Destruction of the North China Craton: a perspective based on receiver function analysis

The North China Craton (NCC), which is composed of the eastern NCC and the western NCC sutured by the Palaeoproterozoic Trans-North China Orogen, is one of the oldest continental nuclei in the world and the largest cratonic block in China. The eastern NCC is widely known for its significant lithospheric thinning and destruction during the Late Mesozoic. Models on the destruction of the eastern NCC can be principally grouped into two: (1) thermal/mechanical and/or chemical erosion, and (2) lower crustal and (or) lithospheric delamination. The erosion model suggests that the NCC lithospheric thinning resulted from chemical and/or mechanical interactions of lithospheric mantle with melts or hydrous fluids derived from the asthenosphere, whereas the delamination model proposes lithospheric destruction through foundering of eclogitic lower crust together with lithospheric mantle into the underlying convecting mantle. However, those models lack seismic evidence to explain the destruction process. Here, we analyse the crustal structure and upper mantle discontinuity by employing the H–k stacking technique of receiver function as well as the depth domain receiver function. Our results indicate deep mantle upwelling and lower crustal delamination beneath the eastern NCC, and suggest that either or both of these processes contributed to the unique lithospheric thinning and destruction of the eastern NCC. © 2013 The Authors. Geological Journal published by John Wiley & Sons, Ltd.


INTRODUCTION
The North China Craton (NCC) refers to the Chinese part of the Sino-Korean Platform, comprising a triangular region covering approximately 1 500 000 km 2 , and bound by faults and younger orogenic belts. The Early Palaeozoic Qilianshan Orogen and the Late Palaeozoic-Late Mesozoic Central Asian Orogenic Belt bound the craton to the west and the north, respectively (Ren et al., 1987;Zorin et al., 2001;Zhao et al., 2001;Xiao et al., 2013). To the south, the craton is bordered by the Qinling-Dabie-Sulu ultrahigh-pressure metamorphic belt that separates the NCC from the Yangtze Craton Wu and Zheng, 2013;Santosh and Somerville, 2013).
The NCC is composed of two major Precambrian blocks, the eastern and western NCC, sutured along a Palaeoproterozoic collision zone known as the Trans-North China Orogen (central NCC) (Zhao et al., 2001Zhai and Santosh, 2011;Zhao and Zhai, 2013;Zheng et al., 2013). The eastern NCC has undergone significant tectonothermal reactivation and destruction with a dramatic change in lithospheric architecture during the Ordovician to the Cenozoic (Chen, 2010). Recent evidence gathered from geological and geophysical studies confirm that the Eastern Block of the NCC has lost a considerable part of its lithospheric keel (Zhang, 1993Cope and Graham, 2007;Li et al., 2007;Gao et al., 2008;Wu et al., 2008;Xu and Zhao, 2009;Chen, 2009;Santosh, 2010).
Models on the destruction of the eastern NCC can be principally grouped into two: (1) thermal/mechanical and/or chemical erosion (e.g. Menzies et al., 1993;Deng et al., 1994;Zheng et al., 1998); and (2) lower crustal and (or) lithospheric delamination (Deng et al., 1994;Gao et al., 1998;Xu et al., 2002;Ling et al., 2009). The erosion model suggests that the thinning of the NCC lithosphere resulted from chemical and/or mechanical interactions of lithospheric mantle with melts or hydrous fluids derived from the asthenosphere. The delamination model proposes that the destruction of the lithosphere was caused by foundering of eclogitic lower crust together with lithospheric mantle into the underlying convecting mantle (Ling et al., 2009). However, the processes of erosion and delamination remain controversial and lack robust evidence from seismic investigation. The depth domain receiver function can effectively delineate the 410 and 660 km interfaces, and 410 and 660 km discontinuity topography displays the thermal state of the upper mantle from which any upwelling mantle plume from the lower mantle can be tracked (e.g. Chen and Ai, 2009;Huerta et al., 2009;He, 2010). The H-k stacking technique can be used to derive the thickness and the average composition of the continental crust, which in turn is used to estimate the composition of the lower crust and its geodynamic implications Ammon, 1995, Christensen, 1996;Niu and James, 2002;Nair et al., 2006;Thompson et al., 2010;He et al., 2013aHe et al., , 2013b. In this study, we analyse the crustal and upper mantle structure by employing the H-k stacking technique (Zhu and Kanamori, 2000) and depth domain receiver function (Yuan et al., 1997), and address their dynamics implications. Based on the results, we suggest an integrated model for the cratonic destruction of the eastern NCC.

