Evaluating Video and Bioacoustics Sensor Performance for a Novel Hadal Water Column Profiler

Abstract

Due to its general inaccessibility, there are considerable gaps in knowledge about the bottom 6000-11,000 meters of the ocean, known as the hadal zone. Even within these extreme depths nearly all investigation has occurred on the seafloor so the water column above it remains almost unsampled. To address this inaccessibility, a new Hadal Water Column Profiler was designed to explore this zone. The goal of this project was to determine how well it performed, specifically, its video camera and bioacoustics sensors. A code-based annotation method was used to extract frames from the video in order to more efficiently identify organisms. The software ESP3 was used to create echograms of water column biomass. Both forms of data were compared with deployment factors such as rate of rotation and velocity on both ascent and descent to determine how these factors affected the quality of the data. As expected, the profilers travel speed had an influence of annotation quality, with an optimal rotation period of 5-10 seconds being determined for video observations and a slower velocity of 0.7-0.9 m/s also having an effect. While velocity does not seem to affect the bioacoustics data, a rotation period of 10 seconds is optimal. The second version of the automated frame detection algorithm substantially increased the number of detections, especially for harder to detect organisms such as fish and medusas. Further confirmation of sensor performance comes from the detection in both sensors of the location of the deep scattering layer in Hawai’i. Additionally, the profiler provided new acoustic profiles below 1000 meters.