| Suspended sediment concentration (SSC) and downwelling diffuse attenuation coefficient (Kd) are among the most important parameters of water quality, which are tightly related to coastal ecosystems. Field measurements of SSC and Kd are money-and time-costly and limited to small temporal and spatial regions. Therefore, an accurate and timely remote-sensing monitoring is desirable.For this target, three different aspects, namely, processing of in situ measured data, water optical modeling, and ocean color remote sensing, were considered in the present study.The field area of the study was the East China inland, estuarine, and coastal waters with extremely wide variations (more than three orders) in SSC and Kd. This obstacle makes a task of development of reliable methods for water quality evaluation an extraordinary one.The one of the largest my concerns was development an accurate remote sensing SSC retrieval algorithm, which is assumed to be sensitive enough to small variations in SSC and stable enough to avoid environmental impacts, for example, aerosol scattering.Another my concern is accurate processing in situ radiometric measurements. With this aim, I have compared six modes of deriving remote-sensing reflectance Rrs(λ) from the field radiometric measurements. Results indicated that all models yield similar spectral features of Rrs(λ), however, numerical values may be quite different.A water optical modeling was concentrated on three separate issues:1) a connection between SSC and inherent optical properties (IOPs),2) a connection between IOPs and the underwater remote-sensing reflectance rrs(λ), and 3) a connection between the rrs(λ) and the above-water remote-sensing reflectance Rrs(λ). I studied here spectral features of Rrs(λ) and how they related with the spectral features of IOPs.Twenty eight models (mostly empirical) for SSC retrieval from the Rrs(λ) spectra were analyzed in the present study with two different purposes:1) To find the models posessing the best performance and 2) To validate semi-empirical models derived from the water optical modeling. Statistical analysys helped find out eight best models, six of them were derived for the East China reagion while two others-for the European waters.An image processing also is the one of the most important processes to perform accurate remote sensing monitoring. Two different approaches were applied in the study. On the one hand, we developed an analytical algorithm (AA) for atmospheric correction (Sokoletsky et al.,2014a; Yang et al.,2015) and simplifiedcompared it to the Second Simulation of a Satellite Signal in the Solar Spectrum (6S). Results showed that the AA-derived Rrs(λ) do have the similar shape, close values and the same main maximum as the 6S-derived Rrs(λ). On the other hand, we have applied the commonly used atmospheric correction procedures with the corresponding NIR-SWIR modification. Such NIR-SWIR approach merges an advantage of the NIR-SWIR approach and the "best fit" strategy. Comparison of obtained results with in situ measurements has proven that this NIR-SWIR approach is reasonable. Although the Rrs(λ) values at the blue, green and purple bands were overestimated, the bands centered in 645.5 nm and 553.6 nm shown to be pretty acceptable, yielding the relative errors of 11.7% and 17.7%, respectively.One more remarkable issue of remote sensing was considered in the present study, namely, an efficiency of image processing. A fast image processing system was designed for all kinds of satellite images. Based on the algorithms we discuss in this study, image processing for the Moderate Resolution Imaging Spectroradiometers (MODIS) was realized by the object-oriented programming language C++. With the rapid development of Graphics Processing Unit (GPU) last decades, the GPU general purpose computing demonstrates a new chance to accelerate substantially an efficiency even for real-time monitoring. GPU is merged to our fast image processing system.Retrieval of SSC and Kd was realized by the MODIS/Terra and MODIS/Aqua satellite sensors. A geographic area located between the 27— 35°N and 119— 125°N was selected to analyze flood and dry seasons from 2000 to 2015. Results showed the large difference between these seasons in the aspect of spatial pattern and SSC levels: in the most part of area the SSC level during a dry season was higher than during a flood season. Correspondingly, the area occupied by the high SSC>80 gm-3, within the Subei Bank of Yellow Sea and Zhejiang Coastal area, was estimated almost twice larger during a dry season than during a wet season. |
PDF Full Text Request