Differential Interferometric Synthetic Aperture Radar-Based

Landslide Monitoring: A Case Study of Wayanad, India

Manvi Kanwar

and Samsung Lim

School of Civil and Environmental Engineering, University of New South Wales, Australia

Associate Professor, School of Civil and Environmental Engineering, University of New South Wales, 2052, Australia

Keywords: Rainfall-Induced Landslides, D-Insar, Sentinel-1, Remote Sensing.

Abstract: On July 30th, 2024, the Wayanad region of Kerala, India, experienced a devastating landslide triggered by

intense monsoon rainfall. This event highlighted the region's need for effective landslide monitoring and early

warning systems. Differential Interferometric Synthetic Aperture Radar (DInSAR) is a powerful and efficient

remote sensing technique for detecting ground deformation and monitoring landslides. This study utilizes

Sentinel-1 SAR images to monitor the Wayanad landslide by processing and analyzing SAR data using the

DInSAR technique to identify ground displacement patterns. The results demonstrate the effectiveness of

DInSAR in capturing pre- and post-event deformations, offering valuable insights into the landslide dynamics.

This study underscores the potential of SAR-based real-time landslide monitoring and risk mitigation in

landslide-prone regions.

1 INTRODUCTION

Wayanad, a district in Kerala's Western Ghats, India,

is highly susceptible to landslides, a hazard

exacerbated by both natural and anthropogenic

factors. The region's topography, characterized by

elevated slopes and a soil cover consisting mainly of

boulders, colluvium, and laterite, combined with the

region's propensity for heavy monsoon rainfall,

makes it particularly vulnerable to landslide events

(Kuriakose et al., 2009). On July 30th, 2024, a massive

rainfall event which brought over 14 cm of rain in a

single day — 493% above the average — triggered

multiple landslides in Chooralmala, Mundakkai, and

Vellarimala villages (Mishra, 2024). These landslides are

not isolated; Wayanad has a history of similar events,

such as the over 3,000 landslides reported during the

2018 floods and the multiple landslides in Puthumala

in 2019 (National Remote Sensing Centre, 2023).

While natural factors such as topography and

rainfall primarily contribute to landslides,

anthropogenic activities have increasingly

significantly exacerbated these events. Studies

following the 2018 and 2019 floods identified

extensive land-use changes, particularly the

https://orcid.org/0009-0006-6680-6068

https://orcid.org/0000-0001-9838-8960

establishment of tea and rubber plantations and the

construction of houses and other infrastructure in

landslide-prone areas (Premlet B, 2019). These

activities obstruct natural drainage systems and

saturate the soil, making it more susceptible to

landslides during heavy rains. Moreover, the velocity

of rivers such as Iruvazhinji, which runs through steep

terrain before descending rapidly in elevation, can

intensify the impact of floods and landslides.

Kerala's Western Ghats, including Wayanad, have

long been recognised as vulnerable. In 2011, the

Madhav Gadgil-led Western Ghats Ecology Expert

Panel recommended that a sizeable portion of the

Western Ghats, including areas in Wayanad, be

designated as Ecologically Sensitive Zones (ESZ)

(Gadgil Madhav, 2011). This designation would have

restricted industrial activities, mining, and

construction in these areas. However, these

recommendations were not implemented, exposing

the region to the ongoing risks of landslides and other

environmental disasters.

In areas affected by excessive rainfall, using

optical images to determine landslide extent can be

severely hindered by cloud cover and adverse weather

conditions. However, microwave-based Synthetic

Kanwar, M. and Lim, S.

Differential Interferometric Synthetic Aperture Radar-Based Landslide Monitoring: A Case Study of Wayanad, India.

