Remote Sensing Retrieval Model For Phytoplankton Chlorophyll-α Concentration Sea Pearl River Mouth Study

Since the reform and opening up, the economic developments quickly, which hadbeen seriously affected the environment and resources of the Pearl River Estuary,worsening environmental pollution, industrial and domestic sewage volume isincreasing, leading to the quality of Pearl River Estuary water is declining; theconcentrations of phytoplankton rising, resulting in big area of red tide frequentlyhappened, not only affected the nearby marine life, even became a threat to the livingenvironment of local residents. Chlorophyll a is the main pigment of phytoplanktonphotosynthesis, the concentration and dynamics of which reflected the concentration,biomass and distribution law of phytoplankton in water, which is an objectivebiological indicators reflect the situation of marine water quality, but also is one of themost important indicators of the eutrophication assessment of marine.In this paper phytoplankton chlorophyll concentration of the pearl river mouth asthe main research object, various types remote sensing image based on source ofphytoplankton chlorophyll-a concentration inversion, the BP artificial neural networkalgorithm, the research of Marine algae chlorophyll remote sensing of theconcentration of the inversion method and the offshore waters for ecologicalenvironment monitoring provides the basis.In this study, by combining optical remote sensing and radar remote sensingimage to extract remote sensing inversion of marine algae chlorophyll-a concentrationof BP artificial neural network model. Research work after a few stages:(1) based onCCD image of phytoplankton chlorophyll spectrum characteristics are analyzed, theselection of the CCD data CCD1, CCD2, CCD3band as the characteristic band;(2) ofthe radar image to the scattering coefficient and use after Cloude-Pottierdecomposition principle through the image preprocessing the incoherent targetdecomposition, get average scattering Angle alpha, scattering entropy H with algaeparameters such as chlorophyll concentration correlation analysis between, final HH,VV polarization of backscatter coefficient and average scattering entropy H;(3) basedon the six parameters are different combinations to establish various model isanalyzed and determined the three layers of BP neural network model as the finalmodel, then build up each parameter combination with actual chlorophyllconcentration of linear model, will get the predicted values and, by fitting and the BPneural network model accuracy of comparison BP artificial neural network model has the highest.The results showed that:(1) between the phytoplankton chlorophyllconcentration and CCD each band spectrum reflectance and radar image to thescattering coefficient and average after scattering entropy H has certain correlation;(2)when using a combination of optical data and radar data building BP artificial neuralnetwork model, which's precision higher than used alone one data set up linear model.(3) BP artificial neural network model of the adaptive ability of organization can verygood simulation chlorophyll concentration and remote sensing between theparameters of the complicated nonlinear relation.