Seafloor Habitat Classification
The Massachusetts Office of Coastal Zone Management (CZM) is studying existing and new approaches to classify and map estuarine and marine habitats in Massachusetts. Habitat classification is the process of identifying habitat types based on a set of standard terms and descriptors. CZM is focusing current classification efforts on the physical environment of the seafloor by examining methodologies to process and interpret acoustic data collected through the Seafloor Mapping Cooperative.
The seafloor habitat classification project began with a CZM review of existing habitat classification frameworks to guide efforts to determine a suitable framework for Massachusetts. This study was followed by a detailed review of four habitat classification frameworks with the objectives of (1) applying four pre-selected habitat classification frameworks to the coastal and ocean environment in northern Massachusetts, (2) displaying and describing results of each framework, (3) comparing results and the relative strengths and weaknesses of each habitat classification framework, and (4) recommending steps required to advance habitat classification efforts in Massachusetts.
Recent habitat classification research in Massachusetts:
- Habitat Classification Feasibility Study for Coastal and Marine Environments in Massachusetts (PDF, 1 MB) - Lund, K. and A.R. Wilbur (2007).
- Mapping Seafloor Surficial Geological Habitat in Massachusetts State Waters - A PowerPoint presentation given at the 2008 Association of American Geographers annual meeting in Boston.
Abstract: The Massachusetts Office of Coastal Zone Management is leading an effort to map the Commonwealth's seafloor environment. In conjunction with the United States Geological Survey, high resolution acoustic imagery, including interferometric sidescan sonar and multibeam bathymetry, were collected to precisely map relative seafloor hardness and water depth for approximately 1,300 square kilometers in Massachusetts waters. These acoustic data are the foundation to better understand the terrain, geologic structure, and ecology of the seafloor. We are currently investigating several GIS/remote sensing-driven semi-automated classification techniques to create a seafloor map showing surficial geological habitat (patches of uniform bottom type). Maps will be based on a combination of physiographic data (bathymetry and derived data) and lithologic data (sidescan sonar interpretations and derivatives). We will process these data using map algebra, multivariate statistics, and supervised classification routines, all common to landscape ecology and terrestrial image classification, to produce a geologic habitat maps. This map will be compared to groundtruth data in an error matrix to quantify the accuracy of select elements. The final surficial geologic habitat map will serve as a basemap that will guide future research, inform decision makers facing increased coastal zone development pressures, and serve as the basis for species-specific habitat suitability modeling.
One of the first and most fundamental steps in creating a seafloor habitat map is to create polygons on the seafloor that encapsulate areas of unique physical characteristics. These polygons are assumed to represent areas of unique habitat. In the above figure, the seafloor off Marblehead, MA has been divided into nearly 5,000 polygons based on differences in lithology (rock type) and rugosity (seafloor "bumpiness"). In subsequent steps, the polygons are assigned habitat characteristics per a seafloor habitat classification scheme.
Preliminary results of potential seafloor habitat off Marblehead, MA. The identified area (the large beige polygon marked by the circled "i") can be described in general terms as a broadly flat area with little relief composed of sand and mud. The alphanumeric code provides the same description with additional detail for use by scientists. Ground truth data testing indicates we can report this classification result as true with 70% confidence. Additional work is required to refine this classification technique and hopefully increase the overall accuracy.