TH1.L09 - ADVANCES IN THE GENERATION OF DEFORMATION TIME SERIES FROM SAR DATA SEQUENCES IN AREAS AFFECTED BY LARGE DYNAMICS

Advances in the generation of deformation time series from SAR data sequences in areas affected by large dynamics

IREA-CNR Napoli, Italy

[email protected]

F. Casu, A. Manconi, A. Pepe, M. Manzo, R. Lanari

IGARSS 2010 – [email protected] 2010 – [email protected]

SummarySummary

This approach has been tested on an ENVISAT ASAR data archive (Track 61, Frames 7173-7191) related to the Galàpagos Islands, focusing on Sierra Negra caldera (Galapagos Islands).

Results have been finally compared to continuous GPS measurements.

In this work, we applied the SBAS strategy to the sequence of the range and azimuth Pixel-Offset estimates, achieved from the same set of small baseline data pairs.

Therefore, to exploit conventional InSAR time series algorithms (such as the PSI or the Small BAseline Subset ones), which are phase based, would become extremely difficult.

In areas characterized by large and rapid deformation dynamics, deformation can not be easily retrieved by exploiting conventional InSAR, due to high fringe rates (possible PhU problems) and misregistration.

A way to overcome this limitation is to exploit the amplitude information.


Key idea of the SBAS algorithm Key idea of the SBAS algorithm (Berardino et al. 2002)(Berardino et al. 2002)

For each coherent pixel, the time series deformation is computed by searching for an LS solution with a minimum norm constraint (the SVD method is applied).


Why small baseline data pairs?Why small baseline data pairs?

Small baseline InSAR interferograms are less affected by noise effects (decorrelation) and are easier to process (registration / unwrapping steps).

ENVISAT Amplitude: Galàpagos

Baseline: 1000mBaseline: 40m


Further constraints: deformation amountFurther constraints: deformation amount

Conventional InSAR Possible PhU problems Co-Registration problems

Eruptive eventEruptive event


Conventional SBAS-InSAR results: GalàpagosConventional SBAS-InSAR results: Galàpagos

26 ASAR Envisat scenes, T61In collaboration with F. Amelung

Azi

mu

th

?

Sierra Negra

Fernandina

IsabelaOctober, 21st

2005

<-3

>3cm

/yr


Pixel-Offset SBAS (PO-SBAS): key ideaPixel-Offset SBAS (PO-SBAS): key idea

We exploit the AMPLITUDE information in order to retrieve deformation in areas characterized by large dynamics.

In particular, we apply the well known SBAS approach to the Pixel-Offsets retrieved via cross-correlation of the amplitude (AMPCOR routine) relevant to small baseline SAR image pairs.

Note that, we may directly use the already registered SAR image pairs computed to generate the SBAS-InSAR time series.


Why (again) small baseline data pairs?Why (again) small baseline data pairs?

Perp. Bas. 40m Perp. Bas. 1000m


Pixel-Offset SBAS (PO-SBAS): “good pixels” vs. baselinePixel-Offset SBAS (PO-SBAS): “good pixels” vs. baseline

Small Baseline Constraint

Quadratic relationship


PO-SBAS results: GalàpagosPO-SBAS results: Galàpagos

RangeAzimuth (mostly North)

Conventional SBAS-InSAR maskPixel-Offset SBAS mask

<-3

>3cm

/yr


Validation of RANGE displacements: SAR vs GPS Validation of RANGE displacements: SAR vs GPS time seriestime series

GPS data from Chadwick, et al. 2006, Geology

GV01GV03

GV05GV02

GV06

GV04

SAR- GPS


Validation of AZIMUTH displacements: SAR vs GPS Validation of AZIMUTH displacements: SAR vs GPS time seriestime series

GPS data from Chadwick, et al. 2006, Geology

GV03

GV02GV04

SAR- GPS


GV03

GV04

GV06

Validation of displacements: SAR vs GPS Validation of displacements: SAR vs GPS baseline changebaseline change

GPS Estimates,from Geist, et al. 2008, Bull. of Volc.


Summary Summary accuracy ca. accuracy ca. 1/201/20 of pixel of pixel


Conclusions and further developmentsConclusions and further developmentsWe present the results achieved by exploiting the amplitude information in order to retrieve deformation in areas characterized by large dynamics (eg. Volcanic and Seismogenic areas).

We show that the well known SBAS approach can be successfully applied also to the Pixel-Offsets retrieved via cross-correlation of the amplitude relevant to small baseline SAR image pairs .

We obtain deformation time series relevant to displacements in both range and azimuth direction in areas where no interferometric phase could be exploited; the achieved accuracy is of about 1/20 of pixel.

The exploitation of higher resolution SAR images (TERRASAR-X, COSMO-Skymed) is envisaged.

The retrieved offset information can be further used in order to properly register the analyzed data pairs, thus “restoring” the possibility of generating interferogram sequences, following Sang-Ho Yun et al., GRL, 2007.


Pixel-Offsets: X-Band case studyPixel-Offsets: X-Band case study

12-Jan-2010 Haiti earthquake, Mw 7.012-Jan-2010 Haiti earthquake, Mw 7.0

COSMO-SkyMED, X-band, COSMO-SkyMED, X-band, wavelength wavelength ~ 3.1cm~ 3.1cm12-Dec-2009 / 21-Jan-2010 Spotlight SAR images12-Dec-2009 / 21-Jan-2010 Spotlight SAR images1 meter ground resolution1 meter ground resolution

Am

plit

ud

e Im

ag

e

Azi

muth

Range


low coherence

1.56

0cm

Ph

ase


Azi

muth

Range


Interferometric PhaseInterferometric Phase


low coherence

>50

<-50

cm

Ran

ge O

ffse

ts


Azi

muth

Range


Range Amplitude Pixel-Offset MapRange Amplitude Pixel-Offset Map


Documents

TH1.L09 - ADVANCES IN THE GENERATION OF DEFORMATION TIME SERIES FROM SAR DATA SEQUENCES IN AREAS AFFECTED BY LARGE DYNAMICS