METHODS AND DATA
This study performs a careful analysis of the receiver functions in the NCC, in order to characterize the bulk seismic properties of the crust with local estimates for the crustal thickness and Vp/Vs ratio (Zandt and Ammon, 1995;Zhu and Kanamori, 2000). For this purpose, we apply the stacking procedure of Zhu and Kanamori (2000) to 314 seismic stations located in the NCC (Fig. 1). These stations have been in operation from July 2008 to the present-day by the China Earthquake network (Zheng et al., 2010). We selected a total of 441 events with magnitude mb ≥ 5.8 recorded by those stations (Figs. 1 and 2 and Supplementary Table S1). For each event-station pair, data were selected within the ranges of 30-95°with reasonable back azimuth coverage and initially windowed 15 s before and 120 s after the P-wave pick. Data are filtered using a zero-phase Butterworth bandpass filter with corner frequencies of 0.03-3 Hz (or 0.05-1 Hz for upper mantle discontinuity study). We then used a modified frequency domain deconvolution (Langston, 1979;Owens et al., 1984;Zhu and Kanamori, 2000), and the Gaussian factor and water level were set to 3 and 0.01 (or 1 and 0.01 for upper mantle discontinuity study), respectively. The stacking is usually done in the time domain for a cluster of events (Owens et al., 1984;Zhu and Kanamori, 2000). Here we chose unequal weights (0.6, 0.3 and 0.1) for Ps, PpPs and PpSs + PsPs phases of the Moho, respectively. Using the Taylor expansion at the maximum and omitting the higher-order terms to quantitatively estimate the uncertainty (Zhu and Kanamori, 2000) of our results, we measured error bars for h and κ, which line up with longitude, by taking into account uncertainties (Fig. 3). We also show two examples of H-k stacking computed at two stations located at the NCC ( Supplementary Fig. S1).
Generally, each amplitude on the radial receiver function is produced by a primary P-to-S conversion at some depth. Depth is determined by the time delay of the amplitude with respect to the direct P using the 1D IASP91 velocity model (Kennett and Engdahl, 1991).
Based on it, we converted the time domain receiver functions into the depth domain after corrections for the incidence angle effect (correcting the move out of Ps conversions to the same incidence or near vertical incidence) (Yuan et al., 1997). By stacking all depth domain receiver functions from different back azimuths at each station respectively, we obtain the average receiver function at each station (Owens et al., 1984). The average receiver function can clearly delineate the upper mantle discontinuity (Zhu, 2000) ( Supplementary Fig. S2).

CRUSTAL THICKNESS AND Vp/Vs RATIO
The crustal thickness of the NCC identified by this study shows a gradual increase in a northwestward direction from 32 to 48 km (Fig. 3a, Fig. 4, arrow a) and in a westward direction in the western part (outside the NCC) ( Fig. 4, arrow b). The central and southern parts of the eastern NCC display a flat Moho interface with a crustal thickness of 32-34 km (Fig. 4). Wei et al. (2013) also obtained similar results in this area. Subduction and collision tectonics may generate local topography on the Moho which may survive for a long time after cessation of the collision and subduction (Cook et al., 2000;Balling, 2000). Stable continental structures can exist over long periods of time (>1.5 Ga) (e.g. de Smet et al., 1998). The dipping Moho topography is considered as an indication of the remnants of collision and deep subduction (Svenningsen et al., 2007). In the Late Mesozoic, only the Kula Plate was rapidly moving north-northwest and was subducting beneath the Eurasian continent (e.g. Hilde et al., 1977;Lee et al., 2005;Hou et al., 2009) with a shallow dipping Benioff zone (Menzies et al., 2007). Therefore, we regard that our results provide a robust image of the fossil architecture of the Kula Plate subducting beneath the NCC.
