ABSTRACT
Interaction between a thermal convective flow and a plume head risen to the base of the continental lithosphere is modeled numerically. Results of this modeling are compared with peculiarities of the Cenozoic plume magmatism in eastern Australia. Correct modeling of the interaction requires a model for thermal convection in the continental upper mantle and a model for gravitational spreading of the plume head at the base of the continental lithosphere in the absence of convection. The described thermal-convection model is based on the idea that the continental lithosphere's a highly stable conductive layer chemically distinct from the upper mantle. Upwelling is typical of zones with ancient thick cratonic lithosphere, and downwelling takes place in zones with younger thin lithosphere. Modeling of gravitational spreading of a plume head in the absence of thermal convection shows that the plume spreads at the base of the continental lithosphere with a velocity of 3 cm/year over 15 Ma and then dissipates through cooling. According to the proposed model, thermal conduction acts on the plume, increasing the velocity of its spreading up to 67 cm/year and decreasing its lifespan to 10 Ma. In the zone of abnormally thin (< 100 km) lithosphere the plume head behaves in an unusual way. Here, in contrast to the "thinspot" model, the plume does not rise to the base of the thin lithosphere because of its interaction with convective flows. However, the plume moves upward for some distance sufficient for decompressional melting and formation of a magma chamber.
Key words: Plume, thermal convection, upper mantle, continent, Australia
- Part 1. Introduction
- Part 2. A Model for Thermal Convection beneath Platform Regions
- Part 3. A Model for Interaction between Plume and Thermal Convection
- Part 4. Discussion
- References
©1998 by Allerton Press, Inc.
Authorization to photocopy items for internal or personal use, or the internal or personal use of specific clients, is granted by Allerton Press, Inc. for libraries and other users registered with the Copyright Clearance Center (CCC) Transactional Reporting Service, provided that the base fee of $50.00 per copy is paid directly to CCC, 222 Rosewood Drive, Danvers, MA 01923.
INTRODUCTION
Thermal convection in the Earth's mantle and formation of plumes (isolated hot jets) are currently considered to be primary mechanisms of the dynamics of the Earth's bowels [17]. Global convective flows in the mantle are related to the movement of large lithospheric plates, and the oceanic lithosphere is commonly regarded as the upper thermal boundary layer of extended upper-mantle cells [810]. Plumes are suggested to provide mantle mechanisms for formation of "hot spots" in oceanic regions and basalt eruptions on continents. Furthermore, it is suggested that superplumes participate in fragmentation of continents and/or in formation of intracontinental rifts [1116]. In moving upward, a plume interacts with thermal convective flows. If its deflection from the vertical exceeds 60°, then the tail of the plume may be detached, and a new head may form at the place of detachment [17, 18]. Rheological stratification of the Earth's mantle also affects the evolution of the plume as it rises from the core-mantle boundary. Using laboratory models, Berkovici and Mahoney [19] demonstrated that on reaching the 660 km discontinuity, the plume rapidly accelerates its rise because the upper mantle is less viscous than the lower mantle. As a result, the head of the plume separates from its tail. A new head is formed at the bottom of the upper mantle from the matter supplied through the tail. After reaching the critical size, the new head starts rising and, like the old head, it appears at the bottom of the lithosphere in 2040 Ma. Support for the described mechanism comes from the so-called provinces of double flood basalts, where two separate episodes of basic eruptions were found, and the duration of the interval between them matched the interval between plume arrivals at the lithosphere bottom. These provinces include: the Kerguelen hot spot (110 and 85 Ma), Shatsky Rise (138 and 100 Ma), Parana basalts (127 and 80 Ma), and Karoo basalts (182 and 86 Ma). The authors claim that the movement of the plate in the period between the eruptions should be taken into account. It follows from [19] that the presence of double eruptions indicate that, according to depth of formation, plumes may be divided into all-mantle and upper-mantle.
The major changes in the structure, dynamics, and composition of the plume occur after it has reached the lithosphere bottom. In the oceanic regions where the direction of a horizontal sublithospheric branch of the convective flow coincides with the plate movement, the head of the arrived plume elongates in the direction of the plate movement, which is reflected in the structure and evolution of arched uplifts of oceanic hot spots [20, 21]. The magmatic events in hot spots forming volcano chains are likely to be caused by the tail of the plume, where the matter is hotter than in the head. It is this matter that, in the opinion of Richards et al. [22], experiences decompressional melting, which results in a hot-spot track. Analysis of factual and modeling data gives a tentative sequence of events of plume effect on the oceanic lithosphere. Approximately 20 Ma before the plume arrived at the lithosphere, a regional arch had formed from the plume passing through the asthenosphere [12]. The arrival of the plume head at the lithosphere somewhat localizes the uplift but increases its amplitude. The spreading of the plume head over the bottom of the lithosphere allows the matter of the tail to rise to the lithosphere. As a result, a melt appears and magmatism begins. Extraction of the melt in the course of volcanic events plays an important role in the evolution of the matter of the rising plume. This process results in a dramatic increase in the viscosity of the plume matter. It sometimes exceeds the viscosity of the asthenosphere, and this allows the rising plume to be considered (after separation of the melt) as part of an oceanic lithosphere plate [23]. The pattern and velocity of spreading of the head matter also depend on the rate of its cooling [24].
