IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO.

7, JULY 2010 3009

Potentials and Limitations of Moon-Borne

SAR Imaging
Gianfranco Fornaro, Senior Member, IEEE, Giorgio Franceschetti, Life Fellow, IEEE,
Fabrizio Lombardini, Senior Member, IEEE, Alessandro Mori, and Mario Calamia

Abstract—Moon exploitation is among the next space mission to its distinguished characteristic from LEO satellites in terms
priorities. Earth observation (EO), which is traditionally imple- of imaging characteristics and potential applications. In fact,
mented on artificial lower Earth orbit satellites, can be, in prin- imaging experiments concerning the Earth and the Moon have
ciple, extended to the platform constituted by the natural Earth already been carried out; in particular, images of the Moon
satellite. With this regard, we investigate the features related to
the EO by a possible Moon-borne synthetic aperture radar system poles have been generated by using SAR systems on the Earth
in terms of imaging characteristics and potential applications, as [4]–[6]. In these experiments, the antennas of the Goldstone
well as of expected limitations. Solar System Radar were used to transmit, via a 70-m-diameter
antenna, a radiation of 500 kW at X-band: The Earth movement
Index Terms—Interferometry, Moon, synthetic aperture
radar (SAR). was used to map the south polar region with a spatial resolution
of 150 m. However, studies concerning the SAR imaging of
I. I NTRODUCTION the Earth from the Moon, at the best of our knowledge, are not
present in the current literature.

A MONG the next space mission goals, the exploitation

of the Moon, the natural Earth satellite, is gaining an
increasing interest. As a matter of fact, calls for ideas are being
One of the major expected problems of lunar EO with respect
to standard LEO satellites is associated with the considerable
increase of the distance between the scene and the sensor. This
recently coordinated by several international space agencies, fact plays an important role for both passive systems and active
in order to evaluate themes and objectives related to lunar SAR remote sensing equipment. In the case of passive systems,
missions. the distance increase leads to an unavoidable resolution loss.
Colonization of the Moon is along the usual track of hu- On the contrary, for SAR systems, the increase of the distance
man civilization and economic expansion; it is also related to could, in principle, drastically impact the power budget [7],
a large number of scientific issues [1], [2], including Earth leading to a transmitted power level demand that could be
observation (EO). hardly provided with standard technology. Nonetheless, there
EO is traditionally carried out by means of sensors mounted are many expected advantages related to the use of a lunar SAR
on artificial satellites mainly orbiting on sun-synchronous low system that make worth further investigations. First of all, the
Earth orbits (LEOs). With respect to standard LEO satellites system would be characterized by a large ground swath, thus
commonly used in remote sensing, Moon has specific features, allowing the imaging of large portions of the Earth surface
such as a longer distance between the sensor and the imaged as well as a reduction of the revisiting time down to one day.
area, a low orbit inclination, a large surface for the installation Second, a steering of antenna in elevation would lead to quasi-
of sensors and equipment, etc. global scene accessibility.
In this paper, we analyze the potentials as well as the In this paper, we mainly investigate the characteristic of lunar
limitations related to active microwave remote sensing with SAR imaging in terms of resolution and power constraints.
synthetic aperture radar (SAR) [3], giving particular emphasis As shown in the following, different from standard stripmap
SAR sensors, a Moon-borne SAR can overcome the limit of
half-antenna-length azimuth resolution [3]–[5] due to a peculiar
imaging effect associated with the Earth rotation, leading to
Manuscript received March 31, 2009; revised August 12, 2009. Date of very high resolution imaging capabilities and hence small pixel
publication April 5, 2010; date of current version June 23, 2010. This work
was supported in part by the Italian Space Agency under the Contract “Italian
size. This feature allows, first of all, the use of large antennas
Vision for Moon Exploration Earth observation from the Moon” (n.I/042/06/0). to partially compensate for power limitations, without paying
G. Fornaro is with the Istituto per il Rilevamento Elettromagnetico the price to lower the resolution down to limits that could be
dell’Ambiente, Consiglio Nazionale delle Ricerche, 80124 Napoli, Italy unacceptable for some specific applications. Second, as shown
(e-mail: fornaro.g@irea.cnr.it).
G. Franceschetti is with the Dipartimento di Ingegneria Biomedica Elet- in Section III, the antenna could be designed to be highly
tronica e delle Telecomunicazioni, Università di Napoli “Federico II,” 80128 directive with no care to the need to implement any electronic
Napoli, Italy (e-mail: gfrance@unina.it). or mechanical steering of the beam along the azimuth direc-
F. Lombardini is with the Dipartimento di Ingegneria della Informazione,
Università di Pisa, 56122 Pisa, Italy (e-mail: f.lombardini@iet.unipi.it).
tion typical of more complex acquisition modes [8] aimed at
A. Mori and M. Calamia are with the Department of Electronics and increasing the integration time to achieve very high resolutions.
Telecommunication, University of Florence, 50139 Florence, Italy (e-mail Finally, for a Moon-borne SAR system, the synthetic aperture
mario.calamia@unifi.it). integration time for full focusing is rather large: This, on one
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. side, can prevent the possibility to effectively image “fast”
Digital Object Identifier 10.1109/TGRS.2010.2041463 moving scenes, such as the sea surface, but on the other side,

0196-2892/$26.00 © 2010 IEEE

Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 3010 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 7, JULY 2010

