1. Introduction
I hope to speak for most of the electromagnetic (EM) induction community in thanking Yardley and Valley [1997] for their interest in clarifying the conditions of existence of fluids and other grain boundary phases which can enhance electrical conductivity of the deep crust. The population of solid-earth EM researchers is relatively small, and we are fully engaged by the formidable problems of data measurement and processing, and of response modeling and inversion, in addition to the relation of conductivity to physical state in the Earth. Outside contributions are gratefully appreciated. My comment is somewhat a personal view of the role of conductivity models in Earth science and an attempt to clarify and update developments in our field. While Yardley and Valley recognize that the seismic community has been divided over adoption of free fluids as an explanation for low-velocity anomalies, there seems to be an assumption that EM researchers are more of one mind on the causes of conductivity. This certainly is not the case and our understanding has advanced beyond some of the early work relied upon by the authors. Discussion here centers around the following specific points. First, the notion that a fixed isotherm in the Earth (i.e., 400oC) represents the top of the conductivity increase in the lower crust was a useful exercise in thought but now clearly is too simplistic. Probable temperatures for the entirety of conductivity models are far from uniform and this should be considered when advocating conductivity mechanisms. Second, the conductivity structure of older, stable regimes can be highly variable laterally even though temperatures appear quite low. Hence, the conductivity structure is difficult to explain by a simple, thermally activated mechanism and graphitic contributions are still attractive. Finally, deep conductive zones in regions long stable are largely, perhaps mostly, in rocks of brittle or semi-brittle rheology and appropriate mechanisms deserve greater consideration.
2. Background to Deep Crustal Conductivity Research
Prior to the 1980's, EM technique researchers tried to infer temperatures deep in the crust by comparing Earth conductivity profiles derived from field EM measurements (usually MT) to laboratory measurements of the solid-state conductivity of dry, representative crustal minerals [Brace, 1971]. As field models accumulated, however, temperatures so implied appeared systematically higher (by 100% or more) than those predicted by geotherms derived from essentially coincident heat flow data [Shankland and Ander, 1983; Haak and Hutton, 1986]. It was concluded that solid-state (semi-conduction) mechanisms of conductivity in dry mineralogy are not completely representative of the deep crust, and that an additional, mineral grain boundary phase of high conductivity is necessary. Due to its common occurence and property as a universal solvent, the conductive boundary phase often has been interpreted as free water [Shankland and Ander, 1983; Haak and Hutton, 1986; Hyndman and Shearer, 1989], a principal topic of this discussion. Even in those early compilations, a distinction was drawn between the lower crusts of Precambrian areas, of stable Phanerozoic areas, and of recently active areas. As noted by Yardley and Valley, attempts have been made to correlate tops of the conductors with an isotherm in the crust, where temperatures around 400oC typically are proposed [e.g., Hyndman et al., 1993]. This correlation is rather tenuous, especially for terrain ages greater than Mesozoic. In fact, the depth to top of the perceived deep conductors below Precambrian areas fully spanned the range 15 to 45 km but with heat flow varying only between 40 to 50 mWm-2. Before proceding, we must understand what aspects of subsurface conductivity structure can be resolved from surface MT measurements. Because EM waves in the Earth propagate diffusively, and because we can have only a finite number of data which contain some error, MT measurements alone cannot resolve sharp structural boundaries or other fine-scale features [Constable et al., 1987; Smith and Booker, 1988]. Related to this, the resolution of MT measurements decreases with depth of investigation in accord with EM similitude or scaling [Wannamaker et al., 1984]. Spatial variations in conductivity (i.e., porosity changes, faults, graphite or sulfide zones) hence may be resolvable as such from a well-sampled MT response if they lie near-surface. However, ensembles of such structures, if buried more than several times their individual characteristic dimensions, will instead have an MT response equivalent within data error to some bulk conductivity average intermediate to the conductive individuals and their resistive host [Madden, 1976]. As a common example, the conductivity and thickness of a buried conductive layer are highly correlated parameters in a structural estimation, such that only their product (conductance) often is resolvable [Jones, 1992]. Moreover, multiple thin layers may have the combined response of an equivalent thicker one. In the face of this resolution limitation, we may solve for optimal values of a few carefully chosen resistivity parameters [e.g., Petrick et al., 1977; Larsen, 1981] or we may estimate a smooth or "minimum" structure representation below the measurement site [DeGroot-Hedlin and Constable, 1990; Smith and Booker, 1991; Mackie and Madden, 1993]. The averaging nature of EM measurements makes us consider networks of conducting elements of a scale much larger than represented in hand specimens or even many outcrops. Finally, the EM community has become increasingly sensitive to the effects of lateral heterogeneity on estimates of conductivity below a given survey area. It is impractical to thoroughly reanalyse all MT survey data in the compilations by Shankland and Ander [1983] and Haak and Hutton [1986], but there have been repeated warnings in the past about inadequate regard for effects of upper crustal inhomogeneity in the data when producing lower crustal models, starting with Porath [1971]. Early interpretation approaches assuming locally 1-D or oversimplified 2-D geometries were argued to be prone to bias errors, not merely model scatter, giving conductor depths too shallow and conductivities too high [e.g., Wannamaker et al., 1984; Wannamaker and Hohmann, 1991]. Consequently, inferred temperatures for the deep conductors commonly were too low. This was part of a vigorous debate in the community during the 1980's spurred in part by the papers of Shankland and Ander [1983] and Haak and Hutton [1986] whose conclusions about the degree of lower crustal conductivity were not universally accepted. Improved accounting for heterogeneous effects has not removed the need for additional conducting phases in the lower crust in our models, but it has substantially refined the character of our models with attendant implications for physico-chemical state.