The main results in the investigation of plume evolution near the bottom of the lithosphere and evaluation of the plume effect on surface features were obtained in oceanic regions [25, 26]. Here the lithosphere thickness does not exceed 100 km. This results in decompressional melting of the plume matter on its anomalous heating (200300 °C higher than in the surrounding mantle) [27, 28]. Thus, the site of plume rise in oceans is always marked by volcanic activity and formation of an arched uplift. The continental lithosphere always has a more complicated structure than the oceanic one and is distinguished by considerable variations in lateral thickness. The lithosphere thickness of Precambrian platforms is typically 200 km, whereas in the surrounding younger platforms or fold belts the lithosphere is no more than 150 km in thickness [29, 30]. The thickness of the lithosphere of sedimentary basins (100125 km) is close to that of the mature oceanic lithosphere [31]. And finally, the maximum thicknesses of the continental lithosphere (250350 km) are observed in the region of Archean cratons [32]. As seen from the above short characteristics of variations of the continental lithosphere in thickness, a plume reaching the Precambrian lithosphere gives rise to an arched uplift with an amplitude of about 1 km. This is not accompanied by basaltic magmatism, contrary to oceanic regions. This type of structure is now observed in the north of the Siberian Platform, where the arched uplift of the Putorana plateau has formed for the last several million years. As mentioned above, transport of plume matter to a depth of decompressional melting (100 km or less) is a necessary condition for basaltic magmatism. The plume matter may be delivered to this depth provided that the continental lithosphere becomes thinner because of spreading, or a zone of the plate with a continental rift passes over the rising plume [16], or the head of a superplume spreads along the lithosphere bottom to an area with thin continental lithosphere (the "thinspot" mechanism) [33, 34]. Obviously, we do not consider the case when the rising plume immediately happens to come to a region with continental lithosphere no more than 100 km thick.
In our study, we propose and investigate another way of plume delivery to the "decompression" depth. Laboratory and numerical models of convective flows beneath a plate of varying thickness demonstrated that the structure of the flow is stabilized by a thickened zone of the plate, which is characterized by a stable ascending flow conveying excess heat to the edge of the thick plate [3538]. Thus, we suppose that the plume rising to a zone with thick lithosphere is driven away by a stable convective flow to the margin of this zone and farther to a zone with thin lithosphere, providing conditions for decompressional melting. It should be mentioned that beneath continents, in contrast to oceans, the direction of movement of the sublithospheric layers of upper-mantle convective flows may differ from the direction of movement of large lithospheric plates [39, 40]. Thus, the relationship between the direction of plate movement and the geometry of the volcano chain marking the location of the plume in continental regions may be more complex. Here, first of all, a model for upper-mantle thermal convection beneath a platform continental plate of varying thickness is put forward. Then a model for the interaction between the plume beneath this plate and thermal convection in the upper mantle is considered. Finally, the results of the modeling are compared with the evolution of Cenozoic plume magmatism in eastern Australia.
