#IMC: September 14 - 18 2025

Machine learning based characterization of landslides in north-western Himalayas

Assigned Session: FS 3.147: Monitoring, Assessment, and Multi-Scale Modeling of Landslides: Approaches for Hazard Analysis

Abstract ID: 3.13315 | Accepted as Poster | Talk/Oral | TBA | TBA

Ankit Singh (0)
Dhiman, Nitesh (1), Praise Shukla, Dericks (1)
Ankit Singh ((0) Mandi, A9 Innovation centre North Campus KAMAND, 175005, Mandi, HIMACHAL PRADESH, IN)
Dhiman, Nitesh (1), Praise Shukla, Dericks (1)

(0) Mandi, A9 Innovation centre North Campus KAMAND, 175005, Mandi, HIMACHAL PRADESH, IN
(1) Indian Institute of Technology Mandi, A9 Innovation centre North Campus KAMAND

(1) Indian Institute of Technology Mandi, A9 Innovation centre North Campus KAMAND

Categories: Hazards
Keywords: Landslide, Characterization

Categories: Hazards
Keywords: Landslide, Characterization

Download
Download

Climate change coupled with global warming has led to an increase in natural hazards in mountainous regions of the world. Among these hazards, landslides are commonly occurring due to the presence of steep and fragile slopes. Minimizing the losses caused by landslides is important for effective planning and prevention strategies. Identifying landslides leads to studies related to landslide susceptibility mapping, which requires a landslide inventory. However, inventories largely do not take into account the types of landslides and their characteristics, resulting in generalized outcomes that only identify probable landslides.

This study aims to characterize landslides using three machine learning methods: logistic regression, bagging, and J48, while also assessing the capability of transfer learning to predict landslides in unknown regions with similar topographical characteristics. The results showed that J48 and bagging performed better in characterizing landslides. Additionally, transfer learning was effective in predicting landslide characteristics for Kullu district (target) based on knowledge acquired from Mandi district (source).

N/A
NAME:
TBA
BUILDING:
TBA
FLOOR:
TBA
TYPE:
TBA
CAPACITY:
TBA
ACCESS:
TBA
ADDITIONAL:
TBA
FIND ME:
>> Google Maps

Limits: min. 3 words, max. 30 words or 200 characters

Choose the session you want to submit an abstract. Please be assured that similar sessions will either be scheduled consecutively or merged once the abstract submission phase is completed.

Select your preferred presentation mode
Please visit the session format page to get a detailed view on the presentation timings
The final decision on oral/poster is made by the (Co-)Conveners and will be communicated via your My#IMC dashboard

Please add here your abstract meeting the following requirements:
NO REFERNCES/KEYWORDS/ACKNOWEDGEMENTS IN AN ABSTRACT!
Limits: min 100 words, max 350 words or 2500 characters incl. tabs
Criteria: use only UTF-8 HTML character set, no equations/special characters/coding
Copy/Paste from an external editor is possible but check/reformat your text before submitting (e.g. bullet points, returns, aso)

Add here affiliations (max. 30) for you and your co-author(s). Use the row number to assign the affiliation to you and your co-author(s).
When you hover over the row number you are able to change the order of the affiliation list.

1
1
Add Affiliation(s)

Add here co-author(s) (max. 30) to your abstract. Please assign the affiliation(s) of each co-author in the "Assigned Aff. No" by using the corresponding numbers from the "Affiliation List" (e.g.: 1,2,...)
When you hover over the row number you are able to change the order of the co-author list.

1
2
1
Add Co-Author(s)
1
2
1
Add Keyword
Close