基于机载激光雷达数据的天山云杉林蓄积量反演模型构建

Construction of the Accumulation Inversion Model of Picea schrenkiana var.tianschanica Forest Volume Based On Lidar Data

构建多变量的天山云杉林Schumacher蓄积收获模型,提高林分蓄积量反演精度,获得便捷、快速提取森林蓄积信息的技术方法,为探索山地天然林精准监测与评价提供技术途径。以2020年激光雷达影像与样地实测数据为研究材料,在激光雷达影像图中提取遥感因子,代入Schumacher蓄积收获模型中,通过再参数化来构建适用于天山云杉林的可变密度收获预估模型,并进行精度检验。结果表明,激光雷达影像分辨率较高,进行点云数据处理后,树高提取精度为89.64%,每公顷株数提取精度为85.13%,坡度提取精度为84.26%,坡向提取精度为84.26%,海拔提取精度为97.25%。结合Schumacher蓄积收获模型构建天山云杉蓄积量反演模型,R^(2)=0.80,将检验数据代入模型中,估测蓄积量与实测蓄积量平均精度为90.22%,模型的拟合度较好。研究将立地因子、林分密度、林龄等变量引入Schumacher蓄积收获模型,对于天山云杉的蓄积估测精度有较大提高,优于以往经验模型,满足新疆山地天然林数字经营管理的标准。

To construct a Schumacher accumulation and harvest model of Picea schrenkiana var.tianschanica with multivariable,to improve the inversion accuracy of stand accumulation,to obtain a convenient and rapid technical method for extracting forest accumulation information,and to provide a technical approach for exploring the accurate monitoring and evaluation of mountain natural forests.Using the 2020 lidar images and the actual measurement data of the sample plots as the research materials,the remote sensing factors were extracted from the lidar images and substituted into the Schumacher accumulation and harvest model,and the variable density harvest forecast suitable for P.schrenkiana var.tianschanica was constructed through reparameterization.The model was estimated,and the accuracy was checked.The resolution of the lidar image was relatively high.After processing the point cloud data,the tree height extraction accuracy was 89.64%,the number of trees per hectare extraction accuracy was 85.13%,the slope extraction accuracy was 84.26%,and the aspect extraction accuracy was 84.26%.The extraction accuracy was 97.25%.Combining the Schumacher accumulation and harvest model to construct the Tianshan spruce accumulation inversion model,R^(2)=0.80,substituting the test data into the model,the average accuracy of the estimated accumulation and the measured accumulation was 90.22%,and the model had a good fit.The introduction of site factors,stand density,forest age and other variables into the Schumacher accumulation and harvest model has greatly improved the accumulation estimation accuracy of P.schrenkiana var.tianschanica,which is better than previous empirical models,and meets the digital management standards of the mountainous natural forests in Xinjiang.