usSEABED

Information about seafloor characteristics from the beach to the deep sea improves the understanding of interactions between land and sea, effects of river discharge and sea level changes, distributions of benthic flora and fauna, location and type of resources, potential consequences of human activities on the oceans, and other critical issues. Large- and small-scale maps of the seabed, as well as reliable data over broad geographical areas, allow for integrated insights into these issues and more.

To assist in addressing these issues, the USGS and the University of Colorado have created usSEABED.  The usSEABED datasets currently hold georeferenced point data for more than 300,000 data sites in U.S. waters from the beach to the deep sea, rivers, lakes, and estuaries. In usSEABED, existing data from the USGS and other research groups are processed and extended to maximize their density and usability creating unified, comprehensive, relationally linked datasets for mapping and analysis.  Source data include surficial and subbottom data from physical sampling equipment (grabs and cores) and virtual sampling such as descriptions from seafloor photographs and videos.

In addition to quantified lab-derived data, the datasets of usSEABED also include estimated numeric values for those typical seabed characteristics—noted above—based on the extensive accumulation of word-based data in U.S. waters. These data are rich in information, but were previously difficult to quantify, map, plot, or use in comparative analyses or models.

These descriptive data—from short sentences, small essays, or single phrases—are treated as a mathematical equation that is considered as a whole. Filters based on fuzzy set theory assign relative weight to each word in the description, and estimate the values of textural and other parameters. In addition, the textural implications of non-textural terms—such as 'broken shells' or Halimeda—are included in the calculation of grain-size parameters.

The resulting numeric data, now useable in a GIS or model, should be considered "fuzzy"; that is, they give an approximation—not a rigorous measurement—of the assessed values.