Kelp, specifically Nereocystis luekeana (bull kelp), form extensive forests in rocky habitats along the subtidal zone of the coast of British Columbia. Kelp forests provide important habitat for juvenile salmon attracting their preferred food and providing protection from predators. Of particular interest to the Salish Sea Marine Survival Project (SSMSP), kelp habitats have shown to provide optimal feeding and refuge conditions for Chinook and Coho.
01. To define the surface extent of bull kelp beds by using satellite imagery (present and historical) associated with sea-kayak surveys for several regions of the Gulf islands (Salt Spring, Saturna, Pender, and Mayne), and Comox and Cowichan estuaries, in collaboration with several environmental stewardship community organizations and First Nation groups.
02. To initiate a robust study on the use of kelp habitat by juvenile salmon in the Salish Sea.
For this project the following locations will be studied: Salt Spring, Saturna, Pender, and Mayne and possibly Comox estuary or Sansum Narrows-Cowichan.
Generally, kelp mapping is conducted manually via transects or delineation (dive and/or kayak and aerial photography (Mayne Island Conservancy Society, 2010; Cavanaugh et al., 2010; Sutherland et al., 2007; Field, 1996). Though effective, the strategy is labour-intensive and time consuming, and is limited to accessible areas. Specifically in BC, the Ministry of Environment has conducted kelp surveys in specific areas along the coast in the 1970s using transects and infrared aerial photos (Sutherland et al., 2007; Lucas et al., 2007). Localized initiatives kelp inventories have begun recently in the southern Gulf Islands area (Mayne Island Conservancy Society, 2010). An associated alternative to inventories is the use of optical remote imagery, whichcaptures data from large and even inaccessible areas, is cost and time effective, and attains high frequency, with the potential for automation (Simis and Duboi, 2001; Cavanaugh et al., 2011; Casal et al., 2011).
As part of a pilot study for SSMSP, we have recently defined a robust method to successfully map kelp beds using satellite imagery from Mayne Island, BC. The most accurate kelp map product attained was produced from the SPOT-6 image, with the selected variable set of principle components after image preprocessing steps of geo-rectification, atmospheric correction, masking of land and depth below 30 m, and the statistical image processing steps of principle components analysis and variable reduction (Costa et al., submitted). The method demonstrated that a remote optical sensor’s ability to identify kelp beds, such as Nereocystis and Macrocystis, relies mostly on the high reflectance in the infrared spectra compared to very low reflectance from the water. Water strongly absorbs light in the infrared spectra due to the vibrational process on the water molecules; on the other hand, floating kelp strongly reflects infrared radiation due to the interaction of infrared light within the cell structure of the kelp. Figure 1 illustrates the different spectral properties of optically deep water, kelp, and other intertidal classes in the BC coast. Note that this curve is for the kelp species Nereocystis leutkeana. Kelp exhibits a characteristic reflectance peak in the red and near-infrared wavelength ranges.