Publication

Abstract : In the field of marine biology, researches reveal that there exists a constant increase in Algal bloom (AB) count, along the coast of India. This work aims at detecting and classifying six most frequently appearing algal blooms in this region (viz.:Trichodesmium erythraeum, Noctiluca scintillans/miliaris, Cocholodinium ploykrikoides, Chattonella marina and Karenia mikimotoi blooms). The uniqueness of ocean's optical properties such as remote sensing reflectance (Rrs) and normalized water leaving radiance (nLw) during bloom period serve as the underlying features on whose grounds classification is performed. These parameters are acquired from Aqua/MODIS sensor measurements and Regularized least squares classifier is used in GURLS library for classification. An overall classification accuracy of 94.37% is obtained using both Rrs and nLw features, which is superior to the previously conducted studies for monitoring ABs using optical properties of water. Given a MODIS image, a map is developed wherein the pixels with ABs are highlighted and the causative species is recognized. A MODIS image is available every two days and hence frequent generation of AB moitoring maps is possible, which is of great significance in the fisheries industry. © 2016 The Authors. Published by Elsevier B.V.