A reappraisal of active faults in central-east Iran(Kerman province)

Mohammad Talebi, Abbas Sivandi-Pour, Ghasem-Ali Ahmadi, Ehsan Noroozinejad Farsangi,?, Shamseddin Esmaeili, Mohammad-Javad Banimahdi-Dehkordi,Hamidreza Safizadeh, Mahdieh Akbarpoor, Ehsan Ebrahimi, Rabe Sharifi Rad and Manoochehr Fallah

1 International Institute of Earthquake Engineering and Seismology (IIEES), Tehran 1953714515, Iran

2 Faculty of Civil and Surveying Engineering, Graduate University of Advanced Technology, Kerman 7631885356, Iran

3 Geological and Geotechnical Department, Padidab Sepahan Consulting Engineers, Esfahan 8166815173, Iran

4 Razi University, Kermanshah 6714115111, Iran

5 Institute for Advanced Studies in Basic Sciences (IASBS), Zanjan 4513766731, Iran

6 Kerman Provincial Gas Corporation, National Iranian Gas Company, Kerman 7616859920, Iran

ABSTRACT Fault lineaments are the main input data in earthquake engineering and seismology studies. This study presents a digitally-based active fault map of the Kerman region in central-east Iran which experienced several devastating earthquakes on poorly exposed and/or not identified active faults. Using Landsat 8 data, we have carried out the image-based procedures of fault mapping,which include applying the contrast stretching technique, the principal component analysis, the color composite technique, the spectral rationing, and creating the false-color composite images. Besides, we have cross-checked the resulting map with the geological maps provided by the Geological Survey of Iran to decrease the associated uncertainties. The resulting map includes 123 fault segments, still, a part of which has been expressed in the previously compiled active-fault maps of Iran. Indeed, the new one is mapping the poorly exposed active faults, so-called secondary faults, which are able to produce strong events. These faults are primarily associated with poorly defined areas that accommodate low levels of seismicity; however, sporadic strong events are likely to occur. It has also been investigated that these kinds of faults are seismogenic and are able to produce destructive events. In total, the outcome of this study can also be jointed with seismic studies for investigating parts of the earthquake activity in central-east Iran, in particular for the fault-based approaches in impending earthquake-resistant buildings.

Keywords: central-east Iran; Kerman; active fault; satellite image processing; remote sensing.

1. Introduction

Kerman region, as the most extended political province in Iran, is situated in the central-east part of Iran between 26.5° and 32° north latitude and between 54.5° and 60°east longitude (Figure 1). The region is seismically active[e.g., it has experienced ～80 main earthquakes with moment magnitudes (MW) equal or greater than 5 in the last century] and has frequently been subjected by strong earthquakes, as exemplified by the events of 1981 SirchMW7.1 (with ～3 000 deaths; Raeesi et al., 2017), 1981 GolbaftMW6.7 (with ～3 000 deaths; Raeesi et al., 2017),1998 FandoghaMW6.6 (Mousavi-Bafrouei and Mahani,2020), 2003 BamMW6.6 (with ～31 000 deaths; Raeesi et al., 2017), 2005 ZarandMW6.4 (with ～600 deaths; Raeesi et al., 2017), 2010 RiganMW6.5 (Mousavi-Bafrouei and Mahani, 2020), 2011 RiganMW6.2 (Mousavi-Bafrouei and Mahani, 2020), and 2017 HojedkMW6.0 (Savidge et al.,2019).

Figure 1. Illustration of the limits for the main seismotectonic zones, i.e., Zagros, Makran, and Central Iran, in and around the Kerman region (black polygon) (Mirzaei et al., 1998). Arrows illustrate the almost N-S convergent rates of the Iranian collision zone (Vernant et al., 2004), and solid red lines show major active faults (Hessami et al., 2003; Oveisi et al., 2016).Circles and stars also display the epicenter locations of main instrumental and historical MW≥5 earthquakes, respectively(Mousavi-Bafrouei and Mahani, 2020).