DOI: 10.5220/0013080700003935

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 11th International Conference on Geographical Information Systems Theory, Applications and Management (GISTAM 2025), pages 113-120

ISBN: 978-989-758-741-2; ISSN: 2184-500X

113

Aperture Radar (SAR) images offer a significant

advantage in such scenarios. Unlike optical imagery,

SAR can penetrate clouds and provide consistent data

irrespective of weather conditions, making it an

invaluable tool for monitoring large and inaccessible

areas quickly and reliably. This advantage is evident

in recent studies (Debevec Jordanova et al., 2024; Lau

et al., 2024), demonstrating DInSAR's utility in

mapping and monitoring landslides. Similarly, this

research applies DInSAR to assess the Wayanad

landslide, a rainfall-induced event, by leveraging the

phase differences between SAR images captured at

different times. While other studies have used

advanced techniques like SBAS or integrated in-situ

monitoring, this study focuses on rapid assessment of

ground deformation using Sentinel-1 SAR data,

which is freely accessible. This approach offers

valuable insights into hazard assessment and disaster

management, allowing response teams to prioritise

the most vulnerable areas and overcome the

limitations of traditional optical methods.

2 STUDY AREA

Wayanad District, located in the northeastern part of

Kerala, India, is known for its diverse geography and

significant natural beauty. The district is

characterized by its hilly terrain, with elevations

ranging from 700 to 2,100 meters above sea level and

is part of the Western Ghats Mountain range.

Wayanad’s landscape includes dense forests, lush

green tea and coffee plantations, and numerous rivers

and waterfalls. The region experiences a tropical

monsoon climate, with an average annual rainfall of

2,500 to 3,000 millimetres (India Meteorological

Department, 2024), contributing to its rich

biodiversity and dense vegetation. It is bordered by

the districts of Kannur to the west, Kozhikode to the

south, and Malappuram to the southwest. To the east,

Wayanad shares its boundaries with the state of

Karnataka. The district's geographical and climatic

conditions make it highly susceptible to landslides,

particularly during the monsoon season, which

highlights the importance of effective monitoring and

disaster management in the region. The present study

focuses on one sub-swath of a Sentinel-1 SAR image,

which encapsulates approximately 20,825 km² of

area, as shown in Figure 1.

Figure 1: Study Area.

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

114

Table 1: Data Characteristics of the Pre-Event Sar Images Using Stack Overview in Snap.

Acquisition date Track Orbit Temporal

Baseline (days)

Perpendicular

Baseline (m)

Modelled

Coherence

8 July 2024 (reference) 165 54662 0 0 1

20 July 2024 (secondary) 165 54837 12 -123 0.89

Table 2: Data Characteristics of the Post-Event Sar Images Using Stack Overview in Snap.

Acquisition date Track Orbit Temporal Baseline

(days)

Perpendicular

Baseline (m)

Modelled

Coherence

August

2024 (reference) 165 55012 0 0 1

13 August 2024 (secondary) 165 55187 12 -142 0.87

3 METHODOLOGY

3.1 Data Acquisition

This research leverages SAR datasets obtained from

the Sentinel-1A and 1B missions, launched in 2014

and 2016. These satellites are equipped with C-band

SAR technology, which offers open-access data

capable of capturing Earth's surface imagery under all

weather conditions, day and night. Sentinel-1

provides data well suited for medium-resolution

applications, with a revisit period of 12 days,

allowing for extensive area coverage. The satellites

operate in the Interferometric Wide (IW) swath mode,

utilizing Terrain Observation by Progressive Scans

(TOPS) to achieve a ground coverage of

approximately 250 km. The geometric resolution of

the SAR products is 5 m by 20 m in range and

azimuth, respectively.

For the application of the D-InSAR method, two

images were acquired before the landslide event

happened, and two were after the Wayanad landslide

to demonstrate the changes in the landform. Data

processing was performed using SeNtinel

Applications Platform (SNAP) version 10.0.0, a free

software provided by the European Space Agency

(ESA). The SAR data was sourced from the Alaska

Satellite Facility (ASF) and provided by the European

Space Agency (2023) as part of the Sentinel-1

mission. Each Sentinel-1 Single Look Complex

(SLC) product consists of three Interferometric Wide

Swaths (IW1, IW2, and IW3) and nine bursts. The

IW3 sub-swath was selected for this study as it

encompasses the area of interest. The acquired images

were in both VH and VV polarization modes. Tables

1 and 2 present the features of the pre-event and post-

event datasets, respectively.

