Based on the conclusions of the generic models, we attempted to design a model that more closely approximated the rifting history of the Newfoundland-Iberia conjugate margins. The generic model suite results suggest that, if the first phase of extension were significant, the second phase of extension would occur at a different location. Because the second rift phase on these margins occurred in roughly the same place as the first, we infer that the first phase of extension must have been sufficiently minor not to create a strong zone in the lithosphere and shift the second phase of rifting elsewhere.
We aimed to approximate only the gross lithospheric features of the Newfoundland and Iberian Margins. Primarily, we were interested in reproducing the crustal thickness profile across the two margins. On the Canadian side, the crust is 35 to 38 km thick across the Grand Banks and thins gradually seaward of the Jeanne d'Arc Basin to an ultrathin 4 to 8 km beneath the Newfoundland Basin. On the Iberian side, the crust on the continent is about 30 km and thins gradually to 5 to 8 km over a broad area. We tried to model the crustal profile, as well as the distribution of upper crust extension and subsidence.
The original width of the model and the amount of extension it underwent were based on values estimated for the Newfoundland and Iberian Margins (Tett, 1993); for simplicity, we rounded the original width to 600 km and the amount of extension to 400 km. (As in the generic models, these parameters reflect our supposition of a continental nature of Newfoundland Basin and Iberia Abyssal Plain crust.) Of that 400 km, we assigned 60 km to the first rifting phase and 340 km to the second. This choice was based on the value of β = 1.1 that Hiscott et al. (1990) cited for the Triassic sequences of the Bristol Channel area and in the Wessex Basin in the British Isles. Although we are uncertain of the method by which this figure was estimated, and even though it applies to the British basins and not the ones examined in this paper, it is the best estimate available. Paleomagnetic data cannot resolve the amount of extension between Iberia and North America (J.E.T. Channell, pers. comm., 1993). The upper Triassic synrift sediments of the Newfoundland and Iberia shelf basins are buried so deeply, and salt tectonics have deformed the sediments so extensively, that reliable estimates of extension for the Triassic are difficult to obtain.
We assigned durations (25 and 45 Ma, respectively) to the first rifting phase (which lasted roughly from 230 to 205 Ma) and the resting phase (which lasted from 205 to 160 Ma) that are similar to those used in the generic model suite. We chose a duration of 35 Ma (instead of 50 Ma) for the second rift phase, however, using the following reasoning. The second phase is though to have begun at about 160 Ma. Our preliminary choice of 110 m.y. for the initiation of seafloor spreading yielded a duration of 50 Ma for the second phase of the generic model suite. Because a date of 125 Ma is a more accurate average date for the rift-drift transition on the transect discussed here, we reduced the estimate of second-phase duration to 35 Ma. (This had the net effect of making the second phase of rifting faster, making less likely a shift in locus of extension.) The choices for distribution and duration of extension during the two rifting phases results in the rifting path shown in Figure 11. The extension rate of 9.7 km/m.y. (half-rate of 4.85 mm/yr) for the second phase is significantly less than the rate of initial seafloor spreading on the southern half of the Newfoundland margin (24 to 26 km/m.y.; half-rate of 12 to 13 mm/ yr) given by Srivastava et al. (1990). Although this discrepancy is somewhat discouraging, it must be remembered that the extension rate used here is an average over a period of 35 Ma, whereas the initial seafloor-spreading rate is an instantaneous one. Though it is not included in our model, it is conceivable that the instantaneous rate of extension increased during the second phase of rifting, resulting in an average rate quite a bit lower than the rate of rifting just before seafloor spreading.
Several parameters for the initial model (Fig. 12) are less well constrained. The original crustal thickness is almost impossible to constrain. The crust beneath Newfoundland itself, in the heart of unextended Appalachian orogen, ranges in thickness from 40 to 45 km. The Paleozoic orogen was enormously complex, and the part of the orogen about which we are most concerned has been obliterated by extension. Lacking any independent constraints on the original crustal thickness profile, we chose original crustal thicknesses between 40 and 46 km.
The crustal weakness in the model (Fig. 12) was chosen with a similar level of speculation. It was designed to mimic the distribution of the rift basins on the continental shelves of the Newfoundland and Iberian Margins. The shelf basins nucleated on weaknesses inherited from different orogenies on both sides. On the Iberian side, the Porto-Badajoz-Cordoba shear zone (PBCZ)—the major Paleozoic transform fault in western Iberia (Lefort, 1989)—became the eastern boundary of the Lusitanian Basin and served as the landward limit of extension. In southeastern Newfoundland, a similar major strike-slip fault, the Dover fault, appears to be the PBCZ's counterpart on the opposite margin (Keen et al., 1986) and would seem to be a location favored for upper crust extension. Basins on the Newfoundland margin, however, nucleated on Precambrian Avalonian weaknesses, hundreds of kilometers seaward of the Dover fault. Thus, it is impossible to attribute with any certainty a single factor that controlled the upper crustal basin formation, and hence that could be chosen as a "crustal weakness" in the margin-specific model. The role of any crustal weakness is a major uncertainty in the model.
The width and location of the crustal welt (the mantle weakness) in the initial model (Fig. 12) are speculative as well. Without exception, attempts to design a model with a wide crustal welt led to "wide rifting"—a style that we do not think is manifested on these conjugate margins. To achieve the narrow rifting style that appears to have occurred here, we had to use a narrow mantle weakness. The position of the crustal welt in the model reflects the estimates of original margin width (Tett, 1993); the Newfoundland Margin (to the left) is designed to be much wider than the Iberian (to the right). Also, the location of shelf basins and the crustal profile indicate a predominantly symmetrical rifting style. Attempts to model these margins using a model with moderate to large asymmetry resulted in extremely poor fits to the observed crustal profile. Thus, the weaknesses in the initial model are roughly symmetrical.