Radar Clutter Modeling

Maria S. GRECO
Dipartimento di Ingegneria dell’Informazione
University of Pisa
Via G. Caruso 16, I-56122, Pisa, Italy
[email protected]
 Clutter modeling 2

What is the clutter?

Clutter refers to radio frequency (RF) echoes returned from

targets which are uninteresting to the radar operators and
interfere with the observation of useful signals.
Such targets include natural objects such as ground, sea,
precipitations (rain, snow or hail), sand storms, animals (especially
birds), atmospheric turbulence, and other atmospheric effects,
such as ionosphere reflections and meteor trails.
Clutter may also be returned from man-made objects such as
buildings and, intentionally, by radar countermeasures such as
chaff.

2/157
Clutter reflectivity: General Concepts 3

We will focus particularly on sea and land

(or ground) clutter
Radar cross section (RCS)
RCS per unit illuminated area, σo (m2/m2)
RCS per unit illuminated volume, η (m2/m3)
Radar equation and pattern-propagation factor F
Sea clutter RCS and spikes
Land clutter RCS
Sea and land clutter statistics
The compound-Gaussian model
Clutter spectral models

3/157
 Clutter reflectivity 4

A perfectly smooth and flat conducting surface acts as a mirror,

producing a coherent forward reflection, with the angle of incidence
equal to the angle of reflection. If the surface has some roughness,
the forward scatter component is reduced by diffuse, non-coherent
scattering in other directions.
radar
specular reflection
θi=θr
θi θr
For monostatic radar,
clutter is the diffuse smooth
backscatter in the surface
direction towards the
radar specular
component

rough surface
diffuse
component

4/157
Radar Cross Section (RCS) 5

The IEEE Standard for RCS in square meters is

Ps
σ = 4π
Ωpi

where
Ps = power (watts) scattered in a specified direction
from the target having RCS σ
Ω = solid angle (steradians) over which Ps is scattered
pi = power density (watts/m2) of plane wave at target

5/157
Radar Cross Section (RCS) 6

Range of RCS Values (dBm2)

Echo power directly

proportional to RCS

Factors that influence

RCS:
size, shape, material
composition, moisture
content, surface coating
and roughness,
orientation, polarization, RCS of common objects
wavelength, multipath
6/157
 Normalized RCS σ0 7

The normalized clutter reflectivity, σ0, is defined as the total RCS, σ,

of the scatterers in the illuminated patch, normalized by the area, Ac,
of the patch and it is measured in units of dBm2/m2.

ψ Radar pulse
Elevation R
ρ σ 0 = σ / Ac
h θel

φgr
R antenna
range resolution
R θaz footprint

Azimuth
θaz
Clutter patch ρ secφgr Ac = αρ Rθ az sec(φ gr )

local grazing angle

The factor α accounts for the actual compressed pulse
shape and the azimuth beamshape
7/157
 Volume reflectivity η 8

The volume clutter reflectivity, η, is defined as the total RCS, σ, of

the scatterers in the illuminated volume, normalized by the volume
itself, Vc, and it is measured in units of dBm2/m3.

Pencil beam R θaz

R
θaz η = σ / Vc
R θel

θel ρ

Radar pulse

Vc = αρ R 2θ azθel one-way 3dB elevation beamwidth

8/157
Radar Equation and propagation factor F 9

For monostatic radar, received power Pr from a target with

RCS σ is

PG λ
2 2
Pr = t
σ F4

( )R
π
3 4
4

Pt = transmit power
G = antenna gain
R = distance of target from antenna
F = the pattern-propagation factor, the ratio of field
strength at a point to that which would be present if free-
space propagation had occurred

A clutter measurement provides either σF4 or σoF4.

Even so, normally the data are reported as being σ or σo.

9/157
Discrete and distributed clutter 10

DISCRETE
RCS depends on aspect angle, multipath environment,
frequency, and polarization
RCS values up to 30 dBm2 are common
RCS above 40 dBm2 rare, except in built-up areas
Nominal RCS values:
60 dBm2 very large ship or building
50/40 dBm 2 large building or ship
30/20 dBm2 small building/house
20/10 dBm2 trucks/automobiles

DISTRIBUTED
Average RCS = so times A, where A is illuminated surface
area (footprint) of a range-azimuth cell

10/157
 Sea clutter: Dependence on grazing angle 11

At near vertical incidence, the backscatter is quasi-specular and

varies inversely with surface roughness with a maximum at vertical
incidence for a perfectly smooth surface.

