Applications: Earthquakes and Tectonics (3)

The ongoing development of constellations of Synthetic Aperture Radar (SAR) satellites with short repeat time acquisitions allows to explore the behavior of active faults with an unprecedented temporal resolution. The improvement from monthly to daily repeat times sheds a new light on the dynamics of near-surface fault creep along continental faults, which has been shown to exhibit various temporal behaviors, from persistent slow silent slip to discrete episodes of aseismic slip. Along the North Anatolian Fault (NAF), an 80 km-long section is creeping at least since the 1944, M7.3 earthquake near Ismetpasa, Turkey. Recent geodetic measurements suggest an average creep rate of about half the total slip rate accommodated by the NAF (8±3 mm/yr vs. 22±3 mm/yr). We take advantage of the dense set of SAR images acquired by the COSMO-SkyMed™ constellation over the creeping section of the NAF to quantify, with a high spatial and temporal resolution, the distribution of aseismic slip along strike and its evolution between August 2013 and August 2014. Over 7 tracks, 3 ascending and 4 descending, we compute 1000+ interferograms from 350+ radar acquisitions using the ISCE software (JPL). We use the Generic InSAR Analysis Toolbox (GIAnT) and the PyAPS library to correct interferograms from the propagation delays due to the stratification of the troposphere, predicted using the ERA-Interim (ECMWF) re-analysis. We use the New Small Baseline (NSBAS) method to derive the spatial and temporal evolution of the near-fault displacements independently for each track. Our results suggest the fault does not creep steadily over the 2013-2014 period but rather releases strain through discrete aseismic events we refer to as bursts of creep. In particular, we identify one major burst, equivalent to a M5.0 earthquake that took place in a maximum of 8 days releasing the equivalent of 1 year of creep at that location. This encouraging detection favors the maintenance and development of very short repeat time acquisitions along active fault zones.  

This contribution will discuss the growth and evolution of fold-thrust belts, focussing on two examples in east Iran. By combining a range of techniques (InSAR, seismology, remote sensing, geomorphology) it is possible to compare the motions during earthquakes with those in the following decades, and with the longer-term deformation preserved in the geomorphology and geological structures. It is therefore possible to examine how repeated seismic cycles result in the production of geological and topographic structures, and to investigate how such structures can be used to infer the characteristics of the earthquake cycles that produced them. Such work in two areas in east Iran (Tabas and Sefidabeh), which experienced large earthquakes in 1978 and 1994, has provided insights into the nature of the earthquake cycle and the development of the thrust belts in these locations. The fault slip during the earthquakes was mostly buried at depths of over 5 km. Using InSAR we have imaged aseismic creep on shallow fault zones in the decades following the earthquakes, which resulted in the formation of short-wavelength (i.e. < 5 km) topographic and geological structures, of the type that dominate the surface geology in these regions. The postseismic fault creep has occurred for over 30 years following the Tabas earthquake; an order of magnitude longer than is commonly observed. We have geodetically observed the formation of classic structural geometries, such as ramp-and-flat thrust faults and hangingwall anticlines, and constrained the properties and behaviours of the different components of these systems. Our results have implications for the geometry and evolution of thrust belts worldwide, and for the assessment of the earthquake hazard posed by such structures.

References:

- A. Copley and K. Reynolds, Imaging topographic growth by long-lived postseismic afterslip at Sefidabeh, east Iran, Tectonics, DOI: 10.1002/2013TC003462, 2014

- A. Copley, Postseismic afterslip 30 years after the 1978 Tabas-e-Golshan (Iran) earthquake: observations and implications for the geological evolution of thrust belts, Geophysical Journal International, doi: 10.1093/gji/ggu023, 2014

Shallow aseismic creep has been observed on many strike-slip faults around the world. However, the initiation process for shallow creep remains unclear, and no clear explanation exists for why creep persists for years to decades after some earthquakes while after others it decays completely within a few months.