The perpendicular baseline was kept within 200

meters to maintain data quality. This restriction helps

ensure accurate interferometric measurements and

reliable results in SAR data processing. The

perpendicular baseline is a critical parameter in SAR

interferometry, affecting the interferograms' spatial

resolution and coherence (Li & Bethel, 2008).

Keeping it within this limit helps minimize errors and

maintain the overall quality of the interferometric

data (Zhang et al., 2005).

3.2 Preprocessing

Preprocessing of the Sentinel-1 SAR data involves

several steps to ensure the accuracy of the DInSAR

analysis:

Orbit File Application: Accurate orbit data are

applied to the SAR images to correct for satellite

position errors. Precise orbit files provided by ESA

were utilized to ensure accurate geolocation and

alignment of the Sentinel-1 imagery.

Radiometric Calibration: This step involves

converting the SAR image data from digital numbers

(DN) to radar backscatter intensity, which is essential

for quantitative analysis.

Multilooking: The SAR images are multilooked to

reduce speckle noise inherent in SAR data by

averaging adjacent pixels. This study used a multi-

look factor of 4 in range and 1 in azimuth, reducing

noise while maintaining spatial resolution suitable for

analyzing landslide deformation.

Differential Interferometric Synthetic Aperture Radar-Based Landslide Monitoring: A Case Study of Wayanad, India

115

Co-registration: Pre and post-event SAR images are

precisely aligned to ensure accurate phase difference

calculations. This step minimizes phase errors due to

misalignment.

Interferogram Generation: An interferogram is

generated by computing the phase difference between

the co-registered SAR images. This interferogram

represents the ground displacement in the radar line

of sight (LOS).

3.3 DInSAR Processing

The core of the DInSAR technique involves the

following steps:

Phase Unwrapping: The interferometric phase is

wrapped between -π and π, necessitating a phase

unwrapping process to retrieve the actual

displacement values. The Statistical-cost Network-

flow Algorithm for Phase Unwrapping (SNAPHU)

algorithm is commonly used. This study employed

SNAPHU with a coherence mask to restrict

unwrapping to areas with a coherence threshold of

0.3, ensuring accurate displacement retrieval in high-

coherence regions.

Topographic Phase Removal: The influence of

topography on the interferometric phase is removed

using a Digital Elevation Model (DEM), such as the

Shuttle Radar Topography Mission (SRTM) DEM, to

isolate the phase related to ground deformation. A 30

m resolution SRTM DEM was used in this study to

ensure the precise removal of the topographic

component in the Wayanad region, characterized by

hilly terrain.

Temporal and Spatial Filtering: Temporal and

spatial filtering techniques reduce noise and enhance

the signal related to ground deformation. With an

adaptive window size of 32 x 32 pixels, the Goldstein

filter preserves deformation-related signals while

reducing noise.

Displacement Map Generation: The unwrapped

phase is converted into displacement maps

representing the ground movement in the radar LOS

direction. These maps are then geocoded to produce

spatially accurate deformation maps. This study

projected the displacement maps onto the WGS84

coordinate system for spatial analysis and visual

interpretation of ground deformation pattern.

4 RESULTS AND DISCUSSIONS

The results obtained after the dataset's preprocessing

demonstrate the landform change before and after the

event. The LOS displacement and coherence maps

were generated. This analysis helps us showcase the

land movement that happened towards or away from

the radar line of sight. P et al. (2025) used machine

learning models to determine the landslide

susceptibility in the Wayanad district and found that

the western part was most susceptible to landslides.

Thus, the coherence and displacement values of these

areas are discussed.

