We take advantage of geological constraints from Ocean Drilling Program drill holes and high-resolution bathymetry to revisit the near-seafloor magnetic anomaly at the Trans-Atlantic Geotraverse hydrothermal site (Mid-Atlantic Ridge, 26°N). The dipolar anomaly associated with the site is better reduced to the pole if the magnetization vector is tilted by 34°, an observation suggesting that the strongly faulted basalt surrounding the site has been rotated by ∼53° along an axis parallel to the Mid-Atlantic Ridge as a probable consequence of the detachment tectonics inferred in this area. The faults and the deeper detachment focus and guide the hot ascending hydrothermal fluid. Magnetic modeling shows that the nonmagnetic stockwork zone is a significant contributor to the observed negative anomaly, the rest being accounted for by a deeper source probably related to thermal demagnetization of an ascending hydrothermal pipe beneath the active part of the site.


Basalt-hosted hydrothermal sites prevailing over oceanic ridges are associated with reduced, and/or a lack of, magnetization, unveiled by the available deep-sea magnetic surveys (Tivey et al., 1993; Tivey and Johnson, 2002; Tivey and Dyment, 2010; Zhu et al., 2010; Caratori Tontini et al., 2012; Nakamura et al., 2013; Honsho et al., 2013; Szitkar et al., 2014a). This observation is interpreted as a consequence of hydrothermal fluid circulation inducing the thermal demagnetization of basalt titanomagnetite and its alteration to less magnetic minerals such as titanomaghemite (Tivey et al., 1993; Tivey and Johnson, 2002). However, a third possible cause has been neglected, i.e., the presence of thick, nonmagnetic sulfides. Standing and fallen hydrothermal chimneys and the surrounding hydrothermal sediments, collectively named the hydrothermal deposit, accumulate on the seafloor, whereas the underlying stockwork zone displays sulfide veins precipitated within a matrix of altered host basalt. Both the hydrothermal deposit and the stockwork zone exhibit a weak magnetization due to the very low content of magnetic minerals in the sulfides and the strong alteration of the host basalt (e.g., Zhao et al., 1998). Such a nonmagnetic layer increases the distance between the underlying magnetized basalt and the magnetometer, producing a negative reduced-to-the-pole magnetic anomaly (Szitkar et al., 2014a).

The Trans-Atlantic Geotraverse (TAG) site at 26°08′N (Mid-Atlantic Ridge [MAR]) is the first hydrothermal site discovered on a slow spreading center (Rona et al., 1986). The first magnetic study of an active basalt-hosted hydrothermal site was conducted there (Tivey et al., 1993), and it is one of the only three hydrothermal sites drilled by the Ocean Drilling Program (ODP) or the Integrated Ocean Drilling Program. We combine information provided by these comprehensive data sets to constrain the tectonic history and magnetic structure of the site. We show that a large (53°) tectonic rotation affected the area, confirming the presence of a core complex. The nonmagnetic hydrothermal material accounts for one-third of the observed negative magnetic anomaly, the remaining part requiring a deeper demagnetized source.


Active basalt-hosted hydrothermal site TAG (Rona et al., 1986) is located at an average depth of 3650 m on the eastern side of the axial valley, on ∼100 k.y. old crust (Tivey et al., 1993), and has been sporadically active for the past 50 k.y. (Lalou et al., 1990). High-resolution bathymetric data (Pontbriand and Sohn, 2014; Fig. 1A) reveal a donut-shaped, 50-m-high hydrothermal deposit made of two superimposed mounds. The largest mound is ∼200 m in diameter, whereas the second mound, located on top of the first and offset to the north, is only half its diameter and corresponds to the currently active hydrothermal area marked by numerous black smokers.

Data from ODP Leg 158 drill holes were used to build a lithological northwest-southeast–oriented section (Humphris et al., 1996; Herzig et al., 1998; Fig. 1B). The hydrothermal deposit, made of massive pyrite and pyrite breccias overlying pyrite and anhydrite and pyrite-silica breccias, ranges in thickness from 0 m at the edges of the mound to 60 m beneath the black smoker area. The stockwork zone, made of sulfide veins within altered basalt, is restricted to an 80-m-diameter central zone beneath the mound. Rock magnetic properties measured on the drilled samples reveal that both the sulfide deposit and the stockwork zone bear a very weak magnetization (several mA/m; Zhao et al., 1998) that is negligible with respect to that of basalt.

High-resolution magnetic data were collected at varying altitudes over the TAG hydrothermal mound using a three-component fluxgate magnetometer fixed to the frame of the deep-sea submersible Alvin (Tivey et al., 1993). These authors quantified and removed the induced magnetic field of the submersible using calibration loops performed during descent and ascent, continued the magnetic anomaly upward to a datum plane at a depth of 3600 m (∼20 m above the highest point of the mound), and inverted the data to equivalent magnetization (see Parker and Huestis, 1974). For the first time, Tivey et al. (1993) unveiled the lack of magnetization characterizing hydrothermal site TAG, and more generally active and inactive basalt-hosted hydrothermal sites.


