1 Introduction

The continental lithosphere of Iran is composed by an assembly of several crustal blocks attached to the Eurasian plate (e.g. Sengör 1990). The seismic activity in Iran generally results from the convergence between the Arabia and Eurasia plates. This predominantly N-S shortening is accommodated by the deformation in the Zagros fold and thrust belt, by the subduction zone of Makran in the SE, by the Alborz, Kopeh-Dagh mountain belts in the N, and by slip along the main fault systems in the central and eastern Iran (Fig.  1 ). Obviously in such a region, high-precision earthquake locations compiled in a complete and consistent seismic catalogue are a prerequisite for tectonic interpretation and seismic hazard assessments. Due to the intrinsically coupled hypocenter-velocity problem, earthquake locations and, in particular, their uncertainty assessment strongly depend on the velocity model. For reasons of faster and more robust ray tracing in 1-D than in 3-D velocity models (Husen et al. 2011) and also due to the lack of reliable 3-D models because of inadequate coverage by high quality data in many regions, most seismological centers employ a priori 1-D models for routine earthquake locations. Ideally, such 1-D velocity models are obtained as solution to the coupled hypocenter-velocity problem, by joint inversion for hypocenter and 1-D velocity model by minimization of arrival time residuals (Kissling et al. 1995).

Fig. 1
figure 1

Seismicity of Iran in the period 2006 through July 2017 according to IRSC catalogue (earthquakes with M3.5+). The small box marks the study region. Seismic stations of the IRSC network, the BIN network, and stations in neighboring countries (http://seismology.ncms.ae/) are marked by green, blue, and orange triangles, respectively. Note that not all stations were operating during the whole period and the station distribution changed significantly with time. Z: Zagros fold and thrust belt; M: Makran subduction zone; L: Lut block; C: Central Iran; A: Alborz; K: Kope Dagh mountain belts

Full size image

In Iran, two separate and independent seismic networks, Iranian Seismic Center, IRSC, and Broadband Iranian National Seismic Network, BIN, record and collect seismological information for the same region (Fig. 1 ). Different ways of recording and data analysis are routinely followed by the two network operators, and they also use different velocity models for routine earthquake location. Consequently, the two earthquake catalogues produced by the two institutions for the same region exhibit important and noticeable differences (Fig.  2 ; Table 1 ). In this study, we aim to establish a combined consistent earthquake catalogue for the period 2006 to July 2017 for southern Iran. To consistently merge information from the IRSC and BIN networks, we applied the approach to calculate the so-called minimum 1-D model (Kissling 1988) with corresponding station corrections that may serve for routine uniform high-precision earthquake location (Husen et al. 2011) and as an initial reference model for 3-D seismic tomography (Kissling et al. 1994).

Fig. 2
figure 2

Comparison of two routine M3+ seismicity catalogues (from 2006 through July 2017) for southern Iran. a IRSC catalogue containing 6422 events, of what are 2069 well locatable, 1687 poorly locatable and 2666 very poorly locatable events (see text for criteriae), b BIN catalogue containing 4333 events, of what 1004 are well locatable, 874 poorly locatable and 2455 very poorly locatable events, and c comparison of 2069 well locatable IRSC and 1004 well locatable BIN events. Note that epicentral maps in general show very similar distributions. Hypocenter depth distributions are shown by inset in upper right corners

Full size image
Table 1 The information of 6422 and 4333 earthquakes with magnitude 3 or more in southern Iran for the period of 2006 to 201,707 recorded by IRSC and BIN networks, respectively
Full size table

2 Method

The travel time of a seismic wave generated by an earthquake and recorded at a station is non-linearly dependent on hypocentral parameters and seismic velocities sampled along the ray path between source and station. This is known as the coupled hypocenter-velocity model problem (e.g. Kissling 1988; Thurber 1992) that is routinely solved for 3-D local earthquake tomography. Ignoring the intrinsic coupling in routine earthquake location processes, however, can lead to systematic errors in hypocenter location (Thurber 1992). To best address the coupling between hypocenters and velocity model, a the so-called minimum 1-D velocity model in combination with station delays is obtained from inversion of a significantly large data set of well-locatable events by an iteratively damped least square technique (Kissling 1988). In order to provide useful information to the solution of the coupled P-velocity hypocenter problem, each seismic event should be observed by at least 6 stations (Husen et al. 1999, 2011). If the largest azimuthal gap between two reporting stations as seen from the epicenter is larger the 180°, the event is actually located outside the station network. In such case, there exists a trade-off between epicenter distance from network, hypocenter depth, origin time and velocity.

