Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://dynamo.geol.msu.ru/personal/vsz/papers/Zakharov_etal_2015_eng.pdf Äàòà èçìåíåíèÿ: Thu May 21 11:34:12 2015 Äàòà èíäåêñèðîâàíèÿ: Sun Apr 10 00:04:44 2016 Êîäèðîâêà:

ISSN 0145 8752, Moscow University Geology Bulletin, 2015, Vol. 70, No. 2, pp. 77­83. © Allerton Press, Inc., 2015. Original Russian Text © V.S. Zakharov, A.L. Perchuk, S.P. Zav'yalov, T.A. Sineva, T.V. Gerya, 2015, published in Vestnik Moskovskogo Universiteta. Geologiya, 2015, No. 2, pp. 3­9.

Supercomputer Simulation of Continental Collisions in Precambrian: The Effect of Lithosphere Thickness
V. S. Zakharova, A. L. Perchuka, S. P. Zav'yalova, T. A. Sinevaa, and T. V. Geryaa,
Department of Geology, Moscow State University, Moscow, 119991 Russia b ETH Zurich, Switzerland e mail: vszakharov@yandex.ru, alp@geol.msu.ru, serhantes91@gmail.com, sinjvf@rambler.ru, taras.gerya@erdw.ethz.ch
Received November 12, 2014
a b

Abstract--Many aspects of Precambrian tectonics are still unclear due to the undetermined influence of sev eral key physical parameters (the temperature of the mantle, the thickness of the lithosphere, etc.) on the geo dynamic processes. The values of these parameters in the Precambrian are known to have been significantly different from current conditions. This work presents the results of two dimensional (2D) numerical petro logical­thermomechanical experiments that simulate the process of plate convergence at a velocity of 5 cm/year depending on the thickness of the continental lithosphere. The model continental lithosphere thickness ranged from 100 to 200 km, the modern mantle temperature exceeded 150°C, while radiogenic heat generation in the continental crust was 1.5 times higher then that at present. The numerical simulations showed that if the lithosphere is 100­160 km thick, the subduction process (closure of an ocean) is terminated by detachment of the oceanic plate from the continental plate (slab break off) followed by the formation of a large igneous province (an oceanic plateau) between the continents instead of orogeny. The thinner the litho sphere is, the earlier and closer to the surface the slab break off occurs. Thus, for a model with a continental lithosphere thickness of 150 km, the slab was detached in 10.3 Ma at a depth of 150 km, whereas for a litho sphere of 100 km thick it occurred in just 5.1 Ma almost at the surface. In the latter case, the magma genera tion area becomes much larger due to the formation of igneous provinces on both sides of the oceanic slab instead of one side, as is proposed in other models. Collision of continents with a very thick lithosphere (200 km or more) is not accompanied by slab break off and significant volcanic activity. Thus, the results of our modeling show that the SLAB PULL mechanism in a subduction zone contributes to the processes at a plate convergent boundary. Keywords: Precambrian, continental collision, subduction, plate tectonics, numerical modeling, supercomputers DOI: 10.3103/S014587521502012X

INTRODUCTION An increased heat flow at the early stages of the Earth's evolution was a cause of the specific character of tectonic and petrological processes, which differs from the modern character. This was a period of the formation of the bulk of the continental crust, includ ing cratons, the most ancient consolidated areas of continents with deep mantle roots (keels). The mech anisms of the formation of cratons and the character of their interaction in convergence zones are still contro versial (Brown, 2006, 2008; Cawood et al., 2006; Dhuime et al., 2012; Condie and Kroner, 2013). Along with the study of ancient complexes, com puter modeling makes a significant contribution to the study of the geodynamic processes in the Precambrian (Gerya, 2014). For example, using computer modeling Sizova et al. (2010, 2014) studied changes in the pattern of subduction and collision processes depending on such parameters of the lithosphere as the geothermal gradi
77

ent, the upper mantle temperature, and the heat radi ation in the continental crust. These studies showed that at the model parameters that correspond to the present conditions, the continental lithosphere sub merges to a depth of more than 100 km and the forma tion of orogenic belts similar to high mountain Phan erozoic fold belts (for example, the Himalayas and Caucasus). In the case of a higher mantle temperature and radiation heat generation that correspond to the Precambrian conditions, the collision process is quite different. Since the continental lithosphere cannot submerge into the mantle, the formation of flat large magma generating areas instead of orogens occurs. In order to study the Precambrian collision process it is necessary to determine what role another impor tant parameter, viz., the thickness of the continental lithosphere, plays. The thickness of the modern conti nental lithosphere varies in a quite wide range from 90­110 km to 200­300 km in cratonic areas (Gung et al., 2003; McKenzie and Priestley, 2008; Artemieva,


78

ZAKHAROV et al.

