features on the surface of the Earth. Menard has shown
that their crests closely correspond to median lines in the
oceans and suggests that they may be ephemeral features.”
Bullard et al. (1956) and Von Herzen (1959) show that
they have unusually high heat flow along their crests.
Historical models of plate creation
The major question posed by the papers of Hess (1962)
and Dietz (1961) (who expanded on the original idea of
Hess) on seafloor spreading was “how did the zones of
ocean floor creation relate to the places where the oceanic
crust was being destroyed and how thick was the region
that was moving?” Wilson (1965) argued that the simple
concept that the Earth is made up of rigid plates of greater
than 50 km thickness created at spreading centers and
destroyed in trenches could explain the huge deep topographic offsets observed on the mid-ocean ridges. Importantly this showed that the region that was moving was
much thicker than just the crust. Almost immediately,
Ewing et al. (1966) and McKenzie (1967) used heat flow
data to test the concept of a thick moving plate. This plate
is now called the lithosphere. Ewing et al. (1966) argued
that it did not, whereas McKenzie (1967) disagreed and
constructed a simple model of plate creation (Figure 2a)
that was compatible with the observations.
Sclater and Francheteau (1970) used this plate model to
show that the global heat flow data and the subsidence of
the ocean floor (Figure 2a), rather than being an argument
against the plate theory, as Ewing et al. (1966) had
suggested, strongly supported it. Because of the scatter
in the heat flow (left-hand side of Figure 2b), it was realized that it was more useful to plot seafloor subsidence
against crustal age. Sleep (1969) and Sclater et al. (1971)
demonstrated that there was a simple relationship for the
entire age of the ocean floor (Figure 2a) that was compatible with the creation of a plate about 100 km thick. Parker
and Oldenburg (1973) showed that for the first 80 my in
the North Pacific, there was a simple relation between subsidence and the square root of seafloor age (i.e., z /√t).
This is to be expected if the creation of the ocean floor
were to be represented by the cooling of a simple thermal
boundary layer (Turcotte and Oxburgh, 1967). Davis and
Lister (1974) showed that for crust <80 Ma, this relation
held for all three oceans considered by Sclater
et al. (1971).
Hydrothermal circulation in the oceanic crust
Values from the initial measurements near the ridge axis in
the oceans, according to Bullard et al. (1956), were highly
scattered. This prevented use of these data for quantitative
analysis (left-hand side of Figure 2b). In addition, the
averaged heat flow values near the ridge crest were much
lower than those predicted by the thermal model that best
matched the subsidence of the older ocean floor (see
Sclater and Francheteau, 1970). The problem was
resolved by Lister (1972) who showed that the observed
heat flow followed the values predicted by simple models
of fluid flow in a permeable crust where upward flow
occurred in narrow vents with the downward flow being
broad and diffuse. Immediately, it was realized that this
hypothesis explained both the variability in the heat flow
values and the low averaged values near the ridge crest.
The hydrothermal hypothesis of Lister (1972) predicted
the output of hot jets of water on the seafloor, and their discovery (Williams et al., 1974) spawned extensive interdisciplinary studies of seafloor hydrothermal circulation (see
Davis and Elderfield, 2004 for a review). Sclater
et al. (1976) showed that filtering heat flow data on young
seafloor, only to include measurements made in wellsedimented areas with no nearby basement outcrops. This
dramatically increased the heat flow to close to that
predicted the models that matched the topography. Davis
et al. (1999) and Davis and Becker (2004) documented
clearly that, over variable but well-sedimented cover, the
heat flow is directly related to the depth to basement
(Figure 2b). This can only be possible if vigorous water
circulation, occuring within the basaltic basement beneath
the sediment, maintains a constant temperature at the sediment/crust interface.
Modern thermal models
Using carefully filtered heat flow data (Sclater et al., 1976)
and specially selected depth-age provinces, Parsons and
Sclater (1977) showed (a) how extraordinarily well the
plate model fits the observational data and (b) that various
parameters of the plate could be obtained by constructing
physical models that fit the data within measurement error
(Figure 3a: top and bottom). Out of these analyses came a
realization (Parsons and McKenzie, 1978) that the lithosphere, even when represented as a flat plate, was still simply the upper boundary layer of a cooling convection
system (Turcotte and Oxburgh, 1967). Parsons and Sclater
(1977) fit the data with a plate thickness of 125 km and a
basal temperature of 1,330
C. In an analysis using a much
more complete data set, Hillier and Watts (2005) showed
that the depth data for the North Pacific filtered to remove
the effects of swells gave results very similar to that of Parsons and Sclater (1977). However, both papers matched
the heat flow data by adding a 4 mW m
À2 to account for
the effect of radioactivity within the ocean crust. Such a
large value is not supported by the overall measured radioactivity of the crust (Staudigal, 2003).
In an alternative observational approach, Stein and
Stein (1992) ignored both filtering and the selection of
special provinces and plotted all the depth and heat flow
data in the Pacific and Northwest Atlantic against age of
the ocean floor. They found a model (GDH1) that best fit
both data sets that had a plate thickness of 95 km and a bottom temperature of 1,425
C (Figure 3b). Though the
basal temperature is too high to explain the rare Earth distribution in the melted rock found at the ridge axis
(McKenzie et al., 2005), GDH1 was the first model that
actually had the predicted heat flow high enough to fit
the data over ocean floor >130 Ma without assuming the
MARINE HEAT FLOW
451
that their crests closely correspond to median lines in the
oceans and suggests that they may be ephemeral features.”
