How Does Multiple Scattering Affect the Spaceborne W-Band Radar Measurements at Ranges Close to and Crossing the Sea-Surface Range?
ABSTRACT A radar simulator capable of treating multiple-breakscattering effects has been upgraded to include the interaction with a Kirchoff surface, which realistically reproduces the effect of water surfaces. Multiple-scattering effects explain in a straightforward way some peculiar features of the first images delivered by the 94-GHz cloud-profiling radar onboard the CloudSat, overpassing precipitating systems. The reflectivity profiles without the usual peaks at surface range are found to be distinctive signatures of strong multiple scattering. Moreover, multiple scattering is responsible for producing long signal tails at apparent ranges far below the surface with a strong sensitivity on the microphysical assumptions of the icy segment of the cloud. The estimates of multiple-scattering enhancement at surface and close to the surface range and the saturation levels for simplified precipitating profiles for both CloudSat and EarthCARE configurations are provided.
- Journal of Atmospheric and Oceanic Technology - J ATMOS OCEAN TECHNOL. 01/2004; 21(9).
- [show abstract] [hide abstract]
ABSTRACT: The mirror image (MI) rain echo, received through the double reflection of the radar pulse from the surface, may provide useful information in estimating the rainfall rate from airborne and spaceborne weather radars. However, because of the complicated scattering mechanisms involving the surface and rain and the relatively small amount of measured data, studies of the MI effect have been few. In this paper, a more rigorous model of the MI return power has been constructed that yields the co-and cross-polarized components of the MI and bistatic returns as a functions of the radar parameters (antenna beamwidth and radar altitude) and the scattering properties of the rain and surface. As a test of the model, the mirror image return, as estimated from theory, is compared with the measured MI range profile, and reasonably good agreement is obtained. For meteorological applications, algorithms for estimation of the rain path attenuation are developed based on the difference of the direct and MI returns at the same distance from the surface. The accuracies of the algorithms are analyzed and compared with those of the surface reference technique (SRT) through the simulations for airborne and spaceborne [the case of the Tropical Rain Measuring Mission (TRRM)] geometriesIEEE Transactions on Geoscience and Remote Sensing 04/1999; · 3.47 Impact Factor
1644IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 46, NO. 6, JUNE 2008
How Does Multiple Scattering Affect the Spaceborne
W-Band Radar Measurements at Ranges Close to
and Crossing the Sea-Surface Range?
Alessandro Battaglia and Clemens Simmer
Abstract—A radar simulator capable of treating multiple-
breakscattering effects has been upgraded to include the inter-
action with a Kirchoff surface, which realistically reproduces
the effect of water surfaces. Multiple-scattering effects explain
in a straightforward way some peculiar features of the first
images delivered by the 94-GHz cloud-profiling radar onboard
the CloudSat, overpassing precipitating systems. The reflectivity
profiles without the usual peaks at surface range are found to
be distinctive signatures of strong multiple scattering. Moreover,
multiple scattering is responsible for producing long signal tails at
apparent ranges far below the surface with a strong sensitivity on
the microphysical assumptions of the icy segment of the cloud. The
estimates of multiple-scattering enhancement at surface and close
to the surface range and the saturation levels for simplified precip-
itating profiles for both CloudSat and EarthCARE configurations
Index Terms—Millimeter-wave radar, radar application, radar
theory, rainfall effects, scattering.
In the meantime, EarthCARE, which is a joint European–
Japanese Earth Explorer mission, has progressed into Phase B
of its development (see http://www.esa.int/esaLPearthcare.
html). In order to achieve cloud-profiling capabilities, both
CloudSat and EarthCARE missions deploy/will deploy 94-GHz
radars with a 0.1◦beamwidth antenna. The EarthCARE cloud-
profiling radar (CPR) will have higher sensitivity (−36 dBZ
versus −26 dBZ) and smaller footprints (due to the smaller
flying altitude: 450 versus 705 km) compared with CloudSat.
The interpretation of the echoes from radar systems is
usually done in terms of the so-called single-scattering (SS)
approximation. Exceptions are hail spikes  and mirror im-
ages (MIs) (e.g., observed by the Tropical Rainfall Measuring
Mission–Precipitation Radar ), which have been interpreted
as three-body scattering with the surface as a scattering target.
The rigorous models of the “flare echoes” and the cross- and
copolarized MI returns have been constructed by Zrniæ 
LOUDSAT  has been successfully acquiring data since
May 2006 (see http://www.CloudSat.cira.colostate.edu).
Manuscript received April 30, 2007; revised September 10, 2007.
