Anomalous free-air gravity signals in and around the Antarctic continent have been reported for some decades. Recent definition of the Antarctic gravity field from field-based oversnow traverses and supporting data from Earth-orbiting satellites reveal discrete regions of both negative and positive free-air gravity anomalies. The data from these observations have enabled us to construct a free-air gravity anomaly map of Antarctica. Negative free-air gravity anomalies are found to occur mainly on the Antarctic continent, in particular, in the Wilkes Land, Ross Sea, central continental, and Weddell Sea sectors. Positive free-air gravity anomalies are found to occur mainly in the offshore circum-continental sectors. While each of these regions of anomalies provides excellent opportunities for further investigation, including identification of the causes of the negative and positive free-air gravity anomalies, special attention is given to the negative free-air gravity anomaly sites of the continent proper. Three potential sources of the negative free-air gravity anomalies are identified: the mantle, lithosphere, and crust. Examination of thermally induced density variations in the mantle based upon seismic tomography, and analysis of mantle-related gravity anomaly wavelengths favor a gravity anomaly source other than the mantle. Examination of the subcrustal lithosphere based on upper-mantle thermal structure, the origin of the lithosphere, and crustal influences on the underlying lithosphere, including radiogenic heat, implies that the source of the negative free-air gravity anomalies is less likely to be the subcrustal lithosphere and more likely to be located in the Antarctic crust. Examination of possible crustal features that might account for these anomalies leads to a consideration of subglacial topography and specific locales of anomalously low rock density.

In recent years geological and geophysical studies of the Antarctic continent have evolved from ground surveys and field work to airborne surveys, and more recently to satellite remote sensing. In each case, considerably more information has been introduced, which enables us to integrate earlier data with the more recent data to form a more comprehensive view of the gravity field of the continental and immediately offshore oceanic lithospheres.

Figure 1, which we have constructed from field-based gravity observations, and, in particular, CHAllenging Minisatellite Payload System (CHAMP) global positioning system (GPS), European Improved Gravity model of the Earth by New Techniques (EIGEN–1S) satellite data (Reigber et al., 2002; Barthelmes, 2002), depicts the complete Antarctic free-air gravity anomaly spectrum. Negative free-air gravity anomalies are prominent on the continent, and include anomalies in the Wilkes Land and northern Transantarctic Mountain sector, the Ross Sea sector, the continental interior, and the Weddell Sea sector. Some of these anomalies are comparable to gravity anomalies in other regions on Earth, but some are surprisingly different. The Ross Sea, for example, displays a “paradoxical” gravity field (Karner, 2004; Karner and Watts, 1983) of unusually large, absolute magnitude negative free-air gravity anomalies uncharacteristic of most other oceanic regions of this kind, and it appears therefore to require a unique explanation. In addition, the gravity data depict a somewhat radial circum-continental field of numerous positive free-air gravity anomalies, and comparatively weak negative free-air gravity anomalies offshore. All are of lesser absolute magnitude than the continental interior negative free-air gravity anomalies. The somewhat radial circum-continental field is a pattern not normally found in other oceanic regions, nor adjacent to most continental land masses.

Finally, the distribution of negative free-air gravity anomalies in interior Antarctica, extending from the Ross Sea to the Weddell Sea, appears to be regionalized to a central axis of the continent. Difficult to explain on the basis of traditional geological structures or Earth's interior structure, it is the purpose of this paper, first, to provide a useful gravimetric map of the Antarctic continent for use by other investigators and, second, to examine the Antarctic gravity spectrum in an effort to explain the likely source of these gravity anomalies as they may relate to the Antarctic mantle, lithosphere, or crust.

