Master Thesis Xavier Banque

Snow Monitoring over
Mountainous
Regions with C-Band
Polarimetric SAR
Author: Xavier Banqué Casanovas

Date: May 2012
Adviser: Carlos López Martı́nez
2
Acknowledgments
The author of the present Master Thesis would like to acknowledge the
company Starlab Barcelona S.L. in which most of the work of this research has
been carried out. Likewise, he would like to acknowledge the master adviser
Carlos López for his guidance and tutorship along the development of the work
described hereafter. Finally, also Spanish CDTI (Centro para el Desarrollo
Tecnológico Industrial) has to be acknowledged since it provided the funding for
the research project that framed this research work carried out by a consortium
of Starlab Barcelona S.L. and UPC-TSC.
4
Abstract
This note summarizes the research work conducted in the frame of the Master
Thesis Snow Monitoring over Mountainous Regions with C-Band Polarimetric
SAR, targeting the polarimetric backscattering response of the snow-pack at
C-band. The scope of the research work carried out in this Master Thesis was
to characterize the polarimetric backscattering response of the snow-pack in
mountainous regions, in order to asses the feasibility of a quantitative monitoring
method based on C-Band polarimetric SAR measurements.
CONTENTS 6
Contents
1 Introduction 7
2 Polarimetry Introduction 9
2.1 SAR Polarimetry Introduction . . . . . . . . . . . . . . . . . . . 9
2.2 Introduction to Polarimetry . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 Polarization Ellipse . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Jones Vector . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Stokes Vector . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.4 Covariance Matrix . . . . . . . . . . . . . . . . . . . . . . 14
2.3 Scattering operators and polarimetric characterization . . . . . . 15
2.3.1 Scattering coefficient . . . . . . . . . . . . . . . . . . . . . 15
2.3.2 Scattering matrix . . . . . . . . . . . . . . . . . . . . . . . 16
2.3.3 Polarimetric Coherency and Covariance Matrices . . . . . 18
2.3.4 Kennaugh Matrix. Huynen Decomposition . . . . . . . . 19
2.3.5 Canonical Scattering mechanisms . . . . . . . . . . . . . . 20
2.4 Polarimetric SAR Speckle Noise . . . . . . . . . . . . . . . . . . . 21
2.4.1 Speckle Noise Introduction . . . . . . . . . . . . . . . . . 21
2.4.2 Polarimetric Speckle Statistics . . . . . . . . . . . . . . . 22
2.5 Polarimetric decomposition theorems . . . . . . . . . . . . . . . . 22
2.5.1 Huynen decomposition . . . . . . . . . . . . . . . . . . . . 23
2.5.2 Freeman and Durden decomposition . . . . . . . . . . . . 24
2.5.3 Eigenvector-based decomposition . . . . . . . . . . . . . . 25
2.5.4 Pauli decomposition. Coherent Decomposition . . . . . . 25
2.6 SAR polarimetry Applications . . . . . . . . . . . . . . . . . . . . 26
3 SAR Polarimetry for Snow Remote Sensing 28

3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.1.1 Snow Water Equivalent . . . . . . . . . . . . . . . . . . . 29
3.1.2 Snow Remote Sensing . . . . . . . . . . . . . . . . . . . . 29
3.1.3 Polarimetric SAR Sensors . . . . . . . . . . . . . . . . . . 31
3.2 Snow interaction with EM waves . . . . . . . . . . . . . . . . . . 34
3.2.1 Snow physical modeling . . . . . . . . . . . . . . . . . . . 34
3.2.2 Snow-pack Microwaves Scattering Signature . . . . . . . . 35
3.2.3 Snow-pack Polarimetric Signature . . . . . . . . . . . . . 38
3.3 Non Coherent PolSAR Data Methods . . . . . . . . . . . . . . . 39
3.3.1 Snow Density Estimations Based on Co-pol SAR Data . . 39
3.3.2 SAR Data Plus Meteo Model Based Methods . . . . . . . 41
3.3.3 Snow-pack Thermal Resistance Based Model for Dry Snow
Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.4 Coherent PolSAR Data Methods . . . . . . . . . . . . . . . . . . 43
3.4.1 Polarimetric Decomposition over Glacier Ice . . . . . . . . 44
3.4.2 Other Polarimetric Decompositions Addressing Volume . 45
3.5 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4 C-band fully Polaimetric SAR measurements over Mountainous

Regions 48
4.1 Indroduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.2 Description of the Experiment . . . . . . . . . . . . . . . . . . . . 49
4.2.1 Experiment Planning and Design . . . . . . . . . . . . . . 49
4.2.2 Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
CONTENTS 7
4.3 Snow Backscattering Results Analysis . . . . . . . . . . . . . . . 53

4.3.1 Preliminary Analysis . . . . . . . . . . . . . . . . . . . . . 53
4.4 Snow Monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.4.1 Forested Zones . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4.2 Alpine meadows . . . . . . . . . . . . . . . . . . . . . . . 62
4.4.3 Urban Area Ground Truth . . . . . . . . . . . . . . . . . 63
4.4.4 Overall Analysis . . . . . . . . . . . . . . . . . . . . . . . 63
4.4.5 Data Interpretation . . . . . . . . . . . . . . . . . . . . . 71
4.5 Preliminary Results . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.5.1 Proposed ∆ Parameter Results . . . . . . . . . . . . . . . 74
4.5.2 Proposed H/ᾱ method Results . . . . . . . . . . . . . . . 78
4.5.3 Preliminary Results Conclusions . . . . . . . . . . . . . . 79
5 Conclusions and Further Steps 81

8
1 Introduction
The current Master Thesis research work is motivated by the idea of Snow
Remote Sensing. During the last decades, Snow Monitoring has been an in-
teresting research topic for both the scientific and the remote sensing commu-
nity, due to the important role of the snow-pack in the hydrological cycle and
its many derivative applications for hydrology, meteorology and climate study.
Snow quantitative remote sensing is indeed still a research hot topic, since the
retrieve of quantitative snow physical parameters, such as Snow Water Equiva-
lent (SWE), remains not yet solved to a satisfactory operational level. Seasonal
snow in mountainous regions is for most of the countries in the north hemi-
sphere the main source of water reservoirs, therefore the estimation of the SWE
contained in the snow-pack would be a highly valuable information for water
management in those countries.
Taking the fact that snow is almost transparent to C-band SAR signal as
a starting point, the research work carried out in this Master Thesis aimed to
evaluate the impact of using fully polarimetric C-band SAR data in the im-
provement of snow monitoring with regards to single polarization performance.
Polarimetric backscattering response of the snow-pack might be more sensi-
tive to snow parameters, provided that polarimetry is sensitive to the different
scattering mechanisms of distributed targets, such as snow-pack volume scatter-
ing. With this purpose, a research project called EOSWAN, funded by CDTI,
has been carried out by a consortium formed by Starlab Barcelona S.L. and
TSC-UPC. Current space-borne C-band polarimetric SAR sensors and future
missions, such as Sentinel-1, would guarantee the feasibility of a potential snow
9
remote sensing service based on polarimetric SAR data.

Therefore this Master Thesis describes the research work realized in the
frame of EOSWAN project summarizing the methodology adopted, the experi-
ments realized and the extracted conclusions and proposed further steps . The
text is organized in five sections as follows:
Introduction The present section, it briefly introduces the Master Thesis

work and describes the information organization in the text
Polarimetry Introduction In this section a theoretical introduction to

SAR polarimetry is provided, presenting the electromagnetics theory basis
on which is based the research work described in further sections. The
basic concepts will be presented for the sake of clarity in the following
sections.
SAR Polarimetry for Snow Remote Sensing This section presents

a state-of-the-art review of Snow Remote Sensing with Polarimetric SAR
data, as well as a review of the Active Microwave theory for snow remote
sensing. The preliminary literature research work resulting in this section
has been the departure point for our own research activities described in
section 4
C-band fully Polarimetric SAR measurements over Mountain-
ous Regions This is the core section of the Master Thesis with a wide
description of the research work carried out. It consists of a description of
the experiments realized, an analysis of the snow backscattering and the
final results on snow monitoring.
Conclusions and Further Steps The final section gathers the conclu-
sions and lessons learned during the development of the research work
and proposes some next steps to deepen in the topic of snow quantitative
remote sensing with C-band polarimetric SAR data.
10
2 Polarimetry Introduction
2.1 SAR Polarimetry Introduction
The SAR polarimetry technique aims to study the polarimetric response of a
scatterer to an incident polarized electromagnetic wave. Then, a scatterer is
characterized by the comparison of the polarimetric properties of the scattered
wave, with regards to those properties of the incident wave. Therefore, SAR
polarimetry is in pursuit of finding out determined parameters of the scatterers
from the knowledge of their polarimetric response.
The use of polarimetry in the frame of a SAR system adds some information
only available under polarimetric interpretation of the scattered signal, so that
we can talk about SAR Polarimetry. In order to take the maximum profit of the
polarimetric signature of a determined scatterer, this has had to be illuminated
by waves with both horizontal and vertical polarization and its scattered waves,
which will be a combination of both polarizations, have to be gathered also in
both horizontal and vertical polarization. In other words a fully polarimetric
system transmits and receives in the two linear polarizations, and there are few
instruments that can operate with such a polarimetric configuration. Therefore,
up to date, the amount of fully polarimetric SAR data available is reduced, and
polarimetric techniques are still a research field in a variety of target remote
sensing.
The polarimetric signature of a scatterer is to be analyzed in order to extract
the desired feature information from the observable. For instance, a scatterer
from whom no more information can be extracted under a non polarimetric
approach, can provide information about new parameters when observed in a
2.2 Introduction to Polarimetry 11
polarimetric approach. SAR polarimetry becomes specially interesting when

finding out that some important geo-phyical parameters in earth observation
can only be extracted when looking at the polarimetric features of the target,
such as the dry snow coverage.
In the last decade, SAR polarimetry has become an important technique for
remote sensing and consequently more in-orbit instruments with polarimetric
capabilities are available or will be soon. Several important missions from the
most significant space field actor countries point out the increasing importance of
SAR polarimetry techniques, that will benefit of the growing amount of available
SAR polarimetric data, to speed up the research and development of new remote
sensing based on them.
2.2 Introduction to Polarimetry

This section intends to introduce the general electromagnetic waves polarization
concept in a generic manner, since Polarimetry is concerned with the control of
coherent polarization properties of electromagnetic waves. An electromagnetic
wave is defined by an electric field vector and an orthogonal magnetic field vector
whose relation and time-space evolution is defined by Maxwell equations.
From the Maxwell equations, under the assumption of a linear medium free
of sources, the general propagation equation is obtained. Among the infinite so-
lutions of the propagation differential equation, the monochromatic plane wave
solution can be studied considering a propagation media free of mobile electric
charges. The expression of this equation can be significantly simplified by con-
→
− −
sidering the complex expression E (→ r ) of the monochromatic time-space electric
→
− −
field E (→
r , t), as follows:
→
− → h→
− − jwt i
E −r , t = Re E →

r e (2.1)
In a general way, a monochromatic plane wave with constant complex amplitude

−→
Eo propagation in the direction of the wave vector k̂ has the complex form:
→
− → −
→ → −→− →
− −
E (−
r ) = Eo e−j k r with E (→
r ) · k̂ = 0 (2.2)
where the wave number k is complex magnitude comprising the propagation

constant β and the medium losses α as it follows:
r
ε00
k = ωµε 1 − j = β − jα (2.3)
ε
The electromagnetic wave can be actually represented by the time dependent
propagation of the electric field vector. Given a plane monochromatic electro-
magnetic wave defined in the orthonormal coordinate space (x,y,z), propagating
in the k̂ = ẑ direction, its electric field vector is defined by the equation:
 
Eox cos(wt − kz + δx )
E(z, t) =  Eoy cos(wt − kz + δy )  (2.4)
0
where w is the angular frequency,k = 2π/λ is the wave number, δx , δy represent

the phase offsets for each component ad we have considered a loss free media,
so that the term α = 0. At a given time t = t0 , the electric field is composed
of two orthogonal sinusoids with different amplitudes and phases in general.
Figure 2.1: Monochromatic plane wave spatial evolution [3]
The relationship between amplitudes an phases of both sinusoids determines

the wave polarization.
Therefore, three types of polarization can be specified:
Linear polarization: δ = δx − δy = 0. The electric field is a sine wave
incribed in a plane with an orientation phi with respect to the x̂ axis.
Circular polarization: phase difference π/2 and equal amplitudes. δ =
δx − δy = π2 + kπ and E0x = E0y . The wave rotates circularly around x
axis with orientation φ(z) describing a circle.
Elliptic polarization: General case, the wave describes an elliptical shape
around the ẑ axis, helical trajectory as space evolves.
2.2.1 Polarization Ellipse

In order to characterize the wave polarization, analyzing the temporal evolution
at a fixed spatial point is preferred, since the spatial evolution has to deal
with helical trajectories, whereas observing the wave at a fixed point as time
evolves leads to an elliptical shape in an equiphase plane (x̂, ŷ) orthogonal to
the direction of propagation ẑ. The resulting ellipse, called polarization ellipse,
that describes the wave polarization, is as follows.
The Polarization ellipse is characterized by means of three parameters that
uniquely describe its shape:
Amplitude A is the ellipse axis:
q
A= 2 + E2
E0x (2.5)
0y
The ellipse orientation with respect to the x̂ axis that ranges from − pi
2 to
pi
2
E0x E0y
tan 2φ = 2 2 − E 2 cos δ with δ = δy − δx
E0x
(2.6)
0y
and |τ | the ellipse aperture or ellipticity that has the expression:

E0x E0y
|sin 2τ | = 2 2 + E 2 |sin δ| (2.7)
E0x 0y
Figure 2.2: Polarization ellipse characterization[3]
→
−
As time goes by the vector E (z0 , t) describes the elliptical shape. The ellipse
→
−
orientation is the angle of the vector E (z0 , t) with the x axis that evolves with
time ξ (t). The sign of the derivative of this magnitude determines the sense
of rotation, which, by convention, is evaluated when looking in the direction of
propagation.
2.2.2 Jones Vector

The plane wave polarization information is condensed in the complex Jones Vec-
tor that describes a polarization state and is defined from the complex electric
→
−
field vector E (z) as
E0x ejδx

→
− →
−
E = E (z) |z=0 = E (0) = (2.8)
E0y ejδy
that can be written in a matrix as a function of the parameters that characterize

the polarization ellipse:

+jα cos(φ) − sin(φ) cos(τ )
E = Ae (2.9)
sin(φ) cos(φ) j sin(τ )
where the α parameter is a constant phase term. In order to ease the calculations
with polarization vectors, the Jones vector can be rewritten by means of the
polarization algebra expressions, constructed from the unitary Pauli matrices
group.
10 10 01 0 −j
σ0 = σ = σ = σ = (2.10)
01 1 0 −1 2 10 3 j0
leading to the definition of three complex rotation matrices (SU(2)) of the special
unitary group whose expressions are:

cos (φ) − sin (φ)
U2 (φ) = = σ0 cos φ − jσ3 sin φ = e−jφσ3 (2.11a)
sin (φ) cos (φ)

cos (τ ) j sin (τ )
U2 (τ ) = = σ0 cos τ + jσ2 sin τ = ejτ σ2 (2.11b)
j sin (τ ) cos (τ )
jα
e 0
U2 (α) = = σ0 cos α + jσ1 sin α = ejασ1 (2.11c)
0 e−jα
obtained from the multiplicative properties of the quaternion formed by the
Pauli unitary matrices. The Jones vector in cartesian coordinates is thus ex-
pressed as a function of the special group matrices and a unit Jones vector
associated with the horizontal polarization x̂ = ûH :

cos(φ) − sin(φ) cos(τ )
E (x̂,ŷ) = Ae+jα = AU2 (φ) U2 (τ ) U2 (α) x̂ (2.12)
sin(φ) cos(φ) j sin(τ )
Two different polarization states are orthogonal if their Jones vectors E1 E2

are so, which means that their hermitian scalar product is equal to 0.
E1 |E2 = E T1 E ∗2 = 0

(2.13)
According to the definition of a Jones vector E (x̂,ŷ) given by 12, an orthog-

onal E (x̂,ŷ)⊥ Jones vector can be obtained in the same way, just changing the
unit Jones vector x̂ by ŷ.
E (x̂,ŷ)⊥ = AU2 (φ) U2 (τ ) U2 (α) ŷ (2.14)

where ŷ is the unit Jones vector associated to the vertical polarization state.
The orthogonality condition establishes that the ellipse parameters of E (x̂,ŷ) and
E (x̂,ŷ)⊥ satisfy:
π
φ⊥ = φ + τ⊥ = −τ α⊥ = −α (2.15)
2
An elliptical polarization basis is composed of the two unit Jones vectors u and
u⊥ where these vectors result from:
u = U2 (φ) U2 (τ ) U2 (α) x̂ and u⊥ = U2 (φ) U2 (τ ) U2 (α) ŷ (2.16)
Once the response of a scatterer has been acquired in a polarization basis, it

can be obtained in any other basis merely by means of a mathematical trans-
formation.
2.2.3 Stokes Vector

The polarization analysis by means of the Jones vector requires the use of a
coherent radar system, which provides with complex measurements (amplitude
and phase). The phase information is not available in those instruments that
can only measure observable power. Consequently, the polarization information
has to be described only by means of real terms with the so-called Stokes vector.
The value of the four Stokes parameters composing the vector, can be inferred
by means of the Pauli decomposition of an outer product of a Jones vector with
its conjugate transpose. The obtained expression for the Stokes vector g E is as
follows:
2 2 
Ex Ex∗ + Ey Ey∗
    
g0 |Ex | + |Ey |
g1   Ex Ex∗ − Ey Ey∗  E |Ex |2 − |Ey |2 
gE = g2  =  Ex Ey∗ + Ey Ex∗  =  2Re Ex E ∗ 
     (2.17)
y
j Ex Ey∗ − Ey Ex∗

g3 −2Im Ex Ey∗
with the folowing relation between the four Stokes parameters:
g02 = g12 + g22 + g32 (2.18)
where each of the parameters is:

g0 is the total power density of the wave
g1 is the power in the linear polarization, either horizontal or vertical
components,
g2 is the power in the linearly polarized components with a tilt angle of
ψ = 45, and
g3 is the power of the circularly polarized component of the plane wave
Provided the 4 Stokes parameters, obtained from power measurements, the am-
plitude and phase of a monochromatic electromagnetic wave can be determined,
hence its polarization is totally known. The Stokes vector can be thus rewritten
in terms of the polarization ellipse parameters as follows:
2 2
A2
     
g0 E0x + E0y
2 2  A2 cos (2φ) cos (2τ )
g1   E0x − E0y
gE = g2  = 2E0x E0y cos δ  =  A2 sin (2φ) cos (2τ ) 
     (2.19)
g3 2E0x E0y sin δ A2 sin (2τ )
2.2.4 Covariance Matrix

Since the radar is illuminating a distributed target, the received power mea-
surement are time-averaged samples of scattering from a set of different single
targets. Distributed targets are those which may fluctuate in time and space
dimensions due to the heterogenous nature of the target in these two dimen-
sions. Therefore, the scattered measured wave won’t be completely polarized
anymore, like the transmitted wave used to be, but just partially polarized. The
wave covariance matrix contains the information of the polarization state of the
received wave. It consists of the complex correlations of the corresponding Jones
vector E time-varying components.
D E hE E ∗ i
E E ∗
T∗
J= E·E =
x x∗
x y∗
Ey Ex Ey Ey

hJ xx i
J xy
=
J ∗yx J yy

(2.20)
D E D E D E D E
1 g + g1 g −j g
= D 0 E D E D2 E D 3 E
2 g2 + j g3 g0 − g1
The diagonal elements of the wave covariance matrix present the intensities,
whereas the off-diagonal elements are the complex cross-correlation between Ex
and Ey , and the trace of the matrix is the total energy of the wave. In the
case that the off-diagonal elements are 0, no correlation exists between Ex and
Ey and therefore the wave is completely depolarized. On the other hand, for
2
|J| = 0 it follows that hJxx i hJyy i = |hJxy i| , which means that the correlation
between the two Jones vector components is maximum. The general case lies
between these two extreme cases, where |J| > 0 indicates a certain statistical
2.3 Scattering operators and polarimetric characterization 16
dependence between Ex and Ey that can be written in terms of the wave degree
of polarization (DoP) as follows:
q
2 2 2 12
hg1 i + hg2 i + hg3 i

hJi
DoP = = 1−4 (2.21)
hg0 i T r (J)
where
DoP = 0 is fully depolarized wave and
DoP = 1 is fully polarized wave.
It is worth to point out that the wave covariance matrix elements depend on the
polarization basis of the Jones vector, and therefore to move the matrix from
one basis to another a special unitary similarity transformation is needed.
Wave Anisotropy and wave Entropy When computing the eigenvectors

and eigenvalues of the 2x2 Hermitian wave covariance matrix J the obtained di-
agonal form of the matrix can be interpreted as statistical independence between
a set of two wave components. The wave covariance is rewritten as:

λ 0
J = U2 1 U−1 T∗
2 = λ1 u1 u1 + λ2 u2 u2
T∗
(2.22)
0 λ2
where U2 is the 2x2 unitary matrix of the SU(2) group containing the two unit
orthogonal eigenvectors u1 and u2 , and λ1 ≥ λ2 ≥ 0 are the two non negative
real eigenvalues. Wave entropy and wave anisotropy can be defined now as
measures of the correlated wave structure of the wave covariance matrix J as
follows:
2
λ1 − λ2 X λi
Aw = Hw = − pi log2 pi with pi = (2.23)
λ1 + λ2 i=1
λ1 + λ2
Both the entropy (Hw ) and the anisotropy (Aw ) range from 0 to 1with:
For a completely polarized wave λ2 = 0 : Hw = 0 and Aw = 1
For a completely unpolarized wave λ2 = λ1 : Hw = 1 and Aw = 0
and for a partially polarized wave λ1 6= λ2 ≥ 0: 0 ≤ Hw ≤ 1 and 0 ≤

Aw ≤ 1
Note that the fact that the eigenvalues λ1 and λ2 are independent from the
polarization basis makes the entropy and the anisotropy two important basis-
independent measurements.
2.3 Scattering operators and polarimetric characteriza-

tion
2.3.1 Scattering coefficient
In radar imaging two different situations are faced by the radar system, depend-
ing on the relative size of the target with regards to the radar footprint.
if the target is smaller than the radar footprint, the power exchange be-
tween the wave and the scatterer is characterized by the radar cross-
section, which is defined using the radar equation as follows:
PT GT (θ, φ) AER (θ, φ)

PR = σ (2.24)
4πrT2 2
4πrR
where PR represents the power detected at the receiver, PT is the transmit-

ted power, GT the transmitting antenna gain, AER the effective aperture
of the receiving antenna, and rT and rR are respectively the distance be-
tween the transmitter and the target and the distance between the target
and the receiver. Finally the radar cross-section σ determines the effect of
the target in the balance of powers of the radar equation; it basically es-
tablishes the relation between the amount of incident power to the target
and the amount of power reflected by the target:
→
− 2

E S
σ = 4πr2 2 (2.25)
→
−
E I
The radar cross-section depends on significant number of parameters, such

as the wave frequency, the wave polarization, the target material and
shape, the measurement geometry...
In case that the radar footprint is smaller than the illuminated target, this
is interpreted as set of statistically identical point targets. Therefore, in
such a case, the total power received is the contribution of all these point
targets in the illuminated area, and thus the power is obtained integrating
in this A0 area:
ZZ
PT GT (θ, φ) 0 AER (θ, φ)
PR = σ ds (2.26)
A0 4πrT2 2
4πrR
where σ 0 is the averaged radar cross-section per unit area, also called the
scattering coefficient. It represents the ratio of the statistically averaged
scattered power density to the average incident power density over the
illuminated surface with

→− 2
E S
hσi 4πr2

σ0 = = → (2.27)
A0 A0 − 2

E I
2.3.2 Scattering matrix

As previously mentioned, the characterization of a scatterer also depends on the
polarization of the incident and the scattered waves. Hence, using p to denote
the incident wave polarization and q to denote the reflected wave polarization,
the scattering coefficient expression can be rewritten as:

→− 2
E Sq
hσi 4πr2

σ0 = = → (2.28)
A0 A0 − 2

E Ip
Although, the polarization is taken into account in the previous expression, this
is only an electromagnetic waves power based parameter, not exploiting the
vector nature of the electromagnetic waves. In order to use this vector nature,
the scattering process in the target of interest can be described, given the Jones
vectors of the scattered and incident waves, as follows:
e−jkr e−jkr S11 S12

ES = SE I = EI (2.29)
r r S21 S22
With regards to the incident and scattered orthogonal bases, the scattering
process can be described in a general form as:
R
Sφs φi Sφs θi EφI

Eφ
= (2.30)
EθR Sθs φi Sθs θi EθI
It has to be noted that in general, polarization unit vectors ûR R

θ and ûφ are
not orthogonal, therefore these two directions do not constitute the cross-pol
channel, although we treat them as so.
As well as the scattering coefficient, the definition of the scattering matrix
depends on the direction of the incident and the scattered waves. There is two
main conventions concerning the polarimetric scattering definition: ”forward
scatter alignment” (FSA) and ”backscatter alignment” (BSA). FSA is defined
regarding the propagating wave, and it is used for bistatic situations, whereas
the BSA is defined regarding the antennas, and it is very useful for monostatic
configurations. Therefore, the convention used leads to different S matrix for-
mulations, that can be switched between each other by means of transformation
matrices.
The total scattered power in a polarimetric system can be obtained from
the scattering matrix as the so-called Span, which is computed the scattering
matrix Frobenious norm as follows for the bistatic backscattering case:
2 2 2 2
Span = T r S S∗T = |S11 | + |S12 | + |S21 | + |S22 |

(2.31)
In the monostatic backscattering case, the Span expression is as follows:
2 2 2
Span = T r S S∗T = |S11 | + 2 |S12 | + |S22 |

(2.32)
The information contained in the scattering matrix S can be written in vectorial
format by means of the vector V() which is defined as follows:

S SXY 1
S = XX ⇒ K = V (S) = T r (SΨ) (2.33)
SY X SY Y 2
where Ψ is a complete 2x2 complex basis matrices. The scattering target vector
depends on the basis matrices set chosen and in the nature of the geometry of
the measurement, i.e, either it is bistatic or monostatic. Two main matrices
sets are used for the definition of scattering vectors:
Pauli spin matrices basis set {Ψp }
√ 1 0 √ 1 0 √ 0 1 √ 0

−j
{Ψp } = 2 2 2 2 (2.34)
0 1 0 −1 1 0 j 0
that results in the K scattering vector:
1 T
k = √ SXX + SY Y SXX − SY Y SXY + SY X j (SXY − SY X )
2
(2.35)
for the bistatic case, whereas in the monostatic case, the scattering matrix
is simetrical due to the reciprocity given by the geometry of the problem
(SXY = SY X ) and therefore, since the Pauli spin matrix basis has only
three matrices now, the resulting scattering k vector is:
1 T
k = √ SXX + SY Y SXX − SY Y 2SXY (2.36)
2
Lexicographic matrix basis set {ΨL }

1 0 0 1 0 0 0 0
{Ψp } = 2 2 2 2 (2.37)
0 0 0 0 1 0 0 1
that results in the Ω scattering vector:
1 T
Ω = √ SXX SXY SY X SY Y (2.38)
2
that corresponds to the bistatic case. Regarding the monostatic case, with
the same idea used in the k vector case, the Ω vector for the monostatic
backscattering case is:
1 √ T
Ω = √ SXX 2SXY SY Y (2.39)
2
These two different scattering vectors are related to each other by means of a
transformation matrix, which is constrained to be unitary due to the fact that
the Span of the scattering matrix has to be the same with independence of the
basis chosen. Hereafter the transformation expressions for both the bistatic an
the monostatic cases.
k = U4(L→P ) Ω and k = U3(L→P ) Ω (2.40)
where the expressions of the transformation matrices escape to the scope of this
polarimetry introduction note.
2.3.3 Polarimetric Coherency and Covariance Matrices

Even under the assumption of distributed targets, which are subject to vari-
ations in time and space, some assumptions on stationarity and homogeneity
are needed. This can be studied introducing the concept of space and time
varying stochastic processes to describe the environment where the target is.
Information about the target is extracted from the second order moment of the
fluctuations of this varying processes, which are extracted from the polarimetry
Coherency and Covariance matrices.
From the scattering vectors introduced in the previous section and depending
again on the geometry case of the polarimetric radar system, Coherency and
Covarinace matrices are obtained respectively from the outer product of k and
Ω vectors as follows:
D E D E
T = k · k∗T C = Ω · Ω∗T (2.41)
For the bistatic backscattering configuration both matrices are 4x4, and their
expressions are:
2 ∗ ∗ ∗
* |k1 | k1 k2 k1 k3 k1 k4 +

D E 2
k2 k1∗ |k2 | k2 k3∗ k2 k4∗ 
T4 = k · k∗T =  k3 k ∗ k3 k ∗ |k3 |2 k3 k ∗ 
 (2.42)
1 2 4
2
k4 k1∗ k4 k2∗ k4 k3∗ |k4 |
2
* |Ω1 | Ω1 Ω∗2 Ω1 Ω∗3 Ω1 Ω∗4 +
 
2
Ω2 Ω∗1 |Ω2 | Ω2 Ω∗3 Ω2 Ω∗4 
D E
C4 = k · k∗T = 

Ω3 Ω∗ 2
 (2.43)
1 Ω3 Ω∗2 |Ω3 | Ω3 Ω∗ 
4
2
Ω4 Ω∗1 Ω4 Ω∗2 Ω4 Ω∗3 |Ω4 |
Regarding the monostatic case, the expressions are analogous, but using the
dimension 3 k and Ω scattering vectors, therefore the obtained Coherency and
Covariance matrices are 3x3.
Both Coherency and Covariance matrices are by construction Hermitian
semidefinite positive matrices with real nonnegative eigenvalues and orthogonal
eigenvectors. The transformation matrices relating the two scattering vectors
can be used to derive one matrix from the other one, as in the following example:
D E D E
∗T
T = k · k∗T = (UL→P Ω) · (UL→P Ω)
D E
= U(L→P ) Ω · Ω∗T U∗T
(L→P )
= U(L→P ) CU−1
(L→P ) (2.44)
These two matrices are in general terms the starting point for the polarimetric
analysis of a target. They contain information of the target themselves, and
they are used in the polarimetric decomposition methods of the targets
These two matrices can be diagonalized calculating their eigenvectors and
eigenvalues, which as stated before are nonegative values. The expressions of
the decomposed matrices are:
T = UP ΣP U−1
P and C = UC ΣC U−1
C (2.45)
where ΣP and ΣC are diagonal matrices containing the eigenvalues and UP and
UC are the unitary matrices with the three unit orthogonal eigenvectors. The
value of the eigenvalues provide information about the scattering mechanism,
precisely the level of polarization of the scatterers.
2.3.4 Kennaugh Matrix. Huynen Decomposition

Target representation in terms of power allows evaluating a physical event in
different ways, as the result of different measurements, phase coherency is not
anymore needed. Those parameters obtained from power measurements become
incoherently additive parameters concerning the phase. The Kennaugh matrix
establishes the same kind of relation between the incident and reflected Stokes
vector, than the S matrix does with the incident and reflected Jones vectors. It
is obtained as follows:
K = A∗ (S ⊗ S ∗ ) A−1 (2.46)
and the incident and reflected Stokes vector are related with the Kennaugh
matrix as:
gER = KgEI (2.47)
The K matrix can be written as a combination of 9 different parameters
known as Huynen parameters, each of them containing physical interpretation
of the target. Again, depending on whether the radar system is monostatic or
bistatic case, the dimension of the K matrix is 3x3 or 4x4 respectively. These
above presented matrix polarimetric operators, such as the scattering matrix

S and its derived coherency and covariance matrices or the Kennaugh matrix,
can be moved from one polarization basis to another by means of basis change
matrices. Therefore, a polarimetric matrix operator resulting from a determined
polarization basis measurements, can be analyzed under any other polarization
basis just by means of an algebraic transformation.
The received power in a radar system can vary depending on the polar-
ization states of the transmitting and receiving antennas, leading to different
measurements and thus target information sources. The received power can be
expressed in terms of either the scattering matrix S as
2
PTR ∝ ĥTR S ĥT (2.48)

or in an equivalent way in terms of the Kennaugh matrix, in such a way that

two received power measurements can be defined:
Copolarized power: The transmitting and receiving antennas have the
same polarization states (ĥR = ĥT )
Crospolarized power: In this case the transmitting and receiving antennas
have orthogonal polarization states (ĥR = ĥT⊥ )
The optimization polarization problem consists of finding such polarization

states of the transmitted and received waves for a known target that maximizes
or minimizes for the different copol or crosspol measurements. The polarization
states obtained out of this optimization process are the so-called target char-
acteristic polarization states. The diagonalization of the Sinclair S matrix, for
instance, is a method to look for the polarization states that null the cross-polar
power. These will be used for further target characterization.
2.3.5 Canonical Scattering mechanisms

Common targets present always a complex scattering mechanism due to their
complex geometrical structure and heterogenous reflectivities signatures. How-
ever a set of canonical scatterers can be defined with well known scattering
mechanisms characterized by its associated Scattering matrix S. These will
be further used as a reference in the interpretation of real targets scattering
mechanisms.
The definition of these set of canonical scattering mechanisms depends on
the polarization basis used to describe the scattering mechanism. It can be
expressed in the three canonical orthogonal polarizations:
Cartesian polarization basis (ĥ, v̂), linear horizontal an vertical polariza-

tion
Linear rotated basis, the 45 rotated liner polarization
Circular polarization basis
The canonical scattering mechanisms or scatterers are: the sphere or flat plate
or trihedral which are equivalent, the horizontal dipole, the oriented dipole, the
dihedral, the right helix and the left helix.
2.4 Polarimetric SAR Speckle Noise 22
2.4 Polarimetric SAR Speckle Noise

2.4.1 Speckle Noise Introduction
Speckle is a multiplicative noise, the more signal power the more noise power,
inherent in SAR images, due to the phase incoherent received waves coming
from the different scatterers. The received waves from each scatterer are no
longer coherent in phase. These multiple wavelets can either add constructively,
leading to a strong signal, or add out of phase, leading to a weak signal. The
speckle effect in the SAR image is the pixel to pixel signal intensity variation.
The addition of the returned complex wavelets from the elemental scatterers
can be written as:
M
X
(xi + jyi ) = x + jy (2.49)
i=1
where xi + jyi is the returned wave from the i th scatterer. This complex plane
vector addition is clearly depicted in the next figure:
Figure 2.3: Single scatterers contribution to the received signal [3]
The real and imaginary components of each return are random variables
statistically independent, and Gaussian distributed with zero mean and variance
σ 2 /2. Therefore the joint probability density function (PDF) is the result of
the product of the two independent
p gaussian random magnitudes, leading to a
received amplitude A = x2 + y 2 with a Rayleigh PDF as follows:
1 2 2
px,y (x, y) = px (x) py (y) = √ e−(x +y2)/σ (2.50)
πσ
2.5 Polarimetric decomposition theorems 23
the same PDF can be written as a function of the complex magnitude amplitude
A and argument θ and integrated over a θ interval of [−π, π] to obtain the PDF
for the amplitude:
2A 2 2
p (A) = 2 e−(A /σ ) A ≥ 0 (2.51)
σ
√
which is a Rayleigh distribution with mean σ π/2 and variance (4 − π) σ 2 /4.
Besides the amplitude of the received signal, the signal intensity I can be defined
as the square of the amplitude I = x2 + y 2 , having a negative exponential PDF
.
1 I
p (I) = 2 e− σ2 I ≥ 0 (2.52)
σ
The classical method to reduce the speckle noise is averaging several inde-
pendent reflectivity estimations, moving from single-look to multi-look images.
This is done dividing the synthetic aperture in N smaller sub-apertures which
are processed independently and averaged a posteriori as N independent sam-
ples. The result of this√ technique is the reduction of the standard deviation
of the speckle by a 1/ N factor, at expense of reducing the azimuth resolu-
tion by a factor of 1/N . The averaged multi-look amplitude, resulting from the
convolution of the N single-look Rayleigh PDFs has a Chi-squared distribution
which is narrower than the original single-look distribution. Another speckle
reduction technique is the averaging of neighboring pixels, that also worsens the
resolution.
For heterogeneous mediums, the Rayleigh model for the amplitude of the
received signal often fails. It is then used k-distribution model which derives
to a Rayleigh as the medium becomes homogeneous. The k-distribution model
is made up as the product of a gamma distribution, accounting for the texture
medium property, and a Rayleigh distribution. For homogeneous mediums the
variance of the gamma distribution decreases, with mean value equal to 1.
2.4.2 Polarimetric Speckle Statistics

Polarimetric SAR data from reciprocal media (monostatic configuration) can be
interpreted as the result of the interactions of three correlated processes between
HH,VV and HV(VH) polarization channels. Therefore, the speckle statistics are
different than for the single polarization case. It can be proven that a multi-look
covariance matrix has a Wishart distribution, and the PDFs of polarimetric
SAR data are derived from this last assumption. The speckle statistics for
polarimetric SAR images, either multi-look or single-look, are affected by the
fact that different polarization mechanisms are correlated with each other, thus
the PDFs obtaining is not trivial and must be studied individually for each case.
As a very simplified idea, one can state that single-look polarimetric images have
complex gaussians PDF, whereas multi-look polarimetric images have complex
Wishart PDF.
Knowing the speckle noise statistics enables the development of effective
algorithms for speckle reduction, which are useful in further image segmentation
and target estimation/classification applications.
2.5 Polarimetric decomposition theorems

In the previous section, the importance of facing the speckle phenomena prior to
the interpretation of the polarimetric data has been introduced. After speckle
filtering or multi-llok averaging the Coherency T and Covariance C matrices are
said to be distributed or incoherent, and cannot be related anymore to a coher-

ent scattering matrix S. The target decomposition theorems aim to analyze the
polarimetric properties from this incoherent polarimetric representation. Co-
herency matrix is the most important observable of these radar systems since it
is the lowest operator suitable to extract polarimetric parameters for distributed
targets in presence of additive and multiplicative noise from the system and the
speckle respectively.
Many targets require a multivariate statistical description due to speckle
noise and random scattering effects, so called distributed targets. It arises then
the concept of average or dominant scattering mechanism for distributed targets
in order to interpret the scattering data. The target decomposition theorems
provide such interpretation of the distributed target scattering mechanism. The
degree of correlation between the coherency matrix elements might be related
to the nature of the dominant scattering mechanism.
The information extracted from the decomposition methods is the input
for the target classification techniques that aim to identify the polarization
scattering mechanisms. These methods are beyond the scope of this document,
being the highest level of processing before obtaining a polarimetric classified
image.
The decomposition theorems can be classified in four types as follows:
Based on the dichotomy of the Kennaugh matrix (ex: Huynen decompo-
sition)
Based on a model-based decomposition of the coherency or covariance
matrices (ex: Freeman and Durden decomposition)
Based on the analysis of the eigenvectors and eigenvalues decomposition
of the coherency or covariance matrix (ex: Cloude and Pottier)
Based on a coherent decomposition of the scattering matrix S (ex: Pauli)
In the following subsections the mentioned examples will be briefly introduced,
describing the main idea of each polarimetric decomposition theorem.
2.5.1 Huynen decomposition

The Huynen decomposition theorem exploits the fact that a target is an object,
always the same, and independent of aspects like radar system geometry, radar
frequency, EM waves polarization state,... and therefore is possible to extract
physical properties of the target. For the case of a single stationary target the
definition of the 9 Huynen parameters of the Kennaugh matrix, relates each
parameter with a physical feature of the target. In case of distributed targets
a statistical averaging process is required, and the interpretation of the Huy-
nen parameters is also different. The main idea of the Huynen decomposition
theorem is to separate from the data a part identified with a single averaged
target and a residual ”N-target”. An important advantage of this decomposition
theorem is that is model independent.
A single stattionary target coherency matrix has 5 independent components,
the diagonal and 2 off-diagonal elements, therefore it has 5 degrees of freedom.
Whereas in the case of a distributed target, its averaged coherency matrix has
lost the dependency relations between components, and therefore it has 9 inde-
pendent elements.
According with the main idea of this decomposition theorem, these 9 inde-
pendent parameters can be interpreted as:
an effective single target T0 , dominant scattering mechanism, which is

described by 5 of the 9 parameters and
a residual target TN described by the remaining 4 parameters.
The information extracted from the parameters is about the total scattered
power from the regular and the depolarized components, the scattered power
from the symmetric and the nonsymetric components and the coupling between
symmetric and nonsymmetric.
2.5.2 Freeman and Durden decomposition

