Escolar Documentos
Profissional Documentos
Cultura Documentos
IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 8, NO. 1, JANUARY 2011 49
Abstract—This letter compares the coseismic deformation maps among the main shock location, the highest damage areas, and
obtained from different synthetic aperture radar (SAR) sensors the aftershock seismicity patterns.
using the well-known differential SAR interferometry technique. The clear identification of the fault was obtained only six
In particular, four deformation maps have been obtained from X-,
days after, when the first differential synthetic aperture radar
C-, and L-band SAR sensors onboard COSMO-SkyMed, Envisat,
and ALOS satellite missions correspondingly. The test case is the (SAR) interferometry (DInSAR) interferograms yielded a high-
April 6, 2009, earthquake (Mw = 6.3). This seismic event struck resolution map of the coseismic deformation which constrained
a densely populated region of the Apennines and was felt all accurate models of the dislocation [4].
over Central Italy. The SAR data set is rather inhomogeneous, In fact, soon after the main shock, the entire “constellation”
since it includes interferograms with three different wavelengths, of SAR and optical satellites was activated to cover the epi-
four acquisition geometries, different spatial resolutions, variable central region. The first SAR data came, with a delay of few
temporal and spatial baselines, and differently emphasized signal
hours only, from Radarsat, but unfortunately, the interferogram
noise. However, we find that the detected displacements are highly
comparable. The outcome of this work is that, even though such was too noisy to give any useful information [5]. The first us-
differences have an impact on the properties of the interferograms, able interferograms were obtained from Envisat and COSMO-
the displacements can be measured with an overall discrepancy of SkyMed data six days later (on April 12). This was, in fact, the
about half the value of the shortest wavelength (COSMO-SkyMed) first time in which an X-band interferogram showed the com-
data set. plete surface displacement field of an earthquake. COSMO-
Index Terms—Abruzzi earthquake, differential interferometry, SkyMed interferometry had been proved earlier, for the 2008
synthetic aperture radar (SAR). Sichuan (China) earthquake, but with no practical use, due to
the limited coverage of the area [6].
I. I NTRODUCTION Further SAR data came in the following days, from COSMO-
SkyMed, ENVISAT, and ALOS, allowing an unprecedented
50 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 8, NO. 1, JANUARY 2011
satellites are equipped with a flexible multimode X-band SAR conditions; thus, the higher coherence of April-12 is justified
(SAR2000), with right- and left-looking imaging capabilities, by the shorter spatial baseline (41 m in April-12 and 237 m in
incidence angle range of 20◦ –60◦ , and 12-h minimum revisit April-15). We note, as a first comparison, that the similarity
time (at full constellation). Due to the high spatial resolution between the bean-shaped concentric fringe pattern from the
(∼3 m in the stripmap mode), the critical baseline, which is ascending and descending interferograms suggests that the
inversely related with the range resolution, is longer than that vertical component of the detected movement is prevailing, in
for ERS and Envisat systems. Indeed, at 40◦ look angle, it is agreement with the normal faulting mechanism.
∼2600 m. More precisely, the critical baseline is the minimum The spatial and temporal decorrelations have a relevant im-
value for which the spatial correlation is zero [9], and it occurs pact on the detected LOS displacement field. Indeed, in the
when the backscattered signal between two interferometric April-15 interferogram, one or more fringes are missing being
passes becomes completely uncorrelated for each pixel. too noisy. In this interferogram, we count nine fringes, for a
On April 6, following the Istituto Nazionale di Geofisica maximum displacement of ∼23 cm, while ten fringes (28 cm)
e Vulcanologia alert for a large earthquake in Abruzzi, the can be outlined in the one from April 12.
