Remote sensing estimation of phytoplankton absorption associated with size classes in coastal waters

As a second-order expression of the phytoplankton absorption coefficient (a_ph(λ)), the specific absorption coefficient of phytoplankton (a_ph*(λ)) is an essential optical parameter that refines the a_ph(λ) normalized by phytoplankton pigment concentration. Based on the in-situ samples collected in the Bohai Sea, Yellow Sea, and the East China Sea, this study presents a combined approach to determine the specific absorption coefficients of micro- (a_m*(λ)), nano- (a_n*(λ)), and picophytoplankton (a_p*(λ)). Together with a_m*(λ) estimated by the two-component assumption model, our method effectively extracted a_n*(λ) and a_p*(λ) through the least square method. Independent in-situ validation datasets tested the performances of the proposed a_m*(λ), a_n*(λ), and a_p*(λ) by modeling a_ph(λ), and generated encouraging and acceptable predictive errors. The derived mean absolute percentage errors ranged from approximately 35%–55% for several typical wavebands (namely, 412, 443, 490, 555, 660, and 680 nm). Validation by using satellite-ground synchronization samples also produced comparative predictive errors. The spatial distribution of a_ph(λ) and the absorption of micro-, nano-, and picophytoplankton were mapped through applying the proposed specific absorption coefficients to Geostationary Ocean Color Imager (GOCI) images. This showed a spatial rule that pico- and nanophytoplankton dominate the total absorption, rather than microphytoplankton. The annual Moderate Resolution Imaging Spectroradiometer (MODIS) a_ph(4 4 3) product from 2002 to 2019 was used to assess the transferability of those specific absorption coefficients and showed a good performance. The accurate acquirement of a_ph(λ) can provide basical datasets for further environment research, such as estimating primary ocean productivity.