TABLE I
MOON ORBITAL AND PHYSICAL PARAMETERS

of these rotation effects, the same side of the Moon is always

facing the Earth; this effect is referred to as tidal locking. Owing
to the tidal locking, an antenna pointed toward the Earth center
approximately maintains its orientation, but for higher order
effects such as lunar libration.
Fig. 1. (a) Detail of the Earth–Moon system. (b) Detail for the calculation of
the minimum elevation of the Moon at 40◦ latitude. (c) Detail for the calculation
The Earth rotates with a period TE of approximately one day
of the maximum elevation of the Moon at 40◦ latitude. at an angular velocity of ΩE = 2π/TE = 7.292 · 10−5 rad/s.
By comparing the angular velocity of the Earth and the angular
it can allow implementation of “fast” moving target detection, velocity of the Moon around the Earth, it can be noted that the
even starting from single antenna data [9]. former is about 30 times larger than the latter; i.e., ΩE  ΩM .
Moon’s orbit may, in principle, allow the implementation In fact, in a Moon fixed frame of reference, the Moon–Earth
of one-day repeat-pass differential interferometry [8]–[10] for direction shows oscillations called librations. In particular, the
very large swath monitoring of slow ground deformation, etc. inclination of the Moon’s equator to its orbital plane is not zero,
We also address such a topic and show that, due to the large causing an oscillation in latitude (optical libration in latitude).
variation of the elevation angle from day to day, the number of Moreover, the Moon moves on a slightly elliptical orbit around
usable coherent interfering images is rather limited. However, the Earth, and its orbital angular velocity changes, causing
lunar surface could allow the installation of multiple SAR an oscillation in longitude (optical libration in longitude). At
sensors (only one transmitting): This solution, although costly last, there are additional small pendulous oscillations (physical
from the hardware viewpoint, allows partially compensating the libration) due to the irregular structure of the Moon. The optical
aforementioned drawback and also implementing single-pass librations occur on the sidereal period, with a magnitude of
interferometry [10] on a systematic basis, as well as along- about 8◦ . The physical librations have a period of one year
track interferometry (ATI) [13] and advanced 3-D imaging [14], in longitude and about six years in latitude, with amplitudes
[15]. In particular, we also briefly investigate the use of ATI to of 0.02◦ and 0.04◦ , respectively. The longitude libration causes
monitor ocean currents [10], [13] or other distributed targets variation of the angular velocity on the order of 4 · 10−6 rad/s.
with quick deformation motion. This paper aims at evaluating the main parameters of a
Moon-borne SAR system. Accordingly, in the following, we
assume that the Moon is fixed and the Earth is rotating, and
II. G EOMETRICAL PARAMETERS
we neglect higher order motions associated with the libration.
The Earth–Moon system is shown in the upper part of Fig. 1. Obviously, such effects must be considered for data processing
Moon and Earth rotate both around their own rotation axes, and in order to achieve fully focused images. We also fix the
these axes are not parallel. Moreover, the Moon orbits around reference system on the Moon center (selenographic coordinate
the Earth in a plane which is tilted by 5◦ from the ecliptic, i.e., system) with the x-axis pointing to the Earth center, the z-axis
the plane containing the mean orbit of the Earth around the Sun. orthogonal to the Moon orbit plane around the Earth.
A first consequence of the inclination of the Earth rotational In the following, we let RE to be the average Earth radius
axis and of the inclination of the Moon orbit with respect to the and DM to be the average distance between the Earth and the
ecliptic is the large variation of the elevation of the Moon over Moon. From Table I, it is apparent that the main disadvantage of
the horizon. Fig. 1(b) and (c) shows such effects for latitude the use of the Moon as a remote sensing satellite is the distance
of 40◦ ; in particular, it results that the minimum elevation of from the Earth surface, which is about 550 times higher than
the Moon over the horizon is 50 − 24 − 5 = 21◦ , whereas the the distance of common LEO satellites. An obvious advantage
maximum elevation is 50 + 24 + 5 = 79◦ , with an excursion of is the availability of a large surface, whose width is almost
more than 50◦ . 3500 km (2RM ), where different sensors could be installed.
Table I summarizes the main orbital and the physical char- The tangential velocity of points on the Earth surface (vE )
acteristics of the Moon, discussed hereafter. Moon is in syn- depends on the latitude of the point: At the equator, such a ve-
chronous rotation around the Earth, and the orbit has a sidereal locity is about vE0◦ = ΩE RE ≈ 1600 km/h (450 m/s), whereas
period TM of about 27 days (sidereal month), implying that at 40◦ latitude, we have vE40◦ = ΩE RE cos 40◦ = 1200 km/h
it rotates around the Earth center with an angular velocity of (340 m/s). It is worth to note that the antenna-to-scene relative
ΩM = 2π/TM = 2.664 · 10−6 rad/s. The Moon also rotates velocity of the Moon-borne is more than one order lower
around its axis with a period equal to TM . As a consequence (about 1/20) than that of standard LEO satellites. For instance,
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 FORNARO et al.: POTENTIALS AND LIMITATIONS OF MOON-BORNE SAR IMAGING 3011

Fig. 2. (Top curves) Aspect and (bottom curves) elevation angles for the first
sidereal month at 40◦ latitude.

ERS and Envisat sensor footprints move at a velocity of about

25 000 km/h (7000 m/s).

III. R ESOLUTION AND C OVERAGE Fig. 3. Moon–Earth reference geometry in the equatorial plane used for the
calculation of the azimuth resolution. We assume the Moon to be fixed and
Range resolution is governed by a transmitted pulse band- the Earth to be rotating. With respect to Fig. 1, we neglect the inclination of
width B. Similarly to standard LEO satellites the ground reso- the Earth rotation axis and of the segment connecting the Earth and Moon
lution depends on the elevation angle that, for the Moon-borne centers with respect to the ecliptic.
SAR case, can vary considerably in the sidereal month (see This angular amplification effect has an important impact on
Fig. 2). For a transmitted bandwidth of 100 MHz, we have a the achievable resolution. For the Moon-borne SAR being an
ground-range resolution (for flat Earth) of 2.3 m at an average inverse SAR, we use the Doppler approach. Hence, we start
value of the elevation of 50◦ at 40◦ latitude. with the computation of the Doppler bandwidth associated to
Evaluation of the azimuth resolution is a more complex the ground scatterer. Let us refer to Fig. 4, where the geometry
issue. In standard SAR imaging, the azimuth resolution is of the lunar SAR system in Fig. 3 is zoomed on the Earth
independent of the distance between the scene and the antenna surface; for comparison, we also show the geometry of the
[3]–[7]. A Moon-borne SAR has specific characteristics that LEO satellite case. In both figures, the unit vectors r̂1 and r̂2
must be analyzed separately. are the scatterer LOS vectors at the start and end instants of
To proceed along this line, referring to Fig. 1, we neglect the illumination time, respectively; similarly, vE1 and vE2 are
the inclination of the Earth and the inclination of the axis the scatterer velocity vectors at the start and end instants of the
connecting the Moon center to the Earth center. The Moon is illumination time, respectively.
still assumed to be fixed, whereas the Earth is rotating along an For a generic instant within the illumination time, with
axis orthogonal to the ecliptic. associated scatterer velocity vector vE and scatterer LOS vector
Fig. 3 shows this simplified Earth–Moon geometry in the r̂, the Doppler frequency shift associated to the movement is
equatorial plane. The aforementioned simplifications allow per- given by
forming simple geometrical derivations for the evaluation of
the azimuth resolution. In the real case, due to the presence of 2
fd = vE · r̂. (2)
additional angular variations between the scatterer line-of-sight λ
(LOS) vector and the scatterer tangential velocity vector, the
The Doppler bandwidth (BD ) is given by
resolution derived by following our approach can be slightly
overestimated. 2
BD = 2 (vE · r̂)max . (3)
Let L be the size of the antenna, mounted on the Moon λ
surface, and pointing the Earth. The 3-dB antenna aperture is
2ϑ = λ/L, where λ is the radiation wavelength; the resulting From Fig. 4, it is evident that the Moon-borne SAR is charac-
illuminated footprint is X ≈ λDM /L. terized by the fact that the scatterer velocity vector changes its
With simple geometrical considerations (see the Appendix), orientation when crossing the beam of the Moon SAR antenna.
we have the following compact relation between the antenna This fact, at variance of the satellite case where the velocity
semiaperture and the ground Earth angular semiaperture corre- vector describing the SAR antenna movement has a constant
sponding to the target semi-illumination time: orientation, leads to an increase of the Doppler bandwidth with
respect to the LEO spaceborne case.
DM For the Moon-borne case, we have
ϕ≈ ϑ. (1)
RE 4 4
BD = vE sin(ϕM + ϑM ) ≈ vE sin ϕM
λ λ
Equation (1) shows that the Earth angular semiaperture (ϕ, 4 DM 2vE DM
see Fig. 3) is amplified by a factor DM /RE with respect to the ≈ vE ϑM = (4)
λ RE L RE
antenna aperture ϑ. For instance, in the case of a C-band system
with an antenna dimension L of 84 m, we have ϑ ≈ .016◦ and where the first approximation follows from (1) and from the fact
ϕ ≈ 1◦ , so that the total ground angular illumination is of 2◦ . that DM  RE , the second follows from the fact that ϕ  1,
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 3012 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 7, JULY 2010

Fig. 4. Reference geometries for the computation of the Doppler bandwidth

in the (left) Moon-borne case and (right) standard LEO space-borne case.