Italian Civil Protection Department requested ASI to do image
acquisitions in crisis mode, using stripmap (40-km swath and C. L-Band Interferometry: ALOS
3-m ground resolution) and spotlight (10-km swath and 1-m
resolution) modes. In the next 20 days, more than 100 images The L-band ALOS PALSAR instrument, due to its longer
were acquired over the area, with many different geometries wavelength (23.6 cm), has the advantage of a deeper penetration
and look angles. of vegetated areas that contributes to reduce the temporal
We applied DInSAR to the April 4 and April 12 stripmap decorrelation. Moreover, its higher spatial resolution leads to
pair. The interferogram spatial resolution was degraded to 20 m an increase of the critical spatial baseline. For example, for a
since an 8 × 8 multilook processing was applied, to enhance the fine beam single-polarization (FBS) data pair, with a spatial
signal-to-noise ratio. The topographic-phase component was resolution of about 9 m/pixel and an average sight angle of
removed using a 5-m-geometric-resolution digital terrain model 38◦ , the critical spatial baseline exceeds 13 000 m. This allows
(DTM), with a vertical accuracy of 1 m. The DTM had been increasing the number of suitable interferometric pairs for
obtained by an airborne interferometry SAR campaign [10]. DInSAR applications. The capabilities of ALOS PALSAR have
The estimated spatial baseline is quite long (about 435 m) but already been proved applying DInSAR in a very impervious
still able to maintain an overall good coherence. The resulting region such as the 2008 Sichuan (China) earthquake epicentral
interferogram shows a northwest–southeast concentric fringe area [11], [12].
pattern that is composed of 12 fringes, with each corresponding We have analyzed the coseismic surface deformation in the
to a range increment of 1.5 cm going from the outer to the L’Aquila area using a multipolarization interferometric pair:
inner fringes. The maximum ground displacement measured the preseismic image (July 20, 2008) is a fine beam double-
along the satellite LOS (i.e. ∼ 36◦ from nadir) reaches 18 cm polarization (FBD) one, while the postseismic one is an FBS
at midway from L’Aquila and Fossa villages [Fig. 1(a)]. The image (April 22, 2009). Both images have been acquired
surface movement is, as expected, from a southwest dipping from ascending orbits, with a perpendicular spatial baseline of
normal faulting mechanism. 182 m. The temporal baseline is about nine months. Due to
the different acquisition modes, the data have different spatial
resolutions; to achieve the same spatial resolution of the FBS
B. C-Band Interferometry: Envisat image, the FBD one has been oversampled. It is worth noting
that the critical baseline for the FBD–FBS interferogram is
The Envisat Advanced SAR sensor is a C-band polarimetric equal to 2/3 of an FBS–FBS one.
system operational since 2002. We exploited a data set of The resulting FBD–FBS interferogram is shown in Fig. 1(d).
images in ERS-like mode provided by the European Space Two clear fringes, plus a fraction of a third one, come out and
Agency (ESA) and acquired along ascending and descending correspond to 24–26 cm of LOS displacement, in agreement
paths. The first postseismic Envisat image (dated April 12, with the C-band interferograms.
2009) was acquired from a descending orbit and matched with
a preseismic image acquired on April 27, 2008, providing a
III. DInSAR DATA C OMPARISON AND R ESULTS
one-year coseismic interferogram, hereafter named April-12
[Fig. 1(b)]. Three days later, on April 15, an Envisat ascending In the previous section, we have described four interfero-
image was acquired and processed with the image acquired on grams: three from the ascending path (each from a different
March 11, 2009, to obtain a further interferogram, hereafter band) and one from the descending orbit. In Fig. 2, we summa-
named April-15 [Fig. 1(c)]. rize their temporal baselines, while in Table I, we report their
The high-resolution DTM mentioned before was also used to characteristics.
remove the topographic phase from the Envisat interferograms, To measure the actual surface deformation, we applied to
and a multilooking operation (2 × 10 range and azimuth pixels, the four interferograms a phase unwrapping procedure based
respectively) was adopted, leading to a 40 × 40 m degraded on the region growing algorithm [13]. Such an algorithm
spatial resolution. handles noisy interferograms without provoking any significant
Notwithstanding the longer temporal baseline, the descend- distortion to the outcoming relative displacements. In order to
ing interferogram maintains a better coherence than the 35-day compare the results, we geocoded the four displacement maps
ascending one. Indeed, both interferograms use a pair of images in a common geographical system (UTM–WGS-84) using the
acquired in the same season, with quite similar backscattering 5-m DTM, with a resulting geometric resolution of 80 m.
Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Marco Chini. Downloaded on July 01,2010 at 16:21:42 UTC from IEEE Xplore. Restrictions apply.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.
Fig. 1. Four differential wrapped interferograms: (a) COSMO-SkyMed; (b) Envisat descending; (c) Envisat ascending; (d) ALOS. The interferograms have been
processed with the SARscape software. The study area is the red square in the top right inset.
52 IEEE GEOSCIENCE AND REMOTE SENSING LETTERS, VOL. 8, NO. 1, JANUARY 2011
TABLE II
DATA C OMPARISON : D ETAILS
Fig. 4. Map of the per-pixel difference of the displacement values. COSMO-SkyMed has been compared to each of the other data after a postprocessing, aiming
at projecting the data onto COSMO-SkyMed geometry.
of COSMO-SkyMed versus Envisat (either ascending or [2] EMERGEO working group, “Evidence for surface rupture associated with
descending pairs) is close to zero (less than 4 mm) although it the Mw 6.3 L’Aquila earthquake sequence of April 2009 (central Italy),”
Terra Nova, vol. 22, pp. 43–51, 2010.
appears more scattered (≈ 34 < σ <≈ 45 mm). The difference [3] S. Pondrelli, S. Salimbeni, A. Morelli, G. Ekström, M. Olivieri, and
of COSMO-SkyMed versus Envisat descending should theo- E. Boschi, “Seismic moment tensors of the April 2009, L’Aquila (Central
retically be less affected by error noise for the short spatial Italy), earthquake sequence,” Geophys. J. Int., vol. 180, no. 1, pp. 238–
242, Jan. 2010.
baseline (41 m), but it is also the only cross-path (ascending [4] S. Atzori, I. Hunstad, M. Chini, S. Salvi, C. Tolomei, C. Bignami,
versus descending) comparison. S. Stramondo, E. Trasatti, A. Antonioli, and E. Boschi, “Finite fault
inversion of DInSAR coseismic displacement of the 2009 L’Aquila earth-
quake (Central Italy),” Geophys. Res. Lett., vol. 36, no. 15, p. L15 305,
IV. C ONCLUSION Aug. 2009.
[5] TRE. [Online]. Available: http://www.treuropa.com/HomeTRE/News
We have investigated the April 6, 2009, earthquake by a Events/News/tabid/193/newsid969/100/Default.aspx
three-band interferometric data set. We focused our study on [6] S. Stramondo, M. Chini, S. Salvi, C. Bignami, S. Zoffoli, and
the analysis of the capabilities and the comparison of X-, C-, E. Boschi, “Ground deformation imagery of the May Sichuan earth-
and L-band DInSAR for the retrieval of surface movements. quake,” EOS Trans. Amer. Geophys. Union, vol. 89, no. 37, pp. 341–342,
Sep. 9, 2008.
It comes out that, whatever the characteristics of the SAR [7] L. Pulvirenti, N. Pierdicca, and M. Chini, “Analysis of COSMO-SkyMed
sensor used, such as wavelength, orbit path, incidence angle, observations of the 2008 flood in Myanmar,” Italian J. Remote Sens.,
and geometric resolution, the resulting deformation maps from vol. 42, no. 1, pp. 79–90, 2010.
[8] F. Covello, F. Battazza, A. Coletta, E. Lopinto, C. Fiorentino,
a normal fault earthquake are comparable and reliable with an L. Pietranera, G. Valentini, and S. Zoffoli, “COSMO-SkyMed an existing
overall accuracy better than 1.4 cm, which corresponds to half opportunity for observing the Earth,” J. Geodynamics, vol. 49, no. 3/4,
of the COSMO-SkyMed wavelength. pp. 171–180, Apr. 2010.