Despite an important topic throughout past researches,central-east Iran has experienced several devastating events, like 2003 BamMW6.6, which occurred on poorly exposed and/or not identified active faults (e.g., Walker,2006; Talebian et al., 2006; Nemati and Gheitanchi, 2011;Savidge et al., 2019). Indeed, before the occurrence of these earthquakes, which are hosted by so-called secondary faults (Walpersdorf et al., 2014), it was generally supposed that such events were not probably to occur where they did (e.g., Reza et al., 2014). This implies that the current view of the distribution of active faults in this region is still incomplete, and therefore novel knowledge is necessary to study seismic hazard posed by possible planar sources properly.

In this study, we have presented a digitally-based active fault map of the region. For a small-scale area,structural features, such as faults can locally be identified based on direct field observations. Anyway, for the purpose of the regional fault delineation associated with extensive areas, the image-based procedures of fault mapping are reasonable. Many studies in literature also argued the significance of remote sensing for regionally structural monitoring (e.g., Mountrakis et al., 1998; Gomez et al., 2005; Yassaghi, 2006; Taylor and Yin A, 2009;Ramli et al., 2010; Yazdi et al., 2011; Shah et al., 2018);especially for developing countries (Bocco et al., 2001).This seems to be mainly related to financial reasons as well as the availability and applicability of satellite image data; still, the outputs may naturally include some sorts of uncertainties (e.g., Shah et al., 2018).

Anyway, in this study for interpreting the results of image processing, the satellite images have also been integrated into geological maps published by the Geological Survey of Iran (e.g., Alavi Naini et al., 1981;Soheili et al., 1981; Vahdati Daneshmand et al., 1990;Sahandi et al., 1992, 1993; Aghanabati et al., 1993a,1993b, 1993c; Samimi Mamin et al., 1982). We have investigated the relationship between mapped faults and seismicity as well as GPS velocity fields (Zarifi et al.,2014; Walpersdorf et al., 2014), as well.

The results show that the newly drawn faults are mostly associated with poorly defined areas; still, they can be considered as immature faults being able to create higher levels of stress drop compared to old faults(Manighetti et al., 2007). The history of Persian earthquakes shows that these kinds of faults can usually be seismogenic and are able to produce destructive events,when the major faults are quiescent (e.g., Talebian et al.,2004; Walpersdorf et al., 2014). The distribution of background seismicity in the study region also implies that most of the mapped faults are active.

In total, it seems that our resulting map can provide a basis for tectonics of the region (e.g., Koike et al., 1995)and be considered as a reference in future geological surveying. It can also supply a basis for seismic risk and hazard zonation assessments. Besides, it is able to join with seismicity studies as well as geophysical field works(e.g., Taylor and Yin A, 2009), so that it can improve our knowledge about the geological structures in the study region.

In the following sections, we first present a summary of the tectonic setting of the region. Then, we briefly describe the framework of the method and materials used for delineating lineaments. Finally, the delineated faults are presented and discussed using observational data(background seismicity and GPS velocity fields).

2. Tectonic setting

The Iranian plateau is tectonically affected by the interaction between Arabian, Indian, and Eurasian plates.Active tectonics in Iran is associated with the distributed shortening over seismotectonic provinces of Zagros,Alborz, Makran, Kopeh Dagh, and central Iran (e.g.,Mirzaei et al., 1998).

The Kerman region, situated in the central-east part of Iran, is tectonically influenced by active shallow crustal regions and subduction zones. The Central Iranian microcontinent is the main seismotectonic zone in the Kerman region, and the Zagros range and the Makran subduction region are around it (Figure 1). It is subdivided into some micro-blocks (Figure 2), namely Lut, Tabas, Kalmard,Posht-e-Badam, Bayaz-Bardsir, and Yazd blocks(Aghanabati, 2004), which may be due to sequential opening and closing of large oceanic systems or narrow marginal basins. These blocks have been divided by a set of large active faults, which have key roles in developing seismic patterns and deformation of the region. Indeed in this region, N-S to NW-SE-trending right-lateral strike-slip basement faults play the most important roles in accommodating the seismic deformation among the structural blocks (e.g., Berberian et al., 2001).

Figure 2. Structural zones within the Kerman region (after Aghanabati, 2004).

