- © 2014 by the Seismological Society of America
A recent paper in the March/April 2013 issue of Seismological Research Letters, “A Unified Seismic Catalog for the Iranian Plateau (1900–2011)” by Mohammad P. Shahvar, Mehdi Zare, and Silvia Castellaro (henceforth referred to as Shahvar et al., 2013) provides a unified homogeneous earthquake catalog, for the period 1900–2011, particularly for seismic‐hazard assessment studies in Iran. They subdivide the territory of the Iranian plateau into two domains only according to seismicity to derive empirical relationships to convert the original magnitudes (Ms, mb, ML, and MN) to a uniform scale (Mw), applying standard least squares (SR) and inverted least squares (ISR) regression methods and orthogonal regression method (OR). Some contradictions in Shahvar et al. (2013) encouraged us to declare a few points.
The distinction between seismotectonic zones is based on all available geologic, tectonic, geophysical, and seismological data. In other words, a seismotectonic zone is considered an area that under present‐day geodynamic regimes has an internally consistent tectonic setting and unified seismicity pattern (Ye et al., 1993, 1995). Shahvar et al. (2013) claim that the Iranian plateau can de facto be divided into two domains only because seismicity in the Alborz and Central Iran zones show statistically undistinguishable features. In contrast, Karimiparidari et al. (2013) distinguish six seismotectonic zones based on Mirzaei et al. (1998) and Tavakoli (1996). It is noteworthy that the second author of the two concurrent inconsistent papers is in common.
Based on Nowroozi (1971) and Bird et al. (1975), it is stated by Shahvar et al. (2013) that the continental Arabian shield subducts beneath the Zagros belt. However, local seismograph networks have found no reliable depths in the Zagros deeper than ∼20 km, even in places where deeper earthquakes were reported to have occurred (e.g., Niazi et al., 1978; Tatar et al., 2004). According to Jackson and Fitch (1979), all apparently deep earthquakes in Zagros are small and badly recorded with poor depth resolution. There is no reliable evidence for subcrustal events, and no tendency for earthquakes to deepen toward the Main Zagros reverse fault (Ni and Barazangi, 1986). This implies that the earthquake occurrences in Zagros are not, presently, primarily related to the northeast‐dipping thrust zone itself, which experienced its major activity (subduction) in Late Cretaceous and Early Paleogene time (Dewey and Grantz, 1973). According to Engdahl et al. (2006), depths for many events in International Seismological Centre (ISC) and U.S. Geological Survey (USGS) National Earthquake Information Center (NEIC) global catalogs, until only recently based entirely on first‐arrival times of P waves, are often not determined with sufficient accuracy for many geological and tectonic purposes.
Figure 2 in Shahvar et al. (2013) shows some nonenclosed areas with dashed lines as borders of the three seismotectonic zones considered. The exact regions of seismotectonic zones are not clear for the reader. Even more importantly, the width of the Zagros mountain belt (zone I) is far away from what the authors quote from Hatzfeld et al. (2003). It is well known that the Main Zagros reverse fault (MZRF) separates the Zagros mountain belt from Central Iran, which is a major structural discontinuity (Stocklin, 1968; Berberian, 1995). The authors should demonstrate why they extend the width of zone I about two times the width of the Zagros fold belt into Central Iran. In addition, the statement, three main seismogenic zones are identified (fig. 2), is misleading; the authors probably mean seismotectonic zones.
In contrast to what Shahvar et al. (2013) state that “displacement occurs mostly through strike‐slip mechanisms parallel to the axis of the Zagros range”; the Zagros belt is a broad zone of continuing compressional deformation that experiences horizontal shortening of the basement on reactivated normal faults that stretched and thinned the basement of a continental margin on which the Mesozoic sedimentary cover was deposited (Jackson, 1980; Berberian, 1981, 1995; Jackson et al., 1981; Jackson and Fitch, 1981; Molnar and Chen, 1982). This idea is supported by the fault‐plane solutions of earthquakes in the Zagros, which consistently show thrusting with comparatively high‐angle (40°∼50°) fault planes distributed across the whole width of the belt (Jackson et al., 1981). The slip vectors of earthquakes in Zagros indicate consistent motion at an azimuth of north to northeast (30°∼40°; Jackson and McKenzie, 1984). Strike‐slip mechanisms mostly appear along different segments of the main recent fault with a prominent northwest‐trending right‐lateral strike‐slip mechanism along the northeast margin of the Zagros and Kazerun–Borazjan approximately north–south‐trending right‐lateral strike‐slip fault system.