and the last equality follows from the basic relation between the Fig. 5. Equivalent sliding spotlight geometry of a Moon-borne SAR in which
angular aperture of an antenna and its length. the Moon is moving and the Earth does not rotate.
For the LEO satellite case, moving with a velocity of v, we
have [3]–[7] latitude, we have that the Doppler bandwidth is almost constant
over the different latitudes, at least at low latitudes.
4 4 2v Another important remark concerns the effects of the long
BD−sat = v sin ϑM ≈ vϑM = . (5)
λ λ L integration time of the Moon-borne SAR, which is evaluated
in Section VI, on the imaging of nonstationary or partially co-
It is thus evident that, due to the presence of the factor herent scatterers, e.g., of sea and vegetation. The significant ef-
DM /RE , the Doppler bandwidth for the Moon-borne case is fects of azimuth misplacement and defocusing [17] that would
much larger than that of a standard spaceborne case. arise apart from very quiet days are discussed in Section VII,
The resolution Δy for the Moon-borne case is given by together with the possible impact in terms of signal-to-clutter
v L RE ratio (SCR). In this framework, a possible processing that is
Δy = = . (6) robust to critical situations, based on presuming [18], is also
BD 2 DM
discussed, and a useful, although costly, system extension in
Thus, it is DM /RE ≈ 60 times better than that of the LEO the multichannel SAR direction [19] is proposed in Section VIII
spaceborne case (L/2). This resolution gain effect has also after discussing the ATI option.
been highlighted in the study of medium Earth orbit SAR
systems [16]. IV. A MBIGUITY C ONSTRAINTS
The presence of the improvement factor DM /RE in the
expression of the azimuth resolution for the Moon-borne case In any SAR system, the pulse repetition frequency (PRF)
can also be explained in view of an equivalent geometrical fp is down- and upbounded. The first limitation on fp is
configuration. In fact, we may adopt an equivalent Earth–Moon related to the sampling of the azimuth bandwidth. Following
system representation where we assume the Earth to be fixed the equivalent geometric configuration, we have recognized that
and the Moon to be rotating as shown in Fig. 5 around the Earth the Moon-borne SAR can be considered as an intrinsic spotlight
(“Tolemaic” vision); in this case, the imaging system becomes system. Accordingly, we start with the analysis of the space-
similar to standard satellite SAR systems. In order to have the borne spotlight mode. A peculiarity of standard spotlight, as
same integration time of the case where the Moon is fixed, we well as sliding spotlight modes implemented on LEO satellites,
must have the footprint moving at the velocity vE . Accordingly, is that, although the Doppler bandwidth increases with respect
in the equivalent rotation around the Earth, the Moon must have to the stripmap case, the PRF can be kept limited to the Nyquist
a tangential velocity whose magnitude is scaled up by a factor frequency associated to the stripmap case [7]. In other words,
(DM + RE )/RE ≈ DM /RE with respect to vE . Similarly, the in a standard sliding spotlight case, fp−sat must satisfy the
aperture synthesis increases by the same factor, thus determin- following relation to guarantee the absence (in the case of an
ing the gain in the azimuth resolution. Moreover, the geometry ideal beam pattern) of azimuth ambiguities:
in Fig. 5 shows that, with this equivalence, the Moon SAR 2v
imaging is recognized to be intrinsically a sliding spotlight [8], fp−sat ≥ (7)
L
and the resolution gain factor is thus fully explained also from
this viewpoint. where v is the relative velocity between the satellite and ground
A remark is dedicated to the evaluated Doppler bandwidth. scatterers.
Due to a compensation effect between the velocity, which This property is based on the fact that, for a fixed antenna
decreases for an increase of the latitude, and the ground angular position, the Doppler bandwidth associated with the scene is
aperture which, on the contrary, increases for an increase of the equal to the expression at the right-hand side of (7). During
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 FORNARO et al.: POTENTIALS AND LIMITATIONS OF MOON-BORNE SAR IMAGING 3013

the acquisition, this bandwidth is translated in frequency as a Moon-borne SAR imaging may be obviously affected by more
result of the antenna steering, thus leading to an increased band- stringent power limitations. We refer to [7] where it is shown
width. However, such a translation does not strictly requires an that, in the absence of topography, the basic constraint on the
increase of the sampling frequency: Proper processing, which is achievable SNR for any satellite SAR system that satisfies the
able to take advantage of this frequency shift and of the fact that constraint on the antenna minimum effective area in (12) is
the ground swath for the spotlight mode is reduced with respect
PT λ 0 v cos ϑel
to the stripmap mode, can efficiently handle such a signal SN R ≥ σ (13)
undersampling (see [7] and [8] for more details). Unfortunately, KT F B πr c sin2 ϑel
0

in the case of a Moon-borne SAR system, this PRF relaxation where r is the distance between the scene and the sensor, PT
cannot be implemented because, in any given time instant, the is the mean transmitted power, T 0 is the absolute temperature,
Doppler bandwidth associated to the ground scene is increased F is the receiver noise figure, σ 0 is the scatterer radar cross
by a factor DM /RE . The resulting Nyquist condition for a section (RCS) normalized to the size of the ground resolution
correct sampling of the scene, and hence for avoiding azimuth cell, ϑel is the elevation angle, and K = 1.38 · 10−23 J/K is the
ambiguities, is therefore Boltzmann constant.
2vE DM 2vMa Taking advantage of the Tolemaic vision of the Moon-borne
fp ≥ ≈ (8) SAR, we apply (13) also to the lunar system; we get
L RE L
PT λ vMa cos ϑel
where vMa is the equivalent Moon velocity in the Tolemaic SN R ≥ σ0
vision (see Fig. 5) in which the Earth is fixed and the Moon KT 0 F B πDM c sin2 ϑel
PT λ 0 vE cos ϑel
rotates with an angular velocity equal to the one associated to ≈ σ . (14)
the Earth rotation: As discussed in the previous section, the KT 0 F B πRE c sin2 ϑel
Moon velocity is almost DM /RE higher than the velocity of Interesting enough, it can be noted from the last equation
the scatterers on the ground. that, in the case of lunar SAR, the SNR lower limit becomes
An upper limit to the PRF is associated with the necessity independent on the distance between the Earth surface and the
to avoid that echoes backscattered by successive pulses are Moon: This fact is a direct consequence of the increase in
received simultaneously. To this end, letting W to be the swath the azimuth resolution due to the spotlight effect. Nonetheless,
extension in the slant-range direction, we must ensure that [7] when vRE /(vE r) ≈ 90, a considerable power increase is in
c any case required in the Moon system. Moreover, in the lunar
fp ≤ . (9)
2W case, the elevation angle ϑel associated to the incidence radia-
tion significantly varies over the passes on a fixed ground scene.
Letting LR to be the antenna size in the direction orthogonal
to the azimuth and slant range, we have
VI. E XAMPLES
λ
W = DM cot ϑel (10) In order to get a feeling of what we should expect in the
Lr case of a lunar SAR system, we consider the following three
that corresponds to a ground-range swath equal to different configurations at L-, C-, and X-bands:
1) L-band (λL = 18 cm) and L = 180 m;
W λ 1 2) C-band (λC = 5.6 cm) and L = 84 m;
Wg = = DM . (11)
cos ϑel Lr sin ϑel 3) X-band (λX = 3.1 cm) and L = 62 m.
From (5)–(7), we have the following lower bound for the The aforementioned parameters have been selected accord-
antenna equivalent area: ing to ratios of the antenna length to the wavelength of 1000,
1500, and 2000 for the L-, C-, and X-bands, respectively. Such
2 values have been chosen in order to have reasonable ground
vMa vE DM
LLr ≥ 4 DM λ cot ϑel = 4 λ cot ϑel . (12) swath widths and, at the same time, to account for possible
c c RE
technological problems related to the design of antennas op-
Equation (12) shows the fact that the minimum antenna area erating at low frequency. The aforementioned values are only
of a lunar SAR increases with respect to standard satellite an example; other choices are obviously possible.
systems for two different reasons: an increase of the distance Table II provides the quantitative parameters of the three lu-
and a scaling of the velocity (vMa in place of vE in the middle nar system configurations with reference to an average value of
term). As a result, the antenna area increases with the square the elevation of 50◦ at 40◦ latitude; the Envisat ASAR satellite
of the distance between the antenna and the scene [see the last of the European Space Agency is reported for comparison.
equality in (12)]. Due to the increase of the distance between the antenna and
the scene, the Moon SAR systems have azimuth footprints that
are considerably larger than that of Envisat ASAR and of any
V. P OWER C ONSTRAINTS
typical satellite SAR system. The same reasoning applies to the
We now proceed with the evaluation of the transmitted aver- integration time T that can reach 14 min for the L-band case.
age power of a lunar SAR system and compare it to the average The Doppler bandwidths are rather comparable, whereas
power of a standard satellite case with the same minimum as anticipated in Section IV, the minimum antenna area is
achievable signal-to-noise ratio (SNR). Due to the substantial much larger than the satellite case, thus raising technological
increase of the distance between the system and the scene, problems for the antenna.
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 3014 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 7, JULY 2010

