Symposium on Remote Sensing for Resources Development and Environmental Management / Enschede / August 1986
121
Measurements of the backscatter and attenuation properties
of forest stands at X-, C- and L-band
D.H.Hoekman
Wageningen Agricultural University, Netherlands
ABSTRACT: The new airborne multi-band scatterometer (DUTSCAT) promises to be a useful tool for research
in the field of active microwave remote sensing. An evaluation of the use of this system for research
in forestry is given. Besides accurate o° values the system can acquire information on the vertical dis
tribution of backscattering. Through inversion of the multi-level model the parameter a 0 can be divided •
in contributions from a number of layers (3 or 4). An experiment with big corner reflectors placed on the
forest floor was conducted in an effort to gain more insight into the attenuating properties of the forest
canopy. The measurements of attenuation properties together with the division of O 0 in contributions from
several layers simplifies the model-making effort considerably.
KEYWORDS: Active microwave remote sensing, forest canopy attenuation, sources of scattering, modelling.
I . INTRODUCTION
The research activities of the working group ROVE-
Forestry concentrate on modelling radar backscatte
ring of forests and the identification of poten
tial applications of radar remote sensing in fores
try.
Until recently experimenting in the Netherlands by
ROVE-Forestry was limited to the use of an X-band,
HH-polarized SLAR.
The introduction of an airborne multi-band scatte
rometer (DUTSCAT) in 1986 promises to be a major
break-through in research possibilities. Some results
in L- and C-band recently obtained with experi
mental (single band) versions of the new system
will be discussed in this paper.
Both sensors (SLAR and scatterometer) are primarily
used for the acquisition of accurate values of O 0
or Y • Other physical parameters like the canopy
attenuation or the vertical distribution of sources
of backscattering in the forest canopy can be mea
sured too. The physical parameter Y is the only
parameter SLR systems normally supply and this para
meter together with the appropriate ground truth
data would in principle suffice to study the poten
tial applications of interest. Therefore it seems
peculiar, at first sight, that the use of the re
search instruments, the X-band SLAR and DUTSCAT, has
not been limited to measurements of Y . The reason
to measure other physical properties of the forest,
as far as possible, is that these data can simplify
the model-making effort considerably.
Applying radar remote sensing in general asks for
methods to interpret the acquired radar imagery. For
some applications the spatial information (patterns,
textures) is relevant, for other applications the
temporal information (changes). For most applica
tions however relevant information is contained in
the absolute value of y .To extract this informa
tion backscatter models are essential. The modelling
can be done in various ways. For specific applica
tions one can start from the assumption that rela
tions between y values and object (forest) parame
ters can follow empirically constructed expressions
well enough. Modelling radar data on a physical
basis however is more complicated but is to be pre
ferred since these models are likely to have a broa
der validity and can be used to extrapolate thus
gained insights to other situations or to explain
observed phenomena still unknown or unexpected.
In general models with a physical basis describe
the radar return y as a sum of direct contributions
of different types of scatterers in the object's
volume and interactions (e.g. multiple scattering)
between these scatterers. In these models the inci
dent power and the received backscattered power of
scatterers in lower layers is attenuated by scat
terers in layers, above. A direct acquisition of data
on attenuation properties or a method to indicate the
sources of scattering directly would be, along with
Y measurements, of great importance in model
research.
2. SENSOR DESCRIPTION
.The X-band SLAR is internally calibrated and has
digital recording facilities for radar and flight
data. Pre-processing software has been developed
to accurately correct the raw radar data both
geometrically and radiometrically. The resulting
imagery indicates relative Y values, the final
pixel size is 7.5 meter square. The specifications
of the SLAR are listed in table 1.
Table 1. Specifications of Dutch digital SLAR.
Frequency
Antenna
Transmitted power
Pulse length
PRF
Polarization
(Two-way) antenna beamwidth
Sample frequency
Dynamic range
Pixel size
Geometric resolution
in range
in azimuth
Independent samples
9.4 GHz
2.5 m slotted wave guide
25 kW
50 ns
200 Hz
HH
10 mrad
50 MHz (3 m) , 12 bits
70 dB
7.5 m x 7.5 m
7.5 m
10 m/km
~ 15 per pixel
The airborne scatterometer DUTSCAT (Attema & Snoeij
1985) is sideways looking and of the coherent pulse
type. The parabolic dish antenna has a two-way
beam width of 13 degrees at L-band and 3 degrees
at C-band, is mechanically steerable during flight
and can be pointed between 0 degrees and 80 degrees
incidence angle. The range resolution of 15 meters