A MODEL FOR THERMAL CONVECTION BENEATH PLATFORM REGIONS
We start description of the model with our understanding of the rheology of the lithosphere and convecting mantle. This model is developed on the assumption of a dramatic difference in the lithosphere nature between continental and oceanic regions. The oceanic lithosphere is a cooler and, consequently, a more rigid layer of convecting mantle. This assumption implies an identity between the chemical compositions of the oceanic lithosphere and upper mantle. In turn, an increasing body of information on the isotopic composition of igneous rocks [41], mantle xenoliths [42], azimuthal anisotropy of S wave velocities [43], and anisotropy of the electric conductivity of the continental lithosphere [44] has been obtained for continents. These data indicate that the mantle part of the mature continental lithosphere remains isolated for a long time and does not participate in the convective stirring after the completion of the last stage of the orogeny forming the crust and the lithosphere as a whole. They also indicate that the composition of the continental lithosphere differs from the matter of the convecting mantle, first of all, in concentrations of volatile and major elements. Thus, it is safe to suggest that the rheology of the continental lithosphere of stable platform regions can be described by a different PT dependence than the rheology of convecting mantle, which is much more rigid than the oceanic lithosphere. As a first approximation, this model admits that the continental lithosphere has infinite viscosity, that is, this layer can be modeled as a rigid, undeformable body of laterally varying thickness with the conductive way of heat transfer. The rheology of the convecting mantle is described in the considered model by a dependence of viscosity on depth only, with the data on postglacial isostatic regulation taken into account [45]. The asthenosphere is specified as a layer 75 km thick, with a viscosity of 1.3·1020 P situated immediately beneath the lithosphere. The mantle viscosity beneath the asthenosphere is 1022 P. This model neglects the temperature dependence of viscosity, which is commonly used in numerical modeling of convective flows [46]. As mentioned above, usage of the same PT dependence of viscosity for the lithosphere and convecting mantle is required only for oceanic regions, for which numerical models using an appropriate dependence give the lithospheric layer automatically. In spite of the prevalent opinion that the asthenosphere beneath continents is also of thermal nature, data on another reason for viscosity reduction in the upper mantle have been recently obtained. According to laboratory experiments and field data, this is caused by the elevated hydrogen concentration beneath the lithosphere (> 1000 ppm H/Si), which corresponds to a layer of abnormally high electric conductivity [47].
Movement of the matter of the convecting mantle and heat redistribution in it are determined by the NavierStokes equations in the Boussinesq approximation:
where T is temperature,
is kinematic viscosity, Ra = 105-106 is the Rayleigh number, u and v are horizontal and vertical components of the velocity of the matter movement, a = 1/2 is the relation between the thermal conductivities in the crust and mantle, A is the rate of generation of radioactive heat in the interior, and
is the stream function.
The distribution of internal heat sources in the Earth's interior for stable continental regions was taken from [48]. The velocity of matter movement was taken to be zero everywhere in the conductive lithosphere. Conditions of adherence and undeformability (v = 0) were specified for the bottom of the lithosphere. The lower boundary of the designed zone corresponded to the depth 670 km. The conditions for slippage were met there, that is, tangential stresses were equal to zero, and a heat flow of 30 mW/m2 was specified. This value was chosen so that the heat flow on the surface of the thick lithosphere modeling a Precambrian platform would not exceed 4050 mW/m2 and for the thin lithosphere and younger fold belts, 6080 mW/m2. The standard conditions were met at vertical boundaries thermal isolation and zero tangential stresses and the component of the velocity normal to the boundary. The initial distribution of temperatures in depth was specified as indicated in [49], and its lateral disturbance was chosen as a sum of ten sinusoids of various periods [46] with a descending flow beneath the thick layer. The numerical implementation of the Navier-Stokes equations is described in [50]. The conductive lithosphere in this model occupies 200 km of the first 1560 km of depth, which models the Precambrian platform. Then, on lateral movement, its thickness decreases to 150 km over a distance of 840 km and corresponds to the younger Paleozoic platform. Finally, the thinnest lithosphere (100 km) within a stretch of 24002800 km corresponds to a region of the ancient active continental margin, Phanerozoic fold belt, or a sedimentary basin with thin lithosphere.
|
Fig.1. Temperature distribution and structure of thermal convective flow in the continental upper mantle beneath a lithosphere plate of varying thickness. Lengths of arrows are proportional to the velocity of matter movement at corresponding points. |
Investigation of time dependence of the structure of thermal convection at a given geometry of the designed zone demonstrated that a quasi-stationary convection structure had formed for 11.5 Ma with upwelling beneath the thick lithosphere and downwelling beneath thin lithosphere. This structure persisted for 10 Ma, the maximum calculation time (Fig. 1). The number of cells and the location of ascending and descending flows remained constant. Bulges appeared at times in the upper and lower thermal boundary layers. This was accompanied by an increase in the velocity of matter movement in the cells and resulted in an increase in the average stirring rate and the Nusselt number. However, the potential energy stored in the bulges at the considered convection regime (Ra < 106) was insufficient for cleavage and crushing of cells. As a result, the bulges dissipated, and the stirring rate decreased to the average value. This regime is known as regime of instability of thermal boundary layers. It is intermediate between the oscillating regime and the soft-turbulence regime, when cells still exist as a closed circulation, but their number changes with time [51]. To conclude the discussion of the convection model, it should be noted that the proposed model meets the entire standard set of observable geophysical parameters (relief, heat flow, gravitational-field anomalies).
