Mixed crystalline basement of Junggar basin revealed by wide-angle seismic evidence*

: A wide-angle seismic reflection / refraction survey along a ~ 600 km long transect through the Junggar basin from Emin to Qitai allows to receive several images near N-S trending blind faults, which are located at the lower part of the upper crust, the middle crust and the lower crust within the basin and cut up the Moho. These faults, with high seismic velocity and without obvious dislocation, are considered as “extensional faults” formed by north-south compression and east-west extension. These deeply rooted faults provide channels via which basic to ultra-basic materials from upper mantle migrate into the crust and mix up with the crustal material causing thin thickness, high seismic velocity, high density and high magnetic intensity after cooling in the crust of the basin.


Introduction
The Junggar Basin is a large-scale intracontinental superimposed basin in northern Xinjiang since Permian [1][2][3]. It is the most promising area for petroleum prospects in China. In the past decades, the researches on the Junggar basin have given profound results dealing with the tectonic background and the evidence for existing theories [4][5][6][7]. Although a lot of geophysical and geological studies have been carried out since the 1980s, there are still many divergences of view in basin's tectonic evolution and the basement structure.

Геология, поиски и разведка месторождений полезных ископаемых Geology, Prospecting and Exploration of Mineral Deposits
Devonian and later Middle Devonian rifting evolution of the Junggar ocean, formed from the rifting of the early Paleozoic folded basement, the relic ocean from Later Devonian to late Carboniferous, and the final accretion orogeny in the late Carboniferous-early Permian. Wu [9] considered that it was a rift in Permian, a depression in Triassic-Old Tertiary, and a contraction and uplift stage after Neogene. You [10] divided the basin into rift stage in Carboniferous-Triassic, central uplift stage in Jurassic and piedmont depression stage after Cretaceous. Zhao [11] divided the basin into fault depression stage in Permian and Triassic, depression stage in Jurassic-Old Tertiary and shrinkage-uplift stage after Neogene. Xiao et al. [12] considered that the Late Carboniferous-Early Permian was a marine foreland basin. Yang et al. [13] also classified the Early Permian as marine foreland, the Late Permian and Neogene-Quaternary as continental foreland, and the Triassic-Paleogene as oscillating continental basin. The main differences in the above-mentioned division opinions lie in the understanding of whether the nature of the basin is tensional or compressive in the Early Permian and the nature of the depression basin in the Late Permian-Old Tertiary. These differences have not narrowed, but have widened in recent years. Sun [14] advocates that we should abandon the understanding that the early basin is collapse or tension and consider Junggar as a Carboniferous-Permian foreland basin based on the Triassic-Jurassic intracontinental further subsidence, since the Cretaceous is mainly the typical pre-revival period continental basin, which unified only in the late Cenozoic. On the contrary, Chen et al. considered that Junggar Basin was a rift basin in the early stage (P), a craton basin in the middle stage (T-E), and a compression basin in the late stage (N-Q). The basin's basement composition is also a controversial topic in the international community. Some researchers proposed that the central part of the Junggar basin is composed of a conjectural craton nucleus [1,15]. Some others interpreted the basin as a trapped oceanic lithosphere [4], a relict back-arc basin [16] or deformed Altai Paleozoic rocks [17].
According to previous geophysical research in northwestern China, the average crustal thickness of the Junggar basin is about 50 km, while those of the Tarim basin and the Qaidam basin are ~ 55 km [18,19] and ~ 60 km [20], respectively. The average P-wave velocity is 6.3 km/s in the Junggar basin [21], about 6.0 km/s in the Tarim basin [19], and about 5.8 km/s in the Qaidam basin [20]. Therefore, the Junggar basin has a thinner crustal thickness and a higher average crustal velocity than the Tarim and Qaidam basins. The formation mechanism for such characteristics, as well as the high geomagnetic anomalies distributed in the central part of the basin, is not well understood.
Recently, a comprehensive geophysical survey along a ~ 600 km long transect through the Junggar basin from Emin to Qitai was carried out (Fig. 1). Along the profile, a wide-angle seismic reflection/refraction was carried out to obtain 2D velocity structure of the crust and upper most mantle. Based on the 2D velocity structure as evidence for designing the starting model of the 2D density structure and 2D magnetic intensity structure and constrained by gravitational and geomagnetic anomalies along the profile, the density and magnetic intensity structure were obtained. Finally, we proposed a comprehensive geological interpretation.

