Small-Scale Mantle Convection Produces Stratigraphic Sequences in Sedimentary Basins

See allHide authors and affiliations

Science  13 Aug 2010:
Vol. 329, Issue 5993, pp. 827-830
DOI: 10.1126/science.1190115

Changes in the Rocks

Changing sea level or major tectonic events, such as continental collisions, shift stratigraphic sequences by changing the depositional environment where certain rock types form. For example, a deep marine environment where limestone formation is favored may shift relatively quickly to a near-shore environment favoring sandstone formation because the relative sea level has dropped several meters. Petersen et al. (p. 827; see the Perspective by Müller), however, suggest that small-scale convection in the mantle may also induce appreciable changes in the sequence of sedimentary deposits. Using a modeling approach, they found that this is possible on a small scale (that is, just a few hundreds of kilometers) over variable time scales. Thus, while the co-occurrence of sedimentary deposit sequences at regional and global scales can allow sedimentary rocks to serve as markers of marine environments, it should be kept in mind that local changes in surface movements may also manifest themselves in the rock record.


Cyclic sedimentary deposits link stratigraphic sequences that are now geographically distant but were once part of the same depositional environment. Some of these sequences occur at periods of 2 to 20 million years, and eustatic sea-level variations or regional tectonic events are likely causes of their formation. Using numerical modeling, we demonstrate that small-scale mantle convection can also cause the development of stratigraphic sequences through recurrent local and regional vertical surface movements. Small-scale convection-driven stratigraphic sequences occur at periods of 2 to 20 million years and correlate only at distances up to a few hundred kilometers. These results suggest that previous sequence stratigraphic analyses may contain erroneous conclusions regarding eustatic sea-level variations.

Stratigraphic sequences are sedimentary units that reflect cyclic changes in deposition rate and depositional environment due to variations in relative sea level (1). It is commonly assumed that such variations are due to global (eustatic) sea-level changes that thus provide a general framework for the correlation of stratigraphic sequences globally or regionally (2). However, the global universality of sequence stratigraphic events has not been proven unambiguously, and regional variations in relative sea level may be caused by a number of other processes (3, 4). For recurrent cycles occurring at periods of 2 to 20 million years (My) [hereafter defined as intermediate period (IP)], no plausible alternative to eustasy has yet been proposed as the cause of sequence development. Small-scale convection in the sublithospheric mantle may be a possible explanation for the formation of such recurrent stratigraphic sequences, but its role is relatively unknown. Although the specific pattern of large-scale mantle convection is debated (5), there seems to be little doubt that convection occurs at two distinct temporal and spatial scales: (i) the creation and subduction of tectonic plates and possibly the movement of plumes originating from the base of the mantle (5) and (ii) small-scale convection operating on regional scales (611). Depending on the amount of crustal heat production (12), small-scale convection (6) must deliver heat at a rate of about half of a typical continental heat flux of about 60 mW m−2 at the base of the lithosphere.

To assess the associated amplitudes and periods of vertical movements of Earth’s surface in response to small-scale convection, we used a numerical two-dimensional (2D) thermomechanical model of the upper 660 km of Earth’s crust and mantle (Fig. 1) (13). The model converges over several hundred My to a near-equilibrium state (fig. S2), characterized by a relatively cold, nondeforming lithosphere of roughly constant thickness overlying an upper mantle with an adiabat of ~0.6°C/km, transporting ~30 mW m−2 (Fig. 1C). At the base of the lithosphere and at the bottom of the model, thermal boundary layers with higher temperature gradients are maintained. The convective system receives and delivers heat by conduction.

Fig. 1

Snapshot of convecting numerical model of the crust and the upper mantle. (A) Temperature. Black curve, horizontally averaged model temperature; gray curve, theoretical isentropic temperature profile for a potential temperature of 1315°C. (B) Temperature anomaly (deviation from horizontal average). (C) Vertical conductive heat flux and its horizontal average. (D) Effective viscosity and its horizontal average. The thermal boundary layer below the lithosphere is dominated by conductive heat flow.