3. Crustal Compositions and Fluid Stability
Unlike five to ten years ago perhaps, I believe that most in the EM community today recognize that nearly pure water cannot exist stably in ductile lower crust with high metamorphic grade mineralogy preserved to temperatures lower than preceding peak conditions. My own analysis of this appears in Wannamaker [1986], about the same time as the oft-cited Yardley [1986], and was influenced greatly by Wayne Burnham's work [Burnham, 1979a,b]. Modified in Figure 1 here is my earlier depiction of melting and fluid regimes in a ductile, intermediate composition, metaigneous lithology with limited H2O-CO2 component. Reactions depicted, which for natural compositions may be quite gradational, apply to rock samples at amphibolite-granulite facies, or at greenschist facies, depending on the volatile content of the sample. They are defined from diverse experimental results published by various authors rather than calculated from a uniform thermodynamic data base, but they should illustrate the basic principles considered here. These relations are relevant approximately for more mafic average compositions also [Rudnick and Fountain, 1995] because the presence of pyroxene and feldspar components is the most crucial consideration and the relative mineral abundances mainly determine the total buffering capacity of the rock to absorb or exsolve fluids. I consider conduction in brittle rocks at a later point. In the absence of CO2, breakdown curves in Figure 1 for hydrate minerals (i.e., chlorite-out or amphibole-out) at temperatures above the H2O-saturated solidus mark the production of water- undersaturated felsic magmas via vapor-absent melting for sufficient increases in temperature. No H<SUB>2</SUB>O H<SUB>2</SUB>O fluid is produced but instead the H<SUB>2</SUB>O is taken up directly by the melt whose a( H<SUB>2</SUB>O) is significantly less than 1. In particular, for amphibole-present melting in high metamorphic grade rocks, a( H<SUB>2</SUB>O) ÷ .2-.25. Conversely, cooling the rock to temperatures well below the breakdown curves would yield a( H<SUB>2</SUB>O) which should be very small indeed, as computed by Yardley and Valley [1997], and pore space in ductile rocks would have collapsed to a vanishingly small volume. There is no argument. A CO2 component to volatile content was discussed also by Wannamaker [1986] because this component had been used extensively in classic experiments in the 1970's as a relatively inert agent to reduce H<SUB>2</SUB>O activity and study hydrate mineral breakdown reactions and water-undersaturated crustal melting, and because it is inferred to be a frequent volatile component of the lower crust [e.g., Frost et al., 1989a; Wickham, 1992]. Dashed curves I in Figure 1 approximate first appearance of H<SUB>2</SUB>O-CO<SUB>2</SUB> fluids (Pf = Pt) with increasing T in the assumed deep crustal composition at medium and high metamorphic grades. The curves I are traces over a range of pressures of invariant points at constant pressure representing joint breakdown of carbonate and hydrate minerals [see Figure 1 of Wannamaker, 1986]. The carbonate-out reaction for curve Iam specifically produces diopside (reaction 12 of Yardley and Valley, their only one involving just CO<SUB>2</SUB> as the volatile phase) while that for Ichl produces anorthite component of oligoclase assuming pyroxene has been depleted [Jacobs and Kerrick, 1981; Bucher and Frey, 1994, p. 243, 276]. The reduction in temperature of first appearance of H<SUB>2</SUB>O-CO<SUB>2</SUB> fluid with reduced metamorphic grade is less than depicted in Wannamaker [1986], which is a correction here. Nevertheless, in principle for granulite facies mineralogy, Iam is considered pertinent because it seems rare that lower crustal rocks contain absolutely no amphibole [Burnham, 1979b]. The reaction indicates free fluid (CO<SUB>2</SUB>-rich) to be stable only for T > 650-700oC for Pt = 8 kbar and somewhat less at lower pressures. I never stated that such fluids would be stable at 400oC in ductile (Pf = Pt), granulite facies deep crustal rocks. In my original version of Figure 1, they would barely appear even in greenschist rocks at 400oC at 5 kbar (Ichl), the current upward revision in temperatures notwithstanding. I believe it would be very useful to the EM community if the authors would offer a modified version of Figure 1 here incorporating the latest, internally consistent thermodynamic data, including volatile consuming reactions other than mine if they are more pertinent. This would provide us with a petrogenetic grid of first appearance of H<SUB>2</SUB>O-CO<SUB>2</SUB> fluid to bound conditions under which such fluid could explain conductivity anomalies. I concur that even a saturated NaCl- H<SUB>2</SUB>O solution fluid likely would be absorbed at lithostatic pressure and 400oC (or even higher T) in granulite mineralogy of the lower crust. Presumably this would happen with halite exsolving as H<SUB>2</SUB>O is taken up [Sanders, 1991], or with increased uptake of chloride and cations directly by the hydrate mineral until all fluid is gone [Frost and Bucher, 1994; Kullerud, 1995]. According to Aranovich and Newton [1996], for saturated NaCl- H<SUB>2</SUB>O solutions, a( H<SUB>2</SUB>O) ÷ X( H<SUB>2</SUB>O) (mole fraction) ÷ 0.5 at Pt = 5 kbar and even less at higher pressures, which is lower than used by Yardley and Valley [1997]. Beyond just NaCl, each additional component introduced to the system (e.g., Fe, K, Ba, Fl, S, P) should lower the temperature of first fluid appearance yet further [Aranovich and Newton, 1996]. Some of these are considered minor components of the crust but, as I will discuss, we only need fractions of a percent by volume to explain most crustal anomalies in fluid terms. Until the thermodynamic data are available and the calculations done, we will never know precisely the minimum possible temperature of an arbitrary fluid as a function of metamorphic grade and pressure. Nevertheless, decreasing a( H<SUB>2</SUB>O) by adding other components has finite effectiveness and the stable existence of fluid below a certain temperature will require a reduced metamorphic grade. In a brittle rheology, fluid-based conduction differs from the ductile regime.
Over time, we have changed from assuming that fractures and intergranular porosity affect geophysical properties only in the upper 2-3 km [e.g., Brace, 1971; Feves et al., 1977]. Now, hydrologic studies have convinced many that fracture permeability increases markedly with scale of measurement perhaps up to 10 km or more [Brace, 1980, 1984], which is the resolving scale for MT in the lower crust. Deep drilling results from the FSU [Kremenetsky and Ovchinnikov, 1986; Borevsky et al., 1995] and most recently from the German KTB [Emmermann and Lauterjung, 1997, and companion papers) have shown frequent fluid entries under nearly hydrostatic pressure over the full extent of the drilling depth (up to 12 km). At the Saatly hole, Borevsky et al. [1995] encountered large-scale downward fluid flow in fractures continuing beyond corehole total depth which caused the lost circulation that stopped further drilling. We do not know yet the ultimate depth extent of fracture permeability, but earthquake focal depths can reach 40 km in stable areas [Smith et al., 1989; Slemmons et al., 1991]. Seismogenic faulting in turn may only be a shallow bound to brittle failure, implying the brittle regime may extend essentially through the crust in such terrains [Rutter and Brodie, 1992; Ranalli, 1997]. This is only a factor of 3-4 greater than the deepest drilling so far.
Can a decrease in fracture porosity with depth be more than offset by the apparent strong increase in fluid salinity [Emmermann and Lauterjung, 1997] and temperature [Nesbitt, 1993], and the increased scale of rock influencing the MT fields? I don't know, but we shall see later how little actual fluid content may be necessary to explain many conductivity models from stable crust.
4. Non-fluid Causes of Conductivity
Personally, I agree with Yardley and Valley [1997] on the finite role of graphite in lower crustal conductivity, although for mostly different reasons. However, I believe the case for its importance in many older terrains remains strong. The likelihood of graphite in a specific survey region will be a function of its possible provenance, which I divide into two fundamental types. The first concerns conditions under which graphite would precipitate from carbon-bearing fluids introduced to meta-igneous lower crust such as considered above for fluid stability. To illustrate, Figure 2 shows f(O2) (oxygen fugacity) trends exhibited by eruptive igneous products from a variety of settings in the western U.S. and compared to those from older regimes in eastern North America. Although there are exceptions, the igneous trends are interpreted to usually reflect the f(O2) of the lower crustal source regions [Carmichael, 1991]. The stability curve of graphite at 5 kbar pressure is superimposed. Heterogeneity in oxidation is obvious, but graphite does not appear stable in most modern, active situations at least for T > 400-500oC. Some convergence of igneous trends with graphite precipitation towards low temperatures is suggested [and see Frost et al., 1989b], so that it is more difficult to overrule graphite stability near the tops of many deep conductive layers. The ambiguity is exacerbated by the previously described correlation between layer conductivity and thickness in many MT models. If the two parameters cannot be resolved separately then we may have trouble disproving graphite stability in the anomalous domain. It is in the lower crust of Archean or Early Proterozoic cratons where the most reduced conditions are characteristic (as low as iron-wustite buffer) [Haggerty, 1990] and pervasive graphite stability through the base of the crust seems likely (Figure 2). Magmas derived from beneath such regions can reflect this [e.g., Frost and Frost, 1997]. We will discuss lower crustal models to show whether broad-scale graphite stability per se has exerted a major control on the conductivity for Archean terrains (apparently not!). Increased oxidation of the crust and upper mantle outboard of the cratons is attributed to plate tectonism, subduction and slab devolatilization at depth [Haggerty, 