VEGETATION CLUTTER SPECTRAL PROPERTIES
IN VHF/UHF BISTATIC DOPPLER RADAR
V. Sizov1, Cheng Hu2, M. Antoniou3 and M. Cherniakov3
1. Moscow Institute of Electronic Technology, 124498, Moscow, Russia; email@example.com
2. Beijing Institute of Technology, 100081, China; firstname.lastname@example.org
3. University of Birmingham, Edgbaston, Birmingham, B15 2TT, UK;
Phone: + (44) 121 414 4286, fax: + (44) 121 414 4291
Keywords: Multistatic scattering, vegetation clutter, Doppler
radar, radar measurements
A simple model of vegetation clutter creation mechanism is
discussed for the bistatic and forward scattering Doppler
radar configuration. The foliage is considered as a set of
separate pendulums oscillating with different resonance fre-
quencies. On the basis of this model, predictions of clutter
spectral properties are made and compared with clutter data
measured on different frequencies in the VHF and UHF
Ground-based bistatic radars (BSR) are widely used for air,
marine and ground targets detection [1-4]. In such systems,
the target echo is received at the background of surface clut-
ter, and many researches debate the experimental clutter
measurements for different conditions and their statistical
Usually, vegetation clutter is considered as a part of surface
land clutter, characterized by its surface reflectivity. The total
clutter power is estimated as the power collected from the
surface illuminated by the radar antenna pattern. This de-
scription is suitable for radar having small resolution cell
normally lying in the far field of the antenna. Surface reflec-
tivity is approximately independent on radar frequency; the
results of many experimental measurements are given in 
for frequencies from the VHF to X bands. The average re-
flectivity observed by radar at low grazing angles is a strong
function of radar frequency due to propagation factors .
This model is good for long range (tens or more km) radar
with the antenna(s) installed high above the ground surface.
In the Microwave Integrated Systems Laboratory, the Uni-
versity of Birmingham, UK, research on a micro sensor for-
ward scattering radar (FSR) network for ground targets de-
tection and classification for situation awareness is at hand
This system utilizes a continuous wave (CW) carrier fre-
quency in the VHF (about 70 or 150 MHz) band for target
detection and rough target classification, as well as a higher
frequency (between 400 and 900 MHz) in the UHF band for
more precise target recognition.
The main peculiarity of FSR is the absence of range resolu-
tion. Target detection is only possible if the target is moving,
by selecting its Doppler signature. When the radar site is
surrounded by vegetation, foliage (branches and leaves) also
sways with the wind and creates clutter masking the useful
For the micro sensors FSR network, the baseline length is
hundred(s) of meters, and sensor antennas are installed di-
rectly on the ground surface. The foliage may grow on or
near the baseline, surround the radar position and overhang
it, so, foliage should be considered as a number of volume
distributed scatterers. The non-directional antennas of the
sensors and the absence of range resolution cause the clutter
collection within big volume. Even foliage situated far away
from radar position may produce clutter exceeding the target
Different volume distributed targets are known in radar the-
ory and practice – chaff, rain or snow, bird flocks or insect
clouds, fog, etc. All of them are moving with the wind, and
thus create a Doppler frequency shift.
The main difference of foliage scatterers is in the nature of
their movement: vegetation does not travel in some direction,
it only sways around a fixed position. It is known that sway-
ing foliage creates an additional noisy signal (clutter) in MTI
radar in the Doppler frequency band, but strictly speaking,
this noise does not have a Doppler Effect nature. This fact is
confirmed in  –forest clutter spectra under windy condi-
tions are practically independent of radar frequency over a
very wide frequency band (VHF to X). The problem has
been discussed in the technical literature since the early
days of radar development, but as far as the authors as
aware, it still has no reasonable explanation.
