On the relation between cratonic lithosphere thickness, plate motions, and basal drag
Seismic, thermal, and petrological data indicate that the local maximum thickness of
Precambrian lithosphere is highly variable (140-350 km), with a bimodal distribution for
Archean cratons (200-220 km and 300-350 km). We discuss the origin of such large
differences in lithospheric thickness, and propose that the lithospheric base can have large
depth variations over short distances. We find a linear correlation between the horizontal and
vertical dimensions of Archean cratons: larger cratons have thicker lithosphere with the
“critical” surface area of >6-8 x106 km2 for cratons to have  deep (>300 km) roots.
We evaluate the basal drag model as a possible mechanism that may thin the cratonic
lithosphere. Inverse correlations are found between lithospheric thickness and (a) fractional
subduction length and (b) the effective ridge length. In agreement with the basal drag model,
we find that lithospheric thickness of Archean cratons is proportional to the square root of the
ratio of the craton length (along the direction of plate motion) to the plate velocity. Large
cratons with thick keels and low plate velocities are less eroded by basal drag than small fast-
moving cratons.
Click on images to enlarge them

1. Lithospheric thickness.
We summarize seismic and thermal data that show that the
thickness of Archean cratonic lithosphere ranges from 140 to
~350 km. As petrological data contradict tomographic and
thermal models, we propose that the lithospheric base can have
large thickness variations over short distances and different
techniques sample different parts of it. The topography of Bryce
Canyon (western USA) may be an inverted analog of such an
eroded structure
The “Bryce Canyon” model of the lithospheric structure. The lithospheric
undulating boundary due to a heterogeneous erosion of the lithospheric
base by an interaction with mantle plumes and its non-uniform
destruction by convection flow due to compositional inhomogeneities.
Xenolith data sample the most shallow parts of the lithospheric base
(200-250 km), where pressures are low enough to initiate low-percentage
melting and produce kimberlite-type magmatism. Seismic tomography
samples the top of the convective mantle, i.e. the lowermost parts of the
lithospheric base. A diffuse character of the seismic lithospheric base,
especially based on surface waves, supports the model. Thermal data
provide a smoothed integrated picture of the lithospheric structure and
thus give values of the thickness of cratonic keels between xenolith and
seismic estimates. The right part of the figure gives an example of the
lithospheric thickness estimates for the Baltic Shield and the Tanzanian
craton (compare with Fig. to the left).
2. Vertical versus lateral dimensions of the cratons.
We use estimates of lithosphere thermal thickness (Artemieva and Mooney, 2001) to examine the correlations between
cratonic size, plate velocities, and lithospheric thickness. The results suggest that the horizontal and vertical dimensions of the
Archean cratons are well correlated: larger Archean cratons have thicker lithosphere. However, this correlations does not hold
for middle-late Proterozoic lithosphere. The extrapolation of the linear trend for cratonic size versus lithospheric thickness to
the total size of all present-day Archean cratons (hypothesized to have formed an early Archean supercontinent) suggests that
the ancient (~4.0 Ga) lithosphere may have been ~450 km thick.
Lithosphere thermal thickness in Archean cratons versus
area of Archean (A), (B) middle-late Proterozoic, (C)
early-late Proterozoic, and (D) combined cratons. Plots A
and D show that Archean cratons with larger size have
thicker lithospheric keels. No correlation exists for
middle-late Proterozoic cratons (B), because lithospheric
thickness is more uniform.
Lithosphere thermal thickness in the Archean cratons versus craton area. The
extension of the linear trend in Figure A (left) provides an estimate of the thickness
of the lithosphere in the early Archean. Assuming that the area of the oldest (e.g.
~4.0 Ga) hypothesized Archean supercontinent was equal to the total area of the
present Archean cratons, its lithospheric thickness could have been 350-450 km.
Similarly, lithospheric thickness at ~550-500 Ma (when Gondwanaland was formed) is
estimated to be about 280-400 km. Black dot shows estimated lithospheric
thickness in the Slave Craton at the time of Gondwanaland (Pokhilenko et al., 2001).
Cartoon depicting the process of erosion of cratonic lithosphere.
An early Archean supercontinent with a keel down to the mantle transition zone (400-450 km) is split by a mantle plume into two parts with non-equal
dimensions. The fate of these supercontinent fragments is determined by their lateral dimensions. The larger craton diverts the mantle heat from its base
(Ballard and Pollack, 1987) and is mostly affected by secondary convection on its margins (Doin et al., 1997). This process promotes the preservation of
a thick keel (~350 km) in the interior of the craton, but leads to erosion of cratonic margins and thus a gradual decrease of the cratonic size.
When the area of the larger craton is reduced to a critical value of about 6-8 x106 km2 (Fig. 4A), it will start to evolve as a smaller craton. The smaller
craton is not effective in diverting the heat because of its smaller lateral dimension and is more subject to erosion from below by mantle convection until
an equilibrium lithospheric thickness of ~220 km is reached.
4. Basal drag model: correlation of plate velocities with lithospheric thickness and cratonic area.
We evaluate the basal drag model, whereby the lithosphere is eroded due to its relative movement with respect to the
underlying mantle (Sleep, 2001), and find that for Archean cratons this model is in excellent agreement with observed data:
lithospheric thickness is proportional to the square root of the ratio of the craton length (along the direction of plate motion) to
the plate velocity. Large cratons with thick keels and low plate velocities (e.g., Eurasia and North America) are less eroded by
basal drag than fast-moving small cratons (e.g., India and Australia). This means that earlier studies (Forsyth and Uyeda,
1975; Stoddard and Abbott, 1996)  have addressed only one aspect of a more complicated relationship, whereby both craton
size and plate velocity correlate with the lithospheric thickness: large Archean cratons tend to have thick lithospheric keels and
very slow plate velocities. We emphasize that these correlations hold only for the Archean cratons, not for middle-late
Proterozoic cratons; for early Proterozoic cratons the correlation is very weak.
Boxes correspond to different cratons. The vertical dimensions of
the boxes show range of lithospheric thickness and the horizontal
dimension shows the error in plate velocity estimates. The plate
velocities are for the hotspot reference frame (Gripp and Gordon,