The bulk Vp/Vs ratio identified in this study shows highervalues for the regions in the northern part of the eastern NCC and the Trans-North China Orogen (>1.76) (Fig. 5), as well as in the southern part of the western NCC, suggesting an intermediate to mafic/ultra-mafic lower crust in these areas (Zandt and Ammon, 1995;Thompson et al., 2010;He et al., 2013b). Compared to the northern part, the bulk Vp/Vs ratio is lower (mostly between 1.76 and 1.74 or <1.74) at the central and southern parts of the eastern NCC (Fig. 5), which imply an intermediate to felsic lower crust in this area (Zandt and Ammon, 1995;Chang and Baag, 2007;He et al., 2013a).

410 AND 660 km UPPER MANTLE DISCONTINUITIES
Our results show a significant deepening of the 410 km discontinuity beneath the northern part of the Trans-North China Orogen and the eastern NCC (410 km discontinuity >400 km) (Fig. 6). A similar feature is also observed for the 660 km discontinuity (Fig. 7) (660 km discontinuity >655 km) in these regions. The average values for the depth of the 410 and 660 km discontinuities in this area are around 406 and~651 km respectively, both of which are lower than the global average values.
Investigations on the deepening of both the 410 and 660 km discontinuities have correlated the feature with a phase transition from majorite garnet to perovskite at 660 km depth (e.g. Deuss et al., 2006;Huerta et al., 2009), characterized by positive pressure-temperature gradients (the Clapeyron slope is positive). Therefore, when influenced by a mantle plume (or upwelling mantle), both the 410 and 660 km discontinuities would display deepening. A similar 660 km discontinuity feature has been observed in other regions, such as beneath Sri Lanka, western North America, West Yunnan and Northeast China (Pathak et al., 2006;Houser and Williams, 2010;He, 2010;He et al., 2013b).
Elevated temperature is predicted to deepen the 410 km discontinuity from its global average value by ∼8 km/100°C. For a plume with a 200-300°C temperature anomaly, a downward deflection of ∼15-25 km is expected (Foulger, 2012). Our results indicate a significant deepening of both the 410 and 660 km discontinuities by around 15 km. We regard this feature as an indication of the vestige of an upwelling mantle plume. Upwelling of the mantle provides a reasonable explanation for this topography of the upper mantle discontinuities in this area.
The mantle plume has a large diameter of around 500 km or more (Albers and Christensen, 1996). Seismic tomography shows that a possible mantle upwelling exists between longitude 110-120°E and latitude 34-44°N (Liu and Chang, 2001;Zhao and Zheng, 2005). In another study, Liang et al. (2004) suggested a thermal mantle 'dome' in the NCC based on the results for Pn velocity inversion. This location is consistent with our proposed region for the upwelling mantle plume. We consider that these geophysical images record the state of the Mesozoic mantle beneath this region .

DISCUSSION
From west to east in the NCC, the thickness of continental lithosphere varies from 200-250 km in the Ordos Basin to 100-150 km in the Taihang Mountains to 60-100 km in the  et al., 2007). However, our studies show that the Kula Plate subducted far into the NCC (Fig. 4, arrow a), and in this case, there is no reason why only the eastern part of the NCC was affected, preserving the keel beneath the western NCC. We therefore exclude the model of low-angle subduction of the Kula or Pacific plates as a major factor for the lithospheric thinning and destruction of the NCC. Because of the smaller size of the craton, as compared to the other large cratons on the globe, the NCC may have been more easily affected by subduction and collision of the surrounding blocks (Zheng et al., 2007). The Cenozoic tectonic evolution of East China is generally correlated to the collision between the Indian and Eurasian plates (J.Z. M. Liu et al., 2004), as well as the subduction of the Pacific Plate beneath the Eurasian Plate (Deng et al., 1996;Allen et al., 1997). The culmination of the decratonization of the NCC has also been related to an eastward jump of Cenozoic subduction of the Pacific Plate and the farfield effect of eastward extrusion of Cenozoic subduction of the Indian Plate .
