Optical Properties Of Chlorphyll-a In Reservoir Water And Its Concentration Inversion Models By Remote Sensing

Qingshan Reservoir was selected in this study for its representativeness in Hangzhou. After field experiment and observation, data of water apparent optical parameters, inherent optical parameters and water components concentrations were observed according to the formation of algal blooms in different periods and satellite observation time. Seasonal variations of water components optical properties were studied and reasons of water spectral reflectance changes were analysed. On this basis, the present aquatic bio-optical models theory were combined to build remote sensing models using in-situ spectrum measurements and image of HJ-1, which could extract concentration of water components at higher retrieval precision. The conclusions of this study would promote the development of models for extracting water chlorophyll a and suspended substance concentrations, and make it better service to meet the demand of water environment monitoring.The major contents in this study and main results are the following:(1) In-situ spectrum measurement and laboratory absorption measurement. Above-water spectrum measurement method recommended by NASA ocean optical measurement criteria was employed to measure the in-situ spectrum of surface water. The radiance of standard plate, water surface, and sky illuminance was measured, respectively, to calculate the remote sensing reflectance of the water surface. In-situ water samples were also collected and measured in laboratory to analyze water components absorption spectrums, with quantitative filtration technology used for absorption spectrum measurements of suspended particle and the absorption coefficients of CDOM were measured the filtrate light attenuation to approximate computation by spectrophotometer. Water components concentrations were measured meantime.(2) Seasonal variation of optical properties for water components. The measured water reflectance of different periods showed large variation. Meanwhile algae characteristics were obvious. Chlorophyll a absorption peak nearby 675nm and fluorescence peak nearby 700nm were characteristic peaks of the remote sensing reflectance spectrum, but the measurement values of remote sensing reflectance in different time varied. The little reflectance peak nearby 810nm caused by suspended particle was important spectral characteristic of suspended inorganic substance existence, the 810nm peaks'height were basically the same of measurements in different time. Water components absorption coefficient of measurements in different periods were different, and their absorption properties had obvious season variation. There were two forms for suspended particle absorption. The first form was that de-pigmented particle dominated the absorption spectrum of suspended particle at the blue-green band, absorption spectrum curve of suspended particle was similar to de-pigmented particle, namely, absorption coefficient decreased along with the wavelength increases. But pigment particle absorption dominated the absorption spectrum of suspended particle at the red band, especially there are obvious chlorophyll a absorption peaks near 675nm. The second form was that pigment particle dominated the absorption spectrum of suspended particle, absorption spectrum curve of suspended particle was similar to pigment particle, obvious chlorophyll a absorption peaks were near 440nm and 675nm. Absorption spectrum of CDOM decreased exponentially along with wavelength increases, which was tend to zero near 700nm. The difference of CDOM absorption spectrum of measurements in different period was mainly in the short wavelength band, but not significant.(3) Remote sensing inversion model of water components concentrations. With the in-situ spectrum measurements and image of HJ-1, a database for analysis was developed based on in-situ spectrum measurements and the chlorophyll a concentrations data. Through the processing of differential and normalization to spectrum data, the correlation relationships between chlorophyll a concentrations and processed spectrum reflectance were analyzed. The bands with high correlation coefficients were selected as sensitive bands to develop statistical regression model. It was concluded that the model of combined bands showed better fitting precision compared than single bands based on normalized remote sensing reflectance, especially the combined bands of R702/R674 made the best fitting result. The model of single bands showed better fitting precision compared to combined bands based on the first order differential remote sensing reflectance, especially the single bands of R648, which the fitting result was the best. Meanwhile, with three-band model as the theoretical basis, the three-band model for retrieving chlorophyll a concentration was proposed based on in-situ spectrum measurements. According to iterative result, three optimal bands (λ1,λ2,λ3) of the three-band inversion model were 680nrm,697nm and 724nm, respectively. After geometric correction and atmospheric correction for HJ-1 image, correlation relationships between chlorophyll a/suspended particle concentrations and remote sensing reflectance of satellite data were analyzed, and inversion models were developed. The results showed that the correlation coefficient of suspended particle concentrations and remote sensing reflectance was significant, while the correlation coefficient of chlorophyll a concentrations and remote sensing reflectance was not significant for some objective reason such as sampling time etc. Finally the combined band of band2/band3 of HJ-1 image remote sensing reflectance was used to develop the remote sensing inversion model of suspended particle concentrations. The model was also applied to HJ-1 image for retrieving suspended particle concentrations. Classified map was generated according to suspended particle concentrations using the software of ArcGIS.