By matching WKM synthetic seismograms with various data, we found and confirmed that (a) the D" beneath the Caribbean varies laterally, and the variation is best revealed with Scd+Sab beyond 88 degrees where Sed exceeds Sab. b) The low-velocity structure beneath Africa is approximately 1500 km high, at least 1000 km wide and has a 3% reduced velocity of S. 96 6.4 The construction of a long-lasting PKP precursor 97 6.5 The construction of a long-lasting PKP precursor, detailed 98 6.6 Synthetic material from PREM that smoothly shows PKP precursor.
List of Tables
Chapter 1 Introduction
Among these branches, waveform modeling is crucial in revealing detailed structures, e.g., such as the velocity jump at D" and the Ultra Low Velocity Zone (ULVZ, which is assumed to have a 10% decrease in velocity P , 30% reduction in velocity S ) have been detected by waveform studies. Typically waveform modeling involves (1) a forward velocity model, (2) an algorithm to construct theoretical seismograms for the model, (3) matching synthetic seismograms with observed data and (4) find the differences, then return to (1) by changing the velocity model until a satisfactory fit is obtained.
Chapter 2 Constructing Synthetics from Deep Earth Tomographic Models
- Abstract
- Introduction
The basis for their differences is that the doublings, as shown in figure 2.2, vary in position between S and ScS from region to region, that is, points b and c are not global (Wysession et al., 1998). Note that FBC is located in eastern North America, while EDM is in western North America (figure 2.14).
Time (Sec)
The WKM Approx imation
The WKBJ method can be derived directly from the asymptotic theory (wavefront expansion), (Chapman) and is based on geometric ray paths of the type shown in Figure 2.4b. An enlarged view of the beam paths is shown in Figure 2.6c, showing the slight adjustment in path when exiting the fast block.
Application
The ScS phase is shifted forward by about the same amount in both approximations relative to the PREM synthetics, as the correction to the ScS trajectory is less severe than for Sed. Doubling the number of layers produces roughly the same synthetics, which is a good test of the process.
JIVL-
Discussion and Conclusion
If the 2D synthetics do not change much due to a small change in azimuth, 2D modeling appears to be useful. However, since the waveform solution can be decomposed into individual beams (dpi/dti), they can be moved and reassembled to simulate adjacent models, as in Song and Heimberger (1998).
Chapter 3 Application of WKM to Low Ve lo city S t ruct ure Beneath Africa
- Abstract
- Introduction
- Analysis
A 2D cross-section of Grand's model from South America to South Africa is shown in Figure 3.2a. These synthetics are based on the 1D reference model, PREM, a 2D section of Grand's model as shown in Figure 3.2a and some modified models derived in this study.
Open symbols are from the Central African stations, AAE and AI, and are delayed relative to the Southern African stations, as predicted by station statics (station traveltime corrections based on global P data [Dziewonski and Anderson , 1981]. South African stations BUL and PRE are slightly negative, while GRM, WIN and SDB are slightly positive in terms of P wave residuals. Stations AAE and AI have large positive P residuals of 2.5 and 2.1 sec respectively, which we divided into S - wave residuals by multiplying by 2.5.
Data
Discussion and Summary
The data in Figure 3.5a is from the event (670117) and is aligned on directS with vertical lines indicating the different arrivals. The data in Figure 3.5b is from the particular event (731025) given in Figure 3.4 showing crossover PRE (before) and BUL (after). Prediction of S delays for the model proposed by [Ritsema et al., 1998a] and our proposed modifications (ALVS) are given in Figures 3.8b and 8a for comparison.
Core-Mantle Boundary
Ab stract
Introduction
While the subducted material in dynamical models typically exhibits complex forms of folding and buckling at intermediate mantle depths, it generally forms a flat-lying layer at the CMB. The strongest eruptions almost always occur at the plate tip, where advective thickening of the thermal boundary layer occurs caused by horizontal creep of the cold plate as shown in Figure 4.1 (top panel). Enticed by these dynamical predictions, we performed a high-resolution study of seismic phases that sampled the CMB at such a possible margin beneath the South Atlantic (bottom panel of Figure 4.1).
Data and Analysis
Included in the bottom panel are some examples of ray paths for S and ScS at large regions, 85 to 95°, showing the similarity of trajectories through the mantle at regions beyond 95°. There is only a small change in S-wave rise time (start to peak amplitude) for these PREM synthetics due to the slightly positive velocity gradient in the lower 200 km of the model. Adding a low-velocity gradient to D" has a strong effect on S, causing shadowing to begin at shorter distances (ie, beyond about 95°) as shown in the second column.
The symbols correspond to those shown on a map (Figure 4.5) plotted at the ScS bounce point. The top panel of Figure 4.5 shows D" tomographic results showing the transition from under South America (relatively high speed) to West Africa (relatively low speed). Most of the red traces (Figure 4.5) indicate relatively delayed ScS pulses weather. to S, varying in delays from 5 to 10 sec.
Epicentral DistanceC)
Discussion
We have shown in Figure 4.6 some array data that suggests that there are significant changes in D" over dimensions of a few degrees. This is a difficult question, but we can go some way to answering it by generating 2D synthetic data (Figure 4.7) for existing tomographic models as shown in Figure 4.1. Resolving ScS from S obviously becomes more difficult (Figure 4.7) if one filters for long periods, as done in most long-period (LP) tomography studies.
