Accuracy assessment of high-resolution global canopy height maps in Patagonian Andean forests

Abstract

Tree canopy height is one of the most important indicators of aboveground biomass density, productivity and ecosystem structure in forests. The newest released global canopy height models as of 2023 are the 10 m spatial resolution High Resolution Canopy Height Model of the Earth (HRCH) and the first Global Map of Tree Canopy Height (GMTCH) at 1 m spatial resolution. The accuracy of both maps was mostly assessed with independent airborne laser scanning (ALS) data in boreal and temperate forests of the USA, Europe and Oceania, with limited validation datasets in tropical regions. This study aimed to provide an accuracy assessment of high and very high spatial resolution global canopy height maps in Andean Patagonian forests by comparing them with independent unmanned aerial vehicle (UAV)-mounted LiDAR datasets. The mean error of HRCH showed a positive vertical offset within a 2.7–6.6 m range, and the root mean square error (RMSE) values varied from 4.4 m to 8.8 m. In the more complex broadleaf native forests with dense canopies and steep slopes, HRCH underestimated canopy heights. Mean errors of GMTCH indicated a negative vertical offset from −2.1 m to −13.3 m at all study sites, and RMSE values ranged from 4.0 m to 14.9 m. HRCH tended to overestimate the height of low canopies and underestimate the tall ones, whereas GMTCH strongly underestimated the height of all canopies, regardless of their height. Global canopy height models represent a significant advancement in mapping forest attributes and offer new opportunities to obtain deeper insight into forests worldwide. According to the findings, we recommend using both global maps for height estimates as ancillary information at a regional level. The increased availability of ALS and UAV-mounted LiDAR databases in poorly surveyed areas, especially in South America and Africa, will strengthen the robustness of accuracy assessments for global canopy height models.