330
BIOMASS P-band). Backscatter of shorter wavelengths is generally from the upper
canopy, while longer wavelengths penetrate farther into vegetation. Combining
shorter and longer wavelength data can be advantageous to detect contributions of
multiple scattering from different vertical portions of the canopy as well as ground
surface scattering and double bounce ground-stem scattering. Transmitted and
received radar signal polarizations [commonly horizontal (H) and vertical (V)] can
provide information on vegetation composition and structure. Co-polarized signals
(e.g., HH) respond more to trunk-ground configurations (double bounce scattering),
particularly at longer wavelengths, while cross-polarizations (e.g., HV) respond
more to canopy woody biomass (Ningthoujam et al. 2017). Combinations such as
ratios (e.g., HH/HV) can therefore enhance spatial differences in scattering types
and magnitude due to varying vegetation density (Mitchard et al. 2011). Similarly,
steeper incidence angles generally penetrate farther into the canopy, particularly in
leaf-on conditions, but multiple angles may enhance differences in vertical structure
(Henderson and Lewis 1998). All of the above characteristics related to the degree
of canopy penetration provide opportunity for analysis of vertical structural
complexity and composition diversity by using multiple bands, polarizations, and
incidence angles. Radar image texture information has also been found to be useful
in classification (Simard et al. 2000) and biophysical modeling (Kuplich et al. 2005).
Polarimetric data, where phase information is preserved (Ulaby et al. 1987),
allows additional analysis of polarization parameters and, through decomposition
analysis, the relative contributions from the various scattering mechanisms (e.g.,
surface, volume, double bounce) that are associated with canopy structural characteristics. Commonly applied decomposition techniques include (van Zyl 1989;
Cloude et al. 1996; Freeman and Durden 1998; Yamaguchi et al. 2005; Touzi 2007).
Several others were designed to build on or correct issues with previous techniques
(Hong and Wdowinski 2014). The Interferometric SAR (InSAR, e.g., Balzter
(2001)), provides a plain language description) incorporates transmission from two
different angles either simultaneously or in repeat passes (less preferable given
potential decorrelation of the response signals between passes). The phase differences between the radar response signals can be used to estimate the scattering
phase height center, which is associated with canopy density and arrangement. They
can also be used to generate digital elevation models (DEMs), with longer wavelengths that penetrate the canopy being more suitable. Using multiple parallel baselines, SAR tomography has been used to construct a 3-D representation of a given
volume such as a forest (e.g., Reigber and Moreira 2000).
Classification and Biophysical Modeling Applications
As with other RS technologies, landscape or vegetation diversity as an indicator of
biodiversity can be mapped through thematic classification. This can be accomplished across a broad gradient from nonvegetated to dense vegetation classes, to
map landscape cover types (e.g., Devaney et al. 2015) that may be associated with
A. Lausch et al.
BIOMASS P-band). Backscatter of shorter wavelengths is generally from the upper
canopy, while longer wavelengths penetrate farther into vegetation. Combining
shorter and longer wavelength data can be advantageous to detect contributions of
multiple scattering from different vertical portions of the canopy as well as ground
surface scattering and double bounce ground-stem scattering. Transmitted and
received radar signal polarizations [commonly horizontal (H) and vertical (V)] can
provide information on vegetation composition and structure. Co-polarized signals
(e.g., HH) respond more to trunk-ground configurations (double bounce scattering),
particularly at longer wavelengths, while cross-polarizations (e.g., HV) respond
more to canopy woody biomass (Ningthoujam et al. 2017). Combinations such as
ratios (e.g., HH/HV) can therefore enhance spatial differences in scattering types
and magnitude due to varying vegetation density (Mitchard et al. 2011). Similarly,
steeper incidence angles generally penetrate farther into the canopy, particularly in
leaf-on conditions, but multiple angles may enhance differences in vertical structure
(Henderson and Lewis 1998). All of the above characteristics related to the degree
of canopy penetration provide opportunity for analysis of vertical structural
complexity and composition diversity by using multiple bands, polarizations, and
incidence angles. Radar image texture information has also been found to be useful
in classification (Simard et al. 2000) and biophysical modeling (Kuplich et al. 2005).
Polarimetric data, where phase information is preserved (Ulaby et al. 1987),
allows additional analysis of polarization parameters and, through decomposition
analysis, the relative contributions from the various scattering mechanisms (e.g.,
surface, volume, double bounce) that are associated with canopy structural characteristics. Commonly applied decomposition techniques include (van Zyl 1989;
Cloude et al. 1996; Freeman and Durden 1998; Yamaguchi et al. 2005; Touzi 2007).
Several others were designed to build on or correct issues with previous techniques
(Hong and Wdowinski 2014). The Interferometric SAR (InSAR, e.g., Balzter
(2001)), provides a plain language description) incorporates transmission from two
different angles either simultaneously or in repeat passes (less preferable given
potential decorrelation of the response signals between passes). The phase differences between the radar response signals can be used to estimate the scattering
phase height center, which is associated with canopy density and arrangement. They
can also be used to generate digital elevation models (DEMs), with longer wavelengths that penetrate the canopy being more suitable. Using multiple parallel baselines, SAR tomography has been used to construct a 3-D representation of a given
volume such as a forest (e.g., Reigber and Moreira 2000).
Classification and Biophysical Modeling Applications
As with other RS technologies, landscape or vegetation diversity as an indicator of
biodiversity can be mapped through thematic classification. This can be accomplished across a broad gradient from nonvegetated to dense vegetation classes, to
map landscape cover types (e.g., Devaney et al. 2015) that may be associated with
A. Lausch et al.