Coincident magnetotelluric, P‐wave and S‐wave images of the deep continental crust beneath the Weardale granite, NE England: seismic layering, low conductance and implications against the fluids paradigm

SUMMARY It has been postulated that deep crustal zones of both enhanced electrical conductivity and seismic layering may be ascribed a single physical explanation, and more specifically that both result from the presence of free, interconnected fluids around the brittle‐ ductile transition zone. We analyse three complementary geophysical data sets— broadband (10’2‐104 s) magnetotelluric (MT) and P-wave and S-wave deep seismic reflection—from coincident profiles traversing the Weardale granite in northeast England and demonstrate that the lowermost crust in this tectonically stable region contains insuY cient free fl uid to explain the highv P /v S ratios associated with very bright, subhorizontal lamellae seen within the lower crust (below 22 km depth) beneath the granite. Temperature‐depth profiles are presented, and the apparent global correlation between the depths to the tops of zones of enhanced conductivity, seismic reflectivity and inferred brittle‐ductile temperatures of 300‐450°C is also re-examined. The seismic layering demonstrably corresponds to higher temperatures than those usually attributed to the brittle‐ductile transition zone. A priori information of diVerent scalelengths is applied to the problem of static-shift correction of the MT data. Gravity models constrain the depth extent of the granite batholith, whilst pre-existing DC measurements and a borehole drilled into the granite provide information concerning the resistivity structure of the uppermost 500 m. A 3-D model study investigates the significance of possible 3-D induction eVects on the MT data, as implied following application of decomposition. Our results challenge the widespread and little-tested assumption that deep crustal conductors and seismic layering are physically interrelated and occur at the same depth. With static shift and 3-D eVects in the MT data accounted for, we demonstrate that below the Weardale granite the onset of any enhanced conductivity is shallower than the top of the seismic layering, and furthermore that the observed deep crustal conductance of less than 200 S is significantly less than that to be expected (>2000 S) if the high v P /v S ratios calculated from the seismic data are to be explained entirely by fluids. The total conductance of the mid to lower crustal region combined is too low to allow for the significant quantities of free fluids required to explain lower crustal seismic layering through this medium without challenging current hypotheses concerning pore geometries and interconnectivity.

relative proximity to the Earth's surface, endures. Particularly 1 INTRODUC TION enigmatic and contentious for geophysicists are the origins It is almost 30 years since man first beheld lunar rocks, yet and preservation of deep crustal conductivity and seismic the tantalizing inaccessibility of the lower crust, in spite of its layering. Typical electrical conductivity profiles, for both active (e.g. Jiracek et al. 1983;Kurtz et al. 1986) and shield regions (e.g. Connerney et al. 1980;Korja et al. 1989;Vanyan et al. surface layer (attributable to ground water) and underlain by sources of deep crustal fluids may exist in active regions, the appeal of similarly ascribing deep crustal conductivity and a conductive layer, usually ascribed to depths of between 15 and 25 km and assigned resistivities spanning from 1 to reflectivity phenomena to the perpetual presence of free, extensively distributed, lower crustal brines in stable regions 100 V m [compared with 106 V m, expected for dry crystalline rocks (Shankland & Ander 1983;Haak & Hutton 1986)].
is weakened by the realization that, being of comparatively low density and therefore gravitationally unstable, such fluids Meanwhile, seismic reflection syndicates, comprising BIRPS in the U.K. (e.g. Smythe et al. 1982;Brewer & Smythe 1983; would tend to percolate upwards, evacuating the hot, ductile lower crust within geologically short time periods. Applying Matthews 1986;Hall 1986;BIRPS & ECORS 1986), ECORS in France (e.g. Bois et al. 1987), DEKORP in Germany Darcy's law and ascribing realistic porosities and permeabilities, Warner (1990) calculated a maximum of 106 years for the (e.g. DEKORP 1991; Wenzel et al. 1987;Fuchs et al. 1987), COCORP in the US (e.g. Allmendinger et al. 1983) and expulsion of interconnected water from the lower crust. Thus, whilst free fluids may contribute to high electrical conductivities geophysicists in Australia (e.g. Mathur 1983), have all imaged laminated structure, of alternating high-and low-velocity and seismic reflectivities in active regions, where a rechargeable source of fluids may be available, they can only provide a layers, within the lower continental crust. This globally significant, lower crustal lamination, sandwiched between a broadly generally viable explanation in light of a convincing mechanism for their long-term lower crustal maintenance. transparent upper crust and upper mantle, is usually illresolved (Reston 1988), but is manifest as a laterally extensive, Several mechanisms whereby the lifetime of free, interconnected, lower crustal fluids may be extended to geologically multitudinous sequence of short, brightly reflective, mostly subhorizontal but occasionally convex or dipping, elementary significant time periods have been proposed. A commonly cited mechanism is permeability sealing, usually expressed as units (Meissner 1989). Some seismic sections also evince banded reflectors in the mid crust (DEKORP 1991), but never an impermeable layer, at mid-crustal depth, created by mineral precipitates. Numerous researchers have hypothesized saturated in the rheologically disjunct mantle.