The authors are with the Meteorological Institute, University of Bonn, Bonn
Color versions of one or more of the figures in this paper are available online
Digital Object Identifier 10.1109/TGRS.2008.916085
and Liao et al. , respectively. Some authors – have
recently stressed the importance of multiple-scattering (MS)
effects when dealing with high-frequency spaceborne radars
like CloudSat. They have suggested that MS can be detected
either by the use of linear depolarization ratio signatures, em-
ploying multifield-view antennas (an example of such concept
is demonstrated in ), or the synergy with measurements
from other sensors. However, as we will demonstrate, evidence
of MS can already be gained by analyzing reflectivity profiles
sensed by the CloudSat CPR. For instance, many profiles
extracted from CloudSat overpasses over tall precipitating sys-
tems show the following features which are not interpretable in
the frame of SS models.
1) At ranges corresponding to the surface, there is no
“discontinuity” peak in the reflectivity signal.
2) At apparent ranges below the surface, the reflectivity
profiles exhibit long tails (which are not explainable in
the frame of MI theory).
An example of the first phenomenon is shown in Fig. 1, which
depicts the reflectivity profiles from an overpass of CloudSat
over a large precipitating system in the East Pacific. Two
regions are highlighted. One is between the black arrows where
huge MS effects are believed to be present, and another one
is between the green arrows where the role played by the
MS is less relevant. The corresponding profiles are plotted
all together in the lower panel in Fig. 1. All black profiles
present no distinctive transition at the surface range (with
reflectivities between −25 and −15 dBZ) with a long tail at
ranges longer than the surface and a layer between 4 and 8 km
with reflectivities up to 15 dBZ. In the green profiles, the
surface echo is clearly detectable, and the reflectivities from the
ice segment are noticeably lower than in the former interval.
Moreover, for ranges below the surface, the reflectivity signal
is rapidly decreasing with a slope much steeper than the black
In some exceptional cases (tall systems with huge quantities
of ice aloft, e.g., the Tropical Storm 06W on July 24, 2006,
see Fig. 2), the MS signal below the surface can prolong that
far below the surface that it appears as a second-trip echo (the
CPR has a pulse repetition frequency that is equal to 4300 Hz,
leading to a maximum unambiguous range of around 35 km).
The correct interpretation of the signal coming from and
below the surface is crucial when estimating path attenuation
by the surface reference technique ,  or MI analyses
0196-2892/$25.00 © 2008 IEEE
BATTAGLIA AND SIMMER: HOW DOES MULTIPLE SCATTERING AFFECT THE W-BAND RADAR MEASUREMENTS?1645
over the East Pacific, August 7, 2006, with longitude between 157.01◦and
157.74◦and latitude between −5.8◦and −2.4◦(granule 1475). Lower panel:
Vertical reflectivity profiles in black/green correspond to the pixels between the
two black/green arrows in the upper panel.
Upper panel: CloudSat reflectivity vertical cross-section for a system
(e.g.,  and ). Therefore, some important questions arise
from the aforementioned examples such as follows.
1) Can we explain the absence of a reflectivity peak at
surface ranges in the frame of the MS theory?
2) In which conditions do we expect the signal from the
surface and the pixels immediately above the surface to
be “contaminated” by the MS?
3) How deep in the observed medium can MS produce a
CloudSat/EarthCARE detectable signal, i.e., which are
the typical saturation levels?
4) What are the leading factors affecting the MS enhance-
ment (defined as the difference between the MS and SS
To answer these questions, we have developed an MS radar
simulator which includes the interaction with the surface
(description in Section II). The model has been exploited to
simulate the CloudSat/EarthCARE echoes in Section III for
Storm 06W occurred on the July 24, 2006 (granule 1272) southwest of Taiwan.
Lower panel: Vertical reflectivity profiles corresponding to the pixels between
the two black arrows in the upper panel.
Upper panel: CloudSat reflectivity vertical cross-section for a Tropical
some simple vertical profiles representing rainy situations.
Results are discussed in Section IV, and conclusions are drawn
in Section V.
