A Method to Validate Airborne LIDAR CHM Producton Individual Tree Level

doi:10.13466/j.cnki.lyzygl.2022.01.014

Abstract

Abstract:

The accuracy and precision of Airborne LIDAR canopy height model (CHM) product was accessedon in dividual tree level,using forest ground measurements with high-precision position. The analysis procedures were based on dominant trees sampling by spatial analysis tool,iterative elimination of outliers,correlation analysis,and rules for error grading statistics.The method was applied in Northeast Tiger and Leopard National Park. The CHM product was collected by LIDAR RIEGL-VQ-1560i with 10 pulse/m². The results showed that: 1) CHM and "real" height had a significant linear correlation,R²=0.97;CHM products were slightly lower than the "real" height,about 0.7m;Roughly,there was an uncertainty of ±2m(2σ) for each tree;2) For each tree species (group),all CHMs were slightly lower than the "real" height too,ranging from 0.3m to 1.3m;Roughly,There was an uncertainty of ±1.4 ~±2.4m(2σ) for different species;Significantly,the CHM height of the larch was about 1.3m lower than the "real" value;3) the result of error grading statistics showed proportion of CHM with 0~ ±1m error was 57.6%; proportion of CHM with 0~±2m error was 79.4%;proportion of CHM with more than +2m error was 18.6%; proportion of CHM with less than -2m error was 2.0 %. It indicated that about 1/5 of the products had lower CHM values,which needed to be further explored. The results showed that the method could effectively identify the quality problems of CHM products,and provide a guarantee for the quality control of airborne LIDAR forest survey projects under complex operating conditions in large-scale and mountainous forest areas.

Key words: tree height, forestry resources monitoring, airborne LIDAR, canopy height model, CHM accuracy

CLC Number:

S771.8

FU Anmin, GAO Xianlian, WU Fayun, GAO Jinping. A Method to Validate Airborne LIDAR CHM Producton Individual Tree Level[J]. FOREST RESOURCES WANAGEMENT, 2022, (1): 114-123.

Figures/Tables 11

Fig.1

Tab.1

Tab.2

Fig.2

Tab.3

Fig.3

Fig.4

Fig.5

Tab.4

Tab.5

Fig.6

References 12

[1]	Erik Næsset, Gobakken T, Holmgren J, et al. Laser scanning of forest resources:The nordic experience[J]. Scandinavian Journal of Forest Research, 2004,18:482-499.
[2]	庞勇, 赵峰, 李增元, 等. 机载激光雷达平均树高提取研究[J]. 遥感学报, 2008(1):152-158.
[3]	Nelson R. How did we get here? An early history of forestry lidar[J]. Canadian Journal of Remote Sensing, 2014,39:6-17.
[4]	李增元, 陈尔学, 高志海, 等. 中国林业遥感技术与应用发展现状及建议[J]. 中国科学院院刊—“从空间看地球:遥感发展五十年”专辑, 2013,28(S0):132-144.
[5]	曾伟生, 孙乡楠, 王六如, 等. 基于机载激光雷达数据估计林分蓄积量及平均高和断面积[J]. 林业资源管理, 2020(2):79-86.
[6]	Pang Yong, Wang Weiwei, Du Liming, et al. Nystrym-based spectral clustering using airborne LiDAR point cloud data for individual tree segmentation[J]. International Journal of Digital Earth, 2021,14(10):1452-1476. doi: 10.1080/17538947.2021.1943018
[7]	李增元, 庞勇, 刘清旺. 激光雷达森林参数反演技术与方法[M]. 北京: 科学出版社, 2015.
[8]	Demetrios Gatziolis, Jeremy S Fried, Vicente S Monleon. Challenges to estimating tree height via LiDAR in closed-canopy forests:A Parable from Western Oregon[J]. Forest Science, 2010,56:139-155.
[9]	Kraus K, Pfeifer N. Determination of terrain models in wooded areas with airborne laser scanner data[J]. ISPRS Journal of Photogrammetry and Remote Sensing, 1998,53(4):193-203. doi: 10.1016/S0924-2716(98)00009-4
[10]	White J C, Wulder M A, Varhola Andrés, et al. A best practices guide for generating forest inventory attributes from airborne laser scanning data using an area-based approach[J]. The Forestry Chronicle, 2013,89(6):722-723. doi: 10.5558/tfc2013-132
[11]	国家林业局,吉林省人民政府,黑龙江省人民政府. 东北虎豹国家公园总体规划[EB/OL].(2018-09-20)[2021-11-10], http://www.forestry.gov.cn/uploadfile/main/2018-3/file/2018-3-9-599430e5ec1249bab08927453227ff14.pdf.
[12]	Mcgaughey R J, Ahmed K, Andersen H E, et al. Effect of occupation time on the horizontal accuracy of a mapping-grade GNSS receiver under dense forest canopy[J]. Photogrammetric Engineering & Remote Sensing, 2017,83:861-868.

