In the beginning (of active tectonics as we know it today) there was H.F. Reid and the 1906 Great San Francisco earthquake. Reid, through his role as part of the “State Earthquake Investigation Commission” and leader of the committee appointed to consider the geophysical problems of the 1906 earthquake, provided a detailed and comprehensive report on the earthquake and formulated the elastic rebound theory, identifying the cause of earthquakes (Reid, 1910). The elastic rebound theory states that earthquakes occur as slip along faults to release previously accumulated elastic strain energy, a process commonly referred to as “stick-slip mechanism”. Because plate-tectonic motion rebuilds elastic strain, this alternation of interseismic strain accumulation and coseismic strain release constitutes a seismic cycle. Active tectonics investigations aim to resolve the details of the seismic cycle (e.g., (Burbank and Anderson, 2011; McCalpin, 2009)) where the term “active” refers to faults that have slipped during the Quaternary and may host earthquake in the future. Of specials interest is the recurrence of the largest earthquakes, reasoned by their potentially large socio-economic impact.
Active tectonics investigations rely on evidence of faulting that finds its expression in the surface or shallow sub-surface and therefore rely on high-resolution topographic and stratigraphic data sets: Sufficiently large and shallow earthquakes disrupt and displace geomorphic landforms and stratigraphic units, leaving evidence of faulting in the respective records (Agnew and Sieh, 1978; Mukherjee, 2015; Reid, 1910; Wallace, 1968). Under favorable geologic site conditions, this evidence of faulting can be preserved over many seismic cycles, enabling investigations into the recurrence characteristics of surface rupturing earthquakes, as well as geologic fault slip rates and their potential variation in space and time. During the late 1970’s and early 1980’s, a number of comprehensive studies used stratigraphic and geomorphic records and synthesized earthquake chronologies into different end-member models for earthquake recurrence and along-fault slip accumulation (Schwartz and Coppersmith, 1984; Shimazaki and Nakata, 1980; Youngs and Coppersmith, 1985). These models continue to serve as a reference for many investigations into the recurrence of surface-rupturing earthquakes. However, the ongoing accumulation of prehistoric earthquake data, numerical investigations, as well as theoretical considerations have demonstrated the inadequacy of those models for characterizing the general behavior of major active faults (Weldon et al., 2004), in part because of the complexity of the rupture process and its sensitivity to the heterogeneous nature of the Earth’s crust (Fialko, 2006; McCalpin, 2009; Weldon et al., 2005).
Technological advances within the last 40 years, especially within the last 20 years contributed significantly to this understanding. For example, the advent of laser scanning techniques and photo-based 3D reconstruction algorithms for high-resolution DEM generation as well as the increasing availability and resolution of space-borne optical and radar imagery have provided a remarkable additional resource to the active tectonics community –largely increasing data resolution as well as spatial and temporal coverage while decreasing data acquisition time (Bemis et al., 2014; Glennie et al., 2014; Nissen et al., 2012; Oskin et al., 2012; Scott et al., 2018; Zielke et al., 2010). Global Navigation Satellite Systems (GNSS) have also improved in accuracy, precision, and overall availability over the last decades, increasing the ability to resolve relative plate motion and enabling the identification of previously unknown inter-, co-, and post-seismic deformation features. (Amos et al., 2014; Bartlow et al., 2014; Gualandi et al., 2017a; Gualandi et al., 2017b; Kreemer et al., 2014; Kreemer and Zaliapin; Reilinger and McClusky, 2011; Wallace et al., 2017). Similarly, radiometric dating methods have improved significantly over the last three decades due to identification of additional chronometers, a better understanding of nuclide production rates, and the availability of more sensitive mass spectrometers. Along with modified frameworks for data interpretation, these technological advances have started to sharpen our view on earthquake recurrence characteristics as well as on- and off-fault deformation that is associated with inter-, co-, and post-seismic phases of the seismic cycle.