At medium grazing angles the reflectivity shows a lower

dependence on grazing angle (plateau region).

Below some critical angle

(~ 10º, depending on the 0 dB
roughness) the reflectivity

10 log σO
reduces rapidly with smaller
grazing angles (interference
region, where propagation is
strongly affected by
multipath scattering and
shadowing). 0o θc θo 90o
GRAZING ANGLE θ

11/157
Empirical model for sea clutter σ0 12

Nathanson tables [Nat69]

The "standard" beginning 1969, updated 1990

HH and VV POLs; 0.1°, 0.3°,1°, 3°, 10°, 30°, 60° grazing
Many data sources, 60 different experiments
UHF to millimeter wavelengths
Reported by sea state up to state 6
Averaged without separating by wind or wave direction
Greatest uncertainties at lower frequencies and < 3°
Reported RCS generally larger than typical because
(a) experimenters tend to report strongest clutter and
(b) over-water ducting enhances apparent RCS

[Nat69] F.E. Nathanson, Radar Design Principles, McGraw Hill, 1969

12/157
 Empirical model for sea clutter σ0 13

GIT model [Hor78]

Variables: radar wavelength, grazing angle, wind speed, wind

direction from antenna boresight, wave height
Employs separate equations: for HH and VV polarization, and
for 1 to 10 GHz and 10 to 100 GHz
The 1-10 GHz model based on data available for grazing angles
of 0.1 to 10o and average wave heights up to 4 m (corresponds
to significant wave heights of 6.3 m)
Few liable σoF4 data available at 3o grazing and below,
and for dependencies on wind and wave directions
Graphs from the model appear to give “best guesses” of σoF4
versus grazing angles less than 10o

[Hor78] M.M. Horst, F.B. Dyer, M.T. Tuley, “Radar Sea Clutter Model”, IEEE International
Conf. Antennas and Propagation, Nov. 1978, pp 6-10.
13/157
Empirical model for sea clutter σ0 14

GIT model: Wind speed dependence

HH POL, 10 GHz
Cross-wave direction,
2 m signif. wave height,
winds 3, 5, 10, 20 m/s

σoF4 increases with wind

speed,
Critical angle unchanged,
because wave height
assumed fixed.

14/157
Empirical model for sea clutter σ0 15

GIT model: Dependence on significant wave

height h1/3
HH POL, 10 GHz
Cross-wave direction,
10 m/s wind speed,
h1/3 = 0.5, 2, and 6 m

In plateau region, σoF4 is

independent of h1/3 (for
fixed wind speed)
σoF4 increases with h1/3
(multipath reduces critical
angle) at angles < 1o

15/157
Empirical model for sea clutter σ0 16

GIT model: Comparison between HH and VV

POL, 10 GHz

Wind speed/wave height in

equilibrium
σoF4 increases with wind
speed and h1/3
HH/VV ratio increases with
increased surface roughness
and reduced grazing angle
HH>VV at small angles under
rough conditions at 1.25 and
10 GHz

16/157
Sea clutter spikes 17

Bishop (1970) classified sea clutter at X-band into four types:

Noise-like-clutter which appears similar to thermal noise

with no apparent sign of periodicity;

Clumped clutter which appears as a mixture of noise-like

returns and discrete clumped clutter returns that often
fades rapidly;

Spiky clutter which consists largely of a collection of

clumped returns with short persistence, e.g., 1 to 2s;

Correlated spiky clutter consisting of persistent clumped

returns. At any one time, an A-scope trace looks like a comb
with semi-randomly spaced teeth. The whole pattern moves
at speeds up to 40 knots. Individual spikes persisted for 10
to 40 s and the clutter level between spikes was virtually
zero.

17/157
 Empirical model for land clutter σ0 18

σo vs incidence angle for rough ground [Ula89]

4.25 GHz data, high moisture

content
σo is insensitive to surface
roughness at 10o (80o grazing)
Same insensitivity to
roughness observed at 80o
grazing in 1.1 GHz and 7.25
GHz data
Same shapes but lower σo for
dry conditions

[Ula89] Ulaby, F.T. and Dobson, M.C, Handbook of Radar Scattering Statistics for Terrain,
Artech House, Norwood MA, 1989
18/157
 Land clutter σ0 at low grazing angles 19

