JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 96, NO. B7, PAGES 11,759-11,777, JULY 10, 1991 Gravity Anomalies Over Extinct Spreading Centers' A Test of Gravity Models of Active Centers JASONJONAS, STUART HALL, AND JOHN F. CASEY Department of Geosciences, University of Houston, Houston, Texas Gravity fields over active mid-ocean ridges display long wavelength free-air gravity highs with superimposed short wavelength anomalies over the spreading axis. These axial anomalies are negative over slow spreading active ridges with amplitudes of typically 30-70 mGal and wavelengths of 40-60 km but decrease in amplitude and wavelength with increasing spreading rate until, at fast spreading centers, the anomalies are positive (10-20 mGal) with wavelengths of about 20-30 km. Similar free air anomalies are observedover extinct spreading ridges and typically show the same paleo-spreading rate dependency, suggesting that a large componentof the anomaly is not dynamically maintained. Calculation and removal of gravitational effects from basement relief resultsin significantresidual gravity lows at the extinct ridges analyzed here, even though the thermal effects associated with active spreading have diminished and no longer contribute to suchshortwavelength gravity lows. The amplitudes of the residual lows at slowpaleo-spreading rates are 25-45 mGal with haft widths of about 20-45 km, and decrease with increasing spreading rate to about5 mGal amplitude and 10-15 km half-width over fast paleo-spreading ridges.The amplitudes and wavelengths of these residualanomalies are nearly identical to those of the residual anomalies observed over active spreading centers. Most previouslyproposed modelsof mid-ocean ridge structure are unable to explain these anomalies andmany are inconsistent with a leastsome aspect of the large body of knowledge currentlyavailablefor mid-oceanridges. This research providesa further constraint on geometric models of accretion, as these models predict vastly different gravity fields. Results indicate that a frozen-in low density accretionary structure is probablyconfined to within about 10 km of either side of the extinct axis. This conclusion is consistent with the presenceof a low-density gabbroic root situated at shallowmantle depths beneath the ridge axis. This root tends to narrow with increasing spreading rate, producing the required residualgravity field variation. INTRODUCTION The formation of oceanic crust and its mode of accretion at mid-ocean ridges is of fundamental importance to thetheory of plate tectonics. Several geometric models of accretion have beenproposed [e.g., Greenbaum, 1972; Cann, 1974;Sleep, 1975, 1978;Orcuttet aI., 1975, 1976; Dewey and Kidd, 1977; Nisbet and Fowler, 1978; Kusznir, 1980; Herron et aI., 1980; Pallister and Hopson, 1981; Lewis, 1981; Casey and Karson, 1981; Elthon et al., 1982;Derrick et aI., 1987]. Many of these modelsfocuson only one or two aspects of the broad range of geological and geophysical data presently available for mid-ocean ridges and ophiolite co•nplexes. Such proposed models therefore often explain some of the observed phenomena associated with the formation of oceanic lithosphere very accurately, but when testedagainst other types of data, do not satisfy the further constraints they impose. Geological and geophysical data collected over mid- ocean ridges worldwide, and studies of ophiolite complexes provideseveralimportant thermal, mechanical, geochemical and structural constraints. Proposed models, therefore, must not only be able to simulate the observed data in a specific study, but also satisfy these further geological and geophysical constraints if they are to havebroad applicability and accurately represent theactual structure of mid-ocean ridges and the geometryof accretion. Extinct spreading centersprovide an excellent means of evaluating proposed models in that such models must both accurately yield the geologicaland geophysical observations at activespreading centers, and alsopredictobservations over Copyright 1991 by American Geophysical Union Paper number 91JB00617. 0148-0227/91/91JB-00617505.00 extinct spreadingcenters as the geometrical model is allowed to lose its dynamicallymaintainedaspects and be brought to extinction. Several geometricalmodels of accretion have been formulated solelyon the basis of gravitystudies at active ridge segments [e.g., McKenzie and Bowin, 1976; Cochran, 1979; Tapponier and Francheteau, 1978]' and some of these were generated from data gathered over spreading centers that representonly fast [e.g., Lewis, 1982; Madsen et aI., 1984] or slow (Linet al., 1990) spreading centers. In such cases,the model specifically attempts to explain geophysical observations at a particular ridge segment, but cannot necessarily be extended to other active or extinct spreading centerswith different spreading rates. In many of the models, large components of the geophysical anomaly are assumed to be dynamicallymaintained, so that it should not be possible to see these components at older extinct spreading ridges.Here we present gravity analyses from various extinct mid-ocean spreading centers that we feel provide constraints on the component of gravity anomaliesat active spreading centers that can be maintained by dynamic aspects of spreading and also provide a reliable means with which to evaluate proposed geometrical and geophysical modelsof accretion. The gravity signature over an active ridge axis is an important geophysical parameter that provides useful constraintson the structure of accretionarymodels of mid- oceanridges. Gravity fields over active slow spreading centers typically show free air gravity anomaly lows with amplitudes of about 30-70 mGal and wavelengths of about 40-60 km [Watts, 1982; Hall et al., 1986]. This free air low generally decreases in amplitude and wavelength with increasing spreadingrate to the point that at fast spreadingcenters such as the East Pacific Rise [ e.g., Lewis, 1982], the anomaly becomespositive with an amplitude of typically 10-20 mGal and wavelength of about 20-30 km. Gravity fields over extinct spreading centers show very similar gravity anomalies to thoseover active ridges [Watts, 1982] and typically follow the 11,759