TABLE II critical. Aside from the possibility to adapt SAR focusing algo-
PARAMETERS OF THREE SELECTED MOON-BORNE SAR SYSTEMS
COMPARED TO THE ENVISAT ASAR SYSTEM
rithms designed for the problem of focusing data acquired by an
airborne platform in the presence of trajectory deviations [24],
more precise solutions are given in papers specifically devoted
to spaceborne high-resolution imaging [25], [26]. The ad hoc
combination of algorithms for trajectory deviation and track
curvature combination could also handle possible problems
related to the lunar libration.
In addition, the long integration time greatly enhances the
well-known imaging artifacts of nonstationary or partially co-
herent scatterers. For a scatterer having a LOS motion with
a velocity of vt , the azimuth position in the final image is
misplaced by

Δey = vt DM /vMa (15)

Some solutions (inflatable structures, membrane antennas, where vMa is the apparent orbital velocity of the Moon in the
etc.) are currently under development to build large antennas fixed Earth-rotating Moon (Tolemaic) framework, i.e., vMa ≈
for space applications (see [20] and [21] and reference therein). vE DM /RE = 27 km/s. Assuming, for vt , a value of 1 m/s, the
In addition, the technology of solar sails could be transferred to scatterer would be misplaced of about 14 km. More generally,
build large membrane microwave antennas [23]. However, for considering a scatterer with limited coherence time τc , the
lunar-base applications, (real) antenna arrays should be used, effective azimuth resolution Δye is degraded by a factor which
resulting in a simpler transportation and assembly. Finally, note is approximately the ratio of the integration time over the
that the derived values of the antenna size at C- and X-bands coherence time
are comparable to those of the antenna system used from the
Earth-based Moon imaging experiment in [4]. Δye = Δy T /τc . (16)
The maximum ground-range swath at C-band is up to four
times larger that the satellite case; however, due to the reduction In the worst case of scatterer coherence time in C-band
of the footprint velocity, the coverage rate is about four times of τc = 60 ms corresponding to a forest canopy with strong
lower. Obviously, an increase of the range swath and, therefore, wind (of 7–13 m/s wind velocity) [27], the azimuth resolution
of the coverage rate can be achieved by reducing the antenna severely degraded from 0.7 m to about 6.4 km (for wind
size in the cross-range direction: An increase of the azimuth velocity not higher than 3 m/s, coherence time can also be
antenna size is, in this case, required to comply with the antenna doubled, and defocusing was limited to 3.2 km). Note that this
area constraint, thus leading to a degradation of the azimuth degraded azimuth resolution can also be obtained by choosing
resolution. a reduced processing integration time limited to τc ; the rest
The transmitted power budget has also been evaluated ac- of the illumination time may be employed for multilooking.
cordingly to (14) by assuming the following: T 0 = 290 K, Similar misplacements and defocusing also cancel the possi-
SN R = 13 dB, F = 3 dB at L-band, F = 4 dB at C-band, bility of measuring signatures of ocean waves, as quantified in
and F = 5 dB at X-band. Moreover, we considered that σ0 = Section IX.
−19 dB at L-band, σ0 = −10 dB at C-band, and σ0 = −7 dB These blurring effects from nonstationary and partially co-
at X-band corresponding to average values of agricultural fields herent scatterers can also cause clutter power problems, with
for incidence angles of 40◦ [22]. The resulting values in Table II the energy of these scatterers being spread over adjacent areas
show that a considerable increase of the average power in the of stable scatterers. Concerning defocusing of a homogeneous
lunar SAR system is also required. Some solutions for the area, it is known that energy does not change, only redistributed
generation of several tens (and several hundreds) of kilowatts of [17]. The impact of blurring from nonhomogeneous areas is
electrical power have been proposed (for instance, photovoltaic rather complex because it depends also on the transition geome-
arrays with regenerative fuel cell energy storage, orbiting solar try between partially coherent and coherent areas. For a simple
power systems, nuclear power systems, and electrochemical yet representative nonhomogeneous scenario with a step tran-
systems) essentially for lunar-base applications. Further investi- sition along a straight line between two locally homogeneous
gation of these technological problems is, however, outside the areas, one partially coherent originating clutter and the other not
scope of this paper. subject to defocusing, characterized by scattering levels σc0 and
σs0 , respectively, the energy leakage from the partially coherent
area to the coherent area in the vicinity of the transition can
VII. E FFECTS OF L ONG I NTEGRATION T IME
be evaluated as σc0 /2. Thus, the SCR would be, in the worst
Concerning the large increase of the integration time com- case (i.e., in the vicinity of the transition), equal to SCR|dB =
pared to typical satellite SAR, this exhibits its main effects on σs0 |dB − σc0 |dB + 3, with the leakage tending practically to zero
the imaging system. First of all, the effects of the orbit curvature at a distance of one degraded effective azimuth resolution
of the Moon-borne SAR increase with respect to the spaceborne cell Δye . Note that if the transition area would be somewhat
case: The SAR focusing algorithm, which generally assumes aligned with the azimuth direction, the effective distance of
the SAR system moving on a straight line, should account for leakage penetration in the coherent area would be reduced. In
this effect. Similarly, effects of the range migration are more addition, it is worth noting that σc0 is not the total normalized
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 FORNARO et al.: POTENTIALS AND LIMITATIONS OF MOON-BORNE SAR IMAGING 3015

