Enigma of the Olyutorskii Earthquake Resolved by SAR Interferometry

The Mw 7.6 Olyutorskii earthquake of April 20, 2006, struck the southern edge of the Koryak Highland, in a region of great complexity at the junction of the North American, Eurasian, and Pacific plates. This seismic event was notable for several remarkable features. Firstly, it had an unexpectedly large magnitude, leading to a reassessment of the seismic hazard in the Northern Kamchatka region. Secondly, the GCMT focal mechanism solution showed two nodal planes corresponding to nearly thrusting displacements on a 40–51° dipping rupture plane. However, geological field studies conducted in the epicentral area allow to map at the surface three primary segments of a 120 km long rupture zone. The central segment showed predominant right-lateral strike-slip displacements, while the NE and SW segments exhibited mostly thrusts steeply dipping SE with smaller strike-slip components. Thirdly, aftershocks were recorded in an area extending over 200 km southwestward from the surface ruptures mapped in the field, with an intense activity observed in a vast area 75 km to the NW from the surface ruptures. Fourthly, although the ruptures found during fieldwork were dipping to the SE, most aftershocks occurred NW of them, with their depth decreasing in the NW direction. SAR interferometry resolved this apparent discrepancy between seismological and field-geology data, indicating that the primary displacements occurred northwest from the central and southwest segments of the ruptures at the earth's surface. We present in our study a new rupture model based on SAR, GPS, and field geology data. The model consists of a major blind thrust extending NW and three rear subvertical ruptures that reached the earth's surface and were mapped during the field survey. Additionally, models of the main Mw 6.6 aftershocks that occurred on April 29, 2006, and May 22, 2006, are presented. Our new model reconciles all collected data, with a direction of movements on the major thrust agreeing with the rotation of the previously suggested Beringia microplate around a pole situated in the North of the Chukotka peninsula.


Introduction
The M w 7.6 earthquake that occurred on April 20, 2006 was located at the southern edge of the Koryak Highland, which is situated in a complex region at the junction of the North American, Eurasian, and Pacific plates (as depicted in Fig. 1).The number of plates in this area and their boundaries are still a subject of debate.According to the general plate-tectonic model, the North American plate (NAP) encompasses the Bering Sea and Chukotka peninsula and extends westward up to the Cherskii ridge at 140°E (about 1000 km westward from the left border of Fig. 1).The Pacific plate moves in the north-west direction along the Komandorski-Aleutian arc and converges under the Kuril Island arc towards the NAP at a rate of about 75-80 mm/yr (Argus et al., 2011;DeMets et al., 1994).Based on this model, no tectonic activity is expected north of the Kamchatskii Peninsula.However, active diffuse seismicity (Mackey et al., 1997) and GPS data on the deformation field of the Northern Kamchatka, Chukotka, and the Bering Sea (e.g., Gordeev et al., 2001) allow to postulate the existence of the rotating clockwise Beringia plate and the moving eastward Okhotsk plate (Cook et al., 1986;Seno et al., 1996).According to these models, the diffuse plate boundary in the region of the Olyutorskii earthquake would be a convergent margin also accommodating a relative left lateral movement.The detailed comparison of regional fault tectonic and seismicity is resented in Mackey et al. (1997).
Therefore, it is crucial to constrain the rupture geometry and coseismic displacement fields for the Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s00024-023-03351-6.construction and further development of the plate tectonic model of this complex region.We present in this paper new fault plane models of the main seismic event that occurred on April 20, 2006, and two of its strongest aftershocks, which we constructed based on SAR interferometry and GPS data.Compared to our previous investigation of this seismic occurrence (Mikhailov et al., 2018), we have improved the processing of the SAR images correcting the line-ofsight (LOS) displacement field, and incorporated data from two GPS stations located in the vicinity of the focal zone.This allowed us to build a more detailed model of the main event and the two main aftershocks.We then discuss the obtained results in the context of the regional tectonics and existing geological and geophysical data.