A MODEL FOR INTERACTION BETWEEN PLUME AND THERMAL CONVECTION
According to modern views, plumes can be conventionally classed into two types. Type I is a continuous jet supplying hot matter to the lithosphere, and type II is an isolated drop rising by virtue of buoyancy. Type I is usually related to long-living hot spots, whereas type II is responsible for flood-basalt effusion without a resulting track. Appearance of a plume in the model, especially as an isolated hot jet, brings about, first of all, a spatial problem of a numerical representation of the process. On the one hand, modeling of upper-mantle thermal convection in large continental regions can be adequately performed in a two-dimensional variant. Application of planar convection models is justified here by the fact that the sizes of continental regions, such as Australian or North-Eurasian platforms, exceed the thickness of the upper mantle by an order of magnitude. For example, the boundary between the Australian Precambrian craton with thick lithosphere and younger artesian sedimentary basin runs meridionally for over 2000 km. For this reason, two-dimensional modeling of upper-mantle convection throughout a section across this boundary is acceptable. Unlike convection, a plume is a three-dimensional structure with a vertical axial symmetry. Therefore, it is presently modeled in a three-dimensional version [21, 52, 53], although two-dimensional models have not become obsolete either [20]. In the two-dimensional model, a plume is represented so that it, would not disturb the convection structure, on the one hand, and possess features characteristic of a plume, on the other. These features are the abnormally high temperature in the plume head and the stationary location of the site of its rise, that is, the site where the plume head arrives at the lithosphere bottom. In our opinion, these conditions are met by the temperature anomaly localized immediately beneath the lithosphere and marking the site of the plume rise. If a plume of type I is modeled, this anomaly is restored at each time step, thereby ensuring the continuous supply of plume matter and the stationary location of the jet. This model does not take into account the thermal decrease in the viscosity of the plume matter, which is referred by some researchers to its increased mobility, determining the evolution of the shape of the plume head near the lithosphere bottom [24, 54]. However, as shown in [13], the velocity of the spreading head in a liquid more viscous than the plume is controlled by the viscosity of the surrounding matter. Therefore, the evolution of the rising plume head is determined here by the rheology of the matter and the structure of mantle current near the hot jet. In our study, we shall restrict our consideration to a simple all-mantle plume referred to type II in our classification.
Before modeling the interaction between the plume and convection, we consider the features of spreading of the plume head along the lithosphere bottom in the absence of convection, which permits us to isolate a pure interaction effect. Advective movements of thermal or compositional density inhomogeneities in a viscous liquid are described by the same Navier-Stokes equations as thermal convection. The difference is that the Rayleigh number is below its critical value in the absence of temperature difference between the upper and lower boundaries. Here it reflects solely the properties of the thermal conductivity and rheology of the matter; in this case, therefore, the liquid movement is initiated only by buoyancy (floating or submergence) of a given inhomogeneity without convection of the entire liquid volume. Let us place the rising plume beneath the lithosphere of a constant thickness of 150 km. The temperature distribution in the lithosphere and concentration of radioactive isotopes correspond to the above-described model for thermal convection. The temperature of the sublithospheric mantle is now constant and equal to 1300 °C. We started modeling the spreading process from the moment when the bulk of the hot matter was near the lithosphere. Modeling of the rise of an all-mantle plume [52], shows that the plume head at that moment had the shape of an ellipse. The plume volume corresponds to the volume of a sphere with a radius of 200 km. Thus, an initial thermal anomaly with
= 300 °C with respect to the mantle
temperature was specified. This anomaly did not resume with time, because an isolated rising drop was modeled. The evolution of the head shape in the course of its simple spreading, referred to as a gravity current, is shown in Fig. 2. As time passes, the plume spreads along the lithosphere bottom under the action of the buoyancy of its hotter matter and becomes thinner rapidly enough. After 15 Ma, the moving edge of the plume, traveling laterally about 400 km, dissipates, that is, its temperature drops to that of the ambient mantle. The average spreading velocity over this period is no more than 3 cm/year. The time dependence of the spreading is presented in Fig. 3 (lower curve).
