Investigation of seismicity of the Astaneh Fault in the East Alborz

The Alborz mountains, as a folded and faulting region, is the northern region of crust shortening in Iran. Dominant mechanism (Harvard, 2002) of Alborz earthquakes is left-lateral strick-slip paralleled to the range. East Alborz is one of the active regions that plays a significant role in the tectonics of its neighbors like South Casbin Basin. Regarding the geological maps of the area, the faults have clearer outcrops in the east than the west. The strike of mountain range is changed from N110°E in the western part to N80°E in the eastern part. The Astaneh is one of the important faults in the Shahroud fault system in the east Alborz. Astaneh fault, the case study of the local networks, with about 150 km length is clearly seen in satellite pictures especially at the eastern segment. and is about 100 km in length, based on 1:250000 geological map of Semnan (Samadian et al., 1975) and Sari (Vahdati and Saidi, 1990). This fault has made a 30-40 km left lateral pull-apart basin near 53.6°E; this is the same with total left lateral offset, found by reconstructing Cambrian sandstone of the Lalun Formation with the slip rate of 3-5 mm/year. There is 45 (m) left lateral alluvial fan deposit displacement near 54°E (Hollingsworth et al., 2008). The offset dated from the last incision event at about 12 ka in the east (e.g., Regard et al., 2005; Fattahi et. al., 2006). The fault had initially introduced thrust with southward dip (Berberian, et al., 1996) and now is considered as left lateral strike slip (Jackson et al., 2002). The local government responsible say that about 2 million people live in and around the mentioned faults in Semnan, Damghan, Astaneh, Kiasar, Firuzkuh and Veresk cities, which may be affected by the activity of the fault in future. The faults in the Shahroud Fault system were mapped and their surface mechanisms are known, but their related geometry at depth and seismicity are not known. Many important historical earthquakes have occurred in the east Alborz. The most destructive earthquake was historical A.D. 856 Qumes with the estimated magnitude of 7.9 (Ambraseys and Melville, 1982). The distribution of the IGUT recorded earthquakes, because of large station spacing, can not show accurate relationship between seismicity and the faults, also IGUT large station spacing is not appropriate for computation of focal mechanism of the local earthquakes. The master event technique helps increase the accuracy of the teleseismic earthquake location (Engdahl, et al., 1998), but there are only a few events have been relocated in the studied area (small circles in fig.1), because this technique requires large earthquakes with the magnitude greater than 6.5 (Jackson and MacKenzie, 1984). Also the World Wide Seismological Station Network (WWSSN) has just been installed since 1966. Several attempts have been made to solve the earthquakes smaller than 6.5 (Shirkova, 1972; Akasheh and Breckhemer, 1984); but their solutions are not stable, so consequently they could not give a correct explanation because the majority of original data were not available. Therefore the remaining reliable solutions are Centriod Moment Tensor (CMT) at Harvard University (Harvard, 2002) and body waveform modeling (Priestley, et al,. 1994) for the earthquakes with a magnitude greater than 5.5. But there is no earthquake in the studied area with known mechanism except the earthquake of 1990/01/20 with left lateral strike slip mechanism related to the Firuzkuh fault at a longitude of 52.9°E and latitude of 35.8°N (Harvard, 2002). This paper investigates crustal velocity structure and micro-earthquake locations to explain kinematics and seismic activity of the faults in the studied area. We used IGUT network with 10 permanent stations and two local networks each with 10 temporary stations to investigate a selected area in the east Alborz. The duration of the local networks, 2007-2008 network and the 2008 network, was in total months. The temporary stations were visited every week for maintenance, checking their power supplies and internal time against the time of the external GPS of the instruments. First, we estimated the Vp to Vs ratio (1.71) by the Wadati method (Wadati, 1933). Then we determined the crust velocity model using local earthquake arrival times by 1D inversion (Kissling, 1988). Totally 121 events recorded with minimum 8 phases, maximum azimuthal gap of 180°, RMS less than 0.3 s and both horizontal and vertical uncertainties less than 2.0 km were used for computing velocity structure. We processed the inversion in two steps, after testing a few thousand multilayer models, in order to see the convergence of the inversion to a unique velocity model, first 50 random models were tested. These models were stacked with 15 layers of 2.0 km thickness from the surface to 30 km depth, with maximum 0.5 kms-1 velocity change for each layer and with the uniform starting velocity of 6.0 kms-1. Those thin layers allowed us to determine the approximate depth and velocity of the real layers. We suggested a three-layer model with two velocity contrasts located at 4 and 12 km depth over a half space. After merging these layers with a similar velocity then the starting model was repeated with the mentioned layers and the same velocity of the first step. The final selected velocity model is 5.4 and 6.0 km/s for the mentioned layers over a half space with a velocity of 6.3 km/s. Because the majority of the well located events were located shallower than 20 km, we could not determine any layer beneath this depth using inversion. We selected 834 events with a minimum of 6 phases, a RMS less than 0.5 s and a horizontal and vertical uncertainty less than 5.0 km from a total of 1443 events that were recorded by the IGUT network during this period. Regarding the distribution pattern of these earthquakes, the seismic activity is located near Astaneh, Firuzkuh, Mosha, Garmsar, Khazar, North Alborz and North Semnan faults. Also the temporary networks totally recorded 1972 earthquakes during this period. The statistics of the local network earthquakes show that the location error and RMS of about 60% of them are less than 3 km and 0.3 s respectively, but only about 40% of them are inside the networks (with an azimuthal gap less than 180°). We selected 339 earthquakes which were recorded with minimum 6 phases, a RMS value less than 0.3 s and both horizontal and vertical uncertainties less than 3.0 km. For this selection the average of location uncertainties are 1.25 km, depth of seismicity is 8.5 km, the number of the phases read for locating is 12 and RMS of time residuals is 1.4 s. Distribution of these earthquakes shows that the horizontal dimension of the two located clusters with the new model is almost the same as the length of the fault, about 100 km, has good correspondence with the fault and shows the activity of Astaneh two segments. The depth histogram of the earthquakes shows that the majority of the events have been located between 4 and 14 km. To constrain the geometry of the Astaneh fault, we plotted two cross-sections both perpendicular (in section points) to the Alborz range tectonic structures and Astaneh fault. All of the seismicity dips concluded from the depth distribution have high angle and southeast dipping but have more vertically beneath the southwest segment of the Astaneh fault than the east segment. Because of having no earthquake deeper than 23 km, the seismo-genic zone in the area was not greater than 23 km deep. We were also able to estimate the sedimentary cover thickness about 4 km.