Evaluating canopy height metrics from spaceborne against airborne lidar and in situ observations in the temperate-boreal transition forests of North America

Forests play an important role in the sequestration of anthropogenic greenhouse gases from the atmosphere. Researchers and industry leaders use a variety of methods to estimate the amount of carbon taken up by and stored in forests. Ground-based forest carbon estimates are often limited by their reduced spatio-temporal coverage. LiDAR instruments such as NASAs Global Ecosystem Dynamics Investigation (GEDI) instrument on the International Space Station (ISS) improve forest carbon estimation by collecting detailed measurements of vegetation structure using laser pulses. Although GEDI provides increased spatial and temporal coverage, additional accuracy assessment for a variety of forest metrics is needed. As the GEDI mission continues to improve over the mission lifetime, it is important for researchers and managers to compare spaceborne LiDAR forest height measurements with other reference datasets such as airborne and in situ observations to quantify the error associated from sources such as geolocation, atmospheric conditions, forest type, and terrain, among others. Here we evaluate forest height metrics acquired from spaceborne LiDAR missions including GEDI and NASAs Ice, Cloud, and Land Elevation Satellite-2 (ICESat-2) against airborne LiDAR (G-LiHT) and in situ measurements across the temperate-boreal transition forests of North America. The method involves identification of coincident observations using a spatial buffer and Gaussian resolution adjustment to overcome the differences in sampling strategies among the data sources and to provide direct comparison of forest height metrics such as relative height. We also compare the overall distributions of forest height metrics at both local and regional scales to investigate the unique characteristics of each data sources measurements in different scenarios (forest types, terrain types, disturbance histories, etc.). Finally, we estimate the accuracy of each spaceborne LiDAR source against newly acquired G-LiHT airborne LiDAR data along with a robust network of in situ field measurements in Maine. These accuracy assessments of spaceborne forest height metrics provide valuable information for forest biomass uncertainty quantification using data from multiple sources integrated through joint modeling frameworks.