The Science of Seafloor Explorer
HabCam (Habitat Mapping Camera System)
HabCam is a cabled optical and acoustic imaging system that is "flown" from a ship traveling at 5 kts at an altitude of 1 to 3 meters off the bottom at depths to 250m while collecting high resolution still images at a rate of six images per second.
Imaging at this rate provides ~50% overlap to allow for construction of image mosaics of the seafloor. A track approximately 100 nautical miles in length and 259,200 m2 in area is imaged each 24 hour day while at sea. Over 30 million images (>30 TB) have been collected in less than one year of sea time, which demonstrates how quickly this quantity of data can accumulate.
Fauna, flora, benthic maps and more
For the past five years, study areas along the northeast continental shelf have been revisited seasonally, with measurements of all visible macrofauna, and characterization of benthic fauna and flora, oceanic properties (salinity, temperature, nutrients) and substrate type, providing the baseline of an exceptional and unique ecological time series.
Key to extracting useful ecological information from this ever expanding library of image data is development of tools for rapid and accurate segmentation and classification of benthic organisms and substrate, together with visualization of images and their metadata through a Geospatially-explicit, queryable database. As scientists and ecosystem managers, we need to be able to ask questions like "what is the current distribution and abundance of sea scallop and yellowtail flounder on Georges Bank, and how have they changed over the past few years?" And, "where is the invasive tunicate Didemnum vexillum currently co-located with gravel substratum and what is the potential for its' spread to new areas".
Enter Seafloor Explorer
The data now exist to answer such ecologically critical questions, but it is buried in hundreds of TBs of images that need to be processed through a defined and scalable pipeline of tools.
Seafloor Explorer is allowing us to capture data on the distributions of sea scallop and other commercially important species as well as defining the substrate and habitat in which they live. This information has never before been acquired on such an expansive (1000s of km) yet high resolution (1mm pixel resolution) scale. While manual classification currently pushes our knowledge and understanding of these distributions, the development of tools for automated segmentation and classification is lagging orders of magnitude behind the rate at which image data are currently being acquired and manually classified.
Using data from Seafloor Explorer we can now begin to build training sets of images and data that will provide the foundation for automated machine vision approaches to target classification from HabCam images. These tools must be developed if the untapped wealth of information available in optical imagery is to be fully realized in Ecosystem Approaches to Management and understanding Essential Fish Habitat.