II. MODEL DESCRIPTION
The MS forward-biased polarized Monte Carlo radar simula-
tor described in  and  has been extended by introducing
a sea-surface scattering model, which simulates a Kirchoff
well describing sea surfaces at 94 GHz. If the surface height is
assumed to be a Gaussian distributed random surface with zero
mean, variance σ2, and correlation coefficient ρ and satisfying
the additional constraints that the surface roughness is isotropic
and the vertical-scale roughness is larger than the wavelength
[17, Sec. 12–4.4], the backscattering coefficients are func-
tions of the mean-squared surface slope (s2≡ σ2|ρ??(0)|) only
1646IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 46, NO. 6, JUNE 2008
and are independent of the shape of the surface correlation
σ(θ,v,λ) ≡ σhh(θ) = σvv(θ) =|Rvv(0,λ)|2e−
where θ is the zenith angle, λ is the wavelength, and |Rvv(0)|2
is the Fresnel reflectivity evaluated at normal incidence (typ-
ical value is 0.409 at 94 GHz). The effective mean-square
surface slope generally increases with the sea-surface wind
speed v; different empirical relationships are discussed by
Li and Nakamura . From (1), it is clear that there is a unique
relationship between the normalized sea-surface radar cross-
section at normal incidence σ0≡ σ(θ = 0,v,λ) and the mean-
squared surface slope s2; in particular, the smoother the surface
(lower s2), the higher is the σ0. As shown in Fig. 6 by Li and
Nakamura , which presents the airborne surface obser-
vations, the σ0 ranges between 11 and 14 dB for nadir-
looking radars at 94 GHz. Mitrescu et al.  analyzed the
CloudSat surface-return probability distribution function for
different surface types (classified by using the International
Geosphere–Biosphere Programme classification map). For wa-
ter surfaces, there are no significant differences during the three
configurations of the CloudSat beam with respect to the local
zenith (1.7◦forward up to July 7, 2006; 0.0◦between July 7
and August 15, 2006; and finally, 0.16◦forward up to now).
The typical values of the water-surface return range between 38
and 42 dBZ at 500-m vertical resolution, which correspond to
the values of σ0that range from 9.4 to 13.4. In the following
computations, a characteristic value of 12 dB is selected for σ0,
but some simulations have been performed at 10 and 14 dB
Under the same assumptions, the bistatic (BIS) scattering
tion matrix Msurf) are completed, which are characterized by
the geometry of scattering, the refractive index of the sea, and
the root-mean-square surface slope [18, Ch. 2].
In the Monte Carlo tracing procedure used in our model,
each time a radiation impinges the surface, a new direction
is sampled by basing on the two random-number method,
according to the probability distribution function
ab[(a,b = h,v) (thus also the bidirectional reflec-
which depends on the incident polarization, as described by
the four components of the incident Stokes vector Iinc. The
scattered Stokes vector is then computed as
where the renormalization factor C adjusts the scattered power
to the value that is proper to the surface at the given in-
coming direction and polarization state. Because of the bias-
ing procedure, the contribution to the received power in the
apparent range bin is evaluated by accounting for the antenna
pattern, the attenuation, and the probability of scattering in
the radar direction (proportional to Msurf/(4π)). Note that the
BIS component (i.e., scattering from the precipitation/surface
to the surface/precipitation and then back to the radar) and
the MI returns (i.e., a double reflection of the surface with
the precipitation acting as intermediate scatterer; see ) are
automatically included; however, in addition, all other MS
contributions are accounted for.
The Monte Carlo formulation is based on the radiative trans-
fer, which cannot intrinsically include the so-called backscat-
tering enhancement (details in ) caused by the cross-terms
in the analytic Green function method. However, some ex-
pedients can be used to partially correct “a posteriori” for
this deficiency by exploiting the fact that the Monte Carlo
stores the different scattering orders and the cocomponent and
cross-component. As demonstrated in , , and , under
steady processes, in the copolar channel, the backscattering
enhancement generally doubles the intensity derived with the
radiative-transfer theory (i.e., ladder terms are equal to the
cross-terms). According to this thumb rule, we have modified
the outputs of the Monte Carlo scheme so that all the copolar
terms (except for the first order) are doubled. The correction of
the cross-polarization terms is much more problematic because
the cyclic term is not always equal to the corresponding ladder
term –. However, for the purposes of this paper, only
the copolar signals have to be corrected because the measured
reflectivities usually refer exactly to copolar reflectivities. The
geometrical configuration of W-band spaceborne radars implies
an angle shift on the order of 0.0025◦with respect to the
exact backscattering which partially reduces the factor-two
enhancement prediction (e.g., [8, Fig. 4]). The improvements
in the correct evaluation of the backscattering enhancement
(even in cross-channels) are expected by further studies that are
capable of including time-dependent processes.
III. SETUP OF MODEL SIMULATIONS
The model has been exploited to simulate the nadir-pointing
CloudSat/EarthCARE radar (705/450-km altitude, 0.1◦antenna
beamwidth, and 250/100-m vertical resolution) echoes, includ-
ing ranges that are longer than the surface range (the novelty
with respect to ). In order to better illustrate and compare
the contributions to the total return of the BIS, MI, and the
surface reflectance, including the MS returns, a very simple hy-
drometeor profile is assumed (Fig. 3). Each profile is character-
ized by a single parameter IWCFL(indicated by the filled circle
in Fig. 3), which is the ice-water content (IWC) at the freezing
level at HFL. The size distribution of ice particles is exponen-
tial, as described in , i.e., N(r) = N0re−Λicerwith N0=
16 × 103m−3· mm−1, whereas Λice1=
The size distribution and rain-water content (RWC) of the rain
particles at the bottom of the melting layer (and below) are dic-
tated by a one-to-one correspondence between the ice particles
melt (i.e., aggregation and breakup processes are neglected).
