192
S. Bernard et al.
a considerable challenge to utilising such data from space-borne ocean colour sensors. The retinue of methodological challenges to ocean colour application in the
coastal environment are compounded by HAB application: atmospheric correction in turbid waters and an aerosol-laden atmosphere; distinction of bio-optical
constituents in optically complex waters; and adjacency effects from land. The phytoplankton dominated waters of the Southern Benguela can in the large part be
conventionally classified as Case 1 (Morel and Prieur 1977), excluding the typically
short-lived case of freshly upwelled waters where there are enhanced concentrations
of small inorganic particulate through the water column (Kudela et al. 2006). However, the considerable phytoplankton biomass observed under bloom conditions in
the Benguela provide the same functional obstacles to atmospheric correction algorithms as sediment-influenced Case 2 waters: enhanced turbidity and resulting
non-zero signal in the near-infrared, thereby compromising atmospheric correction
algorithms assuming null water-leaving radiance signal in NIR bands (Gordon and
Wang 1994; Wang 2006). Past use in the Benguela has shown that the Bright Pixel
Atmospheric Correction (BPAC) algorithm (Moore and Lavender 2011), with an
ability to adjust for non-zero NIR reflectance, performs relatively well on data from
the MERIS in high biomass Benguela waters. In addition, the typical high biomass
reflectance signals of interest here contain a strong signal, if not a distinct peak,
in the 700–730 nm spectral range (Matthews et al. 2012; Pitcher et al. 2008a): the
only recent ocean colour mission carrying an appropriately located spectral band, at
709 nm, is the MERIS (and forthcoming OLCI on the Sentinel 3 series). Reduced
resolution ( ± 1 km ground resolution) data from the MERIS will therefore be exclusively focused on here; in addition to optimal band placement and routine availability
of suitably derived water-leaving reflectance data, the sensor has excellent sensitivity
with a large signal:noise ratio, and suitably high revisit time with global coverage
every 2 days. This study seeks to demonstrate the utility of products derived from
these ocean colour data to identify bloom types based on cell size and intracellular
pigment characteristics, and show ocean colour application to bloom observation at
the event scale.
10.3 Methods: Bio-Optical Data and Algorithms
St Helena Bay in the Southern Benguela is focussed on due to its high biomass levels and the frequent occurrence of HABs within this region. A retentive circulation,
linked with the upwelling dynamics of the bay, plays an important role in the transport, concentration and dissipation of blooms (Pitcher and Calder 2000). As such, St
Helena Bay has been the site of numerous in situ data collection campaigns during
the last decade. In addition to the satellite data and algorithm output presented here,
in situ radiometry from a Satlantic Hyperspectral Tethered Surface Radiometer Buoy
(H-TSRB) and in situ, fluorometric chlorophyll values are provided for comparison.
Absorption measurements were made from filtered samples using a Shimadzu UV2501 spectrophotometer. Identification of species and estimates of effective diameter
from particle size distributions were made through microscopy and use of a Beckman
Coulter Counter respectively. Sampling was generally conducted from Lamberts Bay
S. Bernard et al.
a considerable challenge to utilising such data from space-borne ocean colour sensors. The retinue of methodological challenges to ocean colour application in the
coastal environment are compounded by HAB application: atmospheric correction in turbid waters and an aerosol-laden atmosphere; distinction of bio-optical
constituents in optically complex waters; and adjacency effects from land. The phytoplankton dominated waters of the Southern Benguela can in the large part be
conventionally classified as Case 1 (Morel and Prieur 1977), excluding the typically
short-lived case of freshly upwelled waters where there are enhanced concentrations
of small inorganic particulate through the water column (Kudela et al. 2006). However, the considerable phytoplankton biomass observed under bloom conditions in
the Benguela provide the same functional obstacles to atmospheric correction algorithms as sediment-influenced Case 2 waters: enhanced turbidity and resulting
non-zero signal in the near-infrared, thereby compromising atmospheric correction
algorithms assuming null water-leaving radiance signal in NIR bands (Gordon and
Wang 1994; Wang 2006). Past use in the Benguela has shown that the Bright Pixel
Atmospheric Correction (BPAC) algorithm (Moore and Lavender 2011), with an
ability to adjust for non-zero NIR reflectance, performs relatively well on data from
the MERIS in high biomass Benguela waters. In addition, the typical high biomass
reflectance signals of interest here contain a strong signal, if not a distinct peak,
in the 700–730 nm spectral range (Matthews et al. 2012; Pitcher et al. 2008a): the
only recent ocean colour mission carrying an appropriately located spectral band, at
709 nm, is the MERIS (and forthcoming OLCI on the Sentinel 3 series). Reduced
resolution ( ± 1 km ground resolution) data from the MERIS will therefore be exclusively focused on here; in addition to optimal band placement and routine availability
of suitably derived water-leaving reflectance data, the sensor has excellent sensitivity
with a large signal:noise ratio, and suitably high revisit time with global coverage
every 2 days. This study seeks to demonstrate the utility of products derived from
these ocean colour data to identify bloom types based on cell size and intracellular
pigment characteristics, and show ocean colour application to bloom observation at
the event scale.
10.3 Methods: Bio-Optical Data and Algorithms
St Helena Bay in the Southern Benguela is focussed on due to its high biomass levels and the frequent occurrence of HABs within this region. A retentive circulation,
linked with the upwelling dynamics of the bay, plays an important role in the transport, concentration and dissipation of blooms (Pitcher and Calder 2000). As such, St
Helena Bay has been the site of numerous in situ data collection campaigns during
the last decade. In addition to the satellite data and algorithm output presented here,
in situ radiometry from a Satlantic Hyperspectral Tethered Surface Radiometer Buoy
(H-TSRB) and in situ, fluorometric chlorophyll values are provided for comparison.
Absorption measurements were made from filtered samples using a Shimadzu UV2501 spectrophotometer. Identification of species and estimates of effective diameter
from particle size distributions were made through microscopy and use of a Beckman
Coulter Counter respectively. Sampling was generally conducted from Lamberts Bay