Artificial Intelligence-Based Landslide Studies in Indonesia: A Systematic Review in Recent Years

Abstract

Abstract Landslide is still a hot topic in geological hazard discussion, including Indonesia. Various methods, including Artificial Intelligence (AI), are used to do research development on landslide topics. Therefore, this paper aims to present a comprehensive review of AI-based landslide studies that focus on specific application area, feature engineering method (FEM), and Digital Elevation Model (DEM) sources used in the studies. This research used a qualitative method with a systematic review approach toward recent landslide studies (2012-2022) that investigated systematically in a synthesis. The exploration resulted in 26 papers from national and international indexed journals or proceedings, which filtered into 13 articles that discuss or mention the specific application area, FEM, and DEM sources. The analysis shows that AI applications in landslide studies are dominated for landslide susceptibility mapping and still a few for other applications. It also shows that almost all AI-based landslide studies chose SRTM as the source of DEM. Regarding FEM, only five articles discussed important landslide factor selection. There are four FEMs that were used in those studies, i.e., variable deduction, certainty factor model, C.45 algorithm, and variable importance ranking. From the deep analysis of those 13 articles, it can be concluded that AI-based landslide studies in Indonesia still need to be developed instead of focusing on landslide susceptibility mapping only. Studies to find effective landslide factors and compatible DEM resources using AI also can be new opportunities for landslide experts.