37 Rural Sites [Bil02] 0

-10

Spatial averages of σoF4 for θ -20

MEAN OF σO F4 (dB)
< 8o grazing
Larger spreads in σoF4 at
-30

lower frequencies -40

Resolution 150 & 15/36 m

-50
HH and VV polarizations
-60
At each frequency, the RANGE POL
RES. (M)1
150 H
median spatial average is -70
150
15/36
V
H +

roughly –30 dB
15/36 V X

10,000
1,000
100
Note: σoF4 can be larger at -80

the lower frequencies VHF UHF L- S- X-BAND

FREQUENCY (MHz)

[Bil02] J.B. Billingsley, Low-angle radar land clutter – Measurements and empirical models,
William Andrew Publishing ,Norwich, NY, 2002.
19/157
Clutter statistics: effect of spatial resolution 20

The scattered clutter can be written as the vector sum from

N random scatterers

z = ∑ σ i exp [ jφi ]
N
phase term
RCS of a single scatterer

With low resolution radars, N is deterministic and very high in each

illuminated cell. Through the application of the central limit
theorem (CLT) the clutter returns z can be considered as Gaussian
distributed, the amplitude r =|z| is Rayleigh distributed and the
most important characteristic is the radar cross section.

2r  r2 
p( r ) = 2 exp  − 2  u ( r )
σ  σ 
20/157
 Clutter statistics: effect of spatial resolution 21

This is not true with high resolution systems. With reduced cell
size, the number of scatterers cannot be longer considered
constant but random, then improved resolution reduces the average
RCS per spatial resolution cell, but it increases the standard
deviation of clutter amplitude versus range and cross-range and, in
the case of sea clutter, versus time as well.

Jakeman and Pusey [Jak78] showed that a modification of the CLT

to include random fluctuations of the number N of scatterers can
give rise to the K distribution (for amplitude PDF):

K distributed if N is a negative
1 N
Z= ∑ aie jϕi N
→∞
→ R = Z binomial r.v. (Gaussian distributed if N
N i=1 is deterministic, Poisson, or binomial)
2-D random walk N = E{N},{ai}i.i.d.,{ϕ i}i.i.d.

21/157
 The compound-Gaussian model 22

In general, taking into account the variability of the local power τ,

that becomes itself a random variable, we obtain the so-called
compound-Gaussian model, then
2r  r2 
p(r | τ ) = exp  −  u (r )
τ  τ 
∞
p (r ) = ∫ p( r | τ ) p(τ ) dτ ; 0 ≤ r ≤ ∞
0

According to the CG model:

z(n) = τ (n)x(n)

x(n) = xI (n) + jxQ (n)

Texture: non negative random
Speckle: complex Gaussian
process, takes into account the
process, takes into account the
local mean power
local backscattering
22/157
 The compound-Gaussian model 23

Particular cases of CG model (amplitude PDF):

ν
4ν µ  4ν   4ν 
K (Gamma texture) pR (r ) = ν −1  r  Kν −1  r u (r )
2 Γ (ν )  µ   µ 

νb ∞
2br  ν   r 2
 ν 
b

GK (Generalized ν b−2
pR ( r ) =
Γ (ν )  µ  ∫0 τ exp − τ −  µ τ   dτ
Gamma texture)  

∞
LNT (log-normal r 2 r2 1 2
exp  − − 2 ln (τ δ )   dτ
2 ∫τ 2
pR ( r ) =
texture) 2πσ 0  τ 2σ 

c −1
cr 
W, Weibull pR ( r ) =   exp  −( r b) c  u ( r )
bb

23/157
 The K model 24

The K model is a special case of

the compound-Gaussian model:
K-PDF (amplitude PDF) N = negative binomial r.v.
1 τ (local clutter power) = Gamma distributed
Amplitude R = K distributed
0.8 ν=1

ν=1.5 Gamma-PDF (texture PDF)

pR (r) E{R}

0.6 ν=4.5
3
0.4
2.5
ν=0.5
0.2 ν=0.5
2

ν=30

p τ (τ )
0 1.5
0 1 2 3 4 5 6 ν=1
r/E{R} ν=1.5
ν=10
1

The order parameter ν is a ν=2

0.5
measure of clutter spikiness
The clutter becomes spikier as ν decreases
Gaussian clutter:ν → ∞ 0
0 0.5 1 1.5 2 2.5 3
texture (τ)
24/157
The multidimensional compound-Gaussian model 25