The calculation of the minimum 1-D velocity model is a trial and error procedure that involves various initial models to investigate the solution space (Kissling et al. 1994, 1995). To account for lateral variations in the shallow subsurface, station delays are incorporated in the inversion process.

3 Establishing a data set of well-locatable earthquakes

IRSC operates a combination of telemetered local networks, that are equipped with different kinds of seismometers including short-period (SS1), mid-band (Trillium-40s), and broad-band (CMG3ESP-120, CMG3T-360s, Trillium-120s, and Trillium-360s) sensors. Recorded data in all local networks are transmitted by VPN or satellite communications to the center in Tehran. The International Institute of Earthquake Engineering and Seismology operates the Broad-band Iranian National Seismic network (BIN), a regional network equipped with broad-band (CMG3T-120s) sensors and satellite communication system for receiving raw data. In both centers, arrival times of seismic phases for the earthquakes are manually picked by different operators. The DAN software, developed by the Geological survey of Canada (DAN User’s Guide 1995) that contains HYPO71 program for earthquake location (Lee and Lahr 1975), is used by IRSC for routine processing, whereas the HYPOCENTER program provided by the SEISAN software package (Lienert and Havskov 1995) is used by BIN. IRSC and BIN use different magnitude scales for their catalogues. BIN provides local magnitudes based on calibrated ML for the Alborz region (Askari et al. 2009) that is very close to relations for southern California (e.g. Hutton and Boore 1987). This relation was calculated by IRSC short period seismograms, so station corrections are not usable for BIN’s stations. Basically, local magnitude is based on synthetic Wood Anderson seismograms that is a measure of high frequency content of radiated energy from seismic source and can underestimate this parameter for large earthquakes or earthquakes with longer rise time and duration. Magnitudes in IRSC are calculated based on local magnitude relation for northwest of Iran calibrated (Mn magnitude scale) by reported body wave magnitudes from NEIC and ISC (Rezapour 2005).

Our study region encompasses the southern half of Iran between 25° N to 34° N latitude and 51° E to 63° E longitude (Figs.  1 and 2 ). We used the arrival times of P-phases from earthquakes with magnitudes more than 3 for the time period 2006 through July 2017 which were reported by the two IRSC and BIN networks (Fig. 2 ). Due to the significant number of earthquakes along the southernmost region of Iran and for better azimuthal coverage of those earthquakes, the data set has been complemented by observations from stations in neighboring countries in the south of Iran (Fig. 1 ).

With the goal in mind to merge the seismic catalogues and to establish a minimum 1-D model for southern Iran and corresponding station corrections for the combined data from the IRSC and BIN networks, we selected the 2069 well-locatable (azimuthal gap less than 180o and at least 6 P observations) events of magnitude M3.3+ from the IRSC catalogue for period 2006 to July 2017 and the 1004 well-locatable (azimuthal gap less than 180° and at least 5 P observations) events of magnitude M3+ from the BIN catalogue (Table 2 ). To account for the differences in reported event magnitudes and in numbers of stations between two networks (Fig. 1 ), we adapted the selection criteria accordingly. Since the routine hypocenter locations of these events were based on different velocity models and location procedures, we first calculated updated 1-D models with station corrections for the BIN and the IRSC stations separately (Fig.  3 ) and in a second step a good 1-D model with station corrections for all stations combined using the above selected data set of 3073 well-locatable events.

