运用机载LiDAR数据对橡胶林地上生物量估测的采样尺度效应分析  

Analysis of Sampling Size Effect of Aboveground Biomass Estimation of Rubber Forest Using Airborne LiDAR Data

作　　者：罗洪斌 舒清态 庞勇[2] 王强 王冬玲 LUO Hongbin;SHU Qingtai;PANG Yong;WANG Qiang;WANG Dongling(College of Forestry,Southwest Forestry University,Kunming,650224;Resource Information Institute,Chinese Academy of Forestry,Beijing 100091)

机构地区：[1]西南林业大学林学院,昆明650224 [2]中国林业科学研究院资源信息研究所,北京100091

出　　处：《林业资源管理》2020年第1期136-142,共7页Forest Resources Management

基　　金：国家自然科学基金项目(31860205,31460194);云南“唐守正”院士工作站项目(2018IC066).

摘　　要：气候变化下,森林生物量遥感监测是当前研究的热点,机载LiDAR作为重要的遥感信息源,其采样大小对生物量估测精度有着一定的影响。以机载LiDAR数据为信息源,以44块30m×30m的方形橡胶林实测样地数据为基础,对机载激光雷达数据进行不同尺寸采样(共21个采样尺寸,边长从10m至30m,间隔为1m),提取不同采样尺寸下的激光雷达参数,并与橡胶林地上生物量建立PLSR模型,就机载激光雷达采样大小对橡胶林地上生物量估测精度的影响进行研究。研究表明:当采样尺寸小于18m时,估测精度随着采样尺寸的增大而增大;而当采样尺寸大于18m时,估测精度随着采样尺寸的增大而减小,进而趋于平缓。结果虽然呈现出一定的规律性,但是差异并不是很明显。当采样尺寸为18m时估测效果最佳,模型决定系数(R^2)为0.718,均方根误差(RMSE)为17.830 t/hm^2;交叉验证精度P和RMSEcv分别为82.741%和18.874t/hm^2。相较于实际样地(30m)尺寸下的估测结果,18m采样尺寸下的R^2提高了1.989%,RMSEcv降低了2.611%。因此,生物量的估测精度受机载激光雷达数据采样尺寸大小的影响,在生物量估测过程中需结合研究对象和研究区的实际情况对采样尺寸进行选择,从而提高生物量估测精度。Remote sensing monitoring of forest biomass under climate change is a hot topic of current research.As an important remote sensing information source,airborne LiDAR′s sampling size has a certain influence on the estimation accuracy of biomass.In this paper,the airborne LiDAR data is sampled in different sizes(21 sampling sizes,side length from 10 m to 30 m,interval is 1 m),the LiDAR parameters of different sampling sizes are extracted,and the PLSR model is established with the biomass on the rubber forest.The effect of airborne LiDAR sampling size on the estimation accuracy of biomass on rubber forest was studied.The results show that the estimation accuracy of the aboveground biomass of rubber forest is affected by the sampling size of airborne LiDAR data.The results showed a certain regularity,but the difference was not significant.When the sampling size is less than 18 m,the estimation accuracy increases with the increase of sampling size,while the sampling size is larger than 18 m.The estimation accuracy decreases as the sampling size increases,and thus tends to be gentle.When the sampling size is 18 m,the estimation result is the best.The model determination coefficient R^2 is 0.718,the root mean square error RMSE is 17.830 t/hm^2,and the cross-validation accuracy P and RMSEcv are 82.741% and 18.874 t/hm^2.Compared to the estimated results at the actual sample size(30 m), R^2 increased by 1.989% and RMSE decreased by 2.611% at the 18 m sample size.Thus,the biomass estimation process for an actual case study and research of the sampling area size is selected,thereby increasing the biomass estimation accuracy.

关 键 词：采样 LIDAR 生物量 橡胶林 

分 类 号：S771.8[农业科学—森林工程]