基于GEDI、Sentinel-2和机载激光雷达的森林冠层高度反演研究

Research on Forest Canopy Height Inversion Based on GEDI, Sentinel-2 and DJI-L1 Data

持续且大规模地监测森林冠层高度对于估计与森林有关的碳排放和分析森林退化至关重要。全球生态系统动态调查(GEDI)搭载的激光雷达自2019年4月以来一直在收集与植被垂直结构相关的数据。本研究首先利用Sentinel-2影像、SRTM-DEM数据和GEDI提供的森林高度产品构建森林高度预测模型。然后,利用大疆禅思L1(DJI-L1)无人机系统获取的激光点云数据验证构建模型的精度(R=0.73,RMSE=2.62 m,MAE=2.23 m)。最后,获取了2021年黑龙江省松峰山林区空间分辨率为30 m的全覆盖森林冠层高度数据。本研究证实了利用GEDI数据与时序光学数据联合获取大范围内的全覆盖森林冠层高度变为可能,对于森林碳储量的估计和森林资源的监测具有非常重要的意义。

Continuous and large-scale monitoring of forest canopy height is essential for estimating carbon emissions and analyzing forest degradation. The LiDAR on the Global Ecosystem Dynamics Investigation(GEDI) has been collecting data related to the vertical structure of vegetation since April 2019. In this study, the estimation model of forest canopy height was constructed by using Sentinel-2 images, SRTM-DEM data and canopy products provided by GEDI product. Then the accuracy of the model was verified by comparing with the canopy height obtained by DJI-L1 unmanned aerial vehicle(UAV)(R=0.73, RMSE=2.62 m, MAE=2.23 m). Finally, the continuous forest canopy height data with spatial resolution of 30 m in a forest area of Heilongjiang province in 2021 was obtained. In this study, combining GEDI data with optical data to obtain large-scale forest canopy height for continuous years was possible, which would be of great significance for estimating forest carbon stocks and monitoring forest resources.