Bio-Optical Inversion in Highly Turbid and Cyanobacteria-Dominated Waters

Phytoplankton pigment absorption data from algal-bloom-dominated waters are highly desirable to better understand the primary productivity and carbon uptake by algal biomass in a regional scale. However, retrieving phytoplankton pigment absorption coefficients, in turbid and hypereutrophic waters, from above-surface remote sensing reflectance (R_rs) is often challenging because of the optical complexity of the water body. In this paper, a quasi-analytical algorithm has been parameterized using in situ data to retrieve inherent optical properties from R_rs(λ) in highly turbid productive aquaculture ponds, where the phytoplankton absorption coefficient (3.44-37.67 m^-1 ) contributes 54 % of the total absorption at 443 nm (4.99-47.21 m^-1). The model was validated using an independent data set by comparing the model-derived optical parameters with in situ measured values. The absolute percentage error (assuming no error in the in situ measurements) of the estimated total absorption coefficient at( λ) varied from 15.22 % to 24.13 % within 413-665 nm, and the overall average error was 19.87 %. Maximum and minimum errors occurred at 443 and 665 nm, respectively. Similarly, the percentage error for the phytoplankton absorption coefficient a_φ(λ) varied from 15.9 % to 41.27 % within the 413-665-nm range, and the average error was 27.24 %. The spectral shape of modeled a_φ(λ) matched very well (R² = 0.97) with the measured a_φ(λ). A supplementary method was also developed to retrieve first-order estimates of colored detrital matter absorption coefficients a_CDM( λ) from subsurface remote sensing reflectance r_rs( λ) using an empirical approach. Results reveal that the retrieval accuracy of a_φ(λ) improved after incorporating the first-order estimates of a_CDM(λ) in the algorithm.