to exactly null the signal from an hypothetical scatterer with

a LOS velocity of 1 m/s. Extending the concept in [29], the
presuming time window should equal the time the scatterer
needs to move half a wavelength, i.e., 28 mm. Assuming, for
the sake of simplicity, a PRF at the Nyquist limit (656 Hz), this
means presuming 18 azimuth samples. Note that the azimuth
resolution would be reduced from 0.7 to 12.6 m, which is still
very good. In practice, scatterers with higher velocities, nonuni-
form motions, and also periodic motions (with similar peak
instantaneous velocities) may be attenuated. Another possible
Fig. 6. Mean one-day coherence distribution at C-band evaluated from six remedy to defocus-originated clutter problems may be (partial)
ERS-1/ERS-2 tandem acquisitions over the Abruzzi region in Italy. resolution recovery, which requires more complex (multichan-
nel) hardware and is discussed at the end of Section IX.
RCS of the partially coherent area but, generally, a fraction
1/(1 + rcv ) of it, with rcv being the power ratio of the constant-
to-varying scattering components [27] (e.g., ground and trunk, VIII. C ROSS -T RACK I NTERFEROMETRY
and branches and leaves, respectively). With respect to existing LEO satellites, the main peculiarity
It is easily recognized that, among the possible sources of of Moon-borne SAR imaging is related to the site accessibility
clutter, the sea (or, more generally, water) can be considered to which is reduced to one day. Behind daily imaging potentiality,
be not of great concern, unless for very windy days, with the it is therefore worth to investigate the potential implementa-
scattering level being significantly less (at least, a few decibels tion of repeat-pass interferometry at short temporal separation,
less) than other areas, e.g., urban extensions and (agricultural) which is fundamental for the differential interferometry tech-
fields. For instance, assume that, for water with strong wind nique [12], [30] widely used to monitor ground deformations.
(of 10 m/s wind velocity), σc0 = −16 dB, and that, for fields, In this regard, a Moon-borne SAR could provide information
σs0 = −9 dB in C-band [28]. For the simple transition geometry, complementary in terms of measurable components of ground
we would get SCR|dB = 10 dB in the vicinity of the interface. deformation (i.e., the north–south deformation components)
Conversely, clutter from a strong wind-blown forest might with respect to that commonly provided by existing satellites.
exhibit higher power. Assuming, for C-band, a pessimistic Decorrelation is, however, the primary source of noise in
value of σc0 = −6 dB, the SCR for fields would decrease to SAR interferometry that may prevent the application of interfer-
SCR|dB = 0 dB, although for urban areas (σs0 = −4 dB [28]), ometric processing [10], [11] and, therefore, must be analyzed
a reasonable SCR|dB = 5 dB would still be obtained. On the in detail for the Moon-borne SAR system. Here, we start with
other hand, for wind velocity not higher than 3 m/s, it can the analysis of the decorrelation issue in the framework of
be estimated that even the forest clutter would not constitute repeat-pass interferometry by assuming the presence of a single
a significant limitation: In this condition, we can assume that SAR system located on the Moon surface. We specifically
rcv ≈ 10 [27]. Thus, the effective cluttering σc0 (possibly from refer to the problems associated with the spatial decorrelation
leaves and small branches only) would be an order of magnitude caused by the appearance of the spectral shift effect [31] in
less (≈ −16 dB), and SCRs are correspondingly higher, in the radar response and due to the presence of variations of the
addition to the reduced distance of leakage penetration. Ob- incidence angle between different acquisitions, i.e., the angle
viously, the clutter problem increases at X-band and reduces between the normal to the surface and the direction of the
at L-band where the penetration of the radiation within the radiation propagation vector (LOS). More precisely, we neglect
canopy is higher. To have an idea of the impact of possible wind the presence of topography variations on the Earth and hence
and constant scattering components in forest areas, we show refer to a flat surface. This allows carrying out the analysis
in Fig. 6 the distribution of the one-day coherence at C-band with respect to the elevation angle, which is, in this case, the
computed by using six ERS1/ERS2 tandem acquisitions: In complement of the incidence angle. Other decorrelation sources
particular, we selected an area of about 100 km × 100 km in the such as, for instance, temporal decorrelation [10] are similar to
Abruzzi region in the central part of the Appennini mountains in spaceborne SAR system and thus are not analyzed.
Italy. Even though representative only of a single test area, this We recall that, for a distributed scattering and with reference
plot shows that more than the 60% of the image has coherence to flat surfaces, the radar return shows a complete decorrelation
values which are higher than 0.3. at an elevation angle difference (Δϑ) between the different
To reduce the possible criticalities in the vicinity of non fully passes exceeding the following limit [4]:
coherent areas, a synthetic aperture processing that is robust to
energy leakage may be achieved by resorting to presuming, as λB λ
hinted in [18]. Azimuth presuming reduces the antenna foot- Δϑmax = = (17)
c tan ϑel 2Δr tan ϑel
print and azimuth resolution and is sometimes used for quick-
look processing or data rate reduction. For a slow-moving where B is the transmitted bandwidth and Δr is the slant-range
SAR, the signal from a moving or decorrelating scatterer can resolution. Let us start by considering the ERS and Envisat
be canceled or attenuated, being averaged out during azimuth satellite reference case.
presuming [18], [29], so that null or little residual energy would Assuming the presence of a distributed scattering over a
clutter the coherent areas. ground cell of about 20 m, we have a total (spatial) decorre-
With reference to the C-band parameters of the Moon-based lation for Δϑ ≥ 0.07◦ . Note that, as the scene elevation angle
SAR, consider, as an example, to design the presuming factor does not significantly change over the different passes, this
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 3016 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 7, JULY 2010

however, that the maximum elevation angle difference in (17)

depends itself on the elevation angle. In other words, due to
the variation of the elevation angle, the ground-range resolution
changes; so does the sensitivity of the scattered return to the
angular variations of the interferometric system. For this reason,
in Fig. 7, we have also overlaid the plot of −Δϑmax and
Δϑmax , computed from (17), and the Moon’s orbit information,
in order to locate the regions of useful elevation variation for
the C-band (dashed curves) and L-band (dashed–dotted curves).
From this plot, it is evident that the most favorable conditions
to implement one-day repeat-pass interferometry are obtained
when the Moon elevation angle is minimum: In this case, the
ground-range resolution cell achieves its minimum value, and
the Moon elevation angle variation is closer to zero among all
the other days of the sidereal month.
Nonetheless, at C-band, the acquisitions that are useful for
one-day repeat-pass interferometry are rather limited. Problems
get worse at X-band cases, whereas at L-band, we have almost
that half of the sidereal month is useful for one-day interferom-
etry: However, it should be noted that the implementation of
L-band systems with 100-MHz bandwidth, necessary for a
slant-range resolution of 1.5 m, is a complex task from the
Fig. 7. (Top) Elevation angle in degrees and (bottom) its variation day per hardware viewpoint.
day at 40◦ latitude. Dashed and dashed-dotted lines locate the maximum A possible cure to the angular decorrelation problem could
elevation angle difference curves for repeat-pass interferometry for C and
L band, respectively.
be associated with the use of a frequency-tunable radar system:
The system operates a variation of the transmitted frequency
maximum elevation angle difference is almost constant. This in order to compensate the spectral shift affecting the imaged
means that we are prevented to implement interference between backscattering function [31]. Anyway, such a solution would,
radar images of ground scatterers whenever the elevation angle first of all, assume the backscattering to be distributed on flat
difference exceeds the limit of 0.07◦ , which corresponds to a surfaces, thus limiting the coverage in a real scenario with
separation between the orbits (usually referred to as orthogonal topographic variations.
baseline) of about 1100 m. Another solution, costly from the hardware and payload
In the Moon-borne case, assuming an elevation of 50◦ (the viewpoint, which could allow increasing the number of corre-
mean value of the elevation at 40◦ latitude), the maximum lated acquisitions, is related to the possibility to install more
elevation angle difference at C-band is about 0.9◦ . This increase than a single SAR sensor on the Moon surface. For example, a
is a direct consequence of the increase in the slant-range second SAR sensor, with a lunar displacement of thousands of
resolution. kilometers over the first one, could allow gaining an elevation
To implement repeat-pass interferometry, it is demanding to angle difference of about 0.5◦ in the best case. This means
investigate whether the variation, from day to day of the Moon that the dash and dash–dot curves at the bottom of Fig. 7
elevation, is below such a limit. As already stated, due to the should be widened by 0.5◦ , and therefore, the gain on the
rotation of the Moon around the Earth, the elevation angle has a
useful acquisition would decrease. It should also be pointed out
large excursion in the sidereal month. In the upper plot in Fig. 7,
that the atmosphere represents another problem. As a matter
the elevation angle is shown, and in the lower image, the starred
of fact, despite the fact that the most critical contribution for
curve reports the variation of the elevation angle between
two successive days for the first 100 days (first four sidereal interferometric application is associated to the troposphere, the
months): The reference latitude is 40◦ , and the curves have use of lower elevation angles (i.e., higher incidence angles)
been derived from precise Moon’s orbit information provided implies an increase of atmospheric disturbances. This problem
by Galilean Plus. Obviously, repeat-pass interferometry could is even more pronounced at higher latitudes.
be, in principle, implemented at temporal spans corresponding Indeed, the possibility to install different antennas on the
to the sidereal month (about 27 days) and multiple. On the Moon surface that collect simultaneous radar images from
other hand, it is evident that the presence of a large amount different elevation angles is a unique possibility offered by
of variation of the maximum elevation angle from one day to the use of the natural Earth satellite. Single-pass cross-track
the other may impair implementation of interferometry between interferometry could also be implemented, and even 3-D tomo-
data acquired at few-day differences. graphic imaging [14], [15] could be explored. This is at variance
By analyzing the upper plot in Fig. 7, it could be argued that, of existing SAR technology on LEO satellites that precludes
around the peaks and valleys of the elevation curve, where the the possibility to acquire multiview data simultaneously. This
variation is minimal (see the starred curve in the lower plot), fact poses limitations from the interferometric point of view
one might achieve useful acquisitions to implement one-day because of the presence of temporal decorrelation: Differ-
temporal separation repeat-pass interferometry. We observe, ent views are acquired only with repeat passes. Single-pass
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 FORNARO et al.: POTENTIALS AND LIMITATIONS OF MOON-BORNE SAR IMAGING 3017

