利用无人机激光雷达估算红树林地上生物量

Estimation of aboveground biomass of mangrove forest using UAV-LiDAR

红树林作为热带地区碳储量最高的植被类型之一,面积呈现破碎化、减少趋势,地上生物量(AGB)的空间分布及其动态信息对于温室气体通量、碳储量的估算以及政策制定和实施至关重要。但是常用于AGB估算的光学数据或者SAR数据均存在信号饱和现象,且传统估算红树林生物量的算法对数据要求高、估算精度相对较低。针对该问题,本研究使用无人机激光雷达(UAV-LiDAR)数据对比了4种梯度增强决策树算法对于估算入侵红树林物种无瓣海桑AGB的精度,同时探讨了建模过程中的变量重要性。结果表明:(1)XGBR对于评估红树林AGB具有较高的拟合能力,达到R2=0.8338,RMSE=1.55 Mg/hm^(2);(2)研究区的无瓣海桑预测AGB的值为73.10~190.00 Mg/hm^(2),平均值为109.10 Mg/hm^(2);(3)描述冠层高度特征的激光雷达指标是估计红树林AGB的重要变量。本研究证明了UAV-LiDAR数据与XGBR模型对于估算红树林AGB的可行性,以期为红树林生态系统的蓝碳研究提供数据支撑。

As one of the vegetation types with the highest carbon storage in tropical regions,the area of mangrove forest shows a trend of fragmentation and reduction.The spatial distribution and dynamic information of mangrove biomass are crucial to the estimation of greenhouse gas flux and carbon storage,as well as policy formulation and implementation.However,both optical data and SAR data commonly used for biomass estimation have signal satur-ation phenomenon,and traditional estimation algorithms for mangrove biomass estimation have high data require-ments and relatively low estimation accuracy.In order to solve this problem,this study compared the accuracy of four gradient enhanced decision tree algorithms for estimating aboveground biomass(AGB)of invasive mangrove species Sonneria apetala used UAV-LiDAR data,and discussed the importance of variables in the modeling pro-cess.The results indicate that:(1)XGBR had a high fitting ability for the estimation of mangrove AGB,reaching R²=0.8338,RMSE=1.55 Mg/hm^(2).(2)The predicted AGB in the study area ranged from 73.10 Mg/hm^(2) to 190.00 Mg/hm^(2),with an average of 109.10 Mg/hm^(2).(3)LiDAR index describing canopy height characteristics is an important variable for estimating mangrove AGB.Conclusion:This study proved the feasibility of UAV-LiDAR data and XGBR model for estimating the AGB of mangrove forests,in order to provide data support for the blue carbon research of mangrove ecosystems.