The Simulation And Application Of Terrestrial Lidar System

The plant area index(PAI)of forest canopy was widely used in the fields of forestry,remote sensing,and agriculture.However,accurate estimation of the PAI of forest canopies is still a challenge for traditional optical methods.LiDAR(Light Detection And Ranging)is a new method and shows some merits as compared to traditional optical methods,such as the non-contact measurements of forest canopies,highly precision,not easily to be interfered by the weather conditions,and the ability of penetrating the 3D structures of forest canopies.Therefore,understanding the interaction between LiDAR and the 3-D structure of forest canopy can help us to improve the ability of using LiDAR to retrieve the PAI of forest canopies.In this paper,the terrestrial LiDAR was simulated by using the Monte Carlo ray tracing algorithm based on a typical forest scene library,and the estimation precision of PAI of LiDAR was comprehensively analyzed in the end.The main contents of this thesis are as follows:(1)The construction of a typical virtual forest scene library.With the goal of testing the discrete terrestrial LiDAR instrument,we proposed a new simulation method of clustered distribution tree pattern,in which the 3D spatial distribution of canopy components and the clustering effect in the community of forest ecology were implemented.A typical virtual forest scene library was constructed with different LAI,PAI,tree density,and tree distribution patterns using 14 3D detailed structural tree models.(2)The simulation of discrete terrestrial LiDAR and the analysis of the simulation precision.The Monte Carlo ray tracing algorithm was chosen as the method to simulate the discrete terrestrial LiDAR,and the main source of errors for the simulation were analyzed subsequently,such as the field of view,the threshold of multiple scattering order,and the sampling number of rays in each laser beam.Results showed that the larger field of view of the LiDAR,the greater the contribution of multiple scattering to the total echo signal;the simulation results would be stable if the number of multiple scattering order was set to 7;and the virtual forest scene would be stably sampled if the laser beam was accurately sampled whose sampling number was set to 500.(3)The study on the acceleration schemes of the simulation efficiency of discrete terrestrial LiDAR.Three types of data structure were used to organize and manage the data sets of forest scenes at two scales of single tree models and scenes.Totally 9 schemes of data structures were compared in this thesis,we found that the simulation efficiency would be better than others if the data structure of the single tree models and forest scenes are KD-Tree and BVH separately.Several factors were chosen to evaluate the efficiency of the 9 schemes of data structures during the comparison,such as traversing number of scene nodes,traversing number of single tree model nodes,traversing number of tree instance,traversing number of triangles,and the average ray tracing time needed for each laser beam.(4)The estimation of PAI of forest canopies using discrete terrestrial LiDAR.The gap fraction estimation results from the discrete terrestrial LiDAR method are highly correlated with these from the DHP method,and the correlation coefficient between them is 0.77.Whilst the average gap fraction from the discrete terrestrial LiDAR was lower about 7.67%as compared to estimates from the DHP instrument.Therefore,the PAI estimates derived from the discrete terrestrial LiDAR are generally larger than those from the DHP method.Results showed that the discrete terrestrial LiDAR method was usually overestimated the PAI estimates if its estimates were validated with the true values of the PAI of virtual forest scenes.The estimation precision can be improved if we increase the sampling resolution of the LiDAR,such as the angular resolution.