基于无人机机载激光雷达估算黄河三角洲孤岛刺槐林地上生物量

研究旨在利用无人机机载激光雷达数据来计算黄河三角洲孤岛林场刺槐地上生物量。以黄河三角洲孤岛林场刺槐林为研究对象，利用无人机机载激光雷达数据，采用分水岭分割算法从单木尺度提取孤岛林场刺槐的树高和冠幅；然后利用背包移动雷达数据提取8 个样方单株刺槐胸径；再结合异速生长方程，从单木尺度计算8 个样方的刺槐地上生物量。为了验证结果，用野外实测树高和胸径验证雷达数据中提取的单木结构参数；最后利用无人机机载激光雷达提取的树高和冠幅及两者的对数与乘积的对数形式，构建估算刺槐地上生物量模型，对模型估算过程中存在不确定性进行讨论，获得孤岛林场生物量分布图，并且结合前人研究的孤岛刺槐健康状况进行分析。结果表明：（1）基于分水岭分割算法能较精确地从无人机机载激光雷达数据中提取刺槐的单木结构参数。（2）树高与冠幅乘积的对数模型估算效果（R2=0.82，RMSE=3.66 kg/株）优于非对数模型（R2=0.58，RMSE=6.73 kg/株），也优于两者对数的模型（R2=0.76，MSE=4.52 kg/株）。（3）模型构建过程中有一定的不确定性，这种不确定性主要来自单木识别过程。（4）刺槐地上生物量高低与其健康状况有很强的相关性。

Estimating Aboveground Biomass of Robinia pseudoacacia in the Yellow River Delta Based on UAV Airborne Laser Radar

The aim is to calculate the aboveground biomass of Robinia pseudoacacia in Gudao forest farm of the Yellow River Delta based on UAV airborne laser radar. Taking the R. pseudoacacia forest in the island forest of the Yellow River Delta as the research object, using the UAV airborne laser radar data, the watershed segmentation algorithm was used to extract the height and the crown diameter of the R. pseudoacacia in the island forest from the single wood scale; then the backpack mobile radar was used to fit eight samples. The DBH was combined with the allometric growth equation to calculate the biomass of 8 plots of R. pseudoacacia forests from the single wood scale. To verify the results, the measured tree height and DBH were used to validate the single wood structure parameters extracted from the radar data. The tree height and crown diameter extracted by UAV LIDAR and the logarithmic form of logarithm and product logarithm of the UAV were used to construct and estimate the aboveground biomass model of R. pseudoacacia and discuss the uncertainties in the model estimation process. Biomass distribution map was obtained, and previous research on the health status of the island-specific R. pseudoacacia was analyzed. The results showed that: (1) based on the watershed segmentation algorithm, the single wood structure parameters of the R. pseudoacacia could be extracted from the UAV LIDAR data accurately; (2) the logarithmic model of the product of crown and tree height (R2=0.82, RMSE=3.66 kg/plant) was superior to the non-logarithmic model (R2=0.58, RMSE=6.73 kg/plant) and was better than the logarithm model (R2=0.76, RMSE=4.52 kg/plant); (3) there was a certain degree of uncertainty in the model construction process, this uncertainty mainly came from the single wood recognition process; (4) there was a strong correlation between the aboveground biomass and its health status.