PIXEL BASED LANDSLIDE IDENTIFICATION USING LANDSAT 8 AND GEE

Abstract. Landslide is one of the most common natural disasters triggered mainly due to heavy rainfall, cloud burst, earthquake, volcanic eruptions, unorganized constructions roads, and deforestation. In India, field surveying method used identify potential landslide regions update inventories maintained by Geological Survey but it very time-consuming, costly, inefficient. Alternatively, advanced remote sensing technologies in analysis allow rapid easy data acquisitions help improve traditional detection capabilities. Supervised Machine learning algorithms, for example, Support Vector (SVM), are challenging conventional techniques predicting with astounding accuracy. this research work, we have utilized open-source datasets (Landsat 8 multi-band images JAXA ALOS DSM) Google Earth Engine (GEE) landslides Rudraprayag using machine techniques. a district Uttarakhand state which has always been center attention geological studies its higher density landslide-prone zones. For training validation purpose, labeled locations obtained from inventory (prepared India) layers such as NDVI, NDWI, slope (generated DSM Landsat satellite imagery) were used. The identification performed SVM, Classification Regression Trees (CART), Minimum Distance, Random forest (RF), Naïve Bayes techniques, SVM RF outperformed all other achieving an 87.5% true positive rate (TPR).