|
|
Fig.2. Evolution of the gravity current of the plume matter (white yellow) beneath continental lithosphere of constant thickness in the absence of convection. |
Fig.3. Time dependencies of the velocity of the frontal edge of the plume. The plot "gravity current" -- in the absence of convection; the plot "convective draft" -- under thermal convection in the upper mantle. |
|
Fig.4. Evolution of temperature distribution and current structure in the continental upper mantle beneath a lithosphere plate of varying thickness with a plume (white yellow) near the bottom of the lithosphere (lengths of arrows are proportional to current velocity). |
To investigate the interaction between the plume and thermal convection, we place a plume with the same features beneath the 150 km thick lithosphere region. The structure of the lithosphere is taken from the above-described thermal-convection model (Fig. 4). The figure shows that after 2 Ma, the plume edge will cover 150 km and reach a region with abnormally thin lithosphere, hereafter referred to as a lithosphere trap. From this moment, the velocity of the lateral plume movement will slightly decrease, as shown by the upper curve (Fig. 3). The decrease in the horizontal component of the velocity is conditioned by the appearance of its vertical component. The plume matter starts rising, but the rise is less significant than it was suggested. The amplitude of the rise is determined by the structure of the mantle flow in the trap, which is characterized by the presence of an ascending vertical velocity component near one edge, a horizontal current in the central part, and descending convection movements near the other edge (Fig. 4). Cooling of the plume matter now occurs much faster than in the first case owing to the convective movements which involve the plume. The anomaly of the plume temperature dissipates almost completely after 10 Ma. The region of descending convective movements in the trap will shift to the right, leaving the anomalous zone. As seen from Fig. 4, the plume matter involved in the convection causes heating of the descending branch of the convection and equalization of lateral temperature variations in the mantle region under consideration. Near the opposite edge of the plume, an intense small-scale convection with a counterclockwise matter movement is determined by the existence of a lateral temperature gradient between the relatively cold edge of the lithosphere of the Precambrian platform and the hot plume. This cell distorts the left edge of the plume and then vanishes as the plume anomaly dissipates.
DISCUSSION
|
Fig.5. Evolution of temperature distribution in the area of a lithosphere trap with a plume (white yellow). |
We start our discussion with remarks on the obtained structure of thermal convection in the upper mantle beneath a continental plate of varying thickness. Considering the convection structure in general, we should emphasize that it differs fundamentally from that in the oceanic upper mantle. In the latter, upwellings of global convection gravitate to areas with the thinnest lithosphere, and downwellings, to subduction areas, where the thickness of the oceanic lithosphere is known to reach its maximum. The inverse situation is beneath continents. Stable upwellings are detected beneath the thick lithosphere, whereas areas of descending convection movements form beneath the thinnest areas of the continental lithosphere. This difference in convection patterns is likely to be explained by the nature of the lithosphere in these areas. In oceans, convection creates and determines the lithosphere structure, whereas on continents the convection structure conforms to the existing lateral irregularities of the upper mechanic conductive layer (lithosphere). In spite of the fact that thin lithosphere relates to the areas of descending convective flows, it is here that plumes can be expected, because descending movements in the area of a lithosphere trap form sublithospheric horizontal flows transporting the rising plume matter to the trap within at least 500700 km. Thus, laterally extended, relatively narrow zones of thin continental lithosphere, favorable for plume concentration, could have been places where continents split. This may have taken place in the Cretaceous, where the South Atlantic was formed.
|
Fig.6. Temporal changes in temperature throughout a vertical section near the left edge of a lithosphere trap with a plume (the profile location is shown in Fig.5). |
As seen from the calculations, the velocity of plume transport by convection is twice as high as that of gravity current. Arriving at the trap, the plume distorts its thermal structure. Fig. 5 presents the temperature evolution in the area of a lithosphere trap, and Fig. 6 presents the time variation in temperature throughout a vertical section near the left edge of the structure. Remember that the bottom of the lithosphere in the trap is located at a depth of 100 km. Before the plume arrival, its temperature did not exceed 1130 °C. The most dramatic change in the thermal state of the area under consideration happened two million years after the plume arrival. While the temperature of the lithosphere bottom rose only by 30 °C, the temperature of the matter at depths of 130150 km increased by more than 200 °C. For the next 2 Ma, the temperature rose to the maximum value of 1550 °C. Then the area began to cool, which was related to the dissipation of the plume anomaly. Six million years after the plume had arrived at the trap (this geotherm is not shown in Fig. 6 to make the plot more illustrative), the temperature distribution in it was close to that for 2 Ma. After the next 15 Ma, the plume matter cooled almost to the mantle temperature, and the former plume can be inferred from conductive heating of this area by 100 °C relative to the initial geotherm. On the other hand, the maximum rise in temperature at the bottom of the lithosphere, to 1220 °C, occurred at that time. This thermal state of the trap will last for the next 1520 Ma, until the final equalization of temperature in the area. During this period, the plume will reach the crust, but the maximum rise in the temperature of its bottom will be no more than 50 °C. The situation, however, may develop in another way. Taking into account the model for the separation of the plume head from its tail on crossing the 660 km discontinuity and rise of the second plume after 1520 Ma, it is suggested that the new plume will cause a more intense heating of the area. This may be reflected in the evolution of magmatism in the area and in a more intense heating of the entire lithosphere, including the crust.