High-resolution topographic and imagery data
The morphology of a landscape in a tectonically active region provides valuable information about the spatial distribution and temporal evolution of strain accumulation and release (e.g., (Burbank and Anderson, 2011)). Topographic data are therefore a key data source for mapping (active) faults, folds, and surface ruptures, as well as measuring the deformation associated with the seismic cycle’s different phases. Depending on the specific topic to be addressed in an active tectonics investigation, researchers may choose between a number of different data sets that record surface morphology at different scales and resolutions.
Global data sets
Synthetic Aperture Radar (SAR) imagery, acquired during various satellite missions, is one of the main sources for global-scale digital elevation models (Rosen et al., 2000). For instance, the Shutter Radar Topography Mission (SRTM) started providing ∼ 90 m resolution DEMs in 2003 (Farr and Kobrick, 2000; Farr et al., 2007; Rexer and Hirt, 2014). A more recent SRTM release from 2014 provides ∼30 m resolution DEMs, covering over 80% of the earth’s surface between 60 N and 56 S (https://lta.cr.usgs.gov/SRTM). These data sets are widely used in the active tectonics community, reasoned by the high data quality as well as ready availability (free for educational use). In 2016, TanDEM-X was launched, providing ∼12 m resolution DEMs on a global scale (https://tandemx-science.dlr.de/) for a reasonable charge (Krieger et al., 2007; Zink and Moreira, 2014; Zink et al., 2016). These data sets present an important resource to active tectonics, enabling for instance regional-scale fault mapping, topographic- and hydrologic investigations, as well as addressing questions such as the interaction between climatic and tectonic forcing on topography. Further, SAR interferometry (InSAR) may detect topographic changes that occurred in between the acquisition of individual SAR images, providing valuable information for a range of applications that assess natural hazards such as floods and landslides (Bonn and Dixon, 2005; Joyce et al., 2009; Rott and Nagler, 2006), as well as volcanic deformations (Berardino et al., 2002; Terunuma et al., 2005). In active tectonics, single-event co-seismic displacement and post-seismic relaxation may be resolved using InSAR (Elliott et al., 2016; Walters et al., 2011; Wright et al., 2004).
Aside from SAR-based DEMs, a wide range of satellite-based imaging systems, directed towards the Earth’s surface, including ASTER (e.g., (Fujisada et al., 2001; Fujisada et al., 2012)), ALOS (e.g., (Tadono et al., 2014)) (http://aw3d.jp/en/), WorldView (https://apollomapping.com/imagery/high-resolution-imagery), Pleiades, and others, provide purchasable, high-resolution (sub-meter) optical imagery that in turn may be used to derive high-resolution DEMs, also with sub-meter resolutions, by utilizing photogrammetric approaches. These data sets provide, among other applications, a contextual framework for regional and local studies, respectively.
Observations via Global Navigation Satellite Systems (GNSS), such as GPS, Galileio, or Beidou, have contributed to active tectonics studies in numerous ways. They enable tracking current, relative plate motion as well as its potential variation through time (Bilham et al., 1997; Gan et al., 2007; Liang et al., 2013; Nocquet et al., 2016; Wallace et al., 2017; Zumberge et al., 1997). As such, they inform where strain is accumulating, whether faults are fully or partially locked or creeping, and thus inform our understanding of fault behavior and contribute to seismic hazard assessments (Bendick et al., 2000; Fialko, 2006; Kreemer et al., 2014; Shen et al., 2001; Zheng et al., 2017). Further, global strain rate estimates, derived from GNSS data, have been used in forecasting global seismicity (Kreemer et al., 2014). GNSS measurements are also used in resolving the spatial distribution coseismic deformation and on-fault slip that is produced by strong earthquakes (Banerjee et al., 2005; Hreinsdóttir et al., 2006; Simons et al., 2011). Respective data sets were also used to identify previously unknown strain-releasing phenomena such as slow-slip earthquakes (Sato et al., 2004; Szeliga et al., 2008; Takagi et al., 2016).
