Landslide susceptibility mapping with Geographic Information System (GIS) combines maps of slope, rock type, rainfall, land use, and past landslides to show which hillsides are most likely to fail. For India’s hill roads, where the Geological Survey of India rates 15 percent of about 4.3 lakh sq km of mapped landslide-prone area as highly susceptible, these maps help engineers decide where to build, strengthen, and watch slopes.
Introduction: Why Landslides Are the Biggest Risk on India’s Hill Roads
India has nearly 16,788 km of National Highways in its hill states, according to the Ministry of Road Transport and Highways (MoRTH). Many of these roads are cut into steep, weak slopes that can give way during heavy rain. When a slope fails, it can cut off towns, villages, and pilgrimage sites that depend on a single road.
A 2024 study in the journal Natural Hazards and Earth System Sciences counted more than 300 landslides along the 247 km road between Rishikesh and Joshimath in Uttarakhand, after very heavy rain in September and October 2022. That works out to about 1.25 landslides for every kilometer of road, and sections where the road had been widened had about twice as many landslides after accounting for other factors.This shows that hill roads are not only damaged by landslides. Cutting into slopes to build or widen a road can also make landslides more likely.
| Aspect | Responding After Slopes Fail | Planning with Susceptibility Maps |
| When slopes are assessed | After a landslide blocks the road | Before alignment, cutting, and widening |
| Basis for decisions | Site visits after each event | Past landslides and ground conditions mapped together |
| Coverage | Individual failed slopes | Whole corridors and catchments |
| Setting priorities | Most recent or most visible failures | Slopes ranked by susceptibility class |
| Monitoring | Driven by complaints and closures | Focused on high-susceptibility slopes |
What Is Landslide Susceptibility Mapping with GIS?
Landslide susceptibility mapping shows how each part of a hillside is likely to fail, based on where landslides have happened before and the conditions that caused them. The National Disaster Management Authority (NDMA) describes it as using a record of past landslides to predict which areas are likely to see landslides in the future. It shows where the ground is weaker, not when a landslide will happen.
GIS makes this possible by stacking many maps of the same area, such as slope, rock type, rivers, roads, and land use. Each layer is scored for how much it adds to landslide risk, and the scores are combined into one map with zones from low to very high. ArcGIS Pro can build these maps using the terrain and overlay tools in ArcGIS Spatial Analyst.
Susceptibility is the first step toward a full picture of landslide danger. NDMA’s National Landslide Risk Management Strategy explains that hazard maps build on susceptibility maps by adding an estimate of how often landslides may occur.
India’s Landslide Mapping Framework: GSI’s NLSM, NRSC’s Landslide Atlas, and NDMA’s Strategy
The Geological Survey of India (GSI) has completed National Landslide Susceptibility Mapping (NLSM) at 1:50,000 scale for landslide-prone hill areas across parts of 19 states and union territories, covering about 4.3 lakh sq km. According to an August 2025 reply in Parliament, 56 percent of this area falls in the low zone, 29 percent in the moderate zone, and 15 percent in the high zone. In Uttarakhand, 22 percent of the mapped area is in the high zone, and in Himachal Pradesh, 29 percent.
GSI has also collected information on about 91,000 past landslides, of which 33,904 have been checked in the field. The maps and inventory are free to download from GSI’s National Geoscience Data Repository and Bhukosh portal, and can be viewed on the Bhusanket portal and the Bhooskhalan mobile app. GSI has now moved to more detailed 1:10,000 maps, completing 160 critical sectors by the 2024-25 field season, with a target of 200 by 2028.
The Landslide Atlas of India, prepared by the National Remote Sensing Centre (NRSC), ISRO, maps about 80,000 landslides from 1998 to 2022 and ranks 147 districts in 17 states and 2 union territories by their exposure to landslides. NDMA’s National Landslide Risk Management Strategy, published in September 2019, sets out which map scale suits which purpose, from national planning to the design of roads. In November 2024, a ₹1,000 crore National Landslide Risk Mitigation Project covering 15 states was approved.
National data is also available for quick screening. Indo ArcGIS Living Atlas includes the Indian Landslide Susceptibility Map, a 100 m map developed by the HydroSense Lab at IIT Delhi and described as India’s first national-scale machine learning landslide susceptibility map. It reports 95.73 percent accuracy on test data and classifies 4.75 percent of India’s area as very highly susceptible.
