The second largest deformation values
belong to the Group 2 landslides. The second largest deformation values demon- demonstrated category
is strated
Group category
3 landslides. is GroupThese3 deformations
landslides. These indicate that the maximum
deformations indicate that deformation
the maximum is
fordeformation
the landslides that occurred during the time of PSI monitoring
is for the landslides that occurred during the time of PSI monitoring (Group (Group 2), whereas
the landslides that
2), whereas the landslidesare going that to happen
are goingin the
to future
happen(Group
in the 3) have(Group
future smaller3)deformation
have smaller
compared
deformation compared to Group 2. It is also noted that sites that havelandslides
to Group 2. It is also noted that sites that have experienced experienced in the
land-
past (Group 1) are also showing signs of deformation, which is
slides in the past (Group 1) are also showing signs of deformation, which is indicative ofindicative of reactivation
and/or settlement/compaction
reactivation of landslides. of
and/or settlement/compaction Thelandslides.
West-facing TheKamakshi
West-facingvillage landslide
Kamakshi vil-
with −28.03 mm and the east-facing Adimali–Rajakad Road landslide with +8.66 mm
lage landslide with −28.03 mm and the east-facing Adimali–Rajakad Road landslide with
deformation belongs to Group 3. Lastly, in the category of Group 1 landslides, none of the
+8.66 mm deformation belongs to Group 3. Lastly, in the category of Group 1 landslides,
developmental stages of the shallow-seated landslide were under surveillance, and hence
none of the developmental stages of the shallow-seated landslide were under surveillance,
and hence the group was characterized by the least deformation value. NH-49 Salaf Mas-
jid landslide with −21.07 mm deformation and +8.53 mm for Painavu landslide are in the
Geomatics 2021, 1 14

the group was characterized by the least deformation value. NH-49 Salaf Masjid landslide
with −21.07 mm deformation and +8.53 mm for Painavu landslide are in the Group 1
landslides. Bianchini et al. [48] and Cigna et al. [49] also used PSI data for describing the
state of activity of landslides in Calabria Region, Italy. Similarly, Mateos et al. [50] used PSI
as a measure to assess the damage caused by landslides in Granada, Spain.
An attempt to improve the existing landslide susceptibility map using PSI data was
also attempted in this study. Accuracy of conventional landslide susceptibility maps
needs improvement as it tends to have a large area of false positives and negatives. The
method deployed to improve the accuracy, i.e., melding of GIS-TISSA derived landslide
susceptibility map and PSI data, resulted in the refinement of the landslide susceptibility
map. The ultimate result ended with a reduction in the critical category from 17 to 2
percentage points. This will help in efficient planning of disaster risk reduction activities.
Combining the PSI information with the slope stability model has significant practical
implications. Traditionally, slope stability analysis is performed using physics-based [7] or
statistical models [51]. Moreover, the models developed using these approaches are highly
conservative, leading to several false positives. Because of the large false positives, the
community often loses its confidence in these models and get desensitized to any early
warning or risk predictions based on these models. Therefore, any opportunity to refine the
slope stability model will enhance community confidence in risk estimations based on these
models. The proposed approach that we have demonstrated in this study, combining PSI
and the slope stability modeling, is a valuable step to achieve that goal. The combination
of PSI with the stability model helped to refine the stability model output. The refined
output (by eliminating false positives) also has significant economic implications as well. If
the agency decides to monitor or mitigate the landslide risk at the critical location, then a
refined output based on PSI and stability model would require fewer locations monitored
or stabilized, which will provide cost savings.

6. Conclusions
This study is an attempt to monitor and map shallow-landslides set in a tropical set-up.
The study resulted in the following conclusions:
i. The three types of temporal landslides pertaining to different periods of PSI analysis
resulted in providing three distinct PSI velocity values. These refer to landslides that
occurred before the analyzed time period, during PSI analysis, and after the analysis.
ii. Traditional slope stability analysis methods often lead to significant false posi-
tives. The proposed approach combining PSI with traditional slope stability shows
promise in reducing these false positives. The PSI derived velocity was used to re-
fine the existing landslide susceptibility map. Such refinement will help to identify
areas that require utmost priority in terms of monitoring or mitigation.
iii. Developing countries like India will only have limited financial provisions for
adopting management practices. The freely available radar data and the provisions
of identifying areas requiring urgent management are able to be deciphered through
studies like the one mentioned here.
iv. Such an approach could also help the engineering community gain better commu-
nity confidence as the slope stability model is more targeted and less susceptible to
false positives.