• In practice, radars process M pulses at time, thus, to determine the

optimal radar processor we need the M-dimensional joint PDF
• Since radar clutter is generally highly correlated, the joint PDF cannot
be derived by simply taking the product of the marginal PDFs
• The appropriate multidimensional non-Gaussian model for use in radar
detection studies must incorporate the following features:

1) it must account for the measured first-order statistics (i.e., the

APDF should fit the experimental data)
2) it must incorporate pulse-to-pulse correlation between data samples
3) it must be chosen according to some criterion that clearly
distinguishes it from the multitude of multidimensional non-Gaussian
models, satisfying 1) and 2)

25/157
 The multidimensional compound-Gaussian model26
• If the Time-on-Target (ToT) is short, we can consider the texture as
constant for the entire ToT, then the compound-Gaussian model degenerates
into the spherically invariant random process (SIRP) proposed by Conte and
Longo [Con87] for modeling the radar sea clutter. By sampling a SIRP we obtain
a spherically invariant random vector (SIRV) whose PDF is given by

∞
1  z H M −1 z 
pZ ( z ) = ∫ exp  −  pτ (τ ) dτ
0 ( πτ )
M
M  τ 

where z=[z1 z2 . . . zM]T is the M-dimensional complex vector representing the

observed data.
• A random process that gives rise to such a multidimensional PDF can be
physically interpreted in terms of a locally Gaussian process whose power level
τ is random.
• The PDF of the local power τ is determined by the fluctuation model of the
number N of scatterers.
26/157
 Properties of SIRVs 27

• The PDF of a SIRV is a function of a non negative quadratic form:

q(z) = (z − mz ) H M −1 (z − mz )
• A SIRV is a random vector whose PDF is uniquely determined by the
specification of a mean vector mz, a covariance matrix M, and a
characteristic first-order PDF pτ(τ):

1
pZ ( z ) = hM ( q ( z ) )
(π ) M
M

∞
 q
hM(q) must be positive and hM ( q ) = ∫ τ − M exp  −  pτ (τ ) dτ
monotonically decreasing 0  τ

• First-order amplitude PDF:

σ
 r2 
pR (r ) = 2 h1  2  , σ 2 = E { R 2 } = E z
r
σ 
2
{ }
• A SIRV is invariant under a linear transformation: if z is a SIRV with
characteristic PDF pτ(τ), then y=Az+b is a SIRV with the same characteristic
PDF pτ(τ).
27/157
 Sea and land clutter APDFs 28

First order statistics:

The homogeneity of sea allows to consider its spatial distribution equivalent
to the temporal distribution. The same is not true for land clutter. The most
common distributions for the sea clutter are the K, GK and Weibull.

Because of the large spatial variability of land clutter the statistics in space
and time are different. The most common are in the table below.

1st order

Rayleigh, Rice, Weibull,

Time
Log-normal

Space Weibull, Log-normal

28/157
Power Spectral Density (PSD) models 29

•The question is: How to specify the clutter covariance matrix and the
power spectral density?

• Correct spectral shape impacts clutter cancellation and target detection

performance.
• The clutter spectrum is not concentrated at zero Doppler only,
but spreads at higher frequencies.
• There are several reasons for the clutter spreading:
Wind-induced variations of the clutter reflectivity (sea waves,
windblown vegetations, etc. ).
Amplitude modulation by the mechanically scanning antenna beam.
Pulse-to-pulse instabilities of the radar system components.
Transmitted frequency drift.
• The pulse-to-pulse fluctuation is generally referred to as internal
clutter motion (ICM).

29/157
 PSD models 30

The PSD is often modeled as having a Gaussian shape:

(
 f −m
) 
2

SG ( f ) = S0G exp  − 
f

 2σ 2f 
 
This is usually a mathematical convenience rather than any attempt at
realism. Often the Doppler spectrum will be strongly asymmetric and
the mean Doppler shift, mf, may not be zero. Clearly for land clutter
mf is usually zero, but for rain and sea clutter in general mf ≠ 0 and
will be dependent on the wind speed and direction.

