当前位置：
X-MOL 学术
›
Seismol. Res. Lett.
›
论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Coseismic Displacements and Surface Fractures from Sentinel-1 InSAR: 2019 Ridgecrest Earthquakes
Seismological Research Letters ( IF 3.3 ) Pub Date : 2020-01-15 , DOI: 10.1785/0220190275 Xiaohua Xu 1 , David T. Sandwell 1 , Bridget Smith-Konter 2
Seismological Research Letters ( IF 3.3 ) Pub Date : 2020-01-15 , DOI: 10.1785/0220190275 Xiaohua Xu 1 , David T. Sandwell 1 , Bridget Smith-Konter 2
Affiliation
Cite this article as Xu, X., D. T. Sandwell, and B. Smith-Konter (2020). Coseismic Displacements and Surface Fractures from Sentinel-1 InSAR: 2019 Ridgecrest Earthquakes, Seismol. Res. Lett. XX, 1–7, doi: 10.1785/ 0220190275. Interferometric Synthetic Aperture Radar is an important tool for imaging surface deformation from large continental earthquakes. Here, we present maps of coseismic displacement and strain from the 2019 Ridgecrest earthquakes usingmultiple Sentinel-1 images. We provide three types of interferometric products. (1) Standard interferograms from two look directions provide an overview of the deformation and can be used for modeling coseismic slip. (2) Phase gradient maps from stacks of coseismic interferograms provide high-resolution (∼30 m) images of strain concentration and surface fracturing that can be used to guide field surveys. (3) High-pass filtered, stacked, unwrapped phase is decomposed into east–west and up–down, south–north components and is used to determine the sense of fault slip. The resulting phase gradientmaps reveal over 300 surface fractures, including triggered slip on the Garlock fault. The east– west component of high-pass filtered phase reveals the polarity of the strike-slip offset (right lateral or left lateral) for many of the fractures. We find a small number of fractures that have slip polarity that is retrograde to the background tectonic stress. This is similar to observations of retrograde slip observed near the 1999 Mw 7.1 Hector Mine rupture, but the Ridgecrest observations are more completely imaged by the frequent and high-quality acquisitions from the twin Sentinel-1 spacecrafts. Determining whether the retrograde features are triggered slip on existing faults, or compliant fault deformation in response to stress changes from the Ridgecrest earthquakes, or new Coulombstyle failures, will require additional field work, modeling, and analysis. Introduction TheMw 7.1 Ridgecrest earthquake struck on 5 July 2019 at 8.19 p.m. local time at the China Lake Naval Air Center, 17 km northeast of the city of Ridgecrest, California (U.S. Geological Survey [USGS], 2019a). Thirty-six hours prior, on 4 July 2019, an Mw 6.4 foreshock occurred (10.33 a.m. local time), 11 km southwest of Searles Valley (USGS, 2019b). The two earthquakes ruptured two conjugate faults in the Airport Lake fault zone and Little Lake fault zone, oriented roughly northwest– southeast (right-lateral strike slip) and northeast–southwest (left-lateral strike slip), respectively. Field scientists reported 2–3 m of right-lateral offset along the southern section of the Mw 7.1 rupture. Twin Sentinel-1 satellites operated by the European Space Agency (ESA) were continuously collecting measurements over this region since 2015 (Torres et al., 2012). These satellites collect wide swath data (250 km) using a burst acquisition mode called terrain observation by progressive scan (TOPS). The twin satellites achieve complete coverage in a short-time interval of six days that is well suited for this earthquake sequence. The new wide swath mode requires along-track alignment of better than 1/200 of a pixel (<7 cm), which is possible using the very accurate orbital information provided by ESA (Sansosti et al., 2006; Xu et al., 2017); earthquake displacements greater than ∼7 cm in the along-track direction will cause phase discontinuities at burst boundaries that should be ignored in the interpretation of the maps in the following sections. Moreover, the Sentinel-1 coverage is excellent for these two earthquakes, providing critical high resolution spatially dense deformation observations of the largest earthquake to strike the Eastern California Shear Zone (ECSZ) in nearly 20 yr (Fig. 1). In this article, we focus on mapping coseismic displacement and strain with the objective of serving these products to the field mapping and modeling communities 1. Institute of Geophysics and Planetary Physics, Scripps Institution of Oceanography, University of California San Diego, La Jolla, California, U.S.A.; 2. Department of Geology and Geophysics, University of Hawaii at Manoa, Honolulu, Hawaii, U.S.A. *Corresponding author: xix016@ucsd.edu © Seismological Society of America Volume XX • Number XX • – 2020 • www.srl-online.org Seismological Research Letters 1 Downloaded from https://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220190275/4921308/srl-2019275.1.pdf by UC San Diego Library user on 21 April 2020 Figure 1. (a) Overview map of the topography, faults, and Sentinel-1 frames surrounding the 2019 Ridgecrest earthquake sequence. Red and blue stars denote the epicenter of theMw 7.1 and 6.4 earthquakes, respectively. Black curves are faults mapped by U.S. Geological Survey (USGS). Red box indicates geographic location of the wrapped interferogram maps provided in (b) and (c). (b) Interferogram from the descending track 71 Sentinel-1 Interferometric Synthetic Aperture Radar (InSAR) data. Each fringe represents 2.8 cm of ground displacement away from the satellite. (c) Interferogram from the ascending track 64 Sentinel-1 InSAR data. The color version of this figure is available only in the electronic edition. 