Flood Mapping Analysis for Sentinel-2 Data Using Supervised and Unsupervised Classifications

B. Chandrababu Naik and Prof. B. Anuradha

Floods are the most dangerous and common disaster occurring in many areas of the world. Satellite images provide more information, which is significant for assessing the disaster impact and taking up flood mitigation activities. The Palakkad and Iddukki districts are more effected flood areas during the month of August 2018 in the state of the Kerala. The main objective of this paper is to study the flood mapping in affected areas and improve the flood classification results from Sentinel-2 images with cloud free data. The flooding areas can be recognized using supervised and unsupervised classification techniques and accuracy analysis can be achevied. The most effected flood area Malampuzha Dam and its surrounding flood extent areas were classified before the flood and after flood image analysis. The unsupervised classification (K-mean cluster) technique provides better results than supervised classification (maximum likelihood and minimum distance) techniques. Further research is required for solving the issue of poor detection of flooded sandy regions.

Volume 11 | Issue 10

Pages: 58-65

DOI: 10.5373/JARDCS/V11I10/20193006