Roadmap | The review process | Home
 Why is OJ anomalous?
Why is the uplift and subsidence history of the Ontong Java Plateau anomalous compared to other large igneous provinces?

Julie Roberge1 and Paul J. Wallace2

1Laboratorio Universitario de Petrología, Instituto de Geofisica, Universidad Nacional Autónoma de México (UNAM), Coyoacan, 04510, Mexico, D.F. roberge@geologia.unam.mx
2Department of Geological Sciences, University of Oregon, Eugene, OR  97403-1272, pwallace@uoregon.edu

This webpage is based on:

Overview

The Ontong Java Plateau (OJP) in the western Pacific (Figure 1) is the largest volcanic oceanic plateau and may represent the largest magmatic event on Earth in the last 200 Ma. Relevant statistics are:

  • Age = Cretaceous  (122 Ma)
  • Volume = 40-45 million km3
  • Area > 1.6 million km2
  • The crest of the plateau is ~1700 m below sea level, and elsewhere it is ~2-3 km deep.

Figure 1: Location of the Ontong Java Plateau in the western Pacific and ETOPO5 bathymetric map of the Ontong Java Plateau showing locations of Leg 192 drill sites (stars).  Locations of previous ODP and DSDP drill sites that reached basement (small solid circles) are also shown.  Depth contours are in meters below sea level. Click here or on figure for enlargement.

The OJP is anomalous compared to other oceanic large igneous provinces such as the Kerguelen plateau in that it never formed a subaerial landmass and did not cause extinction (Figure 2) [Ed: see also other pages on Ontong Java].

Figure 2: Comparison of OJP with other large igneous provinces. Click here or on figure for enlargement.

Submarine basaltic glasses preserve information on magmatic volatile contents (Figure 3).  H2O and CO2 concentrations can then be used to estimate paleo-eruption depths and thus constrain better the uplift and subsidence history of the OJP.

Figure 3: Example of unaltered basaltic glass from pillow rims at sites 1183, 1185, 1186, 1187 (left), and glass in non-vesicular glass shards in volcaniclastic rocks from site 1184 (right).

Water in the mantle source region for Ontong Java Plateau basaltic magmas?

During ODP Leg 192, five widely spaced sites were drilled (Figure 1; Mahoney et al., 2001).  Unaltered glass from pillow basalt rims at four locations (ODP Sites 1183, 1185, 1186, and 1187) and from non-vesicular glass shards in volcaniclastic rocks at Site 1184 were analyzed for:

  1. H2O and CO2 using Fourier Transform Infrared (FTIR) spectroscopy (Table 1), and
  2. for major elements using a Cameca SX-50 Electron Microprobe at University of Oregon. 

Our results complement previously published data for glasses recovered from ODP Leg 130 Sites 803 and 807 (Michael, 1999).  For consistency, we reanalyzed the glasses from Sites 803 and 807 because we used a different data reduction procedure for our CO2 analyses.  Our new analyses and data reduction procedure result in CO2 values that are mostly 15 ppm (at lower concentration) to 30 ppm (at higher concentration) lower than those of Michael (1999).

Table 1: Average dissolved H2O and CO2 contents of submarine basaltic glasses from the Ontong Java Plateau (OJP).

Note: H2O and CO2 were analyzed by FTIR spectroscopy using band assignments and absorption coefficients as described in Roberge et al. (2004).  Peak height measurements for CO2 were calculated using a peak fitting program (S. Newman, unpublished).  This method yields values for experimental glasses (Dixon et al., 1995) that are comparable to the reference-glass subtraction and hand-drawn background method upon which the CO2 solubility relations have been established (J. Dixon, written comm.)  The H2O and CO2 values reported for each site are averages (± 2σ in parentheses) of multiple glass chips (see Roberge et al., 2004, for complete data and analytical uncertainties). Vapor saturation pressures were calculated using VolatileCalc 1.1 (Newman & Lowenstern, 2002). Uncertainties (in parentheses) for saturation pressures, eruption depths, and subsidence values are based on propagation of 2σ uncertainties in the average H2O and CO2 values. The subsidence uncertainties also include uncertainties in depths of the samples in the drill hole.

