This project has two parts: (1) to adapt a suite of waveform tomography algorithms based on a 2D acoustic framework developed over the past 18 years by R. G. Pratt and co-workers and thus far used exclusively in an active source environment to the analysis of recordings of earthquakes at teleseismic distances (what are commonly known as ?receiver functions?) and (2) to apply this algorithm to a collection of broad band seismograms collected in the Tien Shan as part of the recently completed MANAS project. The motivation behind the first part is the demonstrated efficacy of this approach; reasonable models can be constructed without resorting to massively parallel computing. The tasks involved include incorporating a visco-elastic version of this code, which we have already developed, to an adjoint inversion scheme similar to that used in the acoustic case. The researchers will also investigate the feasibility of incorporating anisotropic effects and extending the algorithm to three-dimensions. The second part is motivated partly by our familiarity with the MANAS dataset, but more by results from the application of ?traditional? receiver function migration and arrival time tomography techniques. Specifically, arrival time tomography images indicate the presence of two high wavespeed slabs beneath the Tien Shan that suggest a long history of plate-like lithospheric subduction in this region. The receiver functions show an unusual disruption in the Moho in the areas where these plates appear to be consumed. Both these images have first order implications for the evolution of the Tien Shan, but interpretations suffer from insufficient resolution. Based on active source analogues, the waveform tomography approach will significantly improve the quality of the images of these unusual features in the upper mantle. A better understanding of these features will have a significant impact on our estimation of both the evolution of this region and the root cause of earthquake activity in this part of the world.
Arguably the best sources of information we have about the interior of the Earth are derived from seismograph records of earthquakes recorded at long distances (>1000 km) from the source. A particularly fruitful area of research over the past decade has been in modeling those parts of these seismograms that are modified by those parts of the Earth within a few 100 km of the surface beneath a seismic station. In this project the researchers will develop a new technique to extract more information from these types of waves by adapting methodology that has proved to be very useful in subsurface imaging on a much smaller scale using ?active? sources like explosions and vibrators. Despite the difference in scale, the basic physics involved in modeling wave propagation is the samel. The algorithm will be general enough for scientists and engineers to use in any analogous environment to produce subsurface images.
This project had two principal objectives. The first was to develop a numerical technique that could be used to analyze seismograms of earthquakes at teleseismic distances in order to generate images of the subsurface at a resolutions higher than those currently available from standard techniques. A motivation for pursuing this objective is that this type of data is abundant, may be acquired virtually anywhere on the planet, and can be collected at relatively low cost. Also, this type of data provides our best sampling of the Earth’s lithosphere, an understanding of which is essential for understanding fundamental geological processes. Hence, being able to retrieve images with significantly higher resolution should lead to better understanding of the dynamics at work in the Earth’s interior. We also strove to develop a technique that could be used on small to moderate sized computers, so that more people in the community would find it useful. The second objective was to apply this technique to a an actual data set, partly as a demonstration of its capabilities but also to provide new insights into a geologically intriguing part of the world. The Tien Shan was chosen for this purpose partly because of the existence of an appropriate data set but also because, despite its being the archetypical example of active intracontinental mountain building, the fundamentals of its evolution remain controversial. The primary outcomes of this project are, first, that we were successful in creating a viable algorithm for imaging of teleseismic data. This involved both creating an analysis technique and developing a suite of software to implement it. We now have a reasonably stable version of our code that can be used by other seismologists with a modicum of training in this technique. The software can operate well on small computers (it can be run on a laptop or other single processor machines with standard memory and storage resources). At the same time, we developed parallel versions of the code to take advantage of large facilities when they are available. Our original goal was to analyze teleseismic body waves. We achieved that goal and have extended its capability to include surface waves, thereby increasing its range of applicability. The second primary outcome comes from the analysis data from both the Tien Shan and the Cascades (in Oregon). We decided to include the Cascades in this project because, like the Tien Shan, an appropriate data set could be found, but, unlike the Tien Shan, this data had been processed several times before by other investigators and we saw this as an opportunity to compare this technique to other, more established, techniques. We were able to improve upon existing images of the Cascades; the most important finding being that the amount of serpentinite in the upper mantle wedge (sometime referred to as the "cold nose") is much less than previously estimated. This result has important consequences for the release of water from the downgoing slab, the occurrence of nonvolcanic tremor, and the downdip extent of large earthquake ruptures, since all of these depend on the nature of the interface between the slab and this part of the wedge. Our results from the Tien Shan solved a paradox resulting from previous efforts to image the upper mantle beneath this part of central Asia. On the one hand, travel time tomography suggested a uniform low wavespeed upper mantle that was interpreted as evidence for delamination of the lithosphere beneath the mountains. On the other, receiver function studies suggested the presence of a lithospheric lid with a thickness of several 10’s of km. Our image of the Tien Shan shows that, in essence, both were right but were biased towards sampling different parts of the range. An image we generated from modeling ballistic surface waves shows that the southern part of the range, where the highest mountains are located, appears to have lost its original lithosphere (or that lithosphere has warmed considerably, making it weak), while the northern part retains the higher wavespeed lid observed in receiver functions. One important implication of this study is that this type of analysis can significantly improve the lateral resolution available to traditional surface wave imaging. The main reason for this is because we are able to model deviations (or residuals) in the surface wave morphology by local scatterers, which helps to "pinpoint" to location of any heterogeneity. Other important outcomes relate to development and training of personnel. Work on this project constitutes the principal component of the PhD thesis of a graduate student, Ben Baker, who continues to investigate techniques for improving the inverse algorithm. Ben is expected to complete his degree next year (2013). We have demonstrated our technique to colleagues at other universities and are beginning collaborations with them, and have trained their graduate students in the operation of the software.
