Stand diameter distribution is one of the most important indicators reflecting stand structure, aiding in estimating forest biomass, timber yield, and evaluating stand stability. Unmanned aerial vehicle (UAV)-LiDAR technology provides high-precision three-dimensional spatial information, offering a new technical approach for large-scale and continuous predictions of stand diameter distributions. This study utilized UAV-LiDAR derived features and measured diameter distributions as data sources to develop Weibull diameter distribution models using three parameter prediction methods: the parameter prediction method, percentile-based parameter recovery method, and moment-based estimation parameter recovery method. The model’s performance was evaluated through multiple metrics including, R2, RMSE, MAE, EIP, and EIR, ultimately identifying the optimal Weibull model for Korean pine planatation diameter distributions. The precision of Weibull parameter predictions based on UAV-LiDAR data showed an R2 ranging from 0.51 to 0.71. The diameter distribution models established by the three parameter prediction methods can predict the diameter distribution of the stand, and the parameter prediction method performed best (EIR=57.384, EIP=0.319). UAV-LiDAR technology effectively addresses the limitations of conventional surveys, such as low data acquisition efficiency and restricted spatial coverage, providing a robust technical foundation for reliable predictions of stand diameter distributions.
林分直径分布是指林分内各径阶树木的胸径分布密度特征,其反映了林分中不同大小树木的数量分配情况,是表征人工林林分结构的重要属性之一[1-8]。同时,林分结构的动态变化能够有效反映林木在生长过程中受到的环境因子(如气候、土壤等)和经营措施(如间伐、抚育等)的综合影响。这种动态变化不仅体现了林木个体的生长响应,也反映了林分整体的发育过程,为森林经营决策提供了重要的科学依据[9-10]。红松作为东北林区乡土优势树种之一,广泛分布于长白山脉以及小兴安岭等区域,通过建立该树种的直径分布模型可以揭示红松人工林直径结构的发展规律,为开展科学的经营管理提供依据[11]。在传统人工林经营中,森林资源的动态监测过程主要通过长期固定样地调查,虽然该方法可直接获取相对准确的林分调查因子,但往往耗时费力,固定样地调查数据中往往存在同木异号、错测漏测等[12]问题。而无人机激光雷达技术[13]具有穿透植被冠层直接提供高精度三维信息的能力,在评估林分结构中展示出了巨大的优势。基于激光雷达的林分结构研究中,关于直径分布模型的构建上主要采用2种方法[14]:单木检测法(individual tree delineation,ITD)和基于面方法(area-based approach,ABA)。对于ITD方法而言,建立直径分布模型过程是通过单木分割算法进行单木探测,利用所提取的单木参数构建直径分布模型。在复杂林分中单木分割算法的探测精度受限,主要问题在于单木信息提取错误,从而影响ITD方法的估测精度[15]。相比之下,ABA方法是通过对与样地大小一致的每个栅格单元生成一系列林分特征因子来建立直径分布模型。ABA方法在直径分布模型中的预测精度通常高于ITD方法[16]。因此,ITD方法通常作为ABA方法的辅助手段用于提升精度,弥补其在冠层下树木探测精度的不足[17]。
ABA方法通常采用概率密度函数(probability density function,PDF)构建分布模型,该方法先提取样地尺度的激光雷达特征参数来预测林分株数密度及其PDF参数,最后结合实测数据建立PDF模型。近年来,许多学者针对不同树种尝试了多种概率密度函数来构建直径分布模型,其中包括Weibull分布、Johnson′s sb分布和Beta等[18-21]建立相应的直径分布模型。萨其拉[22]研究表明,使用2参数的Weibull函数建立的红松人工林直径分布模型相较于其他概率密度函数具有更高的预测精度。目前,Weibull参数的估计方法,主要分为参数预测法(parameter prediction method,PPM)和参数回收法(parameter recovery method,PRM)[23]。其中PPM通常采用极大似然估计法(maximum likelihood estimate,MLE)估计直径分布的参数,随后基于林分信息采用最小二乘法(ordinary least squares,OLS)、最大似然估计回归(maximum likelihood estimation based regression,MLER)等预测模型的系数[24]。PRM则使用一种间接的方法,首先估计与分布相关的其他参数,然后使用这些参数来预测Weibull参数。PRM可进一步分为基于百分位数的参数回收法(percentile-based parameter recovery method,PPRM)和矩估计参数回收法(moment-based parameter recovery method,MPRM)[25]。PPRM在以往的研究中通常使用直径的24百分位数和93百分位数建立直径分布模型[24,26];MPRM主要通过林分平均胸径和算术平均胸径的转换关系预测Weibull参数[27]。例如,Cao等[26]利用机载激光雷达数据,比较了百分位参数回收法、矩估计参数回收法和参数预测法所构建的亚热带森林的材积分布预测效果,其中PPRM(EIR=37.75)的效果最好。Gobakken等[28]基于机载激光雷达数据分别构建PPRM和PPM的林分直径和断面积分布模型,结果表明PPRM(偏差:-4.8%~2.7%)比PPM方法(偏差:-4.7%~6.6%)更适用于估测林分直径和断面积的分布。Thomas等[29]在加拿大安大略省中部北方森林中基于PPM方法建立了单峰双参数Weibull模型,该方法在特定林型中具有较高精度(=0.65~0.88)。在构建Weibull直径分布模型时,通常采用全子集法对激光雷达特征变量进行筛选,以决定系数和方差膨胀因子等指标来确定最佳特征变量的组合,基于上述特征因子来预测Weibull分布参数和林分的株数密度[22,30]。