*Based on interpretation that Site 1184 volcaniclastic rocks were erupted in a shallow marine environment but deposited subaerially (Thordarson, 2004).
 
An exciting discovery of Leg 192 was that basement at Site 1187 and the upper group of flows at Site 1185 are composed of high-MgO, incompatible-element-poor basalt that are unlike basalts found elsewhere on the OJP (Figure 4).  Relatively low K2O, Na2O, and P2O5 in all glasses suggest that OJP basaltic magmas formed by large extents of melting.  Low-MgO basalts can be derived by fractionation at low to moderate pressure from parental magmas similar to the high-MgO Site 1187 basalts.

Figure 4: Major element compositions of Ontong Java Plateau basaltic glasses. Data from Sites 803 and 807 and the island of Malaita are from Michael (1999). Lines show fractional crystallization paths for a parental magma with 17.6 wt% MgO calculated as described in the text. Crystallization of this parental composition at pressures of 1 bar to 2 kbar can largely reproduce the observed range of major element compositions. Under these conditions, the crystallization sequence is olivine, followed by olivine + plagioclase, followed by olivine + plagioclase + clinopyroxene. Click here or on figure for enlargement.
 

However, H2O concentrations are similar in the two basalt types despite the lower K2O and TiO2 of the high-MgO glasses.  To understand H2O in mantle source regions it is useful to compare H2O/Ce ratios because these elements have a similar incompatibility to one another during mantle melting and fractional crystallization (Michael, 1995). H2O/Ce values for OJP basalt are 355-370 for high-MgO glasses and 270 for low-MgO glasses (Figure 5).  These values are higher than most depleted and enriched MORB (Michael, 1995).  However, the H2O/Ce values of all glasses may be elevated because of assimilation. If uncontaminated OJP magmas have low Cl/K like other mantle-derived magmas, then primary H2O/Ce values may be as low as 125-140.

Figure 5: H2O/Ce versus K/Ti for OJP basaltic glasses (left). Shown for comparison are H2O/Ce ranges for MORB glasses from various regions (MAR = Mid-Atlantic Ridge). Horizontal black bar shows the H2O/Ce ratio estimated as described in Roberge et al. (2004) for uncontaminated OJP magmas. Click here or on figure for enlargement.

In addition, trace element abundances demonstrate that OJP basaltic magmas formed by large degrees of melting (Figure 6).  However, as previously stated, OJP basalts have low H2O, similar to depleted MORB.  Therefore the large degrees of melting must have been caused by high melting temperature (>1550°C).

Figure 6: Graph of Nb vs. Zr showing the extent of melting required to produce OJP basalts (from Fitton & Goddard, 2004).

Estimating paleo-eruption depths

Vapor saturation pressures were calculated for all sites and then converted into eruption depths (1 bar = 10 m water depth) assuming equilibrium solubility of H2O and CO2 at the time of quenching (Figures 7 and 8; Table 1).  As expected, glass shards from the volcaniclastic deposits at Site 1184 have low vapor saturation pressures, indicating an average quenching depth of 540 ± 210 m. Site 1183 glasses, which come from the shallowest water site on the central high plateau, also have relatively low vapor saturation pressures of 107 bars (1070 ± 90 m), whereas Sites 1185, 1186 and 1187 have saturation pressures of 215 to 245 bars, yielding estimated eruption depths of 2150 to 2450 (±100) m.  Samples from Site 803 yield an average saturation pressure of 290 bars (2900 ± 90 m). 

Figure 7: H2O versus CO2 for Ontong Java Plateau basaltic glasses.  Symbols correspond to Ocean Drilling Program (ODP) site numbers.  Vertical lines represent degassing paths for basaltic melts with initial CO2 contents of 200 ppm (solid line) and 2000 ppm (dashed line).  Also shown are vapor saturation curves for basaltic melts at pressures of 10-35 MPa.  All calculations were made using VolatileCalc 1.1 (Newman & Lowenstern, 2002).

Figure 8: Eruption depth estimates (in mbsl) for Ocean Drilling Program (ODP) sites on Ontong Java Plateau. For all sites, present-day depth of top of igneous basement has been corrected for sediment loading.  Corrected basement depth (Dc) is obtained from the equation of Crough (1983): Dc = dw + ts(rs– rm)/( rw– rm), where dw is water depth (in m), ts is sediment thickness (in m), rs is average sediment density (1.9 g/cm3), rm is upper-mantle density (3.3 g/cm3), and rw is seawater density (1.03 g/cm3).