Any explanation of crustal high-conductivity anomalies must solutions of silica rising in the lower crust and induced by decreases in temperature and pressure to precipitate the silica acknowledge the anomalous composition or physical state of the circumjacent rocks. Some explanations cite Proterozoic along grain-edge boundaries (e.g. Etheridge et al. 1984;Haak & Hutton 1986;Hyndman & Shearer 1989). plate boundaries (e.g. the Iapetus Suture zone; Jones & Hutton 1979) or dewatering of sediments within subduction However, whilst an impermeable layer might present a transient barrier to ascending fluid, it would also present a boundary zones (e.g. Kurtz et al. 1986;Jones 1987;Wannamaker et al. 1989). Other, more generally applicable, for the concentration of fluid by gravity such that the boundary pressure might intensify sufficiently for the fluid to egress. inferences include the presence of graphite (e.g. Brown 1979;Mareschal 1992) or black shales (Stanley 1989), free Sanders (1991), drawing on evidence for the preservation of dry anhydrous rock samples circumjacent with wet, low-grade aqueous fluids (e.g. Connerney et al. 1980;Shankland & Ander 1983;Gough 1986;Hyndman & Shearer 1989), partial melts metamorphic conditions, advances the hypothesis of 'selfsealing hydration'. This model assumes extensive, lenticular, (e.g. Gough 1989), or anastomosing, possibly mylonitized or graphitised, shear zones (e.g. Haak et al. 1997). Seismic lamellae lower crustal, granulite lozenges, divided by a network of anastomosing, retrogressively satiated, brine-soaked, ductile, have been variously ascribed to stratified zones of free fluids (e.g. Hall 1986;Hyndman & Shearer 1989;Merzer & Klemperer shear zones. The products of marginal hydration reactions form impermeable envelopes around the granulite lozenges, 1992), mafic sills (e.g. Christensen 1989; Juhlin 1990), horizontal foliations or ductile shear (e.g. Reston 1988), producing arresting further retrogressive advances into the granulites and preserving their interior dryness. However, Frost & fabric reorientation or mylonitic layers, magmatic underplating (e.g. Warner 1990) and, less commonly, variations in Bucher (1994, considering fluid-amphibole equilibria reactions at lower crustal temperatures and pressures, conclude that metamorphic grade (Fountain & Salisbury 1981), anisotropy (Kern & Fakhimi 1975) or alpha to beta quartz transitions such hydration ceases only when all free fluid has been taken up. (Christensen 1989).
Concomitance has been inferred between crustal high-The brittle-ductile transition zone has also been postulated to act as a sealant against ascending fluids (e.g. Gough 1986), conductivity zones and zones of seismic layering. In magnetotelluric studies, however, spatial uncertainty is spawned by the with hydration reactions associated with the top of the greenschist facies possibly creating an impermeable zone (Hyndman lack of constraint imposed on the thickness of the conducting layer and the affliction of static shift, such that poorly con-& Shearer 1989). Apparent support for the hypothesis of fluids trapped around the brittle-ductile transition is provided by strained models of upper and lower conductivity boundaries have rendered such inferences contentious (Jones 1987). global data compilations, which have been interpreted as suggesting a causal relationship, between surface heat flow Of the principal candidates for explaining deep crustal zones of seismic layering and electrical conductivity jointly, free (indicative of temperature at depth) and the depths to the upper boundaries of zones of high conductivity and seismic aqueous fluids, being able to provide a mechanism for explaining both phenomena, are the most frequently cited (e.g. Hyndman reflectivity (Á dám 1978;Hyndman & Shearer 1989), with the tops of conductive and reflective zones corresponding to 1988; Hyndman & Shearer 1989;Marquis & Hyndman 1992;Merzer & Klemperer 1992). However, the notion of jointly temperatures in the range 300-450°C-temperatures which have been associated with the transition from brittle-elastic to explaining deep crustal zones of anomalous reflectivity and conductivity via the presence of deep-crustal, free fluids evokes ductile crustal behaviour (e.g. Hyndman & Shearer 1989). The apparent correlation tends to be disrupted in tectonically active a mass of contradictory evidence. Whilst enhanced near-surface conductivity is easily reconciled with pore fluid conduction regions, where the conductive zones are elevated, probably due to increased porosity (Á dám 1987). associated with ground water and compacting sediments, and L ow conductance versus high v P /v S ratios below the Weardale granite 421 The degree to which the lower crust is pervaded by water is Block. Devonian in age, its intrusion has been associated with the closure of the Iapetus Suture. The granite batholith is of pertinence to the modelling of all geodynamic processes, yet, to date, crustal fluid distributions are poorly constrained.
flanked to the north and south by the Northumberland and Stainmore troughs, respectively. In this paper, we re-examine the hypothesis that both lower crustal conductivity and seismic layering arise as a result of The lateral extent and depth of the granitic intrusion has been delineated using gravity, magnetics and deep seismic the presence of free fluids. Taking complementary seismic reflection (both P and S wave) and magnetotelluric data from reflection profiling. The batholith is elongated in a NE-SW orientation, and extends to a depth of at least 8 km. A depth coincident profiles in NE England, we question how wellconstrained past inferences that seismic layering and deep of less than 1 km to the central granite cupolas, increasing towards the granite's fault-bounded margins, has been deter-crustal conductivity onset at the same depth have been. We further demonstrate that explanation of the layering evinced mined from 2-D gravity modelling (Evans et al. 1988). The contacts slope moderately steeply outwards. A small magnetic by the seismic data in terms of free fluids would require significantly higher conductances (>2000 S) than are actually basement feature is contiguous with the northern margin of the granite beneath Blanchland. A borehole at Rookhope supported by the MT data (<200 S) in our survey area. Temperature-depth relationships are also considered.