Table 2 The information of well-locatable (IRSC priorities: with azimuthal gap less than 180°, at least 6 observations, and with magnitude 3.3 or more, and BIN priorities: with azimuthal gap less than 180°, at least 5 observations, and with magnitude 3.0 or more) and poorly-locatable (IRSC priorities: with azimuthal gap more than 180°, at least 5 observations, and with magnitude 3.3 or more, and BIN priorities: with azimuthal gap more than 180°, at least 5 observations, and with magnitude 3.0 or more) earthquakes recorded by two IRSC and BIN networks between 2006 through July 2017 in southern Iran
Full size table
Fig. 3
figure 3

Overview of the merging process to establish a consistent well locatable event catalogue combining the corresponding event reports of IRSC and BIN for southern Iran. This data set was subsequently used for the calculation of the regional minimum 1-D model. The process is based on the inversion approach proposed by Kissling (1988) for the coupled hypocenter-velocity problem

Full size image

Note that within this data set of 3073 events there are many events that are reported twice, once by BIN and once by IRSC. Identification of these event pairs and combination of their observations in on single event report denotes the task of the next step in the procedure outlined in Fig. 3 . To obtain consistency in hypocenter locations for the event pairing, all 3756 M3.3+ well and poorly locatable events in IRSC catalogue and all 1878 well and poorly locatable M3+ events in BIN catalogue (Table 2 ) were relocated with this common model using VELEST in single event mode (Kissling et al. 1995).

3.1 Event pairing from the two IRSC and bin catalogues

The identification of common events between two seismic catalogues is based on similarities in hypocenter parameters (Solarino et al. 1997), in particular, origin time and epicenter location. Following the procedure applied by Quintero and Kissling (2001), consistently well-located events with origin time differences of less than 10 s and epicenter location differences of less than 40 km are paired. This search resulted in 621 event pairs individually well-located both by BIN and IRSC. Note that the consistency in pairing must be checked by careful analysis of the residuals when locating the new event hypocenter with the minimum 1-D model and station delays.

Subsequently, we tried to pair the remaining well-located events from one catalogue with the poorly locatable events (Table 2 , Fig. 3 ) from the other catalogue. We thus found an additional 417 event pairs combining the remaining 1448 well-located events in the IRSC catalogue with the 874 poorly locatable events in the BIN catalogue and an additional 151 event pairs combining the remaining 383 well-located events from BIN catalogue with the 1687 poorly locatable events in IRSC catalogue (see the procedure in Fig. 3 ). The resulting local earthquake data set for southern Iran contains 1189 well located events that originally were reported by both catalogues (so-called event-pairs) (Fig.  4 ). In addition, the search process yielded 155 event pairs that even after merging the observations from both networks remained poorly locatable.

Fig. 4
figure 4

Epicenter location map for period 2006 to 2017 reported by IRSC and BIN. Based on the relocation of individually reported well-locatable events by IRSC and BIN networks, 1189 event pairs were identified as belonging to the same earthquake. The well-well-locatable and well-poorly-locatable paired events are joined by black and gray lines, respectively. Red and gray circles indicate epicenters which were reported by IRSC and BIN networks, respectively. Small black dots mark those 8377 still unpaired events included in both catalogues

Full size image

3.2 Cleaning of data set for large individual observation errors

As in this study we are using routine arrival time picks obtained by many different observers, the data set may include some large data errors due to mis-picking of arrival time, mis-identification of first arriving phase, possible clock errors or other kind of errors (Fig.  5 ). For well-locatable events, joint hypocenter determination with VELEST provides a valuable method to identify and diagnose observation errors (Kissling 1988). By inspecting the results of coupled hypocenter-velocity inversions with VELEST, we identified several events with unrealistically large RMS residuals and several stations with unrealistically large variation in residuals from different azimuths of more than 1 s. Based on the interstation differences (Fig. 1 ) and, hence, the expected average ray length in combination with the maximum realistic lateral velocity for crustal rocks, we defined the upper limit of an event RMS at 1 s. As causes for these unrealistically large RMS and residuals, we diagnosed individual or groups of problematic phase observations (picks). Consequently, all waveforms belonging to the in total about 150 events in question were re-picked by hand (for example see Fig. 5 ). Then, after replacing or removing all problematic observations (large individual observation errors) from the data set of well-locatable event pairs, a final data set of 1115 well-locatable event pairs with a total number of 24,606 P-wave observations (Table 3 ) at 161 stations was established and used for the calculation of the regional minimum 1-D model for southern Iran. Due to the rather small number and the lower quality of S-wave arrival time observations, we did not include S data in this study.