2011). The existing data on the lithosphere thickness in the Precambrian are controversial because they are based on model concepts and parameters (heat flow, geothermal gradient, radiogenic heat generation in the crust and upper mantle, etc.), which are characterized by a high degree of uncertainty for reconstructing the conditions of the ancient Earth. Accordingly, the esti mates of the lithosphere thickness are also quite diverse. Thus, it is proposed that the Early Proterozoic crust had a thickness in the range of 120 to 260 km, while the Archean crust had a thickness in the range from 140 to 350 km (and even up to 400 km) (Rudnick et al., 1998; Artemieva and Mooney, 2001). Based on studying volcanites, the Archean lithosphere crust (2.7­2.8 Ga) is estimated to have been 80­90 km (Windley and Devis, 1978). Thus, the range of possible values of the thickness of the continental lithosphere is very wide. Hence, it is quite important to investigate the effect of lithosphere thickness on the collision pro cesses in the Precambrian. The results of our research are considered below. The Modeling of the Collision Process During the modeling we used original consistent thermomechanical and petrological two dimensional models by T.V. Gerya (Gerya and Yuen, 2003, 2007; Gerya, 2010, Sizova et al., 2014). These models con sider the deformation of the medium under the action of applied tectonic forces, thus solving the equations of motion, continuity, and heat conduction in a mov ing medium taking the mass forces associated with thermal and chemical heterogeneities into account, as well as thermal effects induced by adiabatic compres sion/extension and viscous friction. In addition, these models take into account the effects of phase transi tions, including partial melting, migration of fluids and melts, and the formation of continental crust (Vasiliev et al., 2004; Gerya, 2010). In this case, it is assumed that the degree of melting of the rocks depends on the pressure, temperature, and water content. In order to achieve the adequacy of the models the real rheological properties of visco plastic rocks are given. The modeling methods we developed allow one to operate with a wide range of deformation parame ters, which makes it possible to investigate the defor mational processes in strike­slip zones in detail, including near fault deformations (Gerya, 2010). In order to describe the lithological structure of the model we used a very dense grid with randomly distrib uted markers (between the nodes of the main grid). In total, from several hundreds of thousands to tens of mil lions of markers were used when modeling. This makes it possible to identify the characteristic features of the dynamics of the collision zone with high resolution. The computer code was based on the Lagrangian marker in cell finite difference method and the multi grid method. Original high performance com

puting programs using the OpenMP parallel process ing technology (Gerya and Yuen, 2003, 2007; Gerya, 2010; Sizova et al., 2014), which was developed for the thermo mechanical modeling of geodynamic pro cesses, were used. The resources of the supercomputer center of Moscow State University were used for the numerical modeling (Voevodin et al., 2012). The Description of the Model We modeled the dynamics of the lithosphere and upper mantle (asthenosphere) in a vertical section; the horizontal size of the model was 4000 km and the ver tical size was 400 km. The number of nodes in the irregular grid equaled 2041 â 201; the average resolution of the model was ~2 km, the resolution of the model in the collision zone was ~1 km, and about 10 million Lagrangian markers were used. The initial state of the model (Fig. 1) consisted of two continents separated by oceanic litho sphere, with varying ages, temperatures, and, corre spondingly, thicknesses. The passive margin connected the oceanic crust and the left hand continent; in different variants of the model the width of the passive margin varied in the range of 20­150 km. The width of the ocean basin was accepted as 600 km; in the model this parameter was variable; the extent of the left continent was 1600 km; it varied depending on that of the oceanic lithosphere and passive margin; the extent of the right continent was 1700 km. The oceanic crust was composed of an upper basalt layer (2 km) and a lower gabbro layer (5 km). The two layer continental crust was represented by an upper layer (15­20 km) composed of acid rocks with rheological properties of wet quartzite and the lower layer (15­20 km) composed of mafic rocks with the rheological characteristics of plagioclase. At the beginning of the subduction the lithospheric mantle and asthenosphere was made of dry peridotite. The migration of the fluid as a result of metamorphic reac tions in the slab then occurred in the zone of the sink ing lithospheric block. During the modeling the following physical prop erties of all rocks were taken into account: density, ther mal conductivity, solidus and liquidus temperatures, the heat of melting, radiogenic heat generation, activation energy, rheology, and the coefficient of friction. Collision was preceded by subduction of the oce anic lithosphere beneath the right hand continent, which was originally caused by the movement of the left hand continent (arrow on Fig. 1) at a convergence velocity that can be varied from 5 to 30 cm/year. On the right side, a prism of sedimentary rocks was located over the subduction zone. The oceanic plate submerged along a dipping weakened zone into the mantle with rheological prop erties characteristic of wet olivine (Ranalli, 1995) and
Vol. 70 No. 2 2015