The initial data for the present study were acquired during the Victoria Land Traverse, a 4 mo, 2400 km seismic and gravity survey of interior East Antarctica (Weihaupt, 1961; Smith et al., 2011) (Fig. 2). Gravity observations were made using a Frost Gravimeter, ID no. C-3–65, operated from a 24 V power supply, thermostatically controlled to maintain a constant 107 °F temperature. The gravimeter was essentially free from drift, showing a variation of 0.1 mGal upon reoccupation of stations. Seismic observations were made using a Texas Instruments 7000B seismograph with a frequency range of 5–500 cycles per second (cps). A 24 trace system was used in which each trace represented a single 20 cps vertical component seismometer. Four banks of six amplifiers each were used in conjunction with the standard 7000B camera oscillograph, control unit, and dynamotor. Commencing at Scott Base on Ross Island, the traverse crossed the Ross Ice Shelf, ascended the Skelton Glacier through the Transantarctic Mountains, onto Victoria Land Plateau. From the head of the Skelton Glacier, the traverse crossed a major portion of East Antarctica northwest to 71°08′S, 139°11′E, and then east to the USARP (United States Antarctic Research Program) Mountains at 72°37.8′S, 161°32′E (Fig. 2). Gravity observations were made at 5.0 km intervals, and seismic soundings were made at 15.0 km intervals for subglacial rock surface control.

Subsequent gravity data were acquired from CHAMP GPS, EIGEN–1S low-altitude remote sensing satellite (Reigber et al., 2002). This EIGEN–1S data set, containing fully normalized spherical harmonic coefficients complete to at least degree/order 100, continuously and precisely tracked by global positioning system (GPS), provides powerful data to test gravity field models (Klokoenik et al., 2004).

Reduction and analysis of the gravity data collected in the field (Weihaupt, 1961, 1976) involved correction of observed gravity, 0.3086 mGal per meter of elevation, and comparison of the result with theoretical sea-level gravity for the latitude of each station using the international formula:
where ϕ is latitude. By assuming ice and subice rock densities of 0.90 g/cm3 and 2.67 g/cm3, respectively, with a resultant density difference of 1.77 g/cm3, 1.0 mGal is found to correspond to a change of 13.5 m in ice thickness on the basis of the gravity effect for a semi-infinite slab, that is,
where τ is the thickness of the plate per mGal, and the gravitational attraction is then,
where gz is the gravitational attraction in mGal, graphic is the gravitational constant, and б is the density difference between 2.67 g/cm3 and 0.90 g/cm3.

This has enabled identification of at least one major negative free-air gravity anomaly in the region of Station 531 (Fig. 2), which represents subglacial basin-shaped crustal topography, and is believed also to represent low lithospheric density. Reduction and analysis of the EIGEN GPS–1S data (Reigber et al., 2002), along with those of the field-based gravity traverse, have enabled identification of several other major negative free-air gravity anomalies in the Wilkes Land, Ross Sea, central continental, and Weddell Sea regions, as well as the somewhat circum-continental oceanic largely positive free-air gravity anomaly field. The continental and oceanic data, representing 43,200 data points from 60°00′S to 90°00′S latitude and 360°00′ of longitude at half degree intervals, were subsequently applied to a stereographic projection of Antarctica and its offshore periphery, and free-air gravity anomaly points were used to construct the gravity contour map (Fig. 1) of this segment of the Southern Hemisphere.

Distributed across Antarctica, the negative free-air gravity anomalies range from −20 mGal to −60 mGal and are accompanied by an apparently random distribution of fewer continental positive free-air gravity anomalies ranging from 0 mGal to +30 mGal, with the exception of a +50 mGal anomaly in central East Antarctica, and another +50 mGal anomaly near the coast of the continent. The circum-continental gravity values range from 0 mGal to +20 mGal in the Atlantic Ocean sector, and from 0 mGal to −20 mGal in the Pacific Ocean sector, except for a −30 mGal anomaly offshore in the Indian Ocean sector (Fig. 1). Specifically, these include three negative free-air gravity anomalies in the Wilkes Land and northern Transantarctic Mountain sectors (A, B, and G), three large negative free-air gravity anomalies in the Ross Sea sector (C, D, and E), four in the continental interior (F, H, I, and J), and five in the Weddell Sea sector (K, L, M, N, and O) (Weihaupt and Van der Hoeven, 2004; Abbott, 2006; Weihaupt et al., 2006, 2010).