Most of the known clutter measurements were made for
monostatic pulse MTI radar with a limited resolution cell
volume [5-7]. The results of land clutter measurements by
Doppler FSR utilizing a 400 MHz CW carrier frequency
were given in . The foliage clutter power spectrum was
shown as rather narrowband (0.3-0.5 Hz), and having weakly
Thus, knowing the nature of vegetation clutter is important
for target detection analysis in BSR or FSR.
In this paper a simple model of vegetation clutter creation
mechanism is proposed. On this basis the predictions of clut-
ter spectral properties are made and compared with the clut-
ter data measured on different frequencies in VHF and UHF
UNDERSTANDING OF FOLIAGE CLUTTER
Moving target Doppler signature creation
Let’s consider the FSR configuration shown on Fig.1 (top).
Even though it is FSR for the target, for the vegetation sur-
rounding the radar position this is a common BSR configura-
Figure 1 – Moving target Doppler signature creation
The transmitter (TX) and receiver (RX) are situated on the
baseline along the y-axis at distances dT and dR from the ori-
gin of the coordinate system. Assume that a small target hav-
ing an isotropic RCS pattern moves along the x-axis and
crosses the baseline in the origin at the time moment t=0. For
a target moving with a constant velocity v, its x-coordinate
depends on time:
Two signals are received and interfered at the RX antenna –
the direct leakage signal from the transmitter,
signal reflected from the target,
u , and the
The leakage power is defined as
and RX antennas, respectively, and
tor or loss on the TX to RX paths.
T P – transmitted power,
G – gains of the TX
RT F− – propagation fac-
In accordance to the bistatic radar equation, the power re-
ceived from the target,
P , is
whereσ – target RCS, λ – radar wavelength and
– propagation factors from the TX-to-target and target-to-RX
1 F and
Assuming unity transmitted power
nitudes may be obtained as
), the leakage and target signal mag-
Let’s also suppose for simplicity that this model is two-
dimensional (2-D); the target height is equal to 0, as well as
both the antennas’ heights, and wave propagation conditions
correspond to free space. Under this condition, equations (4
and 5) may be simplified:
)()() 4 (
where both the TX-to-target (
lengths are functions of time as the target is moving. There-
fore, the amplitude of the target signal depends on time.
1 R ) and target-to-RX (
R ) path
The phases of the target and leakage rays may be found from
the system geometry (Fig.1):
, and (8)
The phase of the target signal also changes with time.
Received power, dBm
Thus, the total interference signal is defined as
The calculated signature for a target crossing a baseline with
at its midpoint is presented in Fig.1 (bot-
tom); the target RCS
and the target veloc-
A more complex, three-dimensional (3-D) model, takes into
account the antennas and target heights, as well as the two-
ray path wave propagation above real ground , but the
target Doppler signature remains similar to the 2-D model
shown here [13, 14].
We should only point out that if the target during its motion
crosses isophase contours (or isophase ellipsoids in 3-D
model) separated byπ , it means that the target signature
amplitude changes from maximum to minimum or vice
versa. When the target goes a long way through the isophase
surfaces (much more thanλ , or π
in the target signature are known as Doppler Effect for
bistatic radar configuration.
2 in phase), the oscillations
Note that the density of isophase surfaces (contours or ellip-
soids) in space depends on radar wavelength; shorter wave-
length gives a proportionally shorter π-phase distance be-
tween the surfaces, so that a moving target Doppler signature
spectrum extends proportionally with the radar carrier fre-
Model of a tree
Looking on the windblown tree or brush, it can be seen that
leaves or branches sway around some stable position. It
seems obvious to model the foliage as a number of separate
Figure 2 – Model of a tree as a number of pendulums
Certainly, unlike the tower clock pendulum, leaves and
branches do not oscillate on ideal trajectories. They sway
stochastically in general, on some random trajectory with its
own average period. In any case, they oscillate. As a first
approximation, we can consider them as pendulums oscillat-
ing with their individual resonant frequencies. The maximum
magnitude of equivalent pendulums swing is not too big; it
lies from millimeters to tens of centimeters depending on the
pendulum length and the wind speed. Very long branches
may swing by 1-2 meters, but practically not more. These
distances are comparable to the radar wavelength in the VHF
and UHF bands (λ varies approximately from 4.3 m to 0.35
m for carrier frequencies from 69 to 869 MHz). Thus, the
Doppler Effect condition (the target way must be much
longer thenλ ) is ruled out. But what is the signal created by
foliage while swaying?