Although there are few historically recorded seismic activities in this region (e.g., Ambraseys and Melville,1982), most of the active faults within which have a clear expression of Quaternary fault displacements, so that the traces of the faults is visible in satellite imagery (e.g.,Walker, 2006; Walker et al., 2010). Thus, the image-based procedures of fault mapping would be beneficial for this region in order to identify regionally geological features,such as faults, their movements, and their structural relations like pull-apart effects (e.g., Berberian et al.,2001). As mentioned by Walker (2006), this procedure is especially applicable in regions like eastern Iran, where usual fieldwork and accessibility to details of structures may be difficult. Moreover, such studies, the distribution of active faults, enable us to improve significantly the estimation of earthquake hazards posed by these faults.

3. Methodology

Utilizing the framework of the digitally-based mapping has strongly been highlighted in the last decades (e.g.,Nalbant and Alptekin, 1995; Süzen and Toprak, 1998;Bocco et al., 2001; Sarp, 2005; Ramli et al., 2010; Ozturk,2008, 2010, 2011; Shah et al., 2018). Besides the geostructural aims, the digitally-based data can be used for other aspects of engineering programs. For instance,Karimzadeh et al. (2017) developed a hybrid GIS-based site condition model for the Iranian plateau by using digital elevation data as well as geological information.

In total, remote sensing can provide very useful information from available satellite images in a relatively simple, rapid, and inexpensive way. In particular, it makes a rather pragmatic approach to survey a large territory,quickly.

As the structural analysis from the satellite images has been presented in detail by the aforementioned references,we just describe a summary of the used procedure and materials.

The fault delineating procedure has been conducted,using Landsat 8 data, covering parts of the frames 159/39,159/40, 160/38, and 160/39. The used images were recorded in 2019 and characterized with a ground resolution of 30 m × 30 m.

The mapping method basically aims to define structural features which have sharply different characteristics against their neighborhoods. Anyway, satellite image processing includes several steps, leading to an increase in the quality of the structural analysis.

We have considered the enhancing procedure to be consisting of preprocessing and progressive steps, which include applying the contrast stretching technique, the principal component analysis, color composite technique,the spectral rationing, and creating the false-color composite images.

Firstly, the preprocessing stage has been conducted through geometric, atmospheric, and radiometric calibrations. Then, the linear contrast stretching technique was used to provide better visual products. In order to create false-color composite images, two groups of spectral bands were used; the bands of 752 in the visible system and the bands of 321, which reflect the near-infrared one.Moreover, we have applied 3×3 directional high pass filters (N-S, E-W, NE-SW, and NW-SE) in order to do the edge sharpening procedure via using a function so-called Sobel operator. Besides, the method is known as “rationing of spectral bands” has also been tested in both grayscale and RGB spectral modes (we note that R, G, and B in RGB stand for red, green, and blue colors, respectively).

Overall, for delineating major lineaments corresponding to main faults, it can empirically be said that the false-color composite images with the spectral bands of 752 are more suitable for regions with rock outcrops.Besides, for regions of sedimentary deposits, implying the presence of concealed faults, using the directional high pass filters with the directions of E-W and NE-SW can be preferred. Figure 3 shows, as an example, a suite of materials used for delineating lineaments.

4. Results and discussion

4.1. Delineated faults in the Kerman region

Active faults in central-east Iran (Kerman province) are shown in Figure 4 and listed in Table 1. It is not surprising that the digitally-based procedures cannot thoroughly provide the ideally structural maps, and they may need to be integrated with supplementary studies such as paleoseismology investigations. Anyway, this leads to high-cost studies and difficulty (e.g., Bocco et al., 2001;Ramli et al., 2010); hence, they seem not to be immediately applicable for wide regions, especially in developing countries. Anyway, in the current study, the geological maps provided by the Geological Survey of Iran have also been included, as they are expected to result in more accurate outputs (e.g., Ramli et al., 2010). Available published sources (e.g., Ambraseys and Melville, 1982;Berberian et al., 1984, 2000, 2001; Aghanabati, 2004;Berberian, 2005, 2014; Bonini et al., 2003; Fielding et al.,2004; Fu BH et al., 2004; Nakamura et al., 2004; Talebian et al., 2004; Walker and Jackson, 2004; Regard et al.,2005; Jackson et al., 2006; Meyer and Le Dortz, 2007; Le Dortz et al., 2009; Mohajjel, 2009; Walker et al., 2009,2010; Walker and Allen, 2012; Foroutan et al., 2014;Sistanipour and Arian, 2015; Rashidi-Boshrabadi et al.,2018) have also been reviewed in order to interpret the satellite images and updating the data.