The observation from the Maps in Figures 2 to 5

from the estimated coherence suggests several

important conclusions:

4.1 Coherence Analysis

Before the Landslide: The pre-landslide surface

stability of the Wayanad region is evident from the

coherence map in Figure 2, where 57.8% of the pixels

exhibit coherence values above 0.6. This indicates

predominantly stable ground. 25.3% of the region

displays high coherence values (0.8–1.0), typically

associated with undisturbed surfaces like rocky

terrain or dense vegetation. Such areas reflect

minimal displacement or deformation.

Conversely, regions with coherence values below

0.4, constituting 15.7% of the pixels, primarily occur

in riverbeds and forested areas, as seen from the

histogram in Figure 3. These zones experience

temporal decorrelation due to vegetation growth, soil

moisture variations, or minor ground movements,

which are unstable precursors. Some areas' relatively

lower coherence values (0–0.53) suggest potential

surface instability or subtle deformations, even before

the landslide event.

After the Landslide: Following the landslide

triggered by intense monsoon rainfall, significant

displacement and surface disruption are observed,

particularly in areas near Mundakkai Hills,

Punchrimattom, Mundakkai, and Chooralmala

(Figure 4). These locations were selected for

monitoring due to their proximity to steep slopes,

historical landslide occurrences, and the presence of

vulnerable settlements and infrastructure.

The coherence values in these regions decreased

notably, reflecting ground instability and

displacement. At Mundakkai Hills, coherence

dropped from an average of 0.75 (pre-event) to 0.42

(post-event). Similarly, Punchrimattom experienced

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

116

a reduction from 0.81 to 0.46, and Mundakkai

decreased from 0.84 to 0.48. The most severe

displacement occurred near Chooralmala, where

coherence dropped from 0.77 to 0.32, indicating

significant material flow and surface disruption.

Figure 5 demonstrates only 34.2% of the pixels-

maintained coherence above 0.6 after the landslide,

compared to 57.8% before the event. Regions with

coherence values in the range of 0.2–0.4 increased

significantly, from 9.8% (pre-event) to 28.6% (post-

event), indicating widespread ground displacement.

Additionally, areas with coherence below 0.2,

representing 12.3% of the pixels, correspond to zones

of extreme surface disturbance caused by material

displacement during the landslide.

The spatial analysis of displacement highlights

the utility of SAR-based techniques in mapping and

quantifying ground movement. The coherence maps

provide a reliable measure of surface stability, with

changes in coherence values reflecting the severity

and extent of the landslide. These findings align with

observations by (Ferretti et al., 2001), who

emphasized the effectiveness of SAR coherence

analysis in monitoring ground deformation following

natural disasters.

From all above, it can be observed that the

coherence analysis underscores the substantial

alterations in surface properties caused by the

landslide, with significant changes in coherence

values reflecting both the extent of the ground

displacement and the areas of severe surface

disturbance. These results are critical for

understanding the landslide's overall impact, as they

provide detailed insights into the spatial distribution

of ground instability and the transformation of surface

conditions. The observed increase in coherence in

certain regions post-event suggests the exposure of

more stable and homogeneous surfaces, likely due to

the compaction and redistribution of material during

the landslide. This analysis highlights the event's

severity and establishes a reliable foundation for

developing future monitoring strategies, enabling

better risk assessment, mitigation efforts, and

informed planning for disaster-prone regions.

4.2 Displacement and Surface Stability

Figures 6 and 7 illustrate the LOS displacement maps

of the Wayanad region before and after the landslide

event, respectively. These maps provide critical

insights into the extent and movement of the landslide

as detected by the radar system. The LOS

displacement, measured in the radar line-of-sight

direction, highlights surface movement both towards

and away from the radar sensor.

Before the Landslide: The displacement map before

the event, shown in Figure 6, indicates stable surface

conditions with relatively uniform displacement

values across the region. Displacement values

predominantly range between -0.0014 m and 0.0048

m, suggesting minor ongoing ground movement

potentially related to environmental or geological

activity. Areas such as Mundakkai Hills,

Punchrimattom, and Chooralmala exhibit minimal

displacement, underscoring the absence of significant

pre-event deformation.