Regional and local data sets
Regional and local active tectonics studies, for example characterizing the geomorphology of a paleoseismic site, measuring along-fault slip accumulation, or resolving on-vs-off-fault strain-release, often require DEMs with higher spatial resolution (e.g., decimeter- to sub-meter scale) than what is provided by the aforementioned global data sets (Fig. 1). Such high resolution DEMs are currently provided by Light Detection And Ranging (LiDAR), photo-based 3D reconstruction techniques (Bemis et al., 2014; James and Robson 2012, 2014; Johnson et al., 2014), and classic photogrammetric approaches. Differential GPS measurements and classic theodolite surveys may also provide high-resolution topographic data sets. The latter are becoming less popular however, in part because of the substantial time required for data acquisition.
Decimeter- to sub-meter resolution topographic data sets either motivated or distinctly contributed to a range of active tectonics studies such as mapping of active faults in forested and urban regions (Arrowsmith and Zielke, 2009; Engelkemeir and Khan, 2008; Harding and Berghoff, 2000; Kondo et al., 2008), coseismic deformation and offset measurement (De Pascale et al., 2014; Hudnut et al., 2002; Nissen et al., 2012, 2014; Oskin et al., 2012; Ren et al., 2014, 2016; Salisbury et al., 2012; Zielke et al., 2010), interseismic deformation monitoring (Karabacak et al., 2011), landslide inventory identification and motion monitoring (Bull et al., 2010; Schulz, 2007; Van Den Eeckhaut et al., 2011), and quantitative geomorphic analysis (Brodu and Lague, 2012; Hilley and Arrowsmith, 2008; Hilley et al., 2010; Lin et al., 2013). Meanwhile, virtual fieldwork and precise deformation measurements in 3D are also developed and widely used in active tectonics studies (Cowgill et al., 2012; Mackenzie and Elliott, 2017).
LiDAR (Light Detection And Ranging)
LiDAR surveying employs laser light to measure distance and relative position to a target (Axelsson, 1999; Wehr and Lohr, 1999), given that the location of the light source is precisely known. LiDAR systems are generally capable of capturing multiple returns from a single outgoing laser pulse. This property facilitates virtual vegetation removal to create bare-earth digital elevation models even in densely vegetated areas (Glennie, 2009; Glennie et al., 2014; Meigs, 2013) –a very important ability for active tectonics investigations. In fact, this ability constitutes one of the main advantages of LiDAR with respect to DEM generation that utilizes optical imagery (i.e., classical photogrammetry and photo-based 3D reconstruction techniques), as it enables to visualize tectono-geomorphic features that are otherwise covered by vegetation and hence effectively invisible to visualization using optical imagery (Fig. 1). Thus, within the last two decades, numerous active fault systems in USA, Europe, New Zealand, Japan, and China have been extensively surveyed via airborne LiDAR. Many of these data sets are available through web-based portals (e.g., www.opentopography.org).
The first application of LiDAR in active tectonics, highlighting its potential in quantifying coseismic displacement, was a survey of the co-seismic surface rupture caused by the 1999 Hector Mine earthquake (Hudnut et al., 2002). Since then, more and more studies employed LiDAR data and advanced our understanding of active tectonics, such as the co-seismic deformation field detection using pre- and post- earthquake LiDAR DEM (Nissen et al., 2014; Oskin et al., 2012), identification of slip accumulation patterns along strike slip faults (Fig. 2 (Elliott et al., 2015; Oskin et al., 2012; Ren et al., 2016; Zielke et al., 2010; Zielke et al., 2015), or detailed fault zone mapping (e.g., (Behr et al., 2010; De Pascale and Langridge, 2012; Dolan et al., 2016; Villamor et al., 2012; Wechsler et al., 2009).
Stereo pairs of remote sensing images and aerial photographs
Deriving DEMs from stereo pairs of remotely sensed images or aerial photographs is a well-established approach in geodesy (Elassal and Carusl, 1983). In recent years, the growing availability and resolution of remotely sensed imagery such as the SPOT 6/7 (∼1.5 m resolution, launched in 2012/2014), Pleiades-1A/B (0.5 m resolution, launched in 2011/2012), or WorldView 1–4 (∼0.5 m resolution, launched from 2007 to 2016), has enabled generation of high resolution, sub-meter DEMs, by correlation of stereoscopic pairs. Recent studies further indicate that its resolution and precision are comparable with airborne LiDAR (Zhou et al., 2015a, 2015b). In regions with little to no vegetation cover, this method presents a powerful, lower-cost approach to provide high-resolution DEM or optical imagery for active tectonic studies (Bi et al., 2017; Middleton et al., 2016a, 2016b; Zhou et al., 2015a, 2015b, 2016a, 2016b).