Olyutorskii Earthquake: Data and Tectonic Setting
According to current geological and geophysical evidence, the southern part of the Koryak Highland and the Northern Kamchatka region were developed by the accretion of multiple terranes to the Asian continental margin during the Late Cretaceous and Cenozoic periods (Fedotov et al., 2011).Among the significant boundaries separating these terrains are the Vyvenskii-Vetveiskii suture and the Vetlovskii fault zone, indicated in Fig. 2 by red lines I and II.The Vyvenskii-Vetveiskii zone marks a transition between rocks of varying geological ages and compositions (Fedotov et al., 2011), and it is clearly underlined by gravity and magnetic anomalies (Moroz et al., 2015).The northern (Vyvenskii-Figure 1 Topography map of the study area (in meters).The red rectangle shows the epicentral area of the Olyutorskii earthquake, as displayed in Fig. 2. The main event and its two strongest aftershocks are marked by the red and blue stars Vetveiskii) suture dips southeastward, while the southern (Vetlovskii) one dips northwestward (Shapiro & Sololv'ev, 2009).These sutures form a bowlshaped structure that was likely reactivated during the seismic crisis that occurred in April and May 2006.
Based on geological and geophysical data, a crustal cross-section indicates the presence of several SE dipping crustal boundaries that continue down to the Moho discontinuity in the region where the Vyvenskii-Vetveiskii suture is exposed (Moroz et al., 2015).In addition, on vertical sections that run NW-SE across the epicenter area, aftershocks cluster along southeastwards dipping lines (Lander et al., 2010).Figure 2 shows that the main Olyutorskii earthquake and almost all its aftershocks occurred between the Vyvenskii-Vetveiskii and Vetlovskii sutures.
The magnitude of the main event reached M w = 7.6,1 considerably above the magnitudes of the seismic events ever instrumentally recorded in the area (Lander et al., 2010).The location of the epicenter was in close proximity to the less severe (M w = 6.6)Khailino earthquake, which occurred on March 8, 1991, and was the largest instrumentally recorded earthquake in the area before the Olyutorskii earthquake (Lander et al., 2010).
The main event, whose epicenter is indicated by a magenta star on Fig. 2, was accompanied by numerous aftershocks, including two major ones of magnitude M w = 6.6 each, which occurred on April 29 and May 22, 2006, and are indicated by blue stars in the same figure.Despite the fact that the epicenter of the April 29, 2006 earthquake was situated at a considerable distance to the southeast of the main event, near the coast of the Olyutorskii Bay, it is classified as an aftershock due to the occurrence of a strong foreshock in its vicinity just 35 min after the Olyutorskii earthquake.The epicenter of the May 22, 2006 aftershock is located within the cluster of aftershocks that mostly took place a few days after the main event.By the end of March 2008, the USGS reported 309 earthquakes of m b [ 3 in the area.Permanent stations of the Kamchatka regional network supplemented by three temporary seismic stations, which operated in the epicentral zone from May 2 to 17, recorded a total of 2529 earthquakes (Lander et al., 2010).Lander et al., (2010) estimated that 99% of aftershocks occurred at depths less than 35 km, despite the large depth uncertainty for many events (about 30%).The confidence localizations for hypocenters extend somewhat into the mantle.Depth of the main event was 22 km.The most reliable depth estimates were obtained for the earthquakes registered by the temporary network.The depths of these earthquakes do not exceed 27 km, with an uncertainty better than 5 km for 78% of them (Lander et al., 2010).Therefore, they were located within the crust, which is estimated to be 40 km thick (Moroz et al., 2015).Although there is a large uncertainty in the depth estimate, the averaged aftershock depth consistently increases from \ 5 km in the NW part of the aftershock's cloud to 20-25 km in its SW part.
