Canopy structure: the link between optical and lidar remote sensing through canopy spectral invariants

Abstract:

Canopy structure and chemistry are the dominant factors that determine the radiation budget of
vegetation. One approach to understand the role of canopy structure and disentangle it from
canopy chemistry is the canopy spectral invariants theory, or p-theory. Using p-theory, the
bidirectional reflectance factor (BRF) recorded by sensors can be simulated using a few
spectrally-invariant variables and leaf single scattering albedo. The p-theory is originally
developed for the optical domain and there are several challenges associated with it, such as the
assumption of black soil, its requirements for narrowband spectral information (e.g.
hyperspectral), and limitations in very dense forests. The main question of this study is can we
extend the concepts of p-theory to lidar to overcome these limitations? To answer this question,
we developed the theoretical framework in which variables associated with p-theory in the
optical domain can be estimated using lidar point clouds and full-waveform information. To
verify this framework, we conduct a series of experiments using the DART Monte Carlo ray-
tracing model and vegetation scenes with known canopy chemistry and structure such as those
offered in the Radiation Transfer Model Intercomparison (RAMI) project. Our preliminary
results show that there is a strong link between information provided by optical and lidar sensors
through p-theory. We show that information derived from lidar and some fixed, universal canopy
chemistry (i.e. dry matter, water, and chlorophyll content) are sufficient to simulate the optical
signature of a canopy with high accuracy. The results of this study advance our theoretical
understanding of light interaction with canopy elements and also have significant implications
for lidar-optical data fusion.