If high-resolution imagery or DEMs that pre- and post-date an earthquake are available, then the deformation characteristics of an earthquake’s co-seismic and early post-seismic phase may be resolved. While high resolution imagery from spaceborne platforms is available only for the last one or two decades, high-resolution imagery from aerial photography spans a distinctly longer time (there are numerous aerial photographs collected during 1950s–1970s worldwide). Hence, researchers use such pre- and post-earthquake imagery to either derive high-resolution DEMs (Milliner et al., 2015; Zhou et al., 2016b) or use the registered images directly (e.g., (Avouac et al., 2006; Leprince et al., 2007) in order to determine on-fault as well as off-fault strain release that occurred in-between times of image acquisition.
Photo-based 3D reconstruction techniques
In the past several years, photo-based 3D reconstruction based on the structure-from-motion (SfM) algorithm (Bemis et al., 2014; Brown and Lowe, 2005; Fonstad et al., 2013; Johnson et al., 2014; Shimon, 1979; Westoby et al., 2012), started to provide low-cost, very high-resolution (cm-scale) topography data to the active tectonics community. This technique combines sufficiently overlapping optical images of a survey target, taken with a generic over-the-counter digital camera, from various view angles and distances. In doing so, the SfM algorithm determines the relative position of the target’s surface features, creating a point-cloud surface model of the target that may in turn be used to derive DEMs and orthophotos. A range of different camera platforms may be utilized (e.g., hand- or pole-held, ballons, or drones) depending on the size and relief of the survey area, required ground resolution, and available funds. The comparison of LiDAR-derived DEMs and these photo-based DEMs suggest that, in sparsely vegetated regions, the latter can have higher resolution and comparable precision than LiDAR (Bi et al., 2017; Johnson et al., 2014). That said, as with other photogrammetric approaches it only works well and potentially better than airborne LiDAR in sparsely vegetated areas. Under these conditions, however photo-based 3D reconstructions have proven to be very valuable and find an increasing number of applications in the active tectonics community (e.g., (Bemis et al., 2014; Johnson et al., 2014; Salisbury et al., 2018)).
Dating methods in active tectonics
Determining the age of disrupted geomorphic and stratigraphic units is a critical step in active tectonics studies, enabling to constrain a fault’s slip-rate or bracket the time of past earthquakes and thus resolving the temporal aspects of the seismic cycle. Considering the temporal scope of active tectonics studies, focusing on faults that have been active during the Quaternary, the commonly used dating methods are radiocarbon dating, in-situ cosmogenic dating and luminescence dating (Burbank and Anderson, 2011; McCalpin, 2009). Improvements in mass spectrometers over the last few decades have increased measurement precision and enabled measurement of smaller nuclide concentrations, hence increasing the number of datable samples and therefore sharpening the temporal resolution of active tectonics investigations. Especially in-situ cosmogenic dating and luminescence dating have exhibited great improvements during the last two decades, reasoned by the aforementioned advances of mass spectrometers technology and a generally better knowledge of the usable radiometric isotopes and their production rates. Additionally to these radiometric dating methods, growth features such as tree rings, (lake) varves, or those present in coral micro-atolls are also available for dating, providing high resolution temporal constraints of past earthquake rupture.
Radiocarbon dating is the most commonly used technique in active tectonics, enabling to date samples with an age of ∼5101–5104 years. This dating method utilizes the radiometric isotope 14C that is formed in the atmosphere by interaction of nitrogen with high-energy cosmic rays. Organisms (e.g., plants) continuously incorporate carbon into their cells, so that their body’s carbon isotopic ratio mimics the isotopic ratio of the atmosphere. Once an organism dies, it no longer incorporates new carbon and radioactive decay takes over, lowering the amount of 14C within the deceased organism. Knowledge of the 14C production rate at the time an organism was alive and measuring the current 14C/12C isotope ratio along with knowledge of the isotope’s decay rate, yields the sample’s age.