Oscillating target signature creation
Let’s consider, for example, a bistatic radar with TX and RX
positions placed symmetrically
, the baseline length being
dimensions correspond to the conditions in which experi-
mental clutter measurements were made. One small target
(pendulum) with initial coordinates x0=10m, y0=30m and
height z0=5m oscillates in the direction of the x-axis with a
and a magnitude of oscillations A, so that tar-
get’s position varies with time as
Calculated target signatures for different magnitudes of os-
cillation (A=0.03, 0.1 and 0.3m) are shown in Fig.3 for three
carrier frequencies (151, 433 and 869 MHz).
Figure 3 – Oscillating target signatures
b) 433 MHz
Received power, dBm
a) 151 MHz
Received power, dBm
c) 869 MHz
Received power, dBm
The calculation was made by applying the two-ray path
propagation model above dry ground, using the equations
given in .
It is seen from the figure that as the magnitude of pendulum
swing increase, the magnitude of oscillations in the target
signature increases proportionally. Then, when the pendulum
during its motion travels a distance corresponding to a phase
more than π
2 the second harmonic of the pendulum reso-
nance frequency is produced. The rise in signature’s magni-
tude is limited (saturated) at this moment, but the magnitude
of the 2nd harmonic becomes bigger. Next, with the pendu-
lum swinging magnitude further increased, the 2nd harmonic
is also saturated and 3rd harmonic appears, and so on.
Obviously, highest harmonics are more visible at higher radar
carrier frequencies, because the density of isorange (and iso-
phase) surfaces increases as frequency increases.
Experimental verification of pendulum’s signa-
The pendulum experiment is described in detail in . We
shall repeat here only the results of the experiment, since it is
significant to our investigation. The signature of a metal
sphere with diameter 21 cm hanged as a pendulum with a
length of 2.35m was collected in an FSR configuration with a
baseline length of
, at a carrier frequency of 869
MHz. The pendulum oscillated with an initial magnitude of
1m which decreased in time due to air resistance. The 3-D
model was used to calculate the expected target signature
with antennas’ and pendulum’s heights of 0.3m above a con-
crete ground surface .
The calculated and measured pendulum signatures are shown
Figure 4 – Pendulum’s signature
There was a strong wind at the time of measurements, so the
pendulum trajectory was not ideal, having some random dis-
tortions caused by trajectory imperfection. Thus, the meas-
ured signal has some visible noisy component. Nevertheless,
we can see a good coincidence between the predicted and
measured target signatures.
When the magnitude of the pendulum swing is small (time
interval after 10 s on the figure) we can see only one har-
monic in the pendulum signature. When the magnitude of the
swing is still big enough (from 0 to 10 s), the second har-
monic of the same frequency appears. The amplitude of this
2nd harmonic depends on the pendulum magnitude at that
moment. A bigger pendulum magnitude creates bigger 2nd
harmonic amplitude. But no other frequency (Doppler fre-
quency) can be observed in the figure, even as the pendulum
speed varies with time; only the pendulum’s resonant fre-
quency and its harmonics.
We can also see a decrease in the signature’s magnitude after
15s in the measured signature, when the pendulum swing
becomes very small. So, pendulum experiment fully con-
firms our predictions about the oscillating target signature.
Expected foliage clutter properties
The total foliage clutter may be considered as the interfer-
ence between a large number of volume distributed foliage
scatterers, or equivalent pendulums,
where the amplitude
lent pendulum are defined by (7) and (9). Because the phases
may be considered as randomly distributed, the total clutter
power will be the sum of all scattered powers:
for each equiva-
) 4 (
, 2 , 1
As the signature creation mechanism for the pendulum is
known, we can predict the clutter properties on its basis:
• Each oscillating part of foliage (equivalent pendulum)
creates a signature with a frequency near its resonant
frequency or its harmonics.