In addition to evolutionary isotherms, Fig. 6 presents curves for mantle-matter melting. The "dry melting" curve was calculated by the method described in [31], and the "wet melting" curve was taken from [55]. Comparison of the geotherms and melting curves demonstrates that 24 Ma after the plume appearance at the lithosphere bottom, the temperature of the trapped matter rises to such a great degree that it reaches the "wet melting" curve. It is supposed that since then, chambers of deep basalt magmatism have existed there. It is worth noting that, contrary to the expectations, the plume matter, coming to an area of thin lithosphere, failed to rise to its bottom, to a depth of 100 km. This is explained by the presence of an intense horizontal flow in the center of the trap. The resulting movement of the hot matter proved to be subhorizontal, because the positive buoyancy of the plume was insufficient to overcome the horizontal convective drift. Thus, the performed modeling claims for updating of the qualitative "thinspot" model described in [33], because the interaction between the plume matter spreading along the lithosphere bottom and convection was not taken into account. The "thinspot" model must be refined as follows: (a) the plume matter is carried by convection twice as fast as by simple spreading; (b) after reaching the trap, the plume does not rise immediately to the bottom of the thin area of the lithosphere; (c) the plume matter undergoes decompressional melting only near the edge of the area of the thin lithosphere; (d) the degree of the plume-matter melting is estimated to be no more than 13%. In McKenzie's opinion, this is sufficient for the formation of magma chambers of alkali basalts [56].
|
Fig.7. An outline of the structure of the East-Australian
folded assemblage (according to [62, 63]) and location of a hot spot
(plume) on the northward movement of the Australian plate during the
last 35 Ma [59]. |
A geological illustration for the model described in our study is provided by the peculiarities of Cenozoic basaltic magmatism in eastern Australia. It is quite possible that the plume transport by convection is the main way of formation of the geometrically intricate chain of Cenozoic central-type volcanoes in this region. They differ from the lava-field basalts here in becoming younger southward and are arranged in two submeridional lines spaced latitudinally 250 km apart (Fig. 7) [57, 58]. Several mechanisms were proposed for the formation of these volcano chains. They include a model for intraplate extension and passive rise of the mantle [59], simultaneous existence of several individual plumes (the "plume swarm" model) [60], and the above-mentioned "thinspot" model [58]. Although the approaches are different, nearly all the authors relate the manifestation of volcanism of this type to the northward movement of the plate, starting from 35 Ma BP. With the first, somewhat extravagant, model set aside, we dwell on the plume mechanism, more popular at present. The "thinspot" model seems to be more appropriate; however, it does not provide the necessary velocity of plume movement. As for the "plume swarm" model, it is difficult to imagine closely spaced stationary plumes. The laboratory modeling of plumes demonstrated that the distance between the neighboring plume heads was no less than their size [61]. For a plume with a head 200 km in radius, which is close to its minimum size, this distance should be no less than 400 km, which is twice as much as the distance between the volcano chains.
In the subsequent discussion we assume that thermal convection beneath Australia plays a crucial role in the eastward shift of the plume beneath the moving plate. The western part of the lithospheric structure of the eastern half of Australia includes the ancient Australian craton, bordering the younger Devonian fold belts Thompson and Lachlan, forming the bottom of a modern artesian sedimentary basin. The eastern shore of Australia is formed by the New England fold belt, running in a 300 km wide stretch along the whole shore. This belt is a subductional complex of an active margin, whose formation ended in the Late Triassic [62, 63]. It is reasonable to suggest that a distinctive feature of the belt is the dramatic thinning of its lithosphere caused by the presence of a mantle wedge of an ancient subduction zone. The plateau Altiplano in South America or the plateau Tibet in the south of Central Asia, where the lithosphere thickness above the wedge was geophysically estimated to be 7090 km, are modern analogs of such structures. As suggested in [64, 65], after the subduction has completed, the matter of the mantle wedge is substituted by the mantle matter, and the lithosphere structure becomes stable. This way of formation of a lithosphere trap provides an alternative to the spreading mechanism discussed in [33]. Thus, the lithosphere of eastern Australia along the latitudinal profile can be represented, as a first approximation, by a conductive, highly rigid layer of varying thickness, whose structure is similar to that of the lithosphere in the above-considered model for thermal upper-mantle convection. The size and location of the plume in the model were also chosen with regard to the situation in eastern Australia. Thus, the results of the modeling, first of all, the velocity of the plume transfer by convection, can be directly related to real settings. Figure 7 demonstrates that in the period t > 27 Ma the plume was located immediately beneath the area of thin lithosphere in the ancient subduction zone, and the northern volcano chain formed along the eastern margin of the plume just after its rise to the lithosphere bottom. Owing to the rapid horizontal northward movement of the Australian plate and the supposed stationary location of the site of plume arrival at t < 25 Ma, the plume appeared beneath the lithosphere of sedimentary basin, whose thickness is 150 km. The rate of the head movement by convection was estimated from the model to be 67 cm/year. Hence, in 2 Ma, the plume, moving eastward, will reach an area with thin lithosphere. The path of the plume relative to the plate, corrected for the northward plate shift, is shown in Fig. 7 by a dotted line. The appearance of the plume in the trap is accompanied, in addition to magmatism, by heating and isostatic rise of the lithosphere. These effects were estimated to have manifested themselves for 1520 Ma. Thus, the pronounced relief and the intense heat flow in the southern half of the New England fold belt [66] may well be related to the effect of the plume on the lithosphere of the region. For the northern volcano chain, the rise of the plume directly to the trap causes more intense heating and, hence, a higher melting degree as compared with the southern chain, where the cooling plume is transported by convection. The different melting degrees should manifest themselves in geochemical and isotopic features of the effused lavas. This can be verified when new relevant information will be available.