interferometry could be used to evaluate accurate digital ele- with a finer ground-range resolution τc can be larger, but the
vation models (DEM) and DEM variations, a possibility that order of magnitude of the defocusing effect still stands. Note
is being pursued also with upcoming next-generation satellite that a similar SAR resolution loss effect exists also for LEO
tandem configurations (see, e.g., [32] and [33]). Application of spaceborne systems, but it is much less severe: For the Envisat
tomography to single-pass multiple acquisitions would allow ASAR satellite, the nominal 5 m azimuth resolution becomes
generating more accurate 3-D images that could allow better around 175 m over the ocean, in the same conditions. We
analysis of the vertical structure of the backscattering. conclude that a Moon-borne SAR is not a viable alternative to
conventional ATI.
Conversely, the potential of lunar SAR ATI shows to be
IX. ATI
more interesting for monitoring distributed targets with quick
ATI exploits the along-track baseline between two SAR deformation motions such as of quick landslides and quick
antenna to obtain a small time lag τATI between the two glacier flows. As an example for a quick landslide scenario,
corresponding SAR image acquisitions. Estimation of the phase assuming that τc can also be larger than 9 min (see [36]), the
difference between corresponding pixels, possibly with coher- defocusing effect is negligible. A τATI up to about 1 min can
ent multilooking, allows radial velocity measurement of fast be used, which is much less than τc , and that is obtained by
moving scatterers, also in the case of distributed scatterers [10] an along-track sensor spacing up to dATI ≈ 3300 km (note
(for possible benefits from the use of more than two channels that a maximum possible spacing, neglecting limitations from
in ATI, see [32] and references therein). Due to the intrinsic Moon libration, is 2RM = 3476 km). The resulting velocity
restriction of the phase difference in a 2π interval, measurable unambiguous range would be at least on the order of some
velocities are restricted in unambiguous velocity range Δvr centimeters per minute (meters per hour).
which depends on τATI : For two-antenna ATI Another useful application of the along-track configuration
to the Moon-borne SAR may be to resort to a (costly) multi-
λ
Δvr = . (18) baseline ATI to get a multichannel SAR with resolution-recover
2τATI capabilities [19] for the partially coherent areas. A multichannel
Assuming the two antennas of the interferometric system to SAR with K antennas (only one active) can be used to produce
be separated by a distance dATI along the azimuth direction K (resolution-degraded) SAR images with multiple (K − 1)
(along-track baseline) and that only one antenna is active, the time lags. By performing an interchannel Fourier analysis, it is
time lag is given by [10] possible to estimate the Doppler spectrum from the multiple
velocity-spread scattering contributions that were misplaced
dATI into the degraded-resolution complex pixel. By Doppler-driven
τATI = (19)
2vMa relocation of the corresponding contributions, a refocused SAR
image can be obtained [19]. By proper design, resolution can
where vMa is the apparent orbital velocity of the Moon in the
be recovered by a factor K. As an example, for K = 5, the
fixed Earth-rotating Moon (Tolemaic) framework.
recovered C-band resolutions over ocean and over strong wind-
We first consider the ability of the system to potentially
blown forest would be Δye /5 = 3800 m and Δye /5 = 1300 m,
measure the velocity of surface ocean currents. Long (SAR
respectively, thus reducing the clutter problem.
resolved) wave orbital motions [35] cannot be imaged because
of the strong defocusing effect mentioned in Section VII. In
X. C ONCLUSION
this case, the time lag should be lower than the correlation time
of the speckle in the SAR ocean image, the so-called ocean This paper has analyzed the potentialities and drawbacks
coherence time τc [13], which is basically determined by the related to microwave EO with SAR from the Moon.
velocity spread of the resonant ripple Bragg waves because of On the one hand, a major problem of lunar EO with respect
modulation from medium (subresolution-scale) gravity waves to standard lower Earth orbit satellites is associated to the
[35]. At C-band, and for a ground-range resolution of 20 m, substantial increase of the distance between the scene and the
the minimum τc (for wind speed higher than about 10 m/s, sensor. In the case of SAR systems, this fact plays an important
corresponding to strong wind) is around 20 ms (see references role in the power budget, thus leading to a transmitted power
in [32]). By using τATI ≈ τc /2, i.e., 10 ms, the two SAR level demand that can be hardly obtained with standard solar
sensors should be thus spaced by dATI = τATI 2vMa ≈ 0.55 km cell energy technology. On the other hand, due to a peculiar
along the track. The resulting horizontal velocity unambiguous imaging effect due to Earth rotation, a Moon-borne SAR can
range would be Δvrg = Δvr / cos(ϑel ) = 4.4 m/s for ϑel = overcome the limit of half antenna length resolution, and thus,
50◦ , which would be enough to handle typical ocean surface it can achieve high-resolution capabilities, up to 1 m azimuth
velocities without the need of phase unwrapping. Note that, in resolution. This fact allows placing large antennas without
the ATI processing, small Moon–Earth distance changes during paying the price to lower the resolution to insufficient limit
the time lag τATI should be possibly compensated. and to handle the aforementioned power limitation problem.
However, when considering the effective spatial resolution Obviously, the processing becomes more complex.
of the Moon-borne SAR and, hence, of the along-track interfer- Aside from the possibility to steer the antenna beam on a
ometer for the marine application, from the considerations in very wide Earth region, Moon’s orbit may, in principle, allow
Section VII, we get for the C-band parameters an effective az- implementation of repeat-pass interferometry. With respect to
imuth resolution of Δye = Δy T /τc = 19.2 km. Unfortunately, standard LEO satellites, orbit maneuvers cannot be imple-
this highly degraded azimuth resolution also severely impacts mented, and this fact calls for a more precise study of the lunar
the practical possibility of measuring ocean currents. Operating orbits. On the other hand, lunar surface allows the installation
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 3018 IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING, VOL. 48, NO. 7, JULY 2010

