不同森林覆盖类型对LiDAR生成DEM的精度影响分析

Analysis of influence of different forest types on accuracy of DEM generated by LiDAR

运用机载激光雷达(LiDAR)获取的高精度数字高程模型(DEM)是目前地学分析应用满足精细化要求的基础数据。目前对森林覆盖下DEM产品的精度影响分析依然有所欠缺。以实时动态载波相位差分技术(RTK)结合全站仪实测的高程碎部点与机载LiDAR数据生成的高精度DEM产品数值做差,进行单因素方差分析(ANOVA)检验以及独立样本T-检验。经多重比较分析(LSD)可知,以马尾松为代表的针叶林和以桉树、茶树、黄连木、荔枝、阔叶混合林为代表的阔叶林之间差值相差较大,说明针叶林和阔叶林两种不同森林覆盖类型的绝对精度存在差异。但Robust检验结果显示阔叶林地表DEM的精度均值相等性为1,说明阔叶林(以桉树、茶树、黄连木、荔枝、阔叶混合林林地为代表)地表DEM精度不受树种的影响。综上,阔叶林与针叶林的地表DEM精度差异显著,说明阔叶林和针叶林对机载LiDAR生成森林区域地表DEM存在一定的影响。

The high-precision digital elevation model(DEM)generated from airborne light detection and ranging(LiDAR)data is the basic data for current geomorphological analysis applications to meet the refinement requirements.At present,the analysis of the influence of the accuracy of DEM products under forest cover is still lacking.A one-way analysis of variance(ANOVA)test and an independent sample T-test are performed on the difference between the measured elevation fragments from real time kinematic(RTK)combined with the total station and the high-precision DEM products generated from airborne LiDAR data.The least significant difference(LSD)multiple comparison analysis shows that the difference between the coniferous forests represented by the Sargasso Pine and the broadleaf forests represented by the eucalyptus,tea tree,pelargonium,Lychee,and broadleaf mixed forests is large.This indicates that there is a difference in the absolute accuracy of the coniferous forests and broadleaf forests.However,the Robust test results show that the mean equivalence(df1)of the accuracy of the surface DEM of broadleaf forests is 1,indicating that the accuracy of the surface DEM of broadleaf forests(represented by eucalyptus,tea tree,yellow woodland,lychee,and mixed broadleaf forests)is not affected by tree species.In summary,the significant difference in surface DEM accuracy between broadleaf and coniferous forests indicates that broadleaf and coniferous forests have some influence on airborne LiDAR generation of surface DEMs of forested areas.