Supplementary Materials: The following is available online at https://www.mdpi.com/2673-7418/

1/1/2/s1. Supplementary material is PSI deformation plots for different landslide locations.
Author Contributions: Conceptualization, A.R. and K.S.S.; methodology, N.L. and C.L.V.; formal
analysis, A.R., N.L. and C.L.V.; investigation, T.O. and V.M.; resources, E.H.B.; writing—original
draft preparation, A.R., N.L. and C.L.V.; writing—review and editing, K.S.S. and T.O.; visualization,
E.H.B. and V.M.; supervision, K.S.S. All authors have read and agreed to the published version of
the manuscript.
Funding: No funding.
Geomatics 2021, 1 15

Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: Data will be shared based on request.
Acknowledgments: The authors thank the handling editor and anonymous reviewers for providing
constructive remarks.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Piciullo, L.; Calvello, M.; Cepeda, J.M. Territorial early warning systems for rainfall-induced landslides. Earth Sci. Rev. 2018, 179,
228–247. [CrossRef]
2. Wieczorek, G.F. Landslides: Investigation and mitigation. In Landslides: Investigation and Mitigation, Special Report 247, Transporta-
tion Research Board, National Research Council; Turner, A.K., Schuster, R.L., Eds.; National Academy Press: Washington, DC, USA,
1996; pp. 76–90.
3. Zamenian, H.; Koo, D.D. Tunnel asset management (TAM) program application for high risk structural components. J. Eng.
Archit. 2014, 2, 1–12. [CrossRef]
4. Martinović, K.; Gavin, K.; Reale, C.; Mangan, C. Rainfall thresholds as a landslide indicator for engineered slopes on the Irish Rail
network. Geomorphology 2018, 306, 40–50. [CrossRef]
5. Naidu, S.; Sajinkumar, K.S.; Oommen, T.; Anuja, V.J.; Samuel, R.A.; Muraleedharan, C. Early warning system for shallow
landslides using rainfall threshold and slope stability analysis. Geosci. Front. 2018, 9, 1871–1882. [CrossRef]
6. Rossi, M.; Kirschbaum, D.; Valigi, D.; Mondini, A.C.; Guzzetti, F. Comparison of Satellite Rainfall Estimates and Rain Gauge
Measurements in Italy, and Impact on Landslide Modeling. Climate 2017, 5, 90. [CrossRef]
7. Smith, D.M.; Oommen, T.; Bowman, L.J.; Gierke, J.S.; Vitton, J. Hazard assessment of rainfall-induced landslides: A case study of
San Vicente volcano in central El Salvador. Nat. Hazards 2015, 75, 2291–2310. [CrossRef]
8. Cruden, D.M.; Varnes, D.L. Landslide types and processes. In Landslides: Investigation and Mitigation, Special Report 247,
Transportation Research Board, National Research Council; Turner, A.K., Schuster, R.L., Eds.; National Academy Press: Washington,
DC, USA, 1996; pp. 36–75.
9. Keaton, J.R.; De Graff, J.V. Surface observation and geologic mapping. In Landslides: Investigation and Mitigation, Special Report 247,
Transportation Research Board, National Research Council; Turner, A.K., Schuster, R.L., Eds.; National Academy Press: Washington,
DC, USA, 1996; pp. 178–230.
10. Varnes, D.J. Slope movement types and processes. In Landslides: Analysis and Control, Special Report 176, Transportation Research
Board, National Research Council; Schuster, R.L., Krizek, R.J., Eds.; National Academy Press: Washington, DC, USA, 1978; pp.
11–33.
11. Thampi, P.K.; Mathai, J.; Sankar, G.; Sidharthan, S. Debris flow in Western Ghats-A regional evaluation. In Final Proceedings Tenth
Kerala Science Congress; Kerala State Council for Science, Technology and Environment: Kozhikode, India, 1998; pp. 73–75.
12. Sajinkumar, K.S.; Anbazhagan, S.; Rani, V.R.; Muraleedharan, C. A paradigm quantitative approach for a regional risk assessment
and management in a few landslide prone hamlets along the windward slope of Western Ghats, India. Int. J. Disast. Risk Reduct.
2014, 7, 142–153. [CrossRef]
13. Blaschke, T.; Lang, S.; Lorup, E.; Strobl, J.; Zeil, P. Object-oriented image processing in an integrated GIS/remote sensing
environment and perspectives for environmental applications. Environ. Inf. Plan. Politics Public 2000, 2, 555–570.
14. He, K.S.; Rocchini, D.; Neteler, M.; Nagendra, H. Benefits of hyperspectral remote sensing for tracking plant invasions. Divers.
Distrib. 2011, 17, 381–392. [CrossRef]
15. Li, X.; Yeh, A.G.O. Analyzing spatial restructuring of land use patterns in a fast growing region using remote sensing and GIS.
Landscape Urban Plan. 2004, 69, 335–354. [CrossRef]
16. Metternicht, G.; Hurni, L.; Gogu, R. Remote sensing of landslides: An analysis of the potential contribution to geo-spatial systems
for hazard assessment in mountainous environments. Remote Sens. Environ. 2005, 98, 284–303. [CrossRef]
17. Escobar-Wolf, R.; Oommen, T.; Brooks, C.N.; Dobson, R.J.; Ahlborn, T.M. Unmanned aerial vehicle (UAV)-based assessment of
concrete bridge deck delamination using thermal and visible camera sensors: A preliminary analysis. Res. Nondestruct. Eval.
2018, 29, 183–198. [CrossRef]
18. Devanthéry, N.; Crosetto, M.; Monserrat, O.; Cuevas-González, M.; Crippa, B. An Approach to Persistent Scatterer Interferometry.
Remote Sens. 2014, 6, 6662–6679. [CrossRef]
19. Ferretti, A.; Prati, C.; Rocca, F. Nonlinear subsidence rate estimation using permanent scatterers in differential SAR interferometry.
IEEE Trans. Geosci. Remote Sens. 2000, 38, 2202–2212. [CrossRef]
20. Ferretti, A.; Prati, C.; Rocca, F. Permanent scatterers in SAR interferometry. IEEE Trans. Geosci. Remote Sens. 2001, 39, 8–20.
[CrossRef]
21. Bouali, E.H.; Oommen, T.; Escobar-Wolf, R. Mapping of slow landslides on the Palos Verdes Peninsula using the California
landslide inventory and persistent scatterer interferometry. Landslides 2018, 15, 439–452. [CrossRef]
Geomatics 2021, 1 16

