Deep geodynamics of far field intercontinental back-arc extension: Formation of Cenozoic volcanoes in northeastern China*

Vol. 17 Supp. (1 ~8) ACTA SEISMOLOGICA SINICA Nov., 2004 Article ID: 1000-9116(2004)Supp.-0001-08 Deep geodynamics of far field intercontinental back-arc extension: Formation of Cenozoic volcanoes in northeastern China* SHI Yao-lin (;~) ZHANG Jian (~ ~) Laboratory of Computational Geodynamics, Graduate School of Chinese Academy of Sciences, Beijing 100039, China Abstract There are three cases of variation of trench location possible to occur during subduction: trench fixed, trench advancing, and trench retreating. Retreat of trench may lead to back-arc extension. The Pacific plate subducts at low angle beneath the Eurasia plate, tomographic results indicate that the subducted Pacific slab does not penetrate the 670 km discontinuity, instead, it is lying fiat above the interface. The flattening occurred about 28 Ma ago. Geodynamic computation suggests: when the frontier of the subducted slab reaches the phase boundary of lower and upper mantle, it may be hindered and turn fiat lying above the boundary, facilitates the retreat of trench and back-arc extension. Volcanism in northeastern China is likely a product of such retreat of subduction, far field back-arc extension, and melting due to reduce of pressure while mantle upwelling. Key words: back-arc basin; Cenozoic volcano; geodynamics; northeastem China CLC number: P541 Document code: A Introduction Northeastem China has the most strong Cenozoic volcanism in China (Liu, 1999), where distributes more than 500 Cenozoic volcanoes, including sleeping volcanoes of Tianchi Lake (Celestial Pond) of Changbai Mountain, and Wudalianchi (Five linked Lakes) (LIU, 1999). Volcano of Tianchi Lake of Changbai Mountain consists of basaltic rocks of shield-forming stage and trachytes and pantellerites in cone-forming stage. It is suggested by study of REE, incompatible elements and Sr, Nd, and Pb isotopes that rocks at different stages have common origin of magma and close relation in evolution (LIU, et al, 1998). Volcanic rocks of different stages all show source characteristics of primary mantle, probably indicating existence of giant magma chamber and persistent supply of magma derived from the mantle (FAN, et al, 2001). Some volcanoes in Wudalianchi erupte on previous shield of lava, and form composite cone volcanoes. The new volcanoes are basaltic, featured by weak eruption and overflowing of lava. LIU, et al. (2001) systematically reviewed the temporal and spatial characteristics of volcanism of northeastern China. The Cenozoic volcanism of northeastern China, and even more general problem of Mesozoic rift-depression basins and related volcanism in continental margins, has drawn worldwide attention (Mantovani, et al, 2001). Chinese scholars also proposed their hypothesis * Received date: 2003-02-24; revised date: 2003-06-30; accepted date: 2003-06-30. Foundation item: National Natural Science Foundation of China (40234042 and 40174027).

2 ACTA SEISMOLOGICA SINICA Vol. 17 tion (Mantovani, et al, 2001). Chinese scholars also proposed their hypothesis based on their own research and in reference of foreign studies such as asthenosphere convection (MA, LIU, 1999), mantle plume (SUN, NIU, 2000), eastward or northeastward flow in the mantle (ZHAO, et al, 2001), active or passive back-arc extension (XU, ZHANG~ 2000), transcurrent motion in eastern edge of Asia (REN, LI, 2000), pull-apart due to lateral offset of NE or NNE strike faults, or combination of some of the above actions (ZHOU, et al, 1997). Although these hypotheses are based upon geological observations, they are mostly short of observation of the deep structure and theoretical analysis of geodynamics. Formation of the basins, especially mechanism of active volcanism in Northeastern China is a problem not yet well solved. Volcanism usually is produced by subduction as the following scenario. The cold oceanic slab is subducted to the deep hot mantle, the slab is heated up during subduction by conduction from the surroundings and by friction during penetration, such increase of temperature in the slab E.~ 50 100 0 50 100 150 200 250 300 Distance/kin Figure 1 Typical distribution of temperature, dehydration and partial melts in the subduction zone leads to dehydration of amphiboles in the subducted crust. The subducted slab itself usually does not melt because of lower temperature than its surroundings (except special case such as active ridge subduction, which may produce adakites). Rocks dry in the mantle wedge above the slab do have higher temperature, but originally do not melt because of lacking of water, Dehydration of crust in the subducted slab at sufficient depth, then can diffuse to the overriding mantle wedge, the spreading of water may reduce the temperature for partial melting. Therefore, the originally solid hot but dry rock now becomes wet and starts to melt. This usually occurs at a depth about 100 km (SHI, ZHANC~ 1998) that is independent of subduction angle, except for some potassium rich volcanoes occurring 200 km above the slab. Therefore, subduction at lower angle usually has volcanic arc at a greater distance to the trench than those of larger subduction angle (Figure 1). However, volcanoes of northeastern China are far from the subduction zones and are located 600 km above the subducted slab, their formation must be different from volcanoes of subduction zones. Their formation has been a riddle. Previous geologic and geophysical studies are mainly limited to the depth of lithosphere, in this paper, we attempt to propose a new model by using seismic tomographic images deep beneath the lithosphere and to analyze based on geodynamic numerical simulation of rollback subduction. The new hypothesis of formation of northeastern China volcanism from deep geodynamics may provide an altemative point of view for further discussion.