A principal result of the Victoria Land Traverse is the discovery of a major negative free-air gravity anomaly in Wilkes Land. Centered at 71°31′S, 140°00′E, this feature is a minimum of 243 km across, with a minimum absolute magnitude of 158.3 mGal (Fig. 3) (Weihaupt, 1961, 1976). The region of the anomaly was traversed twice, confirming the validity of the data. In contrast to this large negative free-air gravity anomaly, free-air gravity on the Victoria Land Plateau traversed elsewhere by the field party ranges from −42.6 mGal to +23.7 mGal, a range of 66.3 mGal over 700 km. This is equivalent to a rate of change of 0.1 mGal/km. In contrast, the Wilkes Land anomaly displays a rate of change in gravitational acceleration more than forty times that observed regionally, suggesting a region of the lithosphere out of isostatic equilibrium (Heiskanen and Vening Meinesz, 1958). Based upon radiosound survey (Steed and Drewry, 1982), this Wilkes Land anomaly profile is seen to reflect also the profile of the subglacial rock topography, although topography does not explain the total magnitude of the anomaly. Some influence other than subglacial topography therefore appears to augment this anomaly.

As the first major negative free-air gravity anomaly recognized in Antarctica, the Wilkes Land anomaly has focused attention on other, more recently observed negative free-air gravity anomalies in Antarctica. Figure 4 lends better definition to the sizes, locations, and patterns of distribution of the negative free-air gravity anomaly sites (blue), and the positive free-air gravity anomaly sites (yellow). This map, which we constructed largely from EIGEN–1S data, reveals at least 15 negative free-air gravity anomalies distributed across the continent (compare with Fig. 1). Although the resolution from satellite remote sensing is inadequate to provide profiles comparable to the ground-based profile of the Wilkes Land anomaly (Fig. 3), we are able to construct profiles of the subglacial surface for anomalies A and B based upon radiosound survey (Steed and Drewry, 1982). These profiles appear in Figure 5 for comparison with the Wilkes Land anomaly (Fig. 3), and are found to be comparable in broad aspect. High and low points, identified as d, c, f, g, h, i, j, and k, while representing radiosound subglacial topographic points, may tentatively be compared to counterpart gravity/seismic points in Figure 3, viz., a, b, and c. While similar radiosound data are not available for the other EIGEN–1S negative free-air gravity anomalies observed in Antarctica, the profiles in Figures 3 and 5 call attention to their similarity and suggest that the profiles of the other negative free-air gravity anomalies may be similar.

While radiosound data reveal important subglacial topographic information, seismic and gravity data are more useful for identification of possible mantle (Guillou-Frottier et al., 1996; Mareschal et al., 1999; Ritsema et al., 2009; Verhoeven et al., 2009), lithospheric (Nyblade and Pollack, 1993), and crustal (Plescia, 1999) mass and density variations.

The major interior positive free-air gravity anomaly (+50 mGal) in East Antarctica (Fig. 1) exhibits a maximum gradient of 0.22 mGal/ km and is associated with subglacial topographic high ground (Lythe and Vaughan, 2000), demonstrating the importance of subglacial topography in accounting for free-air gravity anomalies. Steep negative free-air gravity gradients also occur with the negative gravity anomalies in the Ross Sea, and again at anomalies A and B (Fig. 1) in Wilkes Land. These sites represent negative free-air gravity gradients of, for example, 0.351 mGal/km, and 0.272 mGal/km, respectively. These gravity gradients exceed most gravity gradients elsewhere on Earth, although similar gradients occur over quite different and well-understood structures such as deep-sea trenches and tectonically active mountains. Such features, like those of the Mariana Trench and the Himalayas, bear little geological similarity to geological features in Antarctica, or to the shapes and distributions of Antarctica's negative free-air gravity anomalies. Similarly, the gradient of the Wilkes Land negative free-air gravity anomaly, south of and adjacent to A in Figure 1 (star icon), exhibits an unusually steep maximum gradient of 4.71 mGal/km.

The origin of the anomalies, functions of mass and distance variations, may be related to compositional, structural, or geologic mass or density differences of the mantle, subcrustal lithosphere, or crust (Weihaupt and Van der Hoeven, 2004). Gravity signals from all of these sources may, of course, be somewhat obscured by the overlying continental ice sheet. Mass variations may also result from density differences caused by temperature differences in the mantle, lithosphere, or crust, and may thus reflect radiogenic variations and heat-flow variations in these structural units. Heat-flow variations, manifested in gravity anomalies (Anderson and Dziewonski, 1984) and thermally driven mass movements reflected in density variations, may also account for variations in the gravity data. Tomographic studies, normally focused on variations in seismic velocities, allow density variations to be inferred as well. On these bases, we examine the strengths and weaknesses of the possible explanations for the Antarctic negative free-air gravity anomalies related to mantle, lithosphere, and subglacial crust.