This decomposition method consists of a technique to fit a physical model com-
posed of 3 scattering mechanism components to the polarimetric SAR data.The
three components of the model are a Bragg scatterer from a relatively rough
surface, a double bounce scatterer from a dihedral like structure and a cannopy
scatterer from a cloud of randomly oriented dipols.
For the Bragg scatterer modeling the surface, the cross-polarized compo-
nents are negligible, and therefore the scattering matrix of these compo-
nent is:
RH 0
S= (2.53)
0 RV
This Sinclair matrix leads to a scattering covariance matrix for this com-
ponent called C3S
The double bounce scattering component is determined by a dihedral scat-

terer with a scattering matrix as follows:
2jγ
e H RT H RGH 0
S= (2.54)
0 e2jγV RT V RGV
where the subindex T stands for the Trunk of the dihedral like structure
and G stands for the ground. Notice that the Trunk and Ground different
reflection coefficients are accounted in the scattering matrix expression, as
well as the propagation factors included by the exponentials accounting
for possible attenuation and phase change effects. As well as in the surface
case, this Sinclair matrix leads to a covariance matrix for this component
C3D .
the volume scattering component, generally facing forestry targets, comes
from a cloud of dipoles and its scattering matrix depends on the rotation
angle θ of the horizontal dipole around the radar line of sight. It leads to
the covariance matrix for this volume scattering component hC3V iθ
The advantage of using this decomposition method is that it is based on a
physical model, and not only mathematical. However two strong limitations
reduce its usage: the three components used in the model apply only for a
determined kind of scenes, and the reflection symmetry assumption is not always
∗
fulfilled. Reflection symmetry hSHH SHV i = hSHV SV∗ V i = 0 assumes symmetry
in a plane normal to the line of sight. This is: if there is contribution of a
scatterer P in one side of the plane, there is an exact contribution of a scatterer Q
at the same position at the opposite side of the plane. Some reflection scenarios
do not have this property.
2.5.3 Eigenvector-based decomposition

This is a basis invariant method which represents an important advantage with
respect to the rest of decomposition methods. This decomposition is based
on the eigenvalues of the 3x3 Hermitian averaged coherency T3 matrix. The
eigenvalues and eigenvectors of the averaged T3 matrix are computed in order
to obtain a diagonal form of the matrix that can be physically interpreted as
the contribution of a set of target vectors statistically independent,
T3 = U3 Σ U−1
3 (2.55)
where Σ is a diagonal matrix with the non negative eigenvalues (λ1 ≥ λ2 ≥ λ3 ≥ 0)
and U3 is a unitary matrix of the SU(3) group with the three orthogonal eigen-
vectors U3 = [u1 u2 u3 ].
T3 is therefore expanded into the sum of 3 independent targets {T0i }i=1,2,3
each of which represents an independent scattering mechanism with its equiv-
alent scattering matrix S. The contribution of each scattering mechanism de-
pends on the value of its associated eigenvalue (the biggest eigenvalue determines
the dominant scattering system).
3
X
T3 = λi ui u∗T
3 = T01 + T02 + T03 (2.56)
i=1
2.5.4 Pauli decomposition. Coherent Decomposition

The coherent based decomposition methods aim to express the received scatter-
ing matrix S as the combination of basis matrices corresponding to canonical
scattering mechanisms:
N
X
S= αi Si (2.57)
i=1
The main limitation in the usage of these theorems is that they ignore the effect
of Speckle noise, that as seen in previous section, in a single-look data image
can cause considerable distortion in the visual interpretation. The speckle fil-
tering through data averaging leads to the non-coherency of the SAR data, and
therefore these methods canot be used anymore. However, the coherent de-
composition methods are useful for the case of a strong dominant scattering
mechanism after an eigenvector-based decomposition. The coherent decomposi-
tion can be applied to the obtained dominant scattering system. Therefore this
decomposition makes sense when the scene suggests a very dominant scattering
mechanism such as a dihedral component in a urban environment.
The Pauli coherent decomposition proposes the expression of the scattering
matrix as the sum of the Pauli matrices, with an elementary scattering mecha-
nism is associated with each basis matrix,

S SHV a 1 0 b 1 0 c 0 1 d 0 −j
S = HH =√ +√ +√ +√
SV H SV V 2 0 1 2 0 −1 2 1 0 2 j 0
(2.58)
from the previous equation the expressions of a, b, c, and d are extracted:
SHH + SV V SHH − SV V SHV + SV H SHV − SV H
a= √ b= √ c= √ d= √ (2.59)
2 2 2 2
Notice that for a monostatic configuration where SHV = SV H , the Pauli matrix
basis can be reduced to three since d = 0.These remaining matrix basis represent
2.6 SAR polarimetry Applications 27
three single scattering mechanisms: the first is the single scattering from a
plane surface, the second and the third are double-bounce scatterings from
dihedral structures with relative orientation 0 and 45 respectively. The Pauli
decomposition associates a color to each of the components of the Span of the
scattering matrix that is written as follows:
2 2 2 2 2 2
Span = |SHH | + 2 |SHV | + |SV V | = |a| + |b| + |c| (2.60)
2 2 2
Being |a| red, |b| green and |c| blue, leads to a decomposition result like the
one shown in the following figure:
Figure 2.4: Polarimetric Pauli decomposition (Great Aletsch glacier 2003,

Switzerland, E-SAR/L-band)
2.6 SAR polarimetry Applications

The set of applications of SAR polarimetry is naturally shared in most part
with SAR applications. In the last decade, polarimetric SAR systems have
been proven to be very capable in earth observation applications. In many of
them, they seem to be more powerful than classical single polarization SAR or
optical sensors.
The subsequent list highlight the most typical polarimetric SAR applica-
tions:
Terrain and Land-Use classification
Forest areas mapping and classification. Information such as biomass,

height or vegetation type can be estimated
2.6 SAR polarimetry Applications 28
Snow covered areas monitoring, allowing the retrieving of wetness and

thickness
Surface geophysical parameters estimation such as soil moisture and rough-

ness
Ice covered areas monitoring
Ocean monitoring, estimating parameters like the SWH or currents.
Damage assessment, in for example forecasting floods from frozen rivers.
Finally, an emerging technique has to be mentioned which is the polarimetric

SAR interferometry (PolInSAR), has been proved very promising in forest clas-
sification. PolInSAR combines both polarimetry and interferometry techniques
to lead to the interferometric coherency matrix that contains polarimetric prop-
erties of each individual image of the interferometer, as well as polarimetric and
interferometric correlation between the two images.
29
3 SAR Polarimetry for Snow Remote Sensing

3.1 Introduction
The remote sensing for snow monitoring has been a research topic for the last
three decades. According to [1], monitoring the amount and distribution of
snow, helps in the climate study, improves the weather forecasts, as well as
water supply and flooding predictions for hydrology management. The impor-
tance of the estimation of snow parameters, such as the Snow Water Equivalent
(SWE), which gives an idea of the amount of water contained in the snow-
pack, in hydrology and climate studies, has pushed the scientific community
to explore the possibilities of the Earth Observation techniques for the snow
quantitative estimation. The recent availability of modern space-borne SAR
sensors with polarimetric capabilities has opened a new research field, grouping
together the SAR advantages of good resolution and weather independence and
the polarimetry sensitivity to the features of distributed targets such as snow.
Therefore, polarimetric SAR (PolSAR) is envisioned to contribute significantly
to the quantitative remote estimation of snow.
The current section will present the state-of-the-art of the remote quantita-
tive estimation of snow with PolSAR, as well as a brief introduction about the
SWE parameter and about the interaction of electromagnetic waves with snow.
After the introduction, Section 3.2 will be devoted to the interaction of snow and
electromagnetic waves, as the physical basis to describe the remote estimation
of snow parameters based on microwave remote sensing techniques. Sections
3.3 and 3.4 will describe the remote sensing techniques based on coherent and
incoherent PolSAR data respectively, and finally,Section 3.5 for conclusions and
3.1 Introduction 30
future research steps will close the section.
3.1.1 Snow Water Equivalent

Snow Water Equivalent (SWE) is a common snow-pack measurement. It is the
amount of water contained within the snow-pack. It can be thought of as the
quantity of water that would theoretically result if you melted the entire snow-
pack instantaneously. This parameter is further used in hydrology and climate
studies for the forecasting of water resources, natural diseases such as flooding,
and water resources management in the scientific field, but it is also important
for other activities such as hydro-power generation.
SWE can be computed as the snow depth times the snow volumetric density,
according to the following equation:

Kg
SW E = d · ρ (3.1)
m2
where d is snow depth in meters and ρ is snow density in Kg per cubic meter.
Therefore, in order to estimate the SWE of a snow-pack, two parameters have
to be estimated: the snow depth and the snow density in order to obtain the
SWE estimation as the product of them.
The snow water equivalent of a snow-pack is one of the most important
measurements of a snow-pack from a hydrological standpoint because it is a
direct measure of the liquid water content of a snow-pack. When some or all
of this water is released during the seasonal melt, it becomes an important
component of local surface water and groundwater budgets. In some areas of
the world, snow melt provides the majority of annual groundwater recharge,
so monitoring of snow water equivalent of snow-packs is an important part
of water resources management. However, SWE measurement is tedious and
costly, since it consists of melting a known volume of snow in order to weight the
water contained in it, in field measurement campaigns, that cannot be carried
out in vast areas. Therefore SWE cannot be extensively studied for large scale
applications which are the ones of scientific interest. From this need of mapping
in higher spatial and frequency scales the SWE parameter, arises the interest of
the remote sensing community in the estimation of this snow-pack parameter.
3.1.2 Snow Remote Sensing

Remote sensing of snow can be classified in two main groups, depending on
the nature of the resulting retrieved information: either it can be qualitative,
i.e. binary information about the presence or not of snow, either it is quantita-
tive, which means that the retrieved information is a magnitude, a quantitative
value. The qualitative remote sensing of snow is known as Snow Covered Area
(SCA) and though it is a valuable information it doesn’t provide quantitative
estimations or models. The retrieval of SWE is a quantitative estimation of the
amount of water contained in the snow-pack per unit area in Kg/m2.
In the last decades, snow remote sensing has been carried out with most of
the main earth observation techniques; passive and active:
Optical sensors offer good resolutions, however their nature does not al-
low them to work during night or during no visibility conditions due to
meteorological effects, and in addition, at optical frequency ranges the
penetration phenomena is almost null. Therefore optical sensors can only
3.1 Introduction 31
be used for qualitative measurements under very specific visibility condi-

tions.
Passive microwave sensors, so called radiometers, have a good sensitivity
to the snow parameters and are meteorology independent, but unfortu-
nately, they have a very poor spatial resolution, which is not suitable for
most of the snow remote sensing applications [1].
Finally, microwave active sensors, Synthetic Aperture Radars (SAR), are
meteorology independent, have a good spatial resolution and work within
a frequency range that has a good trade-off between sensitivity and pen-
etration of the snow-pack. The literature about snow remote sensing is
therefore mostly focused on SAR based techniques given its proven appro-
priateness. The recent advances involve the multichannel SAR sensors,
that exploit diversity of sources of information in order to have more than
one channel of information about the observed target. Polarimetric SAR,
which is going to be the scope of this section, is included in this group, and
it is based on using the polarization information of the electromagnetic
waves in order to achieve the multichannel diversity.
In SAR remote sensing, two main groups of targets can be identified: the
point targets and the distributed targets. The point targets are those whose
size fits inside the resolution cell of the radar, whereas distributed targets are
those presenting a larger extent than the radar resolution cell, and to describe
them they are interpreted as an infinite set of identical statistically independent
point targets. Snow is a distributed target, and as so it has to be analyzed with
statistical techniques.
Besides PolSAR snow remote sensing, which is the scope of this Master
Thesis, it has to be mentioned an important contribution in SAR qualitative
remote sensing: the works of Nagler and Rott [23]. They developed a method for
the retrieve of wet snow covered area in alpine regions based on the comparison
of two SAR C-band images of the same scene with wet snow and without snow
(or with dry snow). In their research they found out that the backscattering
intensity of the wet snow covered scene was significantly lower than the one
with no snow (or dry snow), setting a threshold of 3dB of difference for the
identification of wet snow covered pixels. This reference [23] has been the basis
for the development of some operational services to determine Snow Covered
Area (SCA).
PolSAR data comes from a SAR sensor which is sensitive to the polarization
state of the electromagnetic waves. It transmits and receives EM waves with
two orthogonal polarization states. This means that a PolSAR sensor is able to
retrieve the complete scattering matrix of a target for a given geometry and fre-
quency configuration, fully characterizing a target from a polarimetric point of
view. The scattering matrix describes the relation between the incident and the
scattered fields of a target at two orthogonal polarization states, indicating how
the polarization of a wave changes when impinging the target. The scattering
matrix is fully described by four complex magnitudes, that provide information
about the target geometry and dielectric properties of the target [3, 4].
e−jkr Sxx Sxy

ES = EI (3.2)
r Syx Syy
In Section 3.2 a detailed description of the physical basis of the microwaves
interaction with snow will be presented, describing the snow-pack features to
3.1 Introduction 32
which polarimetric SAR sensors can be sensitive.
3.1.3 Polarimetric SAR Sensors

In the history of polarimetric SAR sensors, as for basically any EO instruments,
two different evolutions have to be studied: the airborne sensors and the space-
borne sensors. While on the airborne side there have been fully polarimetric
SAR sensors since the eighties, it was not until 2006 when the first dedicated
Polarimetric SAR sensor payload was launched in the ALOS mission. Hereafter,
the available space-borne polarimetric SARs will be detailed, whereas the air-
borne ones will be solely commented since for the scope of snow monitoring at
large areas scale and on a regular basis, space-borne sensors make more sense,
due to their global coverage, repetition passes and availability in a regular mode
of operation.
Concerning airborne polarimetric SAR sensors, main research centers and/or
institutions around the globe have their own airborne SAR system. The main
fully polarimetric ones are: AIRSAR from JPL(NASA) P,L and C bands, ESAR
from DLR P,L and S bands, RAMSES from ONEREA P, L, S, C, Ku, Ka, W
bands and EMISAR from Danish Center for Remote Sensing (DCRS) L and C
bands. Airborne polarimetric SAR sensors are very useful for research purposes
or for very specific unusual applications,as well as for punctual observations.
Regarding the PolSAR space-borne sensors, their main advantages are the
global and periodic coverage together with very good spatial resolution, that
make possible to use them for stable and regular remote sensing services, that
can eventually become commercial products. Since polarimetric SAR is quite
a new technique, few space-borne facilities are still available. The space-borne
polarimetric SAR systems are listed below:
SIR-C/X SAR
The Shuttle imaging radar-C and X-band synthetic aperture radar, in
Figure 3.1, was a cooperative space shuttle experiment between NASA,
German Space Agency (DARA) and Italian Space Agency (ASI). This was
a SAR instrument mounted aboard NASAs space Shuttle that flew two
campaigns in 1994. It was the first fully polarimetric space-borne SAR,
that led to the obtaining of the first quad-pol image data sets from space
at L and C bands.
Figure 3.1: SIR-C/X mission

3.1 Introduction 33
ENVISAT ASAR
ESAs ENVISAT launched in 2002, in Figure 3.2, carries the Advanced
Synthetic Aperture Radar (ASAR), in addition to other remote-sensing
instruments. ASAR instrument operates at C-band offering very good ca-
pabilities in terms of coverage, range incidence angles, modes of operation
and polarization. Though it does not consist of a fully-polarimetric in-
strument, it provides partially polarimetric products comprising two non-
coherent images of the same scene in the selectable polarization combina-
tion: HH/VV, HH/HV or VV/VH.
Figure 3.2: ENVISAT ASAR
ALOS-PALSAR
The Japanese Aerospace Exploration Agency (JAXA) Advanced Land
Observing Satellite (ALOS), in Figure 3.3, is specially designed for land
remote sensing, with applications such as land coverage observation, dis-
aster monitoring and resource surveying. Launched in 2006, it carries
the phased array type L-band SAR (PALSAR), whose polarimetric ex-
perimental mode provides L-band full polarized images of swath 20-65Km
wide with a resolution of 24-89m. In partially polarization mode it can
achieve improved images with a resolution up to 14 m.
Figure 3.3: ALOS PALSAR
RADARSAT-2
This Canadian Space Agency (CSA) satellite mission matches the main
CSA challenges of monitoring the environment and managing the natural
resources. The RADARSAT satellites are a major data source for remote
sensing scientific as well as commercial applications. RADARSAT-2, in
Figure 3.4, was launched in December 2007 as a public-private partnership
between CSA and McDonald, Dettwiler and Associates Ltd. (MDA). It
3.1 Introduction 34
operates at C-band fully polarimetric and can achieve a very high resolu-
tion imaging of 8m and full flexibility in the selection of the polarimetric
configuration. It consists of the first instrument with quad-pol capabili-
ties providing full polarimetric products in normal configuration modes,
being a valuable tool in the remote sensing of geophysical properties of
the earths surface.
Figure 3.4: RADARSAT-2
TerraSAR-X - TanDEM-X
The german radar satellite TerraSAR-X, in Figure 3.5, was launched in
June 2007 as a public-private partnership between German Ministry of Ed-
ucation and Science, the German Aerospace Center (DLR) and EADS As-
trium GmbH with an estimated life-cycle of 5 years. TerraSAR-X satellite
mission’s objectives are the provision of high quality multi-mode X-band
SAR data for scientific applications as well as establishing a commercial
EO market to develop a sustainable EO services business. In normal oper-
ation modes, TerraSAR-X can provide dual polarized data, and addition-
ally quad polarized data is available in experimental mode. In ScanSAR
operation mode, it provides 100 km wide swath images with a resolu-
tion of 16m. TanDEM-X opens a new era in space borne radar remote
sensing. The first bi-static SAR mission, is formed by adding a second,
almost identical spacecraft, to TerraSAR-X and flying the two satellites in
a closely controlled formation, with the primary mission of generating a
consistent global digital elevation model with an unprecedented accuracy.
The flight formation of two satellites with polarimetric capabilities allows
a set o polarimetric configurations in the acquired images.
Future Instruments
Since SAR polarimetry has been revealed as a promising remote sensing
technique, there will be further polarimetric SAR space-borne instruments
in the near future. An example is the CoreH2O mission of ESA’s Earth Ex-
plorer program, for cold regions hydrology observations, which is currently
in ESA’s phase A. This mission envisions a twin-frequency dual-polarized
SAR instrument specially developed for the forecasting of water supply
from snow cover and glaciers and, in general, for the modeling of water
and energy cycles at high latitudes. ESA’s Sentinel-1 will be the succes-
sor of ENVISAT’s ASAR as a C-band dual-polarization SAR, and is to
3.2 Snow interaction with EM waves 35
Figure 3.5: TerraSAR-X
be launched before the end of ENVISAT activities. In Canadian Space

Agency, the RADARSAT-2 follower is already planned: it will consists of
a constellation of 3 fully polarimetric SARs called Radarsat Constellation
Mission (RCM).
3.2 Snow interaction with EM waves

In radar distributed targets, the scattering coefficient σ 0 , characterizes the scat-
tering radiation being imaged by the radar. The value of this scattering coef-
ficient depends on one side of a set of parameters that belong to the imaging
system: wave frequency, wave polarization and image configuration (incident
and scattered directions) and on the other side of a set of parameters that de-
pend on the target, such as the geometrical or the dielectric properties. Remote
sensing techniques aim to extract information about the imaged target out of the
retrieved scattered electromagnetic waves. The idea is to have a good modeled
environment so that the influence of the target specific factors can be extracted
from the gathered scattered power, and therefore information about the target
can be retrieved.
In the case of the snow-pack, according to what is immersed in [2] and
[1], snow is considerably well modeled by a cloud of ice particles in an air back-
ground. However, the interfaces air-snow and snow-ground add some complexity
to the method to infer the backscattered power, since it results in a combination
of several scattering phenomena.
The back-scattering response of the snow-pack is the result of the contri-
bution of different scattering mechanisms: underlying ground surface back-
scattering, volume scattering from the snow layer, air-snow interface back-
scattering, and multiple contributions from double and triple bounce between
the snow-pack and the soil surface. In terms of microwave interactions with the
snow-pack, the driving electromagnetic parameters are the relative dielectric
constants of ice and liquid water and their geometrical distribution.
3.2.1 Snow physical modeling

