Evaluation of remote sensing algorithms for cyanobacterial pigment retrievals during spring bloom formation in several lakes of East China

Using reflectance data and water sample data from 75 stations in several eutrophic lakes of East China (Lake Taihu, Lake Gehu, Lake Dongjiu) between 23 April and 3 May 2010, we evaluated several recently proposed remote sensing algorithms to estimate chlorophyll-a concentrations (Chla, 1.0–42 μg/L) and phycocyanin pigment concentrations (PC, 0.1–7.7 μg/L). These lakes experience phytoplankton blooms of the cyanobacteria Microcystis aeruginosa every year due to eutrophication. It was found that after local tuning of the algorithm parameterizations, most algorithms yielded acceptable results for Chla retrievals while accurate PC retrievals were more challenging due to changing species composition (PC:Chla ratios) and low PC concentrations. For the data ranges in the study region, the best Chla algorithm yielded RMSErel (Relative Root Mean Square Error) of ~ 46% (R2 = 0.92, n = 75) and the best PC algorithm yielded RMSErel of ~ 83% (R2 = 0.88, n = 75). Based on these observations, it is recommended that local tuning of algorithm parameters should be performed for remote sensing applications, and future efforts should emphasize on application of the algorithms to satellite data.

► Several algorithms of chlorophyll-a (Chla) and phycocyanin (PC) were evaluated.
► PC was more challenging due to changing species composition (PC:Chla) and low PC.
► The best Chla and PC algorithms yielded RMSErel of ~ 46% and ~ 83% respectively.
► Local tuning of algorithm parameterizations should be recommended for applications.