Landslide Vulnerability and Risk Mapping Using Machine Learning Algorithm with MALAVU Software (Machine Landslide Vulnerability): Study Case Jember Regency, Indonesia

Abstract

Landslide is a geological event that occurs due to the movement of rock or soil masses. Typical movements include falling rocks and large lumps of land subsidence. The impact arising from the landslides affects not only to the human fatality, but also the economic and social aspects. One area in Indonesia where landslides often occur is located at Jember Regency, East Java Province. The purpose of this research is to attempt mapping the vulnerability of landslides in the Jember area using machine learning algorithms, the Bayesian Ridge Regression. The type of research data used is collected from government websites and various literatures. Then, the data is processed with ArcGIS and MALAVU software and the results of the data output are landslide vulnerability maps. In preparing landslide hazard zones, several parameters are used, such as rock lithology, land use, soil type, rainfall, and slope. This study shows that that the Jember region has 3 landslide risk zones, namely low, medium, high. Final analysis found that land cover and topography play an important role in causing landslides. This study also suggests that machine learning method can be utilized to map land slide vulnerability by analyzing statistical data and pattern. Further analysis can be used to mitigate land slides incidents in the future.