Mars lacks an appreciable global magnetic field at present ( < 0.5 nT equatorial surface field) but must have had one in the distant past [Acuña et al., 2001]. The field measured at mapping orbit altitude is dominated by sources in the Mars crust (remanent magnetism) and external fields arising from the interaction of the solar wind with Mars' atmosphere. This external field is highly variable, ranging from a few nT in magnitude to as much as (rarely) ~100 nT. The magnetic field due to crustal remanence reaches a maximum of ~220 nT at mapping altitude.
The influence of external fields was minimized by using only those observations obtained over the darkened hemisphere and by selection of the median value, rather than the average, for each bin. The 18 mapping cycles of data used span a period of 504 days or 0.73 of a Mars year (687 days), beginning during northern summer and continuing through winter and spring. The seasonal coverage, combined with selection of night time data, resulted in more samples at high northern latitudes relative to high southern latitudes. Accuracy of the crustal field (few nT) is limited primarily by residual external fields.
These maps, acquired at nearly constant altitude, are a more reliable indication of the global distribution of magnetization than those deduced from aerobraking phase observations. Acuña et al. [1999; 2001] appealed to the absence of magnetization near the Hellas and Argyre impact basins to argue for cessation of the dynamo during the early Noachian period, prior to the end of late heavy bombardment. Our new maps strengthen this chronology and remove objections [Schubert et al., 2000] that arise if one uses Purucker's model [Purucker et al., 2000] to infer the distribution of magnetization. This model used 11,550 radial dipoles, uniformly distributed in the crust, to compute the field on a constant altitude surface. Spurious crustal fields may result where dipoles are poorly constrained by (sparse) observations or where large external fields were encountered.
The most prominent crustal sources are the extensive, east-west trending features in Terra Cimmeria and Terra Sirenum and centered on the 180 meridian. These have been attributed to quasi-parallel bands of intensely magnetized crustal rock [Connerney et al., 1999] like those associated with sea floor spreading on Earth. They may have formed by an early era of crustal genesis and spreading in the presence of a reversing dipole [Connerney et al., 1999, Sprenke and Baker, 2000, Nimmo, 2000], fragmentation and separation of a uniformly magnetized crust [Sprenke and Baker, 2000], or by a moving locus of dike intrusion [Nimmo, 2000]. The magnitude and polarity of the mapping orbit observations are consistent with earlier results [Connerney et al., 1999] if one takes into account the upward continuation of the potential field.
It is not possible to uniquely determine the intensity or direction of magnetization in the crust from observations of the vector magnetic field above the source [Blakely, 1995]. Since the magnetic field above a uniformly magnetized crust is zero, it is the variation of magnetization with location that is apparent in maps of this kind. The spatial resolution afforded by these maps is comparable to the altitude of observation, some 400 km. It would be quite remarkable if the Mars crust contained sources of these dimensions with uniform magnetizations. More likely we observe the average magnetization over these dimensions, which are not simply related to in situ magnetizations.
A number of features with predominately circular geometry appear in the radial field map. The simplest features are the relatively isolated, unipolar features (e.g., -35°, 295°E). This feature resembles that of an isolated source with average magnetization within a few tens of degrees of the radial direction; the positive radial flux just above the source is readily apparent, while the negative return flux, distributed over a much larger area, is nearly undetectable.
This feature could be produced by a dipole source with moment of 2.8 x 1016 A-m2 at a depth of 106 km beneath the surface, oriented within 9 degrees of vertical. The inferred depth is most likely an indication that the source is distributed spatially and closer to the surface. Alternative models with very different source characteristics fit the data equally well; without a constraint on the source location and dimensions, the direction of magnetization of such a source is indeterminable.