The Olyutorskii earthquake has several unique features.Firstly, its unexpectedly high magnitude necessitated a reassessment of the seismic hazard in the Northern Kamchatka region.Secondly, the focal mechanism solution provides two nodal planes (GCMT2 ), with one strike-dip-rake of 207°, 40°, 76°, and the other with 45°, 51°, 101°, suggesting displacement along a rupture plane dipping 40-51°.However, geological field studies conducted in the epicentral area (Pinegina & Kozhurin, 2010;Rogozhin et al., 2007) revealed a different pattern of movements along the three main segments of the mapped 120 km long rupture.The rupture predominantly stretches in the Northeast direction (blue lines on Fig. 2).Its NE segment is shifted 15 km NW of the central one, while the SW segment is shifted SE of the central one, with its southern edge stretching almost EW.As observed in field studies, displacements in these three segments differ, with predominantly right-lateral strike-slip movements on the central segment, but mostly thrusts at steeply SE dipping ruptures with smaller strike-slip components on the NE and SW segments.
Thirdly, the area covered by aftershocks extends southwestwards for more than 200 km, i.e. much further south-west than the surface ruptures mapped in the field.A significant amount of activity was also detected in an area as far as 75 km to the NW, up to the Vyvenskii-Vetveiskii zone (red line marked by number I in Fig. 2).Field studies have shown that ruptures found in this area dip to the SE, but most aftershocks occurred to the NW of them (Lander et al., 2010).Therefore, there is an apparent discrepancy between seismological and field geology data.
The seismic rupture of the Olyutorskii earthquake appears to be complex, as indicated by the focal mechanism and distribution of aftershocks.Lander and Pinegina (2010) proposed a pattern involving a blind thrust dipping at around 45°to the southeast, and extends to the northwest up to the Vyvenskii-Vetveiskii suture, as well as several nearly vertical rear secondary ruptures accommodating strike-slip and smaller thrust displacements.This interpretation was later supported by a SAR interferometry analysis (Mikhailov et al., 2018).Here, we revisit the rupture model using corrected SAR data and incorporating GPS data, which we did not used in this previous study.Since ERS-2 SAR images did not cover the entire epicentral area, the GPS data allowed us to Figure 2 The main event (magenta star), two main aftershocks (blue stars representing the April 29 (#1) and May 22 (#2) aftershocks, respectively), according to the USGS catalogue.Other symbols indicate aftershocks of varying energetic classes K as per the local scale (Fedotov, 1972), which correlates with magnitude as M w = K/2-1.15(e.g.Abubakirov et al., 2018).Blue lines denote ruptures identified at the surface in the Vivenka river valley.The central segment exhibits a right-lateral strike-slip, while the NE and SW segments are predominantly SE steeply dipping thrusts.Red lines represent the Vyvenskii-Vetveiskii (I) and Vetlovskii (II) suture zones.The yellow star marks the location of Tilichiki GPS site determine the extent of the seismic rupture more precisely.Given the remoteness and lack of permanent seismic stations, satellite data is crucial in this region and has shed new light on the structure of this earthquake area.

Displacement Fields Estimated by SAR Interferometry
Forty six satellite images from the European Space Agency's (ESA) ENVISAT and ERS-2 satellites, as well as from the Space agency of Japan's (JAXA) ALOS-1 satellite, cover the area affected by the Olyutorskii earthquake and its aftershocks.The search area for images was limited to longitudes 165.5°and 168°E, and latitude 60°and 61.5°N, spanning from April 2005 to October 2007.Our collection (refer to Supplementary materials) includes 6 ENVISAT images (four from descending 202D and two from ascending 323A orbits), 36 ERS-2 images (18 from 202 D and 18 images from 431A orbits), and only 2 ALOS-1 PALSAR images acquired much later than the main seismic events (June 01, 2006 andJuly 19, 2006).We used the ASTER-2 GDEM topography grid, which covers regions above 60°N, and the ENVI SARScapeÒ software for image processing.
The DInSAR technique relies on the analysis of the coherence between two radar images.The coherence is affected by various factors, including weather conditions, presence of snow cover or dense vegetation.Besides, inappropriate imaging geometry can lead to loss of coherence.Indeed, most of the collected satellite images were unsuitable for interferometric processing.Only two ENVISAT images from the 202D orbit, acquired on May 1st and June 5th, 2006, respectively provided an interferogram containing a deformation signal above the background noise level for the area and time of the May 22nd, 2006 strong aftershock.Another interferogram, based on a pair of ENVISAT images from orbit 323A acquired on May 10th and June 14th, 2006, covered the area of the main event but after its occurrence, and did not reveal any noticeable postseismic deformation.Among the ERS images, only a pair from the 431D orbit, acquired on October 19th, 2005, and July 11th, 2007, showed displacements, covering the area and time interval of the main event and its April 29th, 2006 aftershock (M w 6.6).(see Fig. 3).Coherence maps and unwrapped interferograms are provided in Supplementary materials.
