Application of hyper-spectral remote sensing in monitoring chlorophyll-aconcentration in drinking water source reservoir in North China

Remote sensing has been widely used for water quality monitoring in recent decades. Hyper-spectral remote sensing is a very effective technology for detecting large-scale water eutrophication, which has attracted lots of research in monitoring chlorophyll-a. In this paper, we used hyper-spectral remote sensing technology to present a method for monitoring chlorophyll-a concentration in Huangbizhuang Reservoir in Shijiazhuang, Hebei Province. In-situ hyper-spectral measurements were conducted by using the portable EKO MS-720 spectroradiometer at 10 different points in Huangbizhuang Reservoir, the source of drinking water for Shijiazhuang City and irrigation water for a large area of croplands along Shijin Irrigation Channel. Water samples were also simultaneously collected for laboratory analyses. Sample site position information was recorded via portable GPS. Chlorophyll-a concentration of the water samples were measured in laboratory by Acetone-spectrophotometric. The hyper-spectral data were converted into remote sensing reflectance. Then different band reflectance, reflectance ratio and other reflectance indices were designed and calculated. Linear correlation analysis between chlorophyll-a concentration and spectral reflectance, reflectance ratio and first-order differential of the water sample reflectance were also analyzed and compared. At last, the spectral reflectance ratio model and the first-order differential model were selected based on obtained correlation coefficient and significance. The results showed that Huangbizhuang Reservoir water chlorophyll-a concentration was low, with the highest concentration of 4.55 g·L^-1. It indicated that the reservoir water was in good condition. Spectral reflectance ratio model (R_705nm/R_680nm) showed close correlation with chlorophyll-a concentrations (r² = 0.736 6). On the other hand, the 696 nm first-order differential reflectance model showed a lot more significant correlation with chlorophyll-a concentrations in the entire analytical tests (r² = 0.875 5). This illustrated that the 696 nm first-order differential reflectance model was more effective for chlorophyll-a concentration monitoring in Huangbizhuang Reservoir. Through linear regression estimation, chlorophyll-a concentration in Huangbizhuang Reservoir was generally at the state of oligotrophication. Hence with regard to chlorophyll-a concentration, Huangbizhuang Reservoir water was suitable for domestic, industrial and irrigation use. The method proposed in this work had potential applications in environmental management for improved chlorophyll-a concentration monitoring efficiency in large-scale water bodies. It was also applicable in policy/decision- makings needed for early warning and prevention of water eutrophication.