Orbital magnetic data show that the lunar crust has been strongly magnetized by intense magnetic fields. Rock samples from the Apollo missions indicate that a global magnetic field (50–100 μT), generated by a core dynamo mechanism, existed in ancient times (4.2–2.5 billion years ago) (Garrick-Bethell et al. 2009; Le Bars et al. 2011; Shea et al. 2012; Weiss & Tikoo 2014; Buffett 2016, Mighani et al. 2020). However, the magnetism inherent in endogenous lunar materials is too weak to have recorded the core field. The prevailing theory suggests that extralunar impact meteorites with high thermoremanent susceptibility are the primary source of the strong remanent magnetization anomalies detected in the lunar crust (Wieczorek et al., 2012, 2023; Wakita et al. 2021; Citron et al., 2024).
Nevertheless, the origins of some lunar magnetic anomalies remain elusive. For instance, isolated strong anomalies in the maria, which are not associated with any known impact events, correlate closely with lunar swirl optical anomalies. It has been hypothesized that lunar swirls may form from the crust’s magnetic field due to uneven space weathering from magnetic deflection of incoming solar wind (Hood & Schubert, 1980; Garrick-Bethell et al., 2011) or through hybrid and kinetic plasma interactions (Zimmerman et al., 2015). However, the fine and complex morphology of these swirls challenges our current understanding of their origins. Recent theories propose shallow, narrow iron-rich magmatic dikes or lava tubes as potential sources of the swirls (Hemingway & Tikoo, 2018). Yet, the magnetism of endogenous lunar materials is inadequate to produce anomalies as strong as those observed at orbital altitudes. Lunar Prospector, GRAIL and other orbiters have comprehensively measured the Moon's global magnetic vector field and gravity field, significantly enhancing the development of related models. These datasets are invaluable, containing critical information about the origins of lunar crust magnetization. In this study, we have reanalyzed these datasets from diverse research perspectives using advanced geophysical techniques, as detailed in recent studies (Khatereh et al., 2022; Zuo et al., 2021, 2024).
Lunar crust magnetic anomaly vector data, captured at an altitude of 30 km and derived from a magnetic field model (Wieczorek, 2018), are integrated into a 3D lunar model using a lunar-centric global Cartesian coordinate system (Fig. 1a & 1b), rather than traditional local 2D map projections. This innovative 3D vector model offers a distinctive perspective on lunar magnetic anomalies, illustrating the direction and intensity of crust magnetization anomalies through an intuitive visualization. To address the challenge of displaying only a limited portion of the data effectively, we create Movie S1—an animation that depicts a 360° rotation of the 3D lunar model, ensuring comprehensive visualization of the data.
New features of lunar magnetization anomalies are elucidated by the 3D model. It is now determined that the crustal anomalies are not magnetized by an ancient core field; otherwise, a global magnetic field pattern akin to Earth's geomagnetic field would be evident (Extended Data Fig.1). The 3D model reveals that lunar magnetic anomalies comprise numerous regional clusters (Fig. 1a & 1b), each exhibiting characteristics of a magnetic dipole field. For instance, central anomalies are observed in the Smythii and Humboldtianum basins, with intense anomalies located in highland regions (clusters A & B) and maria (Fig. 1a). Additionally, these intense clusters are concentrated near the edge of the South Pole-Aitken basin on the far side (Fig. 1b). These dipole-like clusters typically align upward or downward, perpendicular to the lunar surface, irrespective of latitude. Only a few anomaly clusters exhibit different directions, likely influenced by subsequent impact events. This pattern indicates that the lunar crust magnetic anomalies are likely magnetized by local magnetic fields, not by the ancient core field.
Secondly, the nature of the dipole-like anomaly clusters suggests that the magnetizing fields are not transient phenomena generated solely during impacts. While basins such as Humboldtianum and Smythii exhibit pronounced anomalies at their centers (Fig. 2a), most lunar impact basins do not feature a regular central anomaly (Fig. 1a & 1b). Numerous intense anomaly clusters are located in the highlands or maria, rather than within the basins themselves—such as clusters A and B in Fig. 1a, and the anomalies along the edge of the South Pole-Aitken basin (e.g., Ingenii Mare in Fig. 2b). If transient fields from impacts were the primary local magnetizing force, the magnetization anomalies recorded would display a closer relationship with basin structures. However, this is not observed in the distributions of anomalies.
