Transient variation in seismic wave speed points to fast fluid movement in the Earth's outer core

Seismological and geophysical observations indicate Earth’s liquid outer core is not pure iron but contains light elements, leading to a 5–10% density deficit relative to pure iron. As the planet cools, iron crystallizes at the inner-core boundary, releasing latent heat and expelling light elements that provide thermal and compositional buoyancy to power the geodynamo. The timescales and spatial patterns of associated thermochemical convection remain uncertain, particularly whether light elements are released and transported heterogeneously enough to affect seismic wave speeds on decadal timescales. This study asks whether the seismic wavespeed in the liquid outer core varies measurably over decades, and if so, what this reveals about transient, localized fluid motions and compositional anomalies. To test this, the author exploits SKS phases from pairs of large earthquakes with closely spaced hypocenters occurring in different decades. By comparing SKS arrival times to co-recorded reference mantle/core-diffracted phases (Pdiff, Sdiff) and surface-reflected phases (PP, SS), the analysis aims to isolate temporal changes within the outer core from confounding effects of earthquake mislocation or focal differences, as mantle convection occurs on much longer timescales.

Prior work has established that the outer core requires substantial light elements (e.g., H, C, N, O, S, Si) to explain its density deficit relative to pure iron, based on seismology and equations of state. Seismic phases sampling the core (e.g., PKP/PCP families, SKS) have been used to infer whole-Earth and core structures, with some studies suggesting outer-core heterogeneity or alternative explanations such as mantle anisotropy. Finite-frequency theory and adjoint methods have clarified how body-wave travel times and amplitudes are sensitive to 3D structure across Fresnel zones. Global studies have posited outer-core wavespeed heterogeneities exceeding ~0.5% to improve data fits, though theoretical constraints argue that sustained lateral density contrasts in the outer core should be very small unless transient. A possible stably stratified layer at the top of the outer core has been discussed, with mixed seismological evidence and dynamo-model implications. Hemispherical inner-core structure and complex inner-core growth have been proposed, potentially modulating thermochemical fluxes into the outer core.

Data selection and preprocessing: From the USGS PDE catalog (1990–2019), earthquakes with M ≥ 6.0 were searched to identify pairs occurring in different decades with epicenters within 100 km. This yielded 21 pairs; two were excluded due to focal-depth differences >35 km. Three-component broadband displacement seismograms were requested at FDSN stations for epicentral distances 98°–145°. Instrument responses were removed; horizontals were rotated to radial/transverse. Records were band-pass filtered for 20–100 s periods using a zero-phase Butterworth filter. Seismograms affected by foreshocks, dissimilar source radiation, or low SNR were removed, leaving seven earthquake pairs and 55 high-quality SKS measurements (most from pairs with a first event in the 1990s and a second in the 2010s). Differential and double-differential timing: For each pair, relative arrivals were measured by cross-correlation for SKS, Pdiff, Sdiff, PP, and SS (SmKS excluded due to depth-phase interference). Double-differential times ΔtSKS − ΔtX (X ∈ {Pdiff, Sdiff, PP, SS}) were formed to minimize uncertainties from origin times and clock errors. Measurement uncertainties were estimated using variable time windows by shifting start/end times up to ±10 s in 1-s steps; a conservative maximum uncertainty of 0.5 s was adopted. Hypocenter mislocation tests: A hypocenter grid search explored alternative locations at 1-km spacing within a 50-km-radius sphere around the USGS locations (extended up to 200 km in tests). For each trial location, predicted ΔtSKS − ΔtX and ΔtPP − ΔtSS (or ΔtSS − ΔtSdiff) were compared to observations; if all matched within ±0.5 s, temporal outer-core change was deemed unnecessary. Mantle heterogeneity assessment: Synthetic seismograms in the 3D mantle model S40RTS were used to estimate mantle contributions to double-differential times, which were found to be ≤0.3 s, with ΔtSKS − ΔtX approximately zero for this dataset. Finite-frequency sensitivity and waveform modeling: Using SPECFEM3D and adjoint methods, finite-frequency sensitivity kernels for SKS travel time and amplitude were computed (example station ULN; SKS turning latitude ~9.96°N) at 30 s period, giving a Fresnel-zone lateral radius ~800 km at turning depth. Spectral-element simulations tested localized outer-core P-wave speed anomalies of different sizes and strengths: a small, strong anomaly (radius ~6°, depth extent ~600 km, +4% VP) and a large, weak anomaly (radius ~12° ≈ 800 km, depth extent ~800 km above the geometric ray, +1.5% VP). Predicted time shifts and diffraction-induced amplitude/shape changes of SKS were compared with observations to constrain anomaly properties.

- Of 55 SKS measurements from seven earthquake pairs, the majority can be explained without invoking temporal changes in the outer core when accounting for possible earthquake mislocations and measurement uncertainties (≤0.5 s). - Five anomalous measurement sets cannot be reconciled by hypocenter mislocations even when expanding the search radius to 200 km. In all five, SKS from the second earthquake arrives earlier by ~0.8–1.5 s when reference phases (Pdiff, Sdiff, PP, SS) are aligned, indicating increased outer-core P-wave speed along the SKS path at the time of the second event (~20 years later for the main pair). - Four of these anomalies come from a single earthquake pair (first in 1997, second in 2018) recorded at multiple stations (ULN, YAK, COLA, INK). A fifth is from a 1998–2015 pair at station CCM. The SKS rays turn in the upper half of the outer core beneath the low-latitude Pacific. - Reference-phase differentials (e.g., δtPP − δtPdiff ≤ 0.4 s) indicate epicentral-distance differences between the paired events are small (<~15 km), further supporting that SKS advances are not source-geometry artifacts. - Finite-frequency modeling shows the observed SKS time shifts and preserved amplitudes/waveforms are consistent with a relatively large, weak anomaly rather than a small, strong one: a VP increase of ~1–1.5% within a localized region with lateral radius ~12° (~800 km) and depth extent ~800 km in the upper outer core. - If compositional, the inferred 1–1.5% VP increase implies a density deficit of ~2–3% associated with elevated light-element concentration (e.g., on the order of 1–2 wt% more sulfur), suggestive of transient, localized upward flows. - A simple distance-over-time estimate (≈800 km over ~20 years) suggests a characteristic flow speed on the order of ~40 km/year; actual speeds could be higher if the change occurred over a shorter interval. - No comparable SKS advances were observed on neighboring SKS paths, implying the affected regions are localized and not global in extent.

The repeated-earthquake SKS analysis isolates decadal-scale temporal variations in outer-core seismic properties along specific ray paths. The consistent early arrivals of SKS for the later events, when contrasted with stable reference-phase timing and small inter-event distance/depth differences, indicate a localized increase in P-wave speed in the sampled outer-core volumes. Waveform modeling favors anomalies comparable in size to the SKS Fresnel zone, avoiding strong diffraction-induced amplitude reductions that would be expected from smaller, stronger anomalies. Mechanistically, a temperature decrease alone is unlikely: sound speed in liquid iron is nearly temperature-insensitive at core pressures, and producing a 1–1.5% VP increase would require an implausibly large local cooling (~1000 K). Instead, a compositional explanation—transient, localized increases in light-element concentration—better fits the observations, implying a ~2–3% density deficit. This interpretation aligns with thermochemical geodynamo simulations wherein cyclonic circulations and compositional plumes can generate heterogeneous light-element release and transport. The anomalous paths, all beneath the low-latitude Pacific and in the upper half of the outer core, hint at regionally focused dynamics. While theoretical limits suggest sustained lateral density contrasts in the outer core should be very small, the observed anomalies can be transient. Alternative or contributing factors may include regional thermal convection influenced by lowermost-mantle heterogeneity, or descending flows containing suspended solids that alter bulk modulus and wavespeed without large net density contrasts. A thin stably stratified layer at the top of the outer core, if responsible, would require very large VP perturbations (3–15%) to produce the observed timing in a 60–300 km layer, which seems less likely given the rapid temporal change and expected stability constraints. Overall, the findings demonstrate that the outer core is not well mixed on decadal timescales and that seismic monitoring can detect its transient heterogeneities.

This study demonstrates decadal temporal changes in seismic wavespeed within localized regions of Earth’s outer core, detected via double-differential timing of SKS relative to reference phases from closely co-located earthquake pairs. Five anomalous measurements reveal ~0.8–1.5 s earlier SKS arrivals for later events, requiring a localized VP increase of ~1–1.5% within volumes of lateral radius ~800 km and depth extent ~800 km in the upper outer core beneath the low-latitude Pacific. If compositional, these imply a ~2–3% density deficit due to elevated light-element content, consistent with transient upward flows with characteristic speeds on the order of tens of km per year. The work introduces a seismic approach to monitor temporal variability in the outer core and suggests that outer-core convection can produce detectable, localized, short-timescale heterogeneities. Future research should expand the global dataset of repeating or closely located earthquake pairs, refine temporal sampling to constrain onset/duration of changes, integrate global core-flow and thermochemical simulations, and probe potential connections to a putative stratified layer and to lowermost-mantle heterogeneity.

- Anomalous signals are few (five measurement sets) and geographically localized; most measurements show no required change, limiting generalizability. - Temporal resolution is constrained by the interval between paired earthquakes (often ~20 years), leaving ambiguity over whether changes were gradual or rapid. - Although extensive hypocenter grid searches and 3D mantle simulations were performed, residual uncertainties in source parameters or unmodeled path effects cannot be entirely excluded. - Inference of anomaly size/strength involves trade-offs based on finite-frequency modeling; estimates depend on assumed periods and model parameterizations. - The study primarily samples the low-latitude Pacific in the upper half of the outer core; coverage is sparse elsewhere, preventing robust global conclusions. - Mechanistic interpretation (composition vs. temperature vs. other processes) remains indirect without independent constraints on composition or thermal state.