Some of the circular features (e.g., -45°, 224°E; -11°, 110°E) may be associated with an impact process that creates a central basin surrounded by uplifted crust. The central basin, filled from below by post-impact intrusion and extrusion of iron-rich magma, may acquire thermoremanent magnetization (TRM) if it cools in the presence of an early dynamo-generated field. Subsequent flooding of magma external to the uplifted rim, followed by cooling in a normal or reversed field, can produce a surrounding ring of either polarity. A large volume of magnetic material (moment ~6 x 1015 A-m2) is required to produce a 20 nT feature at 400 km altitude. A 200 km diameter disc, only 3 km thick, of intensely and uniformly magnetized rock (60 A/m) would suffice. The lack of an association between these features and visible craters suggests an extensively reworked surface.
The magnetic minerals responsible for crustal magnetic remanence at Mars are likely to be among those magnetizing the Earth's crust. The magnetic properties of crustal rocks depend critically upon the mineralogical form of the iron, which is diagnostic of the composition and conditions prevalent in the crust at the time of formation or alteration. The intensely magnetized crust must contain iron in a form that can acquire and preserve, over aeons, a large remanent field.
Magnetic minerals with high remanence include forms of titanomagnetite (solid solution of Fe3O4 - Fe2TiO4), pyrrhotite (Fe7S8), and titanohematite (solid solution of Fe2O3 - FeTiO3). Magnetite and pyrrhotite in effective single domain (SD) size can acquire very large TRM upon cooling in an Earth-like inducing field. These minerals have the high coercivity characteristic of the Shergotty, Nakhlite, and Chassigny (SNC) meteorites [Cisowski, 1986] and long term stability. In contrast, homogeneous multidomain-sized (MD) magnetite and pyrrhotite are less efficient in acquisition of TRM and have low coercivity and low long term stability. MD hematite is an attractive candidate because it acquires TRM very efficiently and it is very stable [Kletetschka et al., 2000]. Production of hematite requires high levels of oxidation. It is not found in SNC meteorites but was identified in IR spectra of the surface [Christensen et al., 2000]. Pyrrhotite is a less suitable candidate because it is easily demagnetized by the high pressures [Vaughan and Tossell, 1973] associated with large meteorite impacts. Evidently, only the very largest of impacts, e.g., those associated with the Hellas and Argyre basins, have erased crustal remanence (presumably by heating above the Curie point).
The magnetic mineralogy of the Mars crust may reflect an oxidation state that is more oxidizing than that of the Mars mantle. Studies of Shergottite meteorites [Wadhwa, 2000, McSween et al., 2001] suggest that the different samples in this group derive from magmas that have been modified by assimilation of crustal material with an oxidized (aqueous) component. The sample (QUE94201) most representative of mantle conditions contains ilmenite and ulvospinel, appropriate oxide compositions for crystallization of basaltic material near the iron-wustite buffer. Rocks crystallizing under these relatively reducing conditions would be non-magnetic in the Mars crust. At the other extreme is Shergotty, more oxidized than the quartz-fayalite-magnetite buffer. Intense crustal remanence indicates that large portions of the Mars crust formed from mantle-derived material modified by assimilation of an aqueous component at crustal depths. This may lead to a magnetic boundary at depth in the crust dictated by the transition to an oxidation state appropriate for crystallization of magnetic iron oxides, e.g., titanomagnetite or titanohematite.
We thank MGS project personnel and M. Kaelberer, D. Vignes, P. Lawton and M. Purucker at GSFC. G. Neumann prepared the MOLA topography map. Research at Berkeley was supported by NASA grant NAG5-959 and at Delaware by NAG5-8827.
The Editor would like to thank the reviewer of this manuscript.
J. E. P. Connerney and M. H. Acuña,
NASA Goddard Space Flight Center,
Greenbelt, MD, 20771.
P. J. Wasilewski and G. Kletetschka, Code 691, NASA Goddard Space Flight Center, Greenbelt, MD, 20771.
N. F. Ness, Bartol Research Institute, University of Delaware, Newark, DE, 19716.
H. Rème, Centre d'Etude Spatiale des Rayonnements, 31028 Toulouse Cedex 4, France.
R. P. Lin and D. Mitchell, Space Sciences Laboratory, University of California, Berkeley, CA, 94720.
(Received May 22, 2001; revised July 20, 2001; accepted September 6, 2001)