首先,对原始点云数据进行航带拼接,随后进行点云去噪,然后利用改进渐进加密三角网滤波算法[31](improved progressive TIN densification,IPTD)进行点云滤波,将点云分为地面点和非地面点,利用地面点进行高度归一化处理,消除地形起伏对植被高度的影响。利用实时动态载波相位差分技术(real time kinematic,RTK)获取的样地坐标从而确定样地边界,并提取每个样地的样地尺度激光雷达特征变量,共计101个,见表2。
GOBAKKENT, NÆSSETE.Estimation of diameter and basal area distributions in coniferous forest by means of airborne laser scanner data[J].Scandinavian Journal of Forest Research,2004,19(6):529-542.
[2]
VALBUENAR, EERIKAINENK, PACKALENP,et al.Gini coefficient predictions from airborne lidar remote sensing display the effect of management intensity on forest structure[J].Ecological Indicators,2016,60(1):574-585.
[3]
MAGNUSSENS, NÆSSETE, GOBAKKENT.Prediction of tree-size distributions and inventory variables from cumulants of canopy height distributions[J].Forestry,2013,86(5):583-595.
[4]
SAARINENN, KANKAREV, VASTARANTAM,et al.Feasibility of Terrestrial laser scanning for collecting stem volume information from single trees[J].ISPRS Journal of Photogrammetry & Remote Sensing,2017,123(6):140-158.
[5]
BOLLANDSASO M, NÆSSETE.Estimating percentile-based diameter distributions in uneven-sized Norway spruce stands using airborne laser scanner data[J].Scandinavian Journal of Forest Research,2007,22(1):33-47.
[6]
MALTAMOM, SUVANTOA, PACKALÉNP.Comparison of basal area and stem frequency diameter distribution modelling using airborne laser scanner data and calibration estimation[J].Forest Ecology and Management,2007,247(1/3):26-34.
[7]
BERGSENGE, ØRKAH O, NÆSSETE,et al.Assessing forest inventory information obtained from different inventory approaches and remote sensing data sources[J].Annals of Forest Science,2015,72(1):33-45.
[8]
COOMESD A, ALLENR B.Mortality and tree-size distributions in natural mixed-age forests[J].Journal of Ecology,2010,95(1):27-40.
[9]
LÄMÅST, SÄNGSTUVALLL, ÖHMANK,et al.The multi-faceted Swedish Heureka forest decision support system:Context,functionality,design,and 10 years experiences of its use[J].Frontiers in forests and global change,2023,6(18):1-19.
[10]
BAŞKENTE Z, KELEŞS.Developing alternative wood harvesting strategies with linear programming in preparing forest management plans[J].Turkish Journal of Agriculture and Forestry,2006,30(1):67-79.
[11]
JINX J, PUKKALAT, LIF R,et al.Developing growth models for tree plantations using inadequate data – a case for Korean pine in Northeast China[J].Silva Fennica,2019,53(4):10217-10234.
[12]
SAQ L, JINX J, PUKKALAT,et al.Developing Weibull-based diameter distributions for the major coniferous species in Heilongjiang Province,China[J].Journal of Forestry Research,2023,34(6):1803-1815.
PACKALENP, PUKKALAT, PASCUALA.Combining spatial and economic criteria in tree-level harvest planning[J].Forest Ecosystems,2020,7(1):1-13.
[15]
HAOY S, WIDAGDOF R A, LIUX,et al.A hierarchical region-merging algorithm for 3D segmentation of individual trees using UAV-LiDAR point clouds[J].IEEE Transactions on Geoscience and Remote Sensing,2022,60:1-16.
[16]
RÄTYJ, PACKALENP, KOTIVUORIE,et al.Fusing diameter distributions predicted by an area-based approach and individual-tree detection in coniferous-dominated forests[J].Canadian Journal of Forest Research,2020,50(2):113-125.
