by

A.L.HALES and S.BLOC
Geosciences Division
Southwest Center for Advanced Studie
Dallas, Texas

The structure of the tipper mantle, as determined from surface wave dispersion, is believed to be characterized by thick (about 200 klm) low velocity layers of characterized by thick velocity contrast and by two regions of relatively rapid increase of velocity at depth of about 350 and 650 km. The upper mantle structures inferred from the data of Toksöz and Anderson for shields, tectonically active and occanic areas are shown in Fig. 1. Press, also fivids relatively thick low velocitv channels. The thick contrast low velocity layers seem to be, it wide feature.

The S velocity structures determined by Ibrahim and Nuttli from studies of body waves are. qualitatively in accord with the data for surface waves.

Studies of P wave travel times made since 1958 tend to show three in tilt, slope of the travel tinie curvo between 5 wid 30' as is shown by Table, 4 of Johnson and by Fig. 19 of Anderson. The determination of the steepnes's ofthe transitions is clopendent on very detailed coverage of the distance between 10º and 35º as illustrated by the calculated times for a series of models in Fig. 2. In model A the transition takes place over 400 km and there is a very small cusp. This cusp could be eliminated by it smoother approach to the transition. As the depth over which the transition takes place is reduced, the cusp extends. The reaclies its maximum extension for the first order discontinuity in model F. The shapes of the cusps, the relative amplitudes along the varion, branches and the distance at which the break in the slope of the travel time curve occurs vary depending on the velocity gradients above and below the regions of rapid change, but the curves in Fig. 2 show all the characteristics of travel time curves for regions of rapid velocity change. It is important to note that rays bottoming between the depths 200 and 360 km on model F are never observed as first arrivals, so that the velocity distribution can be completely determined only if the rate arrival phases can be observed. These hidden regions are characteristic of all structures with cusps.

Studles of P body wave travel times have recently been made in the United States and there was fairly detailed coverage of the 10º to 30º region by Johnson for fit the western United States, by Green and Hales and by Julian and Anderson for the central and south-western, and by others for the north-western United States. The models of these authors are compared in Fig. 3. The transitions differ in detail, but the larger features of the distributions agree well. As far as the regions of rapid change of velocity are concerned, the body wave and surface wave dispersion models are consistent.

Dowling and Nuttlis, and Lehmann, showed that the characteristic effect of the kind of low velocity zones which can be expected in the upper mantle is an offset of the travel time curves and not a shadow zone. The effect is illustrated in Fig. 4, which shows calculated travel times for a number of models. Models LA, LB and LC have almost identical offsets showing that it is difficult to separate out the effects of velocity and thickness on the basis of travel times alone. Given very detailed coverage it might be possible to discriminate between models by determining the position of the cusp. It is clear from the series of models LA, LD, LE and LF that It will be difficult to identify thin low velocity zones on the basis of body wave travel times.

The analyses of Lehmann and of Green and Hales make it clear that the P bodv wave travel times in the regions to the ESE, E and ENE of the Nevada test site require that there should be a P low velocity zone in the upper mantle. Other body wave evidence for the existence of P low velocitY layers in the upper mantle has been discussed by Lehmann, but nowhere else is the evidence as clear as in the western United States. In particular the body waves travel times reported for stable continental areas do not show the offsets which should exist if the low velocity lavers are thick.

Thus there is a marked contrast between the body wave information and the surface wave studies which require a universal world encircling, moderately thick, low velocity zone. Surface wave data are very much more sensitive to the S velocity distribution than to the P wave velocity distribution or the density. Laboratorv measurements of the temperature coefficients of the P and S velocities suggest that the critical temperature gradient for constant S velocity is between half and two-thirds that for P. Thus it is possible that there may be a low velocity zone for S and not for P. Anderson has estimated that if the S velocity decreases by less than 0.1 km/s over 100 km then there would be no P low velocitv zone. Almost all the surface wave studies, however, show decreases of S velocitv of more than 0.1 km/s over 100 km. The Toksöz and Anderson studies covered oceans, shields and tectonic areas and were therefore fairly general. It is significant that the Brune and Dorman study oft he Canadian Shield, a region of high surface wave phase velocity, shows a decrease of more than 0.1 km/s over 100 km for the S wave velocities. Similar studies in southern Africa by S. B., A. L. H. and Landisman (in preparation) show decreases exceeding 0.1 km/s over 100 km if interpreted in terms of thick low contrast low velocity zones. Thus it seems from the surface wave data that there should be a P low velocity layer everywhere. It is possible that it is not observed because it occurs in the later arrival segment of the travel time curve marked PQ on the curves of Fig. 2. The other possibility which must be considered is that the low velocity i layers in the shield areas are very thin and escape detection for this reason. In this connexion it should be noted that Green and A. L. H. considered the possibility that there was a thin low velocitv layer in the upper mantle beneath the central United States. The observations are shown in Fig. 5. Calculated travel times for two models, one with a thin low velocity layer and the other without, are also shown in the figure. Green and Hales remarked that the amplitudes calculated for the model with the low velocity zone fitted the observed amplitudes slightly better than those calculated for the model without a low velocity zone, but the possibility that the offset arose as a result of lateral variation could not be excluded.