At least three aspects of the Antarctic gravity field must be considered in providing an explanation for the Antarctic free-air gravity anomalies. These include (1) identifying possible alternative sources of these anomalies, (2) explaining the unusually large magnitudes of the negative free-air gravity anomalies, and (3) providing a rationale for the offshore largely positive and relatively low-absolute-magnitude negative free-air gravity anomalies around the continent.

Gravitational Effects of the Antarctic Mantle

Variations in Earth's mantle caused by composition, density, and thermal structure have the potential to create variations in gravity at Earth's surface, and therefore the potential to explain the Antarctic free-air gravity anomaly fields. Earth's broad gravity spectrum, for example, can be explained in part by a combination of viscous mantle flow and random density variations in the mantle (Richards, 1984; Steinberger and Holme, 2002). Similarly, seismic tomography (Vogt et al., 1998; Zhao, 2001) reveals that viscous mantle flow, in the form of mantle convection driven primarily by lateral differences in temperature and density (Anderson and Dziewonski, 1984; Van der Hist et al., 1991), accounts for much if not all of the mantle structure. Treating Earth's mantle as a highly viscous fluid, then, yields an excellent approximation for the behavior of the upper mantle, especially when whole mantle flow is assumed (Richards and Engebretson, 1992). Substantial density anomalies in such a fluid may thus drive a flow component both regionally and in the entire mantle (Richards and Hager, 1984; Richards and Engebretson, 1992), components which have the potential to generate gravity anomalies at Earth's surface. Such variations appear, therefore, to offer an explanation for the shapes, sizes, and regional distributions of the negative and positive Antarctic free-air gravity anomaly fields.

However, lower-mantle structural features are normally revealed as broad, large-scale variations (Richards, 1984), which are resolved into long-wavelength gravity anomalies representing features of the order of 2000–3000 km, i.e., much larger than those in the present study. The Antarctic continental interior anomalies as well as the circum-continental radial pattern of anomalies appear unlikely, therefore, to be related to lower-mantle effects. On a more continental or regional scale, massive xenoliths or subducted slabs of contrasting densities in the mantle are capable of producing gravity anomalies with wavelengths closer to the sizes of the Antarctic free-air gravity anomalies. Massive xenoliths (Rudnick and Fountain, 1995) and subducted slabs in the mantle (White et al., 1999; Bina and Navrotsky, 2000; Levin et al., 2002; Bellahsen et al., 2005; Shito et al., 2009; Krien and Fleitout, 2008) remain possible explanations for gravity anomalies in Antarctica; the shapes and locations of such features tend, however, to be quite different from most of those of the Antarctic gravity field.

Considering the Antarctic radial circum-continental gravity anomaly field, it is interesting to note the radial chains of elongate positive gravity anomalies around the passive continental margins of the Arctic Ocean Basin, for which there is “no simple unifying explanation” (Vogt et al., 1998). These Arctic Ocean positive gravity anomalies may be related to postglacial rebound or to rising thermal convection currents in the mantle, which tend to generate positive (rather than negative) anomalies due to accompanying elevation increase at the surface (Anderson and Dziewonski, 1984; Thomas et al., 2009). Thus, at least one other radial gravity anomaly pattern exists in an oceanic region, this one also in a high latitude and polar region, offshore of the continents around the Arctic Ocean basin. These circumstances tend to favor isostatic upper-mantle effects as the explanation for this offshore Antarctic circum-continental radial pattern. Such an explanation for the circum-continental radial field is believed to have merit also because the Antarctic continent and the continents around the Arctic Ocean basin both have experienced glacial depression and postglacial rebound, as the Arctic continues to do today. While revealing little of the negative free-air gravity anomalies across interior Antarctica, the radially elongated anomalies in the Arctic Ocean Basin and around the Antarctic continent imply the influence of upper-mantle flow (Anderson and Dziewonski, 1984; Vogt et al., 1998), especially as related to shallow sublithospheric mantle heat flow and convection (Nyblade and Pollack, 1993; Rolandone et al., 2002).

