Evidence for a ubiquitous seismic discontinuity at the base of the mantle

Science

Volumn 286, Issue 5443, 1999, Pages 1326-1331

  Sidorin, Igor ^a   Gurnis, Michael ^a   Helmberger, Don V ^a  

^a CALIFORNIA INSTITUTE OF TECHNOLOGY   (United States)

Author keywords

[No Author keywords available]

Indexed keywords

MANTLE DISCONTINUITY; MANTLE STRUCTURE; SEISMIC VELOCITY;

AMPLITUDE MODULATION; ARTICLE; GEOGRAPHY; PACIFIC OCEAN; PHASE TRANSITION; PRIORITY JOURNAL; TOPOGRAPHY; VELOCITY;

EID: 0032724666     PISSN: 00368075     EISSN: None     Source Type: Journal    
DOI: 10.1126/science.286.5443.1326     Document Type: Article

References (39)

1. The most comprehensive global seismic velocity inversions are provided by seismic tomography (13). The characteristic features of all recent tomography models of the lower mantle are faster-than-average linear zones along the circum-Pacific belt and slow regions beneath the central Pacific Ocean and Africa. In many regions, the zones of fast velocity anomalies can be traced from the surface to the CMB and are probably associated with subducted slabs. The broad slow anomalies extend from the base of the mantle to at least 1500 km depth and may be indicative of hot upwellings.
2. T. Lay and D. V. Helmberger, Geophys. J. R. Astron. Soc. 75, 799 (1983).
3. Q. Williams and E. J. Garnero, Science 273, 1528 (1996).
4. M. E. Wysession et al., Science 284, 120 (1999).
5. Long-period S-wave studies allow for a 3% velocity increase over as much as 100 km. However, broad-band studies limit the thickness of the transition zone to about 30 km [T. Lay and C. J. Young, Geophys. Res. Lett. 16, 605 (1989)].
6. The D″ seismic discontinuity, first suggested in (2), explains an observed seismic triplication. The triplication is most pronounced in S waves, although it can sometimes be detected for P waves (11).
7. C. J. Young and T. Lay, Phys. Earth Planet. Inter. 49, 37 (1987).
8. J.-M. Kendall and C. Nangini, Geophys. Res. Lett. 23, 399 (1996).
9. J. B. Gaherty and T. Lay, J. Geophys. Res. 97, 417 (1992); C. J. Young and T. Lay, J. Geophys. Res. 95, 17385 (1990).
10. J. B. Gaherty and T. Lay, J. Geophys. Res. 97, 417 (1992); C. J. Young and T. Lay, J. Geophys. Res. 95, 17385 (1990).
11. M. E. Wysession et al., in The Core-Mantle Boundary Region, M. Gurnis, M. E. Wysession, E. Knittle, B. Buffett, Eds., Geodynamics Series, vol. 28 (American Geophysical Union, Washington, DC, 1998), pp. 273-297.
12. I. Sidorin, M. Gurnis, D. V. Helmberger, X. Ding, Earth Planet. Sci. Lett. 163, 31 (1998).
13. A D″ triplication has been detected beneath the central Pacific Ocean, a region which is anomalously slow at the base of the mantle [S. Russell and T. Lay, Eos (Fall Suppl.) 79 (no. 45), 618 (1998)]. However, the triplication is extremely weak and only detectable by data stacking.
14. S. P. Grand, J. Geophys. Res. 99, 11591 (1994); S. P. Grand, R. D. van der Hilst, S. Widiyantoro, GSA Today 7, (no. 4) 1 (1997). Grand's shear velocity tomography model is parameterized by blocks with lateral dimensions of 2° by 2°. The blocks vary between 75 and 240 km in thickness, with the poorest resolution at the bottom of the mantle. For each block, a velocity perturbation with respect to PREM (Preliminary Reference Earth Model) (29) is provided.
15. S. P. Grand, J. Geophys. Res. 99, 11591 (1994); S. P. Grand, R. D. van der Hilst, S. Widiyantoro, GSA Today 7, (no. 4) 1 (1997). Grand's shear velocity tomography model is parameterized by blocks with lateral dimensions of 2° by 2°. The blocks vary between 75 and 240 km in thickness, with the poorest resolution at the bottom of the mantle. For each block, a velocity perturbation with respect to PREM (Preliminary Reference Earth Model) (29) is provided.
16. I. Sidorin and M. Gurnis, in (10), pp. 209-230.
17. p characterizes the temperature dependence of the shear modulus, G. We used Γ = 6 [D. L. Anderson, Phys. Earth Planet. Inter. 45, 307 (1987)].
18. p characterizes the temperature dependence of the shear modulus, G. We used Γ = 6 [D. L. Anderson, Phys. Earth Planet. Inter. 45, 307 (1987)].
19. ph ΔT/p(h)/g. In the last equation, h is depth, g is the acceleration of gravity, and p(h) is the density (29). The value of Z varies between 0 (low-pressure phase) and 1 (high-pressure phase).
20. The tomography model (13) is smoothed by interpolating the anomalies between the centers of the blocks. The smoothed structure is mapped onto a mesh with a 0.2° by 0.2° by 2 km resolution. A 1.5% velocity jump (31) is added to each vertical column of this fine mesh at a depth predicted by the phase transition characteristics (16). The addition of the discontinuity affects the predicted travel times, and to make our composite model consistent with the original tomography model, we applied a negative velocity gradient at the base of the mantle. In dynamic models, such a gradient is produced by a thermal boundary layer above the CMB (14). We used a constant velocity of 7 km/s (Fig. 2B) at the CMB and computed the gradient to preserve the travel time across the vertical column (32).