1990; Balhaus, 1993; Arculus, 1994; Brandon and Draper, 1996]. The second provenance for graphite of interest to conductivity studies is sedimentary deposition of sufficient biotic material and its subsequent metamorphism to at least greenschist facies. This possibility has been reviewed by Boerner et al. [1996] who demonstrate a strong correlation between specific conductivity anomalies and former clastic sediment-starved, foreland basins in Proterozoic accretionary terrains of Laurentian North America. Many other examples of pronounced, semi-continuous conductors beneath former collisional belts of a variety of ages can be found including Baltica, east-central Europe and the FSU [Hjelt and Korja, 1993], the Iapetus suture of Great Britain and Ireland [Brown and Whelan, 1995; Banks et al., 1996], southern Africa [Gough, 1989], the Denali system of Alaska [Stanley et al., 1990], and the Taconic suture of the southern Appalachians [Ogawa et al., 1996; Wannamaker et al., 1996]. In particular, the well-known study by Frost et al. [1989b] on the Laramie intrusive may be complicated as a case of magmatic graphite by organic contributions. That pluton intruded the Cheyenne suture zone at precisely the location where the North American Central Plains (NACP) anomaly (presumably due largely to biogenic graphite) intersects the suture from the north [Boerner et al., 1996]. Most of those conductive trends also were discovered originally by widely-spaced, reconnaissance arrays of magneto-variometers [Gough, 1989], and then received focused study with MT profiling. Because geophysicists commonly are drawn to study anomalous regions, one should be careful about compiling results from such studies and concluding that the models represent "typical" lower crust. The most conductive lower crustal structures from so-called stable regions reviewed by Shankland and Ander [1983], Haak and Hutton [1986] and Hyndman et al. [1993] appear to lie in such collisional zones. I believe we must distinguish such biogenic concentrations in deeply buried meta-sediments, from the concept of regionally interconnected graphite precipitated from magmatic or metamorphic fluids in the lower crust. In the former case, the rich graphite concentration controls the f(O2), while in the latter case the host mineralogy controls the f(O2) and dictates whether graphite is stable as a precipitate from fluid. I.e., what's the cart and what's the horse? Subsequent metamorphic events also may blur the distinction between the two origins. The vein-filling, formerly biogenic graphite documented by Rumble et al. [1986, 1989] may be a singularly spectacular example of remobilization during high grade metamorphism (over 3,000 km2). Nevertheless, graphite thus dispersed in deep Precambrian rocks likely enters an environment amenable to its stability, if not interconnection, as temperature falls again. Yardley and Valley's point about the tendency of graphite to form isolated crystals during high-grade metamorphism, a point which has been made before [e.g., Jodicke, 1992], is well-taken. Moreover, many ancient suture zones possess no discernable conductivity anomaly and many observed anomalies elsewhere are discontinuous along the orogenic belt [Jones, 1993; Korja and Koivukoski, 1994]. Such zones without anomalies may have been thoroughly overprinted in metamorphism, or else never possessed an extensive organic basin in the first place. Nevertheless, the correlation between former sites of collision and the strong, elongate conductors which are known is just too strong to be dismissed. In addition to Boerner's [1996] examples, graphite-bearing deep crustal xenoliths of appropriate textures have been recovered over the conductive Iapetus suture of Great Britain [Banks et al., 1996] and the Rhenish Massif [Haak and Hutton, 1986]. Carbon isotope studies of grain boundary graphite in the Kapuskasing uplift imply an ultimately biogenic origin with remobilization into metaigneous rocks [Mareschal et al., 1994]. Biogenic graphitic conductors up to tens of kilometers in length have been preserved through granulite facies metamorphism (6-9 kbars) in the Lapland Belt [Korja et al., 1996]. Results from the KTB deep drill hole [Emmermann and Lauterjung, 1997] show that conductive graphitic shear zones can be formed during retrograde metamorphism of high- grade rocks containing disconnected graphite crystals. Fluid- mobilized carbon species have precipitated graphite of similar textures elsewhere [Dissanayake, 1994; Naraoka et al., 1996]. I can only presume that some packages of organic rich metasediments escape high metamorphic grade conditions, or the metamorphism does not progress far enough to utterly destroy long-distance interconnectivity in all cases, or connection can be re-established during cooling and retrogression in fractures. I heartily agree with Yardley and Valley [1997] that many laboratory studies of conductivity of Earth materials have not been made under physico-chemical conditions comparable to those of the deep crust. Solid-state mechanisms in feldspars come to mind [Wannamaker, 1986]. Duba [1976] showed early on the importance of adequate equilibration times and defect populations on high- temperature mon-albite conductivity. Even intermediate and high albite are relatively defect rich over about 600oC [Smith, 1983], and K-feldspar over about 450oC [Goldsmith, 1988]. Agreeably, many diffusion and defect-related processes are enhanced by even fractional a( H<SUB>2</SUB>O). Solid-state feldspar conductivity should be measured for P( H<SUB>2</SUB>O) representative of hydrate minerals of the lower crust. Unfortunately, quoting Kronenberg et al. [1996], experiments so far "indicate that neither the concentrations nor the mobilities of alkali point defects responsible for alkali transport are altered by the substitution of alkalis by H<SUB>2</SUB>O". Hence, the similar H<SUB>2</SUB>O dissolution mechanism of Burnham [1979a] for feldspathic melts, whose pronounced effects on their rheology and conductivity are well-known [Burnham, 1979a; Wannamaker, 1986], does not appear effective in the solid-state. It is unclear also whether such mechanisms would procede very far in ordered feldspars at temperatures down to 400oC or lower, even at high pressures. Possibly, H+ defects in crustal minerals have a role [Kronenberg and Kirby, 1987], as has been suggested for mantle olivine. That dissociated H<SUB>2</SUB>O may enhance olivine conductivity was first noted by Duba and Heard [1980], amplified by Karato [1990], and is still being researched [Shankland and Duba, 1996]. 5. Deep Conductivity Structure versus Tectonic Regime In the past five to ten years, we have seen numerous crustal-scale EM surveys of much improved quality. They include more appropriate modeling assumptions and better incorporation of geologic constraints, which themselves have evolved. I offer example conductivity models from Archean, Proterozoic, and Late Cenozoic tectonic settings and my perception of what may define this physical property therein. Yardley and Valley profess no objection to fluid causes of conductivity in active regimes, but it is worth comparing active and shield areas as the latter are typically quite different and should be viewed in a different framework.
5.a Conductivity Structure in Active Regimes
In Figure 3 are plotted deep conductivity models from two MT transects in the Great Basin extensional environment of the western U. S. [Wannamaker et al., 1997a,b]. The two transects lie over crystalline deep crustal rocks which are relatively well studied and similar, and appear remote from known sutures and their geomagnetic anomalies. The more conductive model comes from the actively extending eastern subprovince of western Utah while the less anomalous one is from surveying in the Great Basin interior of northeastern Nevada. The latter region appears to have become much subdued in rate of extension for the past 5-10 Ma. The MT transects were interpreted in a manner optimal for extracting the dominant 1-D component of the lower crustal conductivity structure [Wannamaker et al., 1997b]. The deep conductivity variations are modeled by a few discrete layers, a parsimonious parameterization in terms of interval averages consistent with the resolving power of diffusive EM fields. Parameter standard deviations typically were only several percent due to high data quality and lateral sampling. The conductances of the active and subdued subprovinces are about 3,000 and 750 S respectively. For both areas it was possible to resolve conductivity of the lower crust independent of conductor thickness, which is unusual in MT surveys. Plotted also for the two areas are the geotherms estimated from surface heat flow and Curie depth [Wannamaker et al., 1997b]. Heat flow data almost always possess substantial scatter, but the estimated temperature of the top of both Great Basin conductors is about 550oC. The top of the high conductivity is correlated with high-salinity brines of minimal a( H<SUB>2</SUB>O) in amphibolite (not granulite) facies metaigneous rocks characteristic of the deep basement of both areas presuming that this facies is compatible with depths around 20 km [Christiansen and Mooney, 1995; Rudnick and Fountain, 1995]. At greater depth, water-deficient felsic melts or even mafic melts near 40 km depth in the east appear compatible with the lithology and the geotherms. A temperature of 550oC for the top of the conductor is substantially greater than that of 350-400oC suggested in past survey compilations and fluids at this temperature are less in conflict with experimental petrology and likely crustal composition here. It is consistent with the fluids lying below the brittle-ductile transition [Bailey, 1990], but in more applicable intermediate-mafic compositions [Rutter and Brodie, 1992; Ranalli, 1997]. In the upper hypersaline fluid regime of the conductors, we now understand that the expected salinities of tens of percent by weight yield high fluid conductivities, at least 100 S/m (Seimens/meter), although the values are not very sensitive to salinity beyond about 25 wt. % [Ucok, 1980; Nesbitt, 1993]. Hence, the model values of 0.13 and 0.05 S/m correspond to fluid contents of about 0.4 and 0.15 vol. % respectively assuming interconnection along grain edges [Watson and