We thank Drs. I.V.Ashchepkov, A.G.Vladimirov, and G.G.Chernykh for valuable discussion and comments.
The study was supported by grants 96-05-65970 and 96-05-66227 from the Russian Foundation for Basic Research.
REFERENCES
[1] J.T.Wilson, Can. J. Phys., vol. 41, p. 863, 1963.
[2] X.LePichon, J. Geophys. Res., vol. 73, p. 3661, 1968.
[3] W.J.Morgan, Nature, vol. 230, p. 42, 1971.
[4] D.P.McKenzie, J.M.Roberts, and N.O.Weiss, J. Fluid Mech., vol. 62, p. 465, 1974.
[5] N.-C.Nataf, Tectonophysics, vol. 187, p. 361, 1991.
[6] L.P.Zonenshain and M.I.Kuz'min, Paleogeodynamics [in Russian], Moscow, 1993.
[7] N.L.Dobretsov and A.G.Kirdyashkin, Deep geodynamics [in Russian], Novosibirsk, 1994.
[8] F.M.Richter, Rev. Geophys. Space Phys., no. 11, p. 223, 1973.
[9] E.M.Parmentier and D.L.Turcotte, Phys. Earth Pl. Inter., vol. 17, p. 281, 1979.
[10] G.F.Davies, Geophys. J., vol. 98, p. 461, 1989.
[11] D.A.Yuen and L.Fleitout, Nature, vol. 313, p. 125, 1985.
[12] R.L.Hill, Earth Planet. Sci. Lett., vol. 104, p. 398, 1991.
[13] R.W.Griffiths and L.H.Campbell, J. Geophys. Res., vol. 96, p. 18295, 1991.
[14] D.E.Loper, Tectonophysics, vol. 187, p. 373, 1991.
[15] B.C.Storey, Nature, vol. 377, p. 301, 1995.
[16] R.S.White and D.McKenzie, J. Geophys. Res., vol. 100, p. 17543, 1995.
[17] J.A.Whitehead, Geophys. J. R. Astron. Soc., vol. 70, p. 415, 1982.
[18] M.A.Richards and R.W.Griffiths, Nature, vol. 342, p. 900, 1989.
[19] D.Bercovici and J.Mahoney, Science, vol. 266, p. 1367, 1994.
[20] M Monnereau, M.Rabinowich, and E.Arquis, J. Geophys. Res., vol. 98, p. 809, 1993.
[21] N.M.Ribe and U.R.Christensen, J. Geophys. Res., vol. 99, p. 669, 1994.
[22] M.A.Richards, R.A.Duncan, and V.E.Courtillot, Science, vol. 246, p. 103, 1989.
[23] J.PhippsMorgan, W.JasonMorgan, and E.Prise, J. Geophys. Res., vol. 100, p. 8045, 1995.
[24] D.Bercovici and J.Lin, J. Geophys. Res., vol. 101, p. 3291, 1996.
[25] D.McKenzie, Geophys. J. R. Astron. Soc., vol. 48, p. 211, 1977.
[26] B.Parsons and S.Daly, J. Geophys. Res., vol. 88, p. 1129, 1983.
[27] D.McKenzie and R.K.O'Nions, Nature, vol. 301, p. 229, 1983.
[28] D.McKenzie and R.K.O'Nions, J. Petrol., vol. 36, p. 133, 1995.
[29] Yu.A.Zorin, M.R.Novoselova, E.Kh.Turutanov, and V.M.Kozhevnikov, J. Geodyn., vol. 11, p. 327, 1990.
[30] D.L.Anderson and J.Polet, Geology, vol. 23, p. 205, 1995.
[31] D.McKenzie and M.J.Bickle, J. Petrol., vol. 29, p. 625, 1988.
