2 Rr es ar vat tecture renew ft 28 Rep cha Cogn hie hhh sese he Ba men AB 3 See |e em 1 ace bb an gt ota oC Uti ite ome es 2 row 38 ‘cao gt rh Site he er et ‘Ser oar vn. un 5 21980 eee ng fos na Doce em (erage asl Athan Seaton and Tut neptune aur The importance of small-scale faulting in regional extension John Walsh*, Juan Watterson* & Graham Yielding+ * Fault Analysis Group, Department of Earth Sciences, University of Liverpool PO Box 147, Liverpool LBD 3B, UK ‘ Badoy Ashton & Assorates Ld, Winceby House, Wincaby, Horncastle, LUneainshice LN9 68, UK ‘A RECURRING observation in many studies of extensional basins has been that the amount of extension visible on normal faults (Cor example, on seismic reflection profiles) is significantly less ‘than the amount of extension indicated by crustal thickness and ‘thermal subsidence. One mechanism suggested to account for this diserepancy is small-scale faulting, with offsets too small to be resolved seismically”. But earthquake studies*'” indicate that ‘small faults are responsible for only a small fraction of the total seismic moment in an active area. Scholz and Cowie" have recently attempted to extend this approach to the total strain at the end of a finite deformation interval by combining scaling laws describ {ng the distributions of fault lengths and displacements, Here ne present fault displacement data that directly conflict with Scholz and Cowie's conclusions, and imply that up to 40% ofthe extension may be missed by summing fault offsets on basin profiles. The ult population at the end of a long deformation interval may differ substantially from that responsible forthe earthquake popu lation at any one time. To characterize the fault population in a basin, we have assembled measurements of fault displacements ata wide range Of scales: regional seismic data, oilfield seismic data and oilfield well cores. Figure 1a shows cumulative frequency plots of fault displacements on two regional profiles spanning the Viking Graben (northern North Sea); on each profile, the dip-slip displacementof one horizon was measured ata observed faults NATURE + VOL 351 - 30 MAY 1991 LETTERS TO NATURE Both of the data curves are roughly linear, indicating a power: law relationship between the Fault displacement and the number of readings greater than that displacement. The slopes of the plots are ~-09 and ~08. Both data curves become Mat at displacements below the effective limit of seismic resolution for these profiles (not all offsets less than ~ 100 m are seen) Figure 16 shows cumulative frequency curves for fault throws (vertical component of displacement) measured from a detailed depth structure map of the Troll eilfel, located on the regional profiles plotted in Fig, 1a. Throws were measured on parallel sample lines roughly perpendicular to the repional fault strike Data from a single sample line produce a roughly linear eurve like that for the regional profiles in Fig. 1a. As more sample lines are added to the plot, a steep right-hand segment develops, but the central segment maintains a slope between ~0.7 and =08, representing the average population slope for the individual sample ines across the oilfield. The overall form of the data curve in Fig. 1b is typical of line samples through faulted areas" Measurements of fault density at the subsseismic seale are provided by Gabrielsen and Koestler, who examined well cores fom the Troll field Ina total of ~1,800m of core, the average rn j i FG. 1 a Cumulative frequency plots for fault splacements on two region seismic profes (proties | (squares) and i (crosses) of ref. 4). Oip-sio splacements wore measured on the vapermost Mise Jurassic on depth converted interpretations across the nartn Viking Graben and record the Late Jressic/earlest Cretaceous episode of faulting" Data from both profiles show a roughly linea astibution n ths log pt, with sones. of about ~09 and ~08 indicating @ power-law relationship between ds. placement and cumulative number. & Cumulative frequerey measurements for tout throws (vertical eisplacements), measured on sample lines across 2 depth structure map of a Jurassic horizon in the Troll oitfila (on the Fegjonalprfles of 8) Numer of eample lines is indicated foreach cuve ‘The plot fr single sample line is roughly linear, ke those in a). Adding readings from progressively more sample lines produces a steep right-hand ‘Segment, but the contral part of the plot mainains te same slope, ep ‘resenting the average population slope on al the sample lines, 3901 (© 1991 Nature Publishing GroupLETTERS TO NATURE fracture density was 0.1m“, This figure includes small faults (with displacements down to ~1 mm) and also joints; without fine lithological layering, small faults cannot readily be distin: guished from joints All of the above datasets comprise displacement measure- ments on one-dimensional transects through the same extensional basin. The datasets derived from seismic reflection data correspond to sub-horizontal one-dimensional transects (horizon dips being <10*, and generally d, and @ is a constant that depends on the sample size. The exponent 'S determines the relative contributions of large and small dis: placements to the total amount of displacement affecting one horizon on the profil. {As bedding dips in the datasets studied are low (<10°), we ccan neglect rotations and consider the heave as being a reason- able measure of the amount of extension on a fault". Assuming that the slip direction on the observed faults lies in the plane of the profile, the sum of the heaves represents the faull-related basin extension along the profile. (The total amount of extension may be greater if there is a significant plastic deformation in the rock volume.) Marrett and Allmendinger"” show that the sum of the displacements on the N largest fault offsets along, the one-dimensional transect is given by Xda, A IUSEE MSHI US ens) (1) and that the total displacement (down to a lower limit of zero displacement) is given by oot dol YEH IVS kN NED om) sine Q) which converges rapidly as N is increased. For equation (2) to be finite, S<1. Note that Figs 1 and 2 imply 0.7 100 m are included, then only ~35% of the total fault moment would be recognized The difference between the results outlined here and those presented by Scholz and Cowie" arises from the types of data Used. Whereas our method is based on objective one: dimensional sampling of fault displacements, which sives a direct measure of extension, the method used by Scholz and Cowie requires an estimate of the fault population ina three- NATURE - VOL 351 + 30 MAY 1991 LETTERS TO NATURE Proportion of ext 0.0~frereerseyrrerrreeepreererrr rye} Minimum heave included Mae ne Syma see sro at anes Sacsan ah Pasar & nc ae cs pay sented ea a haa tartare a eaten 393 (© 1991 Nature Publishing Group