penetrated granite at a depth of 390 m. Resistivity logs from the Rookhope borehole (Dunham et al. 1965) indicate that Acquisition, processing and limitations of the MT data are detailed. These aspects of the seismic data have already been the sedimentary overburden has resistivities of 100-200 V m, and that the uppermost granite has resistivities of 1-2 kV m. described elsewhere (Evans et al. 1988;Ward et al. 1992). In particular, static shifts, three-dimensionality and modelling Borings at Allenheads revealed graphitic intrusion contacts (Creaney 1980), expected to be conductive. Following attempts assumptions applied to the MT data are considered in order to facilitate assessment of the interpretational constraints.
at radiometric dating, the Weardale granite has been ascribed a range of ages, but Fitch & Miller (1965) assert that the true age is at least 390 Myr, with lesser ages being anomalously 2 EXPERIMENTAL LOCATION low due to later contamination by mineralising solutions. Enhanced radiogenic heat production elevates the observed The area of northeastern England underlain by the positively buoyant Alston Block manifests a complex, but well-surface heat flow over the granite to 95 mW m−2 (Evans et al. 1988). Thermal conductivities of 3.1 W m−1 K−1 for the granite documented, geological history. The region is bounded on three sides by the Stublick/Ninety Fathom, Pennine and Lunedale/ and 2.2 W m−1 K−1 for the overlying, carboniferous sediments were logged in the Rookhope borehole (Dunham et al. 1965). Butterknowle fault systems (Evans et al. 1988; Fig. 1), and comprises 400-1500 m of Carboniferous sediments, uncon- The Weardale granite, embracing the properties of high electrical resistivity and low seismic attenuation, is an ideal formably overlying a basement of strongly folded, Lower Palaeozoic slates, into which the Weardale granite is intruded.
upper crustal unit through which to probe the deep crust using combined seismic and electromagnetic techniques. Being over-This batholith of low-density, crystalline rock is probably isostatically responsible for the relative buoyancy of the Alston lain towards its centre by less than 500 m of sedimentary cover, this batholith of dry, crystalline rock, intruded into now-stable ( Fig. 2; Ward et al. 1992). The availability of near-normalincidence shear-wave data, coincident with the compressional-continental crust during the Devonian, may be considered as a window to the deep continental crust representative of stable wave data, allows calculation of v P /v S ratios which, together with the relative P-and S-wave reflection strengths, impose tectonic regimes. The tectonic and magmatic stability of the region precludes transient conductivity phenomena, such as tighter constraints on the apportionment of crustal fluids. Granites, constituting a more differentiated extreme of dewatering of subducted sediments [a candidate for explaining data from Vancouver Island  or recent lower crustal composition, may overlie a lower crust which is more mafic than average and which might either, as a result of its crustal magmatic addition [as Oxburgh & O'Nions (1987) proffer to explain the Rhinegraben and Black Forest data more fractionated nature, evince increased reflectivity or, as a result of being depleted of its felsic component, manifest weaker (Fuchs et al. 1987)].
In 1986, bright, layered reflections from the lower crust were reflectivity (Ward 1991). That lower crustal seismic layering is not a feature confined to deep crustal zones underlying granites recorded by the British Geological Survey, in the vicinity of the Weardale granite (Evans et al. 1988). In 1988, in their first only is not a matter of contention. Other seismic profiles around the British Isles and in other locations not sited over on-shore reflection venture (WISPA), BIRPS acquired the first continuous, lower crustal shear-wave (S-wave) reflection granites have also convincingly imaged lower crustal layering. It would therefore seem either that these layered features data, coincident with the highest-quality segments of the BGS compressional-wave (P-wave) reflection profiles. Lower crustal formed after the intrusion of the granite into the upper crust or that the granite was injected laterally in such a way as not to seismic layering is revealed in both P-and S-wave sections Figure 2. (a) P-and ( b) S-wave shot gathers revealing lower crustal seismic layering beneath the Weardale granite. For direct comparison with respect to depth, the two-way traveltime for the shear-wave stack is compressed relative to the compressional-wave stack (Ward et al. 1992) so that the same features occur at approximately the same position on both displays. L ow conductance versus high v P /v S ratios below the Weardale granite 423 disturb the layering. Models of granite emplacement are highly chloride electrodes with dipole lengths of approximately 100 m. conjectural, but given the significant geochemical variations observed in granite cores from the Rookhope borehole and The profiles traverse remote moorland and woodland. 50 Hz and 150 Hz notch filters were employed to remove the given the apparent lack of any significant intrusion-related folding, it seems likely that emplacement of the Weardale fundamental and first odd harmonics of power-line noise. All MT data were processed using the robust scheme of Egbert & granite probably occurred over a time span of several hundred million years and involved several distinct phases of intrusion Booker (1986). Where remote reference stations were not successfully employed, earth response functions were derived (Robson 1980). If the magma welled up from directly below the current location of the batholith, then any water encountered from a weighted average of the up-and down-biased impedance tensors normalized by the squares of their associated errors as it advanced would have been dissolved into the melt. However, it may be that the magma was channelled up through (Sims et al. 1971). Owing to the close proximity of sounding sites, the apparent resistivities and phases are strongly similar discrete vents, mushrooming only as it encroached upon the Earth's surface. This might explain the five interconnected from site to site, though static shift effects are apparent (Fig. 3).