[32] T.H.Jordan, J. Petrol., Special Issue, p. 11, 1988.
[33] R.N.Thompson and S.A.Gibson, J. Geol. Soc., London, vol. 148, p. 973, 1991.
[34] S.A.Gibson, R.N.Thompson, O.H.Geonardos, et al., J. Petrol., vol. 36, p. 189, 1995.
[35] V.P.Trubitsyn, Yu.F.Belavina, and V.V.Rykov, Fizika Zemli, no. 11, p. 3, 1993.
[36] J.P.Lowman and G.T.Jarvis, Geophys. Res. Lett., vol. 20, no. 19, p. 2087, 1993.
[37] A.M.Bobrov and V.P.Trubitsyn, Fizika Zemli, no. 7, p. 5, 1995.
[38] L Guillou and C.Jaupart, J. Geophys. Res., vol. 100, p. 24217, 1995.
[39] D.McKenzie and N.Weiss, Geophys. J. R. Astron. Soc., vol. 42, p. 131, 1975.
[40] S.A.Tychkov, Convection in the mantle and dynamics of platform regions [in Russian];, Novosibirsk, 1984.
[41] K.Bell, J.Blenkinsop, T.J.S.Cole, and D.P.Menagh, Nature, vol. 298, p. 251, 1982.
[42] W.L.Griffin, F.Kaminsky, S.Y.O'Reilly, et al., in: Extended abstracts of Sixth International Kimberlite Conference, Novosibirsk, p. 194, 1995.
[43] P.G.Silver and W.Chan, Nature, vol. 335, p. 134, 1988.
[44] M.Mareschal, R.L.Kellet, R.D.Kurtz, et al., Nature, vol. 375, p. 134, 1995.
[45] W.Fjeldskaar, Earth Planet. Sci. Lett., vol. 126, p. 399, 1994.
[46] U.Christensen, Geophys. J. R. Astr. Soc., vol. 77, p. 343, 1984.
[47] S.Karato, Nature, vol. 347, p. 272, 1990.
[48] A.D.Duchkov, V.T.Balobaev, B.V.Volod'ko, et al., Temperature, cryolithozone, and radiogene generation in the Earth's crust of Northern Asia [in Russian], Novosibirsk, 1994.
[49] L.Fleitout and D.A.Yuen, J. Geophys. Res., vol. 89, p. 9227, 1984.
[50] N.P.Moshkin, E.V.Rychkova, S.A.Tychkov, and G.G.Chernykh, Vychislitel'nye Tekhnologii, no. 4, p. 224, 1995.
[51] B.Travis and P.Olson, Geophys. J. Inter., vol. 118, p. 1, 1994.
[52] C.G.Farnetani and M.A.Richards, J. Geophys. Res., vol. 99, p. 13813, 1994.
[53] P.E.VanKeken and C.W.Gable, J. Geophys. Res., vol. 100, p. 20291, 1995.
[54] H.E.Huppert, J. Fluid Mech., vol. 121, p. 43, 1982.
[55] H.Schmeling and G.Y.Bussod, J. Geophys. Res., vol. 101, no. B3, p. 5411, 1996.
[56] D.McKenzie, Earth Planet. Sci. Lett., vol. 95, p. 53, 1989.
[57] P.Wellman and I.McDougall, Tectonophysics, vol. 23, p. 49, 1974.
[58] S.Y.O'Reilly and M.Zhang, Contrib. Mineral. Petrol., vol. 121, p. 148, 1995.
[59] R.H.Pilger, Jr., J. Geophys. Res., vol. 87, no. B3, p. 1825, 1982.
[60] F.L.Sutherland, J. Volcanol. Geotherm. Res., vol. 9, p. 181, 1981.
[61] J.A.Whitehead and D.S.Luther, J. Geophys. Res., vol. 80, p. 705, 1975.
[62] V.E.Khain, in: Non-Alpine Asia and Australia [in Russian], Moscow, 1979.
[63] W.J.Collins and R.H.Vernon, Tectonophysics, vol. 235, p. 249, 1994.
[64] V.E.Khain, S.A.Tychkov, and A.G.Vladimirov, Collision orogeny: a model for the detachment of a subducted oceanic lithosphere plate as a result of continental collision, Geologiya i Geofizika (Russian Geology and Geophysics), vol. 37, no. 1, p. 5(3), 1996.
[65] S.A.Tychkov and A.G.Vladimirov, Dokl. RAN, vol. 354, no. 2, p. 238, 1997.
[66] J.R.Cull and D.Conley, BMR J. Australian Geology & Geophysics, vol. 8, p. 329, 1983.
Recommended by S.V.Gol'din
1 April 1997