Short scale heterogeneity in the lowermost mantle: insights from PcP-P and ScS-S data

Hrvoje Tkalcic and Barbara Romanowicz

Introduction and Motivation

An important question in global geodynamics is whether the 3D seismic velocity anomalies, as seen in tomographic models of the mantle, are of a thermal or a compositional nature, or a combination of both. While global tomographic S models consistently show predominance of long wavelength structure at the base of the mantle, and in particular two large slow domains in the central Pacific and under Africa, referred to as "superplumes", seismic evidence for shorter scale heterogeneity at these depths is growing. The existence of strong heterogeneity in the vicinity of the two superplumes has been documented through forward modeling of seismic travel times and waveforms. Recent studies have also found evidence for locally rapid variations in other areas such as middle America.

On the other hand, several studies have compared global variations in S and P velocities from tomographic inversions, and some of them found evidence for a strong increase in $R= \partial lnV_s/ \partial lnV_p$ near the bottom of the mantle, as well as a possible decorrelation between shear and acoustic velocity variations below 2000 km depth. Furthermore, while there is evidence for zones of strongly reduced P velocity at the base of the mantle, there is no evidence that these are accompanied with a comparable drop in S velocity.

Through an inversion of PcP-P and PKP(AB-DF) data, we recently obtained maps of D" with lateral P velocity variations that show much shorter wavelength features than seen in S tomographic models (Tkalcic et al., 2002). Here, we consider PcP-P and ScS-S travel time residuals, and analyze their spatial relationship in several regions which are sampled both by PcP and ScS, in comparison to tomographic maps. We discuss implications of our results for the relative variations in P and S velocities in the lowermost mantle, in particular from the stand-point of $\partial lnV_s/ \partial lnV_p$ ratios and their interpretation.

Results

We compared the lateral variations observed in a global dataset of PcP-P differential travel times sensitive to structure in the bottom 500-1000 km of the mantle, to the predictions of global tomographic models, on the one hand, and ScS-S differential travel times, on the other.

Figure 38.1: Best fitting depth profiles of $R= \partial lnV_s/ \partial lnV_p$ obtained by comparing PcP-P travel time data to S tomographic velocity models, using a parameter search for the thickness of the bottom layer and the value of $R$ inside it. Comparison restricted to a subset of PcP-P residuals with PcP reflection points under eastern Eurasia with global S tomographic models $SAW24B16$ (Mégnin and Romanowicz, 2000) (solid line, variance reduction 57%), and models $SB4L18$ (Gu and Dziewonski, 2001) and $S362D1$ (Masters et al., 2000) (thick dashed line, variance reduction 60% and 52%, respectively). The thin dashed line corresponds to the best fitting depth profile of $R$ using the global PcP-P dataset and model $SAW24B16$ (variance reduction 10%). Shaded area represents the uncertainty in the thickness of the lowermost layer for the parametrization used.

This comparison has shown that the different datasets are in good agreement in some regions, in particular in eastern Asia, where fast anomalies over a broad domain are inferred both from tomographic P and S models and from the core-reflected phase data, resulting in a value of $R$ compatible with a thermal origin of heterogeneity at the base of the mantle in this region (solid and thick dashed lines in Figure 38.1). It is interesting to compare this regional result with that obtained by applying the same methodology to the global dataset. The best variance reduction is obtained for model $SAW24B16$ (Mégnin and Romanowicz, 2000) and does not exceed $10\%$. Figure 38.1 also shows the best fitting $R$ profile (thin dashed line) obtained with this model.