We digitized the upward-continued magnetic anomaly contours of Tivey et al. (1993) and built the magnetic anomaly grid used in our study (Fig. 2B). A shift of 27 m to the south and 14 m to the west, determined by adjustment of the topography, was applied to this grid for consistency with the more recent well-navigated high-resolution bathymetric data (Pontbriand and Sohn, 2014). Reduction to the pole using the local inclination and declination predicted by the International Geomagnetic Reference Field (Finlay et al., 2010) (Fig. 2C) presumably moves the anomalies above their causative sources. The hydrothermal area is characterized by a negative reduced-to-the-pole (RTP) anomaly, i.e., a lack of magnetization. However, we note that the RTP appears incomplete, i.e., the remaining anomaly retains some dipolar character: the magnetic low is offset 60 m south with respect to the hydrothermal mound and a strong magnetic high, also present in the equivalent magnetization map of Tivey et al. (1993), is observed north of the mound. We determined the orientation of the magnetization vector that best attenuates this dipolar effect and makes the negative magnetic anomaly as symmetrical as possible, and obtained an inclination of 10° and a declination of 0° (Fig. 2D). The limited extension of the magnetic survey prevents an accurate determination of these parameters, which have estimated uncertainties of ∼10°. This determination is consistent with the ∼20° average natural remanent magnetization inclination determined by Zhao et al. (1998) on basalt samples.

Due to its location within the axial magnetic anomaly (Rona, 1980), the basaltic crust in the TAG area should display a magnetization vector trending north and a geocentric axial dipole inclination of 44°. Any significant departure from this orientation may only result from tectonic motion having affected this crust after its formation. The general strike of the axial valley and faults in the TAG area is N30°E (e.g., Rona, 1980). Assuming a rotation around a N30°E trending horizontal axis and a projected tilt of 34° (the difference between the observed and theoretical inclinations) in the meridian plane leads to a 53° tilt around this axis.

The TAG hydrothermal site is located in a “fault and fissure zone”, or FFZ (Bohnenstiehl and Kleinrock, 1999), a strongly faulted area on the hanging wall of an inferred detachment fault (Tivey et al., 2003; Canales et al., 2007; deMartin et al., 2007). Although important footwall rotations have been proven by paleomagnetic data from such detachment faults (Garcés and Gee, 2007; Morris et al., 2009; MacLeod et al., 2011), little is known about possible rotations encountered by the hanging wall. The FFZ is the part of the hanging wall located just above the area where the footwall rotation occurs, suggesting a causative relationship. It structurally corresponds to the slipped block observed by Cann et al. (1997) on the Atlantis Massif oceanic core complex, that later evolve as rider blocks (Choi and Buck, 2012). Whether they correspond to the abandonment of the detachment fault and the initiation of a new one, or to a local relocation of the existing detachment fault, these blocks are slices of the hanging wall transferred to and rafted with the footwall (Reston and Ranero, 2011). The FFZ could therefore be such a sliver of hanging wall dissected by faults, strongly extended, and consequently rotated as a result of the footwall tectonics (Fig. 3).

Resolving the best inclination of the remanent magnetization vector of basalt surrounding the TAG hydrothermal site reveals important tectonic rotations in relation to the underlying oceanic core complex. Both the deeper part of the detachment fault and the faults associated with these rotations focus the circulation of hydrothermal fluids from the vicinity of the melt zone to the TAG discharge area, as suggested by deMartin et al. (2007). The observed magnetic anomaly does not provide information on the relative timing of the tectonic rotation and the emplacement of the hydrothermal site, because the (rotated) magnetization vector is carried by basalt and hydrothermalism does not modify its direction, only its intensity. The hydrothermal mound is not faulted, suggesting that hydrothermalism continued after the last episode of tectonic rotation. The presence of fossil hydrothermal sites in other parts of the FFZ (e.g., Tivey et al., 1996) suggests that different faults may successively focus the hydrothermal circulation.


We used the new RTP magnetic anomaly (Fig. 2D), together with the high-resolution bathymetry of Pontbriand and Sohn (2014; Fig. 2A) and the depth of the hydrothermal deposit and stockwork zone as defined by the ODP drill holes (Humphris et al., 1996; Herzig et al., 1998; Fig. 1B), to quantify the effect of the nonmagnetic hydrothermal deposit and stockwork zone by a forward modeling approach (e.g., Szitkar et al., 2014a). Modeling assumptions are kept simple, aiming to test physical processes and not to match exactly the anomalies. We assume a 500-m-thick, 14 A/m magnetized basaltic layer, a value consistent with that obtained from surface and near-seafloor magnetics by Honsho et al. (2009) in a nearby section of the MAR. No variation of the magnetization intensity, either vertical or lateral, is considered within the basaltic layer. For the area surrounding the hydrothermal site, the top of this layer is the high-resolution bathymetry (or, outside the survey area, a smooth extrapolation designed to prevent edge effects). For the hydrothermal site, the base of the magnetized layer mimics the interface between the sulfide deposit and the stockwork zone (as constrained by drilling), 500 m deeper, whereas its top is defined by the base of the stockwork. As a result, the magnetized layer thickness decreases to 320 m beneath the site. Anomalies are computed at the pole for consistency (Fig. 2D).