Radiocarbon dating is frequently used in paleoseismic investigations, where fragments of charcoal, peats, or shells are dated (e.g., (McCalpin, 2009)). In concert with stratigraphic evidence of faulting identified in a paleoseismic trench, these dates enable to determine the time of past earthquakes (Fig. 3a). Note however that radiocarbon dating does not provide an earthquake age directly. The latter is constrained by bracketing ages of stratigraphic units that were deposited before and after the occurrence of an earthquake. As a result, earthquake ages are usually associated with substantial uncertainties, in part depending on sedimentation rates, proximity of datable sample to the earthquake’s event horizon, and overall abundance of material for radiocarbon dating (Fig. 3b). These substantial age uncertainties inhibit the correlation between paleoseismic trench sites along a fault (e.g., (Biasi and Weldon, 2009). Bayesian sequence modeling is performed to obtain more reliable paleoseismic dating results, assigning probability density functions to the ages of individual earthquakes (Lienkaemper and Ramsey, 2009; Ramsey, 2008, 2016), Currently, the radiocarbon calibration program OxCal is widely used in this manner (Lienkaemper and Ramsey, 2009; Ramsey, 2008, 2016).
In-situ cosmogenic nuclide dating
In-situ cosmogenic radionuclide (CRN) dating determines –in contrast to radiocarbon dating– the age of a surface and not the age of a stratigraphic unit. In-situ cosmogenic nuclide dating, first proposed by Lal & Arnold in 1985 (Granger et al., 1997; Lal, 1991; Lal and Arnold, 1985; Ma et al., 2014; Mackey and Quigley, 2014), enables to date either the exposure and burial of a surface, based on a range of potential nuclides including 3He, 10Be, 14C, 21Ne, 26Al, 32Si, 36Cl, 39Ar, 41Ca, and 81Kr. Depending on the nuclide or pair of nuclides that is utilized, this dating method can be used to date samples with an age of 103–107 years. This dating method is based on the constant bombardment of the earth’s surface and shallow sub-surface by cosmic rays (Fig. 3d,e). These high-energy rays may generate radionuclides in-situ when they interact with atoms susceptible to this process that are present at the surface (or shallow sub-surface). For surface exposure dating, the reference is the in-situ production rate of cosmogenic radionuclides. For burial dating, the reference is the in-situ decay rate of those nuclides. Ages are in both cases determined by measuring the isotope ratio of nuclides with different half-life time (Balco and Rovey, 2008; Balco and Shuster, 2009; Granger, 2006; Granger et al., 1997, 2013). The most widely used mineral for in-situ cosmogenic nuclide dating is pure quartz, containing both 10Be, and 26Al. Other minerals may also be used, depending on the specific cosmogenic nuclide that is targeted. For example, 3He isotope concentrations in olivines and pyroxenes and 36Cl isotope concentrations in basalts and limestones have been utilized in this regard and created new opportunities for dating volcanic landforms and fault scarps (Mackey and Quigley, 2014; Schlagenhauf et al., 2011; Zreda et al., 1993).