of different sensors, to implement single-pass interferometry, as [6] B. A. Campbell, D. B. Campbell, J. L. Margot, R. R. Ghent, M. Nolan,
well as ATI and tomographic imaging. J. Chandler, L. M. Carter, and N. J. S. Stacy, “Focused 70-cm wavelength
radar mapping of the moon,” IEEE Trans. Geosci. Remote Sens., vol. 45,
Finally, the integration time for full focusing is rather large: no. 12, pp. 4032–4042, Dec. 2007.
This, on one side, prevents the possibility to image “fast” [7] G. Franceschetti and R. Lanari, Synthetic Aperture Radar Processing.
moving targets and leads to an increase of the clutter levels, but Boca Raton, FL: CRC Press, 1999.
on the other side, it allows implementation of moving (nondis- [8] W. G. Carrara, R. S. Goodman, and R. M. Majewski, Spotlight Synthetic
Aperture Radar: Signal Processing Algorithms. Norwood, MA: Artech
tributed) target detection even starting from single-antenna data House, 1995.
by synthetic aperture splitting [9]. [9] J. M. B. Dias and P. A. C. Marques, “Multiple moving target detection
All the aforementioned issues are certainly of interests for and trajectory estimation using a single SAR sensor,” IEEE Trans. Aerosp.
Electron. Syst., vol. 39, no. 2, pp. 604–624, Apr. 2003.
many scientific applications and raise the calls for future more [10] P. A. Rosen, S. Hensley, I. R. Joughin, F. K. Li, S. N. Madsen,
detailed study, possibly based on precise lunar orbit determi- E. Rodriguez, and R. M. Goldstein, “Synthetic aperture radar interferom-
nation, scene coverage, and region accessibility, that might etry,” Proc. IEEE, vol. 88, no. 3, pp. 333–382, Mar. 2000.
consolidate the results of this study and further investigate the [11] R. Bamler and P. Hartl, “Synthetic aperture radar interferometry,” Inverse
Probl., vol. 14, no. 4, pp. R1–R54, Aug. 1998.
potentialities and limitations of an active microwave remote [12] A. K. Gabriel, R. M. Goldstein, and H. A. Zebker, “Mapping small ele-
sensing of the Earth from a lunar base. vation changes over large areas: Differential interferometry,” J. Geophys.
Res., vol. 94, no. 7, pp. 9183–9191, 1989.
[13] R. M. Goldstein and H. A. Zebker, “Interferometric radar measurements
A PPENDIX of ocean surface currents,” Nature, vol. 328, no. 6132, pp. 707–709,
Aug. 1987.
The analysis of the achievable azimuth resolution is carried [14] A. Reigber and A. Moreira, “First demonstration of airborne SAR to-
out by following a Doppler approach, because for the Moon mography using multibaseline L-band data,” IEEE Trans. Geosci. Remote
imaging system, the target is moving as in an inverse SAR. Sens., vol. 38, no. 5, pp. 2142–2152, Sep. 2000.
To evaluate the azimuth resolution, we need to compute the [15] G. Fornaro, F. Lombardini, and F. Serafino, “Three-dimensional multi-
pass SAR focusing: Experiments with long-term spaceborne data,” IEEE
Doppler bandwidth associated to the illuminated footprint, Trans. Geosci. Remote Sens., vol. 43, no. 4, pp. 702–714, Apr. 2005.
whose inverse value gives the temporal resolution of ground [16] K. Raney, “Consideration for SAR image quantification unique to orbital
scatterers moving at a velocity equal to v [3]–[7]. To this end, systems,” IEEE Trans. Geosci. Remote Sens., vol. 29, no. 5, pp. 754–760,
we start by evaluating the relation between ϑ and ϕ in Fig. 2. Sep. 1991.
[17] K. Raney, “SAR response to partially coherent phenomena,” IEEE Trans.
Simple geometrical computations lead to the following: Antennas Propag., vol. AP-28, no. 6, pp. 777–787, Nov. 1980.
[18] F. Prati, F. Rocca, D. Giancola, and A. Monti Guarnieri, “Passive geo-
RE sin ϕ = [DM + RE (1 − cos ϕ)] tan ϑ. (20) synchronous SAR system reusing backscattered digital audio broadcast-
ing signals,” IEEE Trans. Geosci. Remote Sens., vol. GRS-36, no. 6,
For small values of ϕ and ϑ, expansion to the first order leads pp. 1973–1976, Nov. 1998.
[19] B. Friedlander and B. Porat, “VSAR: A high resolution radar system
to the following simple result: for ocean imaging,” IEEE Trans. Aerosp. Electron. Syst., vol. 34, no. 3,
DM pp. 755–776, Jul. 1998.
ϕ≈ ϑ. (21) [20] A. Moussessian, C. Chen, W. Edelstein, S. Madsen, and P. Rosen, “System
RE concepts and technologies for high orbit SAR,” in Proc. IEEE MTT-S Int.
Microw. Symp. Dig., Jun. 12–17, 2005, pp. 1623–1626.
Equation (21) shows that the Earth illumination angular [21] W. Edelstein, S. Madsen, A. Moussessian, and C. Chen, “Concepts and
semiaperture (ϕ) is amplified by a factor DM /RE with respect technologies for synthetic aperture radar from MEO and geosynchronous
orbits,” in Proc. SPIE Remote Sens. Symp., Honolulu, HI, Nov. 8–12,
to the antenna semiaperture ϑ. 2004, pp. 195–203.
[22] F. T. Ulaby, R. K. Moore, and A. K. Fung, Microwave Remote Sensing.
ACKNOWLEDGMENT London, U.K.: Addison-Wesley, 1972.
[23] M. Leipold, H. Runge, and C. Sickinger, “Large SAR membrane antennas
The authors would like to thank C. Ananasso of the Italian with lightweight deployable booms,” Proc. 28th ESA Antenna Workshop
Space Agency for the support and G. Vulpetti of the Galileian Space Antenna Syst. Technol. ESA/ESTEC, May 31–Jun. 3, 2005.
[Online]. Available: http://www.dlr.de/fa/en/Portaldata/17/Resources/
Plus Italy for providing the Moon’s orbit via the computer pro- dokumente/publikationen/2005/11\_leipod.pdf
gram: Spacecraft/Sailcraft Mission Analysis Code (S.M.A.C.), [24] A. Moreira and Y. Huang, “Airborne SAR processing of highly squinted
version R.03.15, February 2007, (Copyright, G. Vulpetti 1983, data using a chirp scaling approach with integrated motion compensa-
2007). The authors would also like to thank the anonymous tion,” IEEE Trans. Geosci. Remote Sens., vol. 32, no. 5, pp. 1029–1040,
Sep. 1994.
reviewers for their valuable and constructive comments that [25] K. Eldhuset, “A new fourth-order processing algorithm for spaceborne
helped improve the quality of this paper. SAR,” IEEE Trans. Aerosp. Electron. Syst., vol. 34, no. 3, pp. 824–835,
Jul. 1998.
[26] D. D’Aria and A. Monti Guarnieri, “High-resolution spaceborne SAR
R EFERENCES focusing by SVD-Stolt,” IEEE Geosci. Remote Sens. Lett., vol. 4, no. 4,
[1] International Conference on Exploration and Utilisation of the Moon pp. 639–643, Oct. 2007.
(ICEUM9), series, [Online]. Available: http://sci.esa.int/ilewg [27] Y. Dong, “Models of land clutter vs grazing angle, spatial distribution
[2] B. J. Eastlund and L. M. Jenkins, “Mission for Planet Earth: Defining a and temporal distribution. L-band VV polarisation perspective,” DSTO
vision for the space program,” in Proc. IEEE Aerosp. Conf., 2004, p. 603. System Sci. Lab., Edinburgh, South Australia, DSTO Research Report
[3] J. C. Curlander and R. N. McDonough, Synthetic Aperture Radar: Systems 0273, Mar. 2004.
and Signal Processing. New York: Wiley, 1991. [28] L. Kenyi and M. Schardt, “Application of Envisat ASAR data for alpine
[4] S. Hensley, E. Gurrola, P. Rosen, M. Slade, J. Jao, M. Kobrick, B. Wilson, land use classification,” in Proc. Envisat ERS Symp., Salzburg, Austria,
C. Chen, and R. Jurgens, “An improved map of the lunar south pole Sep. 2004.
with earth based radar interferometry,” in Proc. IEEE Radar Conf., 2008, [29] L. Cazzani, C. Colesanti, D. Leva, G. Nesti, C. Prati, F. Rocca, and
pp. 1–6. D. Tarchi, “A ground-based parasitic SAR experiment,” IEEE Trans.
[5] J. L. H. Webb, D. C. Munson, and N. J. S. Stacy, “High-resolution plan- Geosci. Remote Sens., vol. 38, no. 5, pp. 2132–2141, Sep. 2000.
etary imaging via spotlight-mode synthetic aperture radar,” IEEE Trans. [30] A. Ferretti, C. Prati, and F. Rocca, “Nonlinear subsidence rate estima-
Image Process., vol. 7, no. 11, pp. 1571–1582, Nov. 1998. tion using the permanent scatterers in differential SAR interferometry,”
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.
 FORNARO et al.: POTENTIALS AND LIMITATIONS OF MOON-BORNE SAR IMAGING 3019

