The Principal Investigators seek funding for collaborative research that aims to develop a new class of tomographic model by explicit integration of observational, theoretical, and computational aspects of seismic data analysis and interpretation. This effort extends their previous work in tomography and involves the measurement of new data sets and the development of wave propagation theory and multi-grid technology that allowsjoint interpretation of data with different sensitivities to Earth's structure. Quantitative integration of thesecomponents is, in our view, a viable and essential step toward the accurate mapping of spatial variations in elastic properties, temperature, and composition. Here the PI's will focus on relative variations in VP and VS in the bottom ~1000 km of Earth's lower mantle, which, they believe, contains critical clues tounderstanding mantle convection and Earth's thermo-chemical evolution over geological time.
Intellectual Merit: Over the past decades global tomography has produced spectacular images of, for instance, mantle flow trajectories and structural complexity near the base of the mantle. However, uneven data coverage and heterogeneous data quality render non-unique, fuzzy images, with substantial spatial variations in reliability. Regularization and the use of inaccurate wave propagation theory probably produce incorrect estimates of elastic parameters even where sampling seems adequate, and the magnitude of wavespeed variations is usually poorly constrained. Moreover, results based on different data sets often disagree in important aspects, and correct joint interpretation of data with different sensitivities to Earth's structure (e.g., body- and surface waves, P or S waves measured at different frequencies) remains a major challenge.
The approximate nature of the "red and blue" images impedes quantitative interpretation and integration with other geophysical constraints and keeps tomography from reaching its full potential as a quantitative probe of Earth's deep interior. This the PI's seek to change. For better parameter estimation they need to exploit the richness of broad-band waveforms and they need more powerful theoretical frameworks for integration and joint interpretation of diverse data sets. The ultimate objective of our approach toward multi-resolution data fusion for global tomography is to produce better 3-D models of Earth's deep interior - on a range of length scales and from a variety of seismological data - by improving (and explicitly linking) three essential aspects of imaging: Data quality and coverage: using automated procedures and multi-resolution concepts (such as time frequency wavelets) they will enhance spatial and spectral data coverage by extracting phase velocity and arrival time information from the vast number of waveforms available through international data centers.
Wave propagation theory: recognizing the need to account for (and benefit from) the different sampling properties of the data considered, and inspired by recent advances in understanding finite frequency effects, they will compute accurate sensitivity kernels for the back-projection of the newly measured data. Parameterization and regularization: to preserve and exploit the localization properties of 3-D sensitivity kernels we will use adaptive multi-grid parameterization and regularization techniques for joint inversion.
The research proposed here focuses on (i) measuring teleseismic P and S type body-wave travel times, (ii) inversion for 3-D variations in .lnVS/.lnVP (or related parameters) in Earth's mantle, and (iii) refining - or refuting - existing views on compositional heterogeneity in the lowermost mantle. They can build on experience in observational seismology and tomography (Van der Hilst, MIT) and wave propagation and inversion theory (De Hoop, CSM), and for the automated data processing they will collaborate with Ritsema (IPGP, France) and involve a postdoctoral associate (for which some fund matching is sought).
Broader Impact: Along with mineral physics data, accurate estimates of elastic parameters are needed to constrain spatial variations in compositon and temperature and, thus, models of mantle dynamics and mineralogy. Furthermore, the concept of and tools for data fusion developed here prepare for the handling and interpretation of large data sets of USARRAY data. The proposed work constitutes the first part of a PhD project at MIT, but students at MIT and CSM will be involved in aspects of the research, either as a Undergraduate Research OPportunity (UROP) or in fulfillment of General Exam requirements.