The model provides a self-consistent representation of the lithosphere and upper mantle, including the asthenosphere, satisfying established knowledge regarding rheology and heat transfer. Small variations within experimentally determined limits for creep parameters in dry olivine (14) allowed us to control the mantle heat flux and the resulting lithospheric thickness. When the lowest possible experimentally constrained value of activation volume for dislocation creep is assumed and the pre-exponential factor of diffusion creep is varied within experimental limits (table S1), the thermal state varies from no convection to vigorous convection transporting heat at a rate of more than 50 mW m−2. Diffusion and dislocation creep occur simultaneously, and the latter dominates in the upper part of the upper mantle. We chose the pre-exponential factor for diffusion creep, so that the transition between dislocation and diffusion creep is consistent with the depth at which seismic anisotropy vanishes (15) and so that the advected heat flux is ~ 30 mW m−2.

In this model, discrete upwellings form at the basal thermal boundary layer and rise rapidly until they slow down because of viscous drag in the boundary layer at the base of the lithosphere (Fig. 1D). Here they spread out laterally while dissipating heat energy, which ultimately diffuses, with the thermal time constant of the lithosphere (~60 My in the present example), to the surface (Fig. 1B). The arrival of a buoyant upwelling at the base of the lithosphere produces a relatively rapid rise of the surface. Mass conservation means that the arrival of an upwelling requires the displacement of material from the boundary layer. This is accommodated by colder, negatively buoyant downwellings that sink into the asthenosphere. The surface expression of this process is subsidence driven by the slow densification of the cooling lithosphere, followed by rapid surface uplift as material gathered from the basal boundary layer suddenly detaches and sinks.

The time scale and amplitude of convection-induced surface movements depend on the chosen creep parameters [within observational constraints (14)] because these affect the scale of convective cells and the lithospheric thickness, and thus its effective elastic thickness. However, as long as the creep parameters result in convection, surface movements occur.

Issues relating to the effects of plate movement, long-term convective stability, and the time and length scale of buoyant loads and resulting vertical surface movements await resolution by 3D high-resolution numerical simulations. For example, the convective pattern may be modified by relative movement between the lithosphere and the asthenosphere, resulting in a 3D convective pattern with Richter rolls parallel to plate movement (6, 9, 10). Nevertheless, comparisons of 2D and 3D convection models have indicated that the resulting thermal state of the lithosphere remains similar in either case (16).

For the present model, vertical surface movements occur with periods spanning two orders of magnitude and with different amplitudes. The long-term component, with periods of up to several hundreds of My and amplitudes of more than 300 m (with no air or sediment filling the evolving depressions), correlates closely with the thermal state of the lithosphere as represented by the surface heat flux (Fig. 2, A and B). These slow changes are caused by transitions between different states of quasi-static long-term organization of convection cells, which last long enough to affect the thermal state of the lithosphere. Although it has been suggested that whole-mantle convection may be required to explain regional stratigraphic events of this time scale (1719), the present results show that convection in the upper mantle may also contribute.

Fig. 2

Temporal evolution of surface heat flux and surface displacements due to small-scale convection. (A) Surface heat flux. (B) Surface displacement. (C) Surface displacements of periods less than 50 My. (D) Correlograms of surface displacements (13). Long-period surface displacements correlate with the thermal structure of the lithosphere. Short-period displacements are caused by buoyancy variations in the thermal boundary layer at the base of the lithosphere.

Faster surface movements with periods of ~2 to 20 My and amplitudes of ~30 m are related to the erratic movements of individual up- and downwellings and their interactions in the thermal boundary layer (Fig. 2C). The buoyancy-induced movements are filtered by the viscoelasticity of the lithosphere (12); a stiffer lithosphere (larger effective elastic thickness) controls the response to relatively fast changes in buoyancy distribution in the thermal boundary layer, whereas a weaker lithosphere (smaller effective elastic thickness) controls the response to slower changes (fig. S3). The short-period movements therefore are correlated at larger distances (some 250 km) than are the longer periods (up to 200 km) (Fig. 2D) (fig. S3).

We suggest that these patterns of surface movements may be the overriding control on the formation of IP stratigraphic sequences that are observed ubiquitously in sedimentary basins and on continental margins and platforms (1). The surface movements cause changes in relative sea level, controlling patterns of erosion and deposition. Stratigraphic sequences are most easily identified on the margins of sedimentary basins, where sea-level changes have substantial effects on sediment lithofacies distribution and sedimentary architecture and hence facilitate the interpretation of stratigraphic sequences by means of seismic and borehole data (1, 20).

Within a quasi-static convective state, individual up- and downwellings occur at time scales of less than 20 My. The resulting vertical surface movements generally show time-transgressive lateral correlations due to horizontal mass transport in the thermal boundary layer (Fig. 2C). Allowing for diachroniety in the range of 2 to 5 My, as may be the case when correlating via condensed sequences of basin centers using seismic data (2), many of the events of Fig. 2C are in fact correlatable at distances up to 800 km (fig. S7).