On the other hand, the PcP-P data indicate shorter scale lateral variations in many other regions. We studied in detail two such localized regions, under central America and south-east Africa, which correspond to downwelling and upwelling regions respectively, as seen in global tomographic models. In central America, lateral variations in P and S velocity appear to track each other. With only a few exceptions, the variations in PcP-P residuals are in good agreement with the slow (resp. fast) regions delineated by ScS-S from the study of Wysession et al. (2001). Resulting estimates of the ratio $R$ in the last 500 km of the lower mantle are not particularly high. Under Africa (Figure 38.2a), one profile (AB) shows similar results (Figure 38.2b), whereas a slightly more northerly neighboring path (profile CD) indicates fast P velocities in the heart of the low S velocity African "plume" (Figure 38.2c). A similar situation is found on the eastern edge of the Pacific plume, where the simultaneous availability of PcP-P and ScS-S data allows us to infer the existence of sharp lateral gradients across compositionally different domains.

Our study documents existence of strong lateral variations at the base of the mantle on scale lengths of several hundred kilometers, implying the existence of compositional variations in D". We note however that the computation of meaningful estimates of $R$ at the base of the mantle, as an indicator of the nature of heterogeneity, remains a challenge. In order to do it correctly, better spatial sampling in both P and S data is needed, as different smoothing of short scale structures can lead to biased estimates.

Collecting more PcP-P and ScS-S data with a compatible sampling of the lowermost mantle on the one hand, and increasing the resolution and accuracy of tomographic models on the other is necessary to gain further insight regarding the nature of heterogeneity in the lowermost mantle.

Figure 38.2: PcP-P travel time residuals plotted at the surface projections of PcP bouncing points for south Atlantic/south Africa region. The largest triangle and circle correspond to residuals of +3 s and -1.5 s, respectively. The background model is $SAW24B16$. Green crosses are ScS reflection points. (b) Cross-section through $SAW24B16$ along profile AB, with P and PcP paths from a South Sandwich Islands region event to SUR, BOSA and LBTB stations and a south Atlantic ridge event to Tanzania network; (c) Cross-section through $SAW24B16$ along profile CD, with P and PcP paths from a South Sandwich Islands region event to BGCA station.

Acknowledgements

Data were obtained from the IRIS Data Management Center. We thank Nicolas Houy, who measured most of the PcP-P travel times used in this study and Michael Wysession for sharing his MOMA S-ScS dataset with us. Figures were made with the General Mapping Tools (P. Wessel and W. H. F. Smith, EOS Trans. AGU 76, 329, 1995).

We thank the National Science Foundation for support of this research.

References

Gu, Y., A.M. Dziewonski, Shear velocity of the mantle and discontinuities in the pattern of lateral heterogeneities, J. Geophys. Res 106, 11169-11199, 2001.

Masters, G. , G. Laske, H. Bolton, and A.M. Dziewonski, Earth's Deep Interior: Mineral Physics and Tomography From the Atomic to the Global Scale. Geophys. Monogr. Ser. (eds Karato. S.-I. et al.) 117, AGU, Washington, DC, 63-87, 2000.

Mégnin, C., B. Romanowicz, The 3D shear velocity structure of the mantle from the inversion of body, surface, and higher mode waveforms, Geophys. J. Int., 143, 709-728, 2000.

Tkalcic, H., B. Romanowicz, N. Houy, Constraints on D'' structure using PKP(AB-DF), PKP(BC-DF) and PcP-P travel time data from broadband records, Geophys. J. Int., 148(3), 599-616, 2002.

Tkalcic, H., B. Romanowicz, Short scale heterogeneity in the lowermost mantle: insights from PcP-P and ScS-S data, Earth Planet. Sci. Lett., 201(1), 57-68, 2002.

Wysession, M.E., K.M. Fischer, G.I. Al-eqabi, P.J. Shore, Using MOMA broadband array, ScS-S data to image smaller-scale structures at the base of the mantle, Geophys. Res. Lett., 190, 167-170, 2001.

Berkeley Seismological Laboratory
215 McCone Hall, UC Berkeley, Berkeley, CA 94 720-4760
Questions or comments? Send e-mail: www@seismo.berkeley.edu
© 2002, The Regents of the University of California.