Fig. 5
figure 5

Two examples of significant outliers arrival time picks in the routine catalogue data by a IRSC and b BIN networks detected with the Minimum1D model data cleaning procedure. RP: routine picks reported by seismological centers, and HP: hand repicks by experienced seismologist

Full size image
Table 3 The information of the final 1115 event pairs which used for 1-D inversion with respect to different magnitude ranges. This data set includes the total number of 24,606 P-observations from two IRSC and BIN national networks and Neighboring stations each with number of 14,680, 7663, and 2263 observations, respectively
Full size table

4 Calculation of a minimum 1-D model for southern Iran

The minimum 1-D P-wave velocity model with appropriate station delays is obtained as result of a series of non-linear inversions of the coupled hypocenter-velocity model problem using the code VELEST (Kissling 1988). Direct, refracted and critically refracted ray paths from source to receivers are calculated for an initial 1-D velocity model that is successively updated during the iterative inversion process. Average station residuals are also included in the inversion to compensate for shallow heterogeneous velocity structure beneath individual stations (Searcy 2003; Husen et al. 2011). These station residuals are calculated with respect to a reference station (with zero delay) and relative to the top layers velocities. Following the strategy outlined in Kissling et al. (1995), we used the data set of the 1115 well-locatable events pairs with a total of 24,606 P observations (Fig.  6 ) and searched the model space for potentially well-performing initial reference 1-D velocity models with fixed layering. Since the highest station is about 3000 m above sea level, the top of the model was set to 5 km above sea level to comprise all stations within the first layer. Since the original event locations are from IRSC catalogue and not in conformity with the chosen initial reference model, hypocenter locations needed to be updated before performing coupled hypocenter-velocity inversions. For this reason, the first VELEST runs were performed with control parameter invertratio = 2 (Kissling et al. 1995) that allows to alternate iterations between one where only hypocenter parameters are inverted for and one where hypocentral parameters, seismic velocities and station delays are simultaneously inverted for.

Fig. 6
figure 6

Distribution of 1115 selected paired high-quality events (red circle) in southern Iran for usage in the Minimum 1-D model calculation. IRSC, BIN and neighboring stations are marked by green, blue, and orange triangles, respectively. Ray paths between epicenters and stations are shown by gray lines

Full size image

The calculation of a minimum 1-D model denotes a trial-and-error process with a series of non-linear inversions with VELEST as a single trial (Kissling 1988). By this process and testing with various initial reference 1-D models, we seek a minimum 1-D model with station delays as the final inversion results for our data set and the coupled hypocenter-velocity problem that yield a stable solution with a significant reduction of data variance.

Before we searched the model space independently of existing models, we first checked what velocity solutions we would obtain from inversion of the newly combined data set for the layering according to the BIN and IRSC routine velocity models. The results clearly documented that an alternative layering was needed. Hence, we did an extensive model search (Kissling et al. 1995) using a large number of initial models with different layering. We began with a series of inversions for relatively fine layers of 5 km thickness on top and subsequently increasing to 10 km with a half space around 100 km depth. Thus we obtained a feeling for the sensitivity of the data set for the overall layering. Note that as usual with a good quality and significant size data set, initial velocities of choice do not cause a bias to the solution unless one chooses a too high velocity in the top layers. The goal of this model search was to obtain a feeling for the velocity gradient (adapted by finer layering on top) in the upper-middle crust from Pg information and on overall crustal thickness, average crustal velocity and sub-Moho velocities from Pn and from deeper earthquakes. Stability of the solution for all unknowns of the coupled problem across iterations was always rated superior than minor further reductions in residual data variance. The results of these tests clearly indicated the most important characteristics of the model families that would yield stable and low residual data variance solutions: the top layers velocities and the upper crustal velocity gradient, the few “prominent” crust-internal velocity interfaces and the Moho depth at 45 km which was surprisingly consistent in depth across the very large region and, hence, exhibits a very pronounced velocity increase in the final model. Note that if the Moho depths would exhibit strong lateral variations across the study region and it was clearly mapped by our data in different subregions, we would obtain a velocity gradient from normal mid-to-lower crustal velocities to uppermost mantle velocities. Overall we tested at least a dozen different model layerings, initial velocity-depth functions and greatly varying the control parameters (damping etc.) before settling on the final layering.