• Big number of pendulums with different length, posi-
tion and moving direction creates a total foliage sig-
nature like Gaussian noise.
• A bigger equivalent pendulum length gives lower fre-
quency components in the clutter power spectrum.
• A bigger equivalent pendulum length has bigger RCS.
• So, the clutter spectral density decreases with fre-
• Most of the equivalent pendulums are less than the
wavelength, so their RCS
Rayleigh diffraction region. Thus, the total power of
foliage clutter is estimated to decrease significantly
σ is defined by the
05 1015 20
Received power, dBm
1st harmonic2nd harmonic
with radar carrier frequency (approximately in 40
• The small dimensions of most scatterers (in compari-
son with wavelength) imply that their RCS has an
omni-directional pattern, and the values of
neither on the direction of incident wave nor the scat-
tering angle. We can predict the independence of clut-
ter power from polarisation for radar wavelengths
which are long enough. For shorter wavelengths, the
scattering from branches (modelled as long cylinder
) and long leaves may cause some dependence on
the radar signal’s polarisation.
• As wind speed increases, the magnitudes of pendu-
lum’s oscillations also increase, but the clutter spec-
tral width remains practically the same. A simple
analogy may be done to the audible foliage rustle.
Under a weak wind the rustle can barely be heard.
Under a stronger wind you can hear a more noisy
sound. But you can not sense any considerable
change in its tonality.
• The clutter spectral width becomes wider only at very
strong winds, where the bigger parts of the trees oscil-
late with a magnitude comparable to the radar wave-
length. At this condition, the low frequency compo-
nent of the clutter power spectrum must have a limita-
tion. The effect of clutter spectra widening is more
visible at higher radar carrier frequencies.
A number of clutter measurements were made at four differ-
ent frequencies in the VHF (69 and 151 MHz) and UHF
9433 and 869 MHz) bands. The measurements were taken in
an urban park area shown in Fig.5.
Figure 5 – Clutter measurements position
The radar baseline (
it was surrounded by trees and brushes. The radiated power
was about 20 dBm at all frequencies. Clutter data were col-
lected on the hard drive of a personal computer.
) was clear from vegetation, but
An example of clutter realization is shown in Fig. 6, meas-
ured at 869 MHz. The increase in clutter magnitude corre-
sponds to an increase in wind speed. The total clutter and
divided in 5 parts according to the average clutter power (the
average clutter power of one part is twice larger than that of
the previous part). The power spectra for each part were cal-
culated and are presented in Fig. 7.
Figure 6 – Example of clutter realization
0 0.51 1.52
2.53 3.54 4.55
MHz One-sided Power Spectrum
Figure 7 – Clutter power spectra for different wind speed
Clutter power seems to increase proportionally with wind
speed, but practically no changes in spectral width can be
observed for different wind speeds, coinciding with the pre-
dictions made above.
Clutter power spectra measured at the same observation pe-
riod for different carrier frequencies are shown in Fig. 8.
The significant dependence of clutter power (28 dB differ-
ence) on radar carrier frequency (from 869 to 151 MHz) cor-
responds approximately to a predicted decay of 40dB/decade,
but the power spectrum width does not increase proportion-
ally to the carrier frequency, remaining practically the same
at all frequencies.
Thus, the results of clutter measurements in the bistatic FSR
configuration confirm the clutter spectral properties expected
on the basis of the pendulum model for foliage. These results
also coincide with those measured using monostatic radar [5,
0 204060 80100 120 140 160
12 3 4
0 0.51 1.52 Download full-text
2.53 3.54 4.55
One-sided Power Spectrum
Figure 8 – Clutter power spectra at different carrier frequencies
The foliage clutter properties in the VHF and UHF bands
were considered for the bistatic radar configuration.
