Long-term persistency of a strong non-dipole field in the South Atlantic

The Earth's geomagnetic field, generated by a self-sustaining dynamo, shields the planet from solar charged particles that can disrupt satellites, communications, and power systems. Although the field is largely axial-dipole in morphology, prominent non-dipole features exist, notably the South Atlantic Anomaly (SAA), a region of weak intensity and high directional variability. Historical, satellite, archeomagnetic, and paleomagnetic records show that some non-dipole structures persist over time. SAA-like features have been recurrent during the past millennium and episodically over Holocene to multimillion-year intervals, and their core-mantle boundary (CMB) expression involves time-varying reversed flux patches. Whether SAA-like anomalies and associated Atlantic–Pacific hemispheric asymmetries persist on million-year timescales remains debated. Recent work suggests enhanced secular variation in the Southern Hemisphere and in the Atlantic relative to the Pacific. This study aims to test the long-term persistence of anomalous geomagnetic behavior in the South Atlantic over the past few million years by adding new high-quality paleodirectional data from Trindade Island, evaluating PSV and time-averaged field (TAF) structure for 0–10 Ma, and using synthetic models to relate CMB flux patch properties to observed latitudinal PSV patterns.

Prior studies have documented recurrent SAA-like features from archeomagnetism in southern Africa and Brazilian records over the past millennium, and from paleomagnetic reconstructions spanning the Holocene, 28–48 ka, 0–100 ka, and 5.3–23 Ma. Miocene and Late Pleistocene data from Saint Helena and Tristan da Cunha reported low paleointensities and elevated VGP dispersion, indicating long-lived anomalous behavior in the South Atlantic. Hypotheses include top-down control via lower mantle thermal heterogeneity (particularly the African LLSVP) that fosters persistent weak surface field minima and flux patches at the CMB. Dynamo models incorporating mantle heterogeneity reproduce persistent longitude bands of weak field and some hemispheric tendencies, though accurately matching the latitude of the SAA is challenging. On million-year timescales, 0–10 Ma PSV compilations allow evaluation of latitudinal dependence of VGP dispersion with limited plate motion effects, but data are sparse in the Southern Hemisphere (~31% of global 0–10 Ma data), underscoring the need for new records to better resolve hemispheric asymmetries.

Field sampling and paleomagnetic measurements: In 2017, 17 sites were sampled on Trindade Island across four geological units ranging from 3.9 Ma to 59 ka. Eight sites belong to the Trindade Complex (mostly phonolitic necks and a dyke), and nine sites are a'a lava flows from the Morro Vermelho, Valado, and Paredão formations. At least five specimens per site (200 total) were subjected to alternating field (AFD) and thermal demagnetization (THD) in a magnetically shielded lab using standard instruments (ASC-TD48 oven, LDA5 AF demagnetizer, 2G cryogenic magnetometer with RAPID system, JR-6A spinner). AFD involved 18 steps (2–100 mT) and THD 17 steps (room temperature to 600 °C). Characteristic remanent magnetization (ChRM) directions were determined via principal component analysis, selecting linear components decaying to the origin with MAD ≤ 10° and DANG ≤ 5°. Site means used Fisher statistics. Data selection and reference frames: Of 15 sites with site-level directions, two were excluded from PSV/TAF analyses due to failing quality thresholds (k ≥ 50, n ≥ 5), yielding 13 high-quality sites (12 normal, 1 reversed). Plate motion corrections used NNR-MORVEL56 to compute paleolocations and VGPs. A mean direction and paleopole were computed and compared to GAD expectations and geomagnetic polarity time scale for reversal accounting. VGP dispersion and Model G fitting: Between-site VGP dispersion S_b was computed with bootstrap 95% confidence bounds accounting for within-site dispersion, following standard formulations. The latitude dependence of PSV was analyzed using Model G, S(λ) = sqrt(a^2 + (bλ)^2), where a and b represent equatorially symmetric/asymmetric contributions. Nonlinear least squares (Levenberg–Marquardt) fits were performed for the global 0–10 Ma dataset and separately for Atlantic (−90°E to 90°E) and Pacific (90°E to 270°E) hemispheres, with and without inclusion of Trindade. Inclination anomaly and TAF models: Inclination anomaly ΔI = I_mean − I_GAD (with I_GAD = arctan(2 tan λ)) was computed for Trindade and for compiled studies. Two-parameter zonal TAF models were fit to ΔI vs latitude to infer axial quadrupole (G2 = g2^0/g1^0) and octupole (G3 = g3^0/g1^0) contributions relative to the axial dipole (g1^0), using nonlinear least squares. Axial dipole dominance: The degree of dipole dominance was estimated from the Model G a parameter via the empirical relation log(AD/NAD_median) = k1 log a + k2, providing median ratios of axial dipole (AD) to non-axial dipole (NAD) components for each hemisphere. Synthetic CMB flux patch scenarios: Idealized radial field models at the CMB were constructed as an axial dipole background plus equatorially antisymmetric Gaussian flux patches (normal and/or reversed). Scenarios varied patch amplitude ratios (γ), latitudes, and longitudinal separations (Δφ) across five cases (A–E). Fields were upward continued to the surface using kernel analysis to compute synthetic declinations/inclinations, from which latitude-binned VGP dispersion curves (5° bins) were derived and compared against observed PSV curves. These experiments probed how flux patch properties control the shape and hemisphericity of S_g(λ).