The best results were obtained with two layers of 5 km thickness for the shallow crust overlying two layers of 10 km thickness representing the upper to middle crust with P-wave velocities up to 6.15 km/s and a lower crustal layer with 6.4 km/s velocity above the pronounced Moho at 45 km depth (Fig.  7 ). Thus the obtained minimum 1-D velocity model in this study (Table 4 ; Fig. 7 ) indicates a gentle increase in P-wave velocity in the upper and middle crust, a Conrad discontinuity at 30 km depth and a sharp increase in seismic P velocity to 8.09 km/s at the Moho. Hypocenter and station distributions allow us to resolve the crust and the uppermost mantle, the latter mainly by Pn phase observations. With only few events below 50 km depth, resolution below the immediate sub-Moho layer is poor (Table 4 ). Note that the minimum 1-D model differs significantly from either IRSC or BIN routine velocity models except for the Moho depth. The strong velocity gradient within the upper crust, in particular, causes the pronounced differences in the hypocenter distribution (Fig. 7 ).

Fig. 7
figure 7

The Minimum 1-D model for southern Iran in comparison with the routinely used 1-D velocity models by IRSC and BIN (left). Right: Focal depth distribution of the 1115 paired earthquakes as relocated by the Minimum 1-D model using appropriate station delays and by routine 1-D models. Note the significantly reduced number of events with shallow hypocentral depth with respect to routine locations

Full size image
Table 4 The computed minimum 1-D P-wave velocity model for southern Iran. NHIT is the number of rays passed through that specific layer
Full size table

Station corrections denote the travel time effects of azimuthally averaged near-surface geology for stations located within the network (Kissling 1988; Husen et al. 2011). However, due to poor sampling with only subvertical rays, the first layer velocity refers to a mathematical average over all lithologies sampled within the study region and to the lithology of the reference station. This reference station should be centrally located within the network and exhibit a significant number of observations and its station delay is a priori set as equal to zero. In this study, the IRSC network KHGB station, located near the southeastern edge of the large central Iranian crustal block, was chosen as the reference station (Fig.  8 ). We determined small corrections for stations in the central part of the study region, nearby the reference station, whereas larger station delays (absolute value more than 0.7 s) were obtained toward its outer part (Table 5 ; Fig. 8 ). The corrections for stations along the edges of the study region are associated with the combined effects of near surface geology, the limited azimuthal coverage, and large scale velocity variations in the crust due to distances of observations at these stations where the difference between resolved and unresolved structure accumulates with longer ray paths (Husen et al. 2011). We observe regionally very consistent values of corrections for group of stations in the different regions. Generally, stations in the center of the study region around and including the reference station, KHGB, and most stations in the southern Zagros folded zone exhibit slower velocities in the subsurface than the stations in other regions (Fig. 8 ).

Fig. 8
figure 8

Distribution of final station delays calculated with respect to the minimum 1-D model for southern Iran. Crosses and open circles indicate the positive and negative values of delays, respectively, for stations with at least 20 P-wave observations, relative to the KHGB reference station that is marked by a green star. For stations with a good azimuthal coverage, the station delay should correspond to the near-surface geology in a relative sense referring to the subsurface of the reference station. The corrections for stations along the edges of the study region are associated with the combined effects of near surface geology, the limited azimuthal coverage, and large scale velocity variations in the mid-to-deep crust due to distances of observations at these stations where the difference between resolved and unresolved structure accumulates with longer ray paths (Husen et al. 2011). Note the generally consistent values in the different regions. Stations in the center of the study region around and including the reference station and most stations in the Zagros folded zone exhibit slower velocities in the subsurface than the stations in other regions (see text for more details). Geological map modified from the structural map of NGDIR (National Geoscience Database of Iran, http://www.ngdir.ir)