Microphysical parameterizations of falling velocities are taken
from . Note that because of the dependence of fall speeds
BATTAGLIA AND SIMMER: HOW DOES MULTIPLE SCATTERING AFFECT THE W-BAND RADAR MEASUREMENTS? 1647
Section III. Immediately below the freezing level located at HFL, a mixed layer
with melting particles is present. The total height of the storm is twice HFL.
The atmosphere is characterized by a temperature profile with a constant lapse
rate γ. The ground is approximated as a Kirchoff surface.
Schematic for the RWC and IWC profiles used for the simulations in
on air density, the RWC increases slightly from the bottom of
the melting layer to the ground. The IWC profile decreases
linearly with height from the freezing level to the top of the
cloud (located at double the freezing level height). Atmospheric
gas absorption is computed by assuming a completely satu-
rated (with respect to water) water vapor profile according to
In order to highlight the effect of the thickness of the raining
systems, we have run simulations with the altitude of HFLset
to 3 and 6 km. These runs are referred to as “midlatitude”
and “tall” (mimicking tropical profiles) systems. Ice density is
assumed to be ρice= 0.1 or 0.4 g/cm3to describe “snowlike”
or “graupellike” particles. The surface is always assumed of the
Kirchoff-type with σ0= 10, 12, and 14 dB.
IV. MODEL RESULTS
The examples of simulated reflectivity profiles with σ0=
10 dB are shown in Fig. 4 in the CloudSat configuration for
tall systems with graupel aloft. The different curves correspond
to the BIS, the hydrometeor SS reflectivity (Z[SS]; which
is obtained by adding contributions that involve only an SS
with the hydrometeors so that, below the surface, it coincides
with the sum of MI + BIS) and the MS reflectivity (Z[MS]).
As already noted by Liao et al. , the (circle symbol line)
BIS contribution is really important to be only just below the
surface (it is rapidly suppressed within a distance that is equal
to the radar-footprint radius) where it tends to mask the MI
return. On the other hand, at longer ranges, the curve labeled
as Z[SS] practically corresponds to the MI contribution. When
thickness; both are reported in the title of each panel), the MS
becomes increasingly important and the Z[MS] curves move
away from the Z[SS] profiles (clear sign of MS enhancement),
which is a result already well understood from former MS
computations. In our four profiles for bins that are close to
the surface, the MS enhancement corresponds to about 4.2,
16, 46, and more than 100 dB for the four cases, respectively.
Due to the much stronger surface SS echoes, results are quite
different when we consider the surface range (0.2, 1.6, 25, and
104 dB, respectively). Therefore, the return at surface range is
generally contaminated at higher rain rates than radar echoes
that are close to the surface, which is a relevant aspect when
using surface reference techniques. In intermediate regimes, it
may happen that the path-integrated attenuation estimated from
surface echoes is the usual SS attenuation, whereas the echoes
just above the surface are contaminated by MS (see discussion
in Section IV-A). The MS effects seem to be even more relevant
at apparent ranges that are longer than the surface distance,
where they cannot be neglected because they are already at very
Z[MS] lines that are above and below the surface in the upper-
left panel in Fig. 4). This means that the information from such
ranges cannot be exploited by using simple MI assumptions.
Finally, note that at ranges corresponding to the surface at
low rain rates (upper panels in Fig. 4), the surface range bin
clearly shows a discontinuity peak like the green lines in the
CloudSat data in Fig. 1. However, the MS effects tend to mask
this discontinuity, and above a critical RRg(e.g., for the values
corresponding to the bottom panels in Fig. 4), the reflectivity
profiles do not present any discontinuity at the surface (like the
black lines in Figs. 1 and 2). Therefore, MS can simply explain
this feature, which is frequently observed in the CloudSat
database. Another obvious consequence of MS is the decrease
of the minimum height of detection (MHD), which is the lowest
altitude (above or below the surface) with a signal above the
is substantially lowered. For instance, for RRg= 15 mm/h
(bottom-right panel in Fig. 4), the MHD computed with the SS
approximation would give a value of 5.7 km while accounting
for the MS contributions; the signal remains above MDT down
to around 8 km below the surface. Obviously, this effect is
enhanced/suppressed when increasing/decreasing σ0but only
at small rain rates; at higher rain rates, the signal coming
apparently from below the surface is actually independent from
the surface properties (see discussion hereafter).