MOSCOW UNIVERSITY GEOLOGY BULLETIN


SUPERCOMPUTER SIMULATION OF CONTINENTAL COLLISIONS Z, km 0 50 5 cm/year 100 150 200 250 300 350 1700 1800 1 2 3 4 5 1900 6 7 8 9 10 2000 2100 11 12 13 14 15 2200 16 17 18 19 20 2300 21 22 23 24 25 2400 2500 2600 2700 X, km 26 27 28
1300

79

100 500
90 0

Fig. 1. The initial parameters of the model: 1, air; 2, water; 3, sediments 1; 4, sediments 2; 5, upper continental crust; 6, lower continental crust; 7, upper oceanic crust (basalts); 8, lower oceanic crust (gabbro); 9, dehydrated "dry" lithosphere mantle; 10, dehydrated ("dry") asthenosphere; 11, hydrated lithosphere mantle; 12, hydrated mantle; 13, serpentinized lithosphere; 14, depleted peridotites; 15, crystallized melts, derived from partially molten metasedimentary rocks; 16, crystallized tonalite trondhjemite granodiorites (TTG), derived from partially molten gabbro; 18, crystallized basalts, derived from partially molten peridotite; 19, partially molten sediments 1; 20, partially melted sediments 2; 21, partially molten upper continental crust; 22, partially molten lower continental crust; 23, partially molten basalts; 24, partially melted gabbro; 25, partially melted lithos phere mantle; 26, partially molten asthenosphere mantle; 27, melt (basalt, gabbro), melted from peridotite; 28, acid melts (TTG), melted from basalts. The direction of the movement of the left hand continent is shown with an arrow. Isotherms, °C.

lower strength during plastic deformation (the internal friction coefficient was equal to 0.1). This zone was located between the oceanic and continental (right) plates and extended from the base of the oceanic crust to the base of the continental lithosphere. The free slip boundary conditions were set. The upper bound ary of the lithosphere was regarded as a free inner sur face that overlies a low viscosity (1018 Pa s) model layer (air or sea water with a density of 1 kg/m3 and 1000 kg/m3, respectively) with a thickness of 18­20 km. A significant difference in the viscosity due to the introduction of a low viscosity boundary layer mini mized the shear stresses at the top of the lithosphere, which allows one to regard it as an effective free sur face (Schmeling et al., 2008). At the upper boundary, processes of erosion and sedimentation were observed. The initial thermal structure of the oceanic plate was determined by ocean geotherms that were calcu lated based on its thermal age and the upper mantle temperature (Turcotte and Schubert, 2002). In the models that are described in this work, the thermal structure was calculated for lithosphere with an age of 40 Ma. The thermal structure of the conti nental lithosphere was determined by radiogenic heat generation in crustal layers and the temperature at the basement. The continental lithosphere thickness in our models varied from 100 to 200 km. The mantle temperature was 1495°C, that is, 150°C higher than the temperature values taken for the modern geody namical conditions (1345°C); the radiogenic heat generation was 1.5 times higher than at present. According to (Abbott et al., 1994; Djomani et al.,
MOSCOW UNIVERSITY GEOLOGY BULLETIN Vol. 70

