The author benefited from many stimulating discussions with past and present staff and students of the Seismology Department. The initial location of the fault coincides with the lower, northernmost edge of the north-dipping main thrust.
INTRODUCTION
The core is subject to more variable interpretations than perhaps any other region of the Earth's interior. The revision of the previous article took into account the developments in the area of the core structure that culminate in Chapter 6.
MANTLE
CORE
These phases are called P'dP', as defined by Whitcomb and Anderson (1970), and indicate a reflection of the P'P' phase at a depth d in kilometers. Distribution of the P'P' reflection points extends about 3° north of the Atlantic-Indian Rise and about 5° west of the Ninety East Ridge as seen in Figure 2.2.
II REFLECTORS
NINETY- EAST RIDGE
ATLANTIC- INDIAN RISE
AFRICAN RIFT
Figures 2.6 and 2.7 show the calculated depths from the precursor measurements as a function of distance to the Atlantic-Indian rise and the Ninety-East ridge (in Figure 2.2, perpendicular distance to lines A-A' and B-B', respectively). The size of the symbol indicates the degree of confidence and the type of symbol indicates the event.
NINETY-ERST RIDGE
EVENT
DISTANCE FR~M RIDGE IN DEGREES
The P'dP' phase arrives at the array from a direction opposite to that of the event epicenter. The depths of the P'dP' phases are calculated as in Chapter 2 and are shown in Figure 3.5 for the ten events.
EPICENTRRL DISTANCE. DEGREES
LATITUDE
The next shallower level of readings is a zone between 360 and 410 km; under Antarctica and the rises the readings are at the extreme depth ranges of this zone, and under the East Indian Ocean they are in the middle of the range. Due to the subjective way in which the meaning of the power peaks is determined, a second method of picking phases is used.
Ill I
ASYMMETRIC P'P'
Asymmetric P'P' phases reflecting from dipping interfaces at or near the Earth's surface would arrive earlier than the main phase for all branches except a special path for the AB branch. Asymmetric reflection types: in the plane of the great circle and out of the plane of the great circle.
Because the amplitudes of P'P' are not as well known as those of P', we translate amplitudes of P'P' are not as well known as those of P', we translate P' into symmetric P'P' amplitudes for use as comparisons . Bullen (1963, p. 127), the energy density due to geometric spreading of a seismic beam on the surface of the earth can be written in the form Now suppose that T1(~) corresponds to P' of the DF branch; for a surface focus, T11(2~) corresponds to P'P' of the same branch.
All of our observations are within 1 to 2 seconds, so for our purposes the amplitude is directly proportional to the square root of the energy. Now we can see that the relationship between the symmetric amplitude P'P' and the corresponding amplitude P' from (4.8) and (4.9). The amplitude data for P' in Fig. 4.3 can now be converted to P'P' amplitudes using equation (4.10), which only involves geometric spreading.
120 Epicentral distance, deg
But at least part of the difficulty lies in assigning the wrong P1P1 branch to a read and in. The problems of asymmetric P1P1 are avoided by isolating the P1P1 reflection from the ocean surface, which is perfectly horizontal for purposes here. Identification of the appropriate P1P1 branch is done by using the LASA array in Montana to measure the dt/d6 of a phase and by determining the largest relative amplitude branch.
In addition, the dt/d6 or apparent wave slowness of some of the phases was determined using the LASA installation in Montana as a beamforming array as in. Map of reflection points on the East Ninety ridge, events 6,8, 11 (southern cluster) and events, 7,9,10 (northern cluster). In all time plots, solid dots represent grade 4, and open dots represent grade 3 (largest dots) to grade 1 (smallest dots).
E PICENTRRL DISTANCE
IN D EGREES
As discussed in Chapter 6, the P'GH phase between P'AB and P'DF is identified as P'Bc. However, the phase and its dt/d~ are well determined and the analysis here is independent of the terminology. It can be seen that the· largest amplitudes sometimes miss the predicted times for that branch by several seconds, such as. To compare all P'P' data with the largest amplitude to each other, regardless of the branch they represent, all of them.
It should be remembered that these predicted times correspond to the continents, since that is where most of the P' data are collected. If it is first assumed that the average velocities of the first mantle from the P' stages do not change, then the main variation of the reduced time P'P' will be due to changes in the reflection point of the crustal structure, the water depth and of course the depth . and reflector dip. Three crustal velocity models shown in Figure 5.6 that differ only in the upper 56 km are used to estimate temporal variations.
