Thoughts on Algorithmic Approaches to Reveal Marine Geomorphology as a Proxy for Habitat

2010 Ocean Sciences Meeting
February 22-26, 2010, Portland, OR

Special Session IT38: Marine Habitat Mapping: Where Physical Oceanographers, Marine Ecologists, GIS Experts and Habitat Suitability Modelers Meet

DJ Wright

Dept of Geosciences, Oregon State Univ

Corvallis, OR 97331-5506

dawn@dusk.geo.orst.edu

WD Heyman

Department of Geography

Texas A&M University

College Station, TX

There is a growing body of data suggesting that the underlying geology and geomorphology of marine environments dictates the location of critical life habitat for a variety marine species, thus serving also to advance the application of ecosystem-based management (EBM) to the design of marine reserve networks. However, with only 5-10% of the world’s seafloor mapped with the resolution of similar studies on land marine geomorphology still represents a persistent gap in our knowledge. Sea bottom geomorphology and habitat information can be gained through various hydrographic surveys (e.g., single beam, multibeam, and side-scan), satellite-based remote sensors, and aircraft-mounted sensors. Water column properties (e.g., salinity, temperature, current speed and direction, chlorophyll content, turbidity, and nutrients) can be measured directly with boat-based or in-situ instruments or remotely with satellite-based sensors. To accompany this physical information, ecological information (e.g., benthic habitat, species composition, richness and abundance) almost always needs to be evaluated by direct observations generally accompanied by still photography or video acquired by SCUBA, ROVs or submersibles. These data are highly variable in space and time so characterization requires multiple observations over various seasons and times and can be prohibitively expensive and time consuming to collect. Therefore, what are the major algorithmic approaches for classifying, integrating and interpreting physical and ecological data sources to reveal geomorphology as a proxy for habitat? Algorithmic approaches are almost always quantitative, usually automatic or at least semi-automatic, and allow the user to refine the classification at certain stages in the process based on visual observation. This mode of classification, while subject to artifacts, is usually more repeatable, less expensive, and with resolution limited only by the source data. Local and regional habitat schemes have been developed over the years, but given the continuing paucity of available marine habitat data, and the need for rapid and large expansion in marine reserves networks coverage, is it feasible to move toward a standardization of algorithmic seafloor classification approaches? These approaches include quantitative assessment of the shape of the seafloor (slope, aspect, position index), the hardness of the bottom, seafloor roughness (rugosity, benthic complexity), as well as a range of approaches for ecological habitat modeling involving biological data in concert with bathymetry extending from the seafloor up into the water column (generalized linear model, generalized additive modeling, Bayesian models, classification and regression trees, and more). In assessing these many algorithmic approaches, standardization via scale (regional size of a hydrographic survey) as well as by resolution of the data (size of grid cells) will be key. This presentation includes a brief review of such approaches with an eye toward continuing discussions of the future development of a standard decision-tree or bio-geomorphology classification approaches, governed by scale, depth, slope, latitude and habitat type.

INDEX TERMS: [0460] BIOGEOSCIENCES / Marine systems, [1928] INFORMATICS / GIS science, [3045] MARINE GEOLOGY AND GEOPHYSICS / Seafloor morphology, geology, and geophysics.

Download PowerPoint file of talk (7.9 Mb)


Return to Dawn's Vita