Pseudosections displaying the phases and resistivities of sites bosses within the batholith's make-up. Or it may be [as is probable in the case of the Cornubian granites (Shackleton contoured against frequency and along profile location can be found in Simpson (1994). et al. 1982)] that the granite was injected laterally, leaving the present, underlying, lower crust undisturbed. Deep crustal Fig. 4 shows how both the traditional skew (Swift 1967) and Groom-Bailey rms misfit parameter (Groom & Bailey 1989) fluids may have been flushed out during granite emplacement and, in the absence of a source of fluids postdating granite soar (disproportionally compared with possible changes in the error structure of the data) at mid to long periods. In this emplacement, deep crustal conductivity may actually have been reduced.
period range, the Groom-Bailey twist and shear parameters ( Fig. 5) also fail to fit the frequency independence criteria of the decomposition hypothesis, whilst at frequencies of 3 MT DATA ACQUISITION AND 100 Hz−1 second, twist and shear are close to 0°and little is PROCESSING gained from decomposition. The azimuth retrieved via decomposition also exhibits a clear frequency dependence (Fig. 6), Three approximately linear profiles-one east-west and two north-south oriented-were sampled at intervals averaging and, following decomposition, the long-period phases remain split by 10-20°. Thus, an inductive rather than purely galvanic 1 km during the summer months of 1991 and 1992 ( Fig. 1). Audiomagnetotelluric (AMT) data, spanning a period range source for the higher conductivities imaged in the north-south polarization compared with the east-west polarization is of #10−2-102 s were recorded at 26 stations using the SPAM (Short Period Automatic Magnetotelluric) system, with remote implied. At longer periods (30-3000 s), the azimuth lies between reference stations, consisting of two (north-south and east-west oriented) horizontal induction coils established approximately N20°E and N20°W, or equivalently N110°E and N70°E. The Weardale granite strikes WNW-ENE, parallel to the Iapetus 200-500 m away from the local configurations. Long-period magnetotelluric soundings, spanning time periods of approxi-Suture (e.g. Klemperer & Matthews 1987) and the regional geological strike is in the range N90°E to N70°E. The survey mately 40-104 s, were recorded at 11 (approximately alternate) stations, using NERC geologgers and EDA fluxgates. Electric area lies approximately 45 km to the west of and 90 km to the east of broadly north-south-trending coastlines. At least two components were measured using non-polarizable, lead/lead-  modes of induction-one arising from the 'local', geological structure, the other due to regional inductive processes in the surrounding seas and underlying, conductive, North Sea graben sediments-can therefore be anticipated to superimpose, generating 3-D, frequency-dependent distortion effects. The significance of these effects is considered in a later section. DC shifts in the apparent resistivities at neighbouring sites evincing similar phases indicate the presence of static-shift effects. Static shift in MT data is analogous to the static time delays encountered in reflection seismology, but has more dire consequences on model predictions because of the multiplicative rather than additive nature of the distortion which arises when resistivity response functions are translated into resistivity-depth models. To date, numerous methods for correcting for static-shift effects have been forwarded. None has proven to be wholly satisfactory. In the following discussion we will examine static-shift effects using a priori data of different scalelengths (approximately 280 m, 400 m and 11 km). Possible influences due to the finite, 3-D extent of the granite batholith are examined in the next section.
Remodelling of Schlumberger spread, DC resistivity data taken from Parr (1991) using an inversion algorithm based on the Hankel transformation produced the three-layer model shown in Fig. 7 with rms misfits of around 8 per cent. Investigation of resolution depths indicates that the DC data are unable to resolve the thickness of the third layer or the underlying resistivity structure. A blind spot therefore remains between the bottom of the three-layer model and the minimum penetration depth of the MT data. The layer resistivities for the three-layer model agree well with the resistivity log from the Rookhope borehole 1500 m away from the centre of the layer underlain by sediments of average resistivity 160 V m to L ow conductance versus high v P /v S ratios below the Weardale granite 425 conductors and average over larger volumes compared to borehole resistivity logs. Our nearest MT site lies approximately 2 km from the Rookhope borehole and approximately 1 km from the centre of the DC spread. The static-shift distribution function obtained shows a positive shift towards more resistive values. The correction factors make no allowance for changes in the resistivities of the sedimentary sequence as we move away from Rookhope (to a maximum distance of 12.5 km from the borehole), although no spatial correlation between higher static-shift factors and distance from Rookhope is apparent (Fig. 8).
