Distinguishing cyanobacteria from algae using bio-optical remote sensing.
This study advances the use of remote sensing for eutrophication and cyanobacterial bloom detectionin inland and near-coastal waters. The hypothesis that prokaryotic cyanobacteria can besystematically differentiated from algae (or eukaryotic species) on the basis of their distinctive bioopticalfeatures is tested using a novel in situ bio-optical dataset and remotely sensed data from theMedium Resolution Imaging Spectrometer (MERIS). The in situ dataset was collected between2010 and 2012 from three optically-diverse South African inland waters. An empirical algorithm,called the maximum peak-height (MPH) algorithm, was developed for operational determinationsof trophic status (chlorophyll-a), cyanobacterial blooms and surface scum from MERIS. The algorithmuses top-of-atmosphere data to avoid the large uncertainties associated with atmosphericallycorrected water leaving refectance data in optically-complex and turbid waters. The detailed analysisof the variability of the optical properties of the three diverse reservoirs provides new knowledgeof the inherent optical properties of South African inland waters which have previously not beendescribed. The study also provides the frst detailed investigation of the effects of intracellular gasvacuoles on the optics of phytoplankton using a two-layered sphere model. The results demonstratehow gas vacuoles impart distinctive bio-optical features to cyanobacteria and cause backscatteringto be enhanced. An advanced inversion algorithm is developed for detecting phytoplankton assemblagetype and size from water leaving reflectance data. The algorithm, based on a direct solution of the equation of radiative transfer using Ecolight-S radiative transfer model, successfully distinguishes between phytoplankton assemblages dominated by small-celled cyanobacteria and those dominated by large-celled dinoflagellate species. It also provides reliable estimates of phytoplankton biomass (chl-a), and the absorption coefficients of phytoplankton and combined nonphytoplankton particulate and dissolved matter. Finally, the application of the MPH algorithm to a time series of MERIS data from 2002 to 2012 in South Africa’s 50 largest reservoirs is likely to be the most comprehensive assessment of eutrophication and cyanobacteria occurrence from earth observation data yet performed. The results confirm that widespread cyanobacterial blooms and eutrophication remain issues of critical concern for water quality in South Africa.