In processing SAR images, it is crucial to mitigate the undesired atmospheric effects.These effects create a smoothed regional component in the unwrapped interferogram that is challenging to distinguish from the ground displacements resulting from a significant earthquake.As depicted in Fig. 3, the LOS displacement field does not exhibit any noticeable regional component, and instead shows both the main event and the April 29, 2006 aftershock (M w = 6.6).This suggests that the atmospheric effects were effectively mitigated.
The differential interferogram technique reveals the relative phase-shift of the radar signals reflected from the same surface element (pixel) between the first and second image acquisitions.The phase changes in the interval [-p, p].To estimate the displacement field towards the satellite (LOS), an unwrapping procedure is required, which involves the addition of an integer number of periods 2pn and multiplication by the wavelength k/2.This procedure suggests that LOS displacement field is smooth, i.e., that the difference of displacements in neighboring pixels does not exceed k/2.Therefore, when the jump between neighboring pixels in the real displacement field exceeds k/2, it will most likely be smoothed out.To ensure successful phase unwrapping for ERS and ENVISAT images, the gradient of the displacement field should not exceed 2.8 cm per 20 m.Consequently, the narrow deformation zones along the ruptures mapped on the surface are not visible in the unwrapped interferograms.The regional (longwavelength) component of the displacement field (Fig. 3) indicates displacements toward the satellite (''uplifts'') northwest from the central and southwest segments of the ruptured area on the earth's surface (blue lines on Figs. 2 and 3).The amplitude of LOS displacements exceeds 0.6 m in the Vivenka river valley.The area of positive LOS displacements extends toward the Vyvenskii-Vetveiskii suture (line I on Fig. 2), which runs along a plateau at an altitude of more than 500 m above the Vivenka river valley, where field surveys have been conducted.Negative LOS displacements are up to -0.25 m, and positive LOS displacements in the area of the M w 6.6 aftershock reach 0.38 m.

Interpretation of the LOS Displacement Fields
When modeling the displacement field at a seismic rupture using surface displacements, it is common to approximate the rupture as a set of rectangular planes, each of which can be further divided into rectangular elements.This leads to an inverse problem in which the objective is to determine the along-dip and strike-slip components for all elements of the fault planes providing the best fit with GPS, SAR interferometry and other available data.Assuming that the medium accommodating the ruptures is purely elastic or viscoelastic, this inverse problem is linear but ill-posed.It should be noted that a successful resolution (chessboard) test does not necessarily guarantee good quality of the solution if the model of the seismic rupture does not properly approximate the real one (Mikhailov et al., 2018).Therefore, to obtain a reliable solution, it is crucial to incorporate as much additional data as possible.Our model is based on the proposal put forward by Lander and Pinegina (2010) that the main coseismic displacements took place on a southeast dipping blind thrust fault extending northwestward from the ruptures found during the field survey.Indeed, the area with the most aftershocks is located north-west of the ruptures, where our SAR interferometry data reveal the main LOS displacements (as shown in Fig. 3).Furthermore, there is a systematic decrease in aftershock depth in the northwest direction.
Our fault plane model includes a blind rupture separated into 5 9 3 elements along strike and dip respectively (plane #1 on Fig. 4).Even this plane almost entirely covers the epicentral region, no solution can fit both the GPS and SAR data simultaneously.To achieve a synthetic displacement close to the actual one at the Tilichiki GPS site (Fig. 4), an additional plane (#2 in Fig. 4) extending southwest from the area covered by measured LOS displacements, where numerous aftershocks occurred (Fig. 2), is required.Neither a model without the three steeply dipping elements (#3, 4, 5, their displacements are indicated by black arrows on Fig. 4), nor a model without the blind element can provide a solution that adequately fits all observed data.Only a solution incorporating the 5 planes (one subdivided into 15 parts) plus one plane (#6) for the M w 6.6 aftershock can provide a good fit to both GPS and SAR data (Fig. 4).