[17]
SHINJ, TEMESGENH.Generating tree-lists by fusing individual tree detection and nearest neighbor imputation using airborne LiDAR data[J].Open Journal of Forestry,2018,8(4):500-531.
[18]
GORGOSO-VARELAJ J, OGANAF N, IGEP O.A comparison between derivative and numerical optimization methods used for diameter distribution estimation[J].Scandinavian Journal of Forest Research,2020,35(3/4):156-164.
[19]
GORGOSO-VARELAJ J, PONCER A, RODRÍGUEZ-PUERTAF.Modeling diameter distributions with six probability density functions in Pinus halepensis mill.plantations using low-density airborne laser scanning data in Aragón (Northeast Spain)[J].Remote Sensing,2021,13(12):2307-2324.
YET, ZANGY.Prediction model of diameter distribution of Cupressus funebris plantations in central Sichuan based on climatic factors[J].Journal of Northwest Forestry University,2024,39(3):59-66.
XIONGZ Y, ZHOUC G.Effects of goal tree thinning on diameter distribution of natural secondary forest in Laoshan forest farm[J].Territory & Natural Resources Study,2024(1):67-71.
[24]
萨其拉.黑龙江省典型人工针叶林直径分布模型[D].哈尔滨:东北林业大学,2023.
[25]
SAQ L.Diameter distribution models for the typical coniferous plantations in Heilongjiang Province[D].Harbin:Northeast Forestry University,2023.
[26]
LEIY.Evaluation of three methods for estimating the Weibull distribution parameters of Chinese pine (Pinus tabulaeformis)[J].Journal of Forest Science,2008,54(12):566-571.
[27]
CAOQ V.Predicting parameters of a Weibull function for modeling diameter distribution[J].Forest Science,2004,50(5):682-685.
[28]
LIUC, ZHANGS Y, LEIY,et al.Evaluation of three methods for predicting diameter distributions of black spruce (Picea mariana) plantations in central Canada[J].Canadian Journal of Forest Research,2004,34(12):2424-2432.
[29]
CAOL, ZHANGZ N, YUNT,et al.Estimating tree volume distributions in subtropical forests using airborne LiDAR data[J].Remote Sensing,2019,11(1):97-124.
[30]
SIIPILEHTOJ.Modelling stand structure in young Scots pine dominated stands[J].Forest Ecology and Management,2009,257(1):223-232.
[31]
GOBAKKENT, NÆSSETE.Weibull and percentile models for lidar-based estimation of basal area distribution[J].Scandinavian Journal of Forest Research,2005,20(6):490-502.
[32]
THOMASV, OLIVERR D,LIM K,et al.LiDAR and Weibull modeling of diameter and basal area[J].The Forestry Chronicle,2008,84(6):866-875.
WANGM,IM J, ZHAOY H,et al.Multi-platform LiDAR for non-destructive individual aboveground biomass estimation for Changbai larch (Larix olgensis Henry) using a hierarchical bayesian approach[J].Remote Sensing,2022,14(17):4361.
[36]
PACKALÉNP, MALTAMOM.Estimation of species-specific diameter distributions using airborne laser scanning and aerial photographs[J].Canadian Journal of Forest Research,2008,38(7):1750-1760.
[37]
REYNOLDSM R, BURKT E, HUANGW C.Goodness-of-fit tests and model selection procedures for diameter distribution models[J].Forest Science,1988,34(2):373-399.
[38]
GOERNDTM E, MONLEONV J, TEMESGENH.Relating forest attributes with area- and tree-based light detection and ranging metrics for western Oregon[J].Western Journal of Applied Forestry,2010,25(3):105-111.
[39]
TIANX L, SUNS C, MOLA-YUDEGOB,et al.Predicting individual tree growth using stand-level simulation,diameter distribution,and Bayesian calibration[J].Annals of Forest Science,2020,77(2):57-74.
[40]
HAOY S, WIDAGDOF R A, LIUX,et al.Estimation and calibration of stem diameter distribution using UAV laser scanning data:A case study for larch (Larix olgensis) forests in Northeast China[J].Remote Sensing of Environment,2022,268:1127-1169.
[41]
PEUHKURINENJ, TOKOLAT, PLEVAKK,et al.Predicting tree diameter distributions from airborne laser scanning,SPOT 5 satellite,and field sample data in the perm region,Russia[J].Forests,2018,9(10):639-654.
WANGJ C, YANM H, HUANGQ L,et al.Diameter distribution of semi-natural mixed forest of Pinus massoniana and broadleaved trees based on stratification[J].Forest Research,2021,34(3):72-80.
[44]
GALLAGHER-DUVALX, VAN LIERO R, FOURNIERR A.Estimating stem diameter distributions with airborne laser scanning metrics and derived canopy surface texture metrics[J].Forests,2023,14(2):287-305.