Evidence for proportionality ofPandSheterogeneity in the lower mantle

Abstract

To date, most shear-velocity heterogeneity models in the lower mantle have been derived using long-period data. Comparatively little use has been made of the vast International Seismological Centre (ISC) data base of shear-wave arrival times. The aim of this study is to use the ISC P and S arrival times to construct global models of P and S heterogeneity in the lower mantle, and then to compare them in order to investigate whether, within the limitations of the data distributions, they might be proportional. The advantages of constructing both compressional- and shear-wave models from the one data set is that they share similar resolution properties, and hence are most reliable in the same areas. We use data from over 21 000 events to derive a data set of P and S summary rays whose residuals we invert jointly along with hypocentral parameters of over 600 summary events. Particular attention is paid to data weighting so that outliers are not given undue influence. Furthermore, summary rays with high internal variances are downweighted. In order to diminish the effect of model parametrization on our conclusion, we derive three sets of P and S models expanded in terms of Legendre polynomials for their depth variation, and spherical harmonics for their lateral dependence. Comprehensive resolution and error analysis is performed. Correlation coefficients between our P and S models are highly significant, averaging approximately 0.7 for our lowest parametrization (245 model coefficients), and 0.5 for our more highly parametrized models. Visual comparisons show strong similarities in areas where resolution is high and error is small. We also conduct an experiment in which we derive compressional- and shear-wave heterogeneity models from data sets which contain P and S arrivals from the same seismograms. These data sets sample the mantle almost identically. The resultant models compare well but correlation drops towards the core-mantle boundary, indicating that there are genuine physical differences in the lowermost mantle. Our models indicate that the ratio of relative S to P heterogeneities is close to 2. This value is based on both our complete and restricted data set models and hence is not highly dependent on data weighting.