New GIS Approach using Machine Learning Algorithm for early floods Detection

Abstract

Floods are one of the most devastating natural hazards. Early intervention plays an important role in saving resources and many lives during the floods. Various technologies are used for rapid response, such as the use of satellite images in geographical information systems. aerial satellite images of areas for candidate flooding, we have proposed a method of classification of aerial images to identify areas likely to be affected by flooding using various classifiers such as SVM, fine tree, KNN and neural networks. Their performance is compared, and it is observed that the SVM classifier exceeds the remaining classifiers with almost 93.1% due to its simplicity, although neural networks minimize the amount of training to a greater extent. This classification of images is then used to identify the areas likely to be affected by the floods in order to identify and manage the extent of the floods. An electronic complement is also proposed to assist the intervention teams in real time for better results.