基于分区暗像元和Spline插值方法估算太湖气溶胶光学厚度

传统暗像元大气校正算法认为研究区域上空的气溶胶光学厚度呈均匀分布状态。对于Ⅱ类水体,尤其是气溶胶类型复杂的内陆湖区,暗像元算法的均匀性假设将不再适用。针对传统暗像元算法的不合理性,本研究将太湖湖区划分为9个子区域,每个子区域利用传统暗像元算法估算其气溶胶光学厚度,然后结合Spline插值算法获取整个太湖的气溶胶光学厚度信息,并以传统暗像元大气校正算法作为参照,探讨与分析分区暗像元算法的精度状况。通过本文的研究可知：气溶胶光学厚度是遥感大气校正的关键参数;在2003年10月28日,受西北风的影响,太湖上空的气溶胶光学厚度呈湖南低,湖北高的分布模式;分区暗像元大气校正算法获取的气溶胶光学厚度平均值为0.79,标准偏差为0.099,标准偏差与平均值的比值为12.58%,与传统暗像元算法相比,分区暗像元算法综合考虑了水体上空气溶胶光学厚度空间分布的不均匀性,进而有利于改善大气校正的精度。

Using the Sub-area Dark Pixel and Spline Interpolation Approach to Estimate the Aerosol Optical Thickness on Taihu Lake

The traditional atmospheric correction algorithms consider that aerosol optical thickness is homogeneous distribution above the research regions.For the caseⅡwaters,especially in the inland lakes whose aerosol types are very complicated,the approximation of homogeneous distribution becomes improper.Taking into account the unreasonable traditional approach,the study divides the Taihu Lake into 9 sub-areas.The aerosol optical thickness was estimated by the traditional dark pixel algorithm.Then,spline interpolation approach was used to calculate the aerosol optical thickness of each pixel on Taihu Lake based on those 9 dark pixels.According to the results of this paper,it was found that the aerosol optical thickness was an important parameter of atmospheric correction algorithm.Influenced by the northwester,the aerosol optical thickness was lower on the south lake and higher on the north lake on 28 October,2003.The mean aerosol optical thickness,calculated by the sub-area dark algorithm,was 0.79,the standard deviation was 0.099,and the ration of standard deviation and the mean aerosol optical thickness was 12.58%.Comparing with the traditional dark pixel algorithm,the sub-area dark algorithm has comprehensive consideration of the inhomogeneity distribution of aerosol optical thickness in space,and could be beneficial to the accuracy improvement of atmospheric correction.