The data for the Archean cratons supports the “basal drag” model,
which predicts the lithospheric thickness to be inverse proportional
to the square root of the plate velocity. The keels of the fast
moving plates are smaller due to erosion by basal drag. Australia,
which is located close to the subduction zone and thus has a high
plate velocity, is the only plate to plot well off the curve.

5. Relations between lithospheric thickness, slab pull, and ridge push.
show an inverse correlations between lithospheric thickness and: (a) fractional subduction length; and (b) the effective ridge
length. These results indicate that lithosphere erosion by mantle drag is proportional to the plate velocity.
6. Preservation of the lithospheric keels.
Our results suggest that the slower the plate moves the weaker is the erosion of the keel. This implies that thick Archean keels
can be preserved for a long time (i.e. 3-4 Ga). Cratons with large sizes are also more stable with respect to basal erosion by
mantle convection due to an efficient deflection of heat from the deep mantle (Ballard and Pollack, 1987; Lenardic and Moresi,
2001). Their stability is further maintained by the depleted (e.g. Jordan, 1988) and dry (e.g. Pollack, 1986) composition of the
Archean lithosphere. However, very thick (~400 km) lithospheric keels could have survived until present probably only locally,
an observation supported by the fact that Archean cratonic lithosphere thickness rarely exceeds 300-350 km.
7. Reworking of Archean keels in Proterozoic.
When erosion due to secondary convection at the margins of a thick (~350 km), large craton reduces its lateral dimension to a
critical value of ~6-8 x106 km2, the keel fails to divert the basal heat efficiently. In this case, the Archean keel is thinned by
mantle convection to an equilibrium thickness of ~ 220 km. Since all Proterozoic cratons have lithospheric keels that are less
than 200 km thick, we infer that only thinned (~220 km) Archean lithosphere is a candidate for the geologically known
reworking into Proterozoic lithosphere.

