SPATIO-TEMPORAL ANALYSIS OF GLOBAL LANDSLIDE DISASTER RISK POTENTIAL BASED ON REMOTE SENSING DATA WITH INTEGRATED MACHINE LEARNING METHODS

• Main Author: Kusuma Cahya Reynaldi, Ifandra
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/73461
• Institution: Institut Teknologi Bandung

<p align="justify">Every year 67,000 people are killed, 26 million people are plunged into poverty, and nearly 200 million people are affected by natural disasters around the world. Landslides are the most common and frequent hazard in mountainous areas of the world that affect the lives of residents. Based on the first priority of the sendai framework, there is an urgency to assess and model disasters so as to reduce losses due to future disasters. This includes landslides. Detecting and mapping landslide vulnerability on a global scale is possible with remote sensing data, GIS, and machine learning algorithms that have developed to extract information related to disaster risk management. It can provide the best results in acquiring, storing, combining, manipulating, analyzing, and displaying information in determining potential landslide vulnerability. This research aims to model global landslide vulnerability based on machine learning algorithms in 2015 – 2020. Global landslide risk based on exposure to population and road infrastructure is also developed. A comparison was also made between global landslide vulnerability and the rate of population change and night light. Based on the results, it was found that the highest landslide vulnerability was in the Eastern Asia region, covering an area of 4,340,857 km2 (32.7%). As for the risk to population, it was also found that Eastern Asia was the region with the highest risk, covering an area of 3,258,264 km2 (24.6%). On road infrastructure, it is obtained that the primary road is the road with the highest risk in the world. From the correlation of landslide vulnerability with changes in populated areas based on population data or night light data, the region with a high level of vulnerability and a positive rate of change in populated areas is Eastern Asia.