22. Casagli, N.; Frodella, W.; Morelli, S.; Tofani, V.; Ciampalini, A.; Intrieri, E.; Raspini, F.; Rossi, G.; Tanteri, L.; Lu, P. Spaceborne,
UAV and ground-based remote sensing techniques for landslide mapping, monitoring and early warning. Geoenviron. Disasters
2017, 4, 1–23. [CrossRef]
23. Cascini, L.; Peduto, D.; Pisciotta, G.; Arena, L.; Ferlisi, S.; Fornaro, G. The combination of DInSAR and facility damage data for the
updating of slow-moving landslide inventory maps at medium scale. Nat. Hazards Earth Syst. Sci. 2013, 13, 1527–1549. [CrossRef]
24. Gullà, G.; Peduto, L.; Antronico, L.; Fornaro, G. Geometric and kinematic characterization of landslides affecting urban areas:
The Lungro case study (Calabria, Southern Italy). Landslides 2017, 14, 171–188. [CrossRef]
25. Lu, P.; Catani, F.; Tofani, V.; Casagli, N. Quantitative hazard and risk assessment for slow-moving landslides from persistent
scatterer interferometry. Landslides 2014, 11, 685–696. [CrossRef]
26. Novellino, A.; Cigna, F.; Sowter, A.; Ramondini, M.; Calcaterra, D. Exploitation of the intermittent SBAS (ISBAS) algorithm with
COSMO-SkyMed data for landslide mapping in north-western Sicily, Italy. Geomorphology 2017, 280, 153–166. [CrossRef]
27. Schaefer, L.N.; Lu, Z.; Oommen, T. Dramatic volcanic instability revealed by InSAR. Geology 2015, 43, 743–746. [CrossRef]
28. Schaefer, L.N.; Wang, T.; Escobar-Wolf, R.; Oommen, T.; Lu, Z.; Kim, J.; Lundgren, P.R.; Waite, G.P. Three-dimensional displace-
ments of a large volcano flank movement during the May 2010 eruptions at Pacaya Volcano, Guatemala. Geophys. Res. Lett. 2017,
44, 135–142. [CrossRef]
29. Sajinkumar, K.S. Geoinformatics in landslide risk assessment and management in parts of Western Ghats, Central Kerala, South
India. Ph.D. Thesis, IIT Bombay, Mumbai, India, 2005, unpublished.
30. Sajinkumar, K.S.; Oommen, T. Landslide Atlas of Kerala; Geological Society of India: Bangalore, India, 2020; p. 49.
31. GSI. Geology and Mineral Resources of Kerala; Geological Survey of India: Thiruvananthapuram, India, 2005; p. 93.
32. Nandakumaran, P.; Balakrishnan, K. Comparison of hydrochemical characteristics of shallow and depp aquifers in a hard rock
terrain in Kerala, India. In Proceedings of the International Conference & Exhibition on Integrated Water, Waste Water and
Isotope Hydrology, Bangalore, India, 25–27 July 2013.
33. ASF: Alaska Satellite Facility Vertex: ASF’s Data Portal V3.04-54. Available online: https://vertex.daac.asf.alaska.edu/ (accessed
on 8 September 2019).
34. Crosetto, M.; Monserrat, O.; Cuevas-González, M.; Devanthéry, N.; Crippa, B. Persistent scatterer interferometry: A review.
ISPRS J. Photogramm. Remote Sens. 2016, 115, 78–89. [CrossRef]
35. Hooper, A.; Zebker, H.; Segall, P.; Kampes, B. A new method for measuring deformation on volcanoes and other natural terrains
using InSAR persistent scatterers. Geophys. Res. Lett. 2004, 31, L23611. [CrossRef]
36. Notti, D.; Herrera, G.; Bianchini, S.; Meisina, C.; Carlos, J.; Zicca, F. A methodology for improving landslide PSI data analysis. Int.
J. Remote Sens. 2014, 35, 2186–2214. [CrossRef]