The estimated eruption depths for all sites should be viewed with caution for several reasons (Figure 9). Submarine basaltic pillow rims, particularly MORB samples, are commonly supersaturated with CO2 (Dixon & Stolper, 1995), so measured CO2 contents in pillow rims could potentially overestimate true eruption depths. However, submarine OJP lava flows are likely to have much larger volumes and longer flow distances than MORB flows. Geochemical data suggest that some OJP flows may have traveled 100s of km (P.J. Michael, written comm.).  This would allow time for dissolved CO2 to reach equilibrium values at the appropriate seafloor depth before final quenching.  In fact, such long downslope flow distances could have caused lavas to be vapor saturated near their eruption (vent) depth, which would be shallower than the final depth of emplacement (Michael, 1999).  Thus we argue that our “eruption” depths calculated from CO2 data are minimum values because true emplacement depths could have been deeper.  This line of reasoning provides a plausible explanation for the large differences in CO2 contents and inferred eruption depths of glasses from Site 807 Unit A and Units C-G (Table 1).  The low CO2 contents of Unit A glasses suggest that this may have been part of a very long lava flow that had an original vent in much shallower water.  In contrast, Units C-G represent multiple flows, all of which have much higher CO2, and their CO2 contents probably more closely represent their original emplacement depth (3041 ± 240 m).

Figure 9: Illustration of the potential problems of using CO2 content to infer paleo-eruption depth.

Uplift and subsidence of the OJP

Mesozoic marine magnetic anomalies in the Nauru Basin (adjacent to OJP) suggest that the OJP formed within ~130 to ~155 Ma oceanic crust. The depth of 10 to 35 Ma oceanic crust lies between 3600 and 4700 m according to global age-depth curves (Ingle & Coffin, 2004). Using the high Plateau (Site 1183) eruption depth of 1170 m and the Eastern Salient (Site 1184) eruption depth of 540 m we calculate that the maximum uplift was 2400-3500 m for the high Plateau and 3100-4200 m for the Eastern Salient.

Dynamic Uplift

The arrival of a hot and buoyant plume at the base of the lithosphere, combined with crustal thickening due to eruption and intrusion of a large volume of basaltic magma, should produce substantial surface uplift (Olson & Nam, 1986).  Dynamic uplift is the thermal doming produced by viscous normal stresses imposed on the lithosphere by the rising of a plume head and has been estimated through experimental and theoretical studies (e.g. Farnetani & Richards, 1994; Hill, 1991; Griffith et al., 1989; Olson & Nam, 1986).  The results show that at first, before the diapir reaches the base of the lithosphere, the surface topography is determined by the diameter, density anomaly and depth of the diapir (plume head).  When the top of the diapir reaches approximately one diapir radius from the surface, an asymmetric surface swell appears increasing in height and decreasing in width (Figure 10A).  When the upper edge of the diapir reaches 0.2 diapir diameters below the crust surface, the height of the surface swell attains a maximum, with a minimum width (Figure 10B).  From this point on, the swell will subside and increase in width as the diapir spreads laterally beneath the surface.  To estimate OJP’s uplift we need to assume a certain plume head volume, which can be estimated from the erupted volume of the plateau and the degree of partial melting needed to produce it.  Here, a plateau volume of 4.5 x 107 km3, crustal and mantle densities of 2.9 and 3.3 g/cm3, respectively, a plume temperature of ~1550°C, and 30% partial melting (Fitton & Godard, 2004) are used.  With these parameters, dynamic uplift models applied to the OJP predict an uplift of ~1000 to ~3000 m above the surrounding seafloor depending on the shape and diameter of the plume head (Neal et al., 1997; Farnetani & Richards, 1994; Hill, 1991; Griffith et al., 1989; Olson & Nam, 1986).

Figure 10: Illustration of theoretical and laboratory dynamic models.  All models reach similar conclusions: Maximum dynamic uplift for a high temperature (~1500°C) plume and 25-30% partial melting is 1000 to 3000 m.