From velocity/Doppler relationship v = fDλ/2,

standard dev. of scatterer velocity is σv = σfλ/2
Wind changes bandwidths, but typical σv are
Rain/chaff σv~1 to 2 m/s
Sea σv~1 m/s
Land σv~0 to 0.5 m/s
30/157
 PSD models: windblown ground clutter 31

The Gaussian PSD model was proposed by Barlow [Bar49] for windblown
spectra for noncoherent radar systems and over limited spectral
clutter spectra,
dynamic ranges (up to a level 20 dB below the peak level and to a maximum
Doppler velocity of 0.67 m/s)

Essentially all modern measurements of ground clutter spectra, with increased

sensitivity compared to those of Barlow, without exception show spectral
shapes wider than Barlow’s Gaussian in their tails

It had become theoretically well understood from 1965-67 on, that branch
motion in windblown vegetation generates spectra wider than Gaussian

In a much referenced later report, Fishbein et al. [Fis67] introduced the

power-law clutter spectral shape:

nsin( π n) 1 n is the shape

Pac ( f ) = parameter
break-point Doppler frequency where the 2πf c 1+ ( f f c )n
shape function is 3 dB below its peak zero-
Doppler level
31/157
 PSD models: windblown ground clutter 32

Common values of power-law exponent n used in PSD modeling are usually on the
order of 3 or 4, but sometimes greater
The evidence that clutter spectra have power-law shapes over spectral dynamic
ranges reaching 30 to 40 dB below zero-Doppler peaks is essentially empirical,
not theoretical.

However, there is no simple physical model or fundamental underlying reason

requiring clutter spectral shapes to be power law.

Recently, Billingsley [Bil91] showed that measurements at MIT-LL of windblown

ground clutter power spectra to levels substantially lower than most earlier
measurements (i.e., 60 to 80 dB below zero-Doppler peaks) indicate spectral
shapes that fall off much more rapidly than constant power-law at the lower
levels, at rates of decay approaching exponential:
exponential

λ is the radar transmission

λβe  λβ  wavelength and βe is the
Pac ( f ) = exp − e f
4  2  exponential shape parameter

32/157
PSD models: windblown ground clutter 33

Then, recent studies have demonstrated that the ground

clutter spectrum of windblown trees consists of three
components:

• coherent component
• slow diffuse component
• fast-diffuse component

The coherent component was the results of radar

returns from steady objects such as buildings, highways
and from movable objects at rest.
The coherent component is at zero Doppler.

33/157
Slow-diffuse & fast-diffuse components 34

The slow-diffuse component is the consequence of motions of

objects with moderate inertia (tree branches).
The slow-diffuse component occupies a relative narrow region
around zero Doppler.
The spectrum is approximately symmetrical and its spectral
density in dB scale decreases linearly with increasing absolute
values of Doppler frequency.
The fast-diffuse component is the result of movements in light
objects such as a tree leaves. This component has a spectral
density similar to a band-limited noise. Its magnitude is usually
compared to other components.
The spectral extent is of the same order as the Doppler shifts
that corresponds to the wind speed.

34/157
PSD models: windblown ground clutter 35

35/157
 PSD models: windblown ground clutter 36

Spectral shapes having equal AC (Fluctuating) Power

- Source: Billingsley (1996)

Each spectrum for wind

speed of about 20 mph
The Gaussian shape
reported by Barlow (1949)
The power-law shape from
Fishbein et al. (1967)
The exponential shape
from Billingsley and
Larrabee ( 1987)

36/157
Ground clutter spectra: X-band 37

37/157
Ground clutter spectra: S-band 38

38/157
Variation of the spectral slope diffuse components 39

S-band

39/157
L-Band forest PSD vs wind speed 40

Approximate linear
dependence of power density
in dB versus velocity, for all
wind speeds
For VHF through X band,
measured spectral shapes
versus Doppler velocity
found to be essentially the
same

Source: Billingsley (1996).

40/157
Sea clutter PSD 41

The relative motion of the sea surface with respect to the

radar causes an intrinsic Doppler shift of the return from
individual scatterers.
Because the motion of the scattering elements have
varying directions and speeds the total echo contains a
spectrum of Doppler frequencies.
Two effects are of interest:
• the spectral shape and width
• the mean Doppler shift of the entire spectrum.

41/157
 Sea clutter PSD 42

The spectrum of sea clutter is sometimes assumed to have

Gaussian shape. An approximate relationship between the
-3dB bandwidth ∆f of the spectrum and sea state S (Douglas
scale) has been derived by Nathanson:

∆f = 3.6 f 0(GHz ) S

The standard deviation of the Gaussian spectrum is related to ∆f

by the expression:

σ f = 0.42∆f

Recently more complex and realistic models have been proposed

for sea clutter PSD. We are going to analyze them later on.

