Some Suggestions on the New Reference Earth Models

Peter Shearer

IGPP, U.C. San Diego, La Jolla, CA

poster/oral:

Recent improvements in global crustal models promise more accurate corrections for the effect of the crust in resolving deeper seismic structure. However, it is important that velocity models be derived that are consistent with the crustal corrections, in the sense that velocity models can be adjusted to specific crustal thicknesses without introducing a baseline shift in travel times. Current models do not always meet this criteria. For example, PREM and IASP91 predict very similar surface-to-surface teleseismic travel times but have different crustal thicknesses. If these models are adjusted to the same crustal thickness, they produce different travel times. Strategies for producing self-consistent models include: (a) agreement on a "standard" model of crustal structure, (b) use of events with known locations and origin times, (c) tests to be sure velocity model + crustal corrections correctly predict PP-P and SS-S times.

Reference Earth Models are used for many purposes. In seismic modeling, it is much easier to add structure as a perturbation to the model than to remove it. Thus, models should err on the side of simplicity if there is any uncertainty regarding the existence of a particular feature.

Following this philosophy, future reference 1-D models should have:
  1. No 220 km discontinuity

  2. No 520 km discontinuity (at least until more groups agree with me that it is a global feature!)

  3. No D" layers (too intermittent and variable in depth)

  4. Transverse isotropy in upper mantle (now well established, but for many applications isotropic models are sufficient, so purely isotropic models should also be provided; ideally these would be designed directly to fit P and S travel times, not simply be the average of the transversely isotropic model).

  5. No D" anisotropy (global extent and amplitude still uncertain)

  6. No inner core anisotropy (it's not spherically symmetric)
Future reference 3-D models should have:
  1. Large-scale topography on transition zone discontinuities (for some models, but not all models)

  2. Inner core anisotropy aligned with the rotation axis (more complex models involving a tilted symmetry axis and/or heterogeneity are not yet well resolved; possible time dependence is an additional complication)
The pace of current research suggests that any reference 3-D model will quickly become obsolete for seismic modeling purposes. Thus, agreement on a common set of data constraints (body wave travel times, surface wave phase velocity measurements, normal model observations) is a more practical goal than consensus on a single model. In contrast, a new 1-D model is likely to remain useful for a longer period because it will be used as a common base for a variety of more complicated models.

Peter Shearer ( pshearer@ucsd.edu)

Go back to List of Participants
Go to main page