On the other hand, regarding the negative free-air gravity anomalies of the continental interior, the arguments that detract from a lower-mantle explanation of the negative free-air gravity anomaly field also apply to the mid- and upper mantle. That is, the magnitudes and wavelengths of anomalies generated in the upper mantle are components of, and resemble, those of the lower mantle (Richards, 1984). Additionally, regarding crustal temperatures and mantle heat flow beneath the Trans-Hudson orogen of the Canadian continental shield, for example, Rolandone et al. (2002) arrived at a mantle heat flow in the range of 11–16 mW m2, permitting only relatively weak convective and density-induced gravity anomalies. Furthermore, only relatively weak convection is evident in Earth's upper mantle, as revealed by seismic tomography (Anderson and Dziewonski, 1984; Van der Hist et al., 1991; Zhao 2001), as well as in the upper thermal boundary (Richards and Hager, 1984; Montelli et al., 2004). So, while the moderate temperature differences at the base of the upper-mantle layer are a function of the underlying mantle, the temperature and density structures at the top of the upper mantle are more a function of lithospheric control (Ritzwoller et al., 2001). Also, while the temperature at the Moho, 1073 K (Rolandone et al., 2002), is significantly above the threshold value for melting, radiogenically undepleted upper-crustal rocks remain below the melting temperature, and therefore dictate a lower bound on mantle heat flow. This, in turn, limits density and heat-flow variations in the upper mantle (Rolandone et al., 2002). Finally, heat-flow determinations and petrology suggest that there are only very modest temperature anomalies, even in the shallow mantle (Breddam, 2002). While a variety of relatively simple and complex models have been applied to the distribution of density anomalies in the mantle, the one that most nearly approximates Earth's observed gravity field is that of Steinberger and Holme (2002). This model prescribes randomly dispersed anomalies that may be applied to continental and larger than continental-scale domains, unlike the pattern of negative free-air gravity anomalies revealed in interior Antarctica.

Because conditions remain largely unknown in the convecting mantle below the continents (Rümpker et al., 2003), and because we are not yet near a quantitative knowledge of the mantle's density anomalies or heat-flow conditions (Steinberger and Holme, 2002), the heat flow at the Moho can be taken as something of a constant for large scale geologic provinces (Rolandone et al., 2002). Additionally, mantle heat-flow estimates, significantly affected by residual errors (Guillou et al., 1994; Jaupart et al., 1998), tend to be small relative to those of the crust, while in the upper mantle layer features tend to be dominated instead by surface tectonic effects (Dziewonski and Anderson, 1984). On the basis of these considerations, we believe that the mantle is unlikely to be the source of Antarctica's interior negative free-air gravity anomaly field. Variation in Moho temperature and heat flow is due mainly to differences in lithospheric radiogenic heat production (Rolandone et al., 2002), and thus differences in crustal geology. The Antarctic negative free-air gravity anomalies appear, then, to be more likely the result of lithospheric or crustal control than of mantle influence.

Gravitational Effects of the Subcrustal Lithosphere

In contrast to gravity signals from the mantle, gravity signals from Earth's lithosphere and crust are normally of greater absolute magnitude due to the shorter distance of the generating source from observing instruments, and due to distinctive geological domains that possess particular structural or density differences (Lucas et al., 1996; Leclair et al., 1997; Ashton et al., 1999). Such signals also derive from associated lithospheric heat-flow domains (Mareschal et al., 1999; Rolandone et al., 2002).

Treating the lithosphere in terms of viscous rheology, assuming a free upper boundary (Richards and Hager, 1984), and treating differences of average heat flow as functions of variations in lithospheric thickness and composition (Nyblade and Pollack, 1993; Jaupart et al., 1998), it is possible to broadly approximate lithospheric thermal conditions. For stable cratonic and shield regions, the maximum possible lithospheric thickness in which such variations can be detected by gravity sensing is taken to be 200–250 km (Jaupart et al., 1998), or at the upper limit, 200–350 km (Jaupart and Mareschal, 1999). Yet, because of lateral diffusion, such approximations are of limited use for heat-flow purposes, and variations of heat supply at the base of the lithosphere are not detectable at Earth's surface over distances of a few hundred kilometers (Rolandone et al., 2002). Furthermore, there is little evidence for significant differences in heat supply from the mantle to the lithosphere throughout Precambrian shields, such as that of North America (Rolandone et al., 2002), due at least in part to the relative lack of thermal structure in the upper mantle (Breddam, 2002; Foulger and Natland, 2003). For these reasons, we conclude that the Antarctic subcrustal lithosphere also is not likely to be the main signal source for the negative free-air gravity anomalies in interior Antarctica. Further, as Rolandone et al. (2002) noted, heat-flow variations at Earth's surface can only be due to the presence of different types of crust. We would modify this to include also the presence of crustal density and topographic variations. For these reasons, variations in Earth's crust are potentially more likely to be the cause of the free-air gravity anomalies of interior Antarctica. We tentatively conclude that while neither mid- nor upper-mantle conditions, nor subcrustal lithospheric conditions, are the likely sources of the interior negative free-air gravity anomaly field, subglacial topographic or crustal density variations are the more likely source of the anomalies.

