A Study On The Chlorophyll_a, Suspended Matter And Gelbstoff Concentrations Based On MODIS Data

Chlorophyll_a, suspended sediment and yellow substance are three important water quality parameters, they play an important role in the marine environment and eutrophication of water. Ocean Color Remote sensing is a method to derive water quality parameters based on the optical properties of biological in use of the spectral characteristics of the three components of ocean color. Waters are divided into case I waters and case II waters. Now, The research on case I waters is very Mature. Because of its complexity and regional, the research on case II waters is very hard. Consequently, there is considerable interest in the use of remotely sensed data to research case II waters.This theesis makes use of MODIS data and measured data about remote sensing reflectance and three components concentrations to establish several neural network models, and the actual data was obtained from the Ocean Color Experiments in Pearl River Estuary. The models were established and validated with the measured data. The Back Propagation neural network model is established to derive three components concentrations Separately. It is shown that the averaged relative error of chlorophyll-a is 28.59% and the correction coefficient is 0.987. The averaged relative error of suspended matter is 27.34% and the correction coefficient is 0.931. The averaged relative error of yellow substance is 26.72% and the correction coefficient is 0.831. The Back Propagation neural network model is established to derive three components concentrations Contemporary, the study demonstrates that the model can get the better prediction results with a simple algorithm, the averaged relative error of chlorophyll-a is 35.52% and the correction coefficient is 0.982. The averaged relative error of suspended matter is 34.07% and the correction coefficient is 0.877. The averaged relative error of yellow substance is 34.19% and the correction coefficient is 0.689. The General Regression Neural Network model is established to retrieve the suspended matter concentration, the averaged relative error is 17.01% and the correction coefficient is 0.965. At last, combining the MODIS data and after atmospheric Correction of MODIS data to derive remote sensing reflectance, the the concentrations of chlorophyll, suspend matter and yellow substance are mapped with the model and remote sensing reflectance.