RTLRNTIC-INOIRN RISE
Wave velocity, km/sec
CIT 208 from
Because the lowest velocity in the models is that of sea water, variations in ocean depth dominate. the variations in time for surface reflections of the models, and the data are best presented as a function of water depth at the reflection point. The best data are from Ninety-East Ridge, events 6, 8, and 11, and Figure 5.7 shows the reduced times as a function of water depth (again, zero time corresponds to a surface reflection under a continent). The dashed lines in Figure 5.7 indicate interpolated times calculated for a reflection from the sea surface to the sea floor between model Oc, with a 5.3 km water depth, and model 90°E, with a 2 km depth.
If the forecast times are increased by 2.5 seconds, a reasonably good fit of the maximum amplitude arrivals to the surface and bottom reflection times can be obtained. This fit, shown as solid lines in Figure 5.7, is made primarily from ocean surface reflections, which should have the smallest scatter. The three late points near 2 km ocean depth are probably due to using the wrong branch to shorten their times; are close to the true time for AB, but less than 62.5° apart, so branch BC was used for reduction.
WRTER DEPTH IN KIL~METERS
Most importantly, there is a gap between ocean floor and ocean surface time, as you would expect; this means that there should be no reflection in the water. The reduced-time data from events 7, 9, and 10 on the Ninety-East Ridge (not shown) do not show a systematic distribution, probably because the reflections are too closely spaced (Figures 5.1 and 5.5) and the bathymetric data in this area too sparse for a useful comparison decreased times with water depth. A plot of Atlantic-Indian upwelling reduced time as a function of water depth is shown in Figure 5.8.
The largest amplitude data do not also fit the slope of the calculated sea surface reflectance. However, the bathymetry data is based on very few ship tracks, and the lower water depths may be systematically plotted too deep, as any smoothing caused by missing data will wipe out the high-bottom features.
WATER DEPTH IN KILCJMFTERS
- CORE STRUCTURE
The nature of the mantle in the two source regions is unclear; both sets of events are associated with trench structures that represent the downgoing lithospheric plate in plate tectonics theory. This is not a large delay and is of the same magnitude as the station delays observed in the Basin and Range Province relative to. The times of the P'P' phase were previously thought to be too widely distributed to be useful for studies of the Earth's structure.
The problems appear to be due to asymmetric P'P', wrong identification of the branch, and lack of the right velocity structure at the reflection point. They are resolved using the horizontal sea surface as a reflector, using the LASA array and largest relative amplitudes as branch identification tools, and using detailed bathymetry maps at the reflection points of P'P' . The motivation for making a new determination of the nuclear velocity structure stems from some major improvements in nuclear data in the areas outlined above.
REGION SAMPLED BY
A one second shift in the time of PKiKP (or in the . travel time through the outer core) gives only a 5 km shift of the inner core radius. When comparing models, the important factors are the sensitivity of the data to velocity at specific levels in the core and the direction of. It is interesting to note that the P'AB-P'DF differential times of the final model shown in Figure 6.4 are generally parallel to the.
However, the time of the PmKP phase is not only sensitive to the velocity in the nucleus, but also very sensitive to the radius of the nucleus. The angular distance from the station to the scattering point at the boundary is approximately 21°, as shown in Figure 6.12. The final core velocity model depends mainly on the dt/dlls of the core stages.
Traveltime differences are on the order of 0.8 seconds for p· waves and 5.7 seconds for S waves. Helmberger, Shear velocities at the base of the mantle from S and ScS observations, submitted to J.
PART II
DATA SET
It is therefore desirable to select the set of aftershocks most representative of the major tectonic stress release. A practical limit is set at the lower end of the size scale due to a limit on the size of the data set and the signal strengths that the stations record. From this reasoning, the aftershock set most representative of the regional tectonic activity includes all events above a certain size.
NORMAL
10 OTHER
The second set is defined as those shocks whose first P-wave motions are clear at most permanent CIT stations, most of which are between 90 and 30'0 km from the epicenters. Again, the homogeneity of the series of aftershocks is compromised by the occasional immersion of the first movements in the coda of the previous shock. We can only assume that the masking is distributed randomly with respect to the aftershock type and that the stress release in the first few hours is essentially indistinguishable from the rest of the set except in rate.
Mt = 3.0 after the first two days have first motions clear enough to be included in the second set, and this can be considered an approximate magnitude cutoff of ~ = 3.3. The stations used for P-wave first motions in this study with their operational agencies, coordinates and periods of operation are given in Table 2. The number of readings for most events ranged from 10-20 for the first twenty hours of the aftershock sequence and 20-30 for the rest of the study period.