Supp. SHI Yao-lin, et al: FORMATION OF CENOZOIC VOLCANOES IN NORTHEASTERN CHINA 3 1 Roll-back subduction and far field intercontinental back-arc extension induced by deep dynamic process Location of trench may not always be fixed, three cases may exist as shown in Figure 2: the trench is fixed in space, the trench is advancing (therefore, producing tectonic erosion to the overriding plate) and the trench is retreating to induce back-arc extension (Taylor, Karner, 1983; Carlson, Media, 1984; Royden, 1993; SHI, WANG, 1993). Assuming the rate of convergence between two plates is U, migration rate of the trench is V 7",, u (positive if moves towards the overriding plate). (,) A B C e~ D I'2o of ~ o / The location of trench is unchanged if V=0, the c~" distance from volcanic arc to trench then is also r, ~ u fixed. The trench is advancing if V>O, part of rocks A B C oe n/,' in the overriding plate could be eroded and (h) ~ o /. wrapped to the subduction zone by the slab, the (;,,/L) rate of the subducting slab relative to the trench is /7 U-V if the erosion rate is V. If the overriding plate v ~ 7" ~s is continental, the core of ancient continent would (c) A " B o D q..~ ~g F be exposed to direct contact with the subducting D,P i" plate at the trench. Older volcanic arcs become ~t:-v~ closer to trench due to tectonic erosion, and new ~ v r volcanic arc is forming at locations about 100 km a ~ '~-~" above the slab, therefore, a migration of arc to- (a) ~. o_. S t~.a ~,, u D G wards the interior of continent appears. Trench retreats if V<O, overriding plate close to the trench moves together with migrating trench under the so called trench suction, however, overriding plate far away from the trench would not move at all, therefore, the segment of overriding plate between these two locations would extend, producing back-arc extension. The rate of subducting slab relative to trench is U+V. In addition, rollback Figure 2 Three possible cases of trench movement during subduction (a) two plates converge at rate U, location of trench is taken as reference; (b) trench is fixed as a normal case; (c) trench advancing-tectonic erosion; (d) retreat of trench-roll-back subduction and back-arc extension subduction tends to induce small-scale mantle convection at back-arc region, leading to upwelling of hot mantle and back-arc extension (SHI, WANG~ 1993). What is the condition for formation of rollback subduction? Figure 3 shows the numerical simulation of Zhong and Gurnis (1997). The model shows rollback subduction occurred in condition of a phase boundary at 670 km, in a non-newtonian flow in a cylindrical coordinate system. After the initiation of subduction, negative buoyancy due to lower temperature and higher density of the slab tends to pull down the slab for continuous subduction (Figure 3a). However, once the slab gets in touch with the 670 km phase boundary between the upper and lower mantle, although the cold slab is heavy, the lower mantle high pressure phase rock is even heavier, therefore, the subducting slab cannot immediately penetrate the boundary, it instead lies flat on the boundary due to the resistance from lower mantle (Figure 3b). The process is characterized by the plate convergence at rate U, slab roll-back and trench retreating rate of V usually not greater than U, slab sub

4 ACTA SEISMOLOGICA SINICA Figure 3 Vol. 17 Numerical simulation of subducting slab (Zhong, Gumis, 1997) (a) 50 Ma after subduction; (b) 82 Ma after subduction; (c) 117 Ma after subduction; (d) 148 Ma after subduction; (e) 192 Ma after subduction; (f) 213 Ma after subduction; (g) 231 Ma after subduction; (h) 247 Ma after subduction \ \ i% Figure 4 Tomography of northeastern China and adjacent area The section loction in (a) is shown by arclike line in (b)

Supp. SHI Yao-lin, et al: FORMATION OF CENOZOIC VOLCANOES IN NORTHEASTERN CHINA 5 duction rate relative to the mantle as U+V. After the slab gets in touch with the boundary, the subduction angle increases, the flat slab can move horizontally at first (Figure 3c), but gradually stops because of increasing resistance. Then, the flat lying slab thickens under compression due to resistance in the frontier and push at the back of the sequentially sinking slab (Figure 3d). After a period of retreating, more and more cold slab accumulates above the boundary, diffusion of heat enables phase transition from spinel to perovskite to occur at the slab right above the boundary. Such transition is in favor of overcoming the resistance of the lower mantle, and slab finally can penetrate to the lower mantle. The penetration may not occur at the frontier of the slab, instead, the middle of the flat lying slab may have thickest accumulation and largest negative buoyancy, therefore, be the location of start of penetration. Such penetration may pull the slab down and stop the rollback subduction (Figure 3e~h). It is noted that the scenario shown in Figure 3 is only one of the possibility, however, it is very suggestive for our discussion on the formation of volcano in northeastem China. 