Brenan, 1987]. Fully saturated salt solutions might even be half that in volume. These values are considerably less than those stated typically in some interpretations and perceived as the norm by Yardley and Valley. They are in fact compatible with grain-edge fluid percolation models of ductile crustal rocks for currently active extension with magmatic fluid release, and extension which has been reduced since 5-10 Ma [Bailey, 1990; Frost and Bucher, 1994; Wannamaker et al., 1997b]. If the model is reasonable, expulsion time for most lower crustal fluid is short relative to the Phanerozoic time span, a concept recognized independently by many others. Conductor temperature may differ from area to area, however, and each survey needs careful estimates of the thermal regime, and crustal composition and metamorphic grade. As an example, Figure 3 contains a conductivity model from the active transpressional regime of the New Zealand Southern Alps [Stern et al., 1997; Davey et al., 1997]. The profile is from a site about 50 km southeast of the Alpine Fault trace and over the area of maximum Late Cenozoic crustal thickening. Numerical modeling assuming a Newtonian rheology implies downward material trajectories below the upper middle crust and depressed isotherms relative to far-field values [Allis and Shi, 1995]. The top of the most pronouced conductivity is only at 250-300oC. The base of the conductor (and the crust) near 40 km is only 550-600oC, which is about the temperature of the top of the Great Basin conductors. However, the lower temperature of the conductor top relative to the Great Basin is in keeping with the zeolite-lower greenschist facies and quartz-rich rheology of the thick South Island greywacke sequence. The translation of low- grade rocks to greater depth in the crust, where they undergo warming, effects prograde metamorphism and H<SUB>2</SUB>O-rich fluid release. Peak conductivities around 0.025 S/m are compatible with only about 0.1 % porosity. If interconnected porosity falls much below 0.1 %, simple percolation models alone predict residence times beyond 100 Ma even at 600oC [Frost and Bucher, 1994], which could correspond to a conductance of 100-200 S. This duration exceeds the thermal time constant of thick crust [Chapman and Furlong, 1992]. For example, heat flow in the Late Proterozoic Grenville and Paleozoic Appalachian belts is indistinguishable within scatter from that of the Canadian Archean [40-42 mWm-2; Costain et al., 1989; Guillou- Frottier et al., 1995], though on average Proterozoic terrain heat flow seems slightly higher than Archean (Nyblade and Pollack, 1993). In domains of activity older than the Mesozoic, we should expect that the brittle-ductile transition has moved downward approaching the base of the crust [Rutter and Brodie, 1992; Ranalli, 1997] and engulfed any pre-existing, residual fluid zone in formerly ductile rocks (geochemical stability and fluid residence times notwithstanding). The entire concept of a deep crustal conductor due to fluids in ductile rocks below a rheological transition probably has limited validity for the Paleozoic and older orogens.
5.b Lower Crustal Conductivity in Stable Environments
In Figure 3 also is plotted an interpreted conductivity profile from an Archean cratonic regime, in particular the Pontiac terrane of east-central Canada [Mareschal et al., 1995]. This example was chosen because the Canadian Shield is a large and coherent craton [Goodwin, 1996], the MT site density is greater than typical, modern methods of response distortion correction have been applied, and the deep crustal inferences are consistent over the survey area. Note that the maximum conductivity remains relatively small (0.002 S/m). At such lower conductances, typical of Archean interiors remote from suture zones [<50 S; Jones, 1992], it is more difficult to separate conductivity from thickness. Nevertheless, these limited conductivities for Archean interiors have been recognized for some time and were documented in the early compilations of Shankland and Ander [1983], Haak and Hutton [1986] and Hyndman and Shearer [1989]. Values of 0.02-0.1 S/m (resistivities of 10-50 ohm-m) are not representative of Archean regions unless all enhanced conductivity is lumped into a pathological thin layer due to limited data quality, or the terrain has been reworked tectonically. Unlike the examples from thermal regimes, the weak enhancement of deep crustal conductivity in Archean terrains probably lies primarily within the brittle rheological regime, in keeping with reduced shield geotherms and mafic rheologies (Figure 3). Leaving aside equilibrium fluid stability for an instant, the corresponding fracture porosity at high salinities [Fritz and Frape, 1987] would be nominally on the order of 0.01 % for Archean interiors, or about two orders of magnitude less than assumed by Yardley and Valley [1997]. While some of these networks may form during fluid migration closely following peak metamorphic conditions, seismicity and strain are ongoing phenomena and we can expect continued fracture formation as the brittle-ductile transition moves downward into the crust and stress fields change over time. Now, the previously discussed constraints on fluid stability in principle are independent of fluid volume so that miniscule porosity alone does not increase fluid probability. The tendency for fluid in the brittle regime to be hydrostatic in pressure and to be salt-saturated at depth could reduce a( H<SUB>2</SUB>O) to about 0.2 - not low enough by itself to stabilize fluid in granulite rocks at equilibrium [Yardley and Valley, 1997]. Still, one may expect some lateral permeability sealing to occur in the hydration process since the volume of hydrated minerals typically exceeds that of the unaltered silicate reactants, thereby shielding the remaining or replenished fluids from reaction. This is similar to the concept of Sanders [1991], but here I apply it to brittle rocks where the possibility of fracture porosity being supported by the strength of the rock is much more evident. Large blocks of amphibolite were traversed by the KTB hole but copious fluid entries to temperatures nearing 300oC occured through fault zones in the blocks. The possibility of disequilibrium, evidently, is a matter of scale [Bucher and Frey, 1994]. Moreover, while cation exchange capacity of chlorite is much less than other clays, alteration mineralogy may inhibit upward permeability and fluid egress but still allow substantial conduction [Batzle and Simmons, 1977]. It was within the above context that I previously have advocated rather more liberal conditions of fluid-based conduction in brittle shield rocks, although some related statements found by the authors in my unreviewed, open-file report [Wannamaker, 1994] were admittedly oversimplified. I deliberately did not plot a conductivity model for Proterozoic regions in Figure 3 because of the much larger range of interpreted profiles. Instead, as a representative I redraw the 2-D SVEKA model of Korja and Koivukoski [1994] along Finland together with 2- D model isotherms and Moho depth from seismic refraction (Figure 4). Despite careful construction, all MT models are subject to some lack of fit, parameter correlations, or susceptibility to 3-D effects, so I just want to emphasize main features. Very strong conductors extending at least to the middle crust near each end of the profile are associated with graphitic schists in the Proterozoic Tampere and Kainuu schist belts. These particular belts come to near surface, but others in Fennoscandia and elsewhere can be followed to or lie at considerable depth in the crust [Korja and Hjelt, 1993; Jones et al., 1993; Wu, 1994]. Away from the sutures toward the center of the model under the granitoid complex, the highest conductivity (lowest resistivity) resides at generally greater depth and the total crustal conductance is diminished. Nevertheless, even the minimum total conductance (~200 S) far exceeds that of the Archean crust (2-40 S, typical of the Archean). Proterozoic lower crust elsewhere also appears to have conductances of 200-800 S [Jones, 1992; Hyndman et al., 1993]. I note too that conductivity first starts to increase at depths of only ~10 km, going from <10-4 S/m at the surface to an intermediate zone perhaps 10 times greater, before the ultimate increase in the lower crust below 25-35 km (Figure 4). The real variation may be smooth, but the protracted decrease is a required model feature to explain the broad frequency range over which the MT apparent resistivity response decreases [Korja and Koivukoski, 1994]. Although not universal, it is resolved as well in the Archean region here and at the Kapuskasing uplift [Mareschal et al., 1995] and in our initial modeling of the Paleozoic Appalachian transect [Wannamaker et al., 1996]. The idea that Archean deep crust is especially resistive appears supported by results from the Lewisian and Ukranian terrains [Haak and Hutton, 1986] and new MT results from the Slave craton [Jones and Ferguson, 1997]. The Russian Archean to the northeast of Figure 4 has an integrated crustal conductance of only about 2 S/m [Korja and Hjelt, 1993]. Reduced f(O2) alone is not sufficient to promote enhanced interconnected graphite, or we would not expect the Archean terrains to be less conductive. On the other hand, we must avoid including Archean protoliths subsequently overprinted orogenically. Higher conductances than typical are seen under northern Wisconsin [Dowling, 1970] and the Siberian craton [Vanyan et al., 1989], but these have been extensively reworked during the Proterozoic [Goodwin, 1996]. South Africa has seen much thermally- related uplift in the Tertiary [Burke, 1996] which may in part explain the higher conductivities sometimes inferred beneath there away from sutures [Haak and Hutton, 1986]. The model geotherms in Figure 4 imply temperatures of about 400oC at the base of the crust of the SVEKA profile. The modest enhancement in temperatures to the south likely would be even smaller if higher heat production in Proterozoic upper crust is even a partial cause for its slightly higher heat flow [Nyblade and Pollack, 1993; Guillou-Frottier et al., 1995]. Since intermediate- mafic lithologies dominate the lower crust here according to seismic velocities [Korja