Remote estimation of chlorophyll a concentrations over a wide range of optical conditions based on water classification from VIIRS observations

The accurate estimate of surface chlorophyll a concentrations (Chla) by remote sensing presents a number of challenges where inherent and apparent optical properties have significant spatial or temporal variability. Indeed, Chla algorithms for Case 2 waters are often lake or region specific, and they are usually highly sensitive to changes in the dominant chromophoric constituents. This study develops and validates an absorption-specific approach to estimating Chla across an optically heterogeneous dataset. The approach is based on the classification of the optically dominant constituent. We tested this approach with in situ data from Taihu Lake, Poyang Lake, Chaohu Lake, Shitoukoumen Reservoir, Pearl River Estuary and Daya Bay as well as using HydroLight simulated data. The results show an improved performance when compared to most single Chla algorithms. We validated the approach with data from the Visible Infrared Imager Radiometer Suite (VIIRS). Results showed that this absorption-specific approach provided good Chla estimates over clear to very turbid waters with a wide range of optical conditions (R-2 = 0.76, r(RMSE) = 35%, n = 230, p < 0.01).