Aseismic creep and seismic potential
In classical view of earthquake faulting, active faults release abruptly years of accumulated strain (Davison, 1901; Reid, 1910, 1911). Within this perspective, the fault is considered mainly locked during the intervening period. On the other hand, some other faults slip (creep) continuously or transiently at slow rates with or without producing any destructive earthquakes (e.g. Tratchet, 1979; Burfold & Harsh, 1980). These steady or transient slip (creep) behaviours are occurring within or around the seismogenic zone along the faults which then may store little to no strain energy (Hussain et al 2016). Better understanding the mechanism of aseismic creep is important because it can influence the size and timing of large earthquakes. The studies over the last decades using classical alignment arrays, creep meters, GPS and InSAR observations have showed the aforementioned fault-creep activities at varying depths and rates in the seismogenic crust. Creeping faults also generate earthquakes but usually less frequently and/or of smaller size than the fully locked fault segments (Schmidt et al., 2005; Jolivet et al., 2013; Kaneko et al., 2013; Cetin et al., 2014; Maurer & Johnson, 2014). Therefore, determining the relative budget of seismic and aseismic slips, as well as their spatiotemporal interactions and relationship with fault properties, is crucial for determining the seismic potential of active faults (Bürgmann et al., 2000, Bürgmann & Thatcher, 2013; Avouac 2015; Harris 2017, Bürgmann, 2018). Aseismic slip can occur with various temporal behaviours: it can be steady (e.g. Fialko, 2006; Motagh et al., 2007), as observed during the interseismic period, or transient (e.g. Hayes et al., 2014), either triggered by a major earthquake and decaying with time (postseismic relaxation known as afterslip, e.g. Lienkaemper et al., 2001) or spontaneously generated by processes still poorly understood (Jolivet et al., 2015; Rousset et al., 2016). Frictional properties, fault geometry and lithology, as well as pore fluid pressure variations were proposed to explain creep behaviour (Avouac, 2015; Burgman, 2018). For example, the presence of clay-rich gouges and high fluid pressure appear as key factors to generate creep (Carpenter et al., 2011, Kaduri et al., 2017).
Observations of aseismically slipping faults thus stimulate critical questions about creep relevance for seismic hazard assessment (Harris et al., 2016). From a mechanical point of view, creeping behaviour of a fault is an important mechanism that plays a role in the releasing of elastic strain energy in the fault. Aseismic slip may thus be considered as a reducing factor of the seismic moment (energy) of future earthquakes. However, several case studies show that aseismic slip may also trigger earthquakes (e.g. Radiguet et al., 2018).
Figure 2.3 shows the forward calculations of interseismic fault-perpendicular velocity and strain rate profiles of a shallow creeping fault for different creeping rates. The fault-perpendicular velocity profile shows two different types of deformation signals: (1) a long-wavelength signal related to the deep tectonic loading rate, and (2) a velocity discontinuity caused by surface creep at the fault. The higher rate and depth of creep and lower strain rate concentrated close to the fault explain the direct relationship between the fault energy released by seismic (earthquakes) and aseismic (creep) processes at the fault.
Recent development and widespread use of advanced spaceborne techniques, particularly InSAR, have revealed that steady or transient surface creep behaviour along major continental faults is more common than previously thought. A recent review by Harris (2017) discusses the earthquake potential of creeping continental faults using worldwide data. Reported cases in the literature include the Hayward fault (Savage and Lisowski, 1993; Schmidt et al., 2005), the Superstition Hills fault (Bilham, 1989; Wei et al., 2009) and the Central San Andreas fault in California (DeMichele et al., 2011, Jolivet et al., 2015b, Khoshmanesh and Shirzaei 2018), the Longitudinal Valley fault in Taiwan (Champenois et al., 2012; Thomas et al., 2014), the Ismetpasa segment of North Anatolian Fault in Turkey (Ambraseys, 1970; Cakir et al., 2005; Kaneko et al., 2013; Cetin et al., 2014; Bilham et al., 2016; Rousset et al., 2016), the Izmit segment of the North Anatolian Fault (Cakir et al., 2012; Ergintav et al., 2014; Hussain et al., 2016), the Haiyuan fault in China (Jolivet et al., 2012, 2013, 2015a), the El-Pilar fault in Venezuela (Jouanne et al., 2011, Pousse-Beltran et al., 2016), and the Chaman fault in Pakistan (Fattahi & Amelung, 2016). These studies show that the spatial patterns (distribution along strike and depth, magnitude) and temporal behavior of creep (steady versus sporadic [creep bursts] creep) vary significantly. These variations may be exploited to develop a better understanding of aseismic slip processes and their relationships with seismic hazard (Avouac et al., 2015).