To validate our findings, we invert the subsurface magnetization models of the Mendel-Rydberg basin, characterized by a regular magnetic anomaly at its center (Fig. 2a), and Ingenii Mare, marked by irregular anomaly clusters (Fig. 2b). This analysis aims to explore the signatures of magnetizing fields, reconstructed by the inverted magnetization models (Zuo et al., 2024), as shown in Figs. 3a and 3b. Clear animated versions are available in the supporting materials, Movies S2 and S3.
As depicted in Fig. 3a, the crust of the Mendel-Rydberg basin is influenced by an isolated, upward dipole-like magnetizing field. The potential source position of this field, indicated by a green arrow, is located in the shallow crust. The 3D magnetization model for Ingenii Mare reveals several weak dipole-like fields at the basin edges, marked with green arrows, and a central, intense toroidal electrical current source generating a strong magnetizing field. It is well-established that a magnetic dipole is analogous to a toroidal current with zero radius, and broad, intense dipole-like fields are typically produced by electrical current loops. This toroidal current induces a poloidal magnetic field, with lines circulating around the current source as illustrated in Extended Data Figs. 2 and 3. We invert the magnetization model for Ingenii Mare and reconstruct the map of ancient electrical currents (Fig. 4a) using the equivalent electrical current method (detailed in supporting online material Text S1).
The morphology of the electrical currents depicted in Fig. 4a closely mirrors the lunar swirls observed in Ingenii Mare, as shown in Fig. 4b. Notably, the sinuous S-shaped current traversing the rim of the Thompson basin and the linear current in the Van de Graaff basin are vividly illustrated. The 3D magnetic field proximal to these swirls is constructed using the inverted magnetization model, as displayed in Fig. 4c, with a cross-section of this field overlaid with the S-shaped swirl in Fig. 4d. Application of Ampère’s law confirms that the morphology of the swirls aligns with the distribution of electrical currents, which in turn generate the surrounding poloidal magnetic field, as detailed in Figs. 4c and 4d. While these magnetizing fields are estimated using the magnetization model, it is important to note that only the distribution of the ancient magnetizing fields is recovered. The intensity of these fields will be addressed separately in a subsequent section.
In this study, we hypothesize that lunar swirls are remnants of electrical currents that flowed through the lunar crust during periods of ancient dynamo activity. These currents attracted plagioclase-rich dust via electric fields, leading to the formation of high-albedo optical swirls. Furthermore, these electrical currents generated intense magnetic fields that subsequently magnetized the lunar crust, resulting in the magnetization anomalies currently observed by orbiters. A detailed description of the magnetizing procedures is provided in the supporting online material (Text S2).
We argue that electrical currents, rather than magnetic field shielding or intrusive magmatic dikes, are more likely responsible for forming lunar swirls. Firstly, electrical currents can generate intricate structures when flowing through regions of high electrical conductivity, such as intrusive tubes or dikes. Although our inverted electrical current models (Fig. 4a) do not exhibit the same fine structure as the swirls (Fig. 4b), this discrepancy is likely due to limitations in the resolution of orbital data. In contrast, the 3D magnetic field, which spreads over a wide area (Fig. 4c), is unlikely to create such detailed structures through shielding effects alone. Furthermore, analogous high-albedo branch anomalies, formed by ejected plagioclase-rich dust from meteoroid impacts (e.g., in the Oceanus Procellarum region, Extended Data Fig.4), are well-preserved without the influence of magnetic field shielding.
The most direct evidence for determining the origin of the lunar swirls—whether from electric currents or magnetic fields—lies in their alignment with the corresponding field structures. If the boundaries of the swirls align with the vertical edges of the magnetic field, it would suggest that magnetic shielding is responsible for their formation. Conversely, if the swirl boundaries are situated at the center of the 3D magnetic field (indicative of potential poloidal current locations), it would imply that the swirls are formed by the accretion of plagioclase-rich dust due to poloidal electric currents. Analysis of the 3D magnetic field slice (Fig. 4d) reveals that the swirls are located at the positions of potential poloidal electric currents, rather than at the vertical boundaries of the magnetic field. This indicates that the swirls are formed by poloidal electric current accretion rather than magnetic shielding.