2 Deep geodynamics of the formation of volcanic rocks in northeastern China The above discussion gives a general description of volcanism related to subduction. Now we proceed for the special case of deep geodynamics of northeastern China. Seismic method, especially seismic tomography, is the major tool for investigating the state of the interior of the earth. Figure 4 from Zhao, et al (1997) shows a tomographic cross section of northeastern China and Japan Sea from 55 N, 110 E (northern end of Baikal Lake) to 30 N, 145 E (southeast to Tokyo, at sea of northern Ogasawara). Usually, the high seismic velocity zones are believed to be the subducted slab of the Pacific plate, which has lower temperature in comparison with the surrounding same kind of rocks at the same depth. The figure shows the subduction occurs out of the east coast of Honshu island at a subduction angle of about 30 (ZANG~ NING~ 1996), the slab dives beneath the Asian continent, but does not penetrate the 670 km discontinuity between upper and lower mantle, instead, it lies flat on the boundary. This result is in agreement with previous observations. For example, Revenaugh and Sipkin (1994) used ScS phase to study mantle structure of northeastern China. They concluded that the subducted slab is likely lying on the 670 km interface; the 410 km discontinuity may be rich of volatiles, and with about 1% partial melting. Long wave length MT exploration also suggested that mantle conductivity 400~600 km beneath northeastern China is an order of magnitude higher than those of normal mantle in the depth, interpreted as a flat lying slab (Ichiki, et al, 2001). Various methods are consistent in showing an image quite similar to the retreating subduction of Figures 3b-d, such an senary reject the hypotheses that volcanoes in northeastern China originates from plume rising from core-mantle boundary. The Eurasia and Pacific plates are converging at a rate of about 90 mm/a (DeMets, et al, 1990), and the Hawaii island chain shows that the Pacific plate have been moving stably relative to the hot spot for 40 Ma (Turcotte, Schuburt, 2001). Seismic data indicates the present overall subduction angle is about 30, horizontal projection of the slab from the trench to the 670 km boundary is about 1 100 km, the slab itself is about 1 300 km in length, and a rock subducting from the trench to the 570 km discontinuity takes a time of about 15 Ma. There is about 1 500 km of slab lying flat on the discontinuity, to reach such a state, the subduction should have lasted at least 17 Ma.

6 ACTA SEISMOLOGICA SINICA Vol. 17 However, it is noticed that the flat lying slab is thickened somehow during horizontal movement. If the thickening reach 50%, the original slab would be 2 300 km, the subduction then has lasted 26 Ma. In this case, from the initiation of subduction to the formation of the flat lying slab of present, it takes 42 Ma of time in total, and retreat of subduction and extension of back-arc may have occurred for 26 Ma. This is in agreement with geological evidences of the 27~28.4 Ma ages of the earliest volcanoes in Changbai Mountains (PEN~ YIN, 1999) with 31 Cenozoic volcanic events recognized (JIN, et al, 2000). It is also in agreement with the time of start of opening of the Japan Sea at time of 28 Ma (REN, LI, 2000) and major opening during time 22~15 Ma (Taira, 2001). The rate of retreat of trench is not well known, it is estimated to be 90 mm/a if the subduction keep a constant subduction angle, frontier of slab is not advancing and the fiat lying slab is not thickening any more. However, if the frontier of the subducted slab can advance somehow, and the fiat lying slab thickens during subduction, the retreating rate will be significantly smaller. If the retreat rate is only 40 ram/a, the total extension in 25 Ma would be 1 000 km. Even if the subduction angle may increase during the process of subduction, the total extension may still reach (a) Figure 5 Schematic show of subduction and back-arc extension (a) Early stage of subduction; (b) present