| Along with the fast development of quantitative remote sensing, the accuracy of atmospheric correction becomes increasingly important. For water color remote sensing, the methods of atmospheric correction for ocean Case-I water has been developing well. However, the atmospheric correction for Case-II water, especially lake, has been a major problem in the development of water color quantitative remote sensing. Due to the complex of water components and aerosol in different areas, using the default aerosol parameter for atmospheric correction will usually cause large errors, and so, the research on aerosol in the study area plays an important role in increasing the accuracy of atmospheric correction.This paper used 6 MODIS LIB image covering the Taihu Lake with 500 meter resolution and three dates:April 19th,2011, September 3rd,2011, and November 13th, 2011. Based on the radiative transfer model, in combination with synchronized water spectral in situ and historical weather data, the paper studied the atmospheric correction of Case-Ⅱ water at Taihu Lake, aiming to build a lookup table for atmospheric correction parameters and thus increasing the accuracy of the atmospheric correction for MODIS imagery in the study area. The results are as followed:1. Using radiative transfer simulation, this research conducted sensitivity analysis for 6S atmospheric correction, including factors such as atmospheric model, aerosol model, aerosol particles, aerosol optical thickness and geometry of observation. The result showed that the water vapor and ozone is not a major effect, and the reflectance change rate is less than 0.2% in different atmospheric correction environments; the aerosol optical thickness could be calculated from atmosphere vertical visibility. With the increased visibility, the reflectance rate decreases after correction and its decrease is mostly obvious when visibility change between 5 and 20km. But when visibility is larger than 20km, the aerosol doesn’t affect too much for the correction result. So it is feasible to use visibility parameter from visual interpretation of imagery to do atmospheric correction. The choice for different aerosol model has a tremendous impact on correction result. Different combination of aerosol particles affects the accuracy and offset in spectra shape.2. The paper had a sensitivity analysis for atmospheric correction parameters which used field spectra, satellite imagery and historical weather data to investigate the atmospheric mode and aerosol characteristics, and produced a look up table for these parameters according to their change in four seasons. The monthly mean water vapor value was calculated from the water vapor in recent three years provided by AERONET Taihu station. The monthly ozone concentration was fitted from 6S model. For the optimization of aerosol mode, this paper used synchronized field spectra and correction results from various combinations of aerosol particles. The optimization is based on the characteristics of seasonaliy and spectra curve, relative error in atmospheric correction. By minor adjustment, finally acquired the aerosol types in spring, summer and fall and established a look up table for regional atmospheric parameters.3. The result of atmospheric correction for MODIS imagery using parameters from look up table compared with the spectra in situ, the after-correction reflection in image had a good linear relationship in all bands and R2 is bigger than 0.8. The mean relative errors in the first 4 bands of MODIS are 17.2%,70.7%,22.4% and 10.6%, respectively, and the mean absolute errors 0.31%,0.4%,0.35% and 0.28%. The band 2 (859nm) had the largest error which was caused by the domination of scattering from this band and the lower reflectance from water.4. The paper demonstrated that correction method and result are more practical and closer to water spectra in situ compared with the top of atmosphere reflectance, FLAASH correction,6S correction using default settings and NIR-SWIR model. |
PDF Full Text Request