To model earth's surface displacement field, we used the Static1D software based on Pollitz (1996) solution.This software allows for the calculation of strain and stress fields within a viscoelastic spherically layered planet induced by normal and strike-slip displacements along a rectangular plane.The solution is linear with respect to strike-slip and normal displacements, resulting in a linear system of equations whose unknowns are the normal and strike-slip displacements at each element of the fault plane model (which comprises 20 elements in our case).Since this inverse problem is ill-posed, we employ a singularvalue decomposition method, cutting singular values smaller than 10 -4 of the largest one, to solve the linear system of equations.
Using the approach proposed by Diament et al. (2020), we assume that the rake of the displacement vectors should be close to the assigned values.Specifically, for planes #1, 2, 3, and 5, we look for displacements close to pure thrust (rake close to 90°), while for planes 4 and 6, we suggest thrust with a leftlateral component (rake at the hanging wall close to 135°).These conditions were incorporated into the functional together with the misfit between real and synthetic SAR and GPS data, and its minimization provided displacements along strike and along dip for the twenty elements of the rupture planes (Table 1).Our model predictions are in good agreement with the observed data, with the maximum difference occurring in the NW part of the LOS displacement map.The limited extension of the SAR image does not allow us to check whether extending part of the plane 1 would reduce this small gap between the observation and the model.
Table 1 presents the parameters of the fault planes of the rupture model of the Olyutorskii earthquake.The dimensions of plane #6 were estimated using ratio between along strike and along dip length suggested by Wells & Coppersmith (1994).The obtained solution (Fig. 4) shows that elements #3-5 are mainly characterized by right-lateral strike-slip (with rakes of 126°, 111°, and 119°, respectively), while pure thrust dominates on plane #1, although a small rightlateral component appears in its eastern elements.Normal thrust dominates on plane #2.The April 29, 2006 aftershock is characterized by a strong rightlateral component.It is important to note that the SAR data only covers a small portion of its epicentral area, and therefore, the obtained parameters should be considered as a first-order approximation.
The amplitude of displacement varies on plane #1, with a maximum value reaching 3.8 m.On the other hand, the displacement amplitude on plane #2 is relatively small, and there are numerous aftershocks of low magnitude registered by temporal seismic network.However, the inclusion of this plane is crucial to obtain the movements recorded at the Tilichiki GPS site.
The last column of Table 1 presents the seismic moment calculated assuming a shear modulus of 32 GPa.The total seismic moment of all the planes, excluding the main aftershock of April 29, 2006, is 3.90*10 2 °N*m, when the global CMT catalog estimated a seismic moment of 2.99•10 2 °N*m based on analysis of seismology records.The difference in the seismic moment estimates is likely due to the fact that the SAR data covers a time interval that includes both the main event and all its aftershocks.
The ENVISAT satellite images obtained from the 202D orbit on May 1st and June 5th, 2006 allow to identify displacement field caused by the M w 6.6 aftershock that occurred on May 22nd, 2006 (Fig. 5).The LOS displacements appear to be relatively small, ranging from -0.25 to 0.35 m.The best-fit model solution for this event has a strike of 323°, a dip of 65°, and a length of 24 km along strike and 10 km along dip, subdivided into three segments.We estimate the average displacement amplitude to be 1.24 m and the seismic moment to be 0.95*10 19 N*m.This estimate is in between of those provided by the USGS and GCMT catalogues, which give a

Conclusion
The M w 7.6 Olyutorskii earthquake of April 20, 2006 occurred at the southern boundary of the Koryak Highland, at the junction of the North American, Eurasian, and Pacific plates.This seismic event is characterized by several notable features, including a surprisingly high magnitude.The GCMT focal mechanism solution indicates two nodal planes corresponding to nearly pure thrusting displacements on a rupture plane with a dip of 40-51°.However, geological field studies in the epicentral area have identified three main segments of the 120 km-long rupture mapped on the earth's surface.The central segment is dominated by right-lateral strike-slip displacements, while the NE and SW segments exhibit primarily thrusting along steeply SE-dipping ruptures with a smaller strike-slip component.