A. General Discussion
Fig. 5 shows the SS and MS returns for the surface range;
top/bottom panels correspond to CloudSat/EartCARE config-
urations. In each panel, we have included the results for tall
and midlatitude systems with snow/graupel ice density. In
the (dash–dotted lines) SS approximation, the surface return
decreases with the increase of the total optical thickness of the
rain system (i.e., with increasing rain rates and passing from
midlatitude to tall systems or from snow to graupel), which is
a parameter that drives the two-way path attenuation. When the
(continuous lines) MS signal is included, the MS effects partly
compensate for the attenuation, starting being appreciable at
a critical value of RRg, which depends on the kind of profile
(e.g., for tall systems at values of RRgaround 5 mm/h) and
the radar configuration (with the EarthCARE configuration,
obviously producing less MS enhancement due to the smaller
footprint). For both radar configurations at high rain rates, the
signal at surface range becomes practically decoupled from
1648IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 46, NO. 6, JUNE 2008
increasing from (upper left) 2 to (lower right) 15 mm/h. An ice density that is equal to 0.4 g/cm3is selected for all cases.
Simulated CloudSat vertical reflectivity profiles for the scenarios illustrated in Fig. 3 with σ0= 10 dB, (tall systems) HFL= 6 km, and with RRg
the surface properties (continuous lines are far away from the
dash–dotted ones), so that the surface range echo return can
actually increase with increasing rain rates. In Fig. 5, the MDT
dashed line defines the region of saturation by attenuation;
whereas the SS theory forecasts a saturation of the CloudSat
signal at RRg above approximately 6 and 13 mm/h, the MS
computations show that such saturation is reached only for “tall
snow” systems in a restricted range of the RRgvalues. On the
other hand, due to the lower MDT and despite the reduced
MS, saturation is never achieved for EarthCARE setup. For
values of σ0= 10, 14 dB, similar results are found; the SS
curves are simply shifted by ±2 dB, whereas in the MS plots,
the surface reflectivity differences, which are present at small
rain rates, become unimportant at high rain rates (after the
local minimum). This indirectly demonstrates that the signal is
basically unaffected by the surface properties, i.e., it is caused
by pulse stretching within the rain medium.
To better assess the range of validity of the SS schemes,
Fig. 6 shows the MS enhancement (defined as Z[MS] − Z[SS])
for the surface range bin and for the bin immediately above it.
The 3-dB dashed line defines a relaxed range of validity for
the SS theory; for instance, in CloudSat configuration, the MS
corrections have to be applied at near-surface range bins that
are already at 2–3 mm/h (for all our different profiles), whereas
values of 5–7 mm/h define an upper limit to the retrieval of
integrated path attenuation from surface reference techniques
for tall systems with low-density ice particles. For EarthCARE,
the thresholds for SS validity are slightly higher (not shown).
In Fig. 6, the results are also presented for the “Graupel Tall”
scenario when using (dashed lines) a surface with σ0= 10,
14 dB. Obviously, the MS enhancement is larger when sea
surfaces with lower reflectivities are employed.
The presence of MS causes the signal to stay above the MDT
(equal to −26/ −36 dBZ for CloudSat/EarthCARE) for ranges
that are much longer than expected from the SS computations
(Fig. 7). While in the SS approximation, MDT is a monotonic
increasing function of RRg, and saturation is achieved above
the surface that is already at small rain rates. The MS-based
BATTAGLIA AND SIMMER: HOW DOES MULTIPLE SCATTERING AFFECT THE W-BAND RADAR MEASUREMENTS?1649
ground evaluated in the (dash–dotted lines) SS approximation and accounting
for the (continuous lines) MS. The results correspond to the scenarios for
tall/midlatitude systems with snow/graupel, as indicated in the legend, and
the Kirchoff surfaces with σ0= 10 dB. The arrows show the trends when
lowering HFL. Top/bottom panels correspond to the CloudSat/EarthCARE
configurations with the dash line representing the MDT (see text for details).
Reflectivity at the surface range as a function of the rain rate at the
MHDs show a completely different trend: First, following the
SS-based MHDs,thenreaching amaximum atintermediate val-
ues of RRg, and then turning downward. For our profiles, MHD
is never higher than 1.4/ − 0.4 km for CloudSat/EarthCARE
configurations, i.e., saturation will never be reached above these
These results are based on simplified precipitating profiles.
However, they clearly show the trends (indicated by the arrows
in Figs. 5–7) expected when lowering the total depth of the
precipitating system. In correspondence to the same rain rates,
taller systems will produce optically thicker profiles; thus,
they are characterized by stronger attenuation (see lower SS
reflectivities close to the surface in Fig. 5) but simultaneously
enhanced the MS both at and close to the surface (Fig. 6)
and by a lowering of the MHDs (Fig. 7). Obviously, not
only a variation of the total thickness of the profile but also
(dash–dotted lines) near surface range as a function of the rain rate at the
ground. The results correspond to the scenarios for tall/midlatitude systems
with snow/graupel, as indicated in the legend, and the Kirchoff surfaces with
σ0= 10 dB. The arrows show the trends when lowering HFL.
