Airborne laser scanning for terrain modeling in the Amazon forest

ABSTRACT Very few studies have been devoted to understanding the digital terrain model (DTM) creation for Amazon forests. DTM has a special and important role when airborne laser scanning is used to estimate vegetation biomass. We examined the influence of pulse density, spatial resolution, filter algorithms, vegetation density and slope on the DTM quality. Three Amazonian forested areas were surveyed with airborne laser scanning, and each original point cloud was reduced targeting to 20, 15, 10, 8, 6, 4, 2, 1, 0.75, 0.5 and 0.25 pulses per square meter based on a random resampling process. The DTM from resampled clouds was compared with the reference DTM produced from the original LiDAR data by calculating the deviation pixel by pixel and summarizing it through the root mean square error (RMSE). The DTM from resampled clouds were also evaluated considering the level of agreement with the reference DTM. Our study showed a clear trade-off between the return density and the horizontal resolution. Higher forest canopy density demanded higher return density or lower DTM resolution.

摘要 针对亚马逊森林的数字地形模型（Digital Terrain Model，DTM）构建的相关研究尚十分匮乏。在利用机载激光扫描（Airborne Laser Scanning）技术估算植被生物量时，数字地形模型发挥着特殊且重要的作用。本研究探讨了脉冲密度、空间分辨率、滤波算法、植被密度以及坡度对数字地形模型质量的影响。研究团队对三处亚马逊森林区域开展了机载激光扫描测量作业，并基于随机重采样流程，将每组原始点云（Point Cloud）进行降采样，使其脉冲密度分别达到每平方米20、15、10、8、6、4、2、1、0.75、0.5及0.25个脉冲。研究人员将重采样点云生成的数字地形模型，与基于原始激光雷达（LiDAR）数据生成的参考数字地形模型进行对比：逐像素计算二者偏差，并通过均方根误差（Root Mean Square Error，RMSE）对偏差进行汇总统计。同时，研究人员还通过与参考数字地形模型的契合程度，对重采样点云生成的数字地形模型进行了评估。本研究结果表明，脉冲返回密度与水平分辨率之间存在显著的权衡关系。森林冠层密度越高，则需要更高的脉冲返回密度，或更低的数字地形模型分辨率。