Monday, September 3, 2007

Article of the Week

Tyrell, M. 2004. Strategic Plan for Mapping Massachusetts’ Benthic Marine Habitats. Massachusetts Office of Coastal Zone Management, Boston, MA.
A thorough overview of mapping methodologies and challenges. This basically explains how Massachusetts can go about mapping its marine waters using remote sensing technologies (aerial/lidar/casi in shallow waters, various acoustic techniques in deeper waters >5 or 10 meters, and associated groundtruthing--SPI, photos, videos, infauna, epibenthic sleds, trawls). Several specific recommendations include:

  • The state should assemble a technical advisory group composed of scientists and managers including representatives from CZM, Massachusetts Division of Marine Fisheries (DMF / Marine Fisheries), NHESP, DEP and the Department of Conservation and Recreation.

  • The goal should be 4 maps: bathymetry, surficial geology, sediment texture, and benthic habitats.

  • Data collection should meet the National Map Accuracy Standards for positional accuracy

  • Data collection should meet the National Park Service 90% thematic accuracy standard

  • A minimum of one groundtruth sample per square kilometer should be the minimum sampling density for statewide benthic habitat mapping.

  • Specific areas that appear highly heterogeneous in the backscatter should have a higher proportion of groundtruth samples.

  • Backscatter values should be the primary basis for stratification of groundtruth samples for this benthic habitat mapping initiative. Bathymetry should be an additional consideration when designating sampling strata because of the strong influence of water depth over the composition of benthic communities.

  • In order to insure that the delineation of various habitat types is as objective and repeatable as possible, multivariate statistical methods should be used.

  • Unsupervised classification of the groundtruthing stations into habitat types should be based upon natural groupings of sampling stations identified through cluster analysis and non-metric multidimensional scaling.

No comments: