The Earth's outer core, composed primarily of liquid iron, is not purely iron but includes a significant amount of lighter elements (5-10% density deficit). These light elements, such as hydrogen, carbon, nitrogen, oxygen, sulfur, and silicon, are released during the crystallization of the inner core as it grows due to the Earth's cooling. This process and its impact on the geodynamo (the mechanism generating Earth's magnetic field) are not fully understood. The distribution and movement of these light elements could influence seismic wavespeeds. Analyzing seismic waves from earthquakes provides a way to probe the outer core's properties. This study aims to determine if there are detectable temporal variations in seismic wavespeeds in the Earth's outer core on decadal time scales, providing insights into the dynamics of outer core convection and light element transport.

Previous research has established the presence of light elements in the Earth's outer core, necessitating a density deficit in global Earth models to match seismological observations. Studies on the accretion of the Earth and core segregation also contribute to the understanding of elemental composition. Seismic waves, especially core-penetrating SKS waves, have been used to constrain lateral heterogeneities within the outer core. However, investigating temporal changes in seismic wavespeeds on decadal timescales using paired earthquakes at similar locations remains a relatively novel approach.

This study analyzes the differential travel times of core-penetrating SKS waves generated by pairs of large earthquakes (magnitude >6.0) occurring in different decades (1990-2019) with close hypocenters (<100km apart). SKS waves travel as shear waves in the mantle and compressional waves in the core. Using pairs of earthquakes helps minimize uncertainties related to source origin times and clock errors. The arrival times of SKS waves were compared to reference waves (Pdiff, Sdiff, PP, and SS) to detect differences attributable to outer core property changes rather than earthquake location variations. A grid search was performed around the USGS-provided hypocenter locations to account for possible earthquake location uncertainties. Waveform modeling using the Spectral Element Method (SPECFEM) was employed to constrain the size and magnitude of potential wavespeed anomalies in the outer core, using finite-frequency sensitivity calculations to identify the most sensitive regions to perturbations along the seismic ray paths. The study focused on double differential travel times between SKS waves and reference waves, effectively minimizing uncertainties associated with source origin times and possible clock errors.

The analysis of 55 high-quality SKS measurements from seven earthquake pairs revealed that most travel time measurements did not require a temporal change in the outer core. However, five anomalous SKS wave measurements showed a significant advance in arrival time (approximately 1 second) at the time of the second earthquake compared to the first earthquake in each pair. This advance couldn't be explained by earthquake hypocenter mislocations. These anomalous SKS waves all propagated through the upper half of the outer core in the low-latitude Pacific. The 1-1.5% increase in P-wavespeed in these localized regions (approximately 800km radius) suggests a 2-3% density deficit, potentially due to a localized increase in the concentration of light elements, possibly sulfur. Wave propagation modeling using SPECFEM indicated that a P-wavespeed increase of 1-1.5% within a region of about 800 km radius and a depth extent of approximately 800 km in the upper outer core is consistent with observations. This suggests a transient process involving upward flows carrying high concentrations of light elements, with an estimated flow speed of around 40 km/year. The specific mechanisms of these flows remain to be fully understood but are consistent with geodynamo simulations featuring thermochemical flows.

The observed temporal changes in seismic wavespeeds challenge the assumption of a uniformly mixed outer core on decadal timescales. The findings directly address the question of temporal variations in outer core properties, demonstrating that localized, transient changes occur. The 2-3% density deficit is more readily explained by a concentration of light elements than by temperature changes due to the weak temperature dependence of sound speed in liquid iron at core pressures. The estimated flow speed of approximately 40 km/year aligns with results from self-consistent geodynamo simulations. The findings suggest that outer core convection is more complex and heterogeneous than previously thought, potentially involving localized, rapidly varying flows. Future studies should investigate the connection between these observed changes and existing models of outer core stratification and the influence of lowermost mantle heterogeneities on core flow patterns.

This study presents compelling evidence for localized, transient variations in seismic wavespeeds in the Earth's outer core, suggesting fast fluid movement possibly driven by the upward transport of light elements. The findings are consistent with theoretical models of core convection and geodynamo simulations, but the precise mechanisms underlying these transient phenomena need further exploration. Future research should focus on expanding the dataset, improving the resolution of seismic imaging, and refining geodynamo models to better understand the detailed dynamics of outer core flows and light element transport.

The study is limited by the availability of suitable earthquake pairs and the inherent uncertainties associated with seismic data analysis. The interpretation of the observed wavespeed changes as solely due to density perturbations requires further confirmation. Although the grid search helped minimize the uncertainties related to earthquake locations, there might be additional small-scale heterogeneities in the mantle that could influence the SKS travel times, though modeling suggests that the effects are negligible in this case. The estimated flow speed of 40 km/year is a rough estimate based on an assumed timescale of two decades; a shorter time interval would imply a faster flow speed.