Seismic data and processing
Recently, a seismic reflection/refraction profile as part of the ANTILOPE project was carried out to constrain the seismic velocity structure and physical properties of the crust and upper most mantle (Fig. 1). The profile crosses the Junggar basin from Emin in the northwest to Qitai in the southeast margins of the basin. The geometry of the profile was designed to optimize recording of refractions at near-critical distance and to provide velocity control for the near vertical reflection experiments, particularly at Moho depths to image the topography of the Moho. Along the profile, wide angle seismic reflection/refraction with 8 chemical explosions (2 tons of TNT for each) has been carried out, and 204 seismic receivers of DAS-1 & 2 models have been used to record the signals (Fig. 2). All receivers were positioned by GPS, with a station spacing of 1.5~3.0 km, depending on the topographic and tectonic complexity of the region. The acquisition system shown in Fig. 2  provide us with needed coverage and seismic illumination to study the structure and properties of the basement, the structure of the crust and uppermost mantle, and some deep faults particularly those cutting up the crust and controlling the tectonic evolution of the basin. The seismic records along the profile are shown in Fig. 3. The ray tracing and theoretic seismogram are shown in Fig. 4. The effective ray coverage is shown in Fig. 5.  Identifying seismic phases is a key step in processing the seismic data. Due to relatively sparse shot and receiver spacing (compared with near vertical reflection), phase identification was combined with wave group (energy parcel) identification to increase accuracy. Five criteria for phase picking are listed as follows:

Fig. 5. The example of seismic recorders (a) and the 2D velocity structure of the crust and uppermost mantle (b) from Emin to Qitai: aseismic signals from SP Emin The vertical axis indicates reduced travel time by T-x/6 in second and the horizontal axis is the distance in km from the shotpoint. The seismic recorders show strong Moho reflection along the profile but no seismic reflection
from the Moho can be found at the two faults, F1 and F2, indicating that there is no wave-resistance interface at and nearby these two faults. These faults, with high seismic velocity and without obvious dislocation, are considered as the "extensional" faults, formed by north-south compression and east-west extension; b -2D velocity structure of the crust and uppermost mantle The vertical axis is depth in km and the horizontal axis is distance in km. It was obtained by ray tracing and theoretical seismograms for each shot point with an improved SEIS83 program under the condition of 2D lateral inhomogeneous medium Рис. 5

