Remote Sensing Inversion Of Leaf Area Index Of Broad-leaved Forest

Leaf area index is an important characteristic parameter of the vertical structure and the complexity of the vegetation,meanwhile the precise estimation of forest canopy leaf area index is of great significance in monitoring forest stand structure and quality.With the strong labor intensity,the applicable scope of traditional method of estimation leaf area index is limited,while terrestrial LiDAR technology could achieve three-dimensional point cloud data and the 3D structure data of the vegetation and forest stand quickly.This research adopted the Laser scanning collecting point cloud data of broad-leaved forest and single tree.The point cloud data and RGB picture as the basic data,this research studied the accuracy of inversion LAI of broad-leaved forest and single tree which used the methods of layering processing,voxelization,correcting coefficient for leaf inclination,supervised classification and colouring point cloud.Then we put forward a rapid,accurate and efficient method for broad-leaved forest leaf area index.The main contents were as follows:(1)According to the basic knowledge of 3D laser scanning system and the characteristics of the instrument,we used Stonex X300 terrestrial Li DAR Instrument to collect multiple station point cloud data of Broad-leaved forest and single tree,and acquired the experimental data after point cloud de-noising and integration.Meanwhile this paper used LAI-2200 Isolated canopy analyzing to obtain the average leaf Angle,and leaf area index structure parameter of single tree and broad-leaf forest in the area.To verify the accuracy of the inversion model.(2)Using 24 trees of a single experiment broad-leaved forest sample as the research object,we obtained the different canopy layers'laser point cloud data with 0.5m as horizontal segmentation scale,which were processed by voxelization,vertical projection and correcting coefficient for leaf inclination.We got six groups leaf area index of the 24 sample trees at the scale of 1,1.1,1.2,1.3,1.4,1.5 factor.The results showed that the scale factor could influence the voxel,then impact estimating LAI,and Change trend was consistent.Under the scale factor of 1.2,the accuracy of leaf area index inversion was the best and the average estimate accuracy was 91.12%.The highest precision was 15 sample tree,of which was 98.94%.The inversion index and the measured index had higher correlation that determination coefficient R2 was 0.9337.(3)The paper used true color photos which obtained by the terrestrial laser radar.Supervised classification and point cloud mapping were done by ERDAS 9.2 software and Stonex Reconstructor.Then we extracted the leaf laser point cloud data that could apply to inverse broadleaved forest leaf area index and compare with the average of broadleaved forest individual tree.The results showed that the average of the broad-leaved forest LAI of 24 sample trees which measured directly was 93.82%,which was 3.37%lower than the estimated average of the LAI of 24 sample trees measured by the foundation of laser radar.The result could greatly improve the efficiency of the foundation of laser scanner.(4)Coupling terrestrial LiDAR and the use of point cloud voxel method,we estimated LAI of six broad-leaved forest sample plots.The results showed that LAI inversion accuracy is the highest in the plot when the scale factor was between 1.3?1.4,while the inversion precision of six sample plots were 99.1%?96.53%?96.67%?95.0%?87.7%?97.62%,respectively.When the scale range was between1.3?1.4,most of the LAI estimation results were smaller than the directional measurement results,which was the opposite to the inversion LAI of single tree with the point cloud voxel.In the study area,the LAI index of broad-leaved forest ranged from 1.439 to 1.786.