Remote sensing of column-integrated chlorophyll a in a large deep-water reservoir

Abstract

The accurate assessment of total algal biomass in large deep-water reservoirs is challenging because the characteristics of thermodynamic stratification in such reservoirs differ noticeably from those of other types of inland waters. To address this challenge, we proposed a novel remote sensing model to derive column-integrated chlorophyll a (CIC), a better indicator than water surface chlorophyll a concentration (Csurf), for evaluating total algal biomass. The proposed algorithm was utilized to construct the patterns of spatiotemporal change in CIC in the Xin’anjiang Reservoir (the largest freshwater human-made reservoir in East China, located in Zhejiang Province) based on 97 cloud-free Sentinel-2 satellite images from 2016 to 2020. The results showed that the best index for developing the CIC remote sensing model was the average chlorophyll a concentration within the euphotic layer, which could be derived from Csurf and euphotic depth. The proposed model demonstrated good performance in deriving CIC in the Xin’anjiang Reservoir from Sentinel-2 images (yielding a mean absolute percent error of 20.7% and a root mean square error of 16.3 mg/m2). The values of the five-year averaged CIC and total algal biomass within the reservoir were 62.38 mg/m2 and 22.31 t, respectively. The CIC in the Xin’anjiang Reservoir displayed noticeable spatial variation, with the highest CIC appearing in the riverine zone and the lowest CIC appearing in the lacustrine zone. Clear seasonal variation in CIC was observed, with CIC exhibiting the order summer > autumn > spring > winter. Although the model was developed for a specific reservoir, we believe that this research can serve as a case study that illustrates the benefits of using CIC for assessing spatiotemporal variation in phytoplankton.