and Koivukoski, 1994], much or all of the deep crustal conductivity should reside in brittle rocks. But, what is its cause? Well, the low deep crustal temperatures of Proterozoic and even Paleozoic areas do not support a thermally activated cause such as defect populations in minerals. There is no evidence for enhanced alkali conduction with a( H<SUB>2</SUB>O). Unless the role of H+ in crustal minerals vastly exceeds that modeled for olivine under similar conditions, even a generous P( H<SUB>2</SUB>O) of 1 kbar would only increase solid-state conductivity at 400oC to no more than 10-4 S/m (Karato, 1990; Bai and Kohlstedt, 1992). Thermal activation also seems unlikely to explain the initial increases in conductivity common to old terrains which are observed first at depths as shallow as 10 km (e.g., Figure 4), where temperatures would barely reach 150oC. Hydrous defects appear more relevant at the higher temperatures and pressures of the upper mantle. Alternately, deep drilling and regional hydrology studies have shown fractures extending to surprisingly great depth in the crust of ancient terrains, and conceivably throughout the brittle regime to nearly the Moho. For this to explain the conductance increase through the Precambrian, there must be much greater fracturing of the Proterozoic areas. The prodigious orogeny of the Proterozoic belts of course is well known [Goodwin, 1996]. The Archean interiors tended to escape most of it, perhaps due to their deep and depleted lithospheric roots. Problematically, Proterozoic orogeny generally included extensive post-orogenic and anorogenic magmatism [Goodwin, 1996], which implies that most original deformation zones would have been ductile and diffuse or thermally annealed below the middle crust of that time [Newton, 1990]. However, modern intraplate earthquake activity tends to cluster in the fossil orogenic belts surrounding the Archean [Slemmons et al., 1991]. Rupturing frequently occurs in or near pre-existing weak zones dating from the Paleozoic or earlier, suggesting secular advancement of brittle failure into formerly ductile but long- inactive deep crust. With the above arguments, I believe the case is strong for a graphitic cause of the modest conductivity increases below old, stable terrains where its geochemical stability commonly is evident. One of the strengths of its candidacy is no need to preserve open fracture space over long geological time; episodic but widespread dispersal of hydrothermal C-O-H fluids by seismic pumping is inferred in many environments [Newton, 1990; Sibson, 1994] and may extend downward across the brittle-ductile transition [Roering et al., 1995]. Plate tectonic collisional processes of the Proterozoic could have increased input of carbon to the deep crust directly as organic-rich sedimentary distributions or from the mantle below via slab devolatilization. Orogenic granulitization may disrupt much of the original long-range conduction, but yet later retrogression in the 250-450oC range not necessarily related to main stabilization may remobilize and interconnect a portion of carbon over a wider volume (e.g., as at KTB). Graphite in such shear zones may define f(O2) locally even though the bulk rock state may be somewhat more oxidizing [Emmerman and Lauterjung, 1997]. This process of low-grade remobilization may develop over 10's or 100's of Ma and affect ever greater depths and lateral extents in the crust during long-term cooling and intraplate deformation. The apparent increase with depth may combine increased scales of conducting elements in the MT response and the decrease in isotherm descent rate over time. There may be no reason to distinguish between Proterozoic and Paleozoic terrains in the process, but this should be testable in future surveys. Of course, it is possible for solid phase and brine conductors to act in series and augment conduction [Shankland et al., 1997]. This is a miniaturized version of the concept of Merzer and Klemperer [1992].
6. Summary and Conclusions
The early syntheses of Shankland and Ander [1983] and Haak and Hutton [1986] I believe were excellent science in their day. Their main point was that solid-state conduction mechanisms in crustal minerals at appropriate temperatures were inadequate to explain field survey results. This conclusion still stands. Even then, a distinction was made between the degree of conductivity of old, stable regimes and active ones for a specific temperature. As I have argued here, there may be a relation between tectonic age and deep crustal conductivity, but it is made up of (at least) a short term, fluid-based component and a much longer term, dynamic component involving solid phases. The concept that the tops of deep crustal conductors of all ages correspond to an isotherm represented only one school of thought and is oversimplified in my view. In part it has been feasible due to wide uncertainties both in the earlier conductivity models and in geotherms estimated from heat flow. Recent models from active regimes show that major variation in conductor temperature is likely which may be related to differences in rheology and metamorphic grade. Several high quality surveys suggest the conductivity increase with depth may be very gradational, calling into question the whole idea of a conductor isotherm. However, if care is taken in both modeling and incorporation of constraints, the conductivity structures can be assigned physico-chemical causes involving fluids which are consistent with external constraints and petrological principles. Each survey area should be considered as an individual rather than assuming causes for conductivity structure a priori. In regimes whose last tectonic event was Paleozoic or older in age, and in several younger ones too, the most outstanding conductors appear associated with fossil suture zones. They should be considered distinct from background crustal conductivity values to the extent the latter can be defined. Although some early interpretations appealed to residual fluids, most workers today call on deeply underthrust graphitic metasediments as a cause. Frequently in those field areas, there is nearby evidence of graphite with appropriate textures. Many suture zones may have no surviving conductivity, but many do and it is to these which EM geophysicists tend to be attracted for study. We do not need all graphitic occurences to be interconnected over large distances, just enough to contribute sufficiently to the conductivity enhancements in accord with the increasing resolution scale of EM fields. In the older regimes also, most of the crust should now lie in the brittle regime and appropriate conductivity mechanisms should be considered. That deep crustal conductivity increases markedly from the early Archean through the Proterozoic, and that it can start to rise at only about 10 km depth, appear difficult to explain by thermally activated causes including hydrous defects since the temperatures appear quite low. Deep drilling shows that fluid- filled fractures can exist to the brittle-ductile transition in elevated heat flow regimes. If this is true generally, perhaps the modest conductivity enhancements deep in stable regimes may be a combination of higher salinity, temperature and ever-widening scales of contributing fractures with depth. However, any degree of increased fracturing of Proterozoic crust relative to the Archean, whether original or reactivated, must meet the posited increase in conductance through the Precambrian to be an explanation. In the end, I lean towards solid-phase conductors, principally graphite, as responsible for most broad-scale, background conductivity increases deep below long-stable geological terrains, although brines may contribute independently or in series. I mainly have in mind distributions of large-scale, conductive elements such as shear zones which are not represented in hand- specimen or even many outcrop scales, as the KTB results have shown. Whether the carbon source is sedimentary or mantle-derived is not of primary importance. In my hypothesis, continental deformation and metamorphism disperse carbon from its original placement as sub-thrust packages and fluid species. It is especially in the late-stage cooling of an orogeny that retrograde fluids traversing the brittle-ductile transition and shallower may mobilize carbon into an ultimate long-range interconnection along fault zones. Once graphite precipitates from a reduced fluid, fracture porosity need not persist. In the 10's to maybe >100 Ma after orogeny, the dispersion continues downward through the crust roughly apace with the brittle-ductile transition in response to evolving stress fields. The polyphase orogeny characteristic of granulite terranes [Mezger, 1992] presumably enhances the spreading. To some extent, the dispersion may continue today as marked by deep seismicity. The increase in deep crustal conductance during the Proterozoic may be caused in part by an increase in organic-rich sediment deposition plus the advent of plate tectonic processes able to carry carbon-bearing material to great depth. The concept of conductivity mechanisms operating at the brittle- ductile transition and higher in stable regimes, rather than below the transition as in active regimes, needs more testing. From MT researchers, it requires increased quality of field observations, appropriate model construction and close attention to data fitting, to accurately define depth distribution and temperature range of conductivity structure. At this point, we could really benefit from the constructive input of petrologists and other geoscientists. What are the minimum temperatures of stability of arbitrary fluids versus metamorphic grade? Can hydrous defects in minerals significantly enhance mineral conductivity even at low temperatures? What is the ultimate depth extent of fluid-filled fractures? Is retrograde remobilization of carbon into widely- spaced fault zones sufficiently common in the lower crust to affect bulk conduction at scales >10 km? What are the physico-chemical properties of the fluids responsible for graphite movement and how are they generated? We need careful and thoroughly documented efforts across numerous geoscience disciplines to bring EM geophysics from the realm of anecdotal evidence to reliable systematics.