After the Landslide: The post-event displacement

map, shown in Figure 7, demonstrates substantial

changes in surface movement triggered by the

landslide. Negative displacement values (presented in

blue) represent movement away from the radar sensor

and are prominently observed in the areas most

impacted by the landslide, such as Chooralmala.

Maximum negative displacement values reach

approximately -0.0118 m, reflecting the material's

downward and outward movement from the hillside

during the landslide's initiation. In contrast, high

positive displacement values (depicted in yellow)

correspond to regions where the surface moved

towards the radar sensor, likely indicating debris

accumulation. This phenomenon is evident in the

towns of Mundakkai and Chooralmala, where

positive displacement values of up to 0.0118 m are

observed, suggesting debris flow and subsequent

compaction. The displacement maps effectively

capture the dynamic nature of the landslide. The pre-

event SAR images reveal small-scale displacements

indicative of ongoing ground activity, while the post-

event deformation map highlights pronounced

changes resulting from the landslide. The stark

contrast between the negative and positive

displacement values represents the landslide’s

impact, illustrating the initiation of material flow

from the hillside and the accumulation of debris in

surrounding areas, including Mundakkai and

Chooralmala. These results underscore the utility of

InSAR-derived LOS displacement data in detecting

and quantifying surface changes. The analysis

demonstrates the radar system’s capability to track

the landslide's initiation and the subsequent debris

flow, providing valuable insights into the event's

impact and progression.

Differential Interferometric Synthetic Aperture Radar-Based Landslide Monitoring: A Case Study of Wayanad, India

117

Figure 2: Coherence monitored before the landslide event.

Figure 3: Pixel-based analysis of coherence before the landslide event.

Figure 4: Coherence monitored after the landslide event.

1837

2491

3246

4105

4952

5701

6310

5473

4295

3019

2000

4000

6000

8000

Pixel Count

Coherence Range

Pixel Coherence Before the Landslide event

0.0 - 0.1 0.1 - 0.2 0.2 - 0.3 0.3 - 0.4 0.4 - 0.5

0.5 - 0.6 0.6 - 0.7 0.7 - 0.8 0.8 - 0.9 0.9 - 1.0

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

118

Figure 5: Pixel-based analysis of coherence after the landslide event.

Figure 6: Displacement towards radar LOS before the landslide event.

Figure 7: Displacement towards radar LOS after the landslide event.

2114

2785

3398

4321

5223

6012

6584

5547

4198

3109

2000

4000

6000

8000

Pixel Count

Coherence values

Pixel Coherence after the Landslide event

0.0 - 0.1 0.1 - 0.2 0.2 - 0.3 0.3 - 0.4 0.4 - 0.5

0.5 - 0.6 0.6 - 0.7 0.7 - 0.8 0.8 - 0.9 0.9 - 1.0

Differential Interferometric Synthetic Aperture Radar-Based Landslide Monitoring: A Case Study of Wayanad, India

119

5 CONCLUSION

This study demonstrated the successful application of

DInSAR using Sentinel-1 SAR data to monitor the

Wayanad landslide of July 30, 2024. By analyzing

four SAR images captured before and after the event,

coherence and LOS displacement maps were

generated, providing a detailed assessment of the

landslide’s extent and dynamics. The pre-event

coherence map revealed stable surface conditions,

with 57.8% of pixels maintaining coherence above

0.6. In contrast, the post-event coherence map

indicated significant ground deformation, with

coherence values above 0.6 dropping to 34.2% and

regions with coherence below 0.2 increasing to

12.3%. These changes highlight the substantial

disturbance caused by the landslide.

The LOS displacement map further quantified

surface movement, with negative values (up to -0.011

m) reflecting downward displacement in the affected

areas and positive values indicating debris

accumulation. This displacement pattern revealed the

spatial dynamics of the landslide, from initial material

displacement to subsequent debris deposition in the

Mundakkai, Punchrimattom, and Chooralmala

regions.

