Deep-Sea Robotic Survey and Data Processing Methods for Regional-Scale Estimation of Manganese Crust Distribution

Abstract

Manganese crusts (Mn-crusts) are a type of mineral deposit that exists on the surface of seamounts and guyots at depths of >800 m. We have developed a method to efficiently map their distribution using data collected by autonomous underwater vehicles and remotely operated vehicles. Volumetric measurements of Mn-crusts are made using a high-frequency subsurface sonar and a 3-D visual mapping instrument mounted on these vehicles. We developed an algorithm to estimate Mn-crust distribution by combining continuous subsurface thickness measurements with the exposed surface area identified in 3-D maps. This is applied to data collected from three expeditions at Takuyo Daigo seamount at depths of ∼1400 m. The transects add to ∼11 km in length with 12 510 m 2 mapped. The results show that 52% of the surveyed area is covered by Mn-crusts with a mean thickness of 69.6 mm. The mean Mn-crust occurrence is 69.6 kg/m 2 with a maximum of 204 kg/m 2 in the mapped region. The results are consistent with estimates made from samples retrieved from the area, showing more detailed distribution patterns and having significantly lower uncertainty bounds for regional-scale Mn-crust inventory estimation.

Guiding Labelling Effort for Efficient Learning With Georeferenced Images

“…A large proportion of automated classifiers have used a combination of hand-picked features chosen based on expert knowledge of the application domain or through a reward-based selection process [12], [13]. In [12] the authors apply a Support Vector Machine (SVM) to texture-and colour-based features designed to classify seafloor images into different substrates types for reef ecology surveys.…”

“…In [14] hand-picked geometric features are combined with SVM for classification of satellite images. In [13] a similar approach is applied for seafloor mineral prospecting. Spatial invariant features such as Local Binary Patterns (LBP) [15] and Spatial Pyramid Matching (SPM) [16] have also been effectively applied to classification problems for land [17], [18] and seafloor imagery [19], [20].…”

Use of UIoT for Offshore Surveys Through Autonomous Vehicles

“…Underwater observation systems like ENDURUNS meet the growing demand for technological improvement in ocean exploration and monitoring aimed at increasing our understanding of the ocean interior for a valuable number of scientific, industrial and political reasons [6,7]. Deep water exploration and new data collection in this domain are of special interest to provide new insights into a variety of geological and ecological processes [8,9], such as underwater hydrocarbons exploration, offshore carbon capture and storage, exploitation of deep-sea mineral resources, as well as for seafloor observatories [10][11][12][13]. At the same time, political drivers push toward the implementation of new coordinated marine monitoring programmes for establishing, preserving and restoring relevant Marine Protected Areas, as stated in the EU Marine Strategy Framework Directive (2008/56/EC).…”

Leveraging Metadata in Representation Learning With Georeferenced Seafloor Imagery

“…Manually engineered feature descriptors have been investigated by several groups for efficient image representation [5], [6], [8], [9], [10], [11]. In [9], [11], colour-based descriptors were designed based on prior knowledge of targets that are of specific scientific interest.…”

“…Manually engineered feature descriptors have been investigated by several groups for efficient image representation [5], [6], [8], [9], [10], [11]. In [9], [11], colour-based descriptors were designed based on prior knowledge of targets that are of specific scientific interest. Generic descriptors such as Local Binary Patterns (LBP) [12] and Sparse Coding Spatial Pyramid Matching (ScSPM) [13] have also been applied to identify spatially invariant patterns that appear at different scales within images of the seafloor [5], [6].…”