[S18-P] Poster

[S18-P-02] Full parameter space search for a layered, anisotropic inner core using the Neighbourhood Algorithm

^○Joanne Stephenson, Hrvoje Tkalcic (Australian National University, Canberra, Australia)

The inner core of the Earth has long been modelled seismically by simple cylindrical anisotropy with the fast axis parallel to the Earth's rotation axis. This has been quantified with the use of the inner core phase PKIKP, utilising both differential and absolute travel times. This simplistic model is popular as, to some extent, it can explain both PKIKP travel time data and anomalous splitting of normal modes. However, recently it has become apparent that such a simple model fails to fit all observed travel time data. Travel time residuals from PKPbc-PKIKP differential times, plotted against the angle with the Earth's rotation axis, form a distinctive “L" shape where polar paths, typically from the South Sandwich Islands, have a residual range of ~0-5 seconds. A simple model of cylindrical anisotropy fails to fit such extreme range of values. Whilst many models with increasing inner core complexity (e.g. the inner most inner core) have been created on an ad hoc or systematic basis, a full parameter space search has not been comprehensively carried out for travel time data. We employ the derivative free search algorithm, the Neighbourhood Algorithm, to explore multidimensional parameter space for the case of an anisotropic, layered inner core. Using a combination of absolute PKIKP travel time data from the ISC catalogue, global high-quality PKPbc-PKIKP differential travel time data hand-picked by cross-correlation, and a high volume of high-quality hand-picked differential travel time data from the South Sandwich Islands earthquakes, we search for the optimal number of layers within the inner core and the associated parameters of the cylindrical anisotropy equation. Results from this process are presented, the outcomes of which have implications for geodynamic and mineral physics interpretations.