The resistive, crystalline granite is intruded into Palaeozoic slates. Gravity data indicate that the batholith must extend to a depth of at least 8 km, and below our survey area the base has been modelled at 11.5 km (Evans et al. 1988). A zone of weak seismic reflections has also been imaged at 12 km along an E-W profile, although it is unclear whether this should be attributed to the local transition from relatively transparent granite to more reflective host rock and thus treated as a marker for the bottom surface of the granite, or to free fluids associated with the regional, brittle-ductile, transition zone. A conductive 'layer', the two polarization modes are similar apart from static shifts (Fig. 3), and a set of static shifts computed one-dimensionally by equating the depths (Schmucker 1970) a depth of 210 m and a more resistive limestone sequence (580 V m) to a depth of 280 m. With the depth to the third, to the onset of decreasing resistivities obtained at each site to the depth of 11.5 km spans a similar range (0.2-7.4) and mean resistive layer fixed to 210 m, the rms misfit to the DC data is 8.8 per cent. The good agreement between the DC data and value (2.6) to that relying on near-surface observations. borehole data suggests that the uppermost sedimentary sequence has fairly uniform layer resistivities, with an average The intention of this study is to ascertain how significantly 3-D structures might distort estimations of conductance and resistivity log is plotted on a linear scale, below 270 m a log scale is used. The borehole intersected a number of conductive, mineralized veins, and accurate estimation of the low resistivities between 280 and 300 m is particularly difficult. The high MT phases (>45°) for the highest sounding frequencies imply a transition from a relatively more resistive layer to a relatively more conductive layer. This is consistent with the assumption that the onset of MT sounding lies in the sandstone/mudstone sequence from 280 to 400 m. With the apparent resistivity of the uppermost 280 m constrained to 200 V m by both DC data and the linear borehole resistivity log, the MT phases require an underlying conductance of the order 1.5 S (7 Vm×10.5 m) to be imaged at the highest frequency. This is not incompatible with the borehole log, which indicates depressed resistivities from 280 to 300 m depth, and implies a high-frequency (128 Hz) apparent resistivity level of the order of 100 V m. This yields static-shift factors in the range 0.4-5.3, with a mean value of 1.9 for MT soundings at all sites (except for two having poor coherency around 100 Hz and higher in the north-south polarization and one having poor-quality data in the east-west polarization). Below 300 m depth, the borehole log suggests integrated resistivities of the order 100 Vm for the remaining sedimentary sequence. The top of the Weardale granite, with resistivities of 1-2 kV m, is reached at 390 m. This transition to a more resistive layer is also consistent with the phase data. EM induction data are less sensitive to resistors underlying depth to a postulated conductor. The data set consists of sites real data in this period range and implies that inductive 3-D distortion effects from the shallow seas and underlying sedi-only on the granite. Thus, 3-D (Mackie et al. 1993) modelling was performed to investigate the distorting effects of the granite mentary basins play a prominent role in determining the responses throughout this 100-1000 s period band. Given that batholith and surrounding shallow seas compared to a 1-D model. The 1-D model consists of 500 m of surface sediments 3-D induction effects are clearly significant over a broad period (100 V m), granite to 10 km (1000 V m), a deep crustal layer range, rendering TE/TM mode terminology inapplicable of conductance 200 S (100 Vm×10 km) and a relatively more (although the predominant geoelectric strikes are broadly resistive mantle (300 V m). Two types of 3-D models are coneast-west and north-south), we continue to label our two sidered. In the first 3-D model we consider only the effect of polarization modes simply as 'north-south polarization' and the granite batholith embedded in the surrounding sedimentary 'east-west polarization' throughout the paper. basins (100 V m×2 km thick). The sedimentary basins are

MODEL STUDY OF THE DISTORTION
The net effect of the distortion effects investigated is that underlain by metamorphic basement rocks with resistivities of the north-south polarization, if considered in a 1-D sense, the order of 500 V m (e.g. Beamish 1986;Banks et al. 1996).
grossly under-resolves the conductance of the deep crustal The geoelectric structure directly below the survey area is conductor, due to the bias exerted by the highly conductive assigned according to the 1-D model. A deep crustal conductor seas in the relevant period range. Depth estimates to the top has been modelled in a number of studies conducted over the of the conductive layer for the north-south polarization are Northumberland Trough (e.g. Beamish 1986; Harinarayana furthermore biased downwards (Fig. 10), due to 3-D surface et al. 1993; Parr & Hutton 1993;Banks et al. 1996), and this effects. Qualitatively, this may explain the apparent deepening feature of the 1-D model is therefore extended into the deep of the deep crustal conductive layer below the granite batholith crust underlying the basins. In the second 3-D model, northcompared with below the Northumberland Trough in the south-trending shallow seas (50-70 m; 0.25 V m) and underregional model of Banks et al. (1996), derived from 2-D, lying, thick, sedimentary basins (to 5 km) are introduced 45 km rapid relaxation inversion (Smith & Booker 1991) of both to the east of and 90 km to the west of the centre of the station polarization modes. In the absence of additional surficial network. The second 3-D model is otherwise identical to heterogeneities giving rise to static-shift effects, the east-west the first.