Bullard et al. (1956) and Von Herzen (1959) show that
they have unusually high heat flow along their crests.
Historical models of plate creation
The major question posed by the papers of Hess (1962)
and Dietz (1961) (who expanded on the original idea of
Hess) on seafloor spreading was “how did the zones of
ocean floor creation relate to the places where the oceanic
crust was being destroyed and how thick was the region
that was moving?” Wilson (1965) argued that the simple
concept that the Earth is made up of rigid plates of greater
than 50 km thickness created at spreading centers and
destroyed in trenches could explain the huge deep topographic offsets observed on the mid-ocean ridges. Importantly this showed that the region that was moving was
much thicker than just the crust. Almost immediately,
Ewing et al. (1966) and McKenzie (1967) used heat flow
data to test the concept of a thick moving plate. This plate
is now called the lithosphere. Ewing et al. (1966) argued
that it did not, whereas McKenzie (1967) disagreed and
constructed a simple model of plate creation (Figure 2a)
that was compatible with the observations.
Sclater and Francheteau (1970) used this plate model to
show that the global heat flow data and the subsidence of
the ocean floor (Figure 2a), rather than being an argument
against the plate theory, as Ewing et al. (1966) had
suggested, strongly supported it. Because of the scatter
in the heat flow (left-hand side of Figure 2b), it was realized that it was more useful to plot seafloor subsidence
against crustal age. Sleep (1969) and Sclater et al. (1971)
demonstrated that there was a simple relationship for the
entire age of the ocean floor (Figure 2a) that was compatible with the creation of a plate about 100 km thick. Parker
and Oldenburg (1973) showed that for the first 80 my in
the North Pacific, there was a simple relation between subsidence and the square root of seafloor age (i.e., z /√t).
This is to be expected if the creation of the ocean floor
were to be represented by the cooling of a simple thermal
boundary layer (Turcotte and Oxburgh, 1967). Davis and
Lister (1974) showed that for crust <80 Ma, this relation
held for all three oceans considered by Sclater
et al. (1971).
Hydrothermal circulation in the oceanic crust
Values from the initial measurements near the ridge axis in
the oceans, according to Bullard et al. (1956), were highly
scattered. This prevented use of these data for quantitative
analysis (left-hand side of Figure 2b). In addition, the
averaged heat flow values near the ridge crest were much
lower than those predicted by the thermal model that best
matched the subsidence of the older ocean floor (see
Sclater and Francheteau, 1970). The problem was
resolved by Lister (1972) who showed that the observed
heat flow followed the values predicted by simple models
of fluid flow in a permeable crust where upward flow
occurred in narrow vents with the downward flow being
broad and diffuse. Immediately, it was realized that this
hypothesis explained both the variability in the heat flow
values and the low averaged values near the ridge crest.
The hydrothermal hypothesis of Lister (1972) predicted
the output of hot jets of water on the seafloor, and their discovery (Williams et al., 1974) spawned extensive interdisciplinary studies of seafloor hydrothermal circulation (see
Davis and Elderfield, 2004 for a review). Sclater
et al. (1976) showed that filtering heat flow data on young
seafloor, only to include measurements made in wellsedimented areas with no nearby basement outcrops. This
dramatically increased the heat flow to close to that
predicted the models that matched the topography. Davis
et al. (1999) and Davis and Becker (2004) documented
clearly that, over variable but well-sedimented cover, the
heat flow is directly related to the depth to basement
(Figure 2b). This can only be possible if vigorous water
circulation, occuring within the basaltic basement beneath
the sediment, maintains a constant temperature at the sediment/crust interface.
Modern thermal models
Using carefully filtered heat flow data (Sclater et al., 1976)
and specially selected depth-age provinces, Parsons and
Sclater (1977) showed (a) how extraordinarily well the
plate model fits the observational data and (b) that various
parameters of the plate could be obtained by constructing
physical models that fit the data within measurement error
(Figure 3a: top and bottom). Out of these analyses came a
realization (Parsons and McKenzie, 1978) that the lithosphere, even when represented as a flat plate, was still simply the upper boundary layer of a cooling convection
system (Turcotte and Oxburgh, 1967). Parsons and Sclater
(1977) fit the data with a plate thickness of 125 km and a
basal temperature of 1,330
C. In an analysis using a much
more complete data set, Hillier and Watts (2005) showed
that the depth data for the North Pacific filtered to remove
the effects of swells gave results very similar to that of Parsons and Sclater (1977). However, both papers matched
the heat flow data by adding a 4 mW m
À2 to account for
the effect of radioactivity within the ocean crust. Such a
large value is not supported by the overall measured radioactivity of the crust (Staudigal, 2003).
In an alternative observational approach, Stein and
Stein (1992) ignored both filtering and the selection of
special provinces and plotted all the depth and heat flow
data in the Pacific and Northwest Atlantic against age of
the ocean floor. They found a model (GDH1) that best fit
both data sets that had a plate thickness of 95 km and a bottom temperature of 1,425
C (Figure 3b). Though the
basal temperature is too high to explain the rare Earth distribution in the melted rock found at the ridge axis
(McKenzie et al., 2005), GDH1 was the first model that
actually had the predicted heat flow high enough to fit
the data over ocean floor >130 Ma without assuming the
MARINE HEAT FLOW
451