2001), such mantle temperatures correspond to the Neoarchean­Paleoproterozoic boundary. The temperature of the surface was taken equal to 0°; the initial geothermal gradient in the sublithospheric mantle was 0.5 degree/km. The thermal structure was then modeled in accordance with the heat transfer model. The velocity of convergence in all the models described in this work was 5 cm/year. After the termi nation of the subduction of all of the oceanic lithos phere, the pushing forces were turned off and the fur ther spontaneous development of the convergence process in the model was considered. Model Results The results of the modeling of the collision process with a varying lithosphere thickness are shown in Figs. 2 and 3. All the models with a lithosphere thickness of 100­160 km (Fig. 2) were characterized by similar trends of evolution. The subduction of the oceanic plate led to its dehydration. This, accordingly, was a cause of the weakening of the mantle wedge and over lying continental lithosphere, which were subjected to extension and accompanying decompression melting. The interaction of the oceanic lithosphere with the hot asthenosphere during the subduction also leads to the significant warming of a sinking slab, causing strength reduction. As a result, slab break off and its quick submergence into the mantle occur. The slab break off occurred at a depth that corresponds approximately to the continental lithosphere base ment.
No. 2 2015


80 Z, km 0 50 100 150 200 250 300 350 2200 2300 Z, km 0 (c) 50
900

ZAKHAROV et al. 10.33 Ma
100 50 0

Z, km 0 50 (b) 100

9.29 Ma
100 50 0

(a)

900

1300

150 200 250 300 2400 2500 2600
100 5 900 00

1300

500
900

9000 50

7.84 Ma

2700 X, km

100 150

200 250 300
00 13

350 2200 2300 2400 2500 2600 2700 X, km Z, km 5.17 Ma 0 90 500 50100 900 0 0 50 (d) 100 150 200 250 300 350 2000 2100 2200 2300 2400 2500 2600 2700 X, km

00 13

500

350 2200

2300

2400

2500

2600

2700 X, km

Fig. 2. The results of the modeling of convergent plate boundary evolution in the Precambrian at varying thickness of the conti nental lithosphere: (a) lithosphere thickness 150 km, 10.3 Ma after the beginning of subduction; (b) lithosphere thickness 140 km, 9.3 Ma; (c) lithosphere thickness 120 km, 7.8 Ma; (d) lithosphere thickness 100 km, 5.1 Ma. The dip of a detached oceanic slab is shown by an arrow. See legend in Fig.1.

Z, km 0 50 100 150

5.2 Ma

. log ­12 ­13 ­14 ­15

200 250 300 350 2000
10 cm/year

­16 ­17 2100 2200 2300 2400 2500 2600 2700 X, km ­18

· Fig. 3. Velocity (vectors) and strain rate (s­1) of rocks, induced by the dip of a cold slab that detached from a passive continental margin (upper left). The thickness of the simulated lithosphere is 100 km, 5.2 Ma after the beginning of subduction. The lithological and thermal structure is shown in Fig. 2d. MOSCOW UNIVERSITY GEOLOGY BULLETIN Vol. 70 No. 2 2015


SUPERCOMPUTER SIMULATION OF CONTINENTAL COLLISIONS Z, km 0 50 100 150 200 250 300 350 2200 10.21 Ma 100 500

81

0 90

900

2250

2300

2350

2400

2450

2500

2550

2600

2650

2700 X, km

Fig. 4. The results of modeling of convergent plate boundary evolution for the continental lithosphere of 200 km thick, 10.21 Ma after the beginning of subduction. See the legend in Fig. 1d.

The slab break off was followed by the retreat of the left hand continent. As the oceanic slab no longer pre vented movement of the left hand continent, the con tinents moved apart and the space between them was filled partially by upwelling molten mantle material, which leads to the formation of a vast magma genera tion area dominated by basaltic volcanism. Since the pushing forces in the model were turned off and the detached lithospheric slab no longer cre ated pulling forces, the movement of the left hand continent stopped. The presence of a zone of decom pressed and partly molten mantle between continents prevented them from further convergence and made further collision impossible in the modern sense. For the model with a continental lithosphere that was 150 km thick, the partial melting of the mantle at the active margin began 7­8 Ma (model time) after subduction and covered an area approximately 100 km wide. The oceanic slab break off occurred in 10.3 Ma (model time) (Fig. 2a). For the models with the continental lithosphere thickness of 140 km (Fig. 2b) and 120 km (Fig. 2c) the melting of the mantle at the active margin began 5­6 and 4­4.5 Ma (model time) after the beginning of subduction, respectively. The vast zone of hot mantle (and the associated magma generation) that formed between the conti nents had a size of 200­250 km. The slab break off and its submergence into the mantle occurred earlier, in 9.3 Ma for the lithosphere with a thickness of
MOSCOW UNIVERSITY GEOLOGY BULLETIN Vol. 70