polarization should recover the correct depth to the conductor Fig. 9 compares the apparent resistivity and phase responses (Fig. 10), whilst the recovered conductance (approximately at the centre of our station network for the 1-D and 3-D 150 S) is also of the correct order of magnitude compared with models described. Gravity modelling indicates that the contacts the input model. between crystalline and sedimentary rock are more gentle, and the 3-D model with its steep contacts may thus exaggerate the contrast between the granite and sediments. Considering its 6 CONDUCTIVITY, REFLECTIVITY AND simplicity, the granite-plus-seas model reproduces charac-BRITTLE-DUCTILE TRANSITION teristics such as the high mid-band phases, mode splitting and TEMPERATURES: COMPARISON OF general shape reasonably well (compare Fig. 3). DEPTHS For the 0.1-10 s period band, both 3-D models produce All P-wave and S-wave seismic sections recorded over similar responses, and the resistivity and phases for the eastthe Weardale granite indicate a band of bright, laminated west polarization mimic those for the 1-D model. Between 0.1 reflections spanning two-way-traveltimes (TWT) of 7.5-10 s and 1 s, the phase for the north-south polarization is down-(P-wave) and 13-17 s (S-wave). Ward et al. (1992) have shown biased by up to 5°relative to the 1-D model, but converges that these traveltimes translate to lower crustal depths, with again at around 10 s, leaving the north-south resistivities the onset of laminations at 24±2 km and termination at the upward-biased ( by a factor of about 1.4 at 5-10 s). We infer Moho at 30±2 km. A mid-crustal zone of weaker-amplitude that the period band from 1 to 10 s probably marks the seismic reflections was also imaged at a depth of 12 km along transition from 3-D inductive to galvanic distortion due to the an east-west profile. granite batholith impinging on the sedimentary overburden in Simpson (1994) presented 2-D models resulting from rapid the surrounding basins. Above 10 s, the responses for the two relaxation inversion (Smith & Booker 1991) of (decomposed) respective 3-D models diverge, such that from about 50 s interpolated apparent resistivities and phases of both polarization inductive effects from the ENE-WNW-trending geological modes, with rms misfits of 1.3 or less. Profiles were modelled structure apparently compete with those from the approxiboth independently and dependently (at the crossing points) mately N-S-trending shallow seas. Qualitatively this explains to obtain relative static-shift factors via a statistical technique the failure of the decomposition hypothesis in this period of smoothing deeper horizontal gradients present in the models, range. For the 100-1000 s period band, north-south and whilst allowing greater near-surface structural complexity. The east-west polarization resistivities for the granite-only 3-D smooth, least-structure models displayed a broad spatial model lie parallel to but above the 1-D response. North-south agreement between an upper crustal resistive zone (extending and east-west polarization phases are equal, but lie 6°above to depths of 8-12 km), interpreted to correspond to 'wet' the 1-D phases. Meanwhile, for the model which includes granite, and the gravity delineation of the granite batholith. the effects of the shallow seas, the resistivities for the two The resistivity values for the sedimentary overburden were polarizations are divergent, with the east-west polarization consistent with those logged in the Rookhope borehole and resistivities biased up relative to both the 1-D model and those for the granite batholith (order of 3000 V m) were granite-only 3D model. North-south polarization phases are consistent with those expected from the Rookhope borehole biased upwards, whereas east-west polarization phases are and with 'wet' granite in general. All models indicated the biased strongly downwards. The resultant phase splitting is approximately 14°, which approaches that exhibited by the onset of high conductivities within the mid crust (12±4 km).
L ow conductance versus high v P /v S ratios below the Weardale granite 427  Fig. 11 shows data and model fits to the (decomposed) north-the top of the seismic layering (22 km), demonstrating that, irrespective of whether the conductive zone is considered as a south polarization (considering both methods of static-shift correction) for a layered as opposed to a smoothly varying sub-smooth transition or as a layer boundary, and in spite of a possible down-bias to greater depth due to the 3-D effects strata and compares the fit obtained if the top of the conductor is placed to coincide with the shallowest estimated depth to described in the previous section, it is unequivocal that the sediments; Dunham et al. 1965) and a recorded surface heat flow value of 95 mW m−2 (Evans et al. 1988). The uncertainties associated with such temperature-depth calculations increase with depth. Temperatures in the brittle-ductile range (350-500°C) should be reached at depths of 12±0.3 km, and extend to 18-20 km. The onset of enhanced conductivities and a discrete zone of weak, mid-crustal seismic reflections distinct from the bright, lower crustal seismic lamellae could therefore lie in the brittle-ductile transition depth range. Greenschist metamorphic facies are to be expected at these depths (Hyndman & Shearer 1989), and such metamorphic facies can be expected to maintain equilibrium with free fluids . The lower crustal zone of seismic layering is certainly deeper and could lie at the transition to amphibolite facies (565±25°C at 22 km depth).