Luminescence dating determines when a sedimentary unit that contains certain minerals (such as quartz or feldspar) was last exposed to sunlight or sufficient heating (Aitken et al., 1963; Bøtter-Jensen et al., 2000; Daniels et al., 1953; Duller, 2003; Hütt et al., 1988; Lowick et al., 2012; Rhodes, 2011; Wintle and Huntley, 1982; Wintle and Murray, 2006). This dating method is based on the excitation of electrons that may be trapped metastably in a higher energy state by crystal structure defects. The energy that brought an excited electron initially into a higher energy state comes primarily from the decay of radioactive isotopes within the sedimentary unit (Fig. 3c). The amount of electrons, trapped in a higher energy state, is therefore a function of radioactive isotope decay rate and of time. Trapped electrons may fall back to their lower energy state if they are properly stimulated –using optical light for optically stimulated luminescence (OSL), infrared light for infrared stimulated luminescence (IRSL) and post-IR-IRSL), or heat for thermoluminescence (TL). As the electrons drop to their lower energy state, they emit photons of light. Luminescence dating measures how much light a sample emits –which converts to an age when a) combined with the aforementioned decay rate of radioactive isotopes within the sedimentary unit (i.e., the production rate of trapped electrons), and b) assuming that a sample has been fully bleached (i.e., it was exposed to sunlight or sufficient heat to drop all electrons to their lower energy state) or that the inherited amount of trapped electrons (i.e., inherited age) is known (Fig. 3c). Depending on the material and stimulation method, luminescence dating may determine ages of 102 to 5 × 105 years (Prescott and Robertson, 1997; Rhodes, 2011).
Luminescence dating usually gives a mixed age of the sedimentary unit as the unit may contain sedimentary fractions that exhibited different burial histories and times since last bleaching (Fig. 3c). The recently developed spatially resolved luminescence dating has the potential to distinguish different dose populations of the dated single grains, hence reveal the different bleaching or burial history, which is more informative in neotectonic studies for our better understanding the dating result and the evolution and transportation of sediments (Greilich and Wagner, 2006).
Dating using growth structures
Growth structures, such as tree rings, varves, or those present in coral micro-atolls (and others) may provide time constraints for past earthquakes or geologic fault slip rates. For example, strong earthquakes might damage or otherwise affect a tree that is positioned in the vicinity of the earthquake’s surface rupture trace. The abrupt change in the tree’s habitat or health condition may find an expression in the annual tree ring pattern which in turn may be used to assign an age to the causative earthquake (Carver et al., 2004; Jacoby et al., 1995; Jacoby, 1997, 2010; Lin and Lin, 1998; Meisling and Sieh, 1980; Sheppard and Jacoby, 1989). Coral micro-atolls may function in a similar manner, recording sudden changes in sea level that may be associated with the sudden uplift during a subduction zone earthquake. Following this approach, a ∼700-year long record of Sumatran earthquake supercycles was discovered (Natawidjaja et al., 2004, 2007; Sieh et al., 2008).
Lacustrine and marine sediment records may also be used to constrain the time of past earthquakes. A disruption in their typically very delicate, laminated stratigraphic sequence (e.g., in form of turbidite flows) may provide evidence for strong shaking, associated with an earthquake along a nearby fault (e.g., (Avşar et al., 2016; Goldfinger et al., 2007; McCalpin, 2009)). Presence of annual varves within the subaqueous sediments, allows constraining the time of the causative event (causing deposition of a turbidite) with 1-year uncertainty. That said, while the temporal resolution of those records may be very high, it is not always possible to a) rule out other factors that may have caused turbidite flows, b) associate a causative fault with the age (e.g., if many potential sources are present nearby).
The ∼100 years of development in active tectonics have modified this research field significantly, transforming it from a qualitative to a more and more quantitative one. While much has been learned about the different aspects of the seismic cycle within the last 40 years, some of the main questions in active tectonics still remain: How is the recurrence of (large and potentially devastating) earthquakes along a given fault characterized? How do faults accumulate and subsequently release tectonically accumulated strains and how do these processes vary in space and time? How much of the accumulated strain is released on-fault vs. off-fault? How do these incremental steps of plate tectonics relate to large-scale accommodation of plate motion and what can they tell us about the driving forces as well as their variation in space and time? What geological processes and mechanical behaviors govern variations strain release and accumulation, earthquake frequency and location? (Huntington et al., 2018). Addressing these questions with observational data requires sufficiently long records with sufficiently high spatial and temporal resolution. The past two decades, with the above-mentioned technological improvements, have paved the way for making these data sets available to the active tectonics community. These data sets have already started to improve our understanding or the various aspects of the seismic cycle and they will continue to do so in the foreseeable future.
Source: Active tectonics in 4D high-resolution
Authors: Zhikun Ren, Olaf Zielke, Jingxing Yu