140 km and in 7.8 Ma for the lithosphere with a thick ness of 120 km. A sinking slab pulled a small part of the continental passive margin, as well as a part of the accretionary prism. Between the continents, a basalt plateau of 200­300 km wide and up to 20­25 km thick formed. For the model with a very thin continental lithosphere (100 km), the melting of the active margin occurred even faster, in 4 Ma after the beginning of subduction. The zone of hot mantle that formed to the right of the subduc tion zone had a size of 250 km (Fig. 2d). In this model, the oceanic slab was detached from the passive margin in ~5.1 Ma (almost on the surface); then, it started to sink into the mantle. The rise of the hot mantle and related magma generation (formation of plateau basalts) were observed on both sides of the slab. After complete subsidence of the slab, both zones of hot man tle collided, forming a unified area of plateau basalts greater than 500 km wide and up to 25­30 km thick. Subsidence of the detached slab into the mantle is a very rapid process that occurred at a velocity that was several times greater than that of the convergence (Fig. 3). The slab is a relatively cold and, therefore, hard and dense block; it pulled a significant volume of oceanic crust and sediment. The character of the collision with a lithosphere thickness of 200 km or more is significantly different from the above discussed models. The subduction of the oceanic lithosphere led to heating of the active margin, but without melting. In this case, the rise of
No. 2 2015


82

ZAKHAROV et al.

the suprasubduction asthenospheric mantle was locally manifested and it was not accompanied by intensive volcanic activity on the surface. By the time that pushing forces were turned off (9 Ma after the beginning of subduction) the ocean basin was not completely closed. After this, the process of conver gence actually came to an end. The state of the model in 10.2 Ma is shown in Fig. 4. As a result of the mod eling, a collision orogenic belt was not formed and the slab was not detached. Subsequently, the slow relax ation of the slab occurred. CONCLUSIONS 1. The modeling showed that the collision process in the "hot" conditions that correspond to the ancient Earth was accompanied by increased heat generation upon the closure of the ocean and largely depends on the lithosphere thickness. 2. The style of the collision did not change for a lithosphere thickness in the range of 100­160 km. During the collision, the oceanic slab was detached in the transition zone between oceanic and continental lithosphere. This occurred due to a rapid viscosity reduction in this zone that resulted from the deviator stress concentration, deformation, and viscous heat ing at the transition from subduction to continental collision. The slab break off was accompanied by uplift of the suprasubduction mantle and its partial (decompression) melting, which led to the formation of a large igneous province in between the continents. At the same time, depending on the lithosphere thick ness, the models have the following distinct features: The thinner the continental lithosphere was, the earlier the formation of the magmatic province and slab break off occurred. The dimensions of the igne ous province and this parameter are also inversely related; The slab break off occurred at different depth levels, which decreased with a decrease in the lithos phere thickness. In the case of a very thin (100 km) continental lithosphere the slab was detached from the passive margin almost on the surface before complete absorption of the oceanic lithosphere in the subduc tion zone; The rise of the hot mantle and related magma generation at convergence of continents with the lithosphere thickness of 100 km were observed on both sides of the slab. This led to the formation of a large (>500 km wide) magmatic province; The rapid subsidence of a relatively cold slab that is detached from the continental plate into the mantle should lead to the formation of UHP metamorphic rocks. 3. Collision of continents with a lithosphere thick ness of 200 km or more did not lead to slab break off. In this case, the rise of the suprasubduction mantle