Isostatic Uplift

Models that explain hotspot uplift by isostatic compensation of thermally expanded mantle rather than the dynamic effects of a rising plume yield similar results (Ito & Clift, 1998).  The isostatic effect of crustal thickening has also been modeled, suggesting an additional isostatic uplift of 2 to 4 km above the adjacent seafloor (Neal et al., 1997).  Gladczenko et al. (1997) calculated the average OJP crustal density to be 2.86 g/cm3 on the basis of combined seismic velocity analyses and gravity modeling. However, given uncertainties in velocities and the non-unique nature of gravity modeling, it is appropriate to calculate isostatic uplift for a range of densities (2.8 to 3.0 g/cm3).  Below sea level, water-corrected isostasy was calculated using

(1)

where Δh is amount of uplift above seafloor, hOJP is Ontong Java Plateau crustal thickness, hOC is thickness of normal oceanic crust (7 km), ρw is water density (1.03 g/cm3), ρm is mantle density (3.3 g/cm3), and ρOJP is Ontong Java Plateau crustal density (2.8 to 3.0 g/cm3). For calculations above sea level, water- and air-corrected isostasy was calculated using

(2)

where hw is water depth for normal 10-35 Ma oceanic crust (4.1 km). Using the high plateau (Site 1183) crustal thickness of ~30 km (Gladczenko et al., 1997), we estimate isostatic uplift ranging from 2400 m (ρOJP = 3.0 g/cm3) to 4700 m (ρOJP = 2.8 g/cm3) above the surrounding seafloor due to the effects of crustal thickening (Figure 11). 

Adding the initial dynamic uplift (2000 ± 1000 m), and taking into consideration how this changes the seafloor water depth (which in turn influences the isostatic effect of crustal thickening), OJP maximum total uplift would range from 4300 to 6100 (±1000) m above the surrounding seafloor (Figure 11).  These estimates are larger than the estimated maximum uplift (2500 to 3600 m) based on H2O and CO2 data for the basaltic glasses from Site 1183, though there is a slight overlap when the uncertainties in dynamic uplift are considered.

Figure 11: Comparison of predicted vs. observed uplift of Ontong Java Plateau (OJP).  Diagram A shows estimated isostatic uplift due to crustal thickening.  Diagram B shows both isostatic and an average of 2000 m of dynamic uplift (dashed lines) based on plateau-specific models (Neal et al., 1997; Ito & Clift, 1998).  Diagonally ruled area shows maximum plateau uplift inferred from paleo-eruption depths based on CO2 data. Click here or on figure for enlargement.

Subsidence of the OJP

The cooling of the oceanic lithosphere causes the density of lithospheric rocks to increase.  Older lithosphere is more dense and cold than younger lithosphere and this causes the lithosphere to subside as it ages (Figure 12A).  Therefore, after their emplacement, oceanic plateaus subside as a result of cooling and contraction of the lithosphere (Figure 12B; Detrick & Crough, 1978; Coffin, 1992).  Arrival of a hot mantle plume affects this subsidence by creating the initial dynamic uplift discussed previously, therefore reducing the subsidence for a certain amount of time (Figure 12C).  Subsidence curves for normal oceanic lithosphere and hotspot-affected lithosphere suggest that the 122 Ma OJP should have subsided ~2700 to 4100 m since its formation (Figure 13). After correcting the present-day depth to the top of the igneous basement for sediment loading, we calculate the total subsidence of OJP by subtracting the corrected present-day basement depth from the original eruption depth estimated from basaltic glass H2O and CO2 data.  The subsidence estimates vary from 900 m (Site 803) to 1900 m (Sites 1184 and 1185) with an average of 1500 ± 400 m over much of the plateau (Figure 14, Table 1). We have excluded Site 807 from our subsidence average because of the large differences in CO2 content between Units A and C-G glasses, but our preferred eruption depth based on the C-G glasses as described above suggests 600 ± 250 m of subsidence at this site.  Our estimated average subsidence for the OJP is lower than previous estimates based on microfossils (Figure 1-16; Ito & Clift, 1998) and CO2 in glasses from Site 807 Unit A (Michael, 1999).

