Forest structure gradient along a Central Amazon catena revealed by ground LiDAR

We used an upward-looking rangefinder-type ground LiDAR to describe differences in structure between Central Amazon terra firme forest types along a topographic/soil gradient. The LiDAR recorded 1000 last-return distances from the ground per second. At a constant walking speed, we sampled six 150m transects per forest type. All were located in the footprint of the LBA micromet tower near Manaus. For each forest type along the gradient we produced (1) a vertical profile of leaf area density (LAD); (2) a frequency histogram of top-of-canopy (TOC) heights at 1m horizontal scale; and (3) a variogram of these fine-scale TOC heights. LAD profiles and TOC histograms show that topography imposes a gradient in canopy height and in the variance of this height. Low-lying riparian and campinarana forests on white sand have a lower and more homogenous canopy surface. Upper slope and plateau forests on well-drained clay and loam have a taller and more irregular canopy surface. Differences between the two topographic extremes were confirmed using a Canopy Height Model from airborne LiDAR. Autocorrelation of ground LiDAR TOC heights reached 40-100m horizontal distance in upper slope and plateau forests, but extended to less than 20m in the sandy lower slope and riparian forests. The long reach of spatial autocorrelation on upper slopes and plateaus may result from (1) a matrix of lower crowns between scattered emergents, (2) larger gaps caused by fallen emergents, and (3) broad crowns of live emergents.