Full size image
Table 5 P-wave corrections of those stations with at least 20 observations in southern Iran calculated in the inversion relative to KHGB station as reference one
Full size table

5 Discussion

The stability of the resulting minimum 1-D velocity model and the precision of the hypocenter locations may be examined by performing a series of inversions with randomly and with systematically shifted hypocenters upon input (Kissling et al. 1995). To indicate robust solution to the coupled hypocenter-velocity problem, the results of this test must show a small variation in the Minimum 1-D velocity model and station corrections, while the hypocenters should approximately be relocated to their original location. Indeed, the minimum 1-D velocity model with its corresponding station delays for southern Iran fulfills these requirements very well. Hypocenter locations were shifted randomly and systematically by ± 10 km taking care to avoid airquakes. The test results document (Fig.  9 ) that in general well-locatable events are located with the minimum 1-D model and station delays with a precision of better than ± 3 km in epicenter and double that uncertainty in hypocenter depth. More poorly recovering of epicentral locations is observed in the southwestern part of the study region only (Fig. 9 a, b). This is related to the significantly less dense station coverage possibly in combination with the generally lower than average signal to noise ratio at these stations. Since for accurate depth determination (Chatelain et al. 1980; Husen et al. 2003, 2011) the epicentral distance to the nearest station should be equal or shorter than the focal depth. The rather large hypocenter depth uncertainty of ± 6 km throughout the region of study is due to the generally large distance between stations in relation to the average focal depth in the Iranian plateau.

Fig. 9
figure 9

Hypocenters stability test results with respect to a 10 km shift in latitude (a), shift in longitude (b), and shift in depth (c). Open circles denote coordinate differences between shifted input and minimum 1-D locations, and black solid circles are the difference of the output inversion with shifted events and minimum 1-D locations. The average remaining shifts between the minimum 1-D locations and the resulting event location and their standard deviations are given on the right side of the figures. This test documents that well-locatable events are located with the Minimum 1-D model and station delays with a precision of better than ± 3 km in epicenter and double that uncertainty in hypocenter depth

Full size image

Negative station delays denote early arrivals and imply local velocities in uppermost crust are higher than in the average 1-D model, whereas positive delays corresponds to lower velocities immediately beneath the station. As special case, we observe a negative value of about 0.1 s at KRBR and a positive value of about 0.2 s at TVBK station although these two stations are in nearly the same location with a distance of less than 500 m between them. Furthermore, both stations are close to the reference station (Table 5 ; Fig. 8 ). TVBK is located on unconsolidated sediments and exhibits an unusually high level of noise and this leads not only to travel time picks with generally assigned larger observation uncertainty but possibly also to significant mispicks and general delay. For the stations in the southeastern Iran, including JASK, CHBR, and NGCH along the coast of Oman sea (Table 5 ; Fig. 8 ), we observe consistently early arrivals. These may result from a shallowing of the Moho (e.g. Shad Manaman and Shomali 2010) since arrivals at these stations are dominated by phases that sample upper mantle velocities. An observed significant negative delay at station KHNJ in the southern part of the study region possibly is caused by somewhat deeper seismicity in the northeast that sample a volume of high-velocity anomaly (Yaminifard et al. 2007). For stations in the same geological unit such as NIAN, BNB, GENO, and BNDS, as expected we observe similar station corrections, but the absolute value of BNDS delays is larger than the other three ones. This is almost certainly related to observation errors due to a specific instrumental problem with the digitizer at this station.

