Water Optical Properties and Water Color Remote Sensing in Optically Deep and Shallow Waters of Lake Taihu, China

In this study, Lake Taihu in Jiangsu Province of China, a typical large freshwater lake, is selected as the study area. Based on the field spectral measurements and laboratory analyses performed in October 2008, water optical properties and water color/quality remote sensing retrieval models in Lake Taihu were investigated. It was recognized that water quality varied a lot in different areas. Waters in Lake Taihu were classified as optically deep waters (ODWs) and optically shallow waters (OSWs). ODWs are the waters where the water depth is more than three times the measured Secchi Disk Depth (SDD), otherwise they are OSWs. Cyanobacteria blooms happen frequently in ODWs and the water is eutrophicated heavily. Whereas water is very clear with rare cyanobacteria blooms but many aquatic plants in OSWs. Focused on the two types of water areas respectively, the inherent optical properties (lOPs), apparent optical properties (lOPs) and reflectance spectra were analyzed, as well as their relationships to water quality parameters. Local optical parameters f and Q, which play significant roles in water quality parameters retrieval models, were also determined.;Measured remote sensing reflectance data were used to establish two-band and three-band models for chlorophyll-a (Chl-a) concentration estimation, results showed both models were suitable in ODWs. However, aquatic plants in OSWs had great influence on spectra, resulting in the inapplicability of the established models at these sites. Absorption and backscattering coefficients were used to remove those influences and simulate new set of remote sensing reflectance based on radiative transfer theory, which were proved reliable to establish Chl-a retrieval algorithms. Three-band model established by simulated spectra showed more satisfactory performance in whole ODWs, and performance of two-band model in OSWs was also enhanced much.;Several models were established to estimate total suspended solids (TSS) concentrations. Single band model, two-band model and first derivative model all showed good results with R2 greater than 0.8, RMSE lower than 8.3 g m-3 and mean RE lower than 35%. There is no obvious difference found between ODWs and OSWs when validating these models, indicating that aquatic plants has little influence on spectral bands which are sensitive to TSS concentrations. Though several empirical bands performed well but were unsteady, thus quasi-analytical algorithm (QAA) was modified on its reference wavelength and several parameters to estimate TSS from backscattering coefficients of total particles . The modified QAA algorithm was more stable than emprirical ones with fair accuracy.;In ODWs, a semi-empirical algorithm proposed by Simis (2005) was used to roughly estimate Chl-a and cyanobacteria phycocyanin (PC) concentrations. Estimated Chl-a was matched with measured Chl-a quite well but measured PC was lacked for validation. However, preliminary results were obtained on the spectral response of PC on reflectance, which would be helpful on algal biomass estimation. In OSWs, the contribution rates of the bottom and aquatic plants on spectra were also analyzed based on different water characteristics. Several groups of measured data with different depth, SDD transparency, aquatic plants height and coverage, were selected and compared thoroughly to estimate approximate contribution of bottom and aquatic plants on measured remote sensing reflectance in OSWs, which make well basis on establishment of accurate Chl-a retrieval models in OSWs of Lake Taihu.;To realize water quality monitoring of Lake Taihu remotely, Landsat-7 ETM+ imagery and MODIS Terra and Aqua data were applied to map the relative distribution of Chl-a and TSS. ETM+ band 4 was the most sensitive band to Chl-a concentrations, and the ratio of ETM+ band2/bandl had the strongest correlation with TSS concentrations. For MODIS data, the ratio of band2/bandl was most correlated with Chl-a concentrations, and band 1 was most sensitive to TSS concentrations. The mapping results showed rational distribution patterns for both Chl-a and TSS in Lake Taihu, except Chl-a distribution in OSWs where aquatic plants overestimated the retrieved Chl-a values. Mapping water quality parameters by combining satellite data could give more understandable and comprehensive explanation on water quality distribution patterns both spatially and temporally, and is also of great significance on cyanobacteria blooms monitoring.