Phaeocystis globosa (P. globosa hereafter) is a phytoplankton species which commonly blooms at high biomass in April–May in the Southern North Sea and forms undesirable foam which accumulates on the beaches. Monitoring of this species is required by EU directives. Measurement of phytoplankton species composition has historically been made by pigment or microscopic analysis of water samples, which is spatially sparse and temporally infrequent e.g. weekly/monthly. In-water instruments such as flow cytometers can provide very high frequency data but at high acquisition and maintenance cost. Automated in situ above water radiometry has the potential to provide very high frequency data at single locations but requires very careful design of processing algorithms in turbid waters with high non-algal absorption. Spaceborne radiometry could provide both very good spatial coverage and moderate/high frequency of data, e.g. daily/weekly, but accurate determination of phytoplankton species composition is considerably more difficult in turbid waters than in open ocean waters. Prior studies based on a limited number of shipborne reflectance measurements suggested feasibility of P. globosa detection in turbid waters from hyperspectral radiometry. The availability of a new autonomous above water hyperspectral radiometer system has enabled further refinement and intensive testing of these techniques. From a time-series of 4356 water reflectance spectra measured near Ostend harbour in Belgian coastal waters from 2020/04/01 to 2020/08/18, two existing algorithms for P. globosa detection were successfully applied. Results show a high biomass P. globosa bloom occurring in late-April/early-May as found every year in water sample analyses for Belgian coastal waters. The high temporal resolution of the radiometric data allows to capture the evolution of the bloom at time scales sufficiently short (hourly and daily) compared to growth/decay and tidal processes. The challenges of extending the methods to future spaceborne instruments are also tested by simulating the impact of errors in sensor inter-band calibration, atmospheric correction and radiometric noise. Results show that because of their spectral coherence, atmospheric correction errors impact only slightly P. globosa detection whereas inaccuracy in inter-band calibration and radiometric noise are much more problematic as they affect each spectral band independently. Because radiometric noise should be reduced in the new generation of hyperspectral sensors and can always be reduced by spatial binning, the inter-band relative calibration uncertainty appears to be the main challenge for spaceborne mission design. Indeed, it was demonstrated that inter-band calibration error should be 0.25\% and ideally 0.1\% at top of the atmosphere highlighting the need for particular attention to inter-band calibration in sensor design and post processing treatments including vicarious calibration.
Located in
Library
/
RBINS Staff Publications 2022