Sedimentary basins are often formed on continental margins and in intracratonic rifts. With an imposed extension of 60 km, our model produces a rift basin 300 km wide, exhibiting initial fault-controlled subsidence followed by thermal and loading-induced subsidence (Fig. 3) (movie S1) (21). However, the thermal decay time is longer than in conventional model predictions, which do not include the feedbacks between advective heat transport and lithospheric thinning (22). Furthermore, convection-induced effects, which result in periods of enhanced and decreased subsidence, complicate the post-rift subsidence pattern. They are stimulated by the rifting process and comprise discontinuous axial upwellings and, below the rift flanks, erosion of the base of the lithosphere (23). This causes the sedimentary fill to exhibit repeated patterns of transgression and regression, unconformities, and lateral migration of sedimentary facies (Fig. 3, A and D).

Fig. 3

Rift model with small-scale convection and constant sediment input and eustatic sea level. (A) Lithofacies (indicated by paleo water depth). Cyclic deposition is evident from unconformities and lateral migration of lithofacies; subsidence curves for locations a and b appear in (D). (B) Lithospheric cross-section showing gently thinned crust and mantle lithosphere and isotherms from 1200° to 1440°C (solid lines). (C) Cross section of the Sverdrup Basin, Canada (29). (D) Chronostratigraphic diagram and short-term vertical basement displacements at locations a and b. The inferred Jurassic global onlap curve (2) is shown in black for comparison. Differences in local vertical movements cause the two opposite basin margins to exhibit asymmetric stratigraphic evolution with different onlap patterns. In both cases, the time scale of cyclicity resembles that of the global Jurassic onlap curve.

A widely held assumption regarding the cause of sea-level changes that control IP stratigraphic cycles is that it is of eustatic origin (2, 24). This widespread view, combined with difficulties in dating rocks and horizons within them, has led to a situation in which sequence boundaries that are well dated in one place are used to date the same sequence boundary where it can be identified and correlated elsewhere (25). However, a mechanism driving globally synchronous eustatic variations with this periodicity has not been demonstrated, and, in any case, global synchronicity of such cycles is controversial (3, 4, 17, 25).

Besides glacio-eustasy, the only mechanism proposed for generating IP stratigraphic cycles is one of changing tectonic stresses within the crust/lithosphere system, giving rise to vertical motions of the depositional surface from changes in its isostatic (mechanical) equilibrium position (26). Sequence boundaries in many sedimentary basins have been inferred to be clearly related to regional tectonic events (3, 27). However, it has never been demonstrated that there should be a systematic oscillation in tectonic stresses in plates with a 2- to 20-My periodicity. On the other hand, there is plentiful evidence to suggest that this occurs at longer and shorter periods, at the scale of 20- to several hundred-My sequences—due to supercontinental cycles and volume changes in oceanic spreading centers—as well as shorter periods of less than 1 My due to glacial eustasy (17, 25, 28). Models in which changing tectonic stresses (26) are invoked to explain IP cycles are thus ad hoc regarding the actual process causing periodic stress changes.

In the model presented here, however, the ubiquitously occurring IP cycles are a natural consequence of the processes that produce and control the evolution of sedimentary basins themselves through the transport of heat out of Earth by small-scale convection. The induced relative sea-level variations are regionally correlatable and are synchronous or nearly synchronous over spatial scales that encompass the dimensions of most shelf-slope systems of sedimentary basins, where sequences are most readily identifiable. This poses a challenge for the use of sequence stratigraphy in regional and global correlation as well as in the construction of global sea-level records. If IP sequences are indeed controlled by sublithospheric convection, they can only be considered locally for dating sedimentary successions. Our study thus provides an explanation for the enigmatic control on IP cycles, between the shorter-period variations caused by Milankovich cycles and those caused by regional in-plane stress changes and other tectonic and nontectonic contributions, including glacio-eustatic sea-level variations.

Supporting Online Material


Figs. S1 to S7

Table S1


Movie S1

References and Notes

  1. See supporting data on Science Online.
  2. We thank M. Huuse, E. Thomsen, N. Christie-Blick, and three anonymous reviewers for constructive suggestions and D. L. Egholm for providing computational facilities.
View Abstract

Navigate This Article