After the cessation of the Kula Plate subduction, the Pacific Plate moved northward and subducted beneath the Eurasian Plate since 53 Ma, with an E to W trend (e.g. Zhang et al., 2003;Parés and Moore, 2005). The shear wave splitting measurements also show a fossil architecture with an E-W trending deformation in the NCC (Zhao and Zheng, 2005). Our results demonstrate that the westward crustal thickening in the western part of the area (outside the NCC) (Fig 4, arrow b), might have been directly influenced by the east to west compression, possibly related to the Pacific Plate  Supplementary Table S2). Arrow a: the gradual increase in crustal thickness in a northwestward direction in the NCC. Arrow b shows the gradual increase in crustal thickness in a westward direction beneath the western region of this area (outside the NCC).
Central and southern parts of the eastern NCC show a flat Moho interface. This figure is available in colour online at wileyonlinelibrary.com/journal/gj  (Kay and Kay, 1993) in this area. The central and southern parts of the eastern NCC with the flat Moho interface and the felsic-intermediate lower crust is markedly different to the other regions in the eastern NCC (Figs. 4 and 5), which suggest that this area underwent lower crustal delamination (Christensen, 1996;Zandt and Ammon, 1995;He et al., 2013a). The delamination resulted in the flat Moho interface following the low-angle subduction of the Kula Plate. This process also led to the upper lithospheric delamination (Kay and Kay, 1993). When lower crustal delamination occurs, the asthenospheric mantle would rise into and above the level of the lithosphere. Once this happens, the lithosphere will peel away and sink (Stern, 2004). Eventually, the lower crustal and (or) lithospheric delamination (Z.J.  would result in lithospheric thinning and craton destruction (Zhang et al., 2011).
However, it is difficult for the delamination process to act on the whole eastern NCC and remove a domain as thick as 80-120 km of the lithosphere. Lower crustal xenoliths in Mesozoic and Cenozoic basalts within the NCC are of two types , suggesting that a hot and fertile lithosphere has replaced the earlier cratonic keel. The regional synthesis suggests that Mesozoic-Cenozoic lithospheric thinning and mantle replacement was heterogeneously distributed across the NCC in space and time . Moreover, it has been generally assumed that delamination is a large-scale process that occurs rapidly (Zhang et al., 2011). Evidence presented in previous studies suggests that destruction of the eastern NCC is probably diachronous (Xu, 2007;Yang et al., 2010).
It has been inferred that westward subduction of the Pacific Plate has played the first-order role in thinning the lithosphere of the NCC . It is possible that both the mechanical delamination and thermo-chemical erosion served only as the second-order mechanisms for the lithospheric thinning (Zheng and Wu, 2009). The upwelling mantle plume can lead to the lower crust underplating and the formation of a mafic/ultramafic lower crust (Pirajno, 2007;Zandt and Ammon, 1995;He et al., 2013b). The mafic and ultra-mafic lower crust partially overlaps with the deepened region of 410 and 660 km discontinuities, as identified by this study in the northern part of the Trans-North China Orogen and the eastern NCC, confirming an upwelling mantle plume and magma underplating. The P-wave velocity and density in the lowermost crust beneath this area are significantly higher than the surrounding region, suggesting a deep process of lower crustal underplating (Z.J. . A comprehensive synthesis of the U-Pb geochronology and Hf isotopes of Phanerozoic zircons reveal plume-induced widespread and episodic magma underplating into the ancient lower crust of the NCC (H.F. . The basalt-borne xenoliths and geophysical observations reveal a higher heat flow value (∼65 mW/m 2 ) in the eastern NCC from the Cenozoic to the present-day (e.g. Chen et al., 1991;Menzies et al., 1993;Griffin et al., 1998;Fan et al., 2000;Xu, 2001;Zhu et al., 2002;Huang et al., 2003;Wu et al., 2005;Chen et al., 2006Chen et al., , 2008Chen, 2010). These data also support deep mantle upwelling and lower crustal underplating in this region (Stel et al., 1993).