Geology, Prospecting and Exploration of Mineral Deposits
2) The amplitude of the wave must be greater than that of the noise; 3) The apparent velocity for a phase must be within a reasonable range; 4) The picked continuous travel-time curve must be greater than a minimum length; 5) Seismic reciprocity must hold for the complete observational system when we interchange the source and receiver positions.
Based on these criteria, we mainly identified the following seismic phases: Pg, P1, Pc, Pm and Pn.
(1) Pg Wave. Pg is a refracted wave that travels in the top part of the crystalline basement. The slope of its traveltime moveout curve yields precisely the P wave velocity in the basement. Generally, Pg can be traced to more than 130 km away from the shot point. Some significant differences in the reduced travel time for each shot point have been observed due to variations in the depth of the crystalline basement. In the Zayier Shan at the western end of the profile and Bogda Shan at the eastern end of the profile where the crystalline basements are almost exposed to the surface, the Pg wave appears much earlier (Fig. 3). However, in the middle segment of our survey line, the sediment is thick and Pg arrives much later. The apparent velocity of the Pg changes significantly at the two ends of the profile especially the transitional zones from basin to range, due to the undulation of the top surface of the crystalline basement, and changes little in the inner basin, showing that the top surface of the crystalline basement of the basin is flat. The apparent velocity of Pg is about 6.0-6.2 km/s.
(2) P1 Wave. The P1 wave is the reflected wave from an interface in the upper crust at a depth of 21.0-28.0 km. The amplitude and continuity of this wave changed significantly along the profile. Generally, the P1 wave is weak in energy and poor in continuity. The average velocity of the P1 wave is 5.8-6.1 km/s.
(3) Pc Wave. Pc is the reflected wave from the interface in the middle crust, at a depth of 36.0-40.0 km. In the Zayier Shan and the Bogda Shan, the Pc waves are weak and can be traced out to about only 40 km. In the Junggar basin, the Pc waves are relatively strong and have a good continuity. The Pc waves from Zayier Shan and Bogda Shan are different. The Pc waves from Zayier Shan (western branch of SP Karamay) are relatively strong in energy and good in continuity. It can be traced as far as 110 km, while that from the Bogda Shan is weak and poor in continuity. The variation of the amplitude and differences in average velocity are related to geological structure and material properties of each tectonic unit, distinct from the depth of the crystalline basement and undulation of the interfaces. The average velocity determined for P2 is about 6.0-6.3 km/s.
(4) Pm Wave. Pm is the reflected wave from the Moho discontinuity. Based on seismic records, the Pm is characterized by its clear appearance, strong amplitude, good continuity, and far-distance tracing. The Pm wave, however, has different waveforms, amplitudes and travel-time curve patterns in different sections, which suggest the existence of lateral crustal heterogeneities in different tectonic units. However, the Pm waves in the sections between 150-185 km and 220-270 km are relatively strong. This indicates that the Moho is undulated. In the basin, the Pm waves from SP Manas, SP Hutub, SP Changji and SP Fukang (western branch) have strong amplitude, excellent continuity and longer traceable distance, suggesting that the Moho of the Junggar basin is highly reflective and sharp. The average velocity of Pm waves is 6.2-6.7 km/s. However, we did not obtain a clear Pm wave from the Bogda Shan.
Even though the seismic Pm wave from the Moho in the Junggar basin is strong in energy and good in continuity, there are some sections where no clear Moho reflections have been observed (i.e., the reflectivity is very poor).
(5) Pn Wave. Pn is the refracted wave from the top part of the upper mantle. It can be used to determine the velocity of the uppermost mantle. Furthermore, it can be used to determine the Pm wave according to the geometric relationship between the two. Pn waves have been recorded at SP Emin (eastern branch), SP Karamay (eastern branch), SP Manas (eastern branch), SP Hutub (eastern branch) and SP Fukang (western branch) (Fig. 3). The Pn wave arrives at a source-receiver distance of approximately 200 km, with a velocity of about 8.0 km/s. Additional wave groups P1′and Pc′were recorded and discriminated in the eastern edges Геология, поиски и разведка месторождений полезных ископаемых Geology, Prospecting and Exploration of Mineral Deposits of the Zayier Shan (see SP Karamay and SP Manas) and western edge of the Bogda Shan (see SP Changji, SP Miquan and SP Fukang), indicating the existence of some additional lithologic interfaces (C1′ and Cc′) in these regions (in this notation, P1 and Pc are reflected waves from interfaces C1 and Cc in the crust). P1′ and Pc′ waves were not observed in the Junggar basin. This suggests that the crust of the Zayier Shan and Bogda Shan is more complicated than previously assumed, and provides some evidence for mantle intrusion. Reflected and refracted waves (Pg, P1, P2, P3, Pm and Pn) have been identified on different shot records and corrected from shot to shot (Fig. 3). These phases featured by high amplitudes provide a basis for dividing the crust into three velocity layers, the upper, middle and the lower crust. The first arrives at offsets of less than 70 km imply lateral variations in velocities at shallow depths (5.0 to 6.0 km/s) which can be correlated with the surface geology ( Fig. 1, b). The velocity is estimated to be 6.2 to 6.5 km/s in the upper crust (depth of 22 to 27 km), about 6.6 to 6.7 km/s within the middle crust (~ 23 to 37 km), and 6.8 to 7.4 km/s in the lower crust (from 30 to 52 km down to the Moho). An increase in the average velocity throughout the crustal column within the crust along the profile is interpreted as an evidence for the increase in mafic content at upper crustal levels and metamorphic grade in the middle to lower crust [18,19]. The proposed layered velocity model is non-unique but generates theoretical travel time branches that are in agree (to within ±0.1 s) with the interpreted arrivals for all of the sections.