Snow back-scattering depends on its features, that are determined by the main
parameters: density, which can be related to the dielectric constant, the water
content, the temperature, which can be related to the liquid water content, and
the crystalline structure of the ice particles present in the snow-pack. Snow is
composed of three elements: ice, air and water, which can be liquid or vapor
depending on the kind of snow, and the variation of these three component
percentages, determines the snow features.
The parameters that determine the nature of each kind of snow are intro-
duced as follows:
Water Content (liquid): Determines the dielectric properties and it is very

attached to the temperature. Snow is considered dry if there is 0% of liquid
water content, and as the temperature raises, the amount of liquid water
content increases leading to the so called wet snow [2].
Density: Snow density is defined as the ratio of snow mass over the water
reference mass. It’s the most important parameter influencing the dry
snow backscattering power. Snow density can be directly related with the
dielectric constant with the semi-empirical Looyenga’s formula [6] .
εs = 1 + 1.5995ρ + 1.861ρ3 (3.3)
Dielectric constant: The snow dielectric constant real and imaginary parts
are obtained from the air and the ice dielectric constant, since snow is an
inhomogeneous air and ice mixture. For dry snow, the dielectric con-
stants is almost independent of the frequency and the temperature at the
microwaves band of the spectrum, and the imaginary component is neg-
ligible whereas the real one is a function of the density. In the wet snow
case, the imaginary component of the dielectric constant is strongly re-
lated with the liquid water content of snow, increasing rapidly with the
presence of liquid water in the snow-pack. The complex dielectric constant
is a weighted average of the dielectric constants of the snow components:
air, ice and liquid water
ε = ε0 + iε00 (3.4)
The real part of the dielectric constant ε0 is strongly dependent on the

snow density, specially for dry snow. As a summary idea, the more liquid
water content, the higher the dielectric constant, and therefore bigger
absorbtion which means less penetration.
Penetration: It depends on the frequency of the incident electromagnetic
wave as well as on the dielectric constant of the snow, which is driven by
the amount of liquid water. The lower the frequency and the amount of
liquid water, the higher the penetration. Penetration is a very important
parameter for the remote sensing of the SWE, since the penetration capa-
bility of the incident electromagnetic waves will determine the potential
retrieve of information of the back-scattered measured waves. In Figure
3.6, the penetration is given as a function of the liquid water content of
the snow, in the X axis, for several bands.
3.2.2 Snow-pack Microwaves Scattering Signature

Snow-pack is a volume of ice particles and therefore volume scattering has to be
addressed when illuminating snow with EM waves. Volume scattering is caused
mainly by the dielectric discontinuities within a volume, whose spatial location
is usually random, and then scattering is expected in all directions [2]. The
backscattered power received at the radar is the result of the addition of all the
volume scatterers contribution within the solid angle defined by the antennas
radiation pattern [1].
Figure 3.6: snow penetration depth vs frequency [2]
Snow is a mixture of air and ice and its permitivity is a function of the
snow density ρ and the relative permitivities of air and ice. Snow is classified
in two types depending on the amount of liquid water content per unit volume,
since its characteristics and the scattering signature of the snow-pack strongly
depends on it.
Dry snow, which has no liquid water content, has a small absorption coef-
ficient allowing microwaves to propagate over long distances, having pro-
nounced scattering effects in the interaction with the snow-pack volume.
The real part of the permitivity ε0 of dry snow only depends on snow
density which in natural conditions ranges [2]:
0.2 < ρ < 0.5g/cm3 (3.5)
On the other hand, wet snow, with presence of liquid water content, has
a strong absorption coefficient attenuating the microwaves in a very short
distance. In such a situation, scattering is limited to a thin layer close to
the surface called surface scattering. This is due to the strong dependence
of the imaginary part of the permitivity ε00 with the amount of liquid water
in the snow volume. Indeed, it can raise an order of magnitude due to
an increase of a 0.5% of the liquid water volume, drastically reducing the
penetration depth [2].
Another interesting analysis involves the incidence angle in a mono-static
radar configuration:
For small incidence angles (close to nadir) surface scattering is the domi-
nating phenomena.
For bigger incidence angles volume scattering becomes more important,
since the surface contribution is moving to forward scattering.
This document will mainly focus on dry snow, since this is the scope of this
research project, and therefore, from now on, when talking about snow, we will
be referring to dry snow.
As already introduced, the back-scattering response of the snow-pack is the
result of the contribution of different scattering phenomena: underlying ground
surface back-scattering, volume back-scattering from the snow layer, air-snow
interface back-scattering, and multiple contributions from double and triple
bounce between the snow-pack and the soil surface. The following Figure 3.7
from [7] gives a clear graphical idea of the multiple scattering mechanisms that
compose snow-pack back-scattering.
Figure 3.7: snow scattering mechanisms [7]
The low dielectric contrast between air and dry snow-pack and the smooth-
ness of the snow surface lead to a snow surface back-scattering coefficient neg-
ligible ([2] and [1]), and therefore the scattering signature of the dry snow pack
is the addition of the volume scattering contribution and the soil surface back-
scattering contribution. This is the idea that drives the most recent research
works in snow remote sensing with PolSAR: understanding the received back-
scattered power as the addition of different scattering mechanisms out of which
we can infer different informations about the snow-pack. The volume scattering
contribution is driven by the size of the ice crystals relative to the wavelength
and snow particles orientation [2]. The back-scattering coefficient of the snow-
pack is insensitive to the snow surface roughness for dry snow [9].
In most of the cases, the soil surface scattering is considered as Brag scat-
tering or Small Perturbation Method (SPM) [1, 17], since it is considered that
surface roughness is slight with regards to the incident wavelength, and snow-
pack volume backscattering is modeled as a Rayleigh scattering resulting of a
random oriented cloud of particles whose scattered intensity in the far field is
the addition of the individual intensities of each particle [1]. As already said,
volume scattering in the snow-pack is mainly governed by the size of the ice
crystals relative to the incident wavelength [2]: the bigger the snow particles
with regards to the incident wavelength, the higher the interaction of the elec-
tromagnetic waves with the snow grains, and therefore the higher the volume
scattering. Therefore, as the incident wave frequency is reduced the volume
scattering effect becomes more important. Another usual assumption in the
snow-pack scattering response modeling is the Born approximation that estab-
lishes that the scattering generated out of the interactions of the particles with
the diffused radiation is negligible. Born approximation can be assumed since
snow-pack medium fluctuation is very little. Snow presents continuously dis-
tributed volume absorption and volume scattering and is therefore described by
the Radiative Transfer Method (RTM), that describes the waves interaction as
they travel across the volume, taking into account all the different scattering
mechanisms.
At this point of the document we can already point out an interesting con-
cept. For quantitative snow remote sensing, snow-pack volume has to be sensed
in order to get information out of it. This means that microwaves have to pen-
etrate the volume and at same time interact with the snow grains so that some
information about the snow-pack can be extracted from the scattered power.
From what has been said in Section 3.2.1 and in previous paragraph of the cur-
rent Section 3.2.2, one can realize that there must be an interesting trade-off
between penetration capacity and sensitivity to the volume of snow particles. In
[7] is stated that at C-band the volume scattering is the dominant phenomena,
whereas at L-band the ground scattering is the dominant mechanism. However,
this statement is contradicted in other literature sources [?].
3.2.3 Snow-pack Polarimetric Signature

From the point of view of the scattering matrix presented in Section 3.1.2,
snow-pack cannot be analyzed, since it consists of a distributed target subject
to spatial and temporal variations, and therefore the received response is the
averaged contribution of a myriad of single scatterers.
Figure 3.8: Distributed Target Response [3].
Snow-pack analysis requires a statistical approach, based on the second order

moment of the scattering matrix. The polarimetric coherency and covariance
matrices offer statistical information of the scattering Pauli and lexicographic
vectors respectively, that contain the vectorized scattering matrix components.
The elements of these two vectors can be physically interpreted according to the
scattering mechanism that contributes to them. Specifically, the HV propor-
tional component is sensible to volume scattering and the HH-VV proportional
to surface scattering. Therefore these components should have some information
about snow volume and snow sub-surface respectively.
From [5] it can be concluded that at C-Band dry snow is almost transparent
to the incident radar waves for any polarimetric configuration. However, it
is proven that the backscatter is higher for snow free surfaces than for snow
covered surfaces for both co-pol and cross-pol configurations. This might be
3.3 Non Coherent PolSAR Data Methods 40
due to the volume scattering that results in a power dispersion and therefore
less power is back to the antennas. On the other hand, C-Band is not sensitive
to re-frost layers of the snow-pack. However, more recent publications [7] do
find a reasonable sensitivity to the snow-pack volume at C-Band. This difference
might be driven by the amount of liquid water content in the snow-pack, since
the extinction coefficient at C-Band increases dramatically with the liquid water
content, decreasing the backscattered power. Effort needs to be put on this issue
in order to clarify the snow behavior at C-Band.
In PolSAR there are a set of decomposition methods called model-based in-
coherent decompositions that try to identify the main physical components of a
back-scattering scenario in order to find out the contribution of each component
in the total retrieved back-scattered power [4, 3].
3.3 Non Coherent PolSAR Data Methods

Plenty of different techniques and methods for snow parameters retrieval come
out of the research on snow quantitative remote sensing based on non coherent
polarimetric SAR data. In this current Section 3.3 the most relevant advances
will be presented.
First of all, non coherent polarimetric SAR data means that the informa-
tion retrieved is only the amplitude values of some or all of the 2x2 scattering
matrix elements. Without the phase information there is no way to apply fully
polarimetric processing techniques and thus most of the research based on non-
coherent data is based on amplitude contrast or comparison techniques. The
most relevant research works can be grouped in three different areas:
The research work carried out in [9, 10] in 2000 fixed the basis for a method
that has been used in many other research works later. This is based on
the fact that although it does not significantly modify the back-scattering
response, the presence of snow affects the underlying ground backscatter
as well as the relation between the backscatter at different polarimetric
configurations. This initial research work has been the inspiration for
further works such as [6] or [14], and even [15], though this last one is a
more complex evolution of the original idea [9].
The Rennes University 1 with a wide experience in SAR polarimetry, has

led a group of research works on snow remote sensing with PolSAR [7]
and [12]. Though most of them have qualitative approach, it is interesting
what they have achieved in the snow estimation by means of the combi-
nation of polarimetric SAR data and meteorological models [12]. It has
to be clarified that [7] also used the full polarimetric data for polarimetric
optimization in the research activities.
In Canada, in the INRS University Quebec, an ambitious research cam-
paign was carried out achieving interesting conclusions about shallow
snow-pack backscattering response [11]. The method out of their research
linked the thermal resistance of the snow-pack with the backscattering
power observed with no need of polarimetric data.
3.3.1 Snow Density Estimations Based on Co-pol SAR Data

Again, the basis of this group of research started with [9] and [10]. They estab-
lished a method to infer snow density from L-Band dual co-polarization data.
Their rationale is stated as follows: the effects of dry snow cover on surface
back-scattering compared to the surface back-scattering are:
Due to refraction with the snow, the incidence angle at snow-ground in-
terface is smaller than the incidence angle at air-snow interface (Snell’s
law).
The incident wavelength at snow-ground interface is shorter than the wave-

length at the air-snow interface because snow is dielectrically thicker than
air.
The snow-pack reduces the dielectric contrast between air and ground,
reducing the reflectivity at snow-ground interface.
The power attenuation at the snow-pack reduces the incident energy at
the snow-ground interface.
The first two effects in the list above result in a change of the sensor observ-
ables. Both effects depend on the dielectric constant of the snow-pack, which
can be related to the density by means of Looyenga’s semi-empirical formula
in 3.3. we can assume that at L-band the volume scattering is negligible, and
therefore the backscattered power can be expressed by
t 2 s
σpp (k0 , θi )) = Tpp (θi )σpp (k1 , θr )) (3.6)
s
where σpp (k1 , θr )) is the ground surface back-scattering given by the Integral
Equation Method (IEM) model, k1 is the incident wave number at the snow-
√
ground interface with k1 = k0 εs , θi is the incidence angle at air-snow interface
which is related to the incidence angle at the snow-ground interface by Snell s
law,
sin2 θi = εs sin2 θr (3.7)

2
and Tpp is the power transmission coefficient at air-snow interface.
It is observed that snow-pack increases the magnitude of the surface backscat-
tering, specially for HH polarization at large incidence angles, and that the
backscattering magnitude increase is significantly bigger at HH than at VV po-
larization [9]. In summary, though a dry snow-pack doesn’t absorb or scatter
the signal at L-band it affects the surface back-scattering magnitude and the
relationship between HH and VV polarization back-scattering coefficients.
The back-scattering measured co-polarized coefficients σhh and σvv are highly
correlated. They depend on the snow dielectric constant in the wavelength shift
and the incidence angle change, but they also depend on the soil surface prop-
erties (roughness and dielectric constant). In order to estimate snow’s dielectric
constant and therefore density, the goal is to develop a method to characterize
the relationship between HH and VV measurements maximizing its sensitivity
to wavelength and incidence angle and minimizing its sensitivity to the surface
parameters. The model developed has to be based in the sensor observable
parameters and not sensitive to the surface features.
The algorithm development starts with the characterization of the depen-
dence of the surface back-scattering on the incidence angle and wavelength by
means of a IEM simulation for several moistures, RMS surface heights, corre-
lation lengths, incidence angle, wavenumber and surface correlation functions.
Then regression analysis is used to characterize the relationship of HH and VV
back-scattering signatures with all the simulation scenarios at each incidence

angle and wavenumber leading to the equation
√ √
log10 ( σhh + σvv ) = a(θr , k1 ) + b(θr , k1 ) log 10 (σhh + σvv ) +

+ c(θr , k1 ) log 10 (σhh ) + d(θr , k1 ) log 10 σσhh
vv
+
2
σhh
+ e(θr , k1 ) log 10 σvv (3.8)
This equation represents the relationship between HH and VV back-scattering
coefficients at a given incidence angle and wavenumber k1 . Note that the coef-
ficients a,b,c,d and e depend only on the incidence angle and the wavenumber.
Further investigation in the scope of this research is needed since the rationale
for the obtaining of the previous equation is not clear in the documentation
found so far.
The algorithm was validated by means of the comparison of the estimations
obtained from SIR-C L-band images and field snow density measurements, show-
ing an absolute RMS of 42 Kg/m3 and a relative error of 13%.
Further research work was carried out with the same idea than the here-
above explained [13] but also moving to C-Band data [6, 14]. The idea of
the techniques based on C-Band polarimetric data is the assumption that the
total back-scattering coefficient for a dry snow-pack is the sum of two main
v sg
contributions: volume scattering σpp and snow-ground backscattering σpp
t v sg
σpp (k, θ)) = σpp (k1 , θr )) + σpp (k1 , θr )) (3.9)
The development of the algorithm for snow density estimation is based on
the first order volume scattering and the Integral Equation Method (IEM). The
effect on the incidence angle change depending on the snow density found at
L-band [9], applies also for C-Band, that has lower penetration capability and
thus is even more sensitive than L-Band to this effect. Therefore, an algorithm
can be performed to estimate snow density from C-Band ASAR alternative
polarized HH and VV data. Using HH and VV backscattering coefficients for the
snow-ground backscattering, and volume backscattering ratios for the volume
backscattering, a model can be obtained with only two unknowns: the dielectric
constant and the incidence angle that can be related to the dielectric constant
with Fresnel Transmission coefficients and Snell’s law. The model validation
was carried out by comparing the snow density estimations obtained from the
ASAR polarimetric images with field measurements, with a mean absolute error
of 21.2 Kg/m3 . One of the main advantages of the model is that it does not
require knowledge about the surface properties.
Finally, in the same research group, [15] also based on C-Band dual co-
polarization data, comes out with an enhanced technique based in the same
idea, that gives as output an estimation of the snow-pack dielectric constant. It
is interesting the fact that the local incidence angle is corrected on a pixel basis
since in a mountainous region, the local incidence angle has a strong variation,
and it has to be estimated for each pixel prior to its processing.
3.3.2 SAR Data Plus Meteo Model Based Methods

In the work described by [12], a method based on the combination of SAR data
and a meteorological model is presented for the characterization of the snow-
pack, specifically to retrieve the liquid water content, that gives an idea on how
wet or dry is the monitored snow. Three very interesting ideas can be extracted
out of this work:
The main idea is the development of a method based in the combination of
Earth Observation data (ENVISAT ASAR) and a meteorological model
in order to enhance the snow-pack characterization. The good spatial
resolution of the SAR data can significantly improve the capacity of snow
parameters estimation.
The use of a multilayer-snow electromagnetic backscatter model allows
a much more accurate modeling of the scattering phenomena since it
takes into account the physical vertical morphology of the snow-pack, that
presents some layers whose presence depends on the atmospheric and me-
teorological conditions at each given time.
The application of the already mentioned idea about the extinction coef-
ficient increasing with the presence of liquid water within the snow-pack
at C-Band. This principle is used in order to identify the wet snow zones
by means of a simple ratio value on the comparison of winter co-pol SAR
images with reference summer co-pol SAR images.
σwinter
< −3dB (3.10)
σwinter
If for each pixel the condition in equation 3.10 is fulfilled, the pixel contains
wet snow.
3.3.3 Snow-pack Thermal Resistance Based Model for Dry Snow

Monitoring
The work carried out in [11] is based on the comparison of the backscattering
coefficient of a snow covered field with the backscattering coefficient of the same
field snow free at C-Band. A model linking the backscattering coefficient with
the snow-pack and underlying soil physical parameters has been developed for
SWE estimation under a semi-supervised method.
This work was focused on shallow snow-packs, and therefore the snow vol-
ume contribution to the backscattering is negligible, being mainly proportional
to the soil backscattering. However, it was observed a relationship between
the amount of liquid water content in the snow-pack which decreases with the
temperature of the soil when being under 0 degrees, and the soil backscattering
coefficient, that decreases proportionally to the dielectric constant of the soil.
Furthermore a relationship was found between the snow thermal resistance R
and the backscattering power ratio, which is used for the SWE estimation.
The SWE estimation is based on the relation between the winter vs spring
backscattering ratio and the thermal resistance of the snow-pack. There is a
proportional increase of the SWE with the thermal resistance R. The method
is semi-supervised since the mentioned relation depends on the snow density,
changing the slope of the relation between SWE and thermal resistance, and
therefore density dependence has to be estimated.
SW E = ρK(ρ)R (3.11)
where ρ is the snow density and ρK(ρ) is the slope of the regression function
relating SWE with R that has to be estimated.
The method applied is as follows:
3.4 Coherent PolSAR Data Methods 44
Out of the ratio winter vs spring backscattering the thermal resistance of

the backscattering is estimated following an empirical found function:
Restimated = e[ratio+4.1/4.1] (3.12)
where ratio is the winter/spring backscattering division.

then the slope of the relationship between SWE and R can be estimated
using some measured values at the site (semi-supervised)
SW Emeasured
a= with a = ρK(ρ) (3.13)
Rmeasured
3.4 Coherent PolSAR Data Methods

PolSAR data is normally referring to coherent polarimetric data. This means
that both the amplitude and the phase magnitudes of any or all the components
of the scattering matrix 3.2 are retrieved by the radar instrument. Having the
complete amplitude and phase information allows the polarimetric characteriza-
tion of the target and therefore the application of polarimetric data processing
such as the mentioned polarimetric decomposition theorems in 3.2.2. Polarimet-
ric processing relies on fully polarimetric data, which means that data from the
radar instrument fills the whole scattering matrix. However, dual-polarization
(having only two of the four elements of 3.2) data still allows polarimetric anal-
ysis, though having less sources of information.
Snow, as a distributed target needs to be approached from a statistical point
of view. The summarized polarimetric statistical processing is described here-
after:
The retrieved data from the radar instrument composes the scattering
matrix 3.2, whose information can be put in a vectorial form under the
assumption of mono-static radar configuration (reciprocity applies Sxy =
Syx ), using the Lexicographic matrix basis set (see [3] and [4]) leading to
the lexicographic target vector:
h √ iT
K = Sxx 2Sxy Syy (3.14)
where x and y represent two orthogonal polarization states.