CONDUCTANCE, V P / V S RATIOS AND FLUID REQUIREMENTS
Lower crustal v P /v S ratios for the region below the Weardale granite have been calculated as 1.86±0.05 (Ward et al. 1992). v P /v S ratios for a band of weak reflections imaged in the mid crust along an EW profile are indeterminable since the reflections are not well-resolved in the S-wave sections. V P /v S ratios are rather more critically dependent on the distribution of porosity (pore aspect ratios) rather than the percentage porosity itself. Numerous authors (e.g. O'Connell & Budiansky 1974;Toksoz et al. 1976;Bruner 1976;Hashin 1988;Hudson 1981) have formulated mathematical models describing the effects of fluid-filled pores on the elastic properties of rock matrices. The model presented in Fig. 12 (derived from Hyndman & Shearer 1989) is representative. Assuming textural equilibrium porosities of at least 6 per cent vol. for the v P /v S ratios to be explained by fluids, whilst higher aspect ratios are incapable of explaining such high v P /v S ratios through fluids. Outside depth to the onset of enhanced resistivities lies at mid-crustal depths rather than in the lower crust, and thus shallower than the range suggested by Hyndman and Shearer, lower aspect ratios of 0.01 can explain the high v P /v S ratios with only 1.5 the top of the seismic layering. Possible rotation errors add uncertainties of less than 1 km. The slight discrepancy in the per cent fluids. The equivalence of thin, highly conducting layers and thick otherwise good agreement between the two static-shift correction techniques may be attributable to the 3-D effects discussed moderately conducting layers is well known in terms of the resolution potential of the MT method (e.g. Jones 1987). previously (compare with Fig. 10), with the technique relying on near-surface data leaving the conductive layer down-biased However, the total conductance of an anomaly may be better resolved (e.g. Schmucker 1970). Our 3-D model study indicates in the north-south polarization. In both 1-D and 2-D model studies, the bottom of the conductor is ill-resolved. Enhanced that the east-west polarization MT data should correctly resolve the deep crustal conductance, whilst the strongest conductivities may therefore extend into and through the zone of seismic layering imaged from a depth of 24±2 km. The reflectivity is also imaged along an east-west strike. For both methods of static-shift correction, the east-west polarization onset depth is less well-resolved by the east-west polarization of the MT data.
MT data yield conductances in the range 100-200 S and a mean value of 150 S, considering all nine MT sites having good-Assuming that the enhancement in surface heat flow over the granite is attributable to enhanced concentrations of radio-quality data in the appropriate period range. The north-south polarization resolves a conductance of order only 20 S. The genic elements within the granite, and that the respective contributions of upper crustal and deeper heat production division of conductance resolved by the two polarizations is in good agreement with our 3-D model calculations with 200 S distribute as described by Pollack & Chapman (1977) and Vitorello & Pollack (1980), a temperature-depth profile was in the deep crust, and implies that the integrated conductance of the mid and lower crust combined does not exceed 200 S. calculated (Chapman 1986) from the thermal conductivities logged in the Rookhope borehole (3.1 W m−1 K−1 for the granite Resistivity-porosity models are most sensitive to parameters concerning fluid-rock geometry (i.e. connectivity) and fluid and 2.2 W m−1 K−1 for the 400 m of overlying carboniferous L ow conductance versus high v P /v S ratios below the Weardale granite 429 Figure 11. 1-D layered models and data fits for north-south polarization of a typical site: open squares, r*-z* data points with static-shift effects corrected using near-surface (DC/borehole/MT phase) observations; closed squares, r*-z* data points with static-shift effects corrected by considering the depth to the bottom of the granite as inferred from gravity modelling; solid lines, mid-crustal layered model and fit; dotted lines, layered model and fit with deep crustal layer perturbed to coincide with seismic layering. resistivity. Fluid resistivities are in turn sensitive to salinity, pressure and temperature. Nesbitt (1993) considers laboratory data of high-salinity brines, and presents resistivity-depth profiles for several KCl (shown to produce similar results to NaCl) equivalent weight percentage concentrations, assuming a geothermal gradient of 30°C km−1 and lithostatic pressures. The overall decrease in resistivity accompanying an increase in temperature from 20°C to temperatures in the brittle-ductile transition range is found to be of order 3.5.
Considering the origins of pore fluids and metamorphic equilibria conditions, Hyndman & Shearer (1989) infer deep crustal pore fluid salinities comparable to (or possibly greater than) sea water (0.5 M), for which, at mid to lower crustal temperatures and assuming lithostatic pore pressures, data from Quist &Marshall (1968) andNesbitt (1993) suggest resistivities of the order 0.04-0.02 V m. Fluids pumped from the KTB deep borehole in Bavaria down to a depth of 8.9 km have salinities of 68 g l−1 NaCl (1.7 M), corresponding to an approximate resistivity of 0.1 V m at 20°C (Huenges et al. 1997). A fluid conductivity of 25 S is therefore a justifiable lower limit to use porosity models. Limiting pore geometries are characterized by isolated spherical pores at one extreme and perfectly interconnected thin films at the other. Higher and lower degrees If due to fluids distributed throughout mid and lower crustal depths, then, with the mid to lower crustal pore geometries of pore interconnection have been parametrized by smaller and larger Archie's Law exponents: 1 for interconnected low-discussed, a conductance of 200 S can be explained by less than 1 per cent vol. fluids. Even squashed into a 6 km thick aspect-ratio cracks; 2 for poorly connected high-aspect-ratio cracks (e.g. Evans 1982). Laboratory data, measured under lower crustal layer (24-30 km) a conductance of 200 S requires no more than 1.2 per cent vol. fluids with an Archie's Law conditions approaching those thought to be representative of the mid crust, suggest the likelihood of intermediate geometries, exponent of 1.5 or 0.2 per cent for complete fluid interconnection. Conversely, a 6 km thick lower crustal layer con-involving pore interconnection via grain boundary tubes (i.e. intermediate between grain boundary films and isolated pores) taining 6 per cent vol. fluids implies a conductance of the order of 6000 S for inferred textural equilibrium geometries or 2200 S and characterized by an Archie's law exponent of 1.5. (e.g. Lee et al. 1983). Theoretical (von Bargen & Waff 1986) and according to Archie's Law with an exponent of 1.5. Even with only 55 per cent liquid bridging the Waff model predicts no empirical (Watson & Brenan 1987) analyses of the wetting or dihedral angles of polyphase, fluid-rock systems suggest that more than 2 per cent vol. fluids to explain a 6 km thick, 200 S layer or conversely a conductance of 600 S in the presence of lower crustal rocks and saline fluids are in textural equilibrium, with dihedral angles below 60°and wetting by grain boundary 6 per cent fluids. Fig. 14 shows the effect on our 3-D model calculations of films above critical porosities of less than 1 per cent. Watson and Brenan's results (1987) indicate that the addition of replacing the 200 S deep crustal layer with a layer of conductance 1000 S. With this model, the phase splitting evinced CO 2 to pure water generally increases the dihedral angle, although no data are presented for the combined case of by the real data is no longer reproduced in the modelled data, the phases descend to longer periods, the shape of the resistivity H 2 O+chloride+CO 2 . The experiments were conducted at temperatures more representative of mantle temperatures.