Figure 12: A) Illustration of the principle of isostasy, which requires the oceanic crust to subside with age to offset the thickening and cooling of the lithosphere. B) Variations of subsidence with time for normal oceanic lithosphere (Parsons & Sclater, 1977; Stein & Stein, 1992).  C) Variations of subsidence with time for a hot-spot-affected oceanic lithosphere (Ito & Clift, 1998).

Figure 13: Subsidence estimates versus age for ODP sites on Ontong Java Plateau.  Subsidence estimates based on microfossils are from Ingle & Coffin (2004). Subsidence estimates for other large igneous provinces (Detrick et al., 1977; Coffin, 1992) are minimum values and assume that these features originally formed at sea level; true subsidence for these could be 1000-2000 m greater than the values plotted.  Subsidence of hotspot-affected lithosphere (Ito & Clift, 1998) is calculated for plume excess temperatures (ΔT) ranging from 200°C (minimum subsidence) to 350°C (maximum subsidence). Symbols correspond to Ocean Drilling Program (ODP) site numbers.

Figure 14: Illustration of the calculation of subsidence.  Using the pressure-dependent solubilities of H2O and CO2, our data suggest original eruption depths (at 122 Ma) varying from ~1100 m below sea level (mbsl) on the central part of the plateau to 2200-3000 mbsl on the eastern edge.  The glass shards from Site 1184 suggest a quenching depth of 500 mbsl. Click here or on figure for enlargement.

Possible explanations for small initial uplift and subsidence
 
1. Dense garnet granulite in lower OJP crust

One possible explanation is that uplift was tempered by the presence of dense garnet granulite and possibly eclogite in the lower OJP crust that formed from cumulates and intruded and underplated gabbros (Neal et al., 1997).

In Favour

Direct evidence for garnet granulite in the lower crust comes from xenoliths in 34 Ma alnöites on the island of Malaita (Neal et al., 1997).

Against

Seismic velocities, gravity data, phase equilibria, and crustal thickness estimates based on geophysical data do not support the widespread presence of eclogite (Gladczenko et al., 1997; Richardson et al., 2000).

These data do not eliminate the possibility of high-density hidden cumulates in the lower crust, but they indicate that the contribution of such rocks to the average crustal density of OJP is significantly less than Neal et al. (1997) estimated. The upper limit that we used for average OJP crustal density (3.0 g/cm3) in our uplift modeling (Figure 14) allows for the presence of significant dense garnet granulite in the lower crust.  Using a density of 2.9 g/cm3 for the first 10 km of the column and 3.0 g/cm3 for the remaining 23 km, and adding as much as 2 km of eclogite (density of 3.6 g/cm3) at the base of the column, the calculated initial isostatic uplift is 3 km, which is equivalent to the isostatic uplift of the entire column at a density of 3.0 g/cm3. Adding the dynamic uplift of 1-2 km, our model still predicts more initial uplift than is observed (Figure 11).

2. Underlying mantle cause?

If crustal characteristics of the OJP are not responsible for the anomalous uplift and subsidence behavior, then the cause may be in the underlying mantle.  The production of an OJP-scale volume of basaltic crust would produce an enormous melt-depleted residuum in the upper mantle consisting of refractory harzburgite with relatively Fe-poor olivine (Neal et al., 1997; Fitton & Godard, 2004) which would be buoyant relative to fertile mantle (Robinson, 1988).

In Favour

Seismic tomography shows the presence of a rheologically strong and seismically slow upper-mantle “root” extending to ~300 km beneath the OJP, and the seismic characteristics of this root suggest it is chemical or mineralogical rather than thermal in origin (Richardson et al., 2000; Klosko et al., 2001; Gomer & Okal, 2003).

Against

The volume of the root is much larger than can be explained by the volume of mantle remaining from melt extraction needed to form OJP basalts (Neal et al., 1997). Given the enigmatic nature of the low-velocity root beneath the OJP, its role in causing the anomalous uplift and subsidence behavior of the plateau is unclear.

3. Other mantle processes

Other mantle processes that might affect subsidence include slow buoyancy flattening of a plume (e.g., Phipps Morgan et al., 1995) and slow cooling of the lithosphere resulting from the thickness of the plateau, but why these would affect the OJP but not other oceanic LIPs is unclear.  Another possibility is that large-scale magmatic underplating of basaltic magma for ~30 Ma after formation of the plateau provided a continued heat source, and thus reduced subsidence (Ito & Clift, 1998). While there is evidence of some younger volcanic events to support this, the lack of voluminous volcanism post-122 Ma seems inconsistent with this hypothesis.

