FLOOD DETECTION SYSTEM USING SENTINEL - 1 IMAGES AND EXTREME LEARNING MACHINE CLASSIFIER
Keywords:Extreme Learning Machine, Flood detection, Sentinel-1, Google Earth Engine
Floods are one of the most common natural disasters, affecting millions of people worldwide. Floods occur when streams burst their banks, generally as a result of significant rainfall, and inundate areas that are not normally flooded. In August 2018, Kerala witnessed the devastating effect of floods which resulted in 13.86% of the land being inundated. Satellite imagery is one of the most effective techniques for assessing the extent of flood-affected areas with high spatial resolution. The approach of monitoring floods by using Sentinel-1 satellite data from Google Earth Engine (GEE) is presented. Using the satellite images, we have created a dataset after applying preprocessing techniques like resizing and thresholding. We have used Otsu thresholding in this work due to its ability to easily distinguish water and non-water pixels. An Extreme Learning Machine (ELM) model is proposed to identify the flood-prone regions in the chosen study area. We have compared our model with existing classifiers such as Decision Tree and Support Vector Machine and found our model performs better with a good consistency and accuracy score of 0.8787. These systems can be used for better preparedness and aid in monitoring the change or reconstructing the progress of a past flood.
How to Cite
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.