Photon recollision probability in heterogeneous forest canopies: Compatibility with a hybrid GO model

[1] Photon recollision probability, or the probability by which a photon scattered from a phytoelement in the canopy will interact within the canopy again, has previously been shown to approximate well the fractions of radiation scattered and absorbed by homogeneous plant covers. To test the applicability of the recollision probability theory to more complicated canopy structures, a set of modeled stands was generated using allometric relations for Scots pine trees growing in central Finland. A hybrid geometric-optical model (FRT, or the Kuusk-Nilson model) was used to simulate the reflectance and transmittance of the modeled forests consisting of ellipsoidal tree crowns and, on the basis of the simulations, the recollision probability (p) was calculated for the canopies. As the recollision probability theory assumes energy conservation, a method to check and ensure energy conservation in the model was first developed. The method enabled matching the geometric-optical and two-stream submodels of the hybrid FRT model, and more importantly, allowed calculation of the recollision probability from model output. Next, to assess the effect of canopy structure on the recollision probability, the obtained p-values were compared to those calculated for structureless (homogeneous) canopies with similar effective LAI using a simple two-stream radiation transfer model. Canopy structure was shown to increase the recollision probability, implying that structured canopies absorb more efficiently the radiation interacting with the canopy, and it also changed the escape probabilities for different scattering orders. Most importantly, the study demonstrated that the concept of recollision probability is coherent with physically based canopy reflectance models which use the classical radiative transfer theory. Furthermore, it was shown that as a first approximation, the recollision probability can be considered to be independent of wavelength. Finally, different algorithms for calculation of the recollision probability from measured or modeled radiation fluxes are presented and discussed in the article.