The resulting magnetic anomaly (Fig. 2E) has a peak to peak amplitude of ∼1500 nT, compared to the ∼2300 nT of the observed RTP anomaly. Because the hydrothermal deposit has a roughly horizontal base at a depth similar to that of the surrounding seafloor (Fig. 1B), it produces no anomaly on the horizontal observation plane considered, and the computed anomaly (Fig. 2E) mainly reflects the effect of the nonmagnetic stockwork zone. Due to the limited lateral extension of this stockwork zone, the computed anomaly extends over a diameter of ∼175 m, compared to the ∼300 m of the observed anomaly. Moreover, the predicted anomaly is shifted ∼40 m southward with respect to the observed anomaly. The effect of the nonmagnetic stockwork zone contributes about one-third of the observed anomaly at the altitude of data reduction. This effect, certainly stronger in the original magnetic data, acquired 2 m and 20 m above the seafloor, has been reduced by the upward continuation to a horizontal datum plane.

To better account for the observed negative RTP anomaly, we considered additional nonmagnetic sources beneath the hydrothermal site. We note that the minimum of the RTP anomaly (Fig. 2D) is not located above the center of the main TAG hydrothermal mound, but above the center of the smaller, active mound, ∼30 m northward. We therefore added a nonmagnetic vertical cylinder of radius 100 m, located beneath the active hydrothermal mound and extending across the entire magnetized basaltic layer. The resulting magnetic anomaly (Fig. 2F) agrees well in amplitude, size, and location with the observed anomaly (Fig. 2D), although the detailed shape of the latter is not fully accounted for. Beyond the intrinsic nonuniqueness of the potential field solution, the reasonable fit of the modeled and observed anomalies along north-south and east-west cross sections (Fig. 2G) confirms that a northward-shifted deeper source is required to explain the observed anomalies.

This additional demagnetized zone likely corresponds to a hydrothermal pipe, in which two competing, nonexclusive processes may explain the basalt demagnetization: alteration, which permanently degrades strongly magnetic titanomagnetite to less-magnetic titanomaghemite and ultimately to nonmagnetic minerals, or heating above the Curie temperature of titanomagnetite (150–200 °C; Kent and Gee, 1996), i.e., temporary thermal demagnetization. Although it is difficult to discriminate among these processes, the slight northward shift of the observed RTP anomaly beneath the active mound favors thermal demagnetization as the dominant effect. A dominant effect of permanent hydrothermal alteration would more likely result in a magnetic anomaly centered on the main mound.


We revisit the near-seafloor magnetic anomaly for the TAG hydrothermal site (Tivey et al., 1993) taking advantage of more recent geological constraints from ODP Leg 158 drill holes across the hydrothermal mounds and high-resolution bathymetry. The dipolar magnetic anomaly associated with the site is better reduced to the pole assuming an inclination of 10° (instead of 44° expected at 26°N) for the magnetization vector. Such an observation suggests that basalt surrounding the site, which belongs to a strongly faulted and fissured zone, has been rotated by ∼53° along a N30°E horizontal axis (parallel to the MAR axis in this area) as a probable consequence of the detachment tectonics inferred in this area (Tivey et al., 2003; deMartin et al., 2007). The FFZ faults, together with the deeper detachment, focus and guide the hot ascending hydrothermal fluid. Magnetic modeling of the site shows that, although insufficient to explain the entire observed negative anomaly, the hydrothermal material, and more specifically the stockwork zone, is a significant cause of missing magnetization that contributes to approximately one-third of the observed anomaly. The rest of the anomaly is accounted for by a deeper source, probably related to thermal demagnetization of an ascending hydrothermal pipe beneath the active part of the site. The significant contribution of the stockwork zone to the magnetic signature of TAG confirms that it is a common characteristic of all type of hydrothermal sites (Szitkar et al., 2014a, 2014b) of potential interest for deep-sea mineral exploration.

Institut de Physique du Globe de Paris (IPGP), Centre National de la Recherche Scientifique–Institut National des Sciences de l’Univers (CNRS-INSU), and Ifremer are gratefully acknowledged for financial support. We thank R. Sohn for providing the bathymetric data displayed in Figure 2A. We also thank two anonymous reviewers and editor Bob Holdsworth for useful comments. This is IPGP contribution 3585.