Due to the viscosity-depth structure of the upper mantle, thinning of the Archean lithospheric keel will reduce basal drag and
therefore resistance to plate motion. This will permit faster movement of the craton with respect to the underlying mantle,
which, in turn, will enhance lithosphere erosion by the basal drag. If an Archean keel is thinned to significantly less than 200
km, the remaining lithospheric column will be a candidate for strong deformation in a collisional environment and for
modification by metasomatism (i.e. invasion by volatiles and relatively enriched mantle magmas).

8. Secular cooling of the mantle and mantle drag.
We recognize that the basal drag model suggests a long-term preservation of thick (300-350 km) Archean keels only if the
cratons have never experienced a period of high plate velocity. In view of the long (~4 Ga) existence of the keels, this scenario
seems unlikely. Though it has been suggested that plate velocities in Archean could have been even slower than at present, it
is difficult to imagine that the Archean cratons of West Africa, Baltica, and Siberia were never a part of a fast-moving plate.
However, if the viscosity of the mantle to a depth of ~450 km was one or two orders of magnitude lower during the Archean,
corresponding to mantle temperature some 100-200oC higher than today, the basal drag, and along with it, basal erosion
would have been much smaller than today. Thus, the thick keels of Archean cratons would have been preserved, even if
Archean plate velocities were high.
web site traffic counters
Last modified December, 2013; irina@geo.ku.dk

3. Interaction between plate motion and lithospheric keels.
We address the question of how  early Archean lithosphere that may have been ~450 km thick was eroded to its present-day
bimodal thickness of 200-220 and 300-350 km (Doin et al., 1997; Artemieva and Mooney, 2001). One would also expect that
such deep lithospheric keels could influence plate motions (e.g., Chapman and Pollack, 1974). This idea is supported by the
results of Stoddard and Abbott (1996) who showed that the Proterozoic part of the keels, located within the low-viscosity
asthenosphere at the depth around 120-180 km, has a weak effect on plate movement, whilst deeper Archean keels resist
plate motion.

Conversely, plate motion also strongly influences the thickness of the Archean keels. We propose that the interaction between
plate motion and keel thickness is double-sided: for plates not attached to subducting slabs, thick Archean keels can slow
plate velocities, while plate motion erodes the keels by basal drag. The first detailed analysis of the driving forces for plate
motions (Forsyth and Uyeda, 1975) have shown the importance of mantle drag, a conclusion that is supported by the inverse
correlation between the area of Archean cratons and plate velocity (Stoddard and Abbott, 1996).
Since absolute plate velocity is proportional to the amount of
subduction pull (Forsyth and Uyeda, 1975) and increases with a
decrease of lithospheric thickness, we expect an inverse
correlation to exist between lithospheric thickness and
subduction pull. While such a trend may be present in Figure
A, an inadequate distribution of subduction length values
prevents definite conclusions, beyond general consistency with
respect to Fig. above. Likewise, a comparison of lithospheric
thickness with effective ridge push (Fig. B) provides evidence
for an inverse correlation between these two parameters. The
correlation is especially strong not for the mean, but for the
maximum values of lithospheric thickness in Archean cratons
(dashed lines). In conclusion, we find that there is an inverse
correlation between lithospheric thickness in Archean cratons
and the two main driving forces of plate motion. This implies
that while subduction pull and ridge push determine velocities
of lithospheric plate, the lithosphere itself is eroded by mantle
drag which is proportional to the plate velocity.
Artemieva I.M. and Mooney W.D.
Tectonophysics, v. 358, 211-231, 2002.
Download the article
Irina Artemieva: Research highlights
Lithosphere research: Current Projects
Thermal model,2001