Out of the lexicographic target vector the polarimetric covariance matrix
is defined as follows:
 D
2
E √
∗

∗

|Sxx | 2 Sxx Sxy Sxx Syy
√
 D
2
E √

C = K · K ∗T =  2 hSxy Sxx
∗ ∗ 

i 2 |Sxy | 2 Sxy Syy  (3.15)


∗
√
∗
D
2
E 
hSyy Sxx i 2 Syy Sxy |Syy |
with ∗T being the conjugate transpose, ∗ being the transpose and h·i being
the ensemble average
From the PolSAR covariance C or coherency T matrix ,the polarimetric

processing is build, resulting in polarimetric decompositions that intend
to separate the different contributions to the total backscatter, like al-
ready approached in 3.9 from [6]. In the case of snow remote sensing, the
idea will be to estimate the snow-pack volume contribution to the total
snow backscatter, so that we can extract information about the snow pa-
rameters. Model-based polarimetric decompositions seem to be the most
appropriate techniques to face the snow remote sensing issue with polari-
metric SAR data.
The model-based decompositions are those polarimetric decomposition meth-

ods based on identifying physical scattering mechanisms contributing to the
total backscattering target response and associating to each method a partial
coherency or covariance matrix, whose addition leads to the total target co-
herency or covariance matrix.
Care must be taken when developing model based decompositions, so that
they do not result in physically impossible covariance matrices. It has to be
ensured that the covariance matrices of the decomposition have non-negative
eigenvalues. In [16] a method is proposed to guarantee such a condition when
developing model based decompositions.
In the following subsections the most relevant works and ideas about fully
polarimetric SAR snow remote sensing will be highlighted. Since, as already
mentioned, snow remote sensing with PolSAR is a quite new research field, the
concepts and ideas presented might not be focused on snow remote sensing, but
the conceptual similarities make them valuable references.
3.4.1 Polarimetric Decomposition over Glacier Ice

The research carried out in [17] is extremely interesting since, though it is fo-
cused on ice, most of the ideas and concepts used could be applied or adapted
for the snow case. It is stated that to infer ice volume features, surface and
volume scattering contributions have to be identified separately.
The snow-pack and ice cases have considerable similarities in the way the
problem is approached. For instance, the aspect of the illuminated target follows
the same layered structure in both cases (see Figure 3.9). It consists of a first
interface air-ice or air-snow, a volume of ice or snow and the interface of the ice
or snow with the underlying soil.
Figure 3.9: Radar signals illuminating snow-pack [20]

In addition, ice and snow have the same physical components. Obviously,
proportions and morphologies are different and thus, so it is the dielectric con-
stant, which is the most important parameter driving the interaction between
EM waves and ice or snow.
In [17] the approach chosen to infer properties of the ice pack consists of
decomposing the polarimetric covariance matrix in three submatrices, repre-
senting three different scattering sources: ground surface scattering, ice volume
and sastrugi (wind-induced features is the snow-firn upper layer). Then the
covariance matrix results as:
Ctotal = Cground + Cvolume + Csastrugi + N (3.16)

where N is a noise contribution. Each scattering mechanism contributes to
different components of the combined covariance matrix. Surface, modeled as a
Bragg scatterer contributes to xx and yy powers and to the complex co-polar
correlation. Volume, which in this research work is either studied as an oriented
or random volume, contributes to all components or to all except the cross com-
ponents depending on the orientation randomness, and sastrugi also contributes
to all components, but only with real values (no phases). In this work, the ap-
proximation of ice as a random oriented volume is a loss of generality, which
would be much less important in the snow case due to its less layered structure.
Ice formation starts with snow being accumulated and pressured, promoting the
formation of layers, which is not the case of snow. Snow-pack is normally fresh,
it has been subject to less external factors originating the layering due to its
youth, and therefore the random orientation of the snow particles should be a
correct assumption.
A very interesting part of this job takes into account the wave propagation
effects in the ice volume using the Transitivity coefficients of the different in-
terfaces in the modeling of the volume covariance matrix. This approach could
also be reused in the snow-pack case.
Finally, several key ideas are pointed out in [17] that might be very useful
for the PolSAR snow remote sensing case:
The assumption of reflection symmetry, which in the case of ice scatter-

ing can be doubtful, should apply for the snow case provided its random
particles orientation nature.
The assumption of a dipole shape for the ice particles might be a too wide
generalization, which will have to be studied for the snow case, since the
same situation applies.
The terrain slope is assumed to be constant in glacier zones. This won’t be
the case in the remote sensing of seasonal snow whose presence is normally
associated with mountainous zones. In such a case, the idea presented in
Section 3.3.1 from [15] about correcting the local incidence angle in the
snow-pack on a pixel basis will be taken into account to deal with the non
constant terrain slope.
3.4.2 Other Polarimetric Decompositions Addressing Volume

Since it seems that dry snow-pack scattering is basically the soil surface plus
the volume contributions, and that to infer information about the snow-pack
the volume scattering has to be estimated, other references on polarimetric de-
compositions addressing volume scattering have been considered. The current
3.5 Conclusions 47
Section intends to gather the most interesting ideas out of that group of refer-
ences, that could apply for the snow case.
Some interesting research work has been done with PolSAR in the field of
vegetation remote sensing [19, 21]. From the physical point of view, vegetation
is seen as a volume, called canopy. This is the aboveground portion of plant
community or crop. Since the snow-pack is also seen as a volume of air particles
in an air background, concepts from the PolSAR vegetation volume remote
sensing might be also interesting for the snow case.
These two works [19, 21] have the same starting point idea, which is to model
separately the ground and the volume contributions to the total backscatter-
ing of the vegetation covered soil. In [21] a Polarimetric Interferometry SAR
(PolInSAR) approach is used, and the vegetation is seen as a random volume
over ground (RVoG). The RVoG is a model used in PolInSAR to asses the two-
layer scattering coherence problem (vegetation and soil). It considers a random
oriented volume, and therefore the propagation is independent of the polariza-
tion and depends only on the extinction coefficient. It is remarkable from this
job the approach of separating ground and volume contributions out of the total
backscattered power in order to infer information about the volume of vegeta-
tion, since this separation idea seems to be present in all the situations where a
volume scattering is present.
The work carried out in [19] is based on PolSAR. The most relevant idea
regarding the snow-pack monitoring is the inclusion of a model for an oriented
volume. They rely in a probability density function to characterize the particles
distribution, which is completely defined by a mean orientation angle and the
power of a cosine-square function. They come out with a decomposition of the
volume covariance matrix which is the addition of the oriented volume contri-
bution plus the random oriented contribution on a weighted manner depending
on the scatterers volume characteristics.
Finally, another interesting decomposition of PolSAR data is proposed in
[22]. This consists of a four-component decomposition to take into account
the cases when reflection symmetry cannot be applied, which is normally in
the case of man-made targets. This decomposition adds to the well known
surface, double-bounce and volume scattering components, a fourth helicoidal
mechanism. It is a suitable decomposition method for those images with man-
made targets as well as natural distributed targets.
3.5 Conclusions
After the state-of-the-art review concerning SAR and polarimetric SAR for snow
remote sensing, it is clear that snow remote sensing is a very interesting research
topic, on-going for the last decades, due to its scientific applications on climate
study as well as hydrology, and water resources management applications, such
as flooding or water reservoirs forecasting. This Section aims to summarize
all the relevant conclusions out of the literature review, as well as the ideas to
initiate the research work on PolSAR for snow quantitative remote sensing.
Concerning conclusions:
It seems, out of the review, that the general idea to face the volume
scattering case with a model-based decomposition combined with an EM
polarimetric model of the snow scattering problem; involving volume un-
derlying surface scattering plus snow-pack volume scattering, plus other
mechanisms scattering if any, is the appropriate one for the scope of snow
3.5 Conclusions 48
quantitative remote sensing. A polarimetric decomposition has to be used

in order to estimate the volume scattering contribution out of the total
backscatter, and therefore, be able to infer snow-pack volume informa-
tion. And a good model for the snow-pack has to be used to extract the
appropriate information of the snow volume scattering.
C-Band is not very sensitive to snow volume, as it was pointed in [11].
Shallow snow-pack is transparent to C-Band. However, it presents good
snow penetration capabilities. This band might be a good trade-off be-
tween penetration and sensitivity.
The dry to wet snow transition should be clearly detectable, since the
presence of liquid water content in the snow volume drastically changes
the imaginary part of the dielectric constant ε00 , increasing the absorption
of the microwaves, and therefore attenuating the backscattered power.
The incidence angle is an important parameter that influences backscat-
tering. In mountainous zones, the local incidence angle will computed on
a pixel by pixel basis in order to account for the geographical variations
that affect the scattering geometry configuration.
Snow seems to be well modeled as a random oriented cloud of ice particles
in an air background.
Out of the conclusions, the ideas for the initial research steps, which should
inspire further research on this topic are described as follows:
C-Band is going to be chosen to carry out the research with satellite
SAR data, due to its reasonable sensitivity vs. penetration trade-off, to
its space-borne availability, and to the forecasting of future space-borne
availability in the short or mid-term.
A model-based decomposition for the seasonal snow-pack in mountainous
zones will be the proposed method to infer information about the snow
volume. The idea is to be able to separate the snow volume backscatter-
ing contribution of the total gathered backscatter power, and establish a
relation-ship, somehow, between the intensity of the volumetric backscat-
ter and the snow density, by means exploiting the dependence of the real
part of the dielectric constant on the snow density. Therefore, the goal is to
find a relationship between the snow volume scattering and the dielectric
constant.
In a first approach it seems reasonable to model the snow-pack as a ran-
domly oriented volume of small spheres. However, it has to be kept in
mind the possibility of modeling the snow as an oriented cloud of par-
ticles, whose shape can go from a sphere to a dipole depending on the
external conditions to which snow-pack has been subject.
With the sake of summarizing the whole section ideas before closing it: C-
band fully polarimetric space-borne data with very good spatial resolution,
seems a suitable Earth Observation data source to face a research on snow
quantitative remote sensing in order to retrieve density as a first step towards
SWE. The core idea is to come out with a model-based polarimetric decompo-
sition that enables the volume scattering estimation out of the total received
backscattered power, in order to estimate, the relationship between the density
and the volume scattering intensity.
49
4 C-band fully Polaimetric SAR measurements over Moun-

tainous Regions
4.1 Indroduction
As it has been already mentioned, the idea behind this Master Thesis is: SAR
Polarimetry might be sensitive to snow volume at C-band, and if so, could it
be a valuable tool for snow remote sensing? Putting it in other words, can
polarimetry overcome the microwaves C-band low frequency limitation to be
sensitive to snow? . Taking this assumption as starting point, a research project
has been carried out in order to asses the validity of the assumption.
According to [2] and as it has been seen in section 3.2.2, microwave pen-
etration depth in snow is a function of frequency and liquid water content in
the snow, and it is obvious that when intending to be sensitive to the volume
of the snow-pack a significant penetration depth is needed. Therefore, it might
seem that a very low frequency microwave would be well suited for our purpose.
However, on the other hand, it is also known from the existing literature [2, 1]
that the smaller the wavelength the bigger the interaction of the microwave with
the snow particles, normally in the range of sub-millimeter size, i.e., the bigger
the interaction with the snow-pack volume. At the end, a trade-off has to be
found in order to guarantee penetration in as many snow types as possible, in
terms of quantity of liquid water, and sensitivity to the volume. In our case,
C-band has been chosen since it offers reasonably good penetration in not 100%
dry snow, and it seems to show sensitivity to snow volume [7, 1]. The chosen
available space-borne sensor providing C-band fully polarimetric SAR data is
the Canadian RADARSAT-2.
4.2 Description of the Experiment 50
The following subsections will describe the research work carried out, from
the experiment planning, to the data processing and analysis, and finally the
snow monitoring features obtained. Summarizing, the work accomplished fits
the classical structure of a remote sensing research project, with a ground mea-
surements campaign providing with ground truth, some satellite data acquisition
in coordination with the ground campaigns, that produce the polarimetric SAR
data to be processed, and finally some ancillary data supporting the assump-
tions during the data analysis and interpretation of the results. The processing
of the data acquired aimed to find some feature sensitive to the snowed scenarios
that could be used to retrieve snow physical parameters. Finally, an effort to
synthesize the lessons learned out of the gathered data and its processing and
analysis leads to the conclusions and further steps section.
4.2 Description of the Experiment

4.2.1 Experiment Planning and Design
Considerable effort was put in the experiment design and planning since the
team was aware that decisions taken at this point of the project would determine
future results and quality of the research work to be accomplished.
First of all, the test site for the data acquisition was chosen. It is well known
that SAR signal in mountainous regions suffers for layover and shadowing effects,
decreasing the number of valid pixels for the geo-physical parameters retrieve.
Therefore, the idea was to find a more or less flat meadow zone in an alpine
environment. In addition, the need of having an on-site measurements campaign
together with the fact that our interest was set in the Pyrenees due to proximity
reasons, the chosen test site was Pla de Beret in Vall d’Aran, Catalonia, in the
North-Est mountainous region of Spain.
Pla de Beret is a quite flat Alpine meadow located at latitude 42 43’ 21.352”
N and longitude 0 57’ 31.462” E, at an altitude of 1860 meters. Its situation
as part of the ski resort Baqueira-Beret guarantees a good access, which is a
valuable asset for the implementation of the ground measurements campaign.
The following Figure 4.1 shows a picture of the zone.
At an altitude of 1860 m., this meadow is normally snow covered during
most of the winter season, from December to March. Data acquisition of
RADARSAT-2 for this test site was coordinated with the ground measure-
ments campaigns and also with the snow conditions on the site, trying to
have significantly different snow features in the different acquisitions. There-
fore, the election of the acquisition dates was dynamically adjusted trying
to match RADARSAT-2 satellite passes, ground truth campaigns availability
and favorable meteorological information and forecasts. Finally, three different
RADARSAT-2 data acquisitions were programmed: one on the 17th February
2011, a second one on the 13th March 2011, both supported with simultaneous
ground measurements campaigns, and a third one on the 15th Nov 2011 in order
to have a no snow satellite image to use it as polarimetric SAR data ground
truth.
Polarimetric SAR data acquisitions from RADARSAT-2 belong to the Fine
Quad Pol configuration of the sensor with ascending orbit, providing fully po-
larimetric information (HH,HV,VH and VV) with a pixel spatial resolution both
at range and azimuth of around 8 meter. In all the three acquisitions the geom-
etry of the acquisition is the same, determined by the FQ28 acquisition mode
of the satellite sensor, fixing an incidence angle of 46.6 degrees, and a covered
Figure 4.1: Pla de Beret North looking November 2011.
area of around 25 Km2 . SAR images were acquired in Single Look Complex
(SLC) configuration, so that no loss of information affects SAR data before
own processing. The Pauli decomposition RGB representation for the Pla de
Beret Zone in the three different acquisitions can be appreciated in the following
Figure 4.2. Henceforth in the document, those figures containing SAR images
or derived products have the azimuth dimension in the vertical axis and range
dimension in the horizontal axis:
Note that blue color corresponds to the polarimetric channel for plane sur-
face scattering, red corresponds to double-bounce scattering and finally green is
associated with the volumetric scattering (the non-diagonal components of the
scattering S matrix [3]). By means of comparing these snapshot images with
a zenithal optical image (see Figure 4.3) from the same zone of Pla de Beret
we can identify different elements in the scenario, such as the parking zone of
the ski resort and the road that gets to it, the small canyons on the right side
of the parking area or the tele-ski path as a straight line crossing the images
horizontally.
Concerning the ground truth measurements, the two campaigns carried out
consisted of snow measurements of snow depth and snow density, as well as a
general assessment of the snow status carried out by a snow expert who was
present in both campaigns. The approximately 30 measurements per campaign
covered the Pla de Beret zone, with more or less equi-spaced measured spots.
Fortunately, snow conditions changed significantly from one date to the other,
with the snow depth increasing a 100% and the snow density increasing 34%
from one scene to the other. In Table 1, a summary of the average values
obtained during the campaigns is presented, the Difference column refers to the
differences between the two snowy acquisitions.
The ground measurements campaign was carried out in the frame of a project
called AGORA, which was a research project led by Starlab Barcelona S.L. and
funded by ACC10 targeting the use of remote sensing data for the improvement
of the hydrological modelling. The author of the current Master Thesis would
like to thank the AGORA team for the collaboration during the planning of the
Figure 4.2: Color-coded image of Pauli decomposition for the 3 acquisitions.

The colormap used is: HH+VV Blue, HH-VV Red and HV green, and axis
units are pixels
campaigns and the SAR data acquisitions and the valuable data provided.
4.2.2 Data Set

After a first approach analyzing the polarimetric SAR data and the ground data
collected, the need of having additional information about the meteorology on
the site turned up, since the correct interpretation of the polarimetric SAR
data required some knowledge about the snow status that can be estimated
Figure 4.3: Ortho image from Pla de Beret (Institut Cartografic de Catalunya).
Table 1: Snow ground measurements summary
Parameter February 2011 March 2011 Nov 2011 Difference

dryness 100% dry snow not 100% dry snow no snow liq. H2 O
density 236 Kg/m3 317 Kg/m3 0Kg/m3 34%
depth 32.24 cm 62.0 cm 0 cm 100%
from temperature and precipitation information.

Therefore, some ancillary data was also collected, specifically meteorological
data from Pla de Beret site. The meteorological information was obtained from
Confederación Hidrográfica del Ebro, who has an automatic station in Pla de
Beret. Daily information about temperature and precipitation from December
2010 to December 2011 was obtained in order to have ancillary information
to better estimate the exact snow status at the time of each acquisition. The
meteorological data set consists of mean, min and max temperature in degrees
Celsius values and precipitation in liters for the entire period. It has been very
useful the fact that the min and the max temperature values are timestamped,
so that we could have an idea of the temperature evolution some days before
each acquisition took place.
Thus, summarizing, the resulting data set of the research experience is listed
as follows:
Three fully polarimetric C-band RADARSAT-2 Fine Quadpol mode im-

ages centered at Pla de Beret zone. The dates of the images acquisition
are 17/02/2011, 13/03/2011 and 15/10/2011.
Results of two ground truth measurement campaigns registering snow

depth and snow density in several geo-referenced spots in Pla de Beret
during the days 17/02/2011 and 13/03/2011.
Meteorological time series data from December 2010 to December 2011
4.3 Snow Backscattering Results Analysis 54
containing temperature and precipitation information.

In the following subsections, a detailed description of the research process,
the methodology, the use of the data set and the conclusions extraction will be
provided.
4.3 Snow Backscattering Results Analysis

As it has been already mentioned, the idea triggering this Master Thesis was
to check the sensitivity of fully polarimetric C-band SAR data to the dry snow-
pack, hopefully to the volume of snow, using a target polarimetric decomposition
theorem. According to [1] snow volume is an important scattering mechanism in
the snow backscattering, as it was also seen in [8]. Likewise, underlaying surface
backscattering of a snow covered scene is also significant and can be different
from the non snow case, as it was introduced in [9] for L-band polarimetric SAR
data, and in [5] for C-band with an incidence angle of 40, which is close to our
incidence angle of 46.6. Therefore, at the beginning our focus was fixed on the
volumetric and the surface scattering mechanisms trying to identify sensitivity
to dry snow.
For the entire polarimetric SAR data processing a set of different SW tools
has been used:
PolSARPro: Is an ESA software specially suited for the processing of po-
larimetric SAR data. It performs all the basic SAR processing operations
oriented to polarimetric data, and allows to compute several polarime-
try products and decompositions. Data extraction from SLC to further
polarimetric products can be carried out with this software.
NEST The Next ESA SAR Toolbox is a software for SAR data process-
ing. It performs all the basic SAR processing operations such as Speckle
filtering, Calibration, Terrain Correction, Co-registering...
Matlab Matlab has been used for the creation of own scripts with specific
functions, such as image patch selection, terrain slope correction on the
polarimetric orientation angle, statistics calculation, ....
4.3.1 Preliminary Analysis

The first approach to the polarimetric data analysis was focused on observ-
ing changes in the Pauli surface and volumetric scattering (HH+VV and HV
channels) with the presence or no presence of snow, and hopefully identifying
a quantitative parameter tied with the quantity and density of snow. For the
whole research work a change detection technique has been applied. Provided
that the polarimetric SAR images were all acquired with equal geometries, the
direct comparison of the polarimetric features of the same zones of the different
images can provide direct information, which in addition is independent of the
topography, since the effect of this it is deleted when performing a ratio of values
of a specific zone from two different SAR acquisitions.
At the beginning, only the flattest zone of Pla de Beret was used for re-
search activities, leading to the Test Zone 1, covering the meadow below the
parking area on the right side of the road getting to it, with a length of around
1800m and a width of 300m. The local incidence angle in this test zone is like
the mean nominal one for the acquisition: 46.6 degrees. A statistical analy-
sis approach was employed provided that Test Zone 1 can be considered an
homogeneous distributed target. Therefore, mean intensity values, and its cor-
responding variance, were computed for the following polarimetric channels:
HH, HV, VV, HH+VV, HH-VV and Span for the three SAR acquisitions. The
a priori expected result according to the literature, was finding bigger Span
in the snowy scenes, due to the increase of the intensity in HV and HH+VV
channels. Obtained mean values in dB for the different channel intensities are
summarized in the following Table 2:
Table 2: First analysis results summary for Pla de Beret zone (Test Zone 1)
Parameter (dB) February 2011 March 2011 October 2011

HH 57.19 52.34 57.01
HV 48.44 44.66 49.80
VV 56.50 51.96 56.42
HH+VV 58.86 53.81 57.0
HH-VV 53.04 49.44 56.43
Span 60.45 55.88 60.54
As it can be observed, results obtained didn’t match the expected figures.