IEEE Trans. Geosci. Remote Sens., vol. 38, no. 5, pp. 2202–2212, Fabrizio Lombardini (S’96–M’97–SM’03) re-
Sep. 2000. ceived the Laurea degree (with honors) in electronic
[31] F. Gatelli, A. Monti Guarnieri, F. Parizzi, P. Pasquali, C. Prati, and engineering and the Ph.D. degree in telecommunica-
F. Rocca, “The wavenumber shift in SAR interferometry,” IEEE Trans. tion engineering from the University of Pisa, Pisa,
Geosci. Remote Sens., vol. 32, no. 4, pp. 855–865, Jul. 1994. Italy, in 1993 and 1997, respectively.
[32] G. Krieger, A. Moreira, H. Fiedler, I. Hajnsek, M. Werner, M. Younis, He was a Postdoctoral Researcher with the De-
and M. Zink, “TanDEM-X: A satellite formation for high resolution partment of Electronic and Electrical Engineering,
SAR interferometry,” IEEE Trans. Geosci. Remote Sens., vol. 45, no. 11, University College London, U.K., from 1998 to 1999
pp. 3317–3341, Nov. 2007. under a Marie Curie Fellowship of the Training and
[33] H. Friedler, G. Krieger, F. Jochim, M. Kirschner, and A. Moreira, “Analy- Mobility of Researchers Program granted by the EC.
sis of satellite configurations for spaceborne SAR interferometry,” in He is currently with the Department of “Ingegneria
Proc. Int. Symp. Formation Flying Missions Technol., 2002. dell’Informazione,” University of Pisa, as an Assistant Professor. His general
[34] F. Lombardini, F. Bordoni, F. Gini, and L. Verrazzani, “Multibaseline ATI- interests are in the areas of statistical signal processing, estimation and detec-
SAR for robust ocean surface velocity estimation,” IEEE Trans. Aerosp. tion theory, adaptive and superresolution spectral analysis, array processing,
Electron. Syst., vol. 40, no. 2, pp. 417–433, Apr. 2004. nonuniform sampling, multidimensional signal processing, and performance-
[35] K. Hasselmann, R. K. Raney, W. J. Plant, W. Alpers, R. A. Shuchman, bounds evaluation with application to radar systems. In particular, his research
D. R. Lyzenga, C. L. Rufenach, and M. J. Tucker, “Theory of synthetic interests include multibaseline and multifrequency interferometric SAR algo-
aperture radar ocean imaging: A MARSEN view,” J. Geophys. Res., rithms and systems for environmental remote sensing, both cross- and along-
vol. 90, no. C3, pp. 4659–4686, 1985. track 3-D SAR tomography, differential SAR interferometry, coherent SAR
[36] C. Werner, T. Strozzi, A. Wiesmann, and U. Wegmüller, “A real- data combination, multisensor data fusion, distributed detection, and radar
aperture radar for ground-based differential interferometry,” in Proc. detection in non-Gaussian clutter. He has given lectures at universities and
IEEE IGARSS, Boston, MA, Jul. 2008, pp. 210–213. institutions in Italy and abroad, has chaired special sessions on SAR multibase-
line/multichannel interferometry/3-D techniques at international conferences,
and was guest Coeditor of the EURASIP Journal of Applied Signal Processing
special issue on Advances in Interferometric SAR Processing in 2005. He has
been Coauthor or Lecturer of tutorials on multibaseline/multidimensional SAR
Gianfranco Fornaro (M’06–SM’08) received the tomography.
Dr. Eng. degree (summa cum laude) in electronic
engineering and the Ph.D. degree from the Univer-
sity “Federico II,” Napoli, Italy, in 1992 and 1997,
respectively.
Since 1993, he has been with the “Istituto per il
Rilevamento Elettromagnetico dell’Ambiente” (for-
merly IRECE), Italian National Research Council
(CNR), where he is currently a Senior Researcher.
He was also an Adjunct Professor of communication
with the University of Napoli “Federico II” and
University of Cassino, Cassino, Italy, and of signal theory with the University
of Reggio Calabria, Reggio Calabria, Italy. He was a Visiting Scientist with the Alessandro Mori received the Laurea degree (cum
German Aerospace Agency (DLR), Oberpfaffenhofen, Germany, also during laude) in electronic engineering and the Ph.D. degree
the SIR-C/X-SAR mission in 1994, and with the Politecnico of Milano, Milan, in telecommunication and information engineering
Italy. He was an invited Lecturer with the Istituto Tecnologico de Aeronautica, from the University of Florence, Florence, Italy, in
Sao José dos Campos, Brazil, and RESTEC, Tokyo, Japan. He was responsible 1995 and 1998, respectively.
of the remote sensing unit with the Campania Regional Center of Competence He is currently an external Collaborator with the
“Analysis and Monitoring of the Environmental Risk.” He has authored more Applied Electromagnetism Lab team, Department
than 100 papers in international peer-reviewed journals and proceedings of of Electronics and Telecommunication, University
international conferences in the synthetic aperture radar (SAR) processing field. of Florence. His research activities include remote
He has served as an Editor of the “Advances in Interferometric SAR Process- sensing and numerical methods in electromagnetic
ing” special issue of the EURASIP Journal on Applied Signal Processing. His scattering and radiation. He is currently doing re-
main research interests include the signal processing field with applications to search on computationally efficient full-wave methods for the analysis of
airborne and spaceborne SAR data processing, SAR interferometry, differential electromagnetic scattering and radiation.
SAR interferometry, and 3-D and 4-D SAR imaging.
Dr. Fornaro was awarded the Mountbatten Premium by the Institution of
Electrical Engineers in 1997.

Giorgio Franceschetti (S’60–M’62–SM’73–F’88–

LF’01) was a Distinguished Visiting Scientist with
Jet Propulsion Laboratory. He was a Lecturer with
Space-Tech Master, Delft University. He was an Ad-
junct Professor with UCLA, in 1992–2008, a Visiting Mario Calamia has been a full Professor in electro-
Professor in Europe, U.S., and Somalia, and a Lec- magnetics with the University of Florence, Florence,
turer in China and India. He is currently a Professor Italy, and an Associate Professor in radar systems
with University Federico II of Napoli, Napoli, Italy. with the Italian Navy Academy, Leghorn, Italy. He
He is the author of over 180 (peer-reviewed) papers was the former General Director of the Italian Space
and ten books. Agency and former Italian Delegate to the UE’s
Prof. Franceschetti was the recipient of several Space Advisory Group. He had been the Italian
awards, culminating with the gold medal from the President of Italy (2000) Project Scientist for three missions dedicated to Earth
and elevation to the grade of Officer of Italy Republic (2001). His most studies from space (SIR-C/X-SAR and SRTM),
recent recognitions have been the prestigious 2007 IEEE GRS-S Distinguished done in collaboration with NASA (U.S.), DGLR
Achievement Award for “outstanding research in Electromagnetics, Propaga- (Germany), and ASI (Italy). He is the author of more
tion, Remote Sensing and Information Data Processing,” and the 2009 NASA than 70 papers on topics regarding radar, microwaves, and electromagnetic
Group Achievement Award, Cassini Radar Team, for “outstanding accomplish- scattering.
ment in the acquisition and analysis of Cassini Radar data, contributing to a Prof. Calamia is a member of the Italian Electrotechnical and Electronic
better understanding of Titan and the Saturn system.” Association.
Authorized licensed use limited to: University of Chinese Academy of SciencesCAS. Downloaded on July 18,2024 at 12:31:25 UTC from IEEE Xplore. Restrictions apply.