New techniques for analyzing long-period seismic data to determine three-dimensional thermal and compositional structure of the Earth's mantle

Manually measuring travel times of long-period phases, while producing data sets that have high quality and reliability, has become unfeasible with the rapidly expanding global seismic network. We have taken advantage of the similarity of long period waveforms to apply a cluster analysis method to S, SS, P, and PP waves. Our cross-correlation determined time shifts achieve the high data quality of fully manual measurements in a fraction of the time. These relative time shifts are used in tomographic inversions to create detailed, robust models of mantle shear and compressional velocity structure.; Our cluster analysis SS arrival times are used to augment an existing SS dataset to measure the relative time of SS to its reflections off the 410 and 660 km seismic discontinuities. The 410 and 660 km discontinuities are expected to be anti-correlated and the transition zone thickness should be correlated to the transition zone velocity anomalies according to the current understanding of mantle composition and mineralogy. However, we find that the 410 km discontinuity is correlated to 660 km discontinuity, is not correlated to velocity anomalies in the transition zone although the transition zone thickness is correlated to the velocity anomalies. These observations indicate that compositional variations may be affecting the 410 km discontinuity.; We compare our shear and compressional velocity anomaly models, finding that large values of the ratio of shear to compressional velocity and negative values of the ratio of shear to bulk sound velocity are concentrated in the lowermost mantle. These observations cannot be explained by thermal affects alone, indicating chemical heterogeneity. We have taken this analysis a step further by actually inverting for thermal and chemical heterogeneity. Using newly determined sensitivities to temperature and mineralogy, the largest dataset of long-period S and P phases to date, and normal mode splitting measurements, we find that most velocity and density anomalies in the lower mantle can be explained by variations in temperature. However, in the central Pacific, compositional variations are also present in the form of an increase in the mole fraction of perovskite and an increase in the mole fraction of iron in the perovskite.