Retrieving Canopy Clumping Index Of Broad-Leaved Forest Based On Airborne And Terrestrial Laser Scanning

The canopy Clumping Index(CI)describes the process of vegetation interception of light and canopy radiative transfer and is a very important canopy structure characteristic parameter,which is important for reflecting the forest canopy distribution and the accurate inversion of the leaf area index.Li DAR technology can quickly and accurately obtain 3D spatial information of the detection target,and thus more efficiently retrieve the structural information of the forest canopy,and has been widely used for canopy gap fraction and clumping index retrieval.However,the current methods for GF and CI inversion based on Li DAR data often suffer from the inability to make full use of 3D point cloud information,the large computational effort due to excessive redundant data,and the scope and accuracy limitations of relying on Beer’s law for parameter inversion.In this thesis,a method to invert the gap rate based on ground-based Li DAR data by building a Laser Point Area(LPA)model and a method to invert the CI based on Poisson distribution and point cloud clustering using airborne and ground-based data are proposed respectively to address the shortcomings of existing methods:(1)A method for estimating canopy GF in broadleaf forests based on an LPA model is proposed,combined with LX’s method to achieve CI inversion for ground-based Li DAR.A method for estimating the gap fraction of broadleaf forest canopy based on the laser point area model is proposed to address the problems of insufficient use of three-dimensional information in the current method of calculating the gap fraction based on Li DAR data.The method uses the laser point area reconstruction process to express the vertical information of the 3D point cloud in the calculation results after hemispheric projection,which overcomes to a certain extent the information loss caused by the insufficient use of the vertical information of the point cloud in the parameter extraction of the traditional method;and introduces the sample scale parameter and density adjustment parameter to correct the distortion generated in the projection process to a certain extent,which ensures the accurate description of the forest canopy structure by the laser point area model can accurately describe the forest canopy structure.By comparing the results with those of the clumping index based on digital hemispheric photographs,it was found that the gap fraction(R~2=0.8551)and clumping index(R~2=0.5087)of the two methods were highly correlated,which proved the effectiveness of the algorithm.(2)A Poisson distribution and point cloud clustering-based CI inversion method was developed to achieve vertical stratification and inversion of the CI in broadleaf forest canopies.The main method currently used to calculate the leaf clumping index is to calculate the canopy gap fracion followed by the log-mean method or the gap size distribution method.However,such methods have the drawbacks of requiring assumptions on canopy spatial distribution,relying on specific measurement instruments,and the results being heavily influenced by irregular large gaps,which have a significant impact on the accurate retrieval of the clumping index.The thesis proposes a clumping index inversion method based on Poisson distribution and K-means clustering,which avoids the increased computational effort and errors caused by intermediate variables,while combining statistical methods and unsupervised clustering to enhance the universality of the method and expand its applicability.By comparing the inversion of the CI based on the laser point area model and the finite length averaging method proposed in this thesis,the correlation between the two is high(R~2=0.9716)and the influence of the selection of the threshold on the method is discussed,and the validity of the method proposed in this thesis for estimating the CI was analysed.The inversion method of the CI proposed in this thesis makes full use of the three-dimensional information of the point clouds to ensure the validity,and the inversion of the CI of pure forests in small sample plots and mixed forests in large sample plots was carried out by the method based on gap fraction and Poisson assumption method(without gap fraction)respectively.The vertical distribution retrieval of the CI at the sample site scale was achieved based on multi-platform data,which laid the foundation for future high-precision CI retrieval models at a larger scale.