Comparison Between Aerial-and Terrestrial Laser Scanning-based Forest Effective Leaf Area Indices

Leaf area index(LAI)is one of the most important forest canopy biophysical parameters and plays a key role in simulating forest biophysical processes and energy and matter exchanges between forest-soil-atmosphere.Quantitatively analyzing the three-dimensional(3-D)spatial distribution of foliage elements is a key step to retrieve forest leaf area index accurately.However,the traditional optical instruments-based methods usually estimate leaf area index from two-dimensional(2-D)perspective within a limited observation range.The time-consuming and labor-intensive nature of the field-based method limits their application.The availability of 3-D Light Detection and Ranging(LiDAR)technique makes it posible to retrieve forest leaf area index from 3-D perspective directly.The 3-D point cloud data(PCD)were acquired with terrestrial laser scanning(TLS)in different forest plots with various tree species,densities,ages,and spatial distribution patterns.An algorithm named "radial hemispherical point cloud slicing(RHPCS)"was used to process the original TLS data to retrieve "angular gap fraction(AGF)"and extinction coefficients.Moreover,the effective leaf area index(LAIe)values at forest plot level were also achieved based on the Beer's law.Compared with the results obtained using optical instruments,it was found that the combination of TLS and RHPCS could effectively estimate forest leaf area index.In additon,due to the obvious differences in data acquisition ways,point densites,and coverages between TLS and airborne laser scanning(ALS)systems,the AGF and LAIe were retrieved based on ALS and TLS data at the forest plot,respectively.The main conclusions of this study could be drawn as follows:(1)TLS data can be used to estimate forest LAIeThe RHPCS algorithm was used to process the original PCD to retrieve AGF and extinction coefficients through dividing the hemispherical volume into many"trapezoid voxels".By reconstructing the normal vector of a point within a certain neighbor region,I computed the leaf mean inclination angle,and further obtained the canopy extinction coefficient using the Campbell's algorithm.Finally,I calculated LAIe of each forest plot based on the Beer's law.My results showed that the TLS-based LAIe captured 70.84%(N=9,p<0.01)and 74.09%(N=14,p<0.01)variations of LAI-2200 and DHP-based ones variations,respectively.The result showed that it was feasible to calculate the LAIe of forest plots by combining the TLS-based PCD with the RHPCS algorithm.I conducted a sensitivity analysis of laser beam angle(LBA)on AGF estimation,which suggested that the AGF of the RHPCS should be determined based on user-predefined laser sampling space which affected by the characteristic size of foliage elements of a forest canopy.(2)TLS-and ALS-based AGF,LAIe has significantly differences in the same forestBased on the "aerial-terrestrial" lidar PCD of the same forest area,the same stereographic projection method was used to calculate the forest AGF and LAIe.Compared with the DHP-based results,I found that the statistical correlation calculated by using TLS-based PCd is higher than calculated by ALS-based ones.The reason for this error may be the difference in point density between ALS and TLS on the one hand,and on the other hand,depending on the orientation of the DHP taken in the same way as TLS,but it was different with ALS.A comparative analysis of forest LAIe obtained by the "aerial-terrestrial" lidar PCD showed that at the sample scale,the coverage of ALS data had a significant impact on the estimate forest AGF and LAIe.When coverage radius increased from 30 meters to 60 meters,the average reduction of AGF was 0.279,and that of LAIe was 0.47.When it increased from 60 meters to 90 meters,the average reduction of AGF was 0.077,and that of LAIe was 0.018.However,the impact on TLS data was smaller.The radius increased from 30 meters to 45 meters and then increased to 60 meters,the maximum reduction of AGF was 0.039,the minimum reduction was 0,and the average increase of LAIe was 0.051.The number of scanning overpass times affected the integrity of the forest ALS data.The ALS scan angle had a significant effect on the estimation of AGF.The average differences between ALS-,TLS-and DHP-based AGF of the plot with a scan angle of 7°-17° were 0.139 and 0.038.The mean differences of ALS-,TLS-and DHP-based AGF of the plot with a scan angle of 23°-29° were 0.527 and 0.522.The differences in vegetation densities had an effect on LAIe estimation based on TLS.When the coverage increased from 30 meters to 60 meters,the average increased value of LAIe in low-density plots was 0.066,that in medium-density plots was 0.041,and that in high-density plots was 0.026.The vegetation density had no effect on LAIe estimation based on ALS.The effects of different vegetation types on TLS-and ALS-based LAIe were negligible.With the increasing of forest coverage,the average increased TLS-based LAIe of broad-leaved forests,coniferous forests and mixed forests were 0.050,0.056 and 0.043,respectively,and the average increased ALS-based LAIe were 0.186,0.302 and 0.232,respectively.