Additionally, the formation of swirls due to magma intruding and flowing on the lunar surface is also unlikely. For instance, the fine striped shape of the swirls in the Van de Graaff basin (Fig. 4b) does not match the expected patterns from surface-flowing magma. If the swirls were formed by such magma, the magnetizing field would have to be either the core field or a field generated by conductive fluid flow within the core field, following magnetohydrodynamic (MHD) processes. However, neither scenario aligns with the observed magnetic data in terms of field direction and intensity.
Rein-Gamma is another prominent large lunar swirl complex located in the Oceanus Procellarum region. Intense magnetic anomalies have been observed above the narrow, striped, high-albedo swirl region. As illustrated in the magnetic field data slice (Fig. 5a), the strongest magnetic anomalies coincide with the brightest part of the swirl, where the magnetic field direction changes rapidly. The relationship between these swirl complexes and the magnetic anomalies remains unresolved. To investigate this, we invert the subsurface magnetization model and recover the distribution of the ancient electric currents, as shown in Fig. 5b.
The morphology of the electrical currents depicted in Fig. 4a closely mirrors the lunar swirls observed in Ingenii Mare, as shown in Fig. 4b. Notably, the sinuous S-shaped current traversing the rim of the Thompson basin and the linear current in the Van de Graaff basin are vividly illustrated. The 3D magnetic field proximal to these swirls is constructed using the inverted magnetization model, as displayed in Fig. 4c, with a cross-section of this field overlaid with the S-shaped swirl in Fig. 4d. Application of Ampère’s law confirms that the morphology of the swirls aligns with the distribution of electrical currents, which in turn generate the surrounding poloidal magnetic field, as detailed in Figs. 4c and 4d. While these magnetizing fields are estimated using the magnetization model, it is important to note that only the distribution of the ancient magnetizing fields is recovered. The intensity of these fields will be addressed separately in a subsequent section.
In this study, we hypothesize that lunar swirls are remnants of electrical currents that flowed through the lunar crust during periods of ancient dynamo activity. These currents attracted plagioclase-rich dust via electric fields, leading to the formation of high-albedo optical swirls. Furthermore, these electrical currents generated intense magnetic fields that subsequently magnetized the lunar crust, resulting in the magnetization anomalies currently observed by orbiters. A detailed description of the magnetizing procedures is provided in the supporting online material (Text S2).
We argue that electrical currents, rather than magnetic field shielding or intrusive magmatic dikes, are more likely responsible for forming lunar swirls. Firstly, electrical currents can generate intricate structures when flowing through regions of high electrical conductivity, such as intrusive tubes or dikes. Although our inverted electrical current models (Fig. 4a) do not exhibit the same fine structure as the swirls (Fig. 4b), this discrepancy is likely due to limitations in the resolution of orbital data. In contrast, the 3D magnetic field, which spreads over a wide area (Fig. 4c), is unlikely to create such detailed structures through shielding effects alone. Furthermore, analogous high-albedo branch anomalies, formed by ejected plagioclase-rich dust from meteoroid impacts (e.g., in the Oceanus Procellarum region, Extended Data Fig.4), are well-preserved without the influence of magnetic field shielding.
The most direct evidence for determining the origin of the lunar swirls—whether from electric currents or magnetic fields—lies in their alignment with the corresponding field structures. If the boundaries of the swirls align with the vertical edges of the magnetic field, it would suggest that magnetic shielding is responsible for their formation. Conversely, if the swirl boundaries are situated at the center of the 3D magnetic field (indicative of potential poloidal current locations), it would imply that the swirls are formed by the accretion of plagioclase-rich dust due to poloidal electric currents. Analysis of the 3D magnetic field slice (Fig. 4d) reveals that the swirls are located at the positions of potential poloidal electric currents, rather than at the vertical boundaries of the magnetic field. This indicates that the swirls are formed by poloidal electric current accretion rather than magnetic shielding.
