Mapping shallow, benthic communities using optical, remotely sensed data: a test of habitat classification accuracy using multiple sources of bathymetry

Patrick Gagnon, William Jones, Herbert Ripley
Hyperspectral Data International
7071 Bayers Road - Suite 119, Halifax, NS, Canada B3L 2C2

Robert Scheibling
Department of Biology, Dalhousie University, Halifax, NS, Canada, B3H 4J1

Bruce Hatcher
University College of Cape Breton, Sydney, NS, Canada, B1P 6L2

ABSTRACT

Optical remote sensing technologies are increasingly used to map shallow, marine benthic communities for further investigation of ongoing marine, large-scale, coastal community changes associated to global phenomenon such as climate change and biological invasions. This approach, however, requires careful examination of the water-induced light attenuation, the effects of which can significantly reduce our ability to discriminate bottom features (e.g., algal species, bottom types), and further derive image-based distributional maps of such features. One basic approach to reduce the effects of water column light attenuation from remotely sensed data is to deploy spectral targets along a depth gradient, and further derive, from the imagery, a light attenuation curve that can be used for calibration. Although relatively inexpensive, the accuracy of this approach is tightly linked to our knowledge of the depth at each pixel, which, in shallow (<10m) coastal zones, is often problematic given the usually large distance separating each datum in low-cost chart products and databases distributed by qualified instances (e.g., coast guards and hydrographic services). In this paper, we compare our ability to generate accurate, classified, distributional maps of benthic (indigenous and invasive) vegetation along the Atlantic coast of Nova Scotia (Canada) using reflective tarps, airborne hyperspectral data, and two sources of bathymetric data with contrasting inter-pixel distances: 1) data provided by the Canadian Hydrographic Service (large inter-pixel distance), and 2) data collected using a common, recreational fish finder equipped with a GPS and a internal data storage unit (small inter-pixel distance). The operational pros and cons of using each type of datum are further discussed.

Last Update: 2005-03-16