7. References
Allis, R. G., and Y. Shi, New insights to temperature and pressure beneath the central Southern Alps, New Zealand, New Zealand J. Geol. Geophys., 38, 585-592, 1995.
Aranovich, L. Ya., and R. C. Newton, H<SUB>2</SUB>O activity in concentrated NaCl solutions at high pressures and temperatures measured by the brucite-periclase equilibrium, Contr. Mineral. Petr., 125, 200-212, 1996.
Arculus, R. J., Aspects of magma genesis in arcs, Lithos, 33, 189- 208, 1994. Bai, Q., and D. L. Kohlstedt, Substantial hydrogen solubility in olivine and implications for water storage in the mantle, Nature, 357, 672-674, 1992.
Bailey, R. C., Trapping of aqueous fluids in the deep crust, Geophys. Res. Lett., 17, 1129-1132, 1990. Balhaus, C., Redox states of lithospheric and asthenospheric upper mantle, Contr. Miner. Petr., 114, 331-348, 1993.
Banks, R. J., D. Livelybrooks, P. Jones, and R. Longstaff, Causes of high crustal conductivity beneath the Iapetus suture zone in Great Britain, Geophys. J. Int., 124, 433-455, 1996. Batzle, M. L., and G. Simmons, Geothermal systems: rocks, fluids, fractures, in The Earth's Crust: Its Nature and Physical Proper- ties, edited by J. G. Heacock, pp. 233-242, Amer. Geophys. Union Monogr. 20, 1977.
Boerner, D. E., R. D. Kurtz, and J. A. Craven, Electrical conductivity and Paleo-Proterozoic foredeeps, J. Geophys. Res., 101, 13,775-13,791, 1996.
Borevsky, L., S. Milanovsky, and L. Yakovlev, Fluid-thermal regime in the crust-superdeep drilling data, in Proc. World Geotherm. Congress, edited by E. Barbier, G. Frye, E. Iglesias, and G. Palmason, pp. 975-981, Int. Geotherm. Assoc., Auckland, New Zealand, 1995.
Brace, W. F., Resistivity of saturated crustal rocks to 40 km based on laboratory measurements, in The Structure and Physical Proper- ties of the Earth's Crust, edited by J. G. Heacock, pp. 243-256, Amer. Geophys. Union Monogr. 14, 1971.
Brace, W. F., Permeability of crystalline and argillaceous rocks, Int. J. Rock Mech. Min. Sci. Geomech. Abstr., 17, 241-251, 1980. Brace, W. F., Permeability of crystalline rocks: new measurements, J. Geophys. Res., 89, 4327-4330, 1984.
Brandon, A. D., and D. S. Draper, Constraints on the origin of the oxidation state of mantle overlying subduction zones: an example from Simcoe, Washington, USA, Geochem. Cosmoch. Acta., 60, 1739- 1749, 1996.
Brown, C., and J. P. Whelan, Terrane boundaries in Ireland inferred from the Irish magnetotelluric profile and other geophysical data, J. Geol. Soc. Lond., 152, 523-534, 1995. Bucher, K., and M. Frey, Petrogenesis of metamorphic rocks, 318 pp., Springer-Verlag, New York, 1994. Burke, K., The African plate, S. Afr. J. Geol., 99, 341-409, 1996.
Burnham, C. W., The importance of volatile constituents, in The Evolution of the Igneous Rocks: Fiftieth Anniversary Perspectives, edited by H. S. Yoder, Jr., pp. 439-482, Princeton University Press, Princeton, N. J., 1979a.
Burnham, C. W., Magmas and hydrothermal fluids, in Geochemistry of Hydrothermal Ore Deposits, edited by H. L. Barnes, pp. 71-136, John Wiley, New York, 1979b.
Carmichael, I. S. E., The redox states of basic and silicic magmas: a reflection of their source regions?, Contr. Mineral. Petr., 106, 129-141, 1991.
Christiansen, N. I., and W. D. Mooney, Seismic velocity structure and composition of the continental crust: a global view, J. Geophys. Res., 100, 9761-9788, 1995.
Constable, S. C., R. L. Parker, and C. G. Constable, Occam's inversion: A practical algorithm for generating smooth models from electromagnetic sounding data, Geophysics, 52, 289-300, 1987.
Costain, J. K., R. D. Hatcher, Jr., C. Coruh, T. L. Pratt, S. R. Taylor, J. J. Litehiser, and I. Zeitz, Geophysical characteristics of the Appalachian crust, in The Appalachian-Ouachita orogen in the United States, edited by R. D. Hatcher, Jr., W. A. Thomas, and G. W. Viele, pp. 233-318, The Geology of North America, F-2, Geol. Soc. Amer., Boulder, CO, 1989.
Davey, F. J., T. Henyey, W. S. Holbrook, D. Okaya, T. A. Stern, D. Eberhart-Phillips, T. McEvilly, R. Urhammer, H. Anderson, G. R. Jiracek, P. E. Wannamaker, G. Caldwell, and N. Christiansen, Preliminary results from a geophysical study across a modern, continent-continent collisional plate boundary - the Southern Alps, New Zealand, Tectonophysics, Special Issue on VIIth Int. Symp. Deep seismic profiling of the continents, in press, 1997.
DeGroot-Hedlin, C., and S. Constable, Occam's inversion to generate smooth, two-dimensional models from magnetotelluric data: Geophysics, 55, 1613-1624, 1990. Dissanayake, C. B., Origin of vein graphite in high-grade terrains, Mineral. Deposita, 29, 57-67, 1994.
Dowling, F. L., Magnetotelluric measurements across the Wisconsin Arch, J. Geophys. Res., 75, 2683-2698, 1970. Duba, A., Are laboratory electrical conductivity data relevant to the earth?, Acta Geod. Geophys. Montan., 11, 485-495, 1976.
Duba, A. G., and H. C. Heard, Effect of hydration on the electrical conductivity of olivine, Eos Trans. AGU, 61, 404, 1980. Emmermann, R., and J. Lauterjung, The German continental deep drilling program KTB: overview and major results, J. Geophys. Res., 102, 18,179-18,201, 1997.
Feves, M., G. Simmons, and R. W. Siegfried, Microcracks in crustal igneous rocks:physical properties, in The Earth's Crust: Its Nature and Physical Properties, edited by J. G. Heacock, pp. 95-117, Amer. Geophys. Union Monogr. 20, 1977.
Frost, B. R., and K. Bucher, Is water responsible for geophysical anomalies in the deep continental crust?: A petrological perspective, Tectonophys., 231, 293-309, 1994.
Frost, B. R., C. D. Frost, and J. L. R. Touret, Magmas as a source of heat and fluids in granulite metamorphism, in Fluid movements - elemental transport and the composition of the deep crust, edited by D. Bridgwater, pp. 1-18, NATO ASI Series C, 281, 1989a. Frost, B. R., W. S. Fyfe, K. Tazaki, and T. Chan, Grain-boundary graphite in rocks and implications for high electrical conductivity in the lower crust, Nature, 340, 134-136, 1989b.