state. Note the frontier of the flat lying slab has moved westward, flat lying slab has thickened, and sunduction angle has increased. Horizontal distance between T and To is the amount of back-arc extension (b) 700 km (Figure 5) Based on the tomographic images and above estimates, a scenario can be described on the volcanoes evolution. Although the Mesozoic history of Songliao basin and volcanism is not clear from seismic tomography to interpret, it is suggested that changes in Pacific subduction occurred after the Indian-Eurasia collision, leading to the termination of the first active period from 96 to 39 Ma defined by LIU, et al (2001). Volcanism during 39-28 Ma show such transition. Seismic tomography images indicate that since 28 Ma the subducting slab was blocked by the 670 km boundary, laying flat on it and thickening while the slab roll-back in subduction, the retreat of trench induced back-arc extension (including typical back-arc extension of Japan Sea, and far-field intercontinental extension in northeastern China). In the second active period of 28-16 Ma defined by LIU, et al (2001), extension of Japan Sea and extensional faults in eastern edge of Songliao basin both started, but extension of Japan Sea was much strong in activity. Japan Sea extension started from 28 Ma with major activity in period of 22~15 Ma, leading to 400 km of opening. After the cessation of Japan Sea opening, volcanism in northeastern China became more active because the rollback process of deep geodynamics was still going on. Volcanism spread from eastern edge to western edge of Songliao basin until the present. 3 Discussion and conclusions The phenomenon of far field intercontinental back-arc extension for basin formation has been discussed in other regions. Giles, et al (2002) suggested another case of basins of 1.8 Ga in northeastern Australia, although the location of the basins was far from trench, it was still produced by back-arc extension of the continent. There are a number of hypotheses on the formation of

Supp. SHI Yao-lin, et al: FORMATION OF CENOZOIC VOLCANOES IN NORTHEASTERN CHINA 7 back-arc extension and back-arc basins. Mantovani (2001) summarized them as: subduction related mechanism as rollback of subduction slab, induced comer flow, anchored slab, and not subduction related mechanism such as squeeze out, pull-apart, etc. Previous studies on evolutionary history usually are based on geological observation, however, recent tomographic technique provide a lot of new data on the image of subduction slab (HE, LIU, 1998). Although these images describe configuration of the present day subduction slab, it is produced by past plate motion and subduction, therefore, carefully analysis of these images may provide a new pathway to understand the geological history. Numerical simulation is of key importance in deciphering the new kind of geophysical data. In this paper, we apply tomographic data and numerical modeling results of different researchers to discuss the formation of volcanism in northeastern China. Three cases may exist during subduction: Location of trench may keep fixed, advance or retreat. Retreat of trench may produce back-arc extension and upwelling of hot mantle materials. Geodynamic modeling suggests that rollback subduction is most likely to occur when low-angle subducted slab reaches the upper and lower mantle boundary; block of the slab tends to produce retreat of trench and extension in back-arc. Tomographic images of the depth indicate that the formation of northeast China is quite likely the case of rollback subduction, back-arc extension, upwelling of hot mantle, and partial melting due to reduced pressure. Three-dimensional spherical modeling is necessary especially for the case of northeastern China. The present study only recall a geological history of the latest 40 Ma, the basins and volcanism in eastem China can be, however, traced to the Mesozoic, therefore, a study of the even older events is necessary. Spatially, differences in seismic tomography of northeast, north and southeast China exist; their geological history and relation to volcanism are problems for future study. In summary, this paper makes an attempt to interpret the formation of Cenozoic volcanos of northeastem China from deep geodynamics, the model should be tested by further integrated study of structural geology, petrology, geochemistry, and geophysics. Acknowiedgement We would like to thank Professor LIU Jia-qi for suggestions and providing references to this study. References Carlson R L, Media P