[9] H. A. Zebker and J. Villasenor, “Decorrelation in interferometric radar
It has also been demonstrated how the used sensors, which echoes,” IEEE Trans. Geosci. Remote Sens., vol. 30, no. 5, pp. 950–959,
are fully representative of the suitable SAR frequencies and Sep. 1992.
spatial resolutions from satellite missions, have been able to [10] Intermap. [Online]. Available: http://www.intermap.com/ifsar
provide consistent measurements, from different sight angle [11] M. Chini, S. Atzori, E. Trasatti, C. Bignami, C. Kyriakopoulos,
C. Tolomei, and S. Stramondo, “The May 12, 2008, (Mw 7.9) Sichuan
and orbital paths, of the ground deformation pattern due to the earthquake (China): Multiframe ALOS-PALSAR DInSAR analysis of
L’Aquila earthquake. We stress that our results are valid for coseismic deformation,” IEEE Geosci. Remote Sens. Lett., vol. 7, no. 2,
ground deformations with a predominant vertical component. pp. 266–270, Apr. 2010.
[12] M. Hashimoto, M. Enomoto, and Y. Fukushima, “Coseismic de-
From an applicative perspective, an effective improvement formation from the 2008 Wenchuan, China, earthquake derived
of a DInSAR observation strategy for coseismic displacements from ALOS/PALSAR images,” Tectonophysics, 2009, TECTO-124726.
during seismic sequences would be the opportunity to narrow DOI: 10.1016/j.tecto.2009.08.034.
[13] A. Reigber and J. Moreira, “Phase unwrapping by fusion of local and
the temporal sampling of the deformation, using all available global methods,” in Proc. IEEE Geosci. Remote Sens. Symp., 1997, vol. 2,
SAR data independently on the wavelengths. Indeed, this is pp. 869–871.
particularly important during long seismic sequences, to in- [14] J. Dawson and P. Tregoning, “Uncertainty analysis of earthquake source
vestigate the incremental ground deformations caused by large parameters determined from InSAR: A simulation study,” J. Geophys.
Res., vol. 112, no. B9, p. B09 406, Sep. 2007.
magnitude aftershocks. [15] S. Quegan and J. Lamont, “Ionospheric and tropospheric effects on syn-
thetic aperture radar performance,” Int. J. Remote Sens., vol. 7, no. 4,
ACKNOWLEDGMENT pp. 525–539, Apr. 1986.
[16] D. Massonnet and K. L. Feigl, “Radar interferometry and its application
ASI provided the COSMO-SkyMed data. ESA made avail- to changes in the Earth’s surface,” Rev. Geophys., vol. 36, no. 4, pp. 441–
500, 1998.
able the Envisat and ALOS data (the latter copyright of the [17] S. Atzori, M. Manunta, G. Fornaro, A. Ganas, and S. Salvi, “Postseismic
Japanese Space Agency—JAXA). displacement of the 1999 Athens earthquake retrieved by the differential
interferometry by synthetic aperture radar time series,” J. Geophys. Res.,
vol. 113, no. B9, p. B09 309, Sep. 2008.
R EFERENCES [18] D. T. Sandwell, D. Myer, R. Mellors, M. Shimada, B. Brooks, and
[1] A. Akinci, F. Galadini, D. Pantosti, M. D. Petersen, L. Malagnini, and J. Foster, “Accuracy and resolution of ALOS interferometry: Vector
D. Perkins, “Effect of time dependence on probabilistic seismic-hazard deformation maps of the Father’s Day intrusion at Kilauea,” IEEE
maps and deaggregation from the Central Apennines, Italy,” Bull. Seismol. Trans. Geosci. Remote Sens., vol. 46, no. 11, pp. 3524–3534,
Soc. Amer., vol. 99, no. 2, pp. 585–610, Apr. 2009. Nov. 2008.
Copyright (c) 2010 IEEE. Personal use is permitted. For any other purposes, Permission must be obtained from the IEEE by emailing pubs-permissions@ieee.org.
Authorized licensed use limited to: Marco Chini. Downloaded on July 01,2010 at 16:21:42 UTC from IEEE Xplore. Restrictions apply.