Shahvar et al. (2013) state that “focal depths range between 7 and 250 km in Zagros, between 10 and 150 km in the Alborz, and between 7 and 120 km in Central Iran.” This statement is in contrast with focal depth studies in Iran. Indeed, quality of instrumental location depend on different factors such as velocity model, station distribution, number and quality of seismic phases, as well as the location procedure (Ahjos and Uski, 1992). Although with the advent of the World‐Wide Standard Seismographic Network (WWSSN) in the early 1960s and of digital networks in the late 1970s, the quality of location determinations have increased; nevertheless, blind acceptance of the locations determined by international agencies, and even national and local networks, can lead to erroneous inferences. In exception of the 1970.11.09 mb 5.4 event, with a depth of 106 km (USGS), 107 km (Kadinsky‐Cade and Barazangi, 1982), in the Zagros–Makran convergence zone, there is no reliable evidence of intermediate focal depth earthquakes in Iran. Evidences of subcrustal focal depths are pointed out by (Jackson and McKenzie, 1984) for the earthquakes of 2 August 1968 (mb 5.7, h 65 km, ISC), and 17 November 1972 (mb 5.2, h 79 km, ISC) near the Bazman–Taftan–Soltan active volcanic arc in eastern Iran and western Pakistan, which is considered to be associated with consumption of the Arabian plate underneath these regions (e.g., Byrne et al., 1992). The most recent major earthquake of 16 April 2013 (Mw 7.5, h 92, Iranian Seismological Center [IRSC]; Mw 7.7, h 82, USGS; Mw 7.7, h 52, Global Centroid Moment Tensor [Global CMT] catalog) in the same place confirms this consideration. In spite of the known inaccuracies in focal depths, such events form a set, apparently existing only north of 27° N and distinct from the shallow shocks to the south (Jackson and McKenzie, 1984). Although concentration of intermediate‐depth earthquakes (between 70∼300 km) can be interspersed among the shallow earthquakes in continental–continental collisions (Reiter, 1990), seismicity in the Zagros appears to be shallow despite some assertions in the contrary (Molnar and Chen, 1982). Reliable waveform modelings (Jackson and Fitch, 1981; Kadinsky‐Cade and Barazangi, 1982; Jackson and McKenzie, 1984; Ni and Barazangi, 1986; Baker et al., 1993) show that large earthquakes in the Zagros usually nucleate in the upper 8–15 km of the Earth’s crust (seismogenic zone). It is supported by the careful microearthquake studies (Savage et al., 1977; Von Dollen et al., 1977; Niazi et al., 1978) which all failed to reveal activity deeper than 20 km in the Zagros. In other parts of Central–East Iran, all the earthquakes are shallow crustal events with focal depths seldom exceeding 30 km (Shoja‐Taheri and Niazi, 1981). Results of waveform modeling and microearthquake studies show that the majority of earthquakes in Central–East Iran nucleate at depths of 8∼20 km. Meanwhile, a coherent and consistent picture of how focal depth distributions vary geographically within Iran is demonstrated by Engdahl et al. (2006), who relocated Iranian earthquakes occurred between 1918 and 2004. The results confirmed that earthquakes that are definitely in the mantle are either certainly (in the Makran) or probably (in the central Caspian) within lithosphere of oceanic origin (Engdahl et al., 2006).
THE EARTHQUAKE DATABASE AND MAGNITUDE CONVERSION
Unacceptable inefficiencies in data collection by Shahvar et al. (2013) make their catalog inappropriate and even misleading for seismicity study. For instance, event number 1420 is reported with Mw 5.8 instead of 5.3; or event number 2031 is reported with Mw 4.6 instead of 5.1. Also, Table 1 contains some events with parameters that do not correspond with those from the referenced agencies.
Shahvar et al. (2013) stated that “two national broadband seismological networks are presently operating in Iran (International Institute of Earthquake Engineering and Seismology [IIEES] and Iranian Seismological Center [IRSC]… MN, a magnitude scale based on Lg wave (Nuttli, 1973), is provided by IRSC (Rezapour, 2005). ML has instead been provided by IIEES since 2000.” In contrast, Karimiparidari et al. (2013) have stated that IRSC, operating with 74 seismographs, is the largest seismic network in Iran, was established in 1957, and provides data online from 2006 to the present. Karimiparidari et al. (2013) also stated the events which are reported by ISC since 1960 should be a dataset provided by IRSC. It is noteworthy that the seismographic station in Tehran, the first to be installed in Iran, came into operation in 1958 by Institute of Geophysics of the University of Tehran, and was followed a year later by the station in Shiraz. In 1964∼1965, the stations in Tabriz, Mashhad, and Kermanshah came into operation; Shiraz, Tabriz, and Mashhad stations were equipped with instruments of the WWSSN. In 1976, the Iranian Long Period Array (ILPA; Akasheh et al., 1976), was established in the region of Saveh, southwest of Tehran, consisting of seven three‐component wideband borehole seismometers in a circular shape with six instruments forming a circle with a diameter of about 60 km and the seventh located in the center. In 1995, the Iranian seismic network was founded. At present, the network consists of 99 stations grouped in 18 subnetworks, equipped with three‐component short‐period and broadband seismometers (see http://irsc.ut.ac.ir, last accessed May 2013).