Recent studies on the sedimentation history of major basins in the NCC have revealed that the western Shandong-Bohai Bay basins, the western Liaoning basins, and the Xishan Basin to the west of Beijing received some detritus from the southeastern Central Asian Orogenic Belt (CAOB) in the Late Triassic-Jurassic (Yang et al., 2006;Meng et al., 2010;Li et al., 2013). These results suggest that the eastern NCC must have witnessed a regional subsidence rather than uplift in the Jurassic-Early Cretaceous interval Li, 2013). The subsidence might be attributed to thermo-chemical erosion of the lithosphere by the underlying convective mantle (e.g. Menzies et al., 1993;Griffin et al., 1998;Xu, 2001;Li et al., 2013).  Supplementary Table S3). The average value of the 410 km discontinuity is~406 km. A significant deepening of the 410 km discontinuity is observed beneath the northern part of the Trans-North China Orogen and the eastern NCC (410 km discontinuity 400 km). This figure is available in colour online at wileyonlinelibrary.com/journal/gj  Menzies and Xu (1998) and Xu (2001) emphasized the importance of thermo-mechanical erosion from the base of the lithosphere and subsequent chemical erosion resulting from asthenosphere upwelling. Deng et al. (2004) correlated the thermal erosion to the mantle plume activity . Associated with lithospheric thinning is upwelling of the convective asthenosphere and modification of the thermal structure of the lithosphere .
Destruction of the eastern and northern parts of the NCC likely started during the Early Mesozoic (Han et al., 2004;Yang and Wu, 2009;Yang et al., 2007Yang et al., , 2010Zhang et al., 2009), with the lithospheric destruction initiated at the northern and eastern margins S.H. Zhang et al., 2012). This inference is consistent with the location of the proposed upwelling mantle plume. Therefore, we suggest that the lithospheric thinning in the eastern NCC is also related to thermal erosion caused by an upwelling mantle, or plume in this area.

CONCLUSIONS
The Kula Plate subduction beneath the NCC with a low-angle Benioff zone is excluded as a major factor for the lithospheric thinning of the eastern NCC. An upwelling mantle or plume might have existed beneath the northern part of the Trans-North China Orogen and the eastern NCC, which resulted in magmatic underplating beneath the lower crust in this area. The westward subduction of the Pacific Plate and resultant compression led to lower crustal and (or) lithospheric delamination in the central and southern parts of the eastern NCC. Eventually, both upwelling mantle plume and lower crust and (or) lithospheric delamination resulted in lithospheric thinning and craton destruction in the eastern NCC. greatly helped us in preparing the revised version of this manuscript for resubmission. Financial support is provided by National Nature Science Foundation of China (41174081) and SinoProbe-Deep Exploration in . This study also contributes to the 1000 Talent Award to M. Santosh from the Chinese Government. Waveform data for this study are provided by Data Management Centre of China National Seismic Network at the Institute of Geophysics, China Earthquake Administration.

SUPPORTING INFORMATION
Additional supporting information may be found in the online version of this article at the publisher's web site: Table S1. Earthquake catalogues used by this study. Table S2. Results of h-k stacking (valid seismic station). Table S3. Results of upper mantle discontinuity. Figure S1. H-κ stacks from stations JS and NIC. Figure S2. Average receiver function.