10 Ocean Colour Remote Sensing of Harmful Algal Blooms in the Benguela System
189
types, with greater confidence and in a wider variety of bloom conditions. There are
several algorithm types that could be used to achieve this (IOCCG 2006): inherent optical property based reflectance inversion algorithms will be focused on here
as amongst the least constrained and therefore most likely to provide output pertinent to a variety of bloom types (ibid.). Such algorithms are necessarily dependent
upon an ecosystem-specific characterisation and parameterisation of the inherent
optical properties (IOPs) of the phytoplankton assemblage, and the effects of these
upon the water-leaving radiance/reflectance. Use of such algorithms requires some
consideration of the major cellular and assemblage-specific parameters causal to biooptical variability; and quantitative consideration of the ability to detect the effects
of such phenomena in the water-leaving radiance signal across a range of phytoplankton biomass—a complex issue, and one as yet unresolved (e.g. Alvain et al.
2012; Dupouy et al. 2011). Nevertheless key assemblage characteristics potentially
identifiable through determination of phytoplankton IOPs from the water-leaving
radiance spectral signal can be considered as:
1. Average assemblage size. Size, as determined through mean assemblage metrics
such as the effective diameter (Bernard et al. 2006), is arguably one of the most
important gross determinants of the optical properties of the algal assemblage
(Morel and Bricaud 1981; Roy et al. 2011; Zhou et al. 2012). In addition, sizebased ocean colour algorithms have shown effective application in both the coastal
(Ciotti and Bricaud 2006) and open ocean (Kostadinov et al. 2010). Detection of
changing assemblage size is of considerable potential value in HAB applications:
bloom onset is often characterised by changes in the gross size of the algal assemblage, and arguably blooms dominated by large dinoflagellate cells are most
commonly associated with harmful impact in the southern Benguela (Pitcher and
Calder 2000).
2. Gross changes in intra-cellular accessory pigments. Whilst the effects of pigment
changes on cellular absorption are significant and well understood (Johnsen et al.
1994; Lohrenz et al. 2003), there is considerable debate as to how detectable such
effects are on the water-leaving reflectance across different biomass and other optical regimes (Dierssen et al. 2006; Jackson et al. 2011; Alvain et al. 2012). One
important determinant in detecting absorption-related effects, in a reflectance
signal, is phytoplankton biomass. There is little doubt that, while a detailed sensitivity study still must be conducted, such effects are much more visible, and
perhaps only detectable, at considerable biomass concentration (Dierrsen et al.
2006). In the southern Benguela, it is highly unlikely that it is possible to distinguish even high biomass blooms dominated by diatoms or dinoflagellates based
only on pigment-related absorption effects, given that fucoxanthin and peridinin
(as the primary respective accessory pigments) are spectrally similar (Bidigare
et al. 1990). However, the photosynthetic ciliate Myrionecta rubra, a large-celled,
bloom-forming species displaying considerable phycoerythrin absorption (Kyewalyanga 2002), is potentially detectable based on absorption effects on the water
leaving reflectance, and such an approach is investigated during this study.
189
types, with greater confidence and in a wider variety of bloom conditions. There are
several algorithm types that could be used to achieve this (IOCCG 2006): inherent optical property based reflectance inversion algorithms will be focused on here
as amongst the least constrained and therefore most likely to provide output pertinent to a variety of bloom types (ibid.). Such algorithms are necessarily dependent
upon an ecosystem-specific characterisation and parameterisation of the inherent
optical properties (IOPs) of the phytoplankton assemblage, and the effects of these
upon the water-leaving radiance/reflectance. Use of such algorithms requires some
consideration of the major cellular and assemblage-specific parameters causal to biooptical variability; and quantitative consideration of the ability to detect the effects
of such phenomena in the water-leaving radiance signal across a range of phytoplankton biomass—a complex issue, and one as yet unresolved (e.g. Alvain et al.
2012; Dupouy et al. 2011). Nevertheless key assemblage characteristics potentially
identifiable through determination of phytoplankton IOPs from the water-leaving
radiance spectral signal can be considered as:
1. Average assemblage size. Size, as determined through mean assemblage metrics
such as the effective diameter (Bernard et al. 2006), is arguably one of the most
important gross determinants of the optical properties of the algal assemblage
(Morel and Bricaud 1981; Roy et al. 2011; Zhou et al. 2012). In addition, sizebased ocean colour algorithms have shown effective application in both the coastal
(Ciotti and Bricaud 2006) and open ocean (Kostadinov et al. 2010). Detection of
changing assemblage size is of considerable potential value in HAB applications:
bloom onset is often characterised by changes in the gross size of the algal assemblage, and arguably blooms dominated by large dinoflagellate cells are most
commonly associated with harmful impact in the southern Benguela (Pitcher and
Calder 2000).
2. Gross changes in intra-cellular accessory pigments. Whilst the effects of pigment
changes on cellular absorption are significant and well understood (Johnsen et al.
1994; Lohrenz et al. 2003), there is considerable debate as to how detectable such
effects are on the water-leaving reflectance across different biomass and other optical regimes (Dierssen et al. 2006; Jackson et al. 2011; Alvain et al. 2012). One
important determinant in detecting absorption-related effects, in a reflectance
signal, is phytoplankton biomass. There is little doubt that, while a detailed sensitivity study still must be conducted, such effects are much more visible, and
perhaps only detectable, at considerable biomass concentration (Dierrsen et al.
2006). In the southern Benguela, it is highly unlikely that it is possible to distinguish even high biomass blooms dominated by diatoms or dinoflagellates based
only on pigment-related absorption effects, given that fucoxanthin and peridinin
(as the primary respective accessory pigments) are spectrally similar (Bidigare
et al. 1990). However, the photosynthetic ciliate Myrionecta rubra, a large-celled,
bloom-forming species displaying considerable phycoerythrin absorption (Kyewalyanga 2002), is potentially detectable based on absorption effects on the water
leaving reflectance, and such an approach is investigated during this study.