摘要
根据2019年Landsat8 OLI光学遥感数据和塞罕坝机械林场地区2020年森林资源二类清查数据以及临时样地数据,编制塞罕坝机械林场华北落叶松林立地指数表,并得到华北落叶松林小班立地指数。在小班尺度上获取影像的光谱、纹理、植被指数、线性变换、地形等6个种类59个特征,利用极限梯度提升算法(XGBoost)提取华北落叶松林小班林分因子信息估测立地指数。结果表明:遥感-立地因子特征立地指数估测模型的精度更高,准确度为0.877376,召回率为0.894318,精确率为0.926923,F_(1)值为0.908221。因此,在立地指数估测中,将多光谱遥感特征与立地因子提取相结合具有应用潜力。
Based on Landsat8 OLI optical remote sensing data in 2019,forest resource inventory data in 2020 of the Saihanba Mechanized Forest Farm,and temporary sample site data,a site index table for Larix principis-rupprechtii forest in the Saihanba Mechanized Forest Farm was compiled and the subcompartment site index of L.principis-rupprechtii plantation was obtained.The 59 features of 6 categories including spectral,texture,vegetation index,linear transformation,and terrain were extracted at the subcompartment scale from the image.The Extreme Gradient Boosting algorithm(XGBoost)was used to extract stand factor information of L.principis-rupprechtii forest in subcompartment,and estimate the site index.The results showed that the accuracy of the remote sensing-stand factor feature site index estimation model was higher,with an accuracy of 0.877376,recall of 0.894318,precision of 0.926923,and F_(1) value of 0.908221.Therefore,combining multispectral remote sensing features with stand factor extraction has potential applications in site index estimation.
作者
李金恬
范文义
Li Jintian;Fan Wenyi(Northeast Forestry University,Harbin 150040,P.R.China)
出处
《东北林业大学学报》
CAS
CSCD
北大核心
2024年第4期72-81,88,共11页
Journal of Northeast Forestry University
基金
国家自然科学基金项目(31971654)。
