Use of hyperspectral remote sensing reflectance for detection and assessment of the harmful alga, Karenia brevis.

We applied two numerical methods to in situ hyperspectral measurements of remote sensing reflectance Rrs to assess the feasibility of remote detection and monitoring of the toxic dinoflagellate, Karenia brevis, which has been shown to exhibit unique absorption properties. First, an existing quasi-analytical algorithm was used to invert remote sensing reflectance spectra, Rrs(lambda), to derive phytoplankton absorption spectra, a(phi)Rrs(lambda). Second, the fourth derivatives of the a(phi)Rrs(lambda) spectra were compared to the fourth derivative of a reference K. brevis absorption spectrum by means of a similarity index (SI) analysis. Comparison of reflectance-derived a(phi) with filter pad measured a(phi) found them to agree well (R2=0.891; average percentage difference, 22.8%). A strong correlation (R2=0.743) between surface cell concentration and the SI was observed, showing the potential utility of SI magnitude as an indicator of bloom strength. A sensitivity analysis conducted to investigate the effects of varying levels of cell concentrations and colored dissolved organic matter (CDOM) on the efficacy of the quasi-analytical algorithm and SI found that a(phi)Rrs(lambda) could not be derived for very low cell concentrations and that, although it is possible to derive a(phi)Rrs(lambda) in the presence of high CDOM concentrations, CDOM levels influence the a(phi)Rrs(lambda) amplitude and shape. Results suggest that detection and mapping of K. brevis blooms based on hyperspectral measurements of Rrs are feasible.