The 1999 (Mw 7.1) Hector Mine, California Earthquake: Near-Field Postseismic Deformation from ERS Interferometry

Interferometric synthetic aperture radar (InSAR) data over the area of the Hector Mine earthquake (Mw7.1, 16 October 1999) reveal postseismic deformation of several centimeters over a spatial scale of 0.5 to 50 km. We analyzed seven SAR acquisitions to form interferograms over four time periods after the event. The main deformation seen in the line-of-sight (LOS) displacement maps are a region of subsidence (60 mm LOS increase) on the northern end of the fault, a region of uplift (45 mm LOS decrease) located to the northeast of the primary fault bend, and a linear trough running along the main rupture having a depth of up to 15 mm and a width about 2 km. We correlate these features with a double left-bending, right-lateral, strike-slip fault that exhibits contraction on the restraining side, and extension along the releasing side of the fault bends. The temporal variations in the near-fault postseismic deformation are consistent with a characteristic time scale of 135 +42/-25 days, which is similar to the relaxation times following the 1992 Landers earthquake. High gradients in the LOS displacements occur on the fault trace, consistent with afterslip on the earthquake rupture. We derive an afterslip model by inverting the LOS data from both the ascending and descending orbits. Our model indicates that much of the afterslip occurs at depths of less than 3-4 km. Introduction The Hector Mine earthquake (Mw7.1, 16 October 1999) ruptured the surface along the Lavic Lake fault, the Bullion fault, and sections of several other faults within the eastern California shear zone (ECSZ) (Fig. 1) [Dibblee, 1966; USGS et al., 2000; Treiman et al., this issue]. The ECSZ contains a series of northwest-trending faults that collectively take up a substantial amount of the motion between the Pacific and North American plates [Dokka and Travis, 1990]. InSAR observations in the area of the nearby 1992 Landers earthquake reveal transient deformation localized near the earthquake rupture, highlighting the importance of monitoring the near-field postseismic deformation [Massonnet et al., 1993; 1994; Zebker et al., 1994; Peltzer et al., 1996]. To obtain adequate InSAR coverage of the crustal deformation following the Hector Mine earthquake, we asked the European Space Agency to continue acquiring SAR data in the region. To date, four high quality postseismic interferograms can be constructed and we expect more will become available as additional ERS-2 SAR data are collected in 2002. Our analysis of the Hector Mine earthquake focused on the near-field (< 50 km from the rupture) postseismic deformation using SAR data collected by the European Space Agency satellite, ERS-2. While a number of GPS measurements collected after the earthquake may provide constraints on the larger-scale deformation (> 50 km) [Agnew et al., 2002], they lack the spatial resolution to characterize the details of postseismic deformation within 10 km of the main rupture. The large-scale postseismic deformation has also been analyzed using two of the four interferograms presented here [Pollitz et al, 2001]. These larger-scale studies have been interpreted in terms of viscoelastic deformation of the lower crust/upper mantle. However, we believe this interpretation is premature because the observed LOS displacements are small (on the order of a few cm), and not correlated between different interferograms, suggesting that they might be atmospheric artifacts. Here we examine postseismic deformation over four time periods: 4-249 days, 39-144 days, 74-319 days, and 39-354 days after the earthquake (Fig. 2). The four interferograms, derived from seven independent SAR acquisitions, all show consistent patterns. This enables us to perform a quantitative analysis of the postseismic decay time. Since there were only 2 large aftershocks (> Mw5.0) in this combined time period and since they occurred well north of the main rupture, the deformations we have observed are aseismic. InSAR Processing We employ "repeat-pass" interferometry [Massonnet and Feigl, 1998; Zebker at al., 1994; Rosen et al., 2000; Price, 1999] to create LOS displacement maps of an area about 110 km by 110 km surrounding the rupture (Fig. 1). The X-band receiving station at Scripps Institution of Oceanography was initially used to acquire these data so they could be processed in near realtime, although later the data were obtained from ESA through its distributors, SpotImage and Eurimage. Figure 2 gives the perpendicular baselines of the relevant data from the descending track across the Hector Mine earthquake area (ERS track 127) using precise orbital information from Delft Technical University [Scharroo and Visser, 1998]. The topographic phase was removed during each interferogram formation to isolate the deformation phase. The USGS 90-m topography was used to help unwrap the phase of a long baseline (195 m) topographic pair. The residual phase of each interferogram was unwrapped [Goldstein et al., 1988] and transformed into LOS displacement. A best-fitting plane was removed from each displacement map to account for possible errors due to imprecise orbits. Such de-trending does not introduce any spurious signal to the LOS data on length scales less than ~50 km. We refer to these maps as postseismic pairs 1, 2, 3, and 4, as shown in the timeline (Fig. 2). The ERS satellites measure LOS displacement at an angle of about 23° from vertical such that the LOS displacements are more sensitive to vertical rather than horizontal motion. The steep look angle of the ERS satellites impedes precise