First of all, surprisingly, February scene (snowy) and October scene (ground
truth without snow) have almost the same Span, with similar values in most of
the channels, therefore, we initially assumed, as already suspected, that snow is
apparently transparent to C-band polarimetric radar signal. One might think
that the shallow snow-pack scene in February (see Table 1) is the cause of
almost no effect of the snow in the statistically analyzed zone with regards
to October ground truth. Following the criteria of snow quantity, the March
scene should appear with more backscattering than February and October, and
probably with bigger surface and volume scattering. However, data shows that
March scene backscattering power is around 5dB lower in all the polarimetric
channels when compared with October ground truth image data. This results,
opposed to what we were expecting, motivated a visit to the test site, in order
to visually identify the site physical characteristics that might be generating one
or another scattering mechanism. On November 14th 2011, a Test Zone 1 visit
was accomplished with the following conclusions:
There is no vegetation present in the grassland zone analyzed that can

generate volume scattering. Soil is covered by mountain grass that doesn’t
have no trunk nor canopy that could be the origin of volumetric scattering.
Surface is not strictly flat, and the roughness might be high enough so
that no Bragg [2] surface scattering can be assumed in our data analysis.
In the next Figure 4.4 , a picture of the soil can be appreciated, with two
kinds of roughness: small scale roughness due to grass and small variations
and a bigger scale roughness with random small hill like perturbations.
After the Pla de Beret site visit no reasonable explanation could be found
to the fact that February and October scenes have very similar backscattering
in most of the channels and that March scene, with a thicker snow-pack, has
around 5dB lower backscattering than the other scenes.
A potential hypothesis was assuming that March scene snow was not com-
pletely dry. Past research results at C-band with wet (Liquid Water Content
Figure 4.4: Pla de Beret surface detail, November 2011
> 1% per unit volume [2]) snow at an incidence angle of 40 degrees [5] show
that from dry to wet snow backscattering power falls around 5dB. In order to
prove this hypothesis, meteorological information of the site from the interest
period was needed for the estimation of the snow status during the satellite
data acquisition. The Confedereación Hidrográfica del Ebro has a permanent
automatic meteorological station at Pla de Beret, and daily basis information
was obtained about temperature and precipitation for the period of interest.
Taking into account the fact that satellite data acquisition for the three scenes
was around 18h, and with the knowledge of the meteorological data about tem-
perature from the acquisition days (see Table 3), the following assumptions
were done. In the February acquisition the temperature was below 0 during all
the day, as well as the two days before, guaranteeing dry snow status of the
snowpack, whereas in the March case, temperature was above 0 during most
of the day, and in addition max temperature time was only 90 minutes before
the acquisition time, meaning no time enough for the snow to refreeze before
acquisition time. Again, March 2 days prior to the acquisition showed similar
meteorological features in temperature. This means that we can assume that
March scene’s snow status was wet.
Table 3: Temperature in Pla de Beret zone for February and March acquisition
days
Temperature (C) February 2011 March 2011

mean -3.7 2.3
min -6.7 -0.3
max -0.7 6.2
min time 4:15h 0:15h
max time 13:30h 16:30h
4.4 Snow Monitoring 57
In order to generalize the robustness of the meteorological conditions in

which we based our assumption of wet snow for March scene, a plot with the
temperature evolution of the period of interest has been obtained, so that we
can enhance the reliability of our hypothesis. Notice that the few days prior to
the two SAR data acquisitions have similar temperature values: mostly negative
temperatures for the days before February acquisition and mostly positive values
for the days before March acquisition, so that we can conclude that the snow in
February acquisition was dry whereas the March snow was probably wet, having
experienced average temperatures above 0 during the 4 days prior to the SAR
acquisition.
Figure 4.5: Temporal evelution of the Temperature in Pla de Beret during the
period of interest)
4.4 Snow Monitoring

After the first approach processing the data, it was realized that more data
spots needed to be processed in order to reach robust polarimetric snow signa-
ture ideas. Therefore, targeting a wider statistical analysis, several test zones
were identified within the limits of the image. The condition was trying to find
homogeneous distributed target zones from which to be able to extract signifi-
cant polarimetric backscatter information concerning snow type and presence.
Finally 6 different zones have been used in the research work described sub-
sequent sections: two alpine meadows reasonably flat at an altitude of 1800
to 1850m, two forested areas with opposed slopes, another alpine meadow at
2200m and finally a urban zone. The summary of the test zones chosen is as
follows:
Pla de Beret
– Number: 1
– Geo-coordinates: lon 0.966◦ and lat 42.732◦
– Altitude: 1860 m
– Area: 60 x 300 px (trapezoid)
– Description: Alpine meadow

– Comments: Quite flat, so that local angle of incidence is 46.6◦ in
average. There is a little water stream vertically crossing the meadow
that may appear in the SAR image
Bosc Er Anherar
– Number: 2
– Area: 40 x 100 px
– Description: Alpine forest in the north-est corner of Pla de Beret
– Comments: A very steep slope facing the sensor azimuth trajectory,
therefore the local incidence angle has been estimated to ba 22.6◦
Bosc de Parros
– Number: 3
– Altitude: 1860
– Area: 40 x 100 px
– Description: Alpine forest in the north-west corner of Pla de Beret
– Comments: A steep slope facing opposite to the sensor azimuth tra-
jectory, therefore the local incidence angle has been estimated to be
68.4◦
Bonaigua - Pla de Moredo
– Number: 4
– Area: 100 x 20 px
– Description: Alpine Meadow
average.
Pla de Moredo
– Number: 5
– Area: 70 x 60 px
– Description: Alpine meadow
average.
Naut Aran
– Number: 6

– Area: 60 x 50 px
– Description: Urban zone (village)
average.
Notice that the two forested zones (2 and 3) are located in the Pla de Beret,
so that all the ground truth snow measurements information as well as the
meteorologic data can be also used in the analysis of these two zones. Concerning
test zones 4 and 5, both quite flat alpine meadows, they were chosen in order
to have, on one side, a meadow (5) similar to Pla de Beret in altitude, so
that assumptions about similar meteorological and snow conditions could apply
provided that the two meadows are separated approximately 4.5 Km, and on
the other side, a 400 m. higher meadow, in which by assuming the rule of the
temperature decrease with the height increase, the temperature can be assumed
to be between 2 and 4 degrees lower than in Pla de Beret, and the quantity
of snow is expected to be higher. Finally, test zone 6 was chosen as ground
truth, being a urban zone its polarimetric backscattering response was expected
to be know a priori. The following Pauli RGB representation of the February
acquisition shows the situation of the different 6 test zones within the image
(see Figure 4.6).
Figure 4.6: Pauli RGB representation of February 2011 PolSAR Acquisition

with the 6 test zones marked on it). Axis units are pixels and colormap is as in
4.2
Analyzing data on a test zone basis, the idea of the forested zones was to
be able to notice the effect of the trees in the volumetric scattering component,
masking the potential snow contribution to the volume scattering component
having in addition two very different local incidence angles in each forested test
zone, allowing the observation in the polarimetric backscattering response of the
change of angle of incidence effect. The estimation of the local incidence angle
in each zone has been done with an easy trigonometric transformation having
topographic basic information from the site: the slope of each forested zone can
be estimated from the ratio of increase in height with the distance. In Figure
4.7 a simple diagram depicting the geography of the north part of Pla de Beret
shows graphically the method employed for the local incidence angle estimation.
Obtaining the topography information from a detailed map (Vissir application
from Institut Cartogràfic de Catalunya http://www.icc.cat/vissir3/), the calcu-
lation of the local incidence angle simply involves an arctangent operation and
a difference as shown in equations 4.1 and 4.2.

70 ◦ 250
α = atan = 21.8 β = atan = 24.0◦ (4.1)
200 560
0 00
θi = 46.6◦ + α = 68.4◦ θi = 46.6◦ + β = 22.6◦ (4.2)
Where α and β are respectively the angle of the terrain slopes in Bosc de
Parros and Er Anherar, obtained from the topographic information, as shown
in Figure 4.7, where it can be seen as well the different local incidence angles
due to the different terrain slopes.
Figure 4.7: Diagram for the different incidence angles according to the terrain
slopes
In order to easily perform a change detection technique approach in the

analysis of the three different PolSAR images, a co-registration of the images
was needed, so that it can be guaranteed that each pixel in the three different
images correspond to the same geophysical point. The images co-registration
was carried out by the radar team members in TSC using a proprietary software
that achieves very good results in residual co-registration spatial error. Having
applied the co-registration, the aspect of the three Pauli representations of the
PolSAR images was as follows.
For each of the presented test zones, a statistical analysis was carried out: the
averaged intensities were computed for each polarimetric channel plus the Pauli
first and second channels as well as the Span. Note that since the acquisitions
Figure 4.8: PauliRGB representation of the co-registerd data images. Axis units
are pixels and colormap is as in 4.2
were in a mono-static configuration, reciprocity constraints Sinclair matrix to

be symmetrical, which means that HV = VH, and therefore VH is not included
in the analysis. The results of this analysis are summarized in the Tables 5, 6
and 7. In parallel to the polarimetric analysis, the polarimetric decomposition
H/A/ᾱ was applied to the co-registered data, using a 7x7 window from the
full resolution Coherence matrix for each of the three acquisitions. The same
statistical analysis was carried out in the 6 test zones for this polarimetric eigen-
based decomposition, i.e. computing average values for Entropy, Anisotropy and
ᾱ for each of the test zones. The statistical analysis data from the polarimetric
decomposition in the six different test zones is summarized in Table 8. The idea
was to scan the polarimetric averaged intensities from the different test zones
looking for significant results or patterns, when comparing data from the three
different acquisitions for each test zone. The detailed analysis for each test zone
is described hereafter.
4.4.1 Forested Zones

Having already analyzed the results from Pla de Beret zone (test zone 1), the
two forested areas were interesting in order to analyze the effect of the snow
presence and type in the two differently oriented slope forests. As stated in [26],
for large incidence angles the volume scattering contribution from the forest
canopy is expected to be dominant, since the surface scattering rapidly decreases
with increasing angle of incidence. The key aspect is checking whether the
effect of snow can be observed in the two differently oriented forested areas.
Concerning test zone 2 (see Tables 5, 6 and 7), with a very low incidence angle, it
is interesting to notice that Pauli first channel, tied with surface backscattering,
is bigger in the February scene (dry snow) than in March scene (wet snow),
probably due to reduction of the wet snow backscatter, that increases absorption
and forward-scattering . On the contrary, the second Pauli channel, associated
with double-bounce scattering, shows bigger values in the wet snow scene than in
the dry snow one, probably due to the increase of forward scattering in the snow
surface that leads to more double bounce on the tree trunks. When comparing
the two snowy scenes with the no snow October scene, it is interesting to point
that the double bounce Pauli component is bigger in the no snow scene. Finally
it is surprising that the Span is nearly the same in the three scenes, maybe
suggesting that the backscattering phenomena in this test zone is dominated by
the forest and the small angle of incidence, having very few sensitivity to the
presence and/or type of snow. Regarding, test zone 3, a forested area with a high
local incidence angle, the analysis showed similar results than test zone 2, but in
this case, the Pauli surface component is similar in the dry and wet snow scenes,
as well as in the no snow scene. This fact might be due to the very high local
incidence angle, that causes a significant drop in the surface scattering regardless
of the snow. Analyzing the two test zones together, backscattering intensity is
clearly bigger in test zone 2 than in 3 probably due to the general behavior of
decreasing backscattering when increasing the angle of incidence. Concerning
the effect of the forest in the volumetric and double bounce contribution of the
backscattered power, a relative analysis has been carried out involving test zones
|2 |HH−V V |2
1, 2 and 3. It consists of evaluating the ratios: A = |HV Span and B = Span
for the three different test zones and temporal scenes. Results are summarized
in the following Table 4. First row contains A ratio results and second row
contains B results:
Table 4: Forested areas volumetric and double bounce scattering influence

1 2 3
Feb Mar Oct Feb Mar Oct Feb Mar Oct
5.3% 5.2% 7.4% 7.8% 9.8% 10.7% 9.8% 11.9% 12%
18% 18.8% 21.5% 14.7% 19.2% 20.8% 21.4% 23.6% 23.8%
Note that test zone 3 is the one where the forest effect on volume and double
bounce is bigger, probably due to the highest incidence angle that leads to a
bigger volumetric contribution since the signal path interacts longer with the
vegetation. It is worth to check the dominance of the double bounce component
in all the scenes and test zones with regards to the volumetric component.
Finally, it is interesting to realize that, regardless of the test zone, both the
volumetric and the double bounce components are bigger in the no snow scene
(October) than in the wet (March) or dry (February) snow scenes. This fact
will be further analyzed in the following sections.

When putting the attention to the polarimetric H/A/ᾱ decomposition re-
sults (see Table 8) for these two test zones a few interesting observations were
done. It can be clearly appreciated the fact that Entropy as well as ᾱ increase
when moving from dry to wet snow scene and from wet snow to no snow scene.
This fact is opposite to what was expected, since it suggests that the presence
of snow reduces the randomness of the backscattering phenomena in the ob-
served test zones. In test zone 2 with a very low incidence angle, the Entropy
and ᾱ values in the two-dimensional H/ᾱ plane correspond to a random surface
without snow, close to the anisotropic particles volume scattering, moving to a
Random surface in the dry snow scene. In test zone 3, Entropy and ᾱ values
indicate a typical volumetric backscattering with very high Entropy values and
ᾱ values in the range of 40◦ to 50◦ , staying in the Anisotropic particles zone.
When comparing with test zone 1, it is interesting to see the quasi deterministic
nature of test zone 2, since being a forested zone, it was expected to have a
more random character than test zone 1. Note that Entropy and ᾱ values for
test zone 2 are clearly lower than those of test zone 1. This is probably due to
the low incidence angle in test zone 2, decreasing the contribution of the trees in
the backscattering response. On the other hand, comparing test zones 1 and 3,
the forested test zone 3 has, as expected, clearly bigger Entropy and ᾱ values.
4.4.2 Alpine meadows

In the analysis of the alpine meadow test zones 1, 4 and 5, the sensitivity to
the snow in the polarimetric backscatter is to be assessed without any distort-
ing effect, since the backscattering response of these meadows is, a priori, well
known. The only changes in these test zones along the temporal scenes can be
due to snow. Concerning test zones 4 and 5, the first one is 400 m higher than
Pla de Beret, and it was therefore assumed that the increase in altitude may
lead to a thicker snowpack in this test zone. Regarding test zone 5, it has very
similar characteristics than Pla de Beret, and therefore results are expected to
be similar.
Proceeding with the comparison (see Tables 5, 6 and 7), when looking at the
Span values of test zones 1 and 4, there is a decrease in both of around 5dB from
February to March scenes, suggesting the presence of wet snow also in the higher
test zone 4. Furthermore, it is very interesting to notice that the Span in test
zone 4 is 3dB higher in the dry snow scene than in the no snow scene, whereas in
test zone 1 Span is almost equal in these two scenes. Probably related with this,
there is also the fact that the HV component intensity is also bigger (around
2dB) in the dry snow scene than in the no snow scene in test zone 4. This may
be due to the thicker snow-pack in test zone 4 than in test zone 1. Concerning
test zone 5, all the polarimetric channels analyzed show very similar values than
those in test zone 1, with the 5dB decrease from the February to March scenes
indicating the presence of wet snow in March acquisition. It is worth to drive
your attention to that fact that, again, both in test zones 4 and 5, the second
Pauli channel (HH-VV) associated with the double bounce scattering, presents
higher values in the no snow scene than in the wet snow scene, as well as in
the wet snow scene than in the dry snow scene. Notice that this fact can be
observed in all the analyzed test zones up to now (1,2,3,4 and 5), suggesting
some interesting feature to further analyze in deep. Finally, a phenomena which
was not expected arose in the data analysis: the double bounce component in
the meadows is significant, with a magnitude which is comparable to the surface
scattering component in the no snow scene for the three meadows analyzed (1,4
and 5). It was not expected to be as high as the surface scattering component,
since the initial idea was that, being a mountain grass covered surface, the
surface scattering would be much higher than the other scattering mechanisms.
In what the polarimetric H/A/ᾱ decomposition is concerned the behavior
of Entropy and ᾱ is the same that was already observed in the test zones 1,2
and 3, suggesting a pattern that varies with the presence and/or type of snow
independently of the kind of target that the radar is illuminating: Entropy and
ᾱ decrease with the presence of snow, and are lower for the dry than for the
wet snow case. This phenomena might be indicating that the presence of snow
reduces the randomness level of the backscattering mechanism.
4.4.3 Urban Area Ground Truth

The idea of selecting a urban test zone was to look for a deterministic backscat-
tering target. It was expected to have dominance of the dihedral backscattering
mechanism, diminishing if any the effect of snow. The problem was the fact
that no information about the presence of snow is available for the PolSAR ac-
quisition days. However, taking into account the lower height of this test zone
and looking at the meteorologic data from Pla de Beret, which is just 5 Km
far from this Naut Aran test zone, one can infer that no or very few snow was
present in this urban area at the time when the PolSAR acquisitions were done.
When analyzing the polarimetric data for this test zone it is worth to point the
fact that opposite to what happened in the rest of test zones, the Pauli channel
HH-VV doesn’t show a decrease in the scenes from February and March with
respect to October scene, and that all the polarimetric channels are similar for
the three scenes. This fits the expected idea that urban structure would be the
dominant scattering mechanism in this test zone. Concerning H/A/ᾱ polari-
metric decomposition, the Entropy parameter indicates a pretty deterministic
scattering phenomena and shows few variation among different scenes. The
ᾱ parameter, lower than expected does not have either a significant variation
among the different scenes.
4.4.4 Overall Analysis

