利用LiDAR离散点云估测针叶林叶面积指数

通过对小光斑激光雷达离散点云数据进行处理，提取了6个变量参数分别用于估测针叶林叶面积指数，为了提高模型估测精度及弥补单变量模型的不足，在单变量模型的基础上尝试多变量组合共同用于估测森林叶面积指数，经过对比得出单变量模型中OGF模型最好，拟合相关性R=0.897，预测精度p=0.959；多变量预测模型结果差异不大，拟合相关性均＞0.905，估测精度均＞0.957。同时为了验证模型的推广性，对点云数据进行随机稀释操作获得4种不同密度的点云数据，分别用于验证点云密度对OGF模型及OGF与LPI组合模型的影响，结果表明点云密度对模型结果的影响不大，即使在0.125倍点云密度时模型仍能较好的估测针叶林叶面积指数，满足生产需要。

Estimation of the Leaf Area Index of Coniferous Forests Using LiDAR Discrete Point Cloud

Based on processing discrete point cloud data of small footprint LiDAR,six variable parameters were extracted and used to estimate coniferous forest leaf area index (LAI).In order to improve the accuracy of the model and make up the deficiency of the univariate model,multi-variable combination was adopted. Compared with other univariate models,the OGF model was the most suitable one featuring in higher fitting correlation (R=0.897) and prediction accuracy (p=0.959).And there was little difference among the multi-variable prediction models whose fitting correlations were bigger than 0.905 and all the prediction accuracies were more than 0.957.In order to validate the generalization of the model,4 different densities of point cloud data which were obtained by random dilution operation,were used to verify the influence of point cloud density on OGF model as well as the combination model of OGF and LPI,respectively.The result suggested that point cloud density had little influence on the model results and could meet the need of practical production.