estimates of pure strike-slip fault offsets [Massonnet and Feigl, 1998]. In addition, the faults in the eastern California shear zone lie at high angles to the look direction of the radar from the descending orbit, and nearly perpendicular to the look direction from the ascending orbit. Since radar data from the ascending orbits is quite limited, determination of vector offsets in postseismic deformation is a considerable challenge. Postseismic Deformation The LOS deformation maps for each of the four post-seismic intervals show long-wavelength deformation patterns that are highly variable and dependent on the plane removed from each map. Although our postseismic interferogram 1 (full interferogram not presented here) shows general agreement with the corresponding interferogram published by Pollitz et al [2001], we do not yet believe that the large-scale pattern is related to crustal deformation since several other interferograms show different large-scale patterns. In contrast, at the smaller scale (< 50 km), each interferogram displays a prominent near-fault signal that is well resolved by the available data (Fig. 3). Conspicuous are a region of subsidence (60 mm LOS increase) on the northern end of the fault, a region of uplift (45 mm LOS decrease) located to the northeast of the primary fault bend, and a linear trough running along the main rupture having a depth of up to 15 mm and a width of about 2 km. Profiles A-A', B-B', and C-C' (Fig. 4) show the decay of the LOS displacement in the vicinity of the surface rupture. A discussion of each of the 4 interferograms is presented in the next section. Qualitatively, we interpret the near-fault postseismic deformation as being due to crustal relaxation following co-seismic displacements on a rightlateral, strike-slip fault with a double left bend (Fig. 5). The curvature of the fault produced a region of uplift near the restraining bend and a region of opening and subsidence near the releasing bend [Crowell, 1974; Davis and Reynolds, 1996]. Analysis of individual interferograms Postseismic 1 (4-249 days, Fig. 3). All three structural features are clearly visible on this map. A broad region of LOS increase lies distinctly on the west side of the fault with a displacement of about -60 mm (profile A-A', Fig. 4). This might represent either true ground subsidence and/or right-lateral afterslip. The area on the northeastern side of the rupture displays a zone of positive LOS displacements (consistent with uplift and/or right-lateral afterslip) up to 45 mm. This LOS "high" occurs about 2 km east of the Lavic Lake fault and appears to decrease gradually toward the west across the fault (profiles B-B' in Fig 4). On the east side of this high is a sharp N-S trending discontinuity suggesting shallow slip on a N-S trending fault. The high also coincides with the approximately 1 m of east-side-up coseismic uplift [Sandwell et al., this issue]. A linear trough runs along the entire rupture zone although it is most prominent along the releasing bend of the fault (profile C-C', Fig. 4). We believe this feature was formed by faultzone collapse with the greatest amplitude of subsidence in the area of extension at the releasing bend. Postseismic 2 (39-144 days, not shown). This second LOS displacement map shows the same features as postseismic pair 1, but with lower amplitude due to its relatively short time interval (105 days). The interval time is less than 1/2 of the time interval of the three other postseismic pairs, i.e. interferograms 1(245 days), 3 (245 days), and 4 (315 days). The total peak-to-trough displacement seen on this map is about 50 mm, whereas postseismic map 1 shows nearly 105 mm. The greatest change between the postseismic pairs 1 and 2 are the disappearance of the near-field trough signature in postseismic pair 2. Postseismic 3 (74-319 days, Fig. 3). This interferogram starts 74 days after the earthquake, and displays decreased amplitudes in all three structural features. Compared to the other three maps, this displacement map is considerably noisier, perhaps due to atmospheric effects. The broad high/low signatures still exist in the same locations and display a peak-to-trough range of about 60 mm. The trough along the rupture is apparent in this map (~10 mm), suggesting that the relaxation time is more than 74 days. Postseismic 4 (39-354 days, Fig. 3). This displacement map has the largest time span. The LOS displacement pattern is very similar to that seen in postseismic pair 1, although the amplitudes are smaller in postseismic pair 4. This suggests that while a large portion of postseismic deformation occurred during the 40 days after the earthquake, the process responsible for the near-fault postseismic deformation was still active beyond 1-2 months following the earthquake. Near-Field Postseismic Relaxation Time All four LOS displacement maps show the characteristic high/low pattern (Fig. 3). Assuming that this pattern is associated with a single exponential relaxation mechanism [e.g., Shen et al., 1994], we wish to determine the relaxation time, τ, that best explains the InSAR data. The LOS displacement, D, at position x and time t after the earthquake is given by where D∞ is the permanent postseismic deformation. The postseismic displacement map Ii measures the LOS component of deformation between the reference time, tref, and the repeat time, trep To eliminate dependence on D∞, we take the ratio of two displacement maps where j and k correspond to postseismic pairs 1 to 4. From the four pairs, we can form 6 independent LOS ratios R I I k j j k = . (4) D t D e t ( , ) x = −       ∞ − 1 τ , (1) I D e e t t ref rep = −       ∞ − − τ τ (2)