BOX-COX变换在遥感数据建模中的应用

Application of Box-Cox Transformation to Remote Sensing Models

以山东省2006年共15景TM数据和2002年的ETM+数据为基础,结合地面542个样地调查数据,采用多元回归估计的方法,建立蓄积量与遥感信息之间的回归方程,对森林蓄积量进行估测。但由于TM1—TM7波段的灰度值和蓄积量值之间的线性关系不够明显,通过对参与建模变量进行了Box-Cox变换,建立森林蓄积量反演模型,提高了模型的估测精度,解决了在大尺度范围内利用TM数据对森林蓄积量估测问题。

A regression equation between forest volume and remote sensing information was established by multiple regression estimation according to TM data from 15 scenes in Shandong Province for 2006 and ETM＋data for 2002, combining with the ground survey data of 542 sample plots. The forest volume was estimated by the regression equation. An inversion model was given by Box-Cox transformation of variables in the model due to the unobvious linear relationship between the gray value in TM1-TM7 and the forest volume, which could improve the model estimation accuracy and solve the problem of using TM data to estimate forest volume on a large scale.