Despite its critical importance to carbon storage modeling, forest vertical structure remains poorly characterized over large areas. Canopy height estimates from current satellite missions such as ICESat-2 (Ice, Cloud, and land Elevation Satellite-2) offer promise to close this knowledge gap, but their validation is critically important to inform their measurement uncertainties and scientific utility. Using existing airborne laser scanning (ALS) data, the agreement of a variety of terrain and aboveground canopy height metrics including summary height statistics and percentiles, from ICESat-2’ Land, Water and Vegetation Elevation product (ATL08) product was assessed in 12 sites across six major biomes in the United States. The agreement between ATL08 and ALS heights was assessed using the mean bias (Bias, ATL08 – ALS), the mean absolute error (MAE) and their percent equivalents, percent bias (pBias) and percent MAE (pMAE), respectively. In general, the agreement between ATL08 and ALS terrain heights was high (Bias 0.18 m, pBias 0.1%) while canopy heights showed lower agreement (Bias −1.71 m, pBias −15.9%). Analyses by biome, time of acquisition and beam strength of the ICESat-2 photon data also showed generally higher agreement for ATL08 terrain than canopy heights. Analyses also showed the performance of ATL08 heights varied with canopy cover with ATL08 terrain heights showing the best agreement when canopy cover was between 40 and 70% while the best performance for ATL08 canopy heights was observed when canopy cover was greater than 80%. This observation, coupled with analyses by biome, indicate that ATL08 canopy heights are more suitable in relatively dense canopy environments such as conifer and broadleaf forests than relatively sparse environments such a temperate grassland and Savannas. Higher level canopy height percentiles (95th and 98th) showed higher agreement (mean Bias −12.5%) with ALS heights than lower percentiles (minimum, 25th^, mean pBias ~39.2%). These findings indicate that ATL08 canopy heights show more promise for routine canopy height characterization using the 95th and 98% percentiles but is limited in characterizing intermediate vertical structure. The observed performance differences between ATL08 terrain and canopy heights are attributed to differences in photon sampling rates over terrain and canopy surfaces which, compounded with background noise in ICESat-2 photon data, led to different effectiveness for ATL08 processing routines in filtering terrain and off-terrain points. This assessment of the impact of a variety of factors provides the vegetation community with an understanding of the capabilities and limitations of height estimates from the ICESat-2 ATL08 product.

尽管对碳储存建模至关重要，但森林垂直结构在大面积范围内的特征仍然很差。来自当前卫星任务（如 ICESat-2（冰、云和陆地高程卫星 2））的冠层高度估计有望缩小这一知识差距，但它们的验证对于告知其测量不确定性和科学效用至关重要。使用现有的机载激光扫描 (ALS) 数据，在 12 个地点评估了来自 ICESat-2' 土地、水和植被高程产品 (ATL08) 产品的各种地形和地上冠层高度指标的一致性，包括汇总高度统计数据和百分位数横跨美国六大生物群落。ATL08 和 ALS 高度之间的一致性使用平均偏差 (Bias, ATL08 – ALS) 进行评估，分别为平均绝对误差 (MAE) 及其等效百分比、偏差百分比 (pBias) 和 MAE 百分比 (pMAE)。一般来说，ATL08 和 ALS 地形高度之间的一致性很高（偏差 0.18 m，pBias 0.1%），而冠层高度显示较低的一致性（偏差 -1.71 m，pBias -15.9%）。ICESat-2 光子数据的生物群落、采集时间和光束强度分析也表明，ATL08 地形的一致性普遍高于冠层高度。分析还表明，ATL08 高度的性能随冠层覆盖而变化，ATL08 地形高度在冠层覆盖率介于 40% 和 70% 之间时表现出最佳一致性，而当冠层覆盖率大于 80% 时，观察到 ATL08 冠层高度的最佳性能。这一观察结果，再加上生物群落的分析，表明 ATL08 冠层高度更适合于相对密集的冠层环境，如针叶林和阔叶林，而不是温带草原和热带稀树草原等相对稀疏的环境。较高水平的冠层高度百分位数（第 95 和第 98）与较低的百分位数（最小值，第 25^,平均 pBias ~39.2%)。这些发现表明，ATL08 冠层高度在使用 95% 和 98% 百分位数进行常规冠层高度表征方面表现出更大的希望，但在表征中间垂直结构方面受到限制。ATL08 地形和冠层高度之间观察到的性能差异归因于地形和冠层表面光子采样率的差异，再加上 ICESat-2 光子数据中的背景噪声，导致 ATL08 处理例程在过滤地形和关闭-地形点。这种对各种因素影响的评估使植被群落了解了 ICESat-2 ATL08 产品高度估计的能力和局限性。