Remote sensing of freshwater cyanobacteria: An extended IOP Inversion Model of Inland Waters (IIMIW) for partitioning absorption coefficient and estimating phycocyanin

Phycocyanin primarily exists in freshwater cyanobacteria. Accurate estimation of low phycocyanin concentration (PC) is critical for issuing an early warning of potential risks of cyanobacterial population growth to the public. To monitor cyanobacterial biomass in eutrophic inland waters, an approach is proposed to partition non-water absorption coefficient (at-w(λ)) into the contribution of colored dissolved matter (CDM), non-phycocyanin pigments, and phycocyanin with the aim of improving the accuracy in remotely estimated PC, in particular for low PC. The proposed algorithm extends the IOP Inversion Model of Inland Waters (IIMIW) that derives at-w(λ) and chlorophyll-a concentration from remote sensing reflectance. The extended IIMIW retrieves absorption spectra of both CDM (acdm(λ)) and phytoplankton (aph(λ)) with R2 ≥ 0.80 and a relative root mean square error (rRMSE) ≤ 31.79% for acdm(412), aph(443), aph(620), and aph(665) when validated with data collected in 2010 from three Indiana reservoirs. In fact, comparison of our algorithm with other partitioning models demonstrates the new algorithm to be more suitable for inland waters. The algorithm also achieved more accurate PC estimation with R2 = 0.81, rRMSE = 33.60%, and mean relative error (RE) = 49.11% than the widely used semi-empirical algorithm with R2 = 0.73, rRMSE = 45.09%, and mean RE = 182.29% for the same dataset. The validation of our algorithm against the data collected in other years shows that the proposed algorithm worked for a wide range of limnological conditions. In particular, low PC (PC ≤ 50 mg m− 3) values of for all datasets used in this study were well predicted by the proposed algorithm.