模糊回归在太湖悬浮物反演中的应用

Application of Fuzzy Regression in the Retrieval of Suspended Sediment in Taihu Lake

模糊回归分析是一种能直接处理不确定性的分析方法,突破了传统的“观测值的不确定性就是随机性”这一基本假设.本文将此方法应用于悬浮物的遥感定量反演问题中,在太湖水质的采样数据和Landsat7ETM遥感图像的基础上,应用模糊回归分析方法,针对悬浮物与遥感反射率数据之间的相关关系,建立了二者之间的模糊回归方程,并将该方法的计算结果与统计学中的最小二乘回归分析结果进行比较.结果表明:模糊回归模型比最小二乘回归模型提供了更多的信息量,更具优越性;用模糊回归模型进行反演,可信度更高,且更具推广价值.

Fuzzy regression analysis is a method directly dealing with uncertainty, which break through the basic hypothesis of “the uncertainty of observation value is randomicity”. This paper applied this method to remote sensing quantitative analysis of suspended sediment. Aiming at the correlation between suspended sediment concentration and remote sensing reflectance, a fuzzy regression equation was built using the water quality samples and Landsat7 ETM image of Taihu Lake. Comparing the result of this method with least squares regression method in statistics, it is obviously that fuzzy regression model provides more information and is more reliable and worth popularizing.