Aftershock activity after the M w 7.6 Olyutorskii earthquake extended over an area of more than 200 km to the southwest of the surface ruptures that were mapped during the fieldwork.Additionally, a significant number of aftershocks was detected in an area 75 km northwest of the surface ruptures, towards the Vyvenskii-Vetveiskii suture zone.The mapped surface ruptures were found to dip to the SE, while the aftershocks mostly occurred to the NW of these ruptures, with decreasing depth to the NW.SAR interferometry resolved this discrepancy between the seismological, geodetic, and field-geology data, revealing that the main displacements occurred northwest of the central and southwestern segments of the surface ruptures.The amplitude of the LOS displacements (''uplifts'') exceeded 0.6 m, suggesting that the main displacements were the result of a blind thrust extending to the NW from the surface ruptures.We have developed a new rupture model by integrating SAR, GPS, and field geology data.This model includes a blind primary thrust fault that extends to the northwest, and three rear subvertical ruptures that have reached the earth's surface and have been identified by field geology studies.Furthermore, we constructed a rupture model for the two major aftershocks, namely the M w 6.6 events that occurred on April 29, 2006, andMay 22, 2006.Our new model successfully reconciles all the data collected, including data from the two previously unused GPS sites.The direction of movements on the major thrust is consistent with the rotation of the previously suggested Beringia microplate (Cook et al., 1986;Gordeev et al., 2001;Mackey et al., 1997;Seno et al., 1996) around a pole located to the north of the Chukotka peninsula.

Figure 3
Figure 3 LOS displacements in meters induced by the main event and its aftershock of April 29, 2006, as unwrapped from the interferogram based on two ERS-2 images taken on October 19, 2005 and July 11, 2007 along the orbit 431D (the frame of the images is indicated by a blue rectangle).The positive LOS displacements are directed towards the satellite.The epicenters of the main event and the two strongest aftershocks are marked with black stars, while the red lines indicate regional suture zones.The blue lines represent the ruptures that have been mapped at the earth's surface in the Vivenka river valley.The blue triangle shows the Tilichiki GPS cite

Figure 4
Figure 4 Rupture model of the Olyutorskii earthquake that occurred on April 20, 2006 and its main aftershock of April 29, 2006.The black rectangles in the figure outline the projection of elements of the fault plane model on the earth's surface.The colored scale in plot A represents the synthetic LOS displacements (in cm).The blue arrows indicate displacements of the hanging wall of the blind rupture, where the up-dip component is dominant.The black arrows show displacements at three steeply dipping elements that approximate ruptures mapped at the earth's surface.As these elements dip to SE, the displacements contain thrust plus right-lateral strike-slip components that agree with the field data as reported by Pinegina and Kozhurin (2010).The black arrow at plane #6 shows calculated displacement at the rupture of the strong aftershock that occurred on April 29, 2006.This plane dips to NW, so the displacement also contains a slight right-lateral offset.The red arrows indicate synthetic, and the black ones show real displacements at two GPS sites near the epicenter.At the Tilichiki site (166.145E, 60.446 N), the arrows perfectly fit.Plot B of the figure presents a comparison between the real (the colored scale) and the synthetic (isolines) LOS displacement fields.Plot C shows the epicenters of seismic events with their magnitude in the inset from April 20 to April 30, 2006, as reported by the USGS catalogue, on the background of the rupture model and real LOS displacements.Colored scales (in cm) on plots A, B, and C are the same

Figure 5
Figure 5 Model of the M w 6.6 aftershock on May 22, 2006.The colored map displays real LOS displacements (in m), while isolines represent synthetic displacements.The green star indicates the epicenter of the aftershock according to the GCMT catalogue, while the black triangles show other aftershocks with magnitudes above 5 during the period covered by the SAR images.The green arrows depict displacement on the seismic rupture, with the maximum arrow representing 1.8 m

Table 1
Parameters of the fault planes and the best-fit solution