Height Extraction Of Spruce Forest In Picea Schrenkiana Var.Tianschanica Based On Different Point Cloud Data

Although traditional optical remote sensing has been widely used in forestry,it cannot satisfy the inversion of forest single-wood parameters due to the relatively low spatial and spectral resolution.In recent years,Unmanned Detection And Ranging-Digital Aerial Photography(UAV-DAP)and Light Detection and Ranging(Li DAR)remote sensing technologies have developed rapidly,and the resulting point cloud data can not only extract forest parameters at the stand scale,but also provide a technical way for inversion of forest parameters at the single wood scale.In this paper,relying on point cloud data generated from UAV aerial photography and airborne Li DAR data,we conducted a study on the number of single trees of Picea Schrenkiana var.tianschanica,tree height extraction and influencing factors,and the main findings are as follows:(1)In terms of single-wood identification,the best neighborhood radius for obtaining single-wood information based on UAV-DAP and UAV-DAP&Li DAR data is 1 m.The best Gaussian smoothing window based on Li DAR data is 5 × 5(dynamic window),in which the single-wood segmentation accuracy is higher using airborne Li DAR point cloud data,while the difference in single-wood segmentation accuracy between UAV-DAP and UAV-DAP&Li DAR point cloud data is not significant,and overall all three data sources can obtain more accurate single-wood segmentation information.(2)In terms of single wood tree height extraction,the average single wood tree heights were extracted with 88.36%,89.67% and 88.63% accuracy based on the point cloud data generated by UAV-DAP,Li DAR and UAV-DAP&Li DAR,respectively,using the local maximum method,with the highest average single wood tree accuracy based on Li DAR data.There was no significant difference in tree height extraction information among the three data sources,and the overall tree height accuracy was above 85%,which reached the standard of Class C(allowable error 15%)of the technical regulations of forest resources planning and survey(GB/T 26424-2010).(3)At the level of densities,the average tree height accuracy of all three data sources was above 85%at low and medium densities,and above 70% at high densities.The average tree height extraction accuracy of the three data sources was highest in the sample with low densities,less satisfactory in the sample with high densities,and more stable in the average tree height with low and medium densities.The results show that there is no big difference between the three data sources in tree height extraction information,and after processing,the effect of Li DAR can be approximated.