POLYMETALLIC NODULE ABUNDANCE ESTIMATION USING SIDESCAN SONAR: A QUANTITATIVE APPROACH USING ARTIFICIAL NEURAL NETWORK

Abstract

Scattered abundantly across the vast regions of the Clarion and Clipperton Fracture Zone (CCFZ) are pockets of Polymetallic Nodules (PMN). These PMN possess high economic potential as they are rich in minerals such as manganese, nickel, copper and rare earth elements. Quantification of such PMN coverage is important for economic feasibility studies and planning of exploitation strategies. Traditional methods for PMN quantification are labour and time intensive as they rely on freefall box corer measurements and/or image processing of seabed photographs. This research thesis explores PMN abundance estimation using a data-driven method based on Artificial Neural Network (ANN). Data used are geotagged Sidescan Sonar (SSS) seabed backscatter images and seabed photographs collected using an Autonomous Underwater Vehicle (AUV) within the CCFZ. Compared to an underwater camera, the SSS provides a much larger area of coverage, effectively increasing the AUV's efficiency in the task of seabed surveying within the limited dive-time. This is the first known published work to elaborate on a data-driven approach in estimating PMN abundance using SSS seabed backscatter data. The trained ANN model yielded an average accuracy performance of 85.36%, demonstrating that it can be an effective tool in estimating PMN abundance from SSS seabed backscatter images. This approach enables faster evaluation of PMN abundance for future deep seabed exploration without the need for underwater cameras.