From MERIS to OLCI and Sentinel 2: Harmful Algal Bloom Applications & Modelling in South Africa
Bernard, Stewart1; Evers-King, Hayley2; Matthews, Mark2; Robertson Lain, Lisl2; Smith, Marie2
1CSIR, SOUTH AFRICA; 2University of Cape Town, SOUTH AFRICA

The high biomass case 1 waters of the Benguela system and the eutrophic inland waters of Southern Africa present some unique challenges for the operational use of satellite radiometry. Routine detection and analysis of regularly occurring and potentially harmful algal blooms is necessary when considering the severe impacts on water quality, ecosystem health as well as commercial and recreational concerns. These optically complex blooms are typically monospecific and of very high biomass, and provide great opportunity for addressing the problems of atmospheric corrections over turbid waters and coastal/land-adjacent environments, product algorithm development, and the advancement of regional bio-optical modelling capabilities. Spectral reflectance signals through a range of chlorophyll a concentrations, algal assemblages and other bio-optical parameters are characterised via both measurement and modelling in eutrophic inland lakes and the high biomass Benguela system, with specific regard to the spectral characteristics of the OLCI and Sentinel 2 sensors. The inherent and apparent optical properties of typical algal assemblages are described for eutrophic conditions in the Benguela and selected inland waters. Sensitivity to phytoplankton functional types is investigated, with focus on optically significant biological characteristics e.g. particle size distribution, intracellular structure (including vacuoles), and chlorophyll density per cell. Progress towards a synthetic marine dataset is presented with a view to a model of spectral classification. This dataset will allow for improved understanding of the sensitivities in inversion methods, particularly with regards to assumptions made in radiative transfer methods and mathematical inversion techniques. To this end an effort is made towards the parameterisation of theoretical optical quantities (e.g. f/Q and the reflectance approximation, and spectrally variant vs constant backscattering ratio phase functions) and their relationships with water-leaving reflectances through the use of Hydrolight modelling. This informs on the suitability of algorithm approaches for these water types' particular requirements. Some detail is given on the need for the development of high-biomass-specific measurement protocols which address the difficulties encountered when performing optical measurements in these conditions. A preliminary evaluation is made of the suitability of available ocean colour data and algorithms, particularly in the context of OLCI following on from MERIS, and the new high spatial resolution opportunities for water quality monitoring presented by Sentinel 2.