The new minimum 1-D model with station delays for all IRSC and BIN stations combined allows to etablish a complete seismic catalogue for period 2006 to 2017 with all events reported by the two network consistently relocated (Fig. 10). This new earthquake catalogue includes 1344 paired (category A), 3104 unpaired well-to-fairly well locatable (category B), and 4914 unpaired very-poorly locatable events (category C). Note that while this catalogue is complete, there is a high probability that some lower magnitude and poorly locatable events might actually be listed twice with corresponding hypocenters so far distant that recognition is not possible. Furthermore, possibly due to noisy stations in some regions of southern Iran, the two catalogues by IRSC and by BIN contain significant numbers of events reported by only 4 or even fewer stations which may not be located with any kind of uncertainty estimates. In our catalogue, we list them as very poorly-locatable events (event category C) and they may only represent the seismicity distribution in a qualitative sense. The new epicenter map (Fig. 10), however, documents that the seismicity distribution is consistent for the three quality categories of events thus allowing to draw representative conclusions about seismicity in southern Iran from the well and fairly well located part of the catalogue.

Fig. 10
figure 10

Final merged M3+ seismicity catalogue 2006 through July 2017 and comparison with tectonic setting of southern Iran (adapted from Aghanabati 2004). a This catalogue contains a total number of 10,362 events of what 1189 and 155 are well-locatable and poorly locatable, respectively, events that were reported by both IRSC and BIN original catalogues (category A, paired events), 3104 are unpaired well-poorly locatable (category B) and 5914 are unpaired very poorly locatable events (category C; see text). b tectonic setting of southern Iran with GPS velocity vector in brown arrows (Vernant et al. 2004) and CMT solutions of earthquake with magnitude more than 5 from Harvard University catalogue in the period of 1976 to 2018 (http://www.seismology.harvard.edu/CMTsearch.html). Hypocenter depth distribution is shown by inset in upper right corner

Full size image

One of the fundamental parameters for earthquake quantification is the magnitude which in combination with hypocenter locations forms the basic information for different geophysical and engineering studies (Kanamori 1983) such as seismic microzonation for earthquake hazard. The reason for magnitude differences reported by IRSC and by BIN networks for the 1344 paired events in our new catalogue (Fig. 11) originates from different magnitude scales that are used by the two networks. Reported magnitudes in IRSC and BIN are based on an adjusted magnitude scale Mn (Nuttli 1973) from Lg waves for northwestern Iran (Rezapour 2005) and on a calibrated local magnitude, ML, for the Alborz region (Askari et al. 2009), respectively. In fact IRSC’s Mn values are calibrated with mb values from local events reported by ISC and NEIC using also teleseismic data. Thus, in some cases they are larger than the ones calculated by the original Nuttli (1973) scale (see Fig. 4 in Rezapour 2005) from Lg waves. With a standard deviation of 0.2 M and on average a magnitude difference of nearly zero for the 1344 paired events in our catalogue (Fig. 11), the observed difference in reported magnitudes by IRSC and BIN on first sight might be interpreted as insignificant for hazard studies. We note, however, that of the 10,755 events in the time period 2006 to 2017 reported by either one of the two observatories as M3+ earthquakes, only 1344 (25%) could actually be paired and only these were used for this magnitude comparison. With respect to the sensitivities and overlap of the two station networks by IRSC and BIN (Fig. 1 ), this is a surprisingly small percentage of paired events that likely denote the top quality data set. As the new M3+ catalogue is based on comparison of the two independently defined M3+ catalogues by IRSC and by BIN, differences in magnitude estimates by the two networks will also affect the success rate of the event pairing. Hence, while there seems not to exist a significant difference in magnitude estimates for the 1344 events, for the overall new seismic catalogue of more than 10,000 M3+ events within 11 years it might well be significant when used for subsequent seismic hazard studies.

Fig. 11
figure 11

Magnitude difference histogram of 1344 paired (1189 well locatable + 155 poorly locatable) events reported independently by IRSC and BIN