How GIS Builds a Landslide Susceptibility Map: From Inventory to Validated Zones
A susceptibility map is built in five steps, each adding a layer of evidence to the one before it.
Building the Landslide Inventory
Every model starts with a map of past landslides, because these show where slopes have actually failed. Landslide points or outlines can come from GSI’s inventory, satellite images, and field surveys. Some landslides are kept aside at this stage so the finished map can be tested later.
Stacking the Conditioning Factors
Conditioning factors are the ground conditions that make a slope more likely to fail. From a digital elevation model (DEM), the Slope, Aspect, and Curvature tools in ArcGIS Spatial Analyst work out how steep each slope is, which way it faces, and how its surface bends. Rock type, distance to rivers, distance to roads, land use, and rainfall are added as further layers, and the Reclassify tool puts every layer on a common scale.
Weighting the Factors
Not every factor matters equally, so each one is given a weight. The table below compares three common ways to do this in ArcGIS Pro.
| Aspect | Expert Weighting | Statistical Weighting | Machine Learning |
| How weights are set | Experts rank the factors, often using the analytical hierarchy process (AHP) | Weights come from how often past landslides fall in each class of a factor, such as the frequency ratio method | A model learns patterns from past landslides |
| ArcGIS Pro tools | Weighted Overlay, Fuzzy Overlay | Raster Calculator | Forest-based and Boosted Classification and Regression, Presence-only Prediction (MaxEnt) |
| Works best when | Few landslides have been recorded | A good landslide inventory exists | Large areas with large inventories are being mapped |
Expert weighting is quick and easy to explain, but it depends on the judgment of the team. Statistical and machine learning methods lean more on data, which makes a strong landslide inventory essential. Machine learning methods can draw on AI capabilities within ArcGIS to find patterns across many factors at once.
Drawing the Zones
The weighted layers are combined into one score for every cell on the map. These scores are then grouped into classes such as low, moderate, high, and very high susceptibility. ArcGIS 3D Analyst can show these zones on a 3D view of the terrain, which helps engineers see how they fall along a road.
Checking the Map
A susceptibility map is only useful if it predicts well. The landslides kept aside earlier are compared with the zones, and a good map places most of them in the high and very high classes. This test is often summarized as the area under the curve (AUC), where a value closer to 1 means a better prediction.
Inside an ArcGIS Landslide Zonation Study in the Uttarakhand Himalaya
A research paper published in ArcIndia News in 2012 showed how this workflow was applied in the Ganeshganga watershed, an area of about 50 sq km in Chamoli district, Uttarakhand. The authors, including researchers from the Department of Geology at Kumaun University and the Defence Institute of Geospatial Information and Training, studied a watershed of rugged mountains and steep slopes covered with loose soil.
The team built 12 factor layers, including slope, aspect, rock type, drainage, land use, and distance to road. Slope and aspect came from a DEM made from Cartosat-1 satellite data, and rock type came from GSI’s geological map. They used ArcGIS 9.3 to manage, combine, and analyze the layers.
Each factor was weighted using a modified information value (InfoVal) method, a statistical approach, and the result was divided into four susceptibility zones. When tested against known landslides, the map gave a prediction accuracy of 91.74 percent. The same workflow can run in ArcGIS Pro today, with the DEM, factor layers, weights, and validation handled in one project.
Large-scale landslide work in India has also been recognized. NRSC received Esri’s Special Achievement in GIS Award in 2023 for its inventory of about 80,000 landslides, and for machine learning susceptibility maps of tourist corridors in Himachal Pradesh and pilgrimage road corridors in Uttarakhand. NRSC’s Landslide Atlas also describes susceptibility zonation along about 2,000 km of routes in the two states, including the roads to Badrinath, Kedarnath, and Gangotri.
From Susceptibility Map to Safer Roads: Alignment, Slope Treatment, and Monitoring
A susceptibility map earns its value when it shapes decisions on the ground, from where a road goes to which slopes are treated and watched.
Choosing the Alignment
MoRTH says GSI’s geological maps and NLSM are now used in the planning and alignment survey stages of national highway projects in fragile hill areas. Placing planned routes over susceptibility zones lets engineers compare options and avoid the weakest slopes where possible. NDMA’s strategy matches map scale to the task, with 1:25,000 to 1:5,000 maps for advanced planning of roads and more detailed maps for their design.