Numerous observations now show that in many earthquakes the maximum slip occurs on the fault deeper in the crust (>8km), with the slip decaying towards the surface. This slip-deficit cannot be kinematically maintained over multiple earthquake cycles; in the long term the slip in the near surface must equate to that observed at depth. We hypothesise that fault creep can occur after such events to allow the shallow portions (<8 km) of the fault to 'catch up' with the slip at deeper sections.

We test this hypothesis with persistent scatterer InSAR analysis of 3 descending and 2 ascending Envisat tracks that together span a region covering the 1999 Izmit and Duzce ruptures in Turkey. These earthquakes were the latest in a series of large events (>magnitude 6.5) that have occurred on the North Anatolian Fault with a dominant westward progression in seismicity during the last century.

Our Envisat data covers an 8 year time window between 2003 – 2010. A section of the Izmit rupture was previously shown by Cakir et al, (2012) to be undergoing shallow creep. They also showed, using InSAR analysis of ERS data that the fault was fully locked with no evidence of creep before the 1999 earthquakes. However the spatial and temporal nature of this creep remains unclear. We use a small baseline processing strategy using the StaMPS software, which allows for checking of unwrapping errors by summing the residuals around closed interferometric loops.

We correct for tropospheric delays in the InSAR results using an ECMWF ERA-interim weather model (Bekaert et al, in review) and combine results with existing GPS measurements to produce a map of horizontal and vertical surface displacements over the 8-year period. From this we calculate the variation in fault creep rates along the earthquake surface ruptures. We show that shallow fault parallel creep occurs at an average rate of about 8 mm/yr reaching a maximum of 12±2 mm/yr around the city of Izmit.

In addition to this, there appears to be a region north-east of Lake Sapanca that is undergoing subsidence. This region appears to be limited to the Sakarya basin and is bounded by high topography to the west, north, and east and by the North Anatolian Fault to the south. We investigate whether this subsidence is tectonic in origin or a result of compaction due to groundwater extraction.

We use elastic dislocation models to determine the creep distribution with depth and compare with the coseismic slip inferred by Muller et al, (2006) in the Izmit earthquake. We find the fault creep rate is highest in the region of maximum shallow coseismic slip deficit.

We use ERS data from the end of 1999 to 2002 to help constrain the early temporal behaviour of the fault creep and combine this with the longer Envisat dataset to constrain the creep displacement timeseries from the end of 1999 to the end of 2010. The creep rate exponentially decays with time after the earthquake to a near steady rate. We model this temporal behaviour using rate and state friction theory and determine variations in frictional properties within the creeping region.

The recent tectonic activity in north-western Iran seems to be dominated by the NW-SE striking right-lateral North Tabriz Fault (NTF) where regional seismicity (historical and modern one) concentrates. This is further supported by the regional GPS-inferred velocity field that markedly changes direction across the NTF from northward motion south of the fault to north-eastward motion north of it. Next to the NTF all other major strike-slip faults of the wider Iranian-Turkish Plateau region run in a similar NW-SE strike and similar also to the thrust faults of the Caucasus in the north. However, on 11^th of August 2012 the region some tens of kilometres north of the NTF was unexpectedly struck by two shallow earthquakes with magnitudes of Mw 6.4 and Mw 6.2 that occurred only 11 minutes apart. The mechanism of the first event is inferred to be a pure right-lateral strike-slip earthquake, which ruptured the surface and produced an about 12 km long EW-striking surface rupture, closely aligned with the teleseismically inferred mechanism. This mechanism appears to be at odds with the so far established tectonic regime of the region. In these tectonic models the influence of smaller faults is neglected, partly, it seems, because there is very little known about them. Furthermore, the mechanism of the second earthquake of the doublet and the causative fault are not robustly inferred so far, due to the overlapping teleseismic wavefield of the earthquake doublet. A closer investigation of the earthquake doublet and its aftershock region therefore could shed more light into the orientation of the activated faults and their interaction, which would contribute valuable information on the recent regional tectonics.