Gravitational Effects of the Subglacial Crust

Geologic variations in the crust of East Antarctica now become central to interpreting the significance of the observed interior Antarctic negative free-air gravity anomaly field, and while much is now known of the Transantarctic and other mountain ranges, and of the geologic periphery of the continent, relatively little is known of the geology, heat-flow, and density variations of the interior beneath the continental ice sheet. Nonetheless, by direct survey (Fitzsimons, 2000; Studinger et al., 2004), and by extrapolation from studies on other continents, continental East Antarctica is considered to represent an ancient continental shield not unlike the shields of North America and Fennoscandia. Both of these have been well investigated, and both have many geological and geophysical properties in common. We believe it is reasonable and even necessary, therefore, to consider the East Antarctic continental crust as representing a generally comparable continental shield and, as Rolandone et al. (2002) reminded us, to use the same arguments, in this case for East Antarctica, as those applied to other similar geological provinces. This includes making similar estimates of, for example, geological composition, structure, crustal radiogenic heat production, and heat flow as those known or estimated for continental shields such as North America and Fennoscandia (Jõeleht and Kukkonen, 1998; Jokinen and Kukkonen, 1999; Russell and Kopylova, 1999). We therefore proceed on the assumption that East Antarctica is broadly geologically and geophysically comparable to these continental shields, and we next examine the crust as the possible source of the interior Antarctic negative free-air gravity anomaly field.

Given the geologic constraints on the mantle and subcrustal lithosphere discussed previously, observed relatively long-wavelength heat-flow variations can only be attributed to differences in the composition and structure of the crust (White et al., 1999), including differences in the major radiogenic heat generators, namely uranium and thorium (Burwash and Cumming, 1976; Bingen et al., 1996; Bea and Montero, 1999; Bizzarro et al., 2003). Current knowledge of heat production in the deep crust is summarized by other investigations (Rudnick and Fountain, 1995; Jõeleht and Kukkonen, 1998), but it is important to note that few thermal conductivity measurements have been made in the 700–1200 K range of relevance to the lower continental crust. Beyond that, there generally is no significant heat-flow change across crustal geologic contacts (Mareschal et al., 2000a), and determination of reliable average heat production on a large scale, for purposes of density-related heat-flow studies, is difficult (Jaupart, 1983). Because the surface heat-flow field does not display a systematic spatial pattern on regional or whole-continent scales (Decker et al., 1980; Jaupart and Mareschal, 1999), no continentwide systematic density-related heat-flow trend (nor related gravity signal) is normally observable (Mareschal et al., 2000b). Furthermore, removing the radiogenic heat contribution of the crust from broad heat-flow measurements is not possible without a geologic knowledge of the lower crust (Jaupart, 1983; Mareschal et al., 1999; Rolandone et al., 2002), and we have seen that relatively little is known of the geologic structure of Antarctica beneath the continental ice sheet. So, while radiogenic heat production is the major component of crustal heat flow, as well as a large component of the heat supplied to the lithosphere (Rudnick and Fountain, 1995), little is known of heat-related subglacial crustal geologic structure and density variations. On the other hand, geological structure, density, and topographic variations may be revealed by gravity and magnetic signals emanating from beneath the ice sheet.

Earth's continental crust is heterogeneous, often on a scale of some ten kilometers on average (Fountain and Salisbury, 1981). Such heterogeneity in geologic structure, composition, density, and topography translates into variations in gravity signals. In continental shields such as those of North America and Fennoscandia, structural, density, and topographic differences are well known (Jokinen and Kukkonen, 1999; Russell and Kopylova, 1999; Jaupart and Mareschal, 1999; Rolandone et al., 2002), as are their consequent geophysical properties. The strong correlation among surface geology, radiogenic heat production, heat flow, and gravity is therefore a function mainly of these geologic variations in the upper crust (Pandit et al., 1998; Zelt and Ellis, 1999; Rolandone et al., 2002).