Table 1. List of the fault segments shown in Figure 4 in accordance with their numbers (No.). In the fault mechanism columns (FM),the letters N, R, DR, DS, and SS stand for normal, reverse (or thrust), dextral reverse, dextral strike-slip, and sinistral strikeslip, respectively. Each letter marked by ‘*’ indicates that the corresponding fault mechanism is suggested by remote sensing(e.g., evidence from abnormally geomorphologic features like displacement of drainage patterns or pull-apart basins), and no particular information has been found, in this regard, on neither official geological maps nor other published references; still,the faults are somehow traceable on the geological maps.

Figure 3. Example of a suite of materials used for delineating lineaments in the area of the Nayband fault system. (a) A clip of geological quadrangle map (1:250 000 series) of Lakar-Kuh (after Alavi Naini et al., 1981). (b) The false-color composite Landsat 8 image with the spectral bands of 752. In this panel, minor lineaments and two segments of the Nayband fault are shown with thin and bold red lines, respectively. (c) The enhanced Landsat 8 image by the E-W directed high pass filter. (d) The enhanced Landsat 8 image by NE-SW directed high pass filter.

Accordingly, some of the faults in Figure 4 are delineated by considering abnormally geomorphologic features, including displacement or disruption of drainage patterns, river direction, or geological features. The Sarvestan fault system is one clear case implying right-slip shear, as explained by offset landforms with an offset of～500 m (Figure 5). The Pedva and South-Lut fault systems also have similar characteristics.

Figure 4. Main faults in central-east Iran (Kerman province). The name and mechanism associated with each fault are presented in Table 1. The interpretation of the satellite images and updating the data were based on available published sources such as Ambraseys and Melville, 1982; Berberian et al., 1984, 2000, 2001; Aghanabati, 2004; Berberian, 2005, 2014;Bonini et al., 2003; Fielding et al., 2004; Fu BH et al., 2004; Nakamura et al., 2004; Talebian et al., 2004; Walker and Jackson, 2004; Regard et al., 2005; Jackson et al., 2006; Meyer and Le Dortz, 2007; Le Dortz et al., 2009; Mohajjel, 2009;Walker et al., 2009, 2010; Walker and Allen, 2012; Foroutan et al., 2014; Sistanipour and Arian, 2015; Rashidi-Boshrabadi et al., 2018, and geological maps (from Geological Survey of Iran).

However, some of the mapped faults (e.g., Anar, Bam,Davaran, Dehshir, Gowk, Jirof, Kuhbanan, Lakarkun,Lalehzar, Nayband, Rafsanjan, Sabzevaran, Shahdad, and Shahr-e-Babak) have previously been identified as parts of the active fault maps of Iran by Hessami et al. (2003),Javadi et al. (2013), and Oveisi et al. (2016). These faults are of the most deep-seated crustal-scale basement faults,which have had significant roles in the present-day kinematics and structural geometry of the region. Indeed,these faults, known as the major faults, are assumed to load most of the strain and to slip at fast rates (Walpersdorf et al., 2014).

In terms of length, the positions for some of the previously mapped faults are overlapped with the traceable lineaments; thus, this implies that these faults can be mapped longer. The Lalehzar fault system, shown in Figure 6, is one of these cases. The Anar, Rafsanjan, and Gowk fault systems also are the other cases whose lengths seem to be longer.

Figure 5. Northward looking view of the Sarvestan fault system implying right-slip shear, as explained by offset landforms with an offset of ～500 m. Data are from Bing(2021).

Figure 6. Example of the structural interpretation for the area of the Lalehzar fault system. As the positions for this previously mapped fault (thick red line) are overlapped with the traceable lineaments (purple lines), it implies that this fault can be mapped longer. This figure is limited to the coordinate ranges ～29.35°N-29.7°N and ～56.5°E-57.4°E. Minor lineaments are also shown with thin red lines.

Figure 7. Seismic map for the studied region. (a) The previously mapped active-faults (Hessami et al., 2003), which are overlain with recent background seismicity since 2006 with MN≥3 (～ 2 900 events) from the Iranian Seismological Center.(b) The newly drawn faults overlain with Harvard centroid moment tensor (CMT) focal mechanisms (since 1977), and the velocity field data from Walpersdorf et al. (2014) and Zarifi et al. (2014).

