Abstract
Landslides are natural disasters that have resulted in the loss of economies and lives over the years. The landslides caused by the 2005 Muzaffarabad earthquake heavily impacted the area, and slopes in the region have become unstable. This research was carried out to find out which areas, as in Muzaffarabad district, are sensitive to landslides and to define the relationship between landslides and geo-environmental factors using three tree-based classifiers, namely, Extreme Gradient Boosting (XGBoost), Random Forest (RF), and k-Nearest Neighbors (KNN). These machine learning models are innovative and can assess environmental problems and hazards for any given area on a regional scale. The research consists of three steps: Firstly, for training and validation, 94 historical landslides were randomly split into a proportion of 7/3. Secondly, topographical and geological data as well as satellite imagery were gathered, analyzed, and built into a spatial database using GIS Environment. Nine layers of landslide-conditioning factors were developed, including Aspect, Elevation, Slope, NDVI, Curvature, SPI, TWI, Lithology, and Landcover. Finally, the receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC) value were used to estimate the model's efficiency. The area under the curve values for the RF, XGBoost, and KNN models are 0.895 (89.5%), 0.893 (89.3%), and 0.790 (79.0%), respectively. Based on the three machine learning techniques, the innovative outputs show that the performance of the Random Forest model has a maximum AUC value of 0.895, and it is more efficient than the other tree-based classifiers. Elevation and Slope were determined as the most important factors affecting landslides in this research area. The landslide susceptibility maps were classified into four classes: low, moderate, high, and very high susceptibility. The result maps are useful for future generalized construction operations, such as selecting and conserving new urban and infrastructural areas. Doi: 10.28991/CEJ-2022-08-02-02 Full Text: PDF
Highlights
Landslides are caused by many natural and human action-induced disruptions to the soil [1,2,3]
Based on the three machine learning techniques, the innovative outputs show that the performance of the Random Forest model has a maximum area under the ROC curve (AUC) value of 0.895, and it is more efficient than the other tree-based classifiers
The model's comprehensibility helps determine the potential association between manipulating Parameters and Landslide Susceptibility Mapping (LSM) from data, which is critical for considering the parameters that influence landslide susceptibility
Summary
Landslides are caused by many natural and human action-induced disruptions to the soil [1,2,3]. Landslides inflict significant economic loss every year, whether caused by natural or human activity [4,5,6]. The Himalayan Mountain Range is vulnerable to frequent and devastating landslides due to robust topography, poor lithology, climate change, soil, fragile sloping infrastructure, and active tectonic action. Muzaffarabad is the capital of Azad Jammu and Kashmir, Pakistan, and lies in the Himalayan Mountains. In Muzaffarabad, the earthquake brought about significant land losses, namely the landslide of Hattian Bala rock, causing the deaths of 595 people [8, 9]. Landslide susceptibility starts with the inventory of past landslides in the area and the collection of optimal causative factors. The landslide and a practical approach for analyzing the association between landslide occurrences and influencing factors are needed, leading to the landslide susceptibility map [13]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