Full size image

In accordance with the previous catalogues by IRSC and by BIN, our new catalogue shows the major seismic activity to be observed within the Zagros mountain belt, in Central Iran and surrounding the Lut block (Fig. 10) and the source mechanisms of the major events correspond well with the mapped fault systems in these regions. We note a significant difference between the more diffuse seismicity of the Zagros collision zone and the concentrated seismicity in the Central and Eastern Iran that is composed of several tectonic blocks of variable size (Berberian and King 1981). The apparent lack of seismicity in the Lut block may be tentatively attributed to its relative rigidity (Berberian et al. 2001). The high level of seismicity along the western limits of this block is considered as a region of especially high seismic hazard (Fig. 12) (Ambraseys and Melville 1982; Berberian and Yeats 1999). Immediately westward of this high seismic hazard zone, the low level of seismicity corresponds with low GPS deformation rates (e.g. Vernant et al. 2004; Masson et al. 2007). Large earthquakes with different mechanisms (mostly strike-slip) occurred in the east and north-east part of the Lut block (Figs. 10 and 12). The pattern of seismicity in the east of the Lut block is diffuse and shows a higher level of seismicity relative to the western part of this block.

Fig. 12
figure 12

Hypocenter distribution of well locatable events in new catalogue in comparison with map of seismic hazard (adapted from Moinfar et al. 2012). a Seismicity map of 1189 and 1250 well locatable events from category A (paired, blue circles) and category B (unpaired, brown circles), respectively, superimposed on map of seismic hazard with region of special focus in Central Iran marked. Hypocenter depth distribution is shown by inset in upper right corner. b Enlargement of seismic hazard map with major fault systems in central Iran region with location of profiles A to E. c Distribution of well locatable earthquakes 2006 to 2017 from bottom to top along profile A to E. major fault lines are from Hessami et al. 2003; GF: Gowk fault; GSF: Golbaf-Sirch fault; SF: Shahdad thrust fault; RF: Rafsanjan fault; KF: Kuhbanan fault; LF: Lakarkuh fault; NF: Nayband fault; AF: Anar fault; DF: Davaran fault; JF: Jafarabad fault; PF: Posht-e Badam fault; SAF: Sarbala fault; Naif: Naini fault; TF: Tabas fault; EF: Esfandiar fault. White squares show main cities

Full size image

The elongated zone of very high seismic hazard in the middle of Central Iran is following the NW oriented Kuhbanan fault (Fig. 12a). While we observe high seismic activity in our catalogue nicely aligned with this high hazard zone, we note a significant epicentral shift of more than 80 km. The difference between the local seismicity and the high hazard zone might be due to large errors in teleseismically calculated locations used by Moinfar et al. (2012) and the result of systematic shifting of epicenters because of using global models.

To further assess the correlation of the zone of high seismic activity in our catalogue along the western side of the Lut crustal block and the mapped major fault systems in the region, we computed 5 SW-NE cross sections striking about perpendicular to the main trend of seismicity (Fig. 12b). These cross sections (Fig. 12c) indicate that most of earthquakes occurred along known faults in depth range of 5 km to 20 km in the upper crust. For example, along profile-A a cluster dipping almost vertically can be observed which is connected at the surface to Golbaf-Sirch strike-slip fault system (GSF) that appears to play a significant role in transforming the remaining shortening from the South to the North (Berberian et al. 2001). The majority of the alignments in depth sections are near vertical, consistent with dominant strike-slip motion along Golbaf-Sirch, Kuhbanan and Posht-e Badam major faults in the Central Iran that accommodate additional N-S shortening by counterclockwise rotation in the horizontal plane (Walker and Jackson 2004; Mattei et al. 2012; Walpersdorf et al. 2014).

6 Conclusions

To consistently merge the seismic catalogues published by the IRSC and by BIN networks, we applied the approach to calculate the so-called Minimum 1-D model with corresponding station delays that may serve for routine uniform high-precision earthquake location (Husen et al. 2011) and as an initial reference model for 3-D seismic tomography (Kissling et al. 1994). The usage of this model with station delays of those stations with at least 20 observations for southern Iran consistently improves precision and results in a new unified seismic catalogue for the period 2006 to 2017.

IRSC and BIN use different magnitude scales for their catalogues thus causing which causing different reported magnitudes for same events in which in most cases IRSC documents larger ones. Reporting different magnitudes by different organizations can misdirect disaster management organizations. Our cleaned and unified database could be used for calculation of a unique magnitude relation in southern Iran. To have consistency in this parameter, it is suggested to calculate a unique magnitude relation using our data set that can be used by both IRSC and BIN.