21. S. Ni, X. Ding, D. V. Helmberger, Geophys. J. Int., in press.
22. ScS-Scd.
23. Scd-S differential travel times (formula presented) where N is the number of considered ray paths.
24. For Eurasia, Alaska, and India, we used the ray paths studied in (7, 9). To explore the spatial variation of the triplication beneath the central Pacific Ocean and Central America, ray paths from three Fiji-Tonga and three South America events (33, 8) to the World Wide Standard Seismograph Network (WWSSN) and the Canadian National Seismograph Network (CNSN) stations in North America in the 68° to 85° distance range were considered.
25. The D″ triplication has also been extensively studied beneath Central America. However, the structure beneath this region appears to be highly variable [for example, (8, 11)], and no single comprehensive data set is currently available.
26. X. F. Liu, J. Tromp, and A. Dziewonski [Earth Planet. Sci. Lett. 160, 343 (1998)] have suggested that the D″ discontinuity may be produced by the local gradients caused by thermal variations alone without a need for a first-order discontinuity. Although such a possibility has long been acknowledged by the seismological community (10), it is unlikely that thermal gradients of sufficient sharpness (5) can be produced. Dynamically consistent seismic velocity gradients constrained by mineral physics data (11) are not capable of producing a triplication consistent with observations. Similarly, no triplication is produced by the seismic velocity gradients obtained from tomography inversions (18).
27. This is not immediately obvious, considering some degree of smoothing that occurs in tomographic inversions [for example, L. Bréger, B. Romanowicz, L. Vinnik, Geophys. Res. Lett. 25, 5 (1998)]. To test if a flat discontinuity combined with larger amplitude anomalies can produce travel time variations consistent with observations, we globally scaled the anomalies by a factor of 1.5. However, this did not significantly improve the fit to the observations.
28. Dynamic modeling incorporating a chemically distinct dense layer at the base of the mantle [for example, (11, 14)] demonstrates that this layer is depressed beneath cold downwellings (likely corresponding to fast seismic velocity regions) and elevated beneath hot upwellings (slow velocity regions).
29. H.-C. Nataf and S. Houard, Geophys. Res. Lett. 20, 2371 (1993).
30. I. Sidorin, M. Gurnis, D. V. Helmberger, J. Geophys. Res. 104, 15005 (1999).
31. 3 may break down into oxides under D″ conditions; A. M. Hofmeister (in preparation) argued that the velocity jump at the top of D″ may be due to a MgO transition to NaCl or CsCl structure.
32. A. M. Dziewonski and D. L. Anderson, Phys. Earth Planet. Inter. 25, 297 (1981).
33. R. D. van der Hilst and H. Kárason, Science 283, 1885 (1999).
34. It has been demonstrated (11) that a jump of as little as 1% in velocity may explain the observations of the shear wave D″ triplication, provided it is accompanied by sufficiently large gradients. Here, we use a slightly higher value on the premise that tomographic inversion smears the structure so that the gradients provided by the tomography models are somewhat lower than in the real structure.
35. Adding the discontinuity and the low-velocity compensation at the base of the mantle only conserves travel times of phases that do not travel through D″ or cross D″ at steep angles. The imposed low-velocity zone would disturb the travel times of phases such as ScS, especially at large (>80°) distances when the phase almost grazes the CMB. At shorter distances however, only a small portion of the path is affected by the basal low-velocity zone and so the travel time perturbation is less significant.
36. E. J. Garnero, D. V. Helmberger, S. Grand, Phys. Earth Planet. Inter. 79, 335 (1993).
37. The seismic 1D reference models for different regions were obtained in (7-9) to approximate the observed differential travel times for each particular region.
38. The normalization is required for a meaningful comparison of the relative strength of Scd phase for paths with different epicentral distances, since Scd/S increases with distance even for a constant seismic velocity structure (7-9). We obtain the distance trend by averaging Scd/S amplitude ratios for all ray paths with a given epicentral distance.
39. We thank S. Grand for access to his tomography model and T. Lay and E. Garnero for providing the D″ triplication travel time data. Two anonymous reviewers provided a number of valuable comments and suggestions. Supported by NSF grant EAR-9809771. This is contribution no. 8664, Division of Geological and Planetary Sciences, California Institute of Technology.