Additionally, the formation of swirls due to magma intruding and flowing on the lunar surface is also unlikely. For instance, the fine striped shape of the swirls in the Van de Graaff basin (Fig. 4b) does not match the expected patterns from surface-flowing magma. If the swirls were formed by such magma, the magnetizing field would have to be either the core field or a field generated by conductive fluid flow within the core field, following magnetohydrodynamic (MHD) processes. However, neither scenario aligns with the observed magnetic data in terms of field direction and intensity.
Rein-Gamma is another prominent large lunar swirl complex located in the Oceanus Procellarum region. Intense magnetic anomalies have been observed above the narrow, striped, high-albedo swirl region. As illustrated in the magnetic field data slice (Fig. 5a), the strongest magnetic anomalies coincide with the brightest part of the swirl, where the magnetic field direction changes rapidly. The relationship between these swirl complexes and the magnetic anomalies remains unresolved. To investigate this, we invert the subsurface magnetization model and recover the distribution of the ancient electric currents, as shown in Fig. 5b.
Beneath the expansive S-shaped swirl in Ingenii Mare, a high-density intrusion extends from the deep lunar mantle to the crust, bifurcating into two branches near the lunar surface (Fig. 6a, section SI). Similarly, a substantial high-density rock intrusion is observed beneath the stride swirl in the Van de Graaff basin (Fig. 6a, another section). In the Reiner-Gamma region, two high-density intrusive tubes emerge from the deep mantle beneath the prominent swirls (Fig. 6b, section SR).
The inverted magnetization model for Ingenii Mare indicates that the magnetization strength reaches 0.5 A/m beneath the S-shaped swirls (Fig. 6a). In the Reiner-Gamma region, a magnetization strength of 0.4 A/m is observed beneath the swirls (Fig. 6b). Considering the diffusion effect inherent in the inversion procedure, the actual magnetization strength may be higher than these values. For the 0.4 A/m magnetization in Reiner-Gamma, the corresponding magnetizing field strength should be approximately 318 μT (thermoremanence susceptibility is set to 1.58 × 10⁻³ SI as mare basalts, Wieczorek et al., 2012). This value significantly exceeds the strength of the core field (40μT - 100 μT).
Here, we propose that intense ancient electrical currents originated from the deep interior of the Moon. Although the movement of lava flows within the ancient lunar magnetic field can generate electrical currents, their density is insufficient. For instance, consider an upwelling lava flow beneath the crust in the Reiner-Gamma region, characterized by high conductivity of 100 S/m and a velocity of 10 m/s. The induced current, as it traverses a magnetic field of 100 μT, would only produce a magnetic field of 3 μT, significantly lower than the observed magnetizing field strength of 318 μT (detailed in the supporting online material, Text S3). Furthermore, even if drastic changes in the lunar core's magnetic field significantly increased the velocity parameters, leading to stronger electrical currents, such a scenario would still not align with the observed patterns of intrusive rocks (Fig. 6a & 6b). The most plausible source of these currents is the powerful electrical activity generated by ancient core dynamo processes. Under the thermal influence of the core, high-density, high-conductivity mafic lava ascends to the crust-mantle boundary. The intense currents produced by lunar dynamo activity are partially channeled through these pathways to the lunar crust. Once at the surface, these currents flow within the crust, forming the swirls and generating the observed magnetic fields.
A 10 km horizontal electrical current with a density of 13 A/m², confined within a cross-sectional area of 3 km by 1 km, is introduced at the crust-mantle boundary at a depth of 10 km and flows horizontally along the crust (Extended Data Fig.7). In this region, the crust's conductivity is set according to the inverted density model, with values of 0.1 S/m for densities greater than 2500 kg/m³ and 0.01 S/m for densities smaller than 2500 kg/m³. The electrical current diffuses upward from the base of the crust to the surface. Fig. 7a shows that the distribution of electrical current on the crust's surface exhibits an elliptical shape, resembling the morphology of Swirl A in Reiner-Gamma (Fig. 5a). This electrical current generates the magnetizing field, which in turn magnetizes the rocks. The magnitude of it are approximately three orders of magnitude higher than the orbital data, recording 22.7 μT at orbital altitude and 469 μT on the lunar surface (Extended Data Fig.8, Movie S8). The rocks have a thermoremanent susceptibility set at 1.58 × 10⁻³ SI (mare basalt). The resulting magnetic field anomaly, simulated based on the magnetization model of these rocks, is presented in Fig. 7b.