Goldsmith, J. R., Enhanced Al/Si diffusion in KAlSi3O8 at high pressures: the effect of hydrogen, J. Geol., 96, 109-124, 1988. Goodwin, A. M., Principles of Precambrian geology, 327 pp., Academic Press, New York, 1996. Gough, D. I., Magnetometer array studies, earth structure, and tectonic processes, Rev. Geophys., 27, 141-157, 1989. Guillou-Frottier, L., J.-C. Mareschal, C. Jaupart, C. Gariepy, R. Lapointe, and G. Bienfait, Heat flow variations in the Grenville Province, Canada, Earth Planet. Sci. Lett., 136, 447-460, 1995. Haak, V., and R. Hutton, Electrical resistivity in continental lower crust, in Nature of Lower Continental Crust, edited by J. B. Dawson, D. A. Carswell, J. Hall, and K. D. Wedepohl, Geol. Soc. Spec. Publ. London, 24, 35-49, 1986. Haggerty, S. E., Redox state of the continental mantle, in Continental mantle, edited by M. A. Menzies, pp. 87-110, Clarendon Press, Oxford, 1990. Hjelt, S.-E., and T. Korja, Lithospheric and upper mantle struc- tures, results of electromagnetic soundings in Europe, Phys. Earth Planet. Inter., 79, 137-177, 1993. Hyndman, R. D., and P. M. Shearer, Water in the lower continental crust: modeling magnetotelluric and seismic reflection results, Geophys. J. Int., 98, 343-367, 1989. Hyndman, R. D., L. L. Vanyan, G. Marquis, and L. K. Law, The origin of electrically conductive lower continental crust: saline water or graphite?, Phys. Earth Planet. Inter., 81, 325-344, 1993. Jones, A. G., Electrical conductivity of the continental lower crust, in Continental lower crust, edited by D. M. Fountain, R. J. Arculus and R. W. Kay, pp. 81-143, Elsevier, New York, 1992. Jones, A. G., Electromagnetic images of modern and ancient subduction zones, Tectonophys., 219, 29-45, 1993. Jones, A. G., and I. J. Ferguson, Early Archean processes revealed through electromagnetic images of the Slave Craton, Proc. 8th Scientific Assembly of IAGA, Uppsala, Sweden, August 2-15, 1997. Jones, A. G., J. A. Craven, G. W. McNiece, I. J. Ferguson, T. Boyce, C. Farquarson, and R. Ellis, North American Central Plains conductivity anomaly within the Trans-Hudson orogen in northern Saskatchewan, Canada, Geol., 21, 1027-1030, 1993. Karato, S., The role of hydrogen in the electrical conductivity of the upper mantle, Nature, 347, 272-273, 1990. Korja, T., and S.-E. Hjelt, Electromagnetic studies in the Fennoscandian Shield - electrical conductivity of Precambrian crust, Phys. Earth Planet. Int., 81, 107-138, 1993. Korja, T., and K. Koivukoski, Crustal conductors along the SVEKA profile in the Fennoscandian (Baltic) shield, Finland, Geophys. J. Int., 116, 173-197, 1994. Korja, T., P. Tuisku, T. Pernu, and J. Karhu, Field, petrophysical and carbon isotope studies on the Lapland granulite belt: implications for deep continental crust, Terra Nova, 8, 48-58, 1996. Kremenetsky, A. A., and L. N. Ovchinnikov, The Precambrian continental crust: its structure, composition and evolution as revealed by deep drilling in the U.S.S.R., Precamb. Res., 33, 11- 43, 1986. Kronenberg, A. K., and S. H. Kirby, Ionic conductivity of quartz: DC time dependence and transition in charge carriers, Amer. Mineral., 72, 739-747, 1987. Kronenberg, A. K., R. A. Yund, and G. R. Rossman, Stationary and mobile hydrogen defects in potassium feldspar, Geochim. Cosmochim. Acta, 60, 4075-4094, 1996. Kullerud, K., Chlorine, titanium and barium-rich biotites: factors controlling biotite composition and implications for garnet-biotite geothermometry, Contr. Mineral. Petr., 120, 42-59, 1995. Lamb, W., and J. W. Valley, Metamorphism of reduced granulites in low-CO<SUB>2</SUB> vapour-free environment, Nature, 312, 56-58, 1984. Larsen, J. C., A new technique for layered earth magnetotelluric inversion, Geophysics, 46, 1247-1257, 1981. Mackie, R. L., and T. R. Madden, Three-dimensional magnetotelluric inversion using conjugate gradients, Geophys. J. Int., 115, 215- 229. Madden, T. R., Random networks and mixing laws, Geophysics, 41, 1104-1125, 1976. Mareschal, M., R. D. Kurtz, and R. C. Bailey, A review of electromagnetic investigations in the Kapuskasing uplift and surrounding regions: electrical properties of key rocks, Can. J. Earth Sci., 31, 1042-1051, 1994. Mareschal, M., R. L. Kellett, R. D. Kurtz, J. N. Ludden, S. Ji, and R. C. Bailey, Archean cratonic roots, mantle shear zones and deep electrical anisotropy, Nature, 375, 134-137, 1995. Merzer, A. M., and S. L. Klemperer, High electrical conductivity in a model lower crust with unconnected, conductive, seismically reflective layers, Geophys. J. Int., 108, 895-905, 1992. Mezger, K., Temporal evolution of regional granulite terranes: implications for the formation of lowermost continental crust, in Continental lower crust, edited by D. M. Fountain, R. J. Arculus and R. W. Kay, pp. 447-478, Elsevier, New York, 1992. Naraoka, H., M. Ohtake, S. Maruyama, and H. Ohmoto, Non-biogenic graphite in 3.8-Ga metamorphic rocks from the Isua district, Greenland, Chem. Geol., 133, 251-260, 1996. Newton, R. C., Fluids and shear zones in the deep crust, Tectonophys., 182, 21-37, 1990. Nesbitt, B. E., Electrical resistivities of crustal fluids, J. Geophys. Res., 98, 4301-4310, 1993. Nyblade, A. A., and H. N. Pollack, A global analysis fo heat flow from Precambrian terrains: implications for the thermal structure of Archean and Proterozoic lithosphere, J. Geophys. Res., 98, 12,207-12,218, 1993. Ogawa, Y., A. G. Jones, M. J. Unsworth, J. R. Booker, X. Lu, J. Craven, B. Roberts, J. Parmalee, and C. Farquharson, Deep electri- cal conductivity structures of the Appalachian orogen in the southeastern U. S., Geophys. Res. Lett., 23, 1597-1600, 1996. Petrick, W. R., W. H. Pelton, and S. H. Ward, Ridge regression inversion applied to crustal resistivity sounding data from South Africa, Geophysics, 42, 995-1005, 1977. Porath, H., A review of the evidence on low-resistivity layers in the Earth's crust, in The Structure and Physical Properties of the Earth's Crust, edited by J. G. Heacock, pp. 127-144, Amer. Geophys. Union Monogr. 14, 1971. Ranalli, G., Rheology of the lithosphere through space and time, in Orogeny through time, edited by J.