Comparison of Different Water Indices for Water Surface Extraction using Remote Sensing & GIS Techniques

Research Supervisor
Journal Title
Journal Title
Journal ISSN
Volume Title
Bangladesh is a country where the climate is categorized as subtropical monsoon climate with drastic seasonal changes in rainfall, high temperature, and humidity and Saint Martin’s Island is at the helm of it. This makes monitoring the state of the land and water bodies crucial for sustaining resources, economy and even human lives. Surface water extraction can play a huge role in fulfilling this role. Remote sensing technology and GIS techniques have enabled us to efficiently conduct surface water extraction without being physically present in the locations. Data of these locations are obtained from Landsat - 8 OLI-TOA datasets available to the public online. The datasets contain multispectral images with 11 bands which can be used to calculate the values of the water indices. Spectral water indices have allowed us to delineate water bodies from land with great accuracy and is the preferred method by most researchers. This paper presents the assessment and comparison of two of the best water index methods: Modified Normalized Difference Water Index (MNDWI) and Normalized Difference Water Index (NDWI). While both are very useful for water surface extraction, MNDWI showed a more superior performance to that of NDWI with an average overall accuracy of 87.5% and average Kappa value of 0.75 on images over eight years (2013 to 2020). So MNDWI should be the preferred method of choice for Landsat - 8 OLI-TOA images to differentiate between land and water, respectively.
Department Name
Electrical and Computer Engineering
North South University
Printed Thesis