The Effect Of Atmospheric Correction On The Estimated Precision Of Remote Sensing In Case â…¡ Water

The quality and precision of remote sensing information is required higher with the fast development of remote sensing. For water remote sensing, ninety percent information received by remote sensor comes from atmospheric effect, and only less than ten percent energy includes water information. It’s the largest difference between water remote sensing and land remote sensing. Separating water-leaving radiance from atmospheric signal is the key for the application of water remote sensing, which helps to improve the precision of the following research.In this article, the principle and method of atmospheric correction in Case II water was explained. Based on MODIS data、MOD03geometrical data and meteorological data, modified Gordon model method and a new algorithm based on the look up table which founded by6S model were used to correct atmospheric effect, and analyzing the results of atmospheric correction with field spectral data. The concentration of Chl-a was estimated by remote sensing reflectance of water after atmospheric correction, and compared with measured value, which helped to research the influence of atmospheric correction on the estimated precision of remote sensing in Case II water.The results in this study were showed as follow:(1) The variation characteristic of aerosol optical depth at550nm in Taihu Lake estimated by dark object method decreased gradually from west to east. Its values were from0.53to0.83, and the average was0.64.(2) The reflectance after atmospheric correction by6S model was compared with measured spectral data, and the average relative error decreased from415.72%to28.15%. The distributive properties of the contribution rates of Rayleigh scattering and aerosol scattering were less in Taihu Lake Center than other areas, and the reflectance of water was opposite. The contribution rate of Rayleigh scattering reached over sixty percent at shorter wavelength, aerosol scattering was larger at shorter wavelength than that at longer wavelength, and reached52%at NIR. The contribution rate of water reflectance was nearly13%at all bands.(3) The results of modified Gordon model method were coincident with measured data, and its relative error decreased to30.49%. The distributive properties of the contribution rates of Rayleigh scattering, aerosol scattering and water reflectance were similar with that of6S model. The contribution rate of whitecap was less in Taihu Lake Center, and its value was from0.036%to0.125%. (4) The variation property of Chl-a was higher in Meiliang Bay, Zhushan Bay and east Taihu Lake. Using the reflectance model r2*(r3-1-r4-1) to estimate the concentration of Chl-a, the relative errors of sampling point13and14corrected by6S model decreased to14.91%and24.38%, and that by modified Gordon model were34.64%and33.33%. Using the reflectance model r1*(r3-1-r4-1), the relative errors of sampling point13and14corrected by6S model decreased to26.48%and33.60%, and that by modified Gordon model were25.74%and17.36%.