指标	定位系统
基站定位精度	水平精度优于5cm,高程精度优于10cm
基站采样间隔	1s
移动站	移动站距离基站小于30km
移动站静态测量	静态30min以上
全站仪测距精度	精度优于0.5cm
指标	林木检尺
检尺范围	胸径大于5cm
胸径测量	胸径等于或大于20cm的树木,胸径测量误差小于5%;胸径小于20cm的树木,胸径测量误差小于0.5cm。
树高测量	树高10m以下误差小于1m;树高10m以上不超过10%
检尺株数	大于或等于8cm的应检尺株数不允许有误差。小于8cm的应检尺株数,允许误差3株。
冠幅测量	误差小于15%(东西冠幅、南北冠幅)
枝下高测量	误差小于20%
郁闭度测量	误差小于0.05

设备名称	设备类型	主要指标	参数指标
飞机	塞斯纳208B	飞行速度	260 km/h
		飞行高度	2300/2100 m
		平均地面高程	400/200 m
激光雷达载荷	RIEGL-VQ-1560i	脉冲频率	1000 kHz
		视场角FOV	58°
		激光发散度	≤0.25 mrad
		激光波长	1064 nm
		扫描频率	207 Hz
		扫描幅宽	2129 m
		点密度	10 pts/m²
		旁向重叠度	25%
IMU系统	POSAV610	采样频率	200Hz
		横滚/俯仰	0.005°
		航偏	0.008°
GNSS定位系统 (飞机/地面)	CORS站	采样频率	1Hz
	中海达GPS	采样频率	1Hz
	机载GPS	采样频率	1Hz

等级	名称	内容	数据量
0级	激光雷达原始数据	从激光雷达载荷(LIDAR)、惯性导航(IMU)系统和GNSS系统(飞机/地面)导出的原始数据	2.3TB
1A级	激光雷达初级定位数据	将机载GNSS和IMU数据,联合地面参照站数据进行差分技术处理,对航线定位定向进行联合精结算;将精解算姿态和定位数据赋给每个激光脉冲。	4.2TB
1B级	激光雷达去噪数据	剔除明显噪声点后的数据。	4.2TB
2级	激光点云几何精纠正数据	以架构航线为基准,经航带间匹配,整体平差后的数据。	2.7TB
3级	激光雷达分类数据	对激光点云数据进行滤波和分类,区分地面点和非地面点的数据。	2.7TB
4级	CSM/DTM/CHM数据	经点云数据空间差值或格网统计获取CSM和DTM数据,两者之差获取CHM数据。	160G

编码	树种(组)	主要树种	株数/株	最小值	最大值	平均值	标准差
1	天然针叶林Ⅰ	红松(包括赤柏松)、云杉(包括鱼鳞云杉、红皮云杉)	675	-1.5	3.0	0.7	1.0
2	天然针叶林Ⅱ	臭松、樟子松(包括樟子松、赤松、黑松、油松、长白松)、其它针叶	488	-1.6	3.0	0.6	0.9
3	天然针叶林Ⅲ	落叶松	953	-1.5	3.0	0.9	0.8
4	天然阔叶林Ⅰ	水曲柳、胡桃楸、黄菠萝	120	-1.5	3.0	0.8	1.1
5	天然阔叶林Ⅱ	椴树、枫桦	419	-1.6	3.0	0.7	1.0
6	天然阔叶林Ⅲ	柞树、黑桦	601	-1.6	2.9	0.3	1.0
7	天然阔叶林Ⅳ	色树、榆树	205	-1.6	3.0	0.6	1.0
8	天然阔叶林Ⅴ	杨树、白桦	1530	-1.6	3.0	0.7	0.9
9	天然落叶林Ⅵ	其它阔叶	286	-1.6	2.8	0.4	1.0
10	人工针叶林Ⅰ	人工红松(包括赤柏松)、人工云杉(包括鱼鳞云杉、红皮云杉)	152	-1.6	2.9	1.0	0.7
11	人工针叶林Ⅱ	人工樟子松(包括樟子松、赤松、黑松、油松、长白松)、人工臭松、其它针叶	N/A	N/A	N/A	N/A	N/A
12	人工针叶林Ⅲ	人工落叶松	410	-1.5	3.0	1.3	0.8
13	人工阔叶林Ⅰ	人工杨树(包括人工朝鲜柳)	N/A	N/A	N/A	N/A	N/A
14	人工阔叶林Ⅱ	人工其他阔叶林	17	-1.5	2.7	0.3	1.2
		总计	5856	-1.6	3	0.7	1.0

抽样方案	缓冲区大小 /m	样地半径 /m	株数参数	小计	-3σ 以下	-3σ	-2σ	±σ	+2σ	+3σ	+3σ 以上
(1)	5	10	N/A	714	17	6	32	389	126	75	69
(1)	5	10	N/A	100%	2.4%	0.8%	4.5%	54.5%	17.6%	10.5%	9.7%
(2)	4	10	N/A	1017	28	13	58	591	153	95	79
(2)	4	10	N/A	100%	2.8%	1.3%	5.7%	58.1%	15.0%	9.3%	7.8%
(3)	3	10	N/A	1601	84	27	112	915	235	124	104
(3)	3	10	N/A	100%	5.2%	1.7%	7.0%	57.2%	14.7%	7.7%	6.5%
(4)	2	10	N/A	3366	454	115	286	1830	377	176	128
(4)	2	10	N/A	100%	13.5%	3.4%	8.5%	54.4%	11.2%	5.2%	3.8%
(5)	2	15	N/A	7873	1302	351	708	4108	787	340	277
(5)	2	15	N/A	100%	16.5%	4.5%	9.0%	52.2%	10.0%	4.3%	3.5%
(6)	2	15	5株	2195	16	29	117	1264	362	206	201
(6)	2	15	5株	100%	0.7%	1.3%	5.3%	57.6%	16.5%	9.4%	9.2%