Continental-Scale Canopy Height Modeling by Integrating National, Spaceborne, and Airborne LiDAR Data

Abstract Canopy height estimates are widely used in forest biomass and carbon assessment modeling applications with the goal of mitigating climate change through the modification of forest sustainability strategies. As a result, large-scale accurate estimates of contemporary forest conditions are required. The current study utilizes Random Forest (RF) algorithms to integrate land cover, vegetation, soil, and other supplementary data with Geoscience Laser Altimeter System (GLAS) data to predict a wall-to-wall canopy height model (CHM) across Australia. Multiple CHMs are predicted from RF models trained from unique permutations of 6 predictor variables. Each 250 m resolution CHM is independently validated against airborne laser scanning (ALS) heights from 18 countrywide sites; the best CHM yielding R2 = 0.72, and RMSE = 7.43 m. The best countrywide CHM is compared against 2 similar products from the literature, both of which are subject to intersecting ALS performance assessment also. The developed CHM product utilizes up-to-date data, and is tailored to Australia, complementing the National Ecosystem Surveillance Monitoring project mandated to the Terrestrial Ecosystem Research Network (TERN) by the Australian Department of Environment. Furthermore, with future altimetry-based Earth observation missions due for launch, the developed CHM will act as a baseline from which monitoring investigations can be executed.