Continued

Continued

Continued

Similar remakes have also been made about the West-Lut fault by Walpersdorf et al. (2014), investigating the present-day of kinematic in eastern Iran by using GPS data. They suggested that the West-Lut fault ranges further south of its Bam termination through the Sabzevaran fault.

The newly compiled fault map can serve wide implications for earthquake engineering studies such as seismic risk and hazard analyses (e.g., Baker, 2015),estimations of maximum credible earthquakes, and fault setback calculations for critical facilities (e.g., Batatian,2002).

Using the framework of deterministic seismic hazard assessments (Baker, 2015), we provide an example of the calculations of peak ground acceleration (PGA) under rock conditions for a location in Faryab city being one of the most populous cities, located in the seismically active part of the region (28.09°N, 57.22°E).

The nearest active fault delineated on previously compiled maps (e.g., Oveisi et al., 2016) is a segment of the Main Zagros Reverse fault, with the maximum considered magnitude of 7.5 and the minimum epicentral distance of ～35 km. The second nearest fault in this regard is the Sabzevaran strike-slip fault with a minimum distance of ～48 km. By assuming ～50% of the total fault length as the rupture length and according to Wells and Coppersmith(1994), this fault is also able to produce a magnitude ～7.5 earthquake. This implies that the former fault produces larger spectral acceleration amplitudes at the location in comparison with the latter one; thus, we neglect the Sabzevaran fault for the following assessment. Using Akkar et al. (2014) attenuation model, a median PGA of～0.16 g is estimated.

However, based on the new fault map, the nearest active fault to the considered site is the Mehr Gol fault with a minimum distance of ～9 km. According to Wells and Coppersmith (1994), this fault is capable of producing anM=6.9 event. It can also be noted that this fault experienced a strongMW=6.6 earthquake in 1999 (e.g.,Mousavi-Bafrouei and Mahani, 2020). Anyway, the estimation of seismic hazard results in a median PGA of～0.44 g, which is approximately two times more than the former estimate.

It seems that the new PGA value may be a more reasonable estimate for a location in such seismically active areas; thus, meaning the new fault map can improve the assessment of seismic hazard. It is also worth mentioning that the more distant larger-magnitude earthquake may produce larger acceleration at long fundamental periods (e.g., ＞0.8 s). Indeed, the estimations may somehow be rough; however, this does not affect the conclusion.

Overall, the fresh mapped faults are mostly neither short (＜10 km) nor long (＞100 km). According to Walpersdorf et al. (2014), this kind of fault seems to be created in response to stresses from block rotations during several million years; thus, they are somehow spatially associated with major ones. They can also be categorized as active(main) faults, based on evidence from geomorphologic features (e.g., Figure 5) and locations of past seismicity(Figure 7); still, they probably accommodate a smaller amount of strain and have a lower slip rate in comparison with the major faults. The history of earthquakes within Iran also indicates that these faults can usually be seismogenic and have produced many destructive events in Iran when the major faults were quiescent. Walpersdorf et al. (2014) highlighted some of these devastating earthquakes that occurred in central-east Iran, the strong events of 2003 BamMW6.6 is a reasonable example for this case(e.g., Talebian et al., 2004). Indeed, these faults may be classified as young or immature ones, which are able to cause higher levels of stress drop compared to old faults(Manighetti et al., 2007).

4.2. Relationship between main faults and observed field data

Figure 7 shows the schematics for the maps of the previously mapped faults and the newly drawn ones.During the instrumental period (i.e., since 1900), the study region has experienced some destructive earthquakes. For instance, the SirchMW7.3 event happened on the Sirch-Gowk fault in 1981 with ～3 000 deaths, and more recently,the devastating BamMW6.6 earthquake occurred on a high angle right-slip fault in the east part of the region(Berberian, 2005; Fu BH et al., 2004). This earthquake is the deadliest event (with ～31 000 deaths) in Iran since the 1990 RudbarMW7.4 earthquake. However, the recent background seismicity distributed in the entire region(Figure 7a) confirms that most of the faults in this region are more or less active. The statistics of seismic catalogs have also shown that earthquakes are mostly distributed in the upper part of the crustal thickness for central-east Iran(e.g., Engdahl et al., 2006).

