Mapping and analyses of two coralline banks in the western continental margin of India using multibeam echosounder data

Abstract

Six and eleven profiles of bathymetric and backscatter data are drawn from the gridded (rasterized) maps of the two banks, the Gaveshani Bank and another, an unnamed bank (at latitude 13 degrees 43.5' N, longitude 73 degrees 42' E) respectively. The existence of six and five classes of data is established with respect to the Gaveshani and the unnamed bank through the use of Artificial Neural Network (ANN) based unsupervised self-organizing map (SOM) architecture, for determining the number of data classes, and corresponding 60 and 209 segments using fuzzy c-means (FCM) algorithm for segmentation. The segmented profiles of each bank are overlaid on the respective gridded backscatter maps to examine the seafloor morphology associated with the distribution of the overlying sediment material. Lower backscatter intensities (-60 to -30 dB) are observed with respect to the unnamed bank, whereas in the case of Gaveshani bank, the backscatter intensities were varying from -30 to -15 dB, indicating relatively higher intensity in comparison to the unnamed bank. This portends reduced sediment deposition on the Gaveshani bank due to erosion.