We use teleseismic and regional seismic waveforms to study the Ahar 2012 earthquake sequence complemented by RADARSAT-2 InSAR data, provided by the SOAR2 supported project #16736. For the first earthquake strike-slip rupture on a E-W oriented fault is evident. However, the identification of the rupture plane for the second event is ambiguous. So far, different authors assume a roughly ENE-WSW oriented fault plane with a significant thrust component, which led to the assumption that both events ruptured the same EW-oriented fault. Our results from regional waveform moment tensor analysis for the second earthquake, while also pointing to a significant thrust component with an orientation of either EW or NNE-SSW, locate the its source about 6 km north-west of the first one. Also the sequence of aftershocks, located with the regional data, show activations mainly north and north-west of the first mainshock and alignments along NNE-SSW structures. The activation of a similarly oriented fault during the second mainshock is therefore a likely scenario to be considered and tested. We exploit regional waveforms to look into wavefield directivity effects and to invert for finite source models of the earthquake doublet. Moreover, we use the InSAR surface displacement measurements to complementary provide estimates on the source orientations and mechanism dislocations. With additional information from geology and tectonics we will present a multi-data analysis of the sequence and discuss technical difficulties that arise when studying this particular earthquake doublet or earthquakes sequences in general.

The question whether continental plates deform internally or move as rigid
blocks has been debated for several decades.  To further address this
question, we use large-scale InSAR datasets to study how eastern Anatolia
and its surrounding plates deform. We find that most of the deformation is focused
at the North and East Anatolian faults and little intra-plate deformation
takes place. Anatolia is therefore moving, at least its eastern part, as a
uniform block. We estimate the slip velocity and locking depth of the
North Anatolian fault at this location to be 20 mm/year and $\sim$14 km,
respectively. High deformation gradient found near the East Anatolian
fault, on the other hand, suggests that little stress is accumulating
along the eastern sections of that fault.

The atmospheric contribution to the phase signal is one of the largest challenges for applying InSAR for small deformation signals. This is also the case in our study over the Hyblean Plateau in south-east Sicily, Italy.

Temporal and spatial filtering are for DInSAR time-series analysis the most common approaches to address the atmospheric contribution totectonic signals or surface deformationsdue to varying water vapor content,mainly in the troposphere.The filtering operation however might be at the cost of smoothing the signal and thus remove slowtectonic deformations.

In this study we processed a time-series of an InSAR small baseline network using the StaMPS software (Hooper 2008).Instead of applying the standard filtering technique, a tropospheric correction of each interferogram was conducted using weather model data. This approach is included in the InSAR Atmospheric Correction Toolbox (Bekaert et al, in review) and converts the pressure, temperature and relative humidity into a phase-delay along the radar line-of-sight. We used the ERA-Interim weather model at a ~75km resolution, provided by the European Center for Medium-Range Weather Forecast (ECMWF).

For this aim we used 49 descending and 58 ascending Envisat SAR images, spanning the time period from 2003 till 2010. In addition, we have processed 30 SAR images of RADARSAT-2 for the period between 2010-2012. Furthermore, we used the different viewing geometries and the integration of GPS data to decompose the single line-of-sight velocities into a 3-dimensional field vector by applying the SISTEM approach (Guglielmino et al. 2011).

Preliminary results are in good agreement with studies based on ERS data along the coast (Canova et al. 2012). More important, however, is that the atmospherically corrected data retain the deformation signal along geological structures whilst the standard filtering approach is cancelling out very slow deformation patterns. Simultaneously, the variability of the signal in space is diminished with regard to the non-filtered stack and thus gives more confidence on the deformation patterns observed by InSAR measurements. Consequently, the decomposition of the line-of-sight velocities and the integration with the GPS data allowed differentiating between horizontal and vertical movement. Due to those two post-processing steps, we might detect a zone of probable shear stress in east-west direction along the north-south oriented Scicli-Ragusa strike-slip fault, which is poorly understood due to the lack of dense field instrumentation.