Extraction Of Forest Parameters From Terrestrial Laser Point Cloud Of Larch Plantation

Extraction of forest parameters was an important research content of forest resources monitoring.Terrestrial laser scanning technology made up for the shortage of traditional manual sampling survey.However,the accuracy and applicability of extracting DBH and tree height from point cloud data obtained by ground-based lidar under different plantation conditions were still unclear.At the same time,the accuracy of extracting tree height from point cloud data was greatly affected by point cloud normalization.Therefore,the accurate extraction of forest parameters under different stand conditions needs systematic research.Taking Larch(Larix principis-rupprechtii Mayr)plantation in Yangjuangou Forest Farm of Wuzhai County of Shanxi Province as the research object,the data collection and extraction of DBH and height were carried out by using ground laser scanner.The main conclusions are as follows:(1)The extraction algorithm of point cloud data processing is proposed and modularized.This paper studies the standardized cloud data collection process of plots,combined with RTK measurement of the absolute coordinates of plot corners.Through the research and derivation of the principle of coordinate transformation,it writes the automatic cloud extraction program of plots,and avoids the human error from man-made operational mistakes.(2)Based on 3D laser scanning technology,the DBH and tree height of larch plantation are accurately extracted.Through the data processing and extraction analysis,the parameters of DBH and height of larch plantation are obtained.The highest R2(coefficient of determination)of DBH is 0.93,RMSE(root mean square error)is 0.98cm,RRMSE(relative root mean square error)is 5.52%,the highest R2 of tree height is 0.74,RMSE is 1.28m,RRMSE is 8.47%.(3)An improved method of extracting tree height from point cloud data is proposed,which has significant improvement compared with the traditional method of extracting tree height from point cloud data.In view of the problem that the traditional method based on normalized point cloud parameter extraction results in the stretch or contraction distortion of single tree point cloud data under certain slope terrain conditions,which affects the accuracy of tree height parameter extraction,this paper proposes an improved method of tree height parameter extraction,which significantly improves the accuracy of tree height extraction,and the maximum R2 increases by 0.18,The RMSE and RRMSE decreased by 0.3m and 5.8 percentage points respectively.(4)Stand density has a significant impact on the accuracy of tree parameters extraction from 3D laser point cloud.According to the accuracy evaluation results of DBH and tree height extracted in this paper,it is confirmed that the lower the stand density,the simpler the stand,the higher the extraction accuracy of forest parameters.In the design of 8 plots,the accuracy distribution does shows a stepped distribution with the change of stand density.