Event2 Event16
Chapter 5 The Low Velocity Structure beneath Africa and Atlantic from
Waveform and Travel Time Modeling
Abstract
5. 2 Introduction
Algorithm and Differential Travel Time Anal- ys1s
To implement the hybrid simulation technique, S, SKS and ScS groups are calculated using GRT, the sum of which is shown in Figure 5.1. Then SKS and ScS are synthetically shifted (also with appropriate amplitude adjustment) with respect to S to obtain optimal matching with an observed seismogram (bottom light traces). This tomographic model predicts a difference in SKS-S and ScS-S times with a maximum delay of 3 seconds and an average delay of 1.5 seconds, significantly smaller than the observed value, but with a similar trend.
Waveform Constraints
For the three improved models, ScS is always delayed due to the low velocity layers. But for model LVZ1, due to the slow transition from low velocity layer to ALVS, the ScS for larger epicentral distances (greater than 92 degrees) samples an effectively thicker low velocity layer, so ScS is too much delayed compared to the data . The slow transition from low velocity layer to ALVS in model LVZ1 delayed ScS too much.
Discussion and Conclusion
For station LSZ, ScS and SKS are in the slow range for all distances, while S samples more anomalies with increasing distances, leading to smaller SKS-8 and ScS-8. Behind the TSUM and BOSA stations, the S raypaths are not in the anomalous region, while the SKS raypaths sample more slow anomalies with increasing distances, leading to an increasing SKS-8. Since the velocity structure (velocity jump) of the fast D" causes a tripling of S, the low-velocity layer can cause diffraction effects on the S waveform.
Chapter 6 Modeling PKP Precursor with WKM
Introduction
The SKS phases from the deep South American events to the African station AAE show distortions caused by P, SKP dS diffraction, which travels along the CMB as shown, [Camero et al., 1993]. Strong PKP precursors have been successfully modeled with ULVZ for the path encountered in the mid-Pacific upwelling. Here, we extend the previous study of the ULVZ structure located beneath central Africa by Heimberger et al.
Waveform Data and Analysis
Since the western boundary of the ULVZ appears to be further away from the ATD, the difference may be due to geometry. For FURl, the SKP dS segments sample the ULVZ for distances from 109 to 115, while for ATD, which is about 5 degrees east of FURl, the SKP dS path does not sample the ULVZ for epicentral distances less than about 109°. However, with increasing distances, the longer SKP dS segments begin to sample the ULVZ and complicate the waveforms as shown.
Modeling Broadband PKP Precursors
A schematic picture to explain the different waveform complexity at AAE and ATD is shown in figure 6.2c. Most of the analog data observed at AAE (same location as FURl) produce results similar to those seen at FURl in figure 6.2b. Record sections of synthetic material based on PREM and dome like ULVZ model are displayed in figure 6.5 and 6.6 respectively.
Time (s)
DOME ULVZ
Time (T -1.80)
Discussion
Essentially, the only part of the Earth where the DF path is significantly separated from dispersal paths (Figure 6.1) is in the lower mantle. Furthermore, the position of the proposed ULVZ as outlined in Figure 6.2d is consistent with strong short-period dispersion as reported recently by Hedlin and Shearer [2000]. This result is generally consistent with local small-scale mantle convection caused by instability of the thermal boundary layer [ Olson et al ., 1987 ].
Appendix A Seismic Evidence for Ultra Low Velocity Zones Beneath Africa and
A .2 Introduction
Analysis
- a) Anomalous Waveform Data Sampling the Base of the Mantle Beneath Iceland and Africa
The paths for SKS, SP dKS and SKP dS are very similar in core and mantle. However, SP dKS and SKP dS also contain short P-diffractions along the base of the mantle (figure A.3). The bifurcation of the SKS can be seen in the PREM synthetics shown on the right in Figure A.4, as the SP dKS and SKP dS form a single pulse (both phases have identical travel time) that is well separated from the SKS beyond about 114°.
The waveform corresponding to the southernmost path to AAE, marked by the black segment in Figure A.10, is not that anomalous, and neither are many of the Tanzania Array waveforms. It appears that the SKS output points are close to maximum slack, a situation not unlike that of the productive column (a) in Figure A.10. The path geometry appears to be consistent with most observations (figure A.14) where SKP dS interference occurs at a relatively greater distance than in AAE.
Discussion
Although the evidence is weaker than under Africa, there appears to be a vertical structure under the southern Mid-Atlantic Ridge, i.e. almost 55° in the top panel of Figure A.12. The first two columns, Syn2Dl and Syn2D2, contain Green's functions as shown in Figure A.lO (a) and (d). The synthetic Mix lD includes the layered approach with PREM on the source side, and a 5% dip in a 30 km thick layer beneath the array (see Figure A.lc).
Bibliography
Heimberger, Further structural constraints and uncertainties of a thin laterally variable ultralow-velocity layer at the base of the mantle, J. Meyer, Seismic evidence of partial melting within a possibly ubiquitous low-velocity layer at the base of the mantle , Science. Heimberger, Dynamics of a phase change at the base of the mantle consistent with seismological observations, J.
