基于GEDI弱波束回波波形的不同地形坡度森林冠层高度估测研究

Estimation of Forest Canopy Height Overestimated by GEDI Weak Beam Received Waveforms in Different Terrain Slopes

摘要:

针对星载LiDAR(Light laser detection and ranging)GEDI（Global ecosystem dynamics investigation）弱波束条件下低能量发射波激光脉冲引起的低信噪比回波以及高坡度地形导致的林分冠层回波与林下地形回波高度重叠以致森林冠层高度估测精度偏低问题。本文通过GEDI L2A产品提供的质量筛选字段对弱波束高信噪比、非延时回波、非降轨回波以及林地回波进行数据筛选，之后对比分析不同百分比波形长度参数rh_aN条件下森林冠层高度估测结果以确定估测模型适用因子，并引入地形坡度参数DTM(Digital terrain model)修正0°~5°、0°~10°、0°~15°、0°~20°、0°~25°、0°~30°、0°~>30°以及0°~5°、5°~10°、10°~15°、15°~20°、20°~25°、25°~30°、>30°地形坡度条件下森林冠层高度估测模型以解决由高坡度地形引起的林分冠层回波与林下地形回波重叠问题。研究结果显示，单字段quality_flag参数与组合字段stale_return_flag、degrade_flag、quality_flag、sensitivity参数筛选结果相同，筛选后光斑点保有率为66.60%；选用波形长度字段参数rh_64可实现研究区森林冠层高度估测整体精度达到最高，R2、RMSE分别为0.5562和4.196m；坡度校正后森林冠层高度整体估测精度R2、RMSE分别为0.5665和4.150m。研究结果表明森林冠层高度估测精度整体随地形坡度增大呈下降趋势，而引入地形坡度因子后能够在一定程度上提高森林冠层高度估测精度，且地形坡度在0°~20°之间GEDI弱波束回波能够有效估测森林冠层高度。

Abstract:

The problem of low signal-to-noise ratio caused by low-energy emission laser pulses under weak beam conditions of spaceborne LiDAR (Light detection and ranging) GEDI (Global ecosystem dynamics investigation), as well as the overlap of forest canopy echoes caused by high-slope terrain with understorey terrain echoes, leading to underestimated accuracy of forest canopy height estimation. This study attempts to use quality screening fields in GEDI L2A product files to select data with high signal-to-noise ratio under weak beams, non-delayed echoes, non-descending echoes, and forest echoes, and then compare and analyze the overestimated forest canopy height under different percentage waveform length parameters rh_aN conditions to determine the applicable factors of the estimation model. Furthermore, terrain slope parameters DTM (Digital terrain model) are introduced to correct forest canopy height estimation models under terrain slope conditions of 0°~5°、0°~10°、0°~15°、0°~20°、0°~25°、0°~30°、0°~>30°, and 0°~5°、5°~10°、10°~15°、15°~20°、20°~25°、25°~30°、>30°, aiming to solve the problem of overlap between forest canopy echoes caused by high-slope terrain and understorey terrain echoes.The research results show that the single field quality_flag parameter is the same as the combination fields stale_return_flag, degrade_flag, quality_flag, and sensitivity parameters filtering results, and the retention rate of footprints after screening is 66.60%. Selecting the waveform length field parameter rh_64 can achieve the highest overall accuracy of forest canopy height estimation in the study area, with R2 and RMSE of 0.5562 and 4.196m respectively. After slope correction, the overall accuracy of forest canopy height estimation R2 and RMSE are 0.5665 and 4.150m respectively. The research results indicate that the overall accuracy of forest canopy height estimation decreases with the increase of terrain slope, while introducing terrain slope factors can improve the accuracy of forest canopy height estimation to a certain extent. Moreover, GEDI weak beam echoes can effectively estimate forest canopy height when the terrain slope is within 0°~20°.