The Determination of in Situ Phytoplankton Spectral Absorption Coefficients: Direct Measurements, Modeled Estimates, and Applications to Bio-Optical Modeling

The determination of absorption by particles using the quantitative filter technique requires correction for the amplified pathlength. A spectrally-dependent correction factor was developed that reduced errors in the estimation of magnitude and spectral shape of particle absorption to less than 20% and 15%, respectively. Previous models, with no spectral dependence, overestimated the magnitude and shape of the absorption by 20-90% at low sample loads. A model was developed to resolve in situ phytoplankton absorption from a measured in situ total absorption spectrum which includes water, dissolved organics, particulate detritus, and phytoplankton. The model was tested on a set of absorption spectra obtained from the productive waters around the San Juan Islands, WA. Results indicated that the model can predict the spectral shape of phytoplankton absorption (r^2 > 0.9) and the total absorption by phytoplankton with <27% error. Application of this model to data collected from the Sargasso Sea resulted in errors less than 5%. A second model was developed to extract in situ phytoplankton absorption from measured in situ spectral reflectance. The model is a spectral mixture approach in which the reflectance spectrum is deconvolved into component backscattering and absorption spectra. Results indicate that phytoplankton spectral absorption coefficients can be predicted (r2 > 0.91) with < 50% error. The direct determination of quantum yield of carbon assimilation requires measurements of carbon uptake rates, phytoplankton spectral absorption coefficient, and scalar spectral irradiance. A sensitivity analysis was performed in which spectral phytoplankton absorption coefficients were approximated by spectrally-averaged absorption coefficients or by constant chlorophyll-specific absorption coefficients and the scalar spectral irradiance was approximated by spectral vector irradiance, spectrally-averaged vector irradiance, and spectrally-averaged scalar irradiance. All approximations resulted in overestimations in quantum yield. Errors associated with the non-spectral approximation were < 50% in coastal green waters but exceeded 50% in blue oligotrophic or yellow eutrophic waters.