4. Large bolide impact

As an alternative to the mantle plume hypothesis, the OJP may have formed as the result of a large bolide impact (Glikson, 1999) [Ed: see also OJ Impact page]. It has been proposed that this could explain the anomalous uplift and subsidence of the OJP because the impact hypothesis does not require a mantle temperature anomaly to generate large degrees of melting (Ingle & Coffin, 2004). The model simulates a ~20 km diameter bolide of chondritic composition impacting a preexisting lithosphere of ~50 km thickness at a velocity ~20 km/s (see Ingle & Coffin, 2004, Figure 4). Vertical impact and instantaneous vaporization of the ~4 km deep water column are assumed. The penetration depth would be about 60 km with an initial crater diameter of ~ 200 km (Ingle & Coffin, 2004).  Massive decompression melting will take place in the upper mantle, to a minimum depth of 300 km, assuming 100% partial melting resulting from the removal of the lithospheric overburden. This model also explains the low shear-wave velocities observed by Richardson et al. (2000) by catastrophic decrease in pressure of the solid asthenospheric mantle, moving laterally inward and upward from below to replace the extracted mantle during its emplacement beneath the OJP. 

However, Tejada et al. (2004) have argued that the impact hypothesis is not consistent with geochemical and other geophysical data for the OJP, or with the Early Cretaceous paleoenvironmental record (see counter arguments in Ingle & Coffin, 2004).  Furthermore, it remains controversial whether the thermal effects of a bolide impact would indeed create surface uplift and subsidence comparable to that of a hot mantle plume.  Korenaga (2005) argues that excavation-induced melting is essentially the same as melting of hotter-than-normal mantle and that the instantaneous depressurization by the formation of the crater is equivalent to increasing the potential temperature of the underlying mantle [Ed: see also OJ Puzzle page].

Summary and Conclusion

Models of multiphase fractionation show that the high MgO Site 1187 samples could be parental to the low MgO groups.

There is no evidence for high magmatic H2O contents that might have increased extents of mantle melting beneath the OJP.  Instead, large extents of melting must have been caused by a relatively high mantle temperature.

Both the initial uplift and post-eruption subsidence of the Ontong Java Plateau are significantly less than predictions from thermal models of oceanic lithosphere and are less than what is observed for other oceanic large igneous provinces.  A few hypothesis are:

  • The uplift was tempered by the presence of dense garnet granulite and possibly eclogite in the plateau’s lower crust that formed from cumulates and intruded and underplated gabbros (Neal et al., 1997).  However, seismic velocities, gravity data, phase equilibria, and crustal thickness estimates based on geophysical data suggest that the contribution of a dense lower crust is significantly less than Neal et al. (1997) estimated (Gladczenko et al., 1997; Richardson et al., 2000).
  • Subsidence was tempered by the production of an enormous volume of melt-depleted, relatively buoyant residuum in the upper mantle that a plateau-scale volume of basaltic magma would produce (Neal et al., 1997; Fitton & Godard, 2004).  However, the volume of the root is much larger than can be explained by melt extraction needed to form the plateau (Neal et al., 1997).
  • The plateau may have been formed by a large bolide impact (Rogers, 1982; Ingle & Coffin, 2004), since it does not require an anomalously high mantle temperature and it would neither buoy the lithosphere nor lead to subsequent lithospheric cooling and contraction. However, whether the thermal effects of a bolide impact are different or the same as those of a mantle plume is not very well defined (J. Korenaga, written communication) and the geochemical aspect of the bolide impact hypothesis has yet to be proven (Tejada et al., 2004).

More work is clearly needed to determine whether this, the world’s largest large igneous province, was formed by a mantle plume, bolide impact, or some other process.  If the OJP was formed by a plume, then there remains a major gap in our understanding of how large plumes interact with the Earth’s lithosphere.

References

last updated 31st December, 2006

:: HOME :: MECHANISMS :: LOCALITIES :: GENERIC ::
© MantlePlumes.org