Unraveling the lithospheric magnetic field masked by the Earth’s main field by estimating the magnetization and magnetic crustal thickness using a statistical approach

Detailed mapping of Earth’s lithospheric magnetic field provides important insights into the composition, dynamics, and geological history of the crust. This field can be modeled using satellite and near-surface magnetic measurements. However, structures larger than approximately 2500 km in scale are obscured by the dominant magnetic signal generated by the Earth’s core. The superposition of core and crustal magnetic fields introduces ambiguities in both geodynamo and crustal magnetic source studies. Previous efforts to address this issue have included statistical estimates of upper and lower bounds on the long-wavelength components of the crustal field, as well as more deterministic predictions based on geophysical priors such as crustal magnetization and seismic Moho depth models. These approaches, however, have often produced contradictory results. In this study, we adopt a two-step strategy. The first step involves a series of regional spherical spectral analysis of the World Digital Magnetic Anomaly Grid (WDMAM2.2), without relying on any prior information from seismic or magnetization models. This approach, applied to the 5 km ×5 km WDMAM2.2 grid across 6000 regions uniformly distributed over the Earth’s surface, allows us to estimate the probability distributions of three key parameters statistically characterizing crustal magnetization in each of the 6000 regions: magnetization amplitude, magnetic layer thickness, and a power-law exponent. The resulting world map of magnetic layer thickness differs from existing Moho depth models but indicates that, statistically, there is no significant evidence of magnetic sources located below the Moho at the studied length scales. In the second step, the ensemble of regional magnetization models is used to generate a set of large-scale spherical harmonic models of the lithospheric magnetic field (degrees 1 to 50). This set allows us to quantify the extent to which the lithospheric field contaminates both the static and time-varying components of the core magnetic field. We find that this contamination is substantial between spherical harmonic degrees 12 and 15 for the static core field, and from degree 21 onward for the secular variation. • We discuss the formalism relating magnetization, crustal thickness, and lithospheric magnetic field in Spherical Harmonics (SH). • We demonstrate the inability of current seismic and magnetization a priori models to predict with enough confidence the lithospheric magnetic field. • We estimate by a spectral approach applied on a global magnetic anomaly map the large-scale magnetization and magnetic crustal thickness, and their uncertainties, to a maximum spatial resolution equivalent to SH degree 50. • We use an ensemble of predictions and find one lithospheric field and its time variations from SH degrees 1 to 50 best fitting the observations. • We see at the core mantle boundary footprints in core field models attributable to lithospheric field signals between SH degrees 12 and 15.