- High-quality paleodirections from 13 sites on Trindade Island yield a mean direction D = 3.2°, I = 37.3°, α95 = 11.1°, statistically indistinguishable from the local GAD expectation (D = 0°, I = 37.0°). The paleopole (Plat = 87.3°N, Plon = 100.0°E, α95 = 10.9°) is compatible with the rotation axis. - VGP dispersion at Trindade is elevated (S_g ≈ 20.6°), exceeding values from comparable latitudes outside the SAA longitudinal sector (e.g., Réunion ~11–12°, French Polynesia ~13.6°, Easter Island ~9.2°), and comparable to Saint Helena (~21.3° at ~8–10 Ma). This supports persistent anomalous PSV in the South Atlantic over the past 10 Myr. - Inclination anomaly at Trindade is negligible (ΔI ≈ −0.3°), consistent with a predominantly dipolar TAF at this location over the sampled interval; global zonal TAF fits for 0–10 Ma suggest small non-axial contributions (G2 ≈ 3.0%, G3 ≈ 2.1%). - Hemispheric PSV asymmetry over 0–10 Ma: Model G fits show Atlantic hemisphere with higher symmetric term and slightly lower antisymmetric term (a ≈ 12.5°, b ≈ 0.24) than the Pacific (a ≈ 10.2°, b ≈ 0.25), indicating stronger long-term geomagnetic variability in the Atlantic. Including Trindade increases Atlantic a to ~12.9°. - Estimated axial dipole dominance is lower in the Atlantic (AD/NAD_median ≈ 9.1 ± 6.3) than in the Pacific (≈ 14.5 ± 9.4), implying stronger non-axial dipole effects in the Atlantic sector over the last 10 Myr; inclusion of Trindade further reduces Atlantic dipole dominance. - Zonal TAF models by hemisphere yield statistically similar G2 and G3 within uncertainties: Atlantic G2 ≈ 3.8%, G3 ≈ 1.4%; Pacific G2 ≈ 2.4%, G3 ≈ 2.7%, with regional ΔI patterns suggesting persistent non-zonal features (positive ΔI in parts of the SW Pacific and Antarctica) but sparse coverage. - Synthetic CMB flux patch experiments demonstrate that the observed latitude dependence and hemisphericity of VGP dispersion can be reproduced by persistent, intense low-latitude flux patches (normal or reversed) and high-latitude patches. Among tested scenarios, a configuration with both high- and low-latitude normal patches and amplitude ratio γ ≈ 0.26 (Case D) best matches the 0–10 Ma S_g(λ). Configurations with low-latitude reversed patches (Case E) can also produce peaks in VGP dispersion consistent with recurrent SAA-like behavior. - Collectively, data and models indicate a long-lived non-dipole field in the South Atlantic and an Atlantic–Pacific PSV dichotomy persisting over multimillion-year timescales.

The new Trindade dataset augments sparse Southern Hemisphere records and, together with Saint Helena, indicates sustained high PSV in the South Atlantic during the last 10 Myr. The mean direction and paleopole align with a dominantly dipolar TAF, but elevated VGP dispersion points to strong time-dependent non-dipole contributions. Global and hemispheric Model G fits corroborate enhanced long-term secular variation in the Atlantic relative to the Pacific, reflected in reduced axial dipole dominance in the Atlantic sector. Synthetic modeling shows that the latitude and intensity of CMB flux patches critically shape the S_g(λ) curves. Observed patterns are consistent with recurrent, intense low-latitude flux patches (either normal or reversed) beneath the South Atlantic, which can give rise to SAA-like features episodically across multimillion-year intervals. The persistence of Atlantic–Pacific asymmetry aligns with a dominant bottom-up control on secular variation by inner core processes that localize SV beneath the Atlantic, while lower mantle heterogeneity likely contributes to maintaining long-term high-latitude flux patch structure and longitudinal preferences. Thus, both inner core dynamics and mantle thermal structure influence the long-term morphology and variability of the geomagnetic field, with inner core effects playing a primary role in the temporal hemisphericity observed over 0–10 Ma.

This study presents new high-quality paleodirectional data from Trindade Island and integrates them with 0–10 Ma global PSV/TAF datasets and idealized CMB flux patch models. The results reveal (i) persistent anomalous secular variation in the South Atlantic over multimillion-year timescales, akin to recurrent SAA-like behavior, and (ii) a long-lived Atlantic–Pacific hemispheric asymmetry in PSV with weaker dipole dominance in the Atlantic. Synthetic scenarios demonstrate that the observed PSV latitudinal structure can be produced by persistent low-latitude flux patches and high-latitude patches, implicating a combination of bottom-up inner core control and secondary mantle influences on long-term field morphology. Future work should expand high-quality paleomagnetic sampling in the Southern Hemisphere—especially in the Pacific mid- to high latitudes—and integrate absolute paleointensity data and more realistic geodynamo simulations to refine constraints on the mechanisms sustaining these anomalies.

- Limited number of high-quality sites from Trindade (13) due to challenging access and rugged terrain may influence dispersion estimates, though selection criteria mitigate bias. - Southern Hemisphere paleomagnetic data remain sparse in the 0–10 Ma database, particularly in the Pacific mid- to high latitudes, reducing spatial resolution of TAF/PSV inferences. - Some comparator studies sample short time intervals, potentially biasing dispersion or inclination anomaly estimates. - Inclination anomaly and zonal TAF analyses rely on simplified parameterizations (G2, G3) and may not capture all non-zonal features. - Synthetic flux patch models are idealized and do not uniquely resolve the relative roles of normal vs reversed patches or exact patch geometries. - The study focuses on directional records; limited absolute paleointensity constraints from Trindade restrict direct assessment of long-term field strength variations.