-P. Burg and M. Ford, Geol. Soc. Spec. Publ. London, 121, 19-38, 1997. Roering, C., D. D. van Reenen, C. A. Smit, and R. Du Toit, Deep crustal embrittlement and fluid flow during granulite metamorphism in the Limpopo belt, South Africa, J. Geol., 103, 673-686, 1995. Rudnick, R. L., and D. M. Fountain, Nature and composition of the continental crust: a lower crustal perspective, Rev. Geophys., 33, 267-309, 1995. Rumble, D., III, E. F. Duke, and T. L. Hoering, Hydrothermal graphite in New Hampshire: evidence of carbon mobility during regional metamorphism, Geol., 14, 452-455, 1986. Rumble, D., III, C. P. Chamberlain, P. K. Zeitler, and B. Barreiro, Hydrothermal graphite veins and Acadian granulite facies metamorphism, New Hampshire, USA, in Fluid movements - elemental transport and the composition of the deep crust, edited by D. Bridgwater, NATO ASI Series C, 281, 117-119, 1989. Rutter, E. H., and K. H. Brodie, Rheology of the lower crust, in Continental lower crust, edited by D. M. Fountain, R. J. Arculus and R. W. Kay, pp. 201-267, Elsevier, New York, 1992. Sanders, I. S., Exhumed lower crust in NW Ireland, and a model for crustal conductivity, J. Geol. Soc. London, 148, 131-135, 1991. Shankland, T. J., and M. E. Ander, Electrical conductivity, tem- perature, and fluids in the lower crust, J. Geophys. Res., 88, 9475-9484, 1983. Shankland, T. J., and A. G. Duba, Effect of hydrogen on electrical conductivity of mantle olivine, Proc. 13th Biennial IAGA Workshop on Electromagnetic Induction in the Earth, p. 56, Onuma, Japan, July 12-18, 1996. Shankland, T. J., A. G. Duba, E. A. Mathez, and C. L. Peach, Increase of electrical conductivity with pressure as an indicator of conduction through a solid phase in midcrustal rocks, J. Geophys. Res., 102, 14,741-14,750, 1997. Sibson, R. H., Crustal stress, faulting and fluid flow, in Geofluids: Origin, Migration and Evolution of Fluids in Sedimentary Basins, edited by J. Parnell, Geol. Soc. Spec. Publ. London, 78, 69-84, 1994. Slemmons, D. B., E. R. Engdahl, M. D. Zoback, and D. D. Blackwell, editors, Neotectonics of North America, 498 pp., Geol. Soc. Amer., Decade Map Vol. 1, 1991. Smith, J. T., and J. R. Booker, Magnetotelluric inversion for minimum structure, Geophysics, 53, 1565-1576, 1988. Smith, J. T., and J. R. Booker, Rapid inversion of two- and three- dimensional magnetotelluric data, J. Geophys. Res., 96, 3905-3922, 1991. Smith, J. V., Phase equilibria of plagioclase, in Feldspar mineralogy, 2nd edit., edited by P. H. Ribbe, Mineral. Soc. Amer. Rev. Mineral., 2, 223-240, 1983. Smith, R. B., W. C. Nagy, K. A. S. Julander, J. J. Viveiros, C. A. Barker, and D. J. Gants, Geophysical and tectonic framework of the eastern Basin and Range-Colorado Plateau-Rocky Mountain transition, in Geophysical framework of the continental United States, edited by L. C. Pakiser and W. D. Mooney, Geol. Soc. Amer. Mem., 172, 205- 233, 1989. Stern, T. A., P. E. Wannamaker, D. Eberhart-Phillips, D. Okaya, F. J. Davey, and the South Island working group, Crustal structure experiments across the Southern Alps of New Zealand, EOS, Trans. AGU, 78, 329, 335-336, 1997. Ucok, H., I. Ershaghi, and G. R. Olhoeft, Electrical resistivity of geothermal brines, J. Pet. Tech., 32, 717-727, 1980. Van Schmus, W. R., M. E. Bickford, J. L. Anderson, E. E. Bender, R. R. Anderson, P. W. Bauer, J. M. Robertson, S. A. Bowring, K. C. Condie, R. E. Denison, M. C. Gilbert, J. A. Grambling, C. K. Mawer, C. K. Shearer, W. J. Hinze, K. E. Karstrom, E. B. Kisvarsanyi, E. G. Lidiak, J. C. Reed, Jr., P. K. Sims, O. Tweto, L. T. Silver, S. B. Treves, M. L. Williams, and J. L. Wooden, Transcontinental Proterozoic provinces, in Precambrian: Coterminous U. S., edited by J. C. Reed, Jr., M. E. Bickford, R. S. Houston, P. K. Link, D. W. Rankin, P. K. Sims, and W. R. Van Schmus, pp. 171-334, The Geology of North America, C-2, Geol. Soc. Amer., Boulder, CO, 1993. Vanyan, L. L., A. P. Shilovsky, B. A. Okulessky, V. Yu. Semenov, and T. A. Sidelnikova, Electrical conductivity of the crust of the Siberian platform, Phys. Earth Planet. Int., 54, 163-168, 1989. Wannamaker, P. E., Electrical conductivity of water-undersaturated crustal melting, J. Geophys. Res., 91, 6321-6327, 1986. Wannamaker, P. E., and G. W. Hohmann, Electromagnetic induction studies: U. S. National Rep. to IUGG, Rev. Geophys., supplement, 405-415, 1991. Wannamaker, P. E., J. M. Johnston, J. A. Stodt and J. R. Booker, Anatomy of the Southern Cordilleran Hingeline, Utah and Nevada, from deep resistivity profiling: Geophysics, 62, 1069-1086, 1997a. Wannamaker, P. E., W. M. Doerner, J. A. Stodt, and J. M. Johnston, Subdued state of tectonism of the Great Basin interior relative to its eastern margin based on deep resistivity structure, Earth Planet. Sci. Lett., 150, 41-53, 1997b. Wickham, S. M., Fluids in the deep crust - petrological and isotopic evidence, in Continental lower crust, edited by D. M. Fountain, R. J. Arculus and R. W. Kay, pp. 391-421, Elsevier, New York, 1992. Wu, N., High resolution images of conductivity structure in the mid-lower crust and upper mantle., Ph.D. Dissertation, University of Washington, 1994. Yardley, B. W. D., Is there water in the deep crust?, Nature, 323, 111, 1986. Yardley, B. W. D., and J. W. Valley, The petrologic case for a dry lower crust, J. Geophys. Res., 102, 12,173-12,185, 1997. 8. Figures Fig. 1. Pressure-temperature projections of melting and fluid production in intermediate composition igneous rock at either upper amphibolite-granulite facies or at greenschist metamorphic facies (modified from Wannamaker, 1986). Traces of invariant points I (dashed lines) represent first appearance of H<SUB>2</SUB>O-CO<SUB>2</SUB> fluid in the presence of chlorite (chl) or amphibole (am). Breakdown of hydrous minerals yielding H<SUB>2</SUB>O fluid below the wet solidus, or else incongruent melting of hydrous minerals with anhydrous components above the wet solidus, is depicted with medium solid lines. Dotted curves are geotherms of shield and magmatically underplated regions. Fig. 2. Characteristic fields of f(O2) (oxygen fugacity) relative to the quartz-fayalite-magnetite (QFM) buffer for various petrological provinces in North America. Provinces include western U. S. calc-alkaline (WCA), Great Basin Bimodal (GBB), Trans- Proterozoic Plutonic (TPP), Utah topaz rhyolites (UTR), peralkaline iron-rich magmas (PIR) and eastern North American Precambrian metamorphics (EPM). Other buffers include nickel-nickel oxide (NNO) and iron-wustite (IW), the latter thought to be characteristic of Archean deep crust. Data from Lamb and Valley (1984), van Schmus et al. (1993), Frost and Frost (1997), and sources in Christiansen et al. (1986). Fig. 3. Sample conductivity profiles from diverse tectonic environments derived using MT measurements, compared to coincident geotherms computed from surface heat flow and thermal models. Regimes include the extensional eastern and central Great Basin areas [EGB and CGB; Wannamaker et al., 1997a,b], the New Zealand Southern Alps [Stern et al., 1997; Davey et al., 1997], and the Pontiac Archean terrane [Mareschal et al., 1995]. Horizontal dashed lines connect the New Zealand and Archean conductor tops with their geotherms for clarity. Fig. 4. Model resistivity cross section of the Fennoscandian shield of Finland from its south coast to its northeastern border with Russia derived from MT soundings. Included also is seismic refraction Moho (labeled M) and thermal contours from 2-D heat flow modeling. Geologic terranes include Southern Finland migmatite area (SFMA), Tampere schist belt (TSB), central Finland granitoid complex (CFGP), Lake Labota-Bothnian shear zone (LBBZ), Iisalmi block (IB), Kainuu schist belt (KSB) and Archean Kuhmo block (KB). Vertical exaggeration (VE) is 2:1. Redrawn from Korja and Koivukoski [1994].