Velocity structure of the crust and upper most mantle
Under the condition of 2D lateral inhomogeneous medium, we conducted ray tracing and constructed theoretical seismograms for each shot point with an improved SEIS83 program [22], which are presented in Fig. 4, and obtained the 2D velocity structure (Fig. 5, b). According to the effective ray coverage (Fig. 6), we have obtained not only the detailed velocity structure of the crust, but also that of the uppermost mantle (see the area within the blue dotted line in Fig. 5, b).

Fig. 6. Seismic ray coverage of the Emin -Qitai profile
The ray coverage of the transect is made by combining record sections of all shot points along the profile with the respective parts of the ray tracing of the corresponding shot points and deleting all calculated arrivals in the modeling which are not observed in the data Рис. 6

. Покрытие сейсмическими лучами профиля Эминь -Цитай
Покрытие поперечного разреза лучами производится путем объединения участков записи всех очагов сейсмического взрыва вдоль профиля с соответствующими частями трассировки лучей соответствующих очагов взрыва и удалением при моделировании всех вычисленных приходов волны, которые не наблюдаются в данных Density and geomagnetic structures Gravity and aeromagnetic data. To constrain the density structure, we need some information about the gravity. Based on the 1:200000 Bouguer gravity anomaly map of Xinjiang (Fig. 7), there are two gravity gradient zones in the northern and southern margins of the Junggar basin, which clearly delineate the shape of the basin. This is in agreement with the landscape and geological framework. The rock magnetic strength is another important factor one can use to distinguish rock types, together with the P-wave velocity and density. The aeromagnetic anomaly map of the Junggar basin shows that the southern part of the basin (Manas terrain) has a stronger magnetism than the rest of the basin.
Density and geomagnetic structures. Based on the Bouguer anomaly and aero-magnetic anomaly maps (1:200,000 scale), the 2D P wave velocity structure, and the faults investigated at the surface of the transect from Emil to Qitai, we constructed a model for the density structure. The starting model for the density in each cell is given by the following velocitydensity relationship [23]: Finally, the 2D density structure (Fig. 8, a) and magnetic intensity structure (Fig. 8, b) were obtained according to the Syn-Source theory and through joint inversion of gravity with geomagnetism and constrained by gravitational and geomagnetic anomalies along the profile (Fig. 7) [24].
Inference on the rock properties of the Junggar basin According to relationships among the velocity, density and rock type (Fig. 9) [25], and based on the velocity distribution along the profile from Xayar to Burjing [21] and from Baicheng to Da Qaidam [19], basement rock types of the Junggar basin were identified and catalogued (Table).
The velocity is obtained from seismic data generated by artificial earthquakes, and the density and geomagnetic intensity are determined by joint inversion of gravity with geomagnetism along the same profile. The results suggest that the basement lithology of the Junggar basin is complicated especially towards its center. It is composed of mafic, ultra-mafic, and acidic rocks, such as granite, schist, granulite, gabbro.
Moreover, the Pb [23, 26,27], Nd, and Sr [28] isotope compositions of post-collisional intermediate and acidic plutons in the Junggar basin show that these plutons originated from Fig. 7. The map showing geology (a), geomagnetism (b) and gravitation (c)

Fig. 8. The density (a) and magnetic intensity (b) of the crust and upper most mantle along the profile from Emin to Qitai The initial density model was designed based on the relation between velocity and density. The density and magnetic intensity were obtained by joint inversion of the gravity and geomagnetic anomaly based on the Syn-Source principle of gravity and geomagnetism. The upper panel of the a is Bouguer's anomaly measured and calculated by fitting. The lower panel is the density in g/cm 3 . It was obtained by forward modeling under a same frame obtained by velocity structure for both the density and magnetic intensity. The upper panel of the b is magnetic anomaly measured and calculated by fitting. The lower panel is the magnetic intensity in 0.01·A/m obtained by the same way as the density modeling. For the detail description of the joint inversion, please see [24]
Рис. 8