The strength and distribution of the simulated rock remanent magnetization anomaly at an altitude of 30 km (Fig. 7b) is 11.7 nT, closely aligning with the actual orbital observation data (11 nT, Fig. 5a). At the lunar surface, the magnetic anomaly reaches 373 nT. This simulation effectively demonstrates that deep interior electrical current sources can generate electric and magnetic fields that closely match the distribution of lunar swirls and orbital magnetic anomalies.
In addition to anomalies in the mare regions, our analysis encompasses three types of mascon basins. Notably, in the Mendel-Rydberg basin, an isolated anomalous cluster (Fig. 2a) is closely linked with the basin's structure. The density inversion model suggests that this density distribution stems from an impact event. It reveals a bowl-shaped mascon with a 170 km radius, encircled by a negative density ring structure with a 340 km radius, exhibiting a density contrast ranging from -316 to -80 kg/m³ (Fig. 8a). The model slice illustrates high-density materials uplifting at the shallow rim, potentially forming a conduit for electrical current conduction that connects to the deep lunar interior (Fig. 8b). The inversion of ancient electrical currents and the associated intense orbital magnetic anomalies, positioned above the rim of the mascon (Fig. 8c & 8d), corroborate this interpretation.
The strength and distribution of the simulated rock remanent magnetization anomaly at an altitude of 30 km (Fig. 7b) is 11.7 nT, closely aligning with the actual orbital observation data (11 nT, Fig. 5a). At the lunar surface, the magnetic anomaly reaches 373 nT. This simulation effectively demonstrates that deep interior electrical current sources can generate electric and magnetic fields that closely match the distribution of lunar swirls and orbital magnetic anomalies.
In addition to anomalies in the mare regions, our analysis encompasses three types of mascon basins. Notably, in the Mendel-Rydberg basin, an isolated anomalous cluster (Fig. 2a) is closely linked with the basin's structure. The density inversion model suggests that this density distribution stems from an impact event. It reveals a bowl-shaped mascon with a 170 km radius, encircled by a negative density ring structure with a 340 km radius, exhibiting a density contrast ranging from -316 to -80 kg/m³ (Fig. 8a). The model slice illustrates high-density materials uplifting at the shallow rim, potentially forming a conduit for electrical current conduction that connects to the deep lunar interior (Fig. 8b). The inversion of ancient electrical currents and the associated intense orbital magnetic anomalies, positioned above the rim of the mascon (Fig. 8c & 8d), corroborate this interpretation.
Although large impact basins with mascons, such as Crisium, typically do not exhibit significant swirls, intense electrical currents have been identified in the Crisium basin (Fig. 9c). To further investigate, we enhanced the optical image of the basin’s central region (Extended Data Fig.9), which revealed weak swirl anomalies corresponding to the locations of the inverted currents. This finding supports the presence of weak electrical currents at the surface, conducted from the shallow rim of the mascon.
As depicted in Fig. 9b, the magnetized rock layer in this region reaches a thickness of 30 km. These basalts required at least 28.5 million years to cool from their molten state to their Curie point (600°C). The magnetizing field must have formed either subsequently or been maintained throughout this period, suggesting that the source of these electrical currents is endogenous, originating from deep within the Moon’s interior, rather than being a direct result of an impact-generated ambient field.
The Leibnitz basin, a unique lunar crater, is distinguished not only by its magnetic anomalies and a central mascon but also by the presence of swirls at its center (Fig. 10a), making it an exemplary case for exploration. We have mapped the electrical currents (Fig. 10b), finding that their distribution closely aligns with the swirl patterns observed in the basin. The source of magnetization in the Leibnitz basin is located at a depth within 10 km, akin to that in the Mendel-Rydberg basin (Fig. 10c). However, the mascon here is shallower compared to those in other basins (Figs. 8b & 9b), with its upper part extending very close to the lunar surface (<5 km). The apex of the mascon is not directly at the center of the current loop but is positioned just below the eastern part of the central current loop. This configuration correlates with the pattern of strong upward currents on the lunar crust surface. Unlike the deeper mascon in the Crisium basins, the high-conductivity mascon in the Leibnitz basin is comparatively shallow, facilitating the movement of powerful currents from the deeper lunar interior to the lunar crust. This process leads to the accretion of plagioclase-rich dust and the formation of distinctive lunar swirl anomalies on the surface. These dynamics suggest that the shallow mascon not only supports but enhances the electrical activity that shapes these surface features.