2 Seismological Research Letters www.srl-online.org • Volume XX • Number XX • – 2020 Downloaded from https://pubs.geoscienceworld.org/ssa/srl/article-pdf/doi/10.1785/0220190275/4921308/srl-2019275.1.pdf by UC San Diego Library user on 21 April 2020 (see Data and Resources). In addition, we highlight the ability to map small spatial scale (∼30 m) fractures having small offsets (>5 mm) using Sentinel-1 data. Data and Methods Here, we construct coseismic interferograms using two preearthquake acquisitions for each track and all data acquired within a month after the earthquakes (Table 1). Unfortunately, there are no acquisitions between the 36 hr that separated the two events. We produce three types of products using opensource Generic Mapping Tools Synthetic Aperture Radar (GMTSAR) (Sandwell et al., 2016) and Generic Mapping Tools (Wessel et al., 2013) software, with the phase unwrapped using the Statistical-cost, Network-flow Algorithm for Phase Unwrapping software (Chen and Zebker, 2002): 1. The standard interferograms shown in Figure 1 were produced using the nearest acquisitions spanning the earthquakes (Table 1). These were Gaussian filtered at 100 m half-wavelength and sampled at 50 m. Unwrapped and subsampled data, suitable for source modeling, are available on our website. Interestingly, the overall interferometric pattern from this sequence resembles the European Remote Sensing satellites interferogram for the 1999 Mw 7.1 Hector Mine earthquake (Sandwell et al., 2000; Fialko et al., 2002). Both events occurred in similar tectonic context in the ECSZ (Savage et al., 1990), yielding similar moment release and rupturing behavior along the E45S direction. 2. To extract information for smaller scale features, we produce phase-gradient maps directly from the real R x and imaginary I x parts of the full resolution interferograms (Sandwell and Price, 1998), in which the position vector x consists of the range r and azimuth a coordinates of the interferogram. Instead of computing the phase gradient from the phase φ x tan−1 I R , which as many 2π discontinuities, one can use the chain rule of differentiation to develop a formula for the phase gradient directly from R and I. The result is ∇φ x R∇I−I∇R R2 I2 , in which the gradient operator is ∇ ∂ ∂r ; ∂ ∂a , with r and a denoting the direction of gradient along range (look) and azimuth (flight). The numerical derivative filter must be designed to avoid aliasing short-wavelength noise at the Nyquist wavenumber to longer wavelengths, so we combined a central difference filter with a low-pass Gaussain filter having 0.5 gain at 30 m half-wavelength. Phase gradients are very small in the far field of the rupture, so we focus on the second subswath of each TOPS frame and process at full resolution (∼15 m). Unlike standard interferograms, these phase gradient maps can be directly stacked without phase unwrapping (Sandwell and Price, 1998). Thus, we applied the same algorithms to every interferogram (Table 1) and averaged them to produce the final phase gradient maps (Fig. 2). The phase gradient maps are essentially strain maps and thus highlight all types of small spatial scale deformation. There are two types of artifacts to consider when interpreting these maps. First, there are artificial linear phase discontinuities at the burst boundaries of the TOPSmode data. To overcome this, one can estimate the associated azimuthal motion by computing an earthquake source model and include this estimate at the Synthetic Aperture Radar (SAR) coregistration step. Second, the random patterns along the major rupture zones are areas of decorrelation due to extreme ground shaking or deformation rates beyond one radian per pixel. 3. To further define the deformation characteristics of each fracture, we unwrapped the full resolution interferogram following Xu et al. (2016), by imposing a coherence mask along the fault and allowing discontinuity in the map. We stacked the unwrapped phase and then highpass filtered using an 800 m Gaussian filter (Fig. 3). The stacking of unwrapped phase reduces the phase noise to ∼1 mm and also reduces atmospheric effect, especially the elevation-dependent component that possess resemblance to deformation pattern. These stacked phase maps are converted to line of sight (LoS) deformation and then decomposed into east–west motion (positive TABLE 1 Interferometric Pairs versus Perpendicular Baseline Direction Dates (yyyy/mm/dd) B⊥ (m) Descending average look vector: [0.633, −0.112, 0.765] 2019/06/22–2019/07/16 87.79 2019/06/22–2019/07/28 38.09 2019/07/04–2019/07/16 (Fig. 1b) 29.68 2019/07/04–2019/07/28 31.15 Ascending average look vector: [−0.636, −0.112, 0.763] 2019/06/28–2019/07/10 63.38 2019/06/28–2019/07/16 35.98 2019/06/28–2019/07/22 12.37 2019/06/28
更新日期：2020-01-15
京公网安备 11010802027423号