42/157
 43

Radar Clutter:
Live recorded data

43/157
IPIX Radar Description 44

Transmitter
• frequency agility (16 frequencies, X-band)
• H and V polarizations, switchable pulse-to-pulse
• pulse width 20 ns to 5000 ns
• PRF=0 to 20 KHz
Receiver
• coherent receiver
• 2 linear receivers; H or V on each receiver
• quantization: 8 to 10 bits
• sample rate: 0 to 50 MHz
• BW=5.5 MHz
Antenna
• parabolic dish (2.4 m)
• pencil beam (beamwidth 1.1°)
• grazing angle <1°, fixed or scanning

Source: Defense Research Establishment Ottawa.

44/157
 Sea clutter temporal behaviour (30 m) 45

The spikes have different behaviour in the two like-polarizations (HH and VV)
The vertically polarized returns appear
to be a bit broader but less spiky

The dominant spikes on the HH

record persist for about 1-3 s.

45/157
Sea clutter 46

Long waves or swells Waves and whitecaps

46/157
Sea clutter temporal behaviour (30 m) 47

Waves 80 cm high

IPIX radar

Waves 2.4 m high

47/157
 Data Description 48

Dataset 19980204_22 19980204_22 19980204_22 19980204_22 19980204_22

3753 0849 3220 4024 3506
Date and time 02/04/1998 02/04/1998 02/04/1998 02/04/1998 02/04/1998
22:37:53 22:08:49 22:32:20 22:40:24 22:35:06
# Range cells 28 28 28 28 27
Start range 3201 m 3201 m 3201 m 3201 m 3201 m
Range res. 60 m 30 m 15 m 9m 3m
Pulse width 400 ns 200 ns 100 ns 60 ns 20 ns
# Sweep 60000 60000 60000 60000 60000
Sample per cell 60000 60000 60000 60000 60000
PRF 1 KHz 1 KHz 1 KHz 1 KHz 1 KHz
RF-freq. 9.39 GHz 9.39 GHz 9.39 GHz 9.39 GHz 9.39 GHz
Radar and
wave geometry

S S S S S

48/157
 Statistical Analysis: Amplitude Models 49

1  1 
LN, log-normal pR ( r ) = exp  − 2 ln ( r δ )  2  u ( r )
r 2πσ 2  2σ 

c −1
cr 
W, Weibull pR ( r ) =   exp  −( r b) c  u ( r )
bb

ν
4ν µ  4ν   4ν 
K (Gamma texture) pR ( r ) =  r  ν −1 
K r u (r )
2 Γ (ν )  µ 
ν −1
 µ 

νb ∞
GK (Generalized Gamma 2br  ν  ν b−2
 r 2
 ν 
b

texture)
pR ( r ) =
Γ (ν )  µ  ∫0 τ exp − τ −  µ τ   dτ
 

∞
LNT (log-normal r  r2
2 1 2
pR ( r ) =
2 ∫τ 2
exp  − − 
2 
ln (τ δ ) 
  dτ
texture) 2πσ 0  τ 2σ 
49/157
Histogram and moments 50

•A histogram is a graphical representation used to plot density

of data, and often for density estimation.

•A histogram consists of tabular frequencies, shown as

adjacent rectangles, erected over discrete intervals (bins),
with an area equal to the frequency of the observations in the
interval. The height of a rectangle is also equal to the
frequency density of the interval, i.e., the frequency divided
by the width of the interval. The total area of the histogram is
equal to the number of data.

•A histogram may also be normalized displaying relative

frequencies. In that case the total area is 1. The bins must be
adjacent, and often are chosen to be of the same size.

50/157
 Statistical Analysis: Results - 15 m 51

-2
10 100
Histo
W
LN
Averaged Clutter Power

10
-3 K
10
LNT
GK

PDF
1

-4
10
VV 0.1
HH
VH VV data
-5
10 0.01
0 4 8 12 16 20 24 28 0 0.05 0.1 0.15 0.2
Range Cell Amplitude (V)
1000
Data
W
• With resolution of 60 m, 30 m, LN
K

Normalized Moments
and 15 m a very good fitting 100 LNT
GK
with the GK-PDF.
• Negligible differences among 10
polarizations

1
1 2 3 4 5 6
Order
51/157
 Statistical Analysis: Results - 3 m 52

100
Histo
W
10 LN
K
• With resolution of 9 m and 3 m LNT
1 GK
histograms with very long tails

PDF
• Not big differences among 0.1

polarizations, but generally HH

data spikier than VV data 0.01

HH data
0.001
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Amplitude (V)
100 10
6

Histo Data
W 5 W
10 10
LN LN
K K

Normalized Moments
4
LNT 10 LNT
1 GK GK
PDF

3
10
0.1
2
10

0.01
10
VV data
0.001 1
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1 2 3 4 5 6
Amplitude (V) Order

52/157
 53

Average spectral models

53/157
Sea clutter average spectra 54

Capillary waves with wavelengths on the order of

centimetres or less. Generated by turbulent gusts of near
surface wind; their restoring force is the surface tension.

