Object-oriented approach for landslide mapping using wavelet transform coupled with machine learning : a case study of western ghats, India
By: Rana, Himanshu
.
Contributor(s): Babu, Sivakumar G. L.
Publisher: New York Springer 2022Edition: Vol.52(3), June.Description: 691-706p.Subject(s): Civil EngineeringOnline resources: Click here
In:
Indian geotechnical journalSummary: The hills nested in the Western Ghats, India, experience recurrent landslides during the monsoon season every year and draw grave concern owing to the damage and disruption of traffic, which necessitates the evaluation of landslide hazard and risk. Preparation of landslide inventory maps is a prerequisite in the assessment of landslide hazards and risks. The conventional methods for acquiring landslide inventory maps involve extensive field surveys and photograph interpretation, making it tedious and highly reliant on expert judgment. To this aim, an object-based approach is proposed to prepare landslide inventory map, which ensures faster acquisition and assimilation of landslide data. As a first step toward the approach, the digital terrain model (DTM) is obtained from unmanned aerial vehicle (UAV) data for the study area. Wavelet transform technique is performed in tandem with machine learning (ML) algorithms (random forest (RF) and support vector machine (SVM)) to measure the texture of DTM and to train and predict the landslide objects. A small area (area = 1.16 km2) from the Western Ghats, India, is selected for this case study. The results indicate that the RF and SVM algorithms predict landslide objects with 88.64% and 87.45% accuracy, respectively. This study also investigated the effect of texture (wavelet coefficient) on the accuracy of ML algorithms. An important observation from this study is that the curvature mean is the most influential object feature to demarcate a landslide event. The proposed approach generates output in the form of landslide inventory map for the study area.
| Item type | Current location | Call number | Status | Date due | Barcode | Item holds |
|---|---|---|---|---|---|---|
Articles Abstract Database
|
School of Engineering & Technology (PG) Archieval Section | Not for loan | 2022-1548 |
The hills nested in the Western Ghats, India, experience recurrent landslides during the monsoon season every year and draw grave concern owing to the damage and disruption of traffic, which necessitates the evaluation of landslide hazard and risk. Preparation of landslide inventory maps is a prerequisite in the assessment of landslide hazards and risks. The conventional methods for acquiring landslide inventory maps involve extensive field surveys and photograph interpretation, making it tedious and highly reliant on expert judgment. To this aim, an object-based approach is proposed to prepare landslide inventory map, which ensures faster acquisition and assimilation of landslide data. As a first step toward the approach, the digital terrain model (DTM) is obtained from unmanned aerial vehicle (UAV) data for the study area. Wavelet transform technique is performed in tandem with machine learning (ML) algorithms (random forest (RF) and support vector machine (SVM)) to measure the texture of DTM and to train and predict the landslide objects. A small area (area = 1.16 km2) from the Western Ghats, India, is selected for this case study. The results indicate that the RF and SVM algorithms predict landslide objects with 88.64% and 87.45% accuracy, respectively. This study also investigated the effect of texture (wavelet coefficient) on the accuracy of ML algorithms. An important observation from this study is that the curvature mean is the most influential object feature to demarcate a landslide event. The proposed approach generates output in the form of landslide inventory map for the study area.
Click on an image to view it in the image viewer
There are no comments for this item.