CloudSat MS enhancement at the (continuous) surface range and the
different ice microphysical assumptions will strongly affect the
MS enhancement. To demonstrate this, we have performed the
following three additional simulations for the tall system.
1) “Warm rain”: No ice is present above HFL;
2) “Frontal snow”: The ice with a snow density of 0.1 g/cm3
is exponentially distributed with an intercept which varies
with temperature T according to
?1/(m3mm)?= 4.0 × 103e
following . This ice size distribution is reasonable for
frontal rainbands because it implies that the dimension
of the particles is rapidly decreasing at cold tempera-
tures (and, with it, the scattering/backscattering cross-
3) “Large graupel”: graupel particles with 0.6 g/cm3density
and exponentially distributed with intercept N0= 4 ×
103m−3· mm−1constant at all altitudes (a situation more
typical for strong convection).
Fig. 8 shows the MHD evaluated with these three different
microphysical assumptions (together with the “Graupel Tall”
and “Snow Tall” profiles in Fig. 7). There is a large difference
in the saturation height, depending on the ice microphysics,
with the extreme results obtained with profiles containing dense
(graupellike) particles and no ice at all (labeled “Warm rain”).
In the former case, the signal is never saturated above the
surface, and long tails above the MDT are produced even at
10 km below the surface for high rain rates. When the systems
that are even higher than those displayed (our profiles are
capped at 12 km, whereas the system depicted in Fig. 2 is
rising to almost 20 km) are simulated, the tails capable of
accounting for second-trip echoes can be reproduced by our
model. In the “Warm rain” case, the signal is saturated at
ranges above the surface that is already at rain rates higher than
about 5 mm/h, and the MHD constantly increases with RRg.
1650IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 46, NO. 6, JUNE 2008
including (continuous lines) MS as a function of the rain rate at the ground.
The results correspond to the scenarios for tall/midlatitude systems with
snow/graupel, as indicated in the legend, and the Kirchoff surfaces with σ0=
10 dB. The arrows show the trends when lowering HFL. Top/bottom panels
correspond to the CloudSat/EarthCARE configurations (see text for details).
MHD evaluated in the (dash–dotted lines) SS approximation and
At RRg= 50 mm/h, practically 2 km of rain is sufficient to
saturate the signal.
Size distributions play an important role as well, particularly
for snow systems; MS is stronger when size distributions fa-
voring large particles are considered (compare in Fig. 8 the
“snow” with “frontal snow” and “graupel” with “large graupel”
curves). These simple examples demonstrate the major role
played by the ice microphysics in affecting MS effects. Finally,
in Fig. 8, the line labeled as “Graupel black surf” corresponds
to the profiles with the same “Graupel” microphysics but with a
perfectly absorbing surface. Because there is no difference be-
tween this curve and the curve labeled as “Graupel” (which has
a surface with σ0= 12 dB) for rain rates higher than 10 mm/h,
this proves a complete decoupling of the signal from the surface
properties at ranges longer than surface range. In practice,
the radiation sensed from those regions has undergone many
the rain rate at the ground for tall profiles with different ice microphysics, as
indicated in the legends (see text for details).
CloudSat configuration MHD (i.e., saturation level) as a function of
MS events within the rain medium only, without reaching the
An extended version (including scattering on a Kirchoff-type
surface, which is suitable for describing water surfaces) of a
Monte Carlo MS radar simulator has been presented. Even for
simple vertical hydrometeor profiles, the model produces re-
flectivity patterns which resemble the CloudSat–CPR observed
profiles at surface range and below (like those shown in Figs. 1
and 2 and frequently occurring in the CloudSat database).
Unlike SS, MS effects can explain the absence of a surface-
echo peak for precipitating systems.
The results of our simulations (tuned to CloudSat and Earth-
CARE configurations) show that the surface range return is
significantly contaminated by the MS contributions that are
already at rain rates above 5/10 mm/h (depending on the system
type, ice density, and configuration) and certainly completely
decoupled from the surface properties for rain rates greater than
15 mm/h (see Fig. 5). In these conditions, the reflectivity profile
has no discontinuity at all at surface range. Thus, “smooth”
profiles at surface ranges are distinctive signatures of strong
MS effects. When a discontinuity of reflectivity is found at
surface range, MS cannot however be ruled out. In fact, the MS
enhancement is stronger for near-surface range bins, and it has
to be accounted for already at rain rates of few millimeters per
hour (see dashed lines in Fig. 6).
