利用RS和GIS的森林蓄积量偏最小二乘估测研究

Estimating Forest Volume Using RS and GLS Data Based on Partial Least Square Regression

摘要:

在运用RS、GIS可提取的因子为自变量的森林蓄积量遥感估测中，本文提出运用偏最小二乘(PLS)估计的数学模型。它能使蓄积量估测模型的稳健性增强，能够很好地克服传统的基于最小二乘(OLS)估计建立起来的模型无法真正克服样地少、各自变量之间相关性强及对蓄积量的解释性不强等问题，并应用于福建漳浦地区。结果表明：该方法有较高的预报精度，同时可大大降低成本。

Abstract:

In using RS and GIS data,factor as forest volume of self-changed quantity can be drawn among remote sensing, the paper proposes to apply a new mathematical model to partial least squares (PLS)regression,which can remarkably improve stability of estimating forest volume and can overcome more improtant problems in traditional ordinary least squares (OLS) for model established that is hardly to overcome less sample plot, strong correlation among self-changed quantity and weak explanation to the volume and etc., and application to Zhangpu district in Fujian province.The result shows that the PLS enjoys good estimation pricision and can cut down cost in application.