Figure 2.3: Profiles of interseismic velocity and strain rate across a strike slip fault using the combination of screw dislocation model (Savage and Burford, 1973) with opposite sense of motion and .Heavyside function (a) Vertical cross section of simple layered elastic block model. The vertical dark black line in the upper gray layer is elastic and seismogenic part of the fault zone. The red line shows the freely slipping part in the lower and upper part of the fault. In this configuration, creep occurs from surface down to a shallow depth D2. (b) and (c) are the results for a fixed slip rate (25 mm/yr), creep depth (5 km) and locking depth (15 km) for the fault-perpendicular velocity and strain rate, respectively, with a varying creep rate. Note how strain rate becomes very localized at the fault for increasing creep rate.
The North Anatolian Fault (NAF)
The Anatolian plate is bounded by the right-lateral North Anatolian Fault and the left lateral East-Anatolian Fault to the north and east-southeast of Turkey, respectively.
This plate is moving westward due to the convergence of the Arabian and Eurasian plates and slab suction from the Hellenic trench (Reilinger et al., 2006). The 1600-km-long North Anatolian Fault is a major intracontinental strike-slip fault and known as one of the most tectonically active faults in the eastern Mediterranean region, with an average slip rate of 24±2 mm/yr (Reilinger et al., 2006). This right-lateral fault has ruptured between 1939 and 1999 in a sequence of eight M>7 earthquakes, with a westward migration that started near the Karliova triple junction in the east and reached the Sea of Marmara in the west. This migration has been explained by the Coulomb stress transfer during and in between the successive earthquakes along strike (Stein et al., 1997). Assessment of seismic hazard in the Marmara region suggests that a large and earthquake (M>7) may occur with a probability of >50% in the next 30 years under the Sea of Marmara, 20 km south of Istanbul (Parsons et al., 2004; Armijo et al., 2005).
Figure 2.4: Anatolia plate’s relative motion with GPS horizontal vectors in Eurasia fixed reference frame (Reilinger et al., 2006).
Postseismic creep along the North Anatolian Fault
The Ismetpasa and Izmit segments of the North Anatolian Fault are the two sections on land that exhibit shallow aseismic creep in the uppermost few km of the upper crust. Creep is also inferred below the Marmara Sea from the study of microseismicity and geodesy (Ergintav et al., 2014; Schmittbuhl et al., 2016a and 2016b; Klein et al., 2017) and recent sea-floor observations have shown the existence of transient creep events (De Michele et al., 2017; Martínez-Garzón et al., 2019). Based on the observations of the slow displacement of a wall at the train station of Ismetpasa in the period 1957-1969, and railway maintenance reports, the fault creeping behaviour of the NAF at Ismetpasa was first documented by Ambraseys (1970), a decade after the first observation of this phenomenon on the San Andreas Fault (Steinbrugge et al, 1960). Various measurements afterward (GPS, InSAR, Light Detection and Ranging -LIDAR-, creep meter, field observations) allowed to better characterize the spatio-temporal properties of creep along the Ismetpasa fault section (Aytun, 1982; Eren 1984; Deniz et al., 1993; Cakir et al., 2005; Ozener et al., 2012; Kaneko et al., 2013; Cetin et al., 2014; Rousset et al., 2016; Bilham et al., 2016; Kaduri et al., 2017). The creep rate is decaying with time following the 1944 rupture (Cetin et al., 2014), so that creep has been interpreted as postseismic relaxation (Cakir et al., 2005). Gouges formed within the fault zone rocks contain low friction minerals that could explain the creeping behaviour of the fault (Kaduri et al., 2017) and microstructures of deformation show the existence of both creep and brittle deformation mechanisms in the fault zone (Kaduri et al., 2018).
Figure 2.5 : Creeping segments of the North Anatolian Fault. The active faults (Emre et al., 2013) and creeping segments are drawn with thin black and bold red lines, respectively.
