Comparing LiDAR-generated above-ground biomass with field data in an old-growth native urban forest in Australia

Urban forests contribute significantly to ecosystem services, notably in the sequestration and storage of carbon. There is a growing interest in quantifying Above Ground Biomass (AGB) to evaluate their contribution to climate mitigation. Recent advancements in remote sensing technology have resulted in publicly accessible data that facilitates the estimation of AGB across various forest types. To assess its applicability to urban forests, we conducted a comparative analysis involving Light Detection and Ranging (LiDAR) data alongside several spectral indices, specifically the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Leaf Area Index (LAI), derived from satellite imagery. This comparison was made against field-based AGB estimates obtained from 136 ha of old-growth native forest in Brisbane, Queensland, Australia. Initially, the height, diameter at breast height, and species of living trees were recorded within 14 plots of 0.1 ha each (two plots from each of seven regional ecosystems) in the forest. This field data was subsequently integrated with wood density information to estimate AGB for each plot. The field estimates were then juxtaposed with a Canopy Height Model (CHM) constructed by calculating the difference between a digital terrain model and a digital surface model, both generated using LiDAR data, as well as with NDVI, EVI, and LAI values obtained from Sentinel-2 imagery for each plot. The results indicate that the CHM provided the most accurate predictions of AGB per plot when compared to field data, while the spectral indices yielded less satisfactory results. Utilizing simple linear regression (AGB = 19.97*CHM – 80.3, r-squared = 0.66, p-value < 0.001), the CHM was employed to model AGB across the entire forest, which included the calculation of total AGB and Above Ground Carbon. This was achieved by summing the estimated AGB across a one-ha grid and subsequently summing AGB based on the average AGB per regional ecosystem multiplied by its respective area. Both estimates (32,210 and 32,062 metric tons, respectively) were significantly lower than the field data estimate for living trees, which was 37,876 metric tons. The findings of this study demonstrate that CHM derived from publicly accessible LiDAR data reliably predicts field-sampled AGB at the plot level and effectively models AGB at the entire forest scale. Furthermore, the model was utilized to predict AGB across distinct regional ecosystem zones, highlighting its capability to capture the variability of AGB within diverse ecosystem settings.