The focal mechanisms shown in Figure 7b can imply that strike-slip faulting is more active in the study region.In particular, the focal mechanisms located at the uppermiddle part of the region are associated with northweststriking slip faults. Anyway, the events that happened in the southwest area have had a major dip-slip component; it seems to be dominantly organized by the compressive forces at the collision zone between Central Iran, Zagros,and Makran.

The newly mapped faults are associated with areas of both high and low levels of seismicity. For instance, the Rigan fault is located in the southwest region, where shows a relatively high seismic rate (Figure 7a). The southeast part of the study region is the other example of the areas with high levels of seismicity and poorly mapped faults.

Anyway, the newly drawn faults are mostly associated with partially defined areas, where accommodate low levels of seismicity. The northwest part of the region is one of these areas, over where the constancy of velocity rates(Figure 7b) can show that the deformation in this area is in a steady-state mode, and most of the structural units within which just move toward the north; still, sporadic strong events are likely to occur in this region (e.g., Foroutan et al., 2012). Based on GPS data (Walpersdorf et al., 2014),the ～NS shortening rate between the most lower and upper parts of the region is ～5 mm/a. To the east, arrows show an overall decrease in the north-directed velocity. The velocity rate is slower near the Lut block; still, this block is not completely rigid (Walpersdorf et al., 2014). For clarity,we have labeled the South and North Lut faults (numbers 117 and 92 in Figure 4, respectively); still, they are approximations and not exact borders. The Gowk-Nayband fault system (Walker and Jackson, 2002; Walker et al.,2010) can also be considered as the west limit of this block. Walpersdorf et al. (2014) also called this ～NS striking fault system the west Lut fault, divided into enechelon segments implying lateral motion on the fault system. According to Naylor et al. (1986), it seems that reactivation of basement faults can yield the pattern of the en-echelon arrangement.

As a kind of geological structure, a fault network is one of the major controlling factors in hazard assessments(e.g., Baker, 2015). These kinds of structural discontinuities, often in active seismic zones, are expected to be areas of high seismic hazard. Indeed, the probability of seismic hazard, through time, can classically be considered to be greater in areas with rather high levels of seismic activity compared to those with low rates of seismicity. However, evidence from paleoseismology studies have expressed that faults in areas with low levels of seismic activity can also give the notably consequent threat of likely forthcoming strong events, depending on the status of their seismic cycles (e.g., Foroutan et al.,2012). Thus, this can support the necessity for geostructural assessments in such regions, especially for the fault-based methodologies in impending earthquakeresistant buildings (e.g., Dehghan-Manshadi et al., 2020).

5. Summary and conclusions

The focus of this investigation was to develop the active fault map in central-east Iran (Kerman province),using an approach so-called satellite image processing. In order to draw digitally-based faults, a suite of enhancing techniques has been applied to Landsat 8 data. As remote sensing cannot provide ideally structural maps and is no replacement for field mapping, we have alternatively verified the resulting map against the geological maps provided by the Geological Survey of Iran; this can reduce the associated uncertainties (Ramli et al., 2010).

The basis of the most faults presented in this study is by no means new, while parts of which have been systematized as parts of the previously compiled activefault maps of Iran. Anyway, the new one is mapping the poorly exposed active faults, so-called secondary faults,being able to produce strong events.

The CMT focal mechanisms can imply that strike-slip faulting is more active in the study region, and the distribution of recent background seismicity approves that most of the regions of the faults are active. Besides, newly mapped faults are associated with areas of both high and low levels of seismicity; still, they are mostly located in poorly defined areas, where experienced low levels of seismicity.

As a final remark, it is believed that this kind of map can provide the basics for tectonics and future geological studies (e.g., Koike et al., 1995; Taylor and Yin A, 2009).It can take a significant impact on seismic risk and hazard zonation assessments. In particular, evidence from supplementary studies, such as paleoseismology investigations,can support the necessity for structural assessments in such a region, especially for the fault-based methodologies in impending earthquake-resistant structures.

Acknowledgments:

The authors would like to acknowledge the financial support of Kerman Provincial Gas Corporation & National Iranian Gas Company to conduct this research via Funding Sources of #062570 and #062571.