The MS produces long signal tails at ranges below the sur-
face, so that MHD is significantly lowered (even to tens of
km below the surface). The effect increases with the ice con-
tents above the liquid precipitation column (Fig. 7). CloudSat
second-trip echoes can be reproduced in our modeling frame,
too. The different behaviors of the MS-based MDH imply
a strong reduction of saturation levels (compared with SS),
as observed in the CloudSat database. Our simulations show
that such a reduction is expected also for the EarthCARE
BATTAGLIA AND SIMMER: HOW DOES MULTIPLE SCATTERING AFFECT THE W-BAND RADAR MEASUREMENTS? 1651
CPR, where the lower MS (due to the footprint reduction) is
compensated by the better MDT.
All MS effects are highly dependent on the ice layer of the
cloud and on its microphysical assumptions, e.g., large/dense
ice particles strongly enhance MS (Fig. 8). Therefore, the
strength of MS can be potentially used to retrieve microphysical
properties of the ice phase. A detailed statistical analysis of
the MS effects in the CloudSat database to better assess this
potential is now in progress.
The authors would like to thank the NASA CloudSat project
for providing the CPR data.
 G. L. Stephens, D. G. Vane, R. J. Boain, G. G. Mace, K. Sassen, Z. Wang,
A. J. Illingworth, E. J. O’Connor, W. B. Rossow, S. L. Durden,
S. D. Miller, R. T. Austin, A. Benedetti, and C. Mitrescu, “The
no. 12, pp. 1771–1790, Dec. 2002.
 L. R. Lemon, “The radar ‘three-body scatter spike’: An operational
large-hail signature,” Weather Forecast., vol. 13, no. 2, pp. 327–340,
 L. Li, G. M. Heymsfield, P. E. Racette, L. Tian, and E. Zenker, “A 94-GHz
cloud radar system on a NASA high-altitude ER-2 aircraft,” J. Atmos.
Ocean. Technol., vol. 21, no. 9, pp. 1378–1388, Sep. 2004.
 D. S. Zrniæ, “Three-body scattering produces precipitation signatures of
special diagnostic value,” Radio Sci., vol. 22, pp. 76–86, 1988.
 L. Liao, R. Meneghini, and T. Iguchi, “Simulations of mirror image re-
turns of air/space-borne radars in rain and their applications in estimating
path attenuation,” IEEE Trans. Geosci. Remote Sens., vol. 37, no. 2,
pp. 1107–1121, Mar. 1999.
 F. S. Marzano, L. Roberti, S. Di Michele, A. Mugnai, and A. Tassa,
“Modeling of apparent radar reflectivity due to convective clouds at
attenuating wavelengths,” Radio Sci., vol. 38, no. 1, 1002, Jan. 2003.
 S. Kobayashi, S. Tanelli, T. Iguchi, and E. Im, “Backscattering enhance-
ment with a finite beam width for millimeter-wavelength weather radars,”
in Proc. SPIE—Microwave Remote Sensing of the Atmosphere and Envi-
ronment IV, G. S. Jackson and S. Uratsuka Eds., Dec. 2004, vol. 5654,
 S. Kobayashi, S. Tanelli, and E. Im, “Second-order multiple-scattering
theory associated with backscattering enhancement for a millimeter wave-
length weather radar with a finite beam width,” Radio Sci., vol. 40, no. 6,
RS6015, Dec. 2005. DOI:10.1029/2004RS003219.
 A. Battaglia, M. O. Ajewole, and C. Simmer, “Multiple scattering effects
due to hydrometeors on precipitation radar systems,” Geophys. Res. Lett.,
vol. 32, no. 19, L19801, Oct. 2005. DOI: 10.1029/2005GL023810.
 A. Battaglia, M. O. Ajewole, and C. Simmer, “Evaluation of radar mul-
tiple scattering effects from a GPM perspective. Part II: Model results,”
J. Appl. Meteorol., vol. 45, no. 12, pp. 1648–1664, Apr. 2006.
 A. Battaglia, M. O. Ajewole, and C. Simmer, “Evaluation of radar mul-
tiple scattering effects in CloudSat configuration,” Atmos. Chem. Phys.,
vol. 7, no. 7, pp. 1719–1730, Apr. 2007.
 D. Bouniol, A. Protat, A. Plana-Fattori, J.-P. Vinson, M. Giraud, and
N. Grand, “Comparison of airborne and spaceborne 95 GHz radar reflec-
tivities and evidence of multiple scattering in CloudSat measurements,”
J. Appl. Meteorol. Climatol., 2007. submitted for publication.
 R. Meneghini, T. Iguchi, T. Kozu, L. Liao, K. Okamoto, J. A. Jones, and
J. Kwiatkowski, “Use of the surface reference technique for path attenu-
ation estimates from the TRMM precipitation radar,” J. Appl. Meteorol.,
vol. 39, no. 12, pp. 2053–2070, Dec. 2000.
 R. Meneghini, J. A. Jones, T. Iguchi, K. Okamoto, and J. Kwiatkowski,
“A hybrid surface reference technique and its application to the TRMM
precipitation radar,” J. Atmos. Ocean. Technol., vol. 21, no. 11, pp. 1645–
1658, Nov. 2004.
