Coseismic Deformation From the Haiti Earthquake of January 13, 2010, Detected by ALOS/PALSAR@(Ver.021710)
New: Result of ScanSAR – ScanSAR interferometry was added.
[Caution: This page will be updated without notification.]
Manabu Hashimoto (DPRI,
On January 13, 2010, a Mw7.0 earthquake hit southern Haiti and caused severe damages in and around Port au Prince. This region has not suffered from strong earthquakes since 18th century, but rather high seismic risk is inferred from GPS observation (Manaker et al., 2008). This earthquake is considered to have occurred on the Enriquillo fault. It is essential to reveal coseismic deformation from this earthquake and estimate fault parameters for the understanding of earthquake generation process and future seismic hazard evaluation. Therefore we utilize synthetic aperture radar images.
PALSAR (Phased Array type L-band SAR) is an L-band synthetic aperture radar onboard the Japanese satellite named Advanced Land Observation Satellite (hereafter ALOS). ALOS made an urgent observation of PALSAR on January 16 from the Path 138 with the off-nadir angle of 34.3 degree. Performing interferometry with the image acquired on October 11, 2007, using GAMMA® software, we obtained an interferogram with CGIAR hole-filled SRTM DEM (Jarvis et al., 2008). Perpendicular baseline is 267m. Precise state vector was available on Jan. 20, 2010. Figure 1 shows the result using the precise state vector without flattening. The eastern edge of the obtained image is about 30km west of the epicenter. This area corresponds to an about 30km wide peninsula which extends westward from Port au Prince. However, we found at least three cycles of fringes (~40cm), which show increase of range toward the center of fringes, on the south side of the Enriquillo fault. We also recognized three cycles of fringes on the north side of the fault, but their gradient is much steeper than the south side.
Figure 1. Coseismic deformation detected by ALOS/PALSAR. Master and slave images of Path 138 (western path) were acquired on Feb. 28, 2009 and Jan. 16, 2010, respectively, from an ascending orbit. Images of Path 136 were acquired on Sept. 12, 2009 and Jan. 28, 2010. The satellite was flying from south to north emitting microwave eastward. Incidence angle of microwave on the surface is approximately 39 deg. 1 cycle of color corresponds to 11.8cm of range (distance between satellite and earthfs surface) changes. The color change in the order of blue-magenta-yellow-blue means range increase. The red star shows the epicenter by USGS(2010). CMT solution is given by the Global CMT Project(2010).
On Jan. 25, 2010, another acquisition of PLASAR image was made from a descending orbit. Figure 2 is an interferogram of the pair of images acquired on Mar., 9, 2009, and Jan., 25, 2010 (Bperp~798m). We used provisional orbit information of the image in 2010 and performed flattening since large orbital error contaminates the interferogram. Concentrated fringes can be recognized northwest of the epicenter. We can count 6 fringes with color changing in the order of yellow, magenta and blue, implying more than 70cm range decrease. Another circular fringe appears just north of the epicenter. On the other hand, 3 cycles of fringes with the opposite sense appear on the northern coast. Broad fringes can be seen on the south side of the epicenter, which corresponds to that observed in the interferogram in Figure 1. These observation impies rather complicated rupture process during the main shock than the results of body wave inversion studies.
Figure 2. Coseismic deformation derived from the pair of PALSAR images acquired on Mar. 9, 2009, and Jan. 25, 2010, from a descending orbit. The satellite was flying from north to south emitting microwave westward. See also the legend of Figure 1.
I estimated the slip distribution and dip angle of the fault plane by inverting PALSAR interferograms with the method developed by Fukahata and Wright(2008). Their method can estimate dip angle and slip distribution simultaneously under the condition of minimum ABIC. The modeled fault plane is assumed to be 100km long and its depth range is 0-50km. Strike is 262deg from the north. The center of the fault plane is close to the epicenter and the fault plane is dipping northward. I searched dip angle within the range of 40 – 90 deg with step of 2 deg. I discarded data in the northern region >19N and from the island.
@Figure 3 is the estimated slip distribution. Optimal dip angle is estimated to be 52 deg, which is much smaller than 70 deg of CMT. There are three asperities, one of which is located near the hypocenter and has maximum of about 2.5m at the depth of 15km. Left lateral strike slip with slight thrust motion is recognized. Another asperity is about 13km west at a depth of 5km and has about 3m slip with significant thrust component. Last asperity is located further west at the depth of about 12km. Figures 4 and 5 show synthesized interferograms, which roughly explain the characteristics in the observed ones in Figures 1 and 2. Geodetic moment is estimated to be 1.03~1020Nm (Mw7.28) with a rigidity of 33GPa.
I will try to search more favorable model by changing strike and/or location with data acquired during the next couple of weeks.
Figure 3. Estimated slip distribution. Slip is projected onto the vertical plane oriented in the direction of N262E seen from the north side.
Figure 4. Synthesized interferogram for the ascending orbit. The rectangle is the surface projection of the modeled fault plane with the contours of slip distribution. The star indicates the epicenter by USGS(2010).
Figure 5. Synthesized interferogram for the descending orbit. See also the legend of Figure 2.
New: Analysis of ScanSAR – ScanSAR interferometry
On February 11, 2010, an acquisition of WB1 mode (ScanSAR) was made. Observation with WB1 mode was made four times before the mainshock for this path 448 and frame 3250. We have made an interferometric analysis of this image with the image acquired on September 26, 2009. Using the program provided by Prof. David Sandwell, SIO, UCSD, we divided original Level1.0 image into 5 images according to off-nadir angle/PRF. After this division, we performed usual interferometric analysis with Gamma. Perpendicular baselines for 5 swaths are estimated to be 259.8, 236.6, 216.7, 200.5, and 190.1m from 1 to 5, respectively, which are the smallest among the analyses we have made. Figure 6 is flattened interferogram. Original interferogram contains large orbital fringes. Coherence is very high due to small Bperp, and continuity is good between swaths, except the westernmost swath5. We observe coseismic fringes similar to those in Figure 2, but they are much clearer. Broad fringe north of Port au Prince might be coseismic one, judging from its shape. On the both ends of this interferogram, short-wavelength regular disturbances appear. This type of fringes sometimes appears in ScanSAR-ScanSAR interferograms. So far we donft know the reason but suspect the problem of coregistraion of images or matching of bursts.
PALSAR level1.0 data were provided by the Japan Aerospace Exploration Agency via the Geographical Survey Institute under the project of the Evaluation of Utilization of Land Observation Satellite for Disaster Mitigation, Earthquake Working Group. The ownership of PALSAR products belong to JAXA and Ministry of Economy, Trade and Industry.
We thank Professor David Sandwell,
Scripps Institute of Oceanography,
Fukahata, Y., and T.J. Wright, 2008, A non-linear geodetic data inversion using ABIC for slip distribution on a fault with an unknown dip angle, Geophys. J. Int., 173, 353-364, doi:10.1111/j.1365-246X.2007.03713.x
Global CMT Project, 2010, http://neic.usgs.gov/neis/bulletin/neic_rja6_hrv.html
Jarvis, A., et al., 2008, Hole-filled SRTM data V4, International Centre for Tropical Agriculture (CIAT),
Manaker, D.M., et al.,
2008, Interseismic plate coupling and strain
partitioning in the northeastern