The Spectral Characteristics Of Inland Water Bodies And Chlorophyll A Concentration Of Remote Sensing Model,

Inland water quality directly affects national product and people's daily water, accurate and efficient water quality monitoring is particularly important. Conventional water quality monitoring and analysis is complex, long-term and needs a lot of financial resources, Human and material resources, and also limited by climatic and hydrological conditions. Remote sensing technology applied gradually inland water quality monitoring, it can reflect the water quality in the spatial and temporal changes, find the Sources of pollution and the characteristics of pollutants transport, which is difficult to reveal using conventional water quality monitoring. There are advantages of monitoring a wide range, fast, low cost and long-term dynamic monitoring.This paper chose Baiyangdian as research Area, analysis spectral characteristics of inland water bodies, found out the best band or band group, and then established quantitative inversion model of Chlorophyll-a concentration based on field measured spectrum and multi-spectral remote sensing image which were suitable for Baiyangdian Area.Semi-empirical method was used to establish ground-based hyperspectral quantitative estimation of Chlorophyll-a concentration. Construction and application of the model were achieved the following steps: 1) based on the analysis of measured spectral characteristics of inland water, it identified wave peaks and trough, and found out the maximum correlation coefficient. 2) Took Ratio of band on maximum positive correlation and band on negative correlation as variables, Chlorophyll-a concentration as the dependent variable, establish regression model with different Mathematical Methods. 3) Select the model of the highest precision as the quantitative inversion model of Chlorophyll-a concentration based on field measured spectrum. It use empirical methods to establish quantitative inversion model of Chlorophyll-a concentration based on CBERS02-CCD multi-spectral remote sensing image: analysis the correlation coefficient between Chlorophyll-a concentration or its natural logarithm and image single band or bands combinations, choose bands combination with the maximum correlation to linear regression. Comparing the measured data and simulation results, it analysis the model accuracy. Following results were obtained: 1) the correlation of Chlorophyll-a concentration and the original reflectance of water is not high,while the correlation of Chlorophyll-a concentration and the normalized reflectance of water was significantly increased. Wavelength of 557nm and 665nm is higher negative correlation, and 722nm and 717nm have a higher positive correlation. 2) Took Ratio of band on maximum positive correlation and band on negative correlation as variables, Chlorophyll-a concentration as the dependent variable, establish regression model with different Mathematical methods (such as exponential, logarithmic, power, linear and quadratic function), and found that the linear and quadratic model were better. After accuracy verification, found that the calculation results of the linear model using R717/R557 for chlorophyll-a concentration quantitative inversion had the smallest relative error, the model was y = 83.248x - 45.239. 3) Analysis the correlation coefficient between Chlorophyll-a concentration or its natural logarithm and image single band or bands combinations, choose bands combination b1/(b2×b3) with the maximum correlation to linear regression, obtained two models: and . After accuracy verification, found that t: poorer accuracy of linear model results,and higher accuracy of the exponential model results, with an average accuracy of about 10%, and ultimately identified as chlorophyll a concentration inversion model based on remote sensing image for Baiyangdian Area.In this paper, the model based on field measured spectra was higher precision, and the accuracy of model based on remote sensing image was s not high, there may be three reasons: 1) the remote sensing image and the time when the water quality measured were not synchronized; 2) atmospheric correction model and parameters for quantitative remote sensing inversion may have a certain impact, 3) there are non-pure pix in image of inland water.