. Плотность (а) и магнитная напряженность (b) коры и верхней части мантии по профилю от Эминя до Цитая
Первоначальная модель плотности была разработана на основе соотношения между скоростью и плотностью. Плотность и магнитная интенсивность были получены совместной инверсией гравитационной и геомагнитной аномалии на основе принципа Syn-Source гравитации и геомагнетизма. В верхней части рис. a представлена аномалия Буге, измеренная и рассчитанная путем аппроксимации. Нижняя часть рис. а показывает плотность, г/см 3 . Данные были получены прямым моделированием в той же системе отсчета, полученной с помощью структуры скорости, как для плотности, так и для магнитной напряженности. Верхняя часть рис. b демонстрирует магнитную аномалию, измеренную и рассчитанную путем аппроксимации. Нижняя часть рис. b показывает магнитную напряженность, 0.01·А/m, полученную тем же способом, что и при моделировании плотности. Подробное описание совместной инверсии см. в [24] 21 Geology, Prospecting and Exploration of Mineral Deposits the sub-oceanic mantle and not from a craton continental basement. No known Precambrian strata have yet been discovered around the basin. However, Paleozoic oceanic rocks and relic oceanic sediments of the middle Paleozoic age are widely distributed. To the south of the Karamay fault the paleocurrent is directed to the north as shown by imbricated clasts and groove marks in turbidities of continental shelf facies from the Silurian age. This also suggests that the provenance of these strata were pre-Silurian continental blocks. Furthermore, the provenances were probably not the small blocks made up of mafic volcanic rocks that date to the Ordovician or Cambrian ages [12]. The aeromagnetic anomaly of the Junggar basin shows that the southern part of the basin (Manas terrane) has much stronger magnetism, which is in agree with the magnetization of Precambrian schist, migmatites, granites, and hornblende schist. However, the results of joint inversion of gravity with aeromagnetism show that these high aeromagnetic anomalies may be produced by ancient blocks or the NNE trending faults resulting from modern north-south compression. These faults potentially provide channels for deep material to move up into the crust, and subsequently intrude laterally into different sub-layers of crust. This would change the contents of the Junggar basement, resulting in the higher average velocity, density, and magnetism observed in the crust of the basin. These intrusions may have originated as oceanic basalt and tholeiite. Based on the 1:200000 gravity Геология, поиски и разведка месторождений полезных ископаемых Geology, Prospecting and Exploration of Mineral Deposits anomaly map of Xinjiang [29], there are two gravity gradient zones in the northern and southern margins of the Junggar basin, which clearly delineate the shape of basin. This is in agreement with the landscape and geological framework. The 2-D velocity and density structures along the Xayar-Burjing profile indicate that the crustal thickness is about 47 km in the Junggar basin. This thickness however, includes the 15 km thick sediments that have accumulated since the Devonian era. The thickness of the Mesozoic and Cenozoic sediments alone is greater than 11 km. Not taking into account the sediments, the crustal thickness of the crystalline basement of the Junggar basin is only 32 km. The 2-D density structure of the Xayar-Burjing profile [21] shows that the density of the basement roof of the Junggar basin is different in the Wulungu and Manasi terrains; the density and velocity of the former is 2.87 g/cm 3 and 5.9 km/s, while that of the latter is about 2.85 g/cm 3 and 5.8 km/s.
Vertically, the crust is composed of three layers: the upper, middle and lower crust, bottomed by the C1, C2 and Moho interfaces, respectively. The upper layer of the upper crust consists the deposited cover bottomed by the interface G, and the lower layer bottomed by the interface C1. The sedimentary cover ranges in thickness from 1 km to 12 km and reaches a maximum depth of 12 km, while the crystalline basement ranges in thickness from approximately 12 km to 20 km and reaches a maximum depth of ~ 30 km. The velocity in the deposited cover undergoes large lateral variations, especially near Karamay, Hutub and Fukang (Fig. 5). In addition, the density and the magnetic intensity also change greatly (Fig. 8). The crystalline basement exhibits high velocity zones, including a region near Hutub, at ~ 20 km depth, where velocities are 0.1~0.2 km/s higher than sur-rounding medium, and so as the density and geomagnetic intensity.
The middle crust, defined by the interfaces C1 and C2, is comparatively thin, with thicknesses ranging from 4 km to 15 km and a maximum depth of 37 km. Slight variation occurs on velocity, except for a low velocity zone beneath Karamay and Fukang, where the corresponding density is low. The middle crust has a dumbbell shape, i.e. thin in the middle and thick at the two ends of the profile.