Shahvar et al. (2013) stated that “MN values, reported by IRSC since 2004…”, whereas in IRSC, the MN magnitudes of earthquakes were reported since 1996 rather than 2004. In order to determine the magnitude of MN, from 1996 to 2006, the original relation (Nuttli, 1973) has been used and afterwards, a modified relation (Rezapour, 2005) is being used. In Shahvar et al. (2013) the relations ML–MN and Mw–MN have been calculated regardless of any change in the MN relation. This does not appear true due to some inconsistency in the values.
In the section for New Magnitude Conversion Relations, in the first paragraph it is stated that standard least squares regression (SR) is still the most‐used method to derive magnitude conversion relations (Mirzaei et al., 1997, for Iran; Bindi et al., 2005, for Italy; and Yadav et al., 2009, for India). Let us recall that the process of estimating earthquake magnitude using different magnitude scales is subject to measurement errors. Standard linear regression (SR) assumes that the independent variable is either error free or the order of its error is very small compared with the measurement error of the dependent variable. Hence, it is obvious that use of SR is not appropriate for magnitude conversion as both magnitude estimates contain errors. General orthogonal regression (GOR) relation takes into account the errors on both the magnitude types (Madansky, 1959; Kendall and Stuart, 1979; Fuller, 1987; Carroll and Ruppert, 1996; Castellaro et al., 2006; Thingbaijam et al., 2008; Ristau, 2009; Das et al., 2011, 2012). GOR is obtained using minimization of the squares of the orthogonal distances to the best‐fit line, whereas SR is derived by minimizing the squares of the vertical offsets (Das et al., 2013). It is clearly stated in Mirzaei et al. (1997) that Ms–mb relationships are based on Wyss and Habermann (1982), which refers to the General Orthogonal Regression of Madansky (1959); it is contrary to what Shahvar et al. (2013) referred to as standard least squares regression.
Relations between magnitude scales mb and Ms depend on several parameters, such as focal depth, focal mechanism, and earth structure (Wyss and Habermann, 1982), and varies significantly from one earthquake region to another (Bath, 1981). The identical magnitude conversion relations for completely different seismotectonic regions is not an adequate reason to let them to be merged together.
Shahvar et al. (2013) derived the magnitude conversion relation Ms–Mw for Ms≥3.6, on the other hand this relation is used later to convert Ms≥2.7 to Mw. In addition, it is not clear that how magnitude uncertainties are assessed to the unified Mw. The total magnitude uncertainty can be defined as , in which σoriginal is the original scale uncertainty and σconversion is the uncertainty after any conversion from one scale to another. The uncertainties reported by Shahvar et al. (2013) do not comply with a rule, for instance, the uncertainty for mb 4.0 and mb 6.1 is considered as 0.3 magnitude units, equally.
We clarified some shortcomings and erroneous statements appeared in a newly published article by Shahvar et al. (2013) in the March/April 2013 issue of Seismological Research Letters. As frequently notified by researchers, compilation of a uniform catalog of earthquakes is of great importance, which fulfills the condition of homogeneity of the basic data. The question of completeness is paramount; nothing of value can be estimated from a statistical analysis with significant amounts of the data missing. This can be the chief reason for the notion that the completeness and reliability of data are the basis of earthquake research. These data, then, will be used for the study of seismicity, for learning general laws of earthquake occurrence with the aim of protecting human society from the disastrous effects of earthquakes. The orthogonal regression applied for magnitude conversion in Shahvar et al. (2013) is a commonly used procedure the limitations and problems of which are discussed by some authors (e.g., Castellaro and Bormann, 2007). We believe that Shahvar et al. (2013) benefited from the results of Castellaro et al. (2006) and Castellaro and Bormann (2007) for magnitude conversion methods; meanwhile, it suffers from shortcomings in data collection, study of completeness, and uncertainty of magnitudes. Our clarification shows that the content of the work of Shahvar et al. (2013) needs revisions to be implemented for seismicity and seismic‐hazard assessment in Iran.