A semi-analytical algorithm for remote estimation of phycocyanin in inland waters

Phycocyanin (PC) is the unique and important accessory pigment for monitoring toxic cyanobacteria in inland waters. In this study, a semi-analytical algorithm combining both three band indices and a baseline algorithm (TBBA) was developed to estimate PC concentrations and then tested in three eutrophic and turbid reservoirs. TBBA does not need to optimize wavelengths as either the traditional baseline algorithm or three-band algorithms does when it is used across different study sites. TBBA evidently corrects some effects of absorptions due to colored detritus matter and other pigments and backscattering of water column. TBBA accurately estimated PC concentrations with R2 = 0.8573 and rRMSE = 31.4% for water samples with the PC range from 1.4 mg m− 3 to 146.1 mg m− 3. Particularly, TBBA outperformed three-band algorithms and a previously proposed semi-empirical algorithm for the prediction of low PC (PC ≤ 50 mg m− 3) concentration. Further analysis reveals that both the variations of PC:Chl-a and PC:TSM are important factors influencing the performance of all PC algorithms examined in this study and more efforts are required to improve the performance of TBBA on water samples with low PC concentration.

► We develop a new model to estimate phycocyanin (PC) without optimizing wavelengths in inland waters.
► The new model bio-optically interprets two widely used three-band and baseline algorithms.
► Accuracy of estimated PC, particularly for low PC, is much improved compared to existing algorithms.
► Our model imposes a significance for early warning of cyanobacterial blooms.
► The model has potentials to be an alternative technique monitoring cyanobacterial growth.