37. USGS United States Geological Survey Earth Explorer. Available online: https://earthexplorer.usgs.gov/ (accessed on 8
September 2019).
38. Holben, B.N. Characteristics of maximum-value composite images from temporal AVHRR data. Int. J. Remote Sens. 1986, 7,
1417–1434. [CrossRef]
39. NBSS: National Bureau of Soil Survey. Soil Map of Kerala; National Bureau of Soil Science and Land Use Planning: Nagpur, India,
1996.
40. JPL: Jet Propulsion Laboratory. Shuttle Radar Topography Mission, California Institute of Technology and the National Aeronau-
tics and Space Administration. Available online: https://www2.jpl.nasa.gov/srtm/ (accessed on 3 July 2014).
41. BIS: Bureau of Indian Standard. Indian Standard-Preparation of Landslide Hazard Zonation Maps in Mountainous Terrains-Guidelines;
IS 14496 (Part 2); BIS: New Delhi, India, 1998; p. 20.
42. Escobar-Wolf, R.; Sanders, J.; Vishnu, C.L.; Oommen, T.; Sajinkumar, K.S.A. GIS Tool for Infinite Slope Stability Analysis
(GIS-TISSA). Geosci. Front. 2020. [CrossRef]
43. Haneberg, W.C. PISA-m Map-based Probabilistic Infinite Slope Analysis: Version 1.0.1 User Manual [Updated March 2007]; Haneberg
Geoscience: Seattle, WA, USA, 2007; p. 20.
44. Stramondo, S.; Moro, M.; Tolomei, C.; Cinti, F.R.; Doumaz, F. InSAR surface displacement field and fault modelling for the 2003
Bam earthquake (southeastern Iran). J. Geodyn. 2005, 40, 347–353. [CrossRef]
45. Lambiel, C.; Delaloye, R.; Strozzi, T.; Lugon, R.; Raetzo, H. ERS InSAR for assessing rock glacier activity. In Proceedings of the
Ninth International Conference on Permafrost, Fairbanks, Alaska, 29 June–3 July 2008; Volume 1, pp. 1019–1025.
46. Hu, J.; Li, ZW.; Ding, X.L.; Zhu, J.J.; Zhang, L.; Sun, Q. 3D coseismic displacement of 2010 Darfield, New Zealand earthquake
estimated from multi-aperture InSAR and D-InSAR measurements. J. Geod. 2012, 86, 1029–1041. [CrossRef]
47. Ciampalini, A.; Raspini, F.; Lagomarsino, D.; Catani, F.; Casagli, N. Landslide susceptibility map refinement using PSInSAR data.
Remote Sens. Environ. 2016, 184, 302–315. [CrossRef]
48. Bianchini, S.; Cigna, F.; Righini, G.; Proietti, C.; Casagli, N. Landslide HotSpot Mapping by means of Persistent Scatterer
Interferometry. Environ. Earth Sci. 2012, 67, 1155–1172. [CrossRef]
49. Cigna, F.; Bianchini, S.; Casagli, N. How to assess landslide activity and intensity with Persistent Scatterer Interferometry (PSI):
The PSI-based matrix approach. Landslides 2013, 10, 267–283. [CrossRef]
Geomatics 2021, 1 17

50. Mateos, R.M.; Azañón, J.M.; Roldán, F.J.; Notti, D.; Pérez-Peña, V.; Galve, J.P.; Pérez-García, J.L.; Colomo, C.M.; Gómez-López,
J.M.; Montserrat, O.; et al. The combined use of PSInSAR and UAV photogrammetry techniques for the analysis of the kinematics
of a coastal landslide affecting an urban area (SE Spain). Landslides 2017, 14, 743–754. [CrossRef]
51. Oommen, T.; Cobin, P.F.; Gierke, J.S.; Sajinkumar, K.S. Significance of variable selection and scaling issues for probabilistic
modeling of rainfall-induced landslide susceptibility. Spat. Inf. Res. 2018, 26, 21–31. [CrossRef]