Treating the Slopes
For slopes already at risk, MoRTH reports that drones, Light Detection and Ranging (LiDAR) surveys, and digital terrain models are used to study them in detail. In Uttarakhand alone, 58 landslide-prone locations have been treated, 96 are under mitigation works, and investigations or project reports are under way at 104 more. ArcGIS Drone2Map and Site Scan for ArcGIS can turn drone images into detailed surface models.
ArcGIS Reality can then build 3D models of a slope face. These models help engineers plan where retaining walls, rockfall nets, or drains are needed.
Watching the Slopes
A susceptibility map shows which slopes are weak, but not which ones are moving now. MoRTH is deploying Interferometric Synthetic Aperture Radar (InSAR)-based landslide monitoring, a satellite radar method that detects small ground movements, along a 100 km stretch of the Char Dham route. It also plans a real-time warning and alert system on the Parwanoo-Solan section of NH-5. ArcGIS Image Server can manage and analyze satellite images over time, so slope changes can be compared from one season to the next.
Tracking the Work
Field engineers can log cracks, seepage, and fresh slips with photos in ArcGIS Survey123. ArcGIS Dashboards can then bring treated sites, active slopes, and susceptibility zones together on one screen for highway engineers and district officials.
Road agencies can explore Roads and Highways solutions, and disaster management teams can explore Emergency Management solutions, from Esri India to plan safer hill corridors.
Challenges and the Road Ahead
Gaps in the Landslide Record
Many landslides in remote valleys are never recorded, and every model depends on this record. GSI adds new landslides each year from its post-disaster studies, but some areas still have few mapped events. Regular reporting by road teams in the field can help fill these gaps before the next model update.
Matching the Map to the Decision
National and regional maps are made for screening, not for designing a single retaining wall. GSI’s 1:50,000 maps and the 100 m IIT Delhi layer help compare corridors, while alignment and slope design need 1:10,000 maps, LiDAR, and drone surveys. Using a map at the wrong scale can hide risks on individual slopes.
Thin Rainfall Data on High Slopes
Heavy rain is a major trigger for landslides in the hills, but it is measured at few points in the mountains. A 2026 GSI presentation at a National Institute of Disaster Management (NIDM) session noted only 518 automatic weather stations in hilly areas, about one for every 830 sq km. Without closer rainfall records, it is harder to judge how much rain a slope can take before it fails.
Maps That Need Updating
Slopes change after every monsoon, and roads change when they are widened or realigned. A map made before widening may miss the added risk the Rishikesh-Joshimath study points to. Updating susceptibility maps after each season keeps them useful for the next one.
Landslides will remain part of life on India’s hill roads, but the slopes most likely to fail can be mapped before work begins rather than discovered after a road is cut off. GSI’s national and sector maps, new monitoring along key routes, and GIS tools that bring these together give road agencies a clearer view of which slopes need attention first. For the people and pilgrims who rely on these roads, that can mean fewer closures and safer journeys through every monsoon.
FAQs
1.What is landslide susceptibility mapping?
It is a map that shows how likely each area is to have landslides, based on past landslides and ground conditions such as slope and rock type. Areas are grouped into zones from low to very high susceptibility.
2.How is GIS used for landslide susceptibility mapping?
GIS stacks layers such as slope, rock type, drainage, roads, and land use, gives each a weight, and combines them into one map. The map is then checked against known landslides to test how well it predicts.
3.Which factors are used in landslide susceptibility mapping?
Common factors include slope, aspect, curvature, rock type, distance to rivers, distance to roads, land use, and rainfall. The exact set depends on the area and the data available.
4.How does landslide susceptibility mapping help in hill road planning?
It shows which slopes along a planned or existing road are most likely to fail. Engineers can use it to compare route options, decide where to treat slopes, and choose where to monitor ground movement.
5.Which ArcGIS tools are used for landslide susceptibility analysis?
ArcGIS Pro with ArcGIS Spatial Analyst offers tools such as Slope, Aspect, Curvature, Reclassify, Weighted Overlay, and Raster Calculator. For machine learning models, Presence-only Prediction and the Forest-based and Boosted Classification and Regression tool can also be used.
Written by
Esri India Marketing