Among the earliest evidence for variation in the geologic structure and density beneath the East Antarctic continental ice sheet and consequent gravity signals is the discovery of the Wilkes Land anomaly. We have seen that these data (Weihaupt, 1961, 1976), when combined with gravity data collected from Rouillon (1960), reveal a total negative free-air gravity anomaly in the region of 158.3 mGal. Representing a geological feature at least 243 km wide, this discovery supports the belief that the negative free-air gravity field in this area is a consequence of subglacial geological structure and, perhaps, geological density variations. Similarly, among the earliest evidence for variation in the topography beneath the ice sheet is that from a radiosound survey of the region bounded by 67°00′S–75°00′S and 105°00′E–155°00′E (Steed and Drewry, 1982). These data, we have seen, reveal a topographic lowland in the lower reaches of the Wilkes Subglacial Basin inland from George V Coast (Fig. 6). Extending well below sea level, this lowland feature in Wilkes Land, presently occupied by coastward-flowing continental ice, which has modified the subglacial topography, confirms the presence of complex subglacial topography. The unusually large absolute magnitude of the negative free-air gravity anomaly in this area suggests that an additional geologic component is required to explain the anomaly, such as low crustal density. The gravity decrease may therefore be related to the presence of a sedimentary basin or of a substantial thickness of glacial drift. On the basis of a sedimentary or glacial drift density of 2.20 g/cm3, and a density contrast between rock and sedimentary or glacial material of 0.47 g/cm3, the thickness (h) of lower-density materials required to account for this anomaly may be approximated by transposing g = 0.04185/h, i.e.,

A sedimentary or glacial drift thickness of 5180 m, however, is considered to be an unreasonably great thickness for low-density materials such as these, and is therefore regarded not to be the source of the negative free-air gravity anomaly in Wilkes Land. Subglacial topographic and subglacial rock density variations therefore appear to be the only remaining explanations. Both requirements could be satisfied by the presence of an impact crater, which would possess topographic basin and low rock density (from impact brecciation) characteristics.

We began by stating that at least three aspects of the Antarctic gravity field must be addressed in attempting to provide an adequate explanation for the presence of the prominent free-air gravity anomalies in Antarctica. Such an explanation must (1) identify possible sources of the anomalies, (2) explain the unusually large magnitudes of the negative free-air gravity anomalies, and (3) provide a rationale for the offshore positive free-air gravity anomalies. Each of these now has been addressed, and reasonable answers provided, and we conclude that the interior Antarctic negative free-air gravity anomaly field is a function not of the Antarctic mantle, but of the lithosphere, particularly the upper lithosphere, namely the Antarctic crust. Topographic variations and crustal density variations, other than those due to a sedimentary basin or glacial detritus, are adequate to account for the geographic size and the magnitude of the negative free-air gravity anomaly in Wilkes Land. We believe, also, that comparable explanations are likely to account for the other negative free-air gravity anomalies in interior Antarctica. The radial circum-continental free-air gravity anomalies offshore of the Antarctic continent, on the other hand, are regarded to be the result of depression and rebound of the oceanic lithosphere resulting from glacial and postglacial variations in the mass of the Antarctic continental ice sheet.

These answers are regarded as a preliminary step in defining and understanding the Antarctic free-air gravity anomaly field. We therefore recommend further detailed study of each of the major free-air gravity anomaly sites.

We wish to thank Geoforschungs Zentrum, Projektbereich 1.3, Potsdam, Germany, for providing pertinent gravity data in support of this study, particularly F. Barthelmes, P. Schwintzer, and Ch. Reigber. We wish to thank also J. Wyckoff for assistance in preparing our free-air gravity contour map (Fig. 1), as well as colleagues who accompanied us in the field: C. Lorius, W.M. Smith, A. Stuart, A. Heine, A. Taylor, L. Roberts, T. Baldwin, and W. Jackman. Finally, we also thank the National Science Foundation for support of the original field investigations, and the University of Colorado–Denver, the American Museum of Natural History, and Delft University of Technology for resource and administrative support.