It is important to recognize that the surface wave data do not exclude the possibility that the low velocity layers are thin. Aki showed that a laminated mantle with 2 per cent of material of velocity 1.1 km/s fitted both the Japanese Love and Rayleigh data very well. The only other models which would fit the observations were anisotropic models. Aki’s results consistent with other studies which showed that laminated layering produced similar effects to anisotropy, and important because there are other cases in which it has been found impossible to find an isotropie shear wave velocity model which would fit both Rayleigh and Love wave dispersion.

Aki remarked that it was not necessary that there should be multiple thin layers. Fig. 6 shows the Rayleigh wave dispersion for three models, one with a thick low velocity layer, the other two with very thin lavers. The differences between the velocities at any period are less than 0.015 km/s; that is, much less than the errors of observational data.

The figure shows the Love wave dispersion for the same models. In this case the dispersion for the thick low velocity zone is not the same as that flor the thin. If dispersion in the real Earth is caused by thin low velocity channels, it not be possible to fit the observed data for both Rayleigh and Love waves with a model with a single thick low contrast low velocity layer. This is what has been found in Japan and the central United States. It therefore seems possible that a substantial proportion of the low velocity effect on the surface wave dispersion is due to thin layers of high contrast.

In regions such as that in the western United States where the P travel times show offsets of 4-6 s the layers must be fairly thick (for example, 20-40 km), even if liquid.

There are probably many more regions the western United States. Studies of the regional variations of teleseismic travel times have shown that the P arrivals are systematically late in the western United States. Similarly it has been shown that the S arrivals are late in this region. Other regions for which the P arrivals are late are the Andes, the Carpathians, the Apennines, the Tibetan Plateau and south-eastern Ausralia. It is probable that the upper mantle velocity distributions in these regions resemble those in the western United States very closely.

The conclusion to be drawn from this discussion of body wave travel times and surface wave dispersion is that both sets of data tend to favour the hypothesis that the low velocity layers are relatively thin zones of high velocity contrast rather than thick zones of love velocity contrast. The effects of replacing a thick low contrast, low velocity layer by a thin near liquid layer on the periods of the free oscillations and on the velocities for very long period surface waves have yet to be studied.

The existence of low velocity lavers probably plays an important part in tectonic movements and relative motion of the continents. Thus better determinations of its properties, especitilly of its thickness, velocity and depth, are important. It is clear from Fig. 6 that careful studies of Love and Rayleigh dispersion in the 50-100 s period range would resolve the question.

This work supported by a grant from the US Aeronautics and Space Administration. We thank Professor Mark Landisman. for helpful discussion, Mr R. Massé for the use of his program for the computation of surface waves dispersion, and Mrs J. L. Roberts for assisteance the computations.

Received October 28; revised December 9, 1968.

References:

Anderson, D. L., The Earth's Mantle, 12 (1967).

Toksöz, M. X and Anderson, D. L J. Geophys. Res., 71, 1649 (1966).

Press, F J. Geophlys. Res., 73, 5223 (1968).

Ibrahim, A. X_ and Nuttli, 0. W., Bull. Seismol. Soc. Amer., 57, 1063 (1967).

Johnson, L. 11- J. Res., 72, 6309 (1967).

Green, It. W. E and Ifales, A. L., Bull. Seismal. Soc. Amer., 58, 207 (1968). itiffian, 13. 11--- and Anderson, D. L., Bull. Seisinol, Soc. Anter., 58, 339

Dowling, J and Nuttli, 0., Bull. Seismot. Soc. Ainer., 54, 1981 (1964).

Lelirnann, L, The Earth's Mantle, 3 (1967).

Brune, J and Dornian. J Bull. Mismol. Soc. Anter., 53, 167 (1963).

Aki, K., J. Coophys. Res., 73, 585 (1968).

McEvilly,T. V., Bull. Seismol. Soc. Amer.,, 54, 1997 (1964).

Ilustraciones

Figura 1 (volver al texto)	Figura 2 (volver al texto)
Figura 3 (volver al texto)	Figura 4 (volver al texto)
Figura 5 (volver al texto)	Figura 6 (volver al texto)