The depth information of the magnetization field provides crucial insights into the origin of the magnetizing forces affecting the lunar crust, revealing whether they stem from extralunar sources such as solar wind, plasma, or from endogenous sources within the Moon itself. To further investigate, we constructed 3D magnetic field models using inverted magnetization data from the Mendel-Rydberg basin, Crisium basin, and Ingenii Mare.
In a typical vertical magnetic dipole field, a reversal in direction is typically observed at the depth of the source (Fig. S2). This reversal feature, identified at a depth of 10 km within the Mendel-Rydberg basin, accurately locates the source of the magnetizing field within the crust (Fig. 3a). Similarly, a reversal is observed in the Crisium basin model (Fig. 11a) at a depth within 15 km. In Ingenii Mare (Fig. 11b), the presence of both dipole and loop sources is evident, with all reversal characteristics indicating that the source is located within about 10 km of the crust. These cases collectively suggest that the ancient magnetizing fields are generated by endogenous sources within the lunar crust, rather than from extralunar influences.
The magnetizing field generated by a current exhibits a characteristic non-uniform distribution, with field strength diminishing rapidly as distance from the source increases (Extended Data Fig.10). For instance, in the synthetic example, the field strength peaks at 2 mT near the current source. By utilizing the inverted magnetization models of the discussed regions and referencing the estimated range of thermoremanent susceptibility for lunar rocks (Wieczorek, 2012) along with Equation S7, we estimate the maximum possible strength of the magnetizing fields in these study areas, as detailed in Fig. 12.
The intensity of the paleomagnetic field in regions with lunar swirls is significantly greater than in regions devoid of such features. For instance, as depicted in Fig. 12, in areas primarily composed of mare basalts like the Humboldtianum basin, the magnetizing field strength is approximately 100 μT. In contrast, this strength escalates to 400 μT in the Reiner-Gamma and Ingenii Mare regions. If these regions predominantly contain basaltic achondrites, the ancient magnetizing field could reach up to 1000 μT.
The significant strengths of magnetizing fields suggest that magnetized materials may include not only extralunar projectile materials with high thermoremanent susceptibility but also indigenous lunar rocks, such as mare basalts, pristine highland rocks, and other native lunar formations. The presence of swirls highlights regions where ancient, intense electrical currents once traversed the lunar crust, unlike areas without swirls, which lack the strong surface electrical currents necessary to generate such optical anomalies. This distinction elucidates the consistent association of strong magnetic anomalies with lunar swirls and underscores the presence of isolated intense magnetic anomalies.
The uniqueness of the estimated 3D magnetic, gravity, and electric fields in this study warrants discussion to determine whether the conclusions drawn are definitive or merely speculative. All inversions involved in this research achieved satisfactory convergence (Extended data Table 2). Firstly, the distributions of the 3D magnetic fields, constructed from orbital data, are considered unique according to classical potential field theory, despite minor convergence errors. Secondly, the derivation of equivalent electrical currents based on the magnetization model is unique. Thirdly, although the inversion of the magnetization model might face non-uniqueness due to depth uncertainties, current geophysical 3D inversion models employing depth function constraints effectively capture depth information. Finally, cross-referencing and validating results through multi-physics inversions emerge as a viable method to enhance the reliability of the conclusions.
Due to the absence of a dense atmosphere and other active layers on the lunar surface, swirls have been well recorded and preserved. In contrast, Earth's surface is significantly influenced by its atmosphere, water, and biosphere layers. If electrical currents were to conduct to Earth's surface, it is unlikely to produce and preserve such swirl anomaly. Currently, electrical discharges have been observed during volcanic eruptions, which may be contributed by this mechanism. Inspired by the phenomenon of lunar swirls, it can be postulated that celestial bodies with core dynamo activity may exhibit surface discharge phenomena. Our study may aid the further exploration and observation of Earth and extraterrestrial bodies.