Seismic Constraints on the African Anomaly in the Lower Mantle

Seismic studies have consistently shown a significant low-velocity anomaly in the lower mantle beneath Africa. For convenience, we term it "African anomaly". While it is now clear that the base of the "African anomaly" is compositionally distinct, it remains unclear whether the portion in the lower-mantle has a same origin as its base. To further understand the origin of the "African anomaly", we study seismic velocity structure and geometry of the "African anomaly" in the lower mantle along the great-circle path from Passage Drake to Hindu Kush. We use the broadband data recorded in two PASSCAL seismic arrays in Africa, the Kaapvaal seismic array (1997-1999) and the Tanzania seismic array (1994-1995), for earthquakes occurring in the southern Pacific Ocean, Passage Drake, the South Sandwich islands, the Hindu Kush area and the Japan sea. The seismic data from those earthquakes sample the "African anomaly" in opposite directions and within a narrow azimuth range of less than 11^o. They thus provide an opportunity to study the lower mantle structure of the "African anomaly" from both directions. The southwest side of the anomaly is sampled by direct SH, ScS and SKS phases from earthquakes occurring in the South Sandwich islands and the south Pacific Ocean. SH phases travel times from the South Sandwich islands earthquakes show no delay with respect to the preliminary reference earth model (PREM) between 44^o~ and 59^o, but they are increasingly delayed from about 3 s at 68^o~ to 11.7 s at 74^o. These direct SH travel times suggest that the "African anomaly" extends about 1300 km above the core-mantle boundary. The ScS travel time delays have a simple trend increasing from 4.5 s at 44^o~ to 20.5 s at 74^o. The SKS phases for an earthquake occurring in the south Pacific Ocean have a uniform delay about 6 s before 101^o, gradually increasing to a uniform delay of 12 s after 106^o. These ScS and SKS travel time observations are consistent with a model that, in the southwest side of the "African anomaly", the basal layer extends further southwest and the anomaly extends into the lower mantle dipping toward its center. The northeast side of the anomaly is sampled by the SKS phases from an earthquake occurring in Passage Drake, direct SH and ScS phases from events in Hindu Kush, and SKS and SKKS phases from an earthquake in the Japan sea. The SKS phases recorded for a Passage Drake earthquake show decreasing delays from 10 s at 86.9^o~ to 5.6 s at 93.5^o. Both the SH and ScS waves from earthquakes occurring in the Hindu Kush area do not exhibit travel time delays with respect to PREM, placing bounds on the extent of the anomaly in northeast side. The SKKS phases from an earthquake in the Japan sea are slightly delayed only after 140^o~ whereas the SKS phases have a uniform delay about 4 s between 129^o~ and 140^o. These travel times delays are consistent with a model that, on the northeast side, the anomaly also dips toward its center. The geometry of the "African anomaly" with both edges dipping toward its center, suggests that the portion of the "Afrian anomaly" n the lower mantle likely has a same origin as its base. It also indicates that the "African anomaly" may represent a stable and long-living compositional anomaly, similar to those in the geodynamical models and the laboratory experiments.