Yann Marcon

LAPMv2: An improved tool for underwater large-area photo-mosaicking

This paper presents a research software solution specifically developed to allow marine scientists to produce geo-referenced visual maps of the seafloor, known as mosaics, from a set of underwater images and navigation data. LAPMv2 is a suite of tools, which provides the users with the current state-of-the-art methods for automatic feature detection and matching, as well as with powerful tools for registering images, constructing and geo-referencing photomosaics, and importing them into geographic information systems such as ArcGIS. The main key strengths of LAPMv2 include (1) its robust false-match rejection method, which allows the application to run fully autonomously without producing aberrant results caused by erroneous matches, (2) the total control of the user over the mosaicking workflow, (3) the possibility to intuitively create matches between unlinked images, and (4) the graphical interface that guides the user through the different steps of the mosaicking workflow.

Mud extrusion and ring-fault gas seepage – upward branching fluid discharge at a deep-sea mud volcano

Submarine mud volcanoes release sediments and gas-rich fluids at the seafloor via deeply-rooted plumbing systems that remain poorly understood. Here the functioning of Venere mud volcano, on the Calabrian accretionary prism in ~1,600 m water depth is investigated, based on multi-parameter hydroacoustic and visual seafloor data obtained using ship-borne methods, ROVs, and AUVs. Two seepage domains are recognized: mud breccia extrusion from a summit, and hydrocarbon venting from peripheral sites, hosting chemosynthetic ecosystems and authigenic carbonates indicative of long-term seepage. Pore fluids in freshly extruded mud breccia (up to 13 °C warmer than background sediments) contained methane concentrations exceeding saturation by 2.7 times and chloride concentrations up to five times lower than ambient seawater. Gas analyses indicate an underlying thermogenic hydrocarbon source with potential admixture of microbial methane during migration along ring faults to the peripheral sites. The gas and pore water analyses point to fluids sourced deep (>3 km) below Venere mud volcano. An upward-branching plumbing system is proposed to account for co-existing mud breccia extrusion and gas seepage via multiple surface vents that influence the distribution of seafloor ecosystems. This model of mud volcanism implies that methane-rich fluids may be released during prolonged phases of moderate activity.

Underwater hyperspectral imaging as an in situ taxonomic tool for deep-sea megafauna

Identification of benthic megafauna is commonly based on analysis of physical samples or imagery acquired by cameras mounted on underwater platforms. Physical collection of samples is difficult, particularly from the deep sea, and identification of taxonomic morphotypes from imagery depends on resolution and investigator experience. Here, we show how an Underwater Hyperspectral Imager (UHI) can be used as an alternative in situ taxonomic tool for benthic megafauna. A UHI provides a much higher spectral resolution than standard RGB imagery, allowing marine organisms to be identified based on specific optical fingerprints. A set of reference spectra from identified organisms is established and supervised classification performed to identify benthic megafauna semi-autonomously. The UHI data provide an increased detection rate for small megafauna difficult to resolve in standard RGB imagery. In addition, seafloor anomalies with distinct spectral signatures are also detectable. In the region investigated, sediment anomalies (spectral reflectance minimum at ~675 nm) unclear in RGB imagery were indicative of chlorophyll a on the seafloor. Underwater hyperspectral imaging therefore has a great potential in seafloor habitat mapping and monitoring, with areas of application ranging from shallow coastal areas to the deep sea.

A participative tool for sharing, annotating and archiving submarine video data

The VIDLIB Deep-Sea Video Platform is an open access web-based tool for marine video data storage, streaming, sharing, and analysis. Using VIDLIB specialists can share, access and annotate the same videos from anywhere, thus accelerating the video annotation and analysis process. This way, scientists can share expert knowledge for video analysis (i.e. species identification) without the need to upload and download large video files. Moreover, the tool has the functionalities for participatory science, and science communication in that nonspecialists can ask questions or comment on what they see, and get answers from scientists. VIDLIB is available at http://vidlib.marum.de. In this paper we describe the structure and workflow of the VIDLIB Deep-Sea Video Platform and present an example of analysis from video data in the Southwest Indian Ridge.