基于机载LiDAR和光学遥感数据的热带橡胶林叶面积指数反演

叶面积指数（LAI）作为表征植被冠层结构的重要参数，一直是气候变化和生态研究中的热点，遥感技术的发展为大范围叶面积指数的获取提供了可能。以景洪市热带橡胶林为研究对象，以机载LiDAR和Landsat8/OLI为信息源，结合44块样地实测数据，使用支持向量机回归(SVR)、BP神经网络(BPNN)和偏最小二乘回归(PLSR) 3种模型,在前期建立基于林分水平的LAI估测模型的基础上，进一步构建区域尺度的LAI反演模型，实现景洪市橡胶林LAI的反演。结果表明，基于LiDAR的林分水平模型中，SVR模型最优，决定系数（R²）为0.76，相对均方根误差（rRMSE）为17%，估测精度（P）为83%；以SVR模型估测结果作为区域尺度遥感反演模型的先验样本，结合Landsat8/OLI数据的BP神经网络模型反演效果最好，估测精度达76%。

Inversion of Leaf Area Index of Tropical Hevea brasiliensis Forest Based on Airborne LiDAR and Optical Remote Sensing Data

As an important parameter to characterize vegetation canopy structure,leaf area index (LAI) has always been a hot topic in climate change and ecological research.The development of remote sensing technology provides an effective way to obtain LAI in a wide range.In this paper,the tropical Hevea brasiliensis forest occurring in Jinghong City was taken as the research object and the airborne LiDAR and Landsat8/OLI were taken as the information sources,combined with the measured data of 44 samples.Support vector machine regression (SVR),BP neural network (BPNN) and partial least-squares regression (PLSR) models were used.On the basis of the early establishment of LAI estimation model based on stand level,the regional scale leaf area inversion model was further constructed to realize the LAI inversion of H.brasiliensis forest in Jinghong city.The results showed that among the stand level models based on LiDAR,SVR model was the best.The R² of SVR model based on LiDAR was 0.76,rRMSE was 17% and P was 83%.Taking the inversion results of stand level model as a prior sample of regional scale remote sensing estimation model,the BP neural network model combined with Landsat8/OLI data had the best inversion effect,and the estimation accuracy was 76%.