Finally, a global analysis of the polarimetric backscattering dataset as whole
was carried out. Analyzing simultaneously all the data contained in Tables 5,
6 and 7 as well as the H/A/ᾱ decomposition results (Table 8 ), some relevant
conclusions were derived as follows:
In March acquisition there was probably wet snow in the area illuminated
by the radar. All the polarimetric channels in all the test zones from
1 to 5 are around 5dB lower, which could be perfectly explained with
the wet snow hypothesis [5] which is reinforced with the interpretation of
the meteorologic data available from Pla de Beret during the acquisitions
period.
From [8], the idea that the presence of snow increases the volume scattering
mechanism intensity was tried to assess with our data. Only in test zone
4, the HV intensity was bigger for the dry snow scene than for the no
snow scene, probably due to a significantly thick snowpack due to the
higher altitude of this test zone. This is aligned with the fact that in [8]
a 2 m depth snowpack was used as input parameter of the simulations.
Therefore, since in the acquisitions used in this research work there was a
shallow snowpack, the volumetric mechanism increase is not an observed
effect.
Concerning Pauli surface component HH+VV, a slight increase of it can
be observed in most of the test zones. It happens for the test zones 1,2 and
2 2
4, 5 that: |HH + V V |drysnow > |HH + V V |nosnow (between 0.4dB and
3.5dB) which could be due to the reduction of the incidence angle in the
surface scattering with the presence of dry snow caused by the refraction
of the incident signal [9]. Test zone 3 shows almost no variation on this
component with the presence or not of dry snow.
Regarding Pauli double bounce component HH-VV, despite it could not
be observed a general trend, it happens for the test zones 1, 2 and 3 that:
2 2
|HH − V V |drysnow < |HH − V V |nosnow (approximately 1dB). It seems
however that this effect is not happening in test zones 4 and 5, but the
opposite.
Therefore, and despite the effect on HH-VV couldn’t be found in all the
test zones, putting together the information of the last two items, it came
up a new idea that could be used for dry snow detection, since with the
presence of dry snow, on one side the HH+VV component increases and
on the other side the HH-VV component decreases. This leads to:
2 2
∆ = |HH + V V | − |HH − V V | (4.3)
and it happens that ∆drysnow > ∆nosnow around 1dB for all the test zones
from 1 to 5 (see diff columns in Table 9 for February and October scenes
respectively). Though these results don’t show big differences between
dry snow case and no snow and therefore they are not robust, they might
be indicating a research field to be explored. Care has to be taken since
HH-VV is sensitive to polarization orientation angle changes, therefore
data should be corrected to compensate this effect.
About the polarimetric H/A/ᾱ decomposition, the most interesting fact
is realizing how the presence of snow reduces the H and ᾱ parameters in
all the test zones from 1 to 5. This suggests that the presence of snow
reduces the randomness of the scattering phenomena regardless of the
kind of underlying target. The reduction of the Entropy in the snowy
scenes can be easily related with the observed slight decrease in the HH-
VV component and the slight increase of the HH+VV component with the
presence of snow, since this involves HH+VV being a relatively stronger
scattering mechanism in the total Span contribution, thus a reduction of
the randomness of the backscattering.
Summarizing, it seems there is somehow sensitivity to the presence of snow,

and it could have arisen a new parameter that can be used as indicator of dry
snow presence. A good way of graphically visualizing the potential sensitivity
of snow in the polarimetric SAR images is comparing Figure 4.8 with following
Figure 4.9, that depicts ∆ in logarithmic scale for the three different scenes
(Feb-dry snow, Mar-wet snow and Oct-no snow). Notice that the differences
between scenes in Figure 4.8 are very slight whereas the differences in the ∆
inter-normalized images between the three different scenes can be clearly visually
appreciated, specially in the March scene. Note that the differences between dry
Figure 4.9: ∆ (dB) representation of the co-registerd data for the three different
scenes. Values range from -13dB to 23dB and axis units are pixels
snow (diff-17) image and no snow (diff-15) can be identified visually, despite not
being as obvious as in the wet snow case (diff-13).
Notice that the images displayed in Figure 4.9 have been obtained from
Speckle filtered Coherency matrices. A 7x7 Boxcar filter was used in order to
be able of constructing statistically robust ∆ images from the first and second
diagonal elements of the filtered Coherency matrix (T11 and T22 )
4.4
Table 5: 17/02/2011 Statistical Analysis

Polarimetric Analysis
Date 17/02/2011
Zone Magnitude HH HV VV HH+VV HH-VV Span
Mean Linear 2,53E+06 4,04E+05 2,70E+06 7,70E+05 2,02E+05 6,04E+06
Snow Monitoring
Var Linear 8,14E+13 1,74E+12 1,20E+14 8,87E+11 5,51E+10 3,94E+14

1 Mean Log (dB) 64,038 56,065 64,311 58,863 53,05 67,811
Var Log (dB) 139,11 122,39 140,81 119,48 107,41 145,96
2 Mean Log (dB) 65,97 58,719 65,538 67,867 61,505 69,553
Var Log (dB) 133,06 118,39 132,06 136,98 123,72 137,69
3 Mean Log (dB) 58,901 52,654 58,183 60,216 55,838 62,56
Var Log (dB) 119,67 107,15 117,94 122,25 113,33 124,01
4 Mean Log (dB) 62,325 53,732 61,459 64,067 57,453 65,538
Var Log (dB) 127,71 111,21 126,07 131,24 118,85 132,6
5 Mean Log (dB) 58,162 50,606 57,367 59,559 54,724 61,554
Var Log (dB) 119,62 103,54 118,91 123,48 111,49 124,75
67
6 Mean Log (dB) 70,655 60,262 68,612 69,983 69,51 73,226

Var Log (dB) 1 153,25 136,27 150,74 153,47 154,22 158,28
4.4

Date 13/03/2011
Snow Monitoring
Var Linear 2,9659E+11 15793202601 81120203381 4,8199E+11 18126328791 1,04046E+12

1 Mean Log (dB) 60,969 51,626 60,603 54,155 48,365 64,297
Var Log (dB) 135,3 111,37 134,04 117,53 98,665 140,58
2 Mean Log (dB) 65,83 59,945 65,75 67,566 62,736 69,805
Var Log (dB) 133,24 121,19 132,91 136,92 126,4 138,16
3 Mean Log (dB) 58,964 53,811 58,579 60,173 56,702 62,988
Var Log (dB) 120,27 109,35 119,94 123,07 115,3 125,53
4 Mean Log (dB) 57,99 49,441 57,324 59,904 52,821 61,289
Var Log (dB) 125,63 102,67 125,08 131,09 110,71 131,37
5 Mean Log (dB) 52,23 46,148 51,731 53,58 49,447 56,004
Var Log (dB) 108,91 97,183 108,44 112,32 102,75 114,67
68
6 Mean Log (dB) 70,174 59,505 68,069 69,85 68,549 72,696

Var Log (dB) 151,62 129,37 147,1 152,39 150,32 155,33
4.4

Date 15/10/2011
Mean Linear 1600100 304370 1638800 701260 240220 3847700
Snow Monitoring
Var Linear 1,9226E+13 7,2763E+11 2,5613E+13 1,0371E+12 88358000000 9,1368E+13

1 Mean Log (dB) 62,042 54,834 62,145 58,459 53,806 65,852
Var Log (dB) 132,84 118,62 134,08 120,16 109,46 139,61
Mean Linear 3663800 914150 3416000 5288100 1791600 8908100
2 Mean Log (dB) 65,639 59,61 65,335 67,233 62,533 69,498
Var Log (dB) 132,99 120,16 132,03 136,24 126,1 137,57
Mean Linear 822250 249840 750880 1080400 492700 2072800
Var Linear 1,0728E+12 99092000000 8,6427E+11 1,9407E+12 3,5009E+11 3,393E+12
3 Mean Log (dB) 59,15 53,977 58,756 60,336 56,926 63,166
Var Log (dB) 120,31 109,96 119,37 122,88 115,44 125,31
Mean Linear 788430 147820 693430 1147900 333990 1777500
Var Linear 1,6782E+12 38656000000 9,6634E+11 2,6501E+12 2,7761E+11 4,1919E+12
4 Mean Log (dB) 58,968 51,697 58,41 60,599 55,237 62,498
Var Log (dB) 122,25 105,87 119,85 124,23 114,43 126,22 9
Mean Linear 507760 125700 475740 707150 276350 1234900
Var Linear 5,2527E+11 21846000000 4,555E+11 1,1452E+12 1,1743E+11 1,6525E+12
5 Mean Log (dB) 57,057 50,993 56,774 58,495 54,415 60,916
Var Log (dB) 117,2 103,39 116,58 120,59 110,7 122,18
Mean Linear 8660300 806180 5649900 8931800 5378400 15923000
69
6 Mean Log (dB) 69,375 59,064 67,52 69,509 67,307 72,02

Var Log (dB) 150,27 128,63 146,92 152,16 147,25 154,26
4.4
Table 8: H/A/ᾱ decomposition statistical analysis

H/A/ᾱ Decomposition Analysis
Date 17/02/2011 13/03/2011 15/10/2011

Zone Magnitude H A ᾱ H A ᾱ H A ᾱ
Mean Linear 0,684 0,334 30,897 0,700 0,359 31,230 0,778 0,289 37,038
1
Snow Monitoring
Var Linear 0,007 0,014 26,570 0,009 0,014 36,875 0,006 0,013 31,059
Mean Linear 0,734 0,264 32,802 0,809 0,260 39,123 0,818 0,242 39,480
2 Var Linear 0,009 0,009 39,201 0,009 0,011 51,726 0,007 0,009 40,410
Mean Linear 0,815 0,266 39,772 0,861 0,258 44,387 0,868 0,264 44,482
3 Var Linear 0,004 0,011 26,810 0,004 0,011 25,697 0,003 0,011 24,190
Mean Linear 0,648 0,290 29,493 0,625 0,322 28,525 0,761 0,252 35,169
4 Var Linear 0,011 0,015 47,734 0,012 0,013 46,057 0,006 0,011 28,584
Mean Linear 0,785 0,298 37,700 0,790 0,338 38,866 0,832 0,260 41,072
5 Var Linear 0,007 0,014 39,170 0,008 0,014 50,032 0,006 0,010 39,224
Mean Linear 0,729 0,465 44,436 0,731 0,475 43,466 0,733 0,450 41,575
6 Var Linear 0,024 0,048 135,760 0,024 0,046 138,260 0,022 0,047 108,560
70
4.4
Table 9: First and Second Pauli Channels difference for the three scens
Date 17/02/2011 13/03/2011 15/10/2011

Zone Magnitude HH+VV HH-VV diff HH+VV HH-VV diff HH+VV HH-VV diff
1 Mean Log (dB) 58,863 53,05 5,813 54,155 48,365 5,79 58,459 53,806 4,653
Var Log (dB) 119,48 107,41 12,07 117,53 98,665 18,865 120,16 109,46 10,7
2 Mean Log (dB) 67,867 61,505 6,362 67,566 62,736 4,83 67,233 62,533 4,7
Snow Monitoring
Var Log (dB) 136,98 123,72 13,26 136,92 126,4 10,52 136,24 126,1 10,14
3 Mean Log (dB) 60,216 55,838 4,378 60,173 56,702 3,471 60,336 56,926 3,41
Var Log (dB) 122,25 113,33 8,92 123,07 115,3 7,77 122,88 115,44 7,44
4 Mean Log (dB) 64,067 57,453 6,614 59,904 52,821 7,083 60,599 55,237 5,362
Var Log (dB) 131,24 118,85 12,39 131,09 110,71 20,38 124,23 114,43 9,8
5 Mean Log (dB) 59,559 54,724 4,835 53,58 49,447 4,133 58,495 54,415 4,08
Var Log (dB) 123,48 111,49 11,99 112,32 102,75 9,57 120,59 110,7 9,89
6 Mean Log (dB) 69,983 69,51 0,473 69,85 68,549 1,301 69,509 67,307 2,202
Var Log (dB) 153,47 154,22 -0,75 152,39 150,32 2,07 152,16 147,25 4,91
71
4.4.5 Data Interpretation

After the snow monitoring data analysis, having suggested a new potential pa-
rameter into which drive our attention ∆ and the H/A/ᾱ decomposition to be
analysed in detail, there was the need of consolidating the preliminary findings
and assessing the robustness of what has been found. Activities were focused
into three different issues:
The polarimetric optimization technique used in [8] to optimize the change
detection based ratio was studied in detail in order to evaluate the possibil-
ity of applying something similar with the data available in this research.
It turned up that it’s not a really useful method when working with a single
frequency dataset, since this method is based on the use of multi-temporal
and multi-frequency images and we just had C-band images.
The zone that the acquisitions are illuminating is clearly mountainous,

which is a feature to be taken into account when analyzing and processing
the PolSAR data aiming to get some information about the snow. On one
side, there are the widely known SAR effects of shadowing and layover,
which can be easily overcame using a mask determining which are those
not affected pixels than can be used for retrieving geo-physical estimations.
Softwares such as ESA’s NEST can be used to geo-reference and obtain
a shadowing and layover mask for the obtained ∆ images in order to
know which are the reliable pixels. On the other side, though, there is
the PolSAR data problem of data correction for changes in radar cross-
sections caused by azimuth slopes. Surface slopes have two main effects
in PolSAR images [24]:
– Change of the radar cross-section per unit area, which is compensated
with the radiometric slope correction (sin (θilocal )) to correct the
backscattering coefficient value. This operation can be easily carried
out with the terrain correction operation with NEST.
– Polarization states are also affected due to rotation of the scattering
matrix. If the facet surface normal is not in the incidence plane of
the radar signal, the horizontal polarization vector is no longer par-
allel to the surface involving a distortion in the polarization state
of the incident wave. The amount of slope-induced shift in the po-
larization orientation angle can be visualized as a rotation of the
vertical-horizontal basis vector about the Line-of-sight (LOS). The
basic principle to correct this effect is to rotate data about the LOS
by the negative of the polarization orientation shift induced by ter-
rain azimuthal slopes.
The circular polarization method for polarization orientation angle com-
pensation proposed in [24] has been implemented and applied to the avail-
able data. It can be summarized as follows: the shift in the polarization
orientation angle Θ has to be estimated in order to be afterwards com-
pensated.
∗
Θ = [Arg (hSRR SLL i) + π] /4 (4.4)
where the Arg() based on the circular polarization copol components can
be obtained from the linear basis as:
 
∗
−4Re (h(SHH − S ) S i)
∗
Arg (hSRR SLL i) = tan−1  D E V V D HV E  (4.5)
2 2
− |SHH − SV V | + 4 |SHV |
In order to compensate a simple rotation is applied to the scattering ma-

trix:

0 cos (Θ) sin (Θ) Shh Shv cos (Θ) −sin (Θ)
S = (4.6)
−sin (Θ) cos (Θ) Svh Svv sin (Θ) cos (Θ)
The above method was applied to the PolSAR dataset using a 3x3 box
pixel averaging in the estimation of the angle shift, in order to reduce
some noise in the estimated values (with this average, slopes of around
24 m2 were obtained). Resulting data analyzed in the presented 6 test
zones for the sake of comparison, was quite similar to the original one.
It was realized that the angle shift about the LOS for the different test
zones is similar. The real change between the test zones is in the incidence
angle and this does not almost affect the polarization state. HV channel
is the only one that suffered a drop of around 1dB in all the test zones
and temporal scenes, indicating that it was probably overestimated in
previous analysis. However, since the change applied the same way for
all the zones and scenes and since the data analysis is based on change
detection techniques, the ratios computed would remain very similar and
therefore the terrain slope correction for PolSAR data was assumed a
second order phenomena, to be taken into account in further research
work with a wider scope.
Finally, in the results interpretation, the main topics of the work carried
out were on one hand looking for a reasonable physical description of the ∆
parameter, showing an increase in the dry snowy scenes, and on the other
hand the deep analysis of the H/A/ᾱ decomposition, and if possible, find-
ing synergies between these two topics. Indeed, the fact that the double
bounce polarimetric component was surprisingly important, with respect
to surface component, in alpine meadow test zones was the initial point
of the survey. Having such significant levels of double bounce scattering
intensity, together with the high Entropy (suggesting some depolariza-
tion) and ᾱ values, that classified the alpine meadows in the H/ᾱ plane
as random surface and/or anisotropic particles volume diffusion, it was
questioned the initial assumption that the snow back-scattering problem
had to be focused mainly on T11 (surface) and T33 (volume). Therefore,
the initial hypothesis that alpine meadows at C-band could be modeled
with Bragg scattering was concluded not to be correct. In fact, this can be
clearly stated having a look at the pairs H/ᾱ for each test zone, figuring
out that none falls into the Bragg surface zone of Figure 4.10, regardless
of the time of the year scene we work with.
Deploying in the H/ᾱ plane the pairs of Entropy and ᾱ values of the 6
different test zones present in Table 8 for the three different scenes, it
came up a remarkable result. As it can be observed in Figure 4.11 all
Figure 4.10: :H/ᾱ plane [3]
the test zones show a similar increasing slope function except the 6th one
(urban zone). Notice that the dots in the bottom left of each function
correspond to the March scene and the ones in the top right part of the
functions are from the October scene. This means that the presence of
dry snow with respect to the no snow scene decreases the Entropy and the
ᾱ parameters at C-band, thus decreases the randomness of the scattering
phenomena and surprisingly moves the ᾱ parameter to a more random
surface scattering phenomena. In yellow, as expected due to its more
deterministic nature, the urban test zone shows no Entropy variation and
almost to ᾱ variation among the different scenes. Recall that scenes 1
to 5 have different characteristics of height, surface coverage and slope,
and therefore the scattering phenomena in each zone might be different.
However, concerning the presence of snow, the resulting H/ᾱ plot is similar
for all the test zones. This remarkable behavior found in the curves for
test zones 1 to 5 might be further investigated as a means of determining
dry snow covered pixel with a change detection based technique.
With the Bragg surface hypothesis discarded, since probably the rough-
ness of the alpine meadows at C-band does not fulfill the Bragg surface
condition of KS < 0.3, where K is the wave number and s is the rms sur-
face roughness, the X-Bragg model hypothesis was then considered. The
model to be used has to be suitable to deal with rougher surfaces and the
X-Bragg [25, 4] consists of a surface scattering model that accounts for
high rms roughness surfaces. It introduces roughness disturbance model-
ing the surface as a reflection symmetric depolarizer by rotating the bragg
coherency matrix [T ] an angle β about the LOS. The width of the proba-
bility distribution function of β of the model is proportional to the surface
roughness. Opposite to the Bragg Coherency matrix case, there is cross-
polarization, and therefore the polarimetric coherence is ≤ 1. Assuming
an X-Bragg surface, which involves rough surfaces enabling more than just
surface scattering mechanisms, the presence of double bounce scattering
in the test zones could be explained.
4.5 Preliminary Results 75
Figure 4.11: :H/ᾱ plane behavior with Snow
As final for this section, it could be stated from the empirical observations
in this work, that the presence of dry snow has a smoothing effect on the
underlying surface, involving the slight decrease of double bounce scatter-
ing mechanism and the slight increase of the surface scattering mechanism,
with a reduction of the Entropy parameter.
4.5 Preliminary Results

In order to evaluate the potential outcome of the introduced ∆ parameter (see
eq. 4.3) and H/ᾱ trend found in Figure 4.11, for its use as dry snow identifiers,
which is a significant challenge due to the reasons already described in pre-
vious sections about C-band microwaves reduced interaction with snow, some
preliminary results will be presented in this section. It will also be presented a
classifier method based on the H/ᾱ trend found in Figure 4.11 for identifying dry
snow, and results of both classifiers will be compared. Results presented here-
after are the most relevant ones after some semi-empirical work carried out with
the reduced dataset and ground truth budget available. In order to asses the
mentioned potential of polarimetric C-band SAR signal for quantitative snow
remote sensing, based on the two ideas which will be presented henceforth, a
deeper research work is needed, which is unfortunately out of the scope of this
master thesis.
4.5.1 Proposed ∆ Parameter Results