curves is clearly significantly different, and both polarizations resolve a conductance of order 700 S. Deep crustal conductances However, it is suggested that a strong temperature dependence is not to be expected because the controlling factor is the in excess of 1000 S can therefore be ruled out. ratio of surface free energies in the solid-solid/solid-fluid system. Conclusive laboratory data to clarify this point do 8 CONCLUSIONS not exist. Fig. 13 shows the theoretical conductivity of a two-phase We have presented results from the first integrated interpretation of MT data, P-and S-wave seismic reflection pro-system parametrized according to Hermance's (1979) modified form of Archie's law (with exponents of 1, 1.5 and 2), Waff 's filing. By attempting to place constraints on the distribution of lower crustal fluids beneath the Weardale granite in NE (1974) model for thin, interconnected grain boundary films (assuming 100, 75, 65 and 55 per cent liquid bridging), the England we have addressed one of the most important enigmas relating to the deep crustal environment. Papers in which Hashin-Shtrikman (1962) upper bound (equivalent to isolated pores), and the Hashin-Shtrikman lower bound (approxi-deep crustal zones of conductivity and reflectivity are jointly attributed to fluids abound. A general problem with the mately equivalent to Waff 's formulation with complete liquid bridging). The resistivity of the rock matrix is taken as 105 V m. fluids paradigm has been the dilemma posed between the low L ow conductance versus high v P /v S ratios below the Weardale granite 431 permeabilities required to maintain fluids in the lower crust ACKNOWLEDGMENTS and the pore interconnection required to create low resistivities.
The research described in this paper was made possible Our results may help to resolve this paradox.
by Natural Environment Research Council research grant Our research challenges the widespread and little-tested number GR3/7817. Fieldwork equipment and technical support assumption that deep crustal conductive 'layers' and lower were provided by the NERC Geophysical Equipment Pool, crustal seismic layering are concomitant. We have demonstrated Edinburgh. We are grateful to the farmers, gamekeepers and that below the Weardale granite enhanced conductivities are landowners who granted permission for our measurements, supported at mid-crustal depths, corresponding to a probable to Kenny MacDonald for help with fieldwork and to Dean minimum temperature of 350°C and therefore greenschist Livelybrooks for help with fieldwork and programming whilst metamorphic facies conditions (Hyndman & Shearer 1989), at the University of Edinburgh. F. Simpson acknowledges which are expected to maintain chemical equilibrium with free support from a NERC studentship whilst at Imperial College. water (e.g. . A band of weak seismic reflections John Booker, Gary Egbert, Ross Groom, Randy Mackie, is also imaged in this temperature-depth range, but only along Torquil Smith, Hermann Hamel and Matthias Seichter are the profile running parallel to geological strike (east-west).
thanked for making their data processing, decomposition and The zone of highly reflective lamellae is significantly deeper, modelling codes available. Constructive criticism of an earlier possibly at the transition from greenschist to amphibolite facies draft of this paper by Pascal Tarits and an anonymous reviewer (565±25°C at 22 km). The base of the deep crustal conductor have helped to improve its content. is ill-constrained, and slightly enhanced conductivities may thus pervade the zone of lower crustal seismic layering. However, the maximum inferred conductance (conductivity-thickness REFERENCES product) of the deep crust of less than 200 S for the east-west polarization mode of the MT data requires significantly less Á dám, A., 1978. Geothermal effects in the formation of electrically free fluid than do the high v P /v S ratios to be explained entirely conducting zones and temperature distribution in the earth, Phys.
by saline fluids, even with a relatively low conductivity of Earth planet Inter., 17, 21-28. 25 S m−1. [Hyndman & Shearer (1989)  regarding the applicability of lower crustal textural equilibrium, L ow conductance versus high v P /v S ratios below the Weardale granite 433