A simple model describing the foliage clutter mechanism as a
number of oscillating pendulums has been proposed and used
to predict clutter properties.
A number of clutter data collections were taken at different
frequencies (151, 433 and 869 MHz), the main properties of
measured clutter coinciding with our expectations.
The clutter power spectrum width slightly increases with the
radar carrier frequency and does not exceed 0.5-2 Hz at the -
10 dB level, smaller values assigned for the VHF band, and
gradually increasing as we move towards the UHF band. The
power spectrum density decreases fast with frequency.
The clutter power increases proportionally with the wind
velocity. In the VHF and UHF bands no clutter power satura-
tion was observed as the wind velocity increased. But theo-
retical prediction shows this is possible at higher radar carrier
frequencies, coinciding with the results known for
monostatic radar clutter measurements .
The dependence of clutter power on the radar carrier fre-
quency F complies approximately to the F4 or 40dB/decade
Clutter data collection is in progress now for the 69 MHz
carrier frequency. The results of these measurements will be
included in the final version of the paper.
It seems interesting to formulate a more strict mathematical
description of foliage clutter, perhaps describing the tree us-
ing a fractal model. However, this is the topic of a future
The work reported in this paper is funded by the Electro-
magnetic Remote Sensing Defence Technology Centre
(EMRS-DTC), UK, Project no. 2/65.
 M. Cherniakov (ed.), Bistatic Radar, Principles and
Practice: John Wiley & Sons, 2007.
 N. J. Willis, Bistatic Radar: SciTech Publishing, 2005.
 V. S. Chernyak, Fundamentals of Multisite radar Sys-
tems: Gordon and Beach Science Publishers, 1998.
 J. N. Briggs, Target Detection by Marine Radar: The
Institution of Electrical Engineers, 2004.
 J. B. Billingsley, Low-Angle Radar Land Clutter - Meas-
urements and Empirical Models: William Andrew Publish-
 M. Sekine and Y. Mao, Weibull Radar Clutter: Peter
Peregrinus Ltd., 1990.
 D. K. Barton and S. A. Leonov (editors), Radar technol-
ogy Encyclopedia, Artech House, 1998.
 R. Abdullah, M. Cherniakov and P. Jancoviĉ, Automatic
Vehicle Classification in Forward Scattering Radar, in
Proc. First International Workshop on Intelligent Trans-
portation, 2004, Hamburg, Germany, pp. 7-12.
 M. Cherniakov, R. Abdullah, P. Janĉoviĉ, M. Salous
and V. Chapursky, “Automatic ground target classification
using forward scattering radar”, IEE Proceedings Radar,
Sonar and Navigation, Vol. 153, Issue: 5, pp. 427- 437.
 M. Cherniakov, M. Salous, P. Janĉoviĉ, R. Abdullah
and V. I. Kostylev, "Forward scattering radar for ground
targets detection and recognition ,'' in Proc. of the 2d Annu-
lar Technical DTC Conference , Edinburgh, UK, 2005
 M. Cherniakov, M. Salous, V. I. Kostylev and R. Abdul-
lah, "Analysis of forward scattering radar for ground targets
detection,'' in Proc. of the 2d European Radar Conference,
Paris, France, 6-7 Oct. 2005, pp. 145-148.
 V. I. Sizov, M. Cherniakov and M. Antoniou, “Forward
Scatter RCS Estimation for Ground Targets”, in Proc. of the
EuRAD2007, Munich, Germany, 10-12 Oct. 2007, (accepted
 V. Sizov, M. Cherniakov and M. Antoniou, “Forward
scattering radar power budget analysis for ground targets” in
IET Radar, Sonar & Navigation, 2007 (accepted for publica-
 Cheng Hu, M. Antoniou, M. Cherniakov and V. Sizov,
“Quasi-Optimal Signal Processing in Ground Forward Scat-
tering Radar”, submitted to RadarCon2008.