Scaling lidar-derived rainforest canopy metrics across a Mesoamerican landscape

A critical consideration when remotely sensing forest structure is how data resolution impacts signal interpretation. Grain size and spacing between signals affect the discrimination and convolution of canopy components whose spatial distribution results from micro- to macro-scale ecological processes. Remote sensing provides methods to assess scaling patterns of canopy measurements which are expected to vary with tree architecture and the physical environment. For the complex, broad-leaved moist forests of western Belize, standard canopy metrics, i.e. mean height, maximum height, rugosity, vertical diversity, and openness, obtained from airborne laser scanning were highly correlated at scales between 10 m and 100 m and autocorrelation patterns were similar for grain sizes from 10 m to 250 m across a range of lag distances to >1000 m. However, scaling behaviours were metric dependent. Values for mean height and canopy openness were stable, maximum height and rugosity increased, and vertical diversity exhibited a concave, parabolic response as grain size coarsened from 10 m to 1000 m. Because the differences may have implications for ecosystem classification and biophysical modelling, researchers need to quantify these patterns for different forested systems.