Longer gravity waves (sea or swell) with wavelengths ranging

from a few hundred meters to less than a meter. Swells are
produced by stable winds and their restoring force is the
force of gravity.

54/157
Sea clutter average spectra 55

In the literature, it has been often assumed that the sea

clutter has Lorentzian spectrum (i.e., autoregressive of
order 1).

Autoregressive (AR) models with the order P ranging

from 2 to 5 have also been proposed for modelling radar
clutter.

For the sea surfaces some experimental analysis at small

grazing angle, C and X-bands, indicate that the sea Doppler
spectrum cannot be expressed by the Bragg mechanism
only, but also by wave bunching (super-events).

55/157
 Sea clutter average spectra 56

Lee et al. showed that the spectral lineshapes can be

decomposed into three basis functions which are
Lorentzian, Gaussian, and Voigtian (convolution of the
Gaussian and Lorentzian):

Γ 2π 2
S( f ) =
( f − f L ) + ( Γ 2π )
2 2

a
∞ exp ( − x 2 )
S( f ) =
π ∫f−f  2
dx
−∞

V
− x + a
peak of the Lorentzian function fVe
 
Γ −1 = characteristic scatterer lifetime

a = Γ 2 πf Ve shape parameter
fV = centre of the Voigt function

56/157
How to estimate the clutter PSD 57

The PSD can be estimated parametrically (or model-based) or

non-parametrically, without any hypothesis on the model.

Non parametrically, we used the periodogram defined as:

2
Z( f )
P( f ) =
M

where Z ( f ) is the Fourier Transform of the data and M the

number of samples.

There are many variants of the periodogram (for instance,

method of Welch, Blackman and Tukey) [see e.g. Stoica and
Moses book on Power Spectral Analysis].

57/157
 Sea clutter average spectra 58

The spectrum is the sum of two basis

functions among: the Gaussian, the
HH polarization
-3
5 10 Lorentzian, and the Voigtian, with
different Doppler peaks.
-3 non-param
4 10
V+G non-par: periodogram
Gauss and Voigtian basis functions
-3
3 10
PSD

-3
5 10
VV polarization
-3
2 10
-3
non-param
4 10 V+G
-3
1 10
-3
3 10
0

PSD
-1000 -800 -600 -400 -200 0 200 400 600 800 1000 -3
2 10
Frequency (Hz)

-3
High peak (Voigtian): 450 Hz 1 10
Low peak (Gaussian): 320 Hz
0
High peak (Voigtian): 410 Hz -1000 -800 -600 -400 -200 0 200 400 600 800 1000
Low peak: (Gaussian): 250 Hz Frequency (Hz)

58/157
 AR modelling 59

An Autoregressive process of order P, AR(P), is characterized by the difference equation:

P
Z ( n) = ∑ aP ,k Z ( n − k ) + W ( n)
k =1
where the coefficients aP ,k are the process parameters, and W(n) is white noise
For estimating the AR(P) parameters, we use the Yule-Walker equations

 Rz (0) Rz (1) ⋯ Rz ( P − 1)   aP ,1   Rz (1) 

 R (−1) R (0) ⋮   a   R (2) 
 z z   P ,2  =  z 
 ⋮ ⋱ Rz (1)   ⋮   ⋮ 
  a   
R
 z (1 − P ) ⋯ Rz ( −1) R z (0)   P,P   z R ( P ) 
Ra = r
In our case, we don’t know the “true” correlation of the clutter, so we estimate it as

N −1− m
1
Rˆ Z ( m ) = ∑ z ( n) z ∗ ( n + m)
N k =0

59/157
 AR modelling 60

We replace the estimated correlation to the true one and we solve the linear
system
ˆ −1rˆ
aˆ = R
so obtaining an estimate of the characteristic parameters of the PSD.