The Izmit creeping section of the North Anatolian Fault was the locus of the Mw 7.4 Izmit earthquake on 17 August 1999. This earthquake ruptured 145 km of the western section of this fault, along five segments separated by releasing step-overs. The rupture extended from Düzce in the east to the submarine segment of the Hersek delta in the Marmara Sea in the west (Barka et al., 2002; Figure 1). The 1999 Izmit rupture propagated at a supershear rupture velocity (5 km/s) over a large part of the fault zone (Bouchon et al., 2001). This earthquake was followed by the Mw 7.2 Düzce earthquake 86 days after, which ruptured a 45 km-long segment of the North Anatolian Fault (Cakir et al., 2003a; Konca et al., 2010). The Izmit earthquake modified the stress distribution on this fault and played an important role in the occurrence of the Düzce event by promoting stress change and postseismic deformation transients (Hubert-Ferrari et al. 2000; Cakir et al., 2003b). First observation of creep along the central segment of the Izmit rupture was documented by Cakir et al. (2012), using GPS and a Persistent Scatterer Interferometry (PSI) analysis of 24 ERS SAR images, acquired between 1992 and 1999, and 36 Envisat SAR images, acquired between 2003 and 2009. These authors investigated pre- and post-earthquake surface deformation along the supershear segment of the Izmit rupture and concluded that, as no creep was measured before the 1999 Izmit earthquake, the surface creep is related to postseismic afterslip, with a rate of 8-10 mm/yr. Using GPS data and additional Envisat images acquired between 2003 and 2012, from both ascending and descending orbits that allowed to better separate the horizontal and vertical components of creep, Hussain et al. (2016) reported that aseismic slip was confined to shallow depths (<10 km) and confirmed its decaying creep rate through time. Their results showed that creep extends spatially from the Gulf of Izmit in the west to the east of Sapanca Lake in the east, with an average horizontal, fault parallel slip rate of 6 mm/yr and a maximum rate of 11±2 mm/yr, which is nearly %40 of the annual tectonic loading rate (Reilenger et al., 2006, Hussain et al., 2018). Such slip deficit implies that elastic strain is still being accumulated on the fault along the Izmit section. Different models have been proposed to explain such postseismic deformation after the 1999 Izmit earthquake. Postseismic slip partly consists of afterslip on the fault plane and could be controlled by the frictional properties of the fault zone after a stress perturbation (Marone, 1998). Using a three-dimensional viscoelastic finite element method, Hearn et al., (2009) and Wang et al., (2009) proposed that the postseismic deformations in the region of the 1999 Izmit and Düzce events were the result of a combination of shallow afterslip and viscoelastic relaxation of lower crust and/or upper mantle.
The most recent radar satellite constellations, with greater spatial and temporal resolution, offer the opportunity to refine the measurement of aseismic slip phenomena and to further understand the underlying processes. As mentioned before, one goal of my thesis was to show the contribution of high-resolution TerraSAR-X and Sentinel-1 InSAR data sets to better characterize in space and time the aseismic slip along the Izmit section of the North Anatolian Fault.
Ground subsidence is lowering of surface elevation. It represents widespread problems in urban areas built on thick, unconsolidated loose soil layers consisting of clay, silt and peat formations (Zeitoun et al, 2013; Changxing et al., 2007; Evinemi et al., 2016). Ground subsidence can severely and adversely impact the surface structures and infrastructure due to differential settlement of the ground elevation, ground slope and through the development of surface cracks (Fergason et al., 2015). Observation of subsidence is important for understanding the underlying geological phenomena or human-induced causes. Therefore, ground deformation measurements can provide very critical information for the protection of people’s safety and property and for the prevention of geohazards.
Subsidence has been correlated to strata lithology, formation and structure (Wang et al., 2013; Tornqvistetal, 2008; van Asselen et al., 2011; Higgins et al., 2014). However, excessive groundwater pumping is by far the largest cause of subsidence. Dewatering of loose soil layers or highly porous rocks in which the water provides structural support produces volumetric contraction and hence surface subsidence.
Since the 1960s, a considerable amount of literature has been published on land subsidence using various methods of investigations, including conventional practices and advanced ground sensing systems. Data acquisition techniques depend on the spatial extent of the anticipated deformation and results were interpreted to characterize deformation mechanism and their environmental consequences (Dang et al., 2014). These conventional techniques include optical levelling surveys (Bitelli et al., 2000), GPS surveys (Sato et al., 2003; Gourmelen et al., 2007), groundwater monitoring (Taylor and Alley 2002) and tape-extensometers (Ireland et al., 1984; Poland, 1984; Evseev et al., 2018). Advanced techniques include airborne and space based observations such as InSAR (Hoffman et al., 2001; Rucker et al., 2013) and LiDAR (Froese and Mei, 2008), which can provide high-resolution imagery of the subsiding signal.
