Solar radiation is a driver for most biophysical and biochemical processes occurring in plant ecosystems including photosynthesis and transpiration.
Radiative transfer models (RTMs) simulate the scattering and absorption of radiation and are used in a wide range of applications, including sensor data simulations, interpretation of remote sensing images, and sensitivity studies of remote sensing optical signals.
For some RTM applications a detailed description of canopy structure is required, which can be a time-consuming process to generate. A new paper details an algorithm that can reconstruct a variety of trees across and within given species from a relatively small amount of input data.
Růžena Janoutová, scientist at Global Change Research Institute of the Czech Academy of Sciences and her colleagues developed a method to created detailed reconstructions of trees from terrestrial laser scans and ancillary field measurements. According to the article published by in silico Plants, they tested their methods using three tree species with different crown architecture: Norway spruce (Picea abies), European beech (Fagus sylvatica), and white peppermint (Eucalyptus pulchella).
Terrestrial laser scans (TLS) emit laser beams and record the amount and intensity of returned pulses to collect information from surfaces in 3D. Each tree was scanned from at least two positions.
The algorithm comprises four main steps. The first step was segmentation of the terrestrial laser scan tree point cloud to separate wooden parts from foliage (Figure 1A). Second, they reconstructed trunks and branches from the point cloud of wooden parts (Figure 1B). Third, they established the distribution of the foliage (i.e., leaves or needle shoots) from ancillary measurements of leaf angle distribution and foliage point cloud as attractors (Figure 1C). Last, for spruce trees only, they separated foliage into two age categories on ancillary data of percentage of current-year and older needle shoots.
Four individuals of each tree species were successfully reconstructed despite significant variation in trunk and branch shapes and foliage spatial and angular distribution within and among species (see Figure 2).
As different RTMs use different levels of abstraction for describing radiative transfer in vegetation, the authors tested how differences in the detail of tree 3D architecture affect forest reflectance simulations. To do this, the authors used the existing and very complex 3D Discrete Anisotropic Radiative Transfer (DART) model. Using DART, they built virtual forest scenes from their detailed tree models and abstracted simple tree models and compared the impact tree 3D structure detail affected canopy reflectance.
For all three tree species, the reflectance signatures of the abstracted tree models departed from those of the detailed model as abstraction increased. Forest reflections in the NIR region was overestimated by up to 130%, and in the green region by up to 135%. Reflectance in the red chlorophyll absorption region was underestimated by up to -69% and blue region by -40% (Figure 3).
According to Janoutová et al., “Our method of producing detailed 3D representation of trees is robust enough to be applied on species with complex and very different crown architecture, using input TLS scans of different quality. Detailed 3D tree representations can be used to improve existing remote sensing application and allow for new sensitivity studies that were unfeasible using previous, much more simple tree abstractions. Moreover, we found that optimizing the 3D complexity of simulated forest stands was crucial to achieving desirable accuracy while keeping a reasonable simulation computational time of all the thousands of input combinations necessary for inverse retrieval applications”
Růžena Janoutová, Lucie Homolová, Jan Novotný, Barbora Navrátilová, Miroslav Pikl, Zbyněk Malenovský, Detailed reconstruction of trees from terrestrial laser scans for remote sensing and radiative transfer modelling applications, in silico Plants, 2021;, diab026, https://doi.org/10.1093/insilicoplants/diab026
This manuscript is part of in silico Plant’s Functional Structural Plant Model special issue.