65
et al. 2008; Darvishzadeh et al. 2008; Feret et al. 2011; Banskota et al. 2015;
Shiklomanov et al. 2016) or in hybrid approaches where statistical algorithms are
developed based on RTM simulations (e.g., Asner et al. 2011). RTMs encapsulate
our best mechanistic understanding of the coordination among leaf properties,
canopy structure, and resulting spectral signatures at the leaf and canopy scales,
but abstracted to operate with different degrees of complexity and assumptions
(Bacour et al. 2002; Nilson et al. 2003; Kobayashi and Iwabuchi 2008; See also
Morsdorf et al., Chap. 4; Ustin and Jacquemoud, Chap. 14).
At the leaf scale, RTMs were generally spawned from earlier work that identified
the relationships between fresh and dried leaf reflectance and a range of foliar traits,
including pigments, water content, nitrogen, dry matter, cellulose, and lignin. The
realization that leaf optical properties were fundamentally tied to the concentration
and distribution of leaf traits led to the development of models that could closely
mimic the spectral patterns across the shortwave spectral region (0.4–2.5 microns)
based on select leaf properties, such as chlorophyll and water content, as well as
structural variables. By far the most widely and commonly used leaf-level RTM is
the PROSPECT model (Jacquemoud and Baret 1990; Feret et al. 2008), which simulates leaf directional-hemispherical reflectance (R) and transmittance (T), allowing
for the calculation of leaf absorption (1-R+T) based on leaf biochemical and morphological properties, primary and accessory pigments, water content, LMA, or dry
matter content, brown material, and an approximation of the thickness of the internal leaf mesophyll layer (Féret et al. 2008; Féret et al. 2017). PROSPECT then
simulates leaf optical properties based on a generalized plate model describing
leaves as a stack of N homogenous absorbing layers that are calculated based on the
values of input leaf traits and their corresponding spectral absorption coefficient.
Other prominent leaf models include the Leaf Incorporating Biochemistry
Exhibiting Reflectance and Transmittance Yields (LIBERTY) model (Dawson et al.
1998) and LEAFMOD (Ganapol et al. 1998). In particular, LIBERTY is notable
given its original application focusing on improving the modeling of needle-leaf
evergreen conifer species and their leaf optical properties based on several leaf
traits, similar to PROSPECT, but also including foliar lignin and nitrogen content.
Moving to the canopy scale, RTMs are far more numerous with a wide variety of
complexities, assumptions, and requirements (Verhoef and Bach 2007; Widlowski
et al. 2015; Kuusk 2018). Most canopy RTMs leverage leaf-scale models, such as
PROSPECT, to provide the leaf optical properties (i.e., leaf single-scattering albedo)
needed to simulate canopy directional-hemispherical reflectance across select
wavelengths, simulated spectral bands, or specific SVIs. Generally, the soil boundary layer is either prescribed or simulated using a simple model of soil BRDF (e.g.,
Hapke model, Verhoef and Bach 2007), and stem or woody material reflectance and
transmittance (when used) is prescribed. Canopy RTMs can be separated into two
main classes, homogenous and heterogenous models. Homogenous models assume
the canopy to be horizontally unlimited and treated as a turbid medium of sufficiently large number of phytoelements (leaves, stems, other materials). For example, the Ross–Nilson model of plate medium (Ross 1981) assumes these elements
to be composed of small bi-Lambertian “plates” described by their reflectance and
transmittance properties with a specific leaf angle distribution (LAD). Leaves are
3 Scaling Functional Traits from Leaves to Canopies
et al. 2008; Darvishzadeh et al. 2008; Feret et al. 2011; Banskota et al. 2015;
Shiklomanov et al. 2016) or in hybrid approaches where statistical algorithms are
developed based on RTM simulations (e.g., Asner et al. 2011). RTMs encapsulate
our best mechanistic understanding of the coordination among leaf properties,
canopy structure, and resulting spectral signatures at the leaf and canopy scales,
but abstracted to operate with different degrees of complexity and assumptions
(Bacour et al. 2002; Nilson et al. 2003; Kobayashi and Iwabuchi 2008; See also
Morsdorf et al., Chap. 4; Ustin and Jacquemoud, Chap. 14).
At the leaf scale, RTMs were generally spawned from earlier work that identified
the relationships between fresh and dried leaf reflectance and a range of foliar traits,
including pigments, water content, nitrogen, dry matter, cellulose, and lignin. The
realization that leaf optical properties were fundamentally tied to the concentration
and distribution of leaf traits led to the development of models that could closely
mimic the spectral patterns across the shortwave spectral region (0.4–2.5 microns)
based on select leaf properties, such as chlorophyll and water content, as well as
structural variables. By far the most widely and commonly used leaf-level RTM is
the PROSPECT model (Jacquemoud and Baret 1990; Feret et al. 2008), which simulates leaf directional-hemispherical reflectance (R) and transmittance (T), allowing
for the calculation of leaf absorption (1-R+T) based on leaf biochemical and morphological properties, primary and accessory pigments, water content, LMA, or dry
matter content, brown material, and an approximation of the thickness of the internal leaf mesophyll layer (Féret et al. 2008; Féret et al. 2017). PROSPECT then
simulates leaf optical properties based on a generalized plate model describing
leaves as a stack of N homogenous absorbing layers that are calculated based on the
values of input leaf traits and their corresponding spectral absorption coefficient.
Other prominent leaf models include the Leaf Incorporating Biochemistry
Exhibiting Reflectance and Transmittance Yields (LIBERTY) model (Dawson et al.
1998) and LEAFMOD (Ganapol et al. 1998). In particular, LIBERTY is notable
given its original application focusing on improving the modeling of needle-leaf
evergreen conifer species and their leaf optical properties based on several leaf
traits, similar to PROSPECT, but also including foliar lignin and nitrogen content.
Moving to the canopy scale, RTMs are far more numerous with a wide variety of
complexities, assumptions, and requirements (Verhoef and Bach 2007; Widlowski
et al. 2015; Kuusk 2018). Most canopy RTMs leverage leaf-scale models, such as
PROSPECT, to provide the leaf optical properties (i.e., leaf single-scattering albedo)
needed to simulate canopy directional-hemispherical reflectance across select
wavelengths, simulated spectral bands, or specific SVIs. Generally, the soil boundary layer is either prescribed or simulated using a simple model of soil BRDF (e.g.,
Hapke model, Verhoef and Bach 2007), and stem or woody material reflectance and
transmittance (when used) is prescribed. Canopy RTMs can be separated into two
main classes, homogenous and heterogenous models. Homogenous models assume
the canopy to be horizontally unlimited and treated as a turbid medium of sufficiently large number of phytoelements (leaves, stems, other materials). For example, the Ross–Nilson model of plate medium (Ross 1981) assumes these elements
to be composed of small bi-Lambertian “plates” described by their reflectance and
transmittance properties with a specific leaf angle distribution (LAD). Leaves are
3 Scaling Functional Traits from Leaves to Canopies