文献类型：期刊文献

中文题名：基于机载激光雷达冠层高度模型的小班区划

英文题名：Forest Sub-compartment Delineation Based on Airborne LiDAR Canopy Height Model

作者：熊昊[1] 庞勇[1] 荚文[1] 李春干[2]

第一作者：熊昊

机构：[1]中国林业科学研究院资源信息研究所,国家林业和草原局林业遥感与信息技术重点实验室,北京100091;[2]广西大学林学院,广西南宁530004

年份：2022

卷号：35

期号：2

起止页码：28-36

中文期刊名：林业科学研究

外文期刊名：Forest Research

收录：CSTPCD;;Scopus;北大核心:【北大核心2020】；CSCD:【CSCD2021_2022】；

基金：“十三五”国家重点研发计划“落叶松高效培育技术研究”项目(编号:2017YFD0600404)。

语种：中文

中文关键词：机载激光雷达;CHM;树种类别;小班区划;多尺度分割

外文关键词：airborne laser scanner;canopy height model;tree species;forest sub-compartment delineation;multi-resolution segmentation

分类号：S757.4

摘要：[目的]探究如何有效利用机载激光雷达冠层高度模型(CHM)自动区划小班,提高小班区划工作效率。[方法]在高光谱影像树种信息的辅助下,使用机载激光雷达数据生成的CHM进行两种空间尺度的分割和优化来自动区划小班。先对1 m空间分辨率CHM数据进行过分割,再对降尺度处理并平滑后的5 m空间分辨率CHM数据进行欠分割,结合两种尺度分割结果并优化得到最终区划结果。将自动区划结果与人工区划小班、数字正射影像(DOM)屏幕勾绘小班以及主伐作业小班为三类参考小班对比,采用最终测量精度(UMA)准则的圆度(RO),紧致度(CO),形状指数(SI),最小包络圆短半径(RE),椭圆度(EF)和形状因子(P2A)8个指标,及自动区划小班与参考小班的交并比(IOU)指标,定量评价自动区划小班边界勾绘的准确程度。并利用样地实测数据和CHM数据计算自动区划结果平均胸径、平均树高和冠层平均高的可解释性方差,验证自动区划结果的内部一致性和外部差异性精度。[结果]自动区划结果与参考小班的UMA形状、面积等特征较接近,与人工区划小班最相近。自动区划小班与人工区划、屏幕勾绘、主伐作业小班交并比大于70%的比例分别为46%,37%,43%,交并比大于50%的比例分别为61%,54%,55%。自动区划结果平均胸径可解释性方差为97%,平均树高可解释性方差为98%,和人工区划小班相同,说明其内部一致性高且和相邻小班差异大。冠层平均高可解释性方差为84.81%,比人工区划小班提高了1.77%。[结论]利用两种空间尺度的CHM与高光谱树种分类图的分割和优化方法自动区划的小班在内部一致性及边界的精准度方面有明显优势,更符合小班边界处林木的分布情况,小班边界准确,且工作效率高,有助于森林的精细化管理。
[Objective]Based canopy height model(CHM)to automatically delineate the forest sub-compartments of forest resources management inventory similar to the manual delineation.[Method]Supported by tree species derived from hyperspectral image,the CHM generated from airborne LiDAR data was used for multi-resolution segmentation and optimization.First,the segmentation was applied on 1 m resolution CHM to obtain the over-segmentation results.Then 1 m resolution CHM was down sampled to 5 m resolution and segmented to get the under-segmentation results.By combining and optimizing these two results,the final sub-compartment delineation was obtained.The manual sub-compartments,the sub-compartments delineated based on 0.1 m spatial resolution Digital Orthophoto Map(DOM)and the Logging sub-compartments were used as reference data.The Ultimate Measurement Accuracy(UMA)rule was used to validate the accuracy of boundary drawing of stand segmentation results in its consistency with the reference sub-compartments.The UMA has 8 indexes,which are Roundness,Compactness,Shape index,Radius of smallest enclosing ellipse,Elliptic fit,P2A etc.And the Intersection Over Union(IOU)ratio was also introduced to quantify the consistency between the automatic segments and reference sub-compartments.The explained variance of mean height and mean DBH was calculated using the field data to validate the internal consistency and external variability accuracy of automatic segments.[Result]The automatic segments were similar to that of the reference sub-compartments in shape,area and other characteristics of UMA,and were most similar to the manual sub-compartments.The proportions of automatic segments whose IOU ratio was more than 70%with manual sub-compartments,the DOM sub-compartments and the Logging sub-compartments were 46%,37%,and 43%,respectively.The proportions of automatic segments whose IOU ratio more than 50%were 61%,54%,and 55%.The explained variances of mean DBH and mean height of automatically delineated sub-compartments were 97%and 98%,indicating the high internal consistency and distinct differences with adjacent sub-compartments of automatic sub-compartments.The explained variances of mean canopy height was 84.81%,which was 1.77%higher than that of the manual sub-compartments.[Conclusion]The sub-compartments automatically delineated by multi-scale segmentation method with CHM and tree species derived from hyperspectral image have obvious advantages in terms of internal consistency and boundary accuracy,and are more consistent with the distribution of trees at sub-compartments boundaries.This method saves time,increases efficiency,and increases the accuracy of forest sub-compartment delineation,which can support forest delicacy management planning.