Integrating GEDI Observations with a Forest Model Predicts Stand-level Attributes

Abstract:

Full waveform LiDAR remote sensing, such as the Global Ecosystem Dynamics Investigation (GEDI), provides high-resolution observations of vegetation structure. While many studies use statistical modeling to derive ecological attributes from such observations, comparatively less attention has focused on the nature of the relationship between forest structure and the ecological information it provides, and specifically the assumption that there is a unique relationship between forest structure metrics and an ecological attribute. Here, we use the GEDI Simulator to develop a fusion between GEDI observations and an individual-based forest model (FORMIND) to characterize the relationship between forest structure, aboveground biomass density (AGBD), and a stand’s stem-size distribution. We evaluate the extent to which forest stands generated by FORMIND’s Forest Factory approach produce congruent GEDI waveforms that may have different AGBDs and stem-size distributions, and conversely, the extent to which stands with comparable AGBD values and stem-size distributions may produce dissimilar waveforms. To validate these findings, we match GEDI waveforms simulated from FORMIND output to GEDI waveforms simulated from ALS LiDAR over stem-mapped plots, and compare modeled AGBD and stem-size distributions to the observed values calculated from the forest inventories. We find variability in the amount of ecological information embedded in LiDAR waveforms, and demonstrate waveform matching as a compliment to traditional, statistical approaches that derive ecological attributes from structural metrics. This work provides new insight into the relationship between forest structure and ecological attributes, and further, it innovates on a simple method by which to integrate GEDI data with the growing number of ecosystem models capable of simulating LiDAR data, which is of substantial value to this field.