Machine Learning based Landslide Prediction System for Hilly Areas
| Autor: | R Archana Reddy, g Shyamala, Chilupuri Saloni Khanna, R Gobinath |
|---|---|
| Rok vydání: | 2020 |
| Předmět: | |
| Zdroj: | IOP Conference Series: Materials Science and Engineering. 981:032084 |
| ISSN: | 1757-899X 1757-8981 |
| Popis: | The recent decade had seen many natural hazards across the globe claiming numerous human lives and caused severe damage to infrastructure creating havoc. Landslide is one such natural disaster which not only creates irreversible damage but also proved to be frequently occurring in hilly areas. Several parts of the world suffer from landslides, and numerous research works were conducted across the globe to predict and manage landslides. In these works, researchers had used a specific Machine Learning-based prediction system to provide early warning before potential landslide occurs. Bountiful research works conducted on landslide generation proved that elevated water content in the soil, which may increase due to continuous and prolonged rainfall occurring in the slopes, leads to most of the landslides. It is evident that measuring the amount of rainfall is the key to predict landslide generation. This is an area where we looked upon and had developed a mechanism to predict landslides using machine learning models. This study uses seven machine learning algorithms that are trained and tested with integrated rainfall and landslide data for 36 meteorological subdivisions of India from the years 2009 to 2019. The results obtained were consistent and reliable, the algorithm that outperformed other algorithms is Logistic Regression with an accuracy of about 94.6 %. This predictive model shows better performance than the conventional rainfall threshold method. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|