Using repeated small-footprint LiDAR acquisitions to infer spatial and temporal variations of a high-biomass Neotropical forest

Abstract

In recent years, LiDAR technology has provided accurate forest aboveground biomass (AGB) maps in several forest ecosystems, including tropical forests. However, its ability to accurately map forest AGB changes in high-biomass tropical forests has seldom been investigated. Here, we assess the ability of repeated LiDAR acquisitions to map AGB stocks and changes in an old-growth Neotropical forest of French Guiana. Using two similar aerial small-footprint LiDAR campaigns over a four year interval, spanning ca. 20km², and concomitant ground sampling, we constructed a model relating median canopy height and AGB at a 0.25-ha and 1-ha resolution. This model had an error of 14% at a 1-ha resolution (RSE=54.7Mgha^-1) and of 23% at a 0.25-ha resolution (RSE=86.5Mgha^-1). This uncertainty is comparable with values previously reported in other tropical forests and confirms that aerial LiDAR is an efficient technology for AGB mapping in high-biomass tropical forests. Our map predicts a mean AGB of 340Mgha^-1 within the landscape. We also created an AGB change map, and compared it with ground-based AGB change estimates. The correlation was weak but significant only at the 0.25-ha resolution. One interpretation is that large natural tree-fall gaps that drive AGB changes in a naturally regenerating forest can be picked up at fine spatial scale but are veiled at coarser spatial resolution. Overall, both field-based and LiDAR-based estimates did not reveal a detectable increase in AGB stock over the study period, a trend observed in almost all forest types of our study area. Small footprint LiDAR is a powerful tool to dissect the fine-scale variability of AGB and to detect the main ecological controls underpinning forest biomass variability both in space and time.