occurred locally and it was not accompanied by inten sive volcanic activity on the surface. ACKNOWLEDGMENTS This work was supported by Russian Foundation for Basic Research (projects nos. 13 05 01033 and 12 05 01093). REFERENCES
Abbott, D., Burgess, L., Longhi, J., and Smith, W.H.F., An empirical thermal history of the Earth's upper mantle, J. Geophys. Res., 1994, vol. 99, no. 7, pp. 13835­13850. Artemieva, I.M. and Mooney, W.D., Thermal thickness and evolution of Precambrian lithosphere; a global study, J. Geophys. Res., 2001, vol. 106, pp. 16387­16414. Artemieva, I.M., Lithosphere: an Interdisciplinary Approach, Cambridge: Cambridge Univ. Press, 2011. Brown, M., Duality of thermal regimes is the distinctive characteristic of plate tectonics since the Neoarchean, Geology, 2006, vol. 34, pp. 961­964. Brown, M., Characteristic thermal regimes of plate tecton ics and their metamorphic imprint throughout Earth history. When did Earth first adopt a plate tectonics mode of behavior? Geol. Soc. Am., Spec. Pap., 2008, vol. 440, pp. 97­128. Cawood, P.A., Kroner, A., and Pisarevsky, S., Precambrian plate tectonics: criteria and evidence, GSA Today, 2006, vol. 16, no. 11, pp. 4­11. Condie, K.C. and Kroner, A., The building blocks of conti nental crust: evidence for a major change in the tec tonic setting of continental growth at the end of the Archean, Gondwana Res., 2013, vol. 23, pp. 394­402. Dhuime, B., Hawkesworth, C.J., Cawood, P.A., and Sto rey, C.D., A change in the geodynamics of continental growth 3 billion years ago, Science, 2012, vol. 335, pp. 1334­1336. Djomani, Y.H.P., O'Reilly, S.Y., Griffin, W.L., and Mor gan, P., The density structure of subcontinental lithos phere through time, Earth Planet. Sci. Lett., 2001, vol. 184, pp. 605­621. Gerya, T.V. and Yuen, D.A., Characteristics based marker in cell method with conservative finite differences schemes for modeling geological flows with strongly variable transport properties, Phys. Earth Planet. Int., 2003, vol. 140, pp. 295­320. Gerya, T.V. and Yuen, D.A., Robust characteristics method for modeling multiphase visco elasto plastic thermo mechanical problems, Phys. Earth Planet Int., 2007, vol. 1. doi:10.101/j.pepi.2007.04.015 Gerya, T.V., Introduction of Numerical Geodynamic Model ling, Cambridge: Cambridge Univ. Press, 2010. Gerya, T., Precambrian geodynamics: concepts and models, Gondwana Res., 2014, vol. 25, pp. 442­463. Gung, Y., Panning, M., and Romanowicz, B., Global anisotropy and the thickness of continents, Nature, 2003, vol. 422, pp. 707­711.
Vol. 70 No. 2 2015

MOSCOW UNIVERSITY GEOLOGY BULLETIN


SUPERCOMPUTER SIMULATION OF CONTINENTAL COLLISIONS McKenzie, D. and Priestley, K., The influence of lithos pheric thickness variations on continental evolution, Lithos, 2008, vol. 102, pp. 1­11. Ranalli, G., Rheology of the Earth. London: Chapman, 1995. Rudnick, R.L., McDonough, W.F., and O'Connell, R.J., Thermal structure, thickness and composition of conti nental lithosphere, Chem. Geol., 1998, vol. 145, pp. 395­411. Schmeling, H., Babeyko, A.Y., Enns, A., et al., Benchmark comparison of spontaneous subduction models-- towards a free surface, Phys. Earth Planet. Int., 2008, vol. 171, nos. 1­4, pp. 198­223. Sizova, E., Gerya, T., Brown, M., and Perchuk, L.L., Sub duction styles in the Precambrian: insight from numer ical experiments, Lithos, 2010, vol. 116, pp. 209­229. Sizova, E.V., Gerya, T.V., and Brown, M., Contrasting styles of Phanerozoic and Precambrian continental

83

collision, Gondwana Rec., 2014, vol. 25, pp. 522­545. http://dx.doi.org/10.1016/j.gr.2012.12.011. Turcotte D.L. and Schubert G., Geodynamics, Cambridge Univ. Press, 2002. p. 456. Vasilyev, O.V., Gerya, T.V., and Yuen, D.A., The application of multidimensional wavelets to unveiling multi phase diagrams and in situ physical properties of rocks, Earth and Planet Sci. Lett., 2004, vol. 223, pp. 49­64. Voevodin, Vl.V., Zhumatiy, S.A., Sobolev, S.I., et al., Prac tice of "Lomonosov" Supercomputer. Open Systems J. Moscow: Open Systems Publ., 2012, no. 7, pp. 36­39. Windley, B.F. and Devis, F.B., Volcano spacings and lithos pheric, Earth Planet. Sci. Lett., 1978, vol. 38, pp. 291­ 297.

Translated by D. Voroschuk

MOSCOW UNIVERSITY GEOLOGY BULLETIN

Vol. 70

No. 2

2015