Following the development of new satellite technologies and advanced processing techniques, InSAR has been successfully used to investigate deformation resulting from subsidence due to withdrawal of groundwater or other fluids (Béjar-Pizarro et al., 2017; Massonnet et al., 1997; Fielding et al., 1998; Wright et al., 1999; Crosetto et al., 2018), soil consolidation (Kim et al., 2010; Aslan et al., 2018), subsurface mining or collapse of old mines (Carnec et al., 2000; Abdikan et al., 2014), thawing subsidence (Rykhus et al., 2008), tectonic activity (Temtime et al., 2018), and soluble earth materials (Baer et al., 2018). Besides, InSAR has also been used to infer aquifer hydraulic properties such as compressibility and storage coefficient as well as the mechanics of aquifer system (Galloway et al., 1998; Béjar-Pizarro et al., 2017; Hoffmann et al., 2001; Tomás et al., 2009; Calderhead et al., 2011; Ezquerro et al., 2014).
We demonstrate here (Chapter 5 and 6) the contribution of Synthetic Aperture Radar (InSAR) time series analyses to better characterize the ground subsidence particularly in urban areas in northwest Turkey.
Table of contents :
2. BRIEF REVIEW OF LITERATURE
2.1 The Earthquake Cycle
2.1.1 Interseismic period
2.2.1 Aseismic creep and seismic potential
2.3 The North Anatolian Fault (NAF)
2.3.1 Postseismic creep along the North Anatolian Fault
2.4 Ground Subsidence
3.2 Principles of InSAR
3.2.1 Contributors to signal phase
3.3 Overcoming Limitations due to Decorrelation Noise
3.4 Persistent Scatterer Method (PS)
3.4.1 Procedures: from SLC to geocoded interferograms
3.4.2 Procedures: from Interferograms to PS time-series
4. SHALLOW CREEP ALONG THE 1999 IZMIT EARTHQUAKE RUPTURE (TURKEY) FROM GPS AND HIGH TEMPORAL RESOLUTION INSAR DATA
4.2 GPS Data and Processing
4.3 InSAR Data and Methodology
4.3.1 Multi-sensor insar time-series
4.3.2 PS-InSAR Processing
4.4.1 Mean Line-of-Sight Velocity Fields
4.4.2 Fault-parallel horizontal velocities versus vertical velocities
4.4.3 Horizontal Creep Rate Distribution Along Fault
4.4.4 Vertical Motion Along the 1999 Rupture
4.5 Modelling Profile Velocities
4.6 Temporal Evolution of Creep
4.7.1 Spatial Distribution of Creep and Geometrical and Lithological Control
4.7.2 High Temporal Resolution InSAR Data Reveal Burst-Like Behavior of Creep
5. ANALYSIS OF SECULAR GROUND MOTIONS IN ISTANBUL FROM LONG-TERM INSAR TIME_SERIES (1992-2017)
5.2 Background of Study Areas
5.2.1 Geological setting
5.2.2 Study Area 1: Haramidere valley and Avcilar neighborhood
5.2.3 Study Area 2: Ayamama river valley
5.4 Results and Discussion
5.4.1 Main deformation patterns
5.4.2 Site 1 : Haramidere valley
5.4.3 Site 2 : Ayamama river
5.4.4 Sites 4, 6 and 3
5.4.5 Site 5
6. INVESTIGATING SUBSIDENCE IN THE BURSA PLAIN, TURKEY, WITH SATELLITE RADAR INTERFEROMETRY
6.2 Study Area
6.4 Deformation Decomposition of Sentinel -1 Data
6.5.1 InSAR-derived land subsidence maps
6.6.1 Impact of the looking geometry on self-consistency checking between the ascending and descending InSAR observations
6.6.2 Water table variations and InSAR time series
6.6.4 Compressibility of the aquifer in the Bursa plain
6.6.3 Lithological and tectonic control over land subsidence
7. CONCLUSIONS AND PERSPECTIVES