 W. S. Olson, P. Bauer, N. F. Viltard, D. E. Johnson, W. Tao, R. Meneghini,
and L. Liao, “A melting-layer model for passive/active microwave remote
sensing applications. Part I: Model formulation and comparison with
observations,” J. Appl. Meteorol., vol. 40, no. 7, pp. 1145–1163, Jul. 2001.
 A. Battaglia, M. O. Ajewole, and C. Simmer, “Evaluation of radar mul-
tiple scattering effects from a GPM perspective. Part I: Model descrip-
tion and validation,” J. Appl. Meteorol., vol. 45, no. 12, pp. 1634–1647,
 F. T. Ulaby, R. K. Moore, and A. K. Fung, Microwave Remote Sensing:
Active and Passive. Boston, MA: Artech House, 1986.
 L. Tsang, J. A. Kong, and R. T. Shin, Theory of Microwave Remote
Sensing. New York: Wiley-Interscience, 1985.
 J. Li and K. Nakamura, “Vertical distribution of the mirror image returns
observed by TRMM PR and estimated for a 35-GHz radar,” J. Atmos.
Ocean. Technol., vol. 22, no. 11, pp. 1829–1837, Nov. 2005.
 C. Mitrescu, J. Haynes, T. L’Ecuyer, S. Miller, and J. Turk, “Light rain
retrievals using CloudSat94-GHz radar data-preliminary results,” in Proc.
87th AMS Annu. Meeting, San Antonio, TX, 2007.
 M. I. Mishchenko, L. D. Travis, and A. A. Lacis, Multiple Scattering
of Light by Particles: Radiative Transfer and Coherent Backscattering.
Cambridge, U.K.: Cambridge Univ. Press, 2006.
 M. I. Mishchenko, “Enhanced backscattering of polarized light from dis-
crete random media: Calculations in exactly the backscattering direction,”
 S. Kobayashi, T. Oguchi, S. Tanelli, and E. Im, “Backscattering enhance-
ment on spheroid-shaped hydrometeors: Considerations in water and ice
particles of uniform size and Marshall-Palmer distributed rains,” Radio
Sci., vol. 42, no. 2, RS2001, 2006. DOI:2006RS003503.
 T. Oguchi and T. Ihara, “Computer simulation of enhanced backscattering
from randomly distributed spherical scatterers at 30 GHz and compari-
son with measurement,” Radio Sci., vol. 41, no. 6, RS6002, Dec. 2006.
J. Meteorol., vol. 5, no. 4, pp. 165–166, Aug. 1948.
 A. Battaglia, C. Kummerow, D.-B. Shin, and C. Williams, “Toward char-
acterizing the effect of radar bright bands on microwave brightness tem-
peratures,” J. Atmos. Ocean. Technol., vol. 20, no. 6, pp. 856–871, 2003.
 H. Liebe, “An updated model for millimeter-wave propagation in moist
air,” Radio Sci., vol. 20, no. 5, pp. 1069–1089, Sep./Oct. 1985.
 G. Cox, “Modelling precipitation in frontal rainbands,” Q. J. R. Meteorol.
Soc., vol. 114, no. 479, pp. 115–127, Jan. 1988.
Alessandro Battaglia received the B.S. degree from
the University of Padova, Padova, Italy, with a thesis
in particle physics, and the Ph.D. degree in physics
from the University of Ferrara, Ferrara, Italy.
He is experienced in theoretical research scat-
tering computations by populations of nonspherical
particles, radiative transfer in clouds, precipitation
modeled with 3-D structures, and preferentially ori-
ented nonspherical particles. He is currently an
Assistant Professor with the Meteorological Insti-
tute, University of Bonn, Bonn, Germany, where he
mainly works on polarimetric radars and radiometry.
Clemens Simmer received the B.S. and Ph.D.
degrees in meteorology from the University of
Cologne, Cologne, Germany, in 1981 and 1983,
Since 1996, he has been with the Meteorological
Institute, University of Bonn, Bonn, Germany, where
is the Head of the working group on remote sens-
ing and mesoscale modeling. He has been a Prin-
cipal Investigator (PI) in Framework Programme 4
(FP4) (Cloud Retrieval Validation Experiment), FP5
(European Land Data Assimilation System and
Cloud Liquid Water Network), FP6 (GEOLAND and African Monsoon Multi-
disciplinary Analysis), and several European Space Agency projects with work
packages in remote sensing of clouds, precipitation, and soil moisture and
satellite data assimilation. He has been the Leader of many national projects
on remote sensing of precipitation and soil moisture, mesoscale hydrological
and precipitation modeling, and anthropogenic climate changes and climate