Triggered by slope instability due to intensified

rainfall, the Wayanad landslide demonstrates the

utility of DInSAR for rapid and accurate assessment

of rainfall-induced landslides. The technique

effectively detected and quantified ground

deformation, providing critical insights into the

event's mechanisms and extent. These results

emphasize the importance of integrating SAR-based

remote sensing into landslide monitoring systems,

enabling timely assessments for emergency response

and long-term hazard mitigation. Future studies could

incorporate advanced InSAR techniques to monitor

the annual rate of slope movement and improve early

warning systems in regions prone to recurring

landslides. This research highlights the potential of

DInSAR to enhance our understanding of landslide

dynamics and support disaster management efforts in

vulnerable areas.

REFERENCES

Debevec Jordanova, G., Popović, Z., Yastika, P. E.,

Shimizu, N., Oštir, K., & Verbovšek, T. (2024). SBAS

DInSAR and in situ monitoring of the Šumljak

landslide (SW Slovenia) dynamics driven by rainfall

and piezometric-level fluctuation. Landslides.

https://doi.org/10.1007/s10346-024-02408-4

European Space Agency (ESA). SNAP-Sentinel

Application Platform, v.10.0.0. https://step.esa.int/

main/toolboxes/snap.

Ferretti, A., Prati, C., & Rocca, F. (2001). Permanent

scatterers in SAR interferometry. IEEE Transactions on

Geoscience and Remote Sensing, 39(1), 8–20.

https://doi.org/10.1109/36.898661

Gadgil Madhav. (2011). Report of the Western Ghats

Ecology Expert Panel Part I.

India Meteorological Department. (2024). Cumulative

Rainfall For Kerala-Sw Monsoon 2024.

Kuriakose, S. L., Sankar, G., & Muraleedharan, C. (2009).

History of landslide susceptibility and a chorology of

landslide-prone areas in the Western Ghats of Kerala,

India. Environmental Geology, 57(7), 1553–1568.

https://doi.org/10.1007/s00254-008-1431-9

Lau, R., Seguí, C., Waterman, T., Chaney, N., & Veveakis,

M. (2024). InSAR-informed in situ monitoring for

deep-seated landslides: insights from El Forn

(Andorra). Natural Hazards and Earth System

Sciences, 24(10), 3651–3661. https://doi.org/10.5194/

nhess-24-3651-2024

Li, Z., & Bethel, J. (2008). Image Coregistration In Sar

Interferometry.

Mishra, A. (2024, August 1). Explained with maps: How

topography, human density led to devastation in

Kerala’s Wayanad. The Print. https://theprint.in/the

print-essential/explained-with-maps-how-topography-

human-density-led-to-devastation-in-keralas-wayanad/

2202631/

National Remote Sensing Centre, I. S. R. O. (2023).

Landslide Atlas of India (Mapping, Monitoring and

R&D studies using Remote Sensing data).

https://www.nrsc.gov.in

P, L., C, M., Mathew, A., & Shekar, P. R. (2025). Machine

learning and deep learning-based landslide

susceptibility mapping using geospatial techniques in

Wayanad, Kerala state, India. HydroResearch, 8, 113–

126. https://doi.org/10.1016/j.hydres.2024.10.001

Premlet B. (2019). Landslides In Kerala 2018 Chair,

Educational activities IEEE Kerala Section (Vol. 23).

Zhang, H., Zeng, Q., Liu, Y., Li, X., & Gao, L. (2005). The

Optimum Selection of Common Master Image for Series

of Differential SAR Processing to Estimate Long and

Slow Ground Deformation. Master Image for Series of

Differential SAR Processing to Estimate Long and Slow

Ground Deformation. http://dx.doi.org/10.1109/IGAR

SS.2005.1526688.

GISTAM 2025 - 11th International Conference on Geographical Information Systems Theory, Applications and Management

120