J. 1984. Subduction hinge migration [J]. Tectonophysics, 102:399-411. DeMets C, Gordon R G; Argus D E et al. 1990. Current plate motions [J]. Geophys J Int, 101: 425-478. FAN Qi-cbeng, SUI Jian-li, LIU Ruo-xin. 2001. Sr-Nd isotopic geochemistry and magmatic evolutions of Wudalianchi volcano, Tianchi volcano and Tengchong volcano [J]. Acta Petrologica et Mineralogica, 20(3): 232~238 (in Chinese). Giles D, Betts E Lister D. 2002. Far field continental back arc setting for the 1.80-1.67 Ga basins of northeastern Australia J]. Geology, 30(9): 823-826. HE Jian-Hun, LIU Fu-tian. 1998. Relationship between the morphology of subducted slabs and the tectonic evolution in the active continental margins [J]. Progress in Geophysics, 13(2): 15~25 (in Chinese). Ichiki M, Uyeshima M, Utada H, et al. 2001. Upper mantle conductivity structure of the back-arc region beneath northeastern China [J]. Geophys Res Lett, 28(19): 3 773-3 776. JIN Ke, PENG Yu-jing, WANG Yan-sheng. 2000. The divide and contrast of volcanic events of Cenozoic era of Changbaishan area between China and Korea [J]. Journal of Changchun University of Science and Technology, 30(2): 125-130 (in Chinese). LIU Jia-qi, HAN Jing-tai, Fyfe W S. 2001. Cenozoic episodic volcanism and continental rifting in northeast China and possible link to Japan Sea development as revealed from K-Ar geochronology [J]. Tectonophysics, 338: 385-401. LIU Jia-qi. 1999. The China Volcano [M]. Beijing: Science Press, 9-126 (in Chinese). LIU Ruo-xin, FAN Qi-cheng, ZHEN Xiang-shen. 1998. Magmatic evolutions of Tianchi volcano in Changbai Mountain [J]. Science in China (Series D), 28(3): 226-231 (in Chinese). LIU Xiang. 1999. Tectonic control of Cenozoic volcanism in northeastern China [J]. The World Geology, 2:23-29 (in Chinese). MA li, LIU De-lai. 1999. The origin and evolution of Songliao basin and it's relation with asthenosphere convection model [J]. Scientia

8 ACTA SEISMOLOGICA SINICA Vol. 17 Geologica Sinica, 34(3): 365~374 (in Chinese). Mantovani E, Viti M, Babbucci D, et al. 2001. Back arc extension: which driving mechanism?[j] Journal of the Virtual Explorer, 3: 17-44. PENG Yu-jing, YIN Chang-jian. 1999. Stratum age and tectonic environment of Tertiary-lava in Jiling Province [J]. Geological Review, 45(Suppl.): 204~214 (in Chinese). REN Jian-ye, LI Si-tian. 2000. Spreading and dynamic setting of marginal basins of the western Pacific [J]. Earth Science Frontiers, 7(3): 203~213 (in Chinese). Revenaugh J, Sipkin S A. 1994. Mantle discontinuity structure beneath China [J]. J Geophys Res, 99(11): 21911-21 927. Royden L H. 1993. Evolution of retreating subduction boundaries formed during continental collision [J]. Tectonics, 12: 629-638. SHI Yao-lin, WANG Qi-yun.1993. Roll-back subduction and back-arc opening [J]. Chinese Journal of Geophysics, 36(1): 37~43 (in Chinese). SHI Yao-lin, ZHANG Jian. 1998. Thermal modeling of active ridge subduction and arc volcanic [J].Chinese Journal of Geophysics, 41(2): 180-187. SUN Ai-qun, NIU Shu-yin. 2000. The mantle plum and its geothermal consequnce: The deep tectonic setting of geothermal anomaly in the North Chian [J]. Acta Geoscientia Sinica, 21(2): 182~189 (in Chinese). Taira A. 2001. Tectonic evolution of the Japanese Island arc system [J]. Annu Rev Earth Planet Sci, 29: 109-134. Taylor B, Kamer G D. 1983. On the evolution of marginal basins [J]. Rev Geophysics, 21:1 727-1 741. Turcotte D L, Schubert G. 2001. Geodynamics (2nd edition) [M]. London:Cambridge University Press. XU Jun-yuan, ZHANG Lin-yuan. 2000. Genesis of Cenozoic basins in northwest Pacific ocean margin(l): Comments on basin-forming mechanism [J]. Oil & Gas Geology, 21(2): 93~98 (in Chinese). ZANG Shao-xian, NING Jie-yuan. 1996. Study on the subduction zone in western Pacific and its implication for the geodynamics [J]. Chinese Journal of Geophysics, 39(2): 188~202 (in Chinese). ZHAO Da-peng, XU Ying-biao, Wiens D A, et al. 1997. Depth extent of the Lau back-arc spreading center and its relation to subduction processes [J]. Science, 278: 254-257. ZHAO Hui-min, LU Bing-quan, SUN Hong-bin. 2001. Formation and evolution of marginal basins in the western Pacific [J]. Marine Geology & Quaternary Geology, 22(1): 57~62 (in Chinese). ZHONG S, Gurnis M. 1997. Dynamic interaction between tectonic plates, subducting slabs, and the mantle [J]. Earth Interactions, 1(1): 3 ZHOU Zu-yi, DING Xiao, LIAO Zong-ting. 1997. The mechanism of margin basin forming and its revelation for the geologic study of the south-east China [J]. Advance in Earth Sciences, 12( 1): 7-14 (in Chinese).