Concerning ∆ parameter, results correspond to two different levels of filtering
for Speckle reduction in the obtaining process of the parameter. The first trials
in the computation of the ∆ parameter for the three temporal different scenes
(February, March and October) were done using a 7x7 pixel Boxcar filter for
Speckle noise reduction. Results were already displayed in Figure 4.9. Using
these three ∆ data matrices in logarithmic scale, two difference products were
obtained:
∆a = ∆0217 − ∆1015 ∆b = ∆0313 − ∆1015 (4.7)

where ∆a represents the difference between the dry snow scene and the non
snow ground truth scene, and ∆b represents the difference between the wet snow
scene and the non snow ground truth scene. See in Figure 4.12 what these two
differences look like.
Figure 4.12: :7x7 Boxcar obtained ∆ differences in dB in Pla de Beret Zone for
dry and wet snow scenes respect to no snow ground truth scene. Axis units are
pixels.
First of all note the difference between the dry and the wet scenes differences,
suggesting sensitivity to the kind of snow. Notice as well, concerning differences
between ∆a and ∆b , how the wet snow scene difference shows values lower than
those of the dry snow scene difference. Individually analyzed, looking in the
∆a image at Pla de Beret zone, around the pixel (x=1700, y=2700), with an
upper right diagonal shape, it can be noticed that most values are above 1dB,
indicating dry snow covered area. Indeed, having a look at the ∆0217 image
in Figure 4.9 it can be sensed the shape of the mountains and the valleys in
between them, in the zone illuminated by the radar, as it can be confirmed if
attention is put in the comparison of ∆ in Figure 4.9 and Figure 4.13, which
is the topographic map of the PolSAR data covered zone. In order to better
appreciate whether this topography like shape exists, indicating snow in the
higher parts of the map, i.e. the mountains, a magnitude filter was applied to
∆a with a binary mask showing those pixels with a value higher than 1dB, that
might be dry snow covered. The result of this masking can be seen in Figure
4.14.
It is worth to point that the threshold at 1dB has been decided as such with
a very few amount of ground truth data. Recall that this value was adopted
out of the results in Table 9, and further work should be done with a complete
ground truth covering the whole image, in order to optimize this threshold for
the classifier. In addition some empirical essays have been conducted looking
for the value of the threshold showing better visual results with regards to the
topography map in Figure 4.13. Again, it is worth to mention that a layover-
shadowing mask should be applied to this obtained above-1dB filtered image
(Figure 4.14), in order to assess which are the pixels with reliable values and
which are not. Nonetheless, it can be clearly identified in Figure 4.9 the valley
Figure 4.13: :Topography map of the imaged zone in the research
Figure 4.14: :Diff-a image filtered to those pixels with a value higher than 1dB,
indicating the presence of dry snow. White and black pixels indicate snow and
no snow respectively. Axis units are pixels.
Vall d’Aran in the bottom-left area of the figure images, as well as the shape of
the mountains that we suppose covered by dry snow.
As the reader might have already seen, the results displayed in the previous
images 4.12 and 4.14 are quite noisy. In order to try to reduce the noise level,
the ∆ parameter was recomputed using a 11x11 pixel Boxcar filter. Results
from the more filtered ∆ are shown as follows in Figure 4.15.
Figure 4.15: :∆ (11x11 Boxcar used) representation of the co-registerd data for
the three different scenes. Values range from -13dB to 23dB and axis units are
pixels
Computing again the differences of ∆ as in Eq 4.7, for the dry and wet snow
scenes with respect to the no snow ground truth scene, the results, shown in
Figure 4.16, are quite similar to those in Figure 4.12 obtained with a smaller
Boxcar filter. In this case, the whole image is shown so that the geographical
valleys can be appreciated in the bottom-left part of the ∆a image on the left
of the Figure 4.16. Again, manually adjusting a threshold so that the masked
image looks like similar to the topography of the illuminated area, it has been
optimized the dry snow mask shown in Figure 4.17. Notice that its aspect is
less noisy than the one in Figure 4.14, as expected taking into account that
more averaging has been applied to compute the ∆ matrices that led to this
difference image, however the dry snow mask still looks too noisy. Probably a
better Speckle filtering method would lead to better results, since Boxcar kind
of filter has very poor results in the non-homogeneities of the image. Further
word should include better Speckle filtering techniques in order to compute more
reliably the ∆ parameter.
Figure 4.16: :11x11 Boxcar obtained ∆ differences in dB for dry and wet snow
scenes respect to no snow ground truth scene. Axis units are pixels.
Figure 4.17: :Dry snow covered pixels mask. White and black pixels indicate
snow and no snow respectively. Axis units are pixels.
4.5.2 Proposed H/ᾱ method Results

Concerning the proposed methodology for dry snow identification based on the
H/ᾱ pairs of values, it consists of exploiting the trend observed in Figure 4.11 for
test zones 1 to 5, in which the presence of dry snow seems to reduce the Entropy
as well as the ᾱ values. After some empirical observation of the decrease of the
Entropy and ᾱ values in these test zones, an initial proposed decision rule has
been used to create a dry snow identifier. This preliminary decision rule is
applied to each pixel of the image in order to obtain a dry snow cover map,
shown in Figure 4.18. The rule applied to determine whether a pixel is covered
with dry snow or not is described in Eq 4.8. The values of the no snow to
dry snow thresholds for Entropy (0.925) and ᾱ (0.9) have been decided as such
observing the values represented in Figure 4.11.
if (Hdrysnow ≤ 0.935Hnosnow and αdrysnow ≤ 0.91αnosnow ) then dry snow

(4.8)
Notice that for all the test zones except the urban one, the values of Entropy
decrease more or less a 6.5% in the transition from no snow to dry snow cover.
In the case of the ᾱ values, the decrease is a little bit stronger up to a 9%. Note
that this proposed method needs temporal series data, since it is based on the
principle of change detection. Needless to say that these two ratio values have
been adjusted as the result of a reduced dataset and further work should include
optimization of these threshold values, as well as assessing the validity of such a
decision rule, by means of a wider dataset and/or a wider ground truth image of
snow covered area. Nevertheless, results obtained in Figure 4.18 seem coherent,
aligned with those obtained with the ∆ parameter method, suggesting the non
snowed valleys in the bottom-left part of the image.
Figure 4.18: :H/ᾱ based dry snow covered pixels mask. White and black pixels
indicate snow and no snow respectively. Axis units are pixels.
4.5.3 Preliminary Results Conclusions

Analyzing the overall preliminary results it seems clear that the ∆ parameter
presents sensitivity to dry snow presence. It is important to point out that
there is a significant difference in ∆ depending on whether illuminating dry
or wet snow (see Figures 4.16 and 4.12). However, the lack of a whole image
ground truth together with the small dataset doesn’t permit a necessary further
survey in order to assess the potential of this ∆ parameter, optimizing detection
and classification thresholds, developing resulting masks filtering and further

investigating if there is any kind of quantitative approach potential.
On the other hand, the H/ᾱ based classifier seems to provide also coherent
results in the identification of dry snow covered pixels as it has been shown in
Figure 4.18. Indeed, notice the similarity of both methods results comparing
images 4.18 and 4.17. As in the ∆ parameter method case, the reduced dataset
and ground truth do not permit a deeper assessment on the potential of this
H/ᾱ change detection method for dry snow cover determination.
Thus, the final outcome of the preliminary results could be the proposal of
two new polarimetric data based methods that suggest sensitivity to dry snow
in C-band: ∆ parameter and H/ᾱ method, which might deserve further research
work to better explore the possibilities of remotely sensing snow with PolSAR
C-band data. Actually, it would be interesting to explore the synergies of both
methods as a means for snow remote sensing in C-band.
82
5 Conclusions and Further Steps

After the research work carried out in the frame of this Master Thesis, the
lessons learned and conclusions about the main outcome and the future steps
are listed hereafter in this section. It is worth to recall that the triggering idea
of this research work was to assess the potential of C-band polarimetric SAR
in snow remote sensing. Despite it is known that C-band is not optimal for
snow applications, there are spacecrafts currently available and planned for the
future carrying C-band polarimetric SAR sensors, and therefore any step further
on this topic could relatively easily come off an operational remote sensing
application or service. As a general idea after the work carried out, it can be
stated that C-band fully polarimetric data has potential in the snow remote
sensing field. A wider research campaign should be carried out in order to have
a richer dataset, with more heterogeneous scenes in terms of snow depth and
density, as well as test zone morphologies, that would allow to assess the real
potential of this research thread.
The main conclusions and ideas for future work are summarized as follows:
There seems to be some sensitivity to the presence of dry snow in the

polarimetric backscattering data, which can be materialized in the two
presented parameters/methods: ∆ and H/ᾱ based classifier. There is
also, obviously, sensitivity to the presence of wet snow, though this has
nothing to do with polarimetry in its first order of magnitude, since it is
well know from the literature that wet snow presence at C-band involves
a significant drop of the total backscattered power.
83
Better Speckle filtering techniques are needed to further assess the poten-
tial trajectory of the presented approaches. Using less noisy Coherency
Matrix diagonal components T11 and T22 , the resulting parameters would
be much less noisier than they are right now. Advanced polarimetric
speckle filtering techniques are needed in order to be able to filter a very
in-homogeneous data such as the one used for this research work, and
being sure that the source data used to compute ∆ and H/A/ᾱ decom-
position is optimized from the noise reduction point of view. In such a
case, results in detecting dry snow covered area and differentiating with
wet snow would give a more realistic idea of the potential extent of these
polarimetric presented parameters or methods.
It might be appropriate as well to think about some kind of post-filtering
technique for the obtained results using simple decision thresholds. Some
computational intelligence techniques for classification might be studied
in order to cluster in different well defined zones, the obtained boolean dry
snow maps. Neural Networks might be useful in this case since probably
some patterns of ancillary data variables can be found in the contour
conditions of each pixel bringing valuable information for deciding about
the presence and kind of snow.
A wider dataset and ground truth would allow a much more realistic as-
sessment of the potential of the classification results. This means having
more fully polarimetric C-band SAR images, covering many different con-
ditions snow-packs, as well as an extended ground truth information about
the snow-pack and the snow covered area, allowing a robust validation of
the results obtained. An interesting first approach could be using MODIS
hyper-spectral data of the Pyrenees zone to compute the Normalized Dif-
ferential Snow Index (NDSI) with MODIS bands 4 and 6. This would
provide with an initial ground truth map of Snow Covered Area (SCA),
although with very poor spatial resolution of 500m per pixel.
Finally, if it comes up that the presented approaches show good results in

the qualitative field, used as a classifiers. A wider experimental campaign
could be specially designed to evaluate the quantitative potential of the
classifying methods. It might happen that the magnitude of the presented
parameters depends on some snow feature such as density and or depth,
paving the way for a quantitative snow remote sensing with polarimet-
ric C-band SAR data, which would guarantee the feasibility of a stable
snow monitoring service given the fact that polarimetric space-borne SAR
continuity is guaranteed by Radarsat-2 future constellation and Sentinel-1
(dual-pol coherent).
Getting back to the initial goal of this Master thesis, recall that this targeted
Snow Water Equivalent (SWE). The idea was quantitatively sensing some snow
parameter related with SWE, such as density or permittivity, using Polarimet-
ric SAR data at C-band. Unfortunately no robust conclusion can be extracted
about this main goal, since despite C-band polarimetric SAR data presented
sensitivity to the presence of snow, the research carried out so far can not assess
the potential for quantitative snow remote sensing. Further research needs to
be conducted on this issue, with a more ambitious approach, in order to have a
much wider dataset, so that snow remote sensing potential with C-band polari-
metric SAR data can be correctly determined. The global conclusion achieved
84
after this Master thesis research work is that, despite the widely extended idea
that C-band microwaves are not specially sensitive to the presence of snow,
Polarimetry adds some additional information whose contribution is yet to be
carefully investigated because it seems that it increases the snow remote sensing
capabilities of C-band SAR.
REFERENCES 85
References
[1] Christian Matzler, Applications of the interaction of Microwaves with the
Natural Snow Cover, Remote Sensig Reviews, Vol 2, 1987. 28, 30, 34, 35,
37, 48, 53
[2] Fawwaz T. Ulaby, Richard K. Moore, Adrian K. Fung, Microwave Remote
Sensing: Active and Passive, from Theory to Applications,Artech House,
1986. 34, 35, 36, 37, 48, 54, 55
[3] Jong-Sen Lee, Eric Pottier, Polarimetric Radar Imaging. From basics to
Applications, CRC Press, 2009. 11, 12, 21, 30, 38, 39, 43, 50, 73
[4] S. R. Cloude, Polarisation. Applications in Remote Sensing , Oxford Uni-
versity Press, 2010. 30, 39, 43, 73
[5] Tazio Strozzi, Shi, Christion Matzler, Backscattering Measurements of
Alpine Snowcovers at 5.3 and 35 GHz, IEEE Transactions on Geoscience
and Remote Sensing, May 1998. 38, 53, 55, 63
[6] Singh, G. Venkataraman, G., Snow density estimation using polarimetric
ASAR data, Recent Advances in Microwave Theory and Applications, 2008.
MICROWAVE 2008. International Conference on . 35, 39, 41, 44
[7] Audrey Martini, Teledetection d’un couvert neigeux en milieux alpins a par-
tir de donnes SAR polarimtriques multi-frquentielles et multi-temporelles,
PhD Thesis, Universite Rennes, 2005. 37, 38, 39, 48
[8] A. Martini et al., Polarimetric Study of Scattering from Dry Snow Cover
in Alpine Areas, Proceedings IGARSS 2003. 53, 63, 71
[9] Jiancheng Shi, Jeff Dozier, Estimation of Snow Water Equivalent using
SIR-C/X SAR Part I: Inferring snow density and subsurface properties,
IEEE Transactions on Geoscience and Remote Sensing, Nov 2000. 37, 39,
40, 41, 53, 64
[10] Jiancheng Shi, Jeff Dozier, Estimation of Snow Water Equivalent using
SIR-C/X SAR Part II: Inferring snow depth and particle size, IEEE Trans-
actions on Geoscience and Remote Sensing, Nov 2000. 39
[11] Monique Bernier et al., The Potential of Time Series of C-Band SAR Data
to Monitor dry Shallow Snow Cover, IEEE Transactions on Geoscience and
Remote Sensing, Jan 1998. 39, 42, 47
[12] Nicolas Longepe et al., snow-pack Characterization in Mountainous Regions
Using C-Band SAR Data and a Meteorological Model, IEEE Transactions
on Geoscience and Remote Sensing, Feb 2009. 39, 41
[13] Zhen Li et al., Estimation of Snow Density with L-Band Polarimetric SAR
data, IEEE IGARS Proceedings, 2000. 41
[14] Singh, G. Venkataraman, G., Snow density estimation using Polarimetric
ASAR data, IEEE IGARS Proceedings, 2009. 39, 41
[15] Snehmani et al., Development of an inversion algorithm for dry snow
density estimation and its application with ENVISAT-ASAR dual co-
polarization datal, Geocarto International Taylor & Francis, Dec 2010. 39,
41, 45
REFERENCES 86
[16] Jacob J. van Zyl et al, Requirements for model-based polarimetric decom-
positions, IEEE IGARS Proceedings, 2008. 44
[17] Jayanti J. Sharma et al., Polarimetric Decomposition Over Glacier Ice Us-
ing Long-Wavelength Airborne PolSAR, IEEE Transactions on Geoscience
and Remote Sensing, Jan 2011. 37, 44, 45
[18] Jayanti J. Sharma et al., Characterization of Oriented Volumes in Glacier
Ice and Extinction Inversion with PolInSAR, Proceedings PolInSAR, 2009.
[19] Motofumi Arii et al., A general Characterization for Polarimetric Scat-

tering From Vegetation Canopies, IEEE Transactions on Geoscience and
Remote Sensing, Sep 2010. 46
[20] Helmut Rott et al., Snow Mass Retreival by Means of SAR Interferometry,
Proceedings of Fringe03 ESA, Dec 2003. 44
[21] Maxim Neuman et al., Forest Parameter Retreival using a General Repeat-
pass Polarimetric Interferometric Vegetation Mode, IEEE IGARS Proceed-
ings, 2009. 46
[22] Yoshio Yamaguchi et al., Four-Component Scattering Model for Polari-
metric SAR Image Decomposition, IEEE Transactions on Geoscience and
Remote Sensing, Aug 2005. Geophys. Res. Lett., Vol. 31, 2004. 46
[23] Nagler, T. and Rott, H., Retrieval of wet snow by means of multitemporal
SAR data, IEEE Transactions on Geoscience and Remote Sensing, Aug
2002. 30
[24] Jong-Sen Lee et al., Polrimetric SAR Data Compensation for Terrain Az-
imuth Slope Variation, IEEE Transactions on Geoscience and Remote Sens-
ing, Sep 2000. 71
[25] Irena Hajnsek et al., Inversion of Surface Parameters from Polarimetric
SAR, IEEE Transactions on Geoscience and Remote Sensing, Apr 2003. 73
[26] Fawwaz T. Ulaby,M. Craig Dobson, Handbook of Radar Scattering Statistics

for Terrain,Artech House, 1989. 61

Master Thesis Xavier Banque

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Master Thesis Xavier Banque

Uploaded by

Copyright:

Available Formats

Snow Monitoring over

Author: Xavier Banqué Casanovas

3 SAR Polarimetry for Snow Remote Sensing 28

4 C-band fully Polaimetric SAR measurements over Mountainous

4.3 Snow Backscattering Results Analysis . . . . . . . . . . . . . . . 53

5 Conclusions and Further Steps 81

remote sensing service based on polarimetric SAR data.

 Introduction The present section, it briefly introduces the Master Thesis

 Polarimetry Introduction In this section a theoretical introduction to

 SAR Polarimetry for Snow Remote Sensing This section presents

polarimetric approach. SAR polarimetry becomes specially interesting when

2.2 Introduction to Polarimetry

In a general way, a monochromatic plane wave with constant complex amplitude

where the wave number k is complex magnitude comprising the propagation

where w is the angular frequency,k = 2π/λ is the wave number, δx , δy represent

Figure 2.1: Monochromatic plane wave spatial evolution [3]

The relationship between amplitudes an phases of both sinusoids determines

2.2.1 Polarization Ellipse

 and |τ | the ellipse aperture or ellipticity that has the expression:

Figure 2.2: Polarization ellipse characterization[3]

2.2.2 Jones Vector

that can be written in a matrix as a function of the parameters that characterize

Two different polarization states are orthogonal if their Jones vectors E1 E2

According to the definition of a Jones vector E (x̂,ŷ) given by 12, an orthog-

E (x̂,ŷ)⊥ = AU2 (φ) U2 (τ ) U2 (α) ŷ (2.14)

u = U2 (φ) U2 (τ ) U2 (α) x̂ and u⊥ = U2 (φ) U2 (τ ) U2 (α) ŷ (2.16)

Once the response of a scatterer has been acquired in a polarization basis, it

2.2.3 Stokes Vector

with the folowing relation between the four Stokes parameters:

g02 = g12 + g22 + g32 (2.18)

where each of the parameters is:

2.2.4 Covariance Matrix

 DoP = 1 is fully polarized wave.

Wave Anisotropy and wave Entropy When computing the eigenvectors

 For a completely unpolarized wave λ2 = λ1 : Hw = 1 and Aw = 0

 and for a partially polarized wave λ1 6= λ2 ≥ 0: 0 ≤ Hw ≤ 1 and 0 ≤

2.3 Scattering operators and polarimetric characteriza-

PT GT (θ, φ) AER (θ, φ)

where PR represents the power detected at the receiver, PT is the transmit-

The radar cross-section depends on significant number of parameters, such

2.3.2 Scattering matrix

It has to be noted that in general, polarization unit vectors ûR R

2.3.3 Polarimetric Coherency and Covariance Matrices

2.3.4 Kennaugh Matrix. Huynen Decomposition

above presented matrix polarimetric operators, such as the scattering matrix

or in an equivalent way in terms of the Kennaugh matrix, in such a way that

The optimization polarization problem consists of finding such polarization

2.3.5 Canonical Scattering mechanisms

 Cartesian polarization basis (ĥ, v̂), linear horizontal an vertical polariza-

 Circular polarization basis

2.4 Polarimetric SAR Speckle Noise

Figure 2.3: Single scatterers contribution to the received signal [3]

2.4.2 Polarimetric Speckle Statistics

2.5 Polarimetric decomposition theorems

said to be distributed or incoherent, and cannot be related anymore to a coher-

2.5.1 Huynen decomposition

 an effective single target T0 , dominant scattering mechanism, which is

 a residual target TN described by the remaining 4 parameters.

2.5.2 Freeman and Durden decomposition

 The double bounce scattering component is determined by a dihedral scat-

2.5.3 Eigenvector-based decomposition

2.5.4 Pauli decomposition. Coherent Decomposition

Figure 2.4: Polarimetric Pauli decomposition (Great Aletsch glacier 2003,

Introduction The present section, it briefly introduces the Master Thesis

Polarimetry Introduction In this section a theoretical introduction to

SAR Polarimetry for Snow Remote Sensing This section presents

and |τ | the ellipse aperture or ellipticity that has the expression:

DoP = 1 is fully polarized wave.

For a completely unpolarized wave λ2 = λ1 : Hw = 1 and Aw = 0

and for a partially polarized wave λ1 6= λ2 ≥ 0: 0 ≤ Hw ≤ 1 and 0 ≤

Cartesian polarization basis (ĥ, v̂), linear horizontal an vertical polariza-

Circular polarization basis

an effective single target T0 , dominant scattering mechanism, which is

a residual target TN described by the remaining 4 parameters.

The double bounce scattering component is determined by a dihedral scat-

Terrain and Land-Use classification

Forest areas mapping and classification. Information such as biomass,

Snow covered areas monitoring, allowing the retrieving of wetness and

Surface geophysical parameters estimation such as soil moisture and rough-

Ocean monitoring, estimating parameters like the SWH or currents.

Damage assessment, in for example forecasting floods from frozen rivers.

Water Content (liquid): Determines the dielectric properties and it is very

The Rennes University 1 with a wide experience in SAR polarimetry, has

The incident wavelength at snow-ground interface is shorter than the wave-

Out of the ratio winter vs spring backscattering the thermal resistance of

From the PolSAR covariance C or coherency T matrix ,the polarimetric

The assumption of reflection symmetry, which in the case of ice scatter-

Three fully polarimetric C-band RADARSAT-2 Fine Quadpol mode im-

Results of two ground truth measurement campaigns registering snow