Obvious lateral variations are obtained in the lower crustal velocity. The most complex regions lay beneath the Karamay and Fukang, which are two active tectonic zones in the Junggar basin 1 . Here, two low velocity zones with low densities have developed, near which developed several faults cutting up the crust. The lower crust is further characterized by uniform thickness of roughly 15 km, despite having heterogeneous composition [21]. The crustal thickness of the Junggar basin ranges from 46 to 56 km.
Below the Moho interface, the uppermost mantle of the Junggar basin varies in shape of the velocity contours. The contours at the uppermost mantle are convex, indicating that the velocity is higher at the center of the basin than that at its either side [30]. It may further suggest that the upper mantle material may inject into the crust of the Junggar basin, because of the similar rock distribution along the profile (Fig. 10, Table). The Earth's topography at short wavelengths results from active tectonic processes, whereas at long wavelengths it is largely determined by isostatic adjustment for the density and thickness of the crust [30].
On these interfaces, we found several sections from which the seismic waves are very weak, indicating a lower reflection coefficient (Fig. 3). This may reflect the presence of deep faults, which could be confirmed by the combined results of seismic velocity (wide angle reflection/refraction profiling, (Fig. 5), density, geomagnetic intensity (joint inversion of gravity with aeromagnetism (Fig. 8), P-wave imaging 2 and seismic converted wave profile crossing the northern part of the basin [24]. All these faults, with little reflectivity and no obvious dislocation, __________________________________________ 1 Xu Y. Seismic tomography for northwestern China and its tectonic implication: Ph.D dissertation. Institute of Geophysics, CAS, 1999. 2 Ibid. are considered as "extension" faults. They developed with a high angle in the middle and lower crust and extended upwards into the lower parts of upper crust. They were formed probably by the collision of the Junggar basin with the Tarim basin at the end of Permian [21].
Horizontally, the southern Junggar basin along the profile can be divided into three sections, which are the western section (0-150 km), the central section (150-420 km) and the eastern section (420-600 km), based on the velocity, the density and geomagnetic structures (Fig. 8).
In the western section, i.e. the western mountain area, the tectonic activities are more active, where developed many faults. The velocity distribution is complicated with several low velocity zones formed in the middle crust and lower crust, respectively. The densities in the low velocity zone are lower, but the magnetic intensities are complex, featured by high frequency of geomagnetic anomalies. In the central section, mostly the Junggar basin, the velocity, the density and geomagnetic intensity are consistently high, especially in the lower part of the upper crust and the middle crust. In this section, the Moho interface sinks down and the uppermost mantle uplifted. In the eastern part of our profile, the velocity, the density and geomagnetic intensity are relatively low at the eastern margin of the basin, with two lower velocity zones developed in the middle and lower crust, respectively.
Geologically, the evolution of the Tianshan Orogen and Junggar basin can be divided into several stages: the early Paleozoic intra-oceanic arc evolution, the Early Devonian and later Middle Devonian rifting evolution of the Junggar ocean, formed from the rifting of the early Paleozoic folded basement, the relic ocean from Later Devonian to late Carboniferous, and the final accretion orogeny in the late Carboniferous-early Permian [8]. We infer that this final accretion orogeny in the late Carboniferousearly Permian was so strong that some deep faults with S-N direction were formed by S-N compression and E-W extension. In such process, the hot materials from upper mantle migrated to the crust through the deep faults and mixed up with the crustal materials in the central basin, which could be further approved by the high velocity, high density and high geomagnetic intensity beneath the basin. The two low velocity zones at both sides of the basin may be attributed to the heat from upper mantle.
The exchange of the materials or energy of the uppermost mantle with those in the crust is manifested by the heat evolution of the basin. Based on the heat flow variations of the basins in northwestern China since Paleozoic time [31], the heat flow of the Junggar basin since Carboniferous-Permian reduced gradually from 120~150 mW.m -2 to 52 mW.m -2 at present, while that of the Tarim basin, the Qaidam basin and the Sichuan basin almost keep the same value of 45 mW.m -2 , even with a tendency of increase. Therefore, we infer that the high heat value of the Junggar basin may have been caused by the hot materials from the upper mantle mixing the crystalline basement of the Junggar basin.