-1
10 Periodogram We tried AR(P) with P=1 up to
AR (3)
16.
-2
10

AR(3) model shows good fitting

Normalized PSD

10
-3
with data and seems to capture
physical phenomena.
-4
10

-5
10 Good compromise between
model complexity and fitting
10
-6
accuracy.
-400 -200 0 200 400
Frequency (Hz)

Example of periodogram calculated on 60,000 HH polarized data.

60/157
 61

Ground clutter data

61/157
Data recorded at Wolseley site with MIT-LL Phase One radar 62

PHASE ONE radar parameters

Source: MIT-LL, courtesy Mr. J. B. Billingsley

• Frequency Band (MHz) VHF UHF L-Band S-Band X-Band

165 435 1230 3240 9200
• PRF 500 Hz
• Polarization (TX/RX) VV or HH
•Range Resolution 150, 36, 15 m
•Azimuth Resolution 13° 5° 3° 1° 1°
•Peak Power 10 KW (50 KW at X-Band)
•Antenna Control Step or Scan through Azimuth Sector
•Tower Height 60´ or 100´
•10 Km Sensitivity -60 dB
•Amount of Data 25 Tapes/Site
•Acquisition Time 2 Weeks/Site

62/157
 X-band ground clutter data in open agricultural terrain63
703

The illuminated area was covered by

agricultural crops (83%), deciduous
azimuth trees (11%), lakes (4%), and rural
farm buildings (2%).

2D clutter map

(360°)
703
1

range

azimuth
samples
3D clutter map
316
VV polarization

Black areas: high reflectivity

(buildings, fencelines, trees, bushes
aligned along roads)

(270°)
White areas: low reflectivity

1
(field surfaces) 316 range 1
(5.7 Km) samples (1 Km)
63/157
 Ground clutter data analysis64
3
10
VV polarization
2 4th range interval
histogram The analysis, performed on each range
10 Gauss
interval has shown that I and Q PDFs
10 deviate considerably from Gaussian:
the clutter amplitude is not Rayleigh
1
distributed
PDF

-1
10
0.4
10
-2 VV polarization histogram
0.35 4th range interval uniform
-3
10
0.3
-4
10 0.25
-0.05 -0.04 -0.03 -0.02 -0.01 0 0.01 0.02 0.03 0.04 0.05

PDF
I component 0.2

0.15

0.1
The phase is uniformly
distributed 0.05

(it may be not for DC offset, 0

-4 -3 -2 -1 0 1 2 3 4
quantization effects, etc.) Phase (radians)

64/157
 Ground clutter data analysis 65

12
10
1st and 2nd range intervals:
estimate
10
10 K the Weibull distribution
LN provides the best fitting
Normalized moments

10
8 Weib.

HH polarization
6
10 4th range interval
3
10
4 HH polarization histogram
10
4th range interval
2 K
10
2 LN
10
Rayleigh
1 Weibull
0 10
10
2 3 4 5 6

PDF
Moment order 0
10

3rd and 4th range intervals: -1

10

the data show a behaviour that

-2
is intermediate between 10
0 0.05 0.1 0.15 0.2
Weibull and log-normal Amplitude

65/157
 Ground clutter data analysis: windblown vegetation 66

Analyzed clutter data: - recorded at Katahdin Hill site by Lincoln Laboratory.

- Phase One X-band stationary antenna.
- HH-polarization, PRF=500 Hz, 76 range gates.

3D power map
2
10
histogram
Rayleigh
Windblown trees
10

PDF
1 1

Number of 76
pulse
-1
repetition 10
time intervals

Range cells
30720 -2
1 10
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08
Amplitude
Data set courtesy of Barrie Billingsley
of MIT – Lincoln Laboratory These data are Gaussian.
66/157
Spectral model on windblown vegetation 67
1
WP
-1
10 Exp
Gauss
-2 PL2
10
PL3
-3
10

PSD
-4
10

PSD, 35th range cell. -5

10

-6
10

-7
10
-250 -200 -150 -100 -50 0 50 100 150 200 250
Frequency (Hz)

Cell #35 Exp Gauss PL2 PL3

β/σ/fc/fc 5.95 (Hz m)-1 23.63 Hz 1.02 Hz 6.33 Hz

Non-Linear Least Squares

θ Log − NLLS = arg min ∑ Log10 Pac ( f i ,θ) − Log10S( fi )
2
(NLLS) method is used
θ
for parameter estimation: i

67/157