- Email: [email protected]
Weakly magnetic crust in the Canadian Cordillera
Earth and Planetary Science Letters 248 (2006) 476 – 485 www.elsevier.com/locate/epsl
Weakly magnetic crust in the Canadian Cordillera Mark Pilkington a,⁎, David B. Snyder a , Kumar Hemant b,1 a
Geological Survey of Canada, 615 Booth Street, Ottawa, ON, Canada K1A 0E9 b GeoForschungsZentrum, Telegrafenberg, 14473 Potsdam, Germany
Received 11 April 2006; received in revised form 6 June 2006; accepted 8 June 2006 Available online 11 July 2006 Editor: S. King
Abstract Current models of continental crust favor an increase in magnetization with depth. Here we report a counter example from the Canadian Cordillera where almost a full thickness of non-magnetic continental crust is suggested by joint interpretation of magnetic and seismic data. The magnetic field over the Cordillera is characterized by complex, short-wavelength (< 100 km) anomalies associated with intrusive, metamorphic and volcanic rocks that occur at shallow depths (< 5 km) within accreted terranes. The longwavelength (>100 km) portion of the Cordilleran field is subdued and mainly featureless, and suggests a lack of magnetic sources at greater depths. Seismic reflection and refraction data from three major transects in the Yukon and British Columbia, Canada support this interpretation and indicate that sedimentary-like formations make up the majority of the crust. The dominance of shallow, upper crustal magnetization in the Canadian Cordillera contrasts with the generally-held view that the lower continental crust is the primary source for long-wavelength magnetic anomalies. Sources for these anomalies are often assumed to be located in the lower crust when surface magnetizations are insufficient to produce such anomalies or no correlation exists between the magnetic field and the mapped surface geology. The Canadian Cordillera appears to be an example of a non-magnetic lower crust overlain by a more magnetic upper crust that is, however, not magnetized strongly enough to produce significant long-wavelength magnetic anomalies. © 2006 Elsevier B.V. All rights reserved. Keywords: continental crust; magnetic field; susceptibility
1. Introduction The magnetic properties of continental crust are dependent on its composition from the surface down to the Curie isotherm, specifically on the distribution of magnetite, the most commonly occurring ferrimagnetic mineral. Generally, crystalline igneous and metamorphic rocks are the main contributors to anomalous magnetic ⁎ Corresponding author. Fax: +1 613 952 8987. E-mail address: [email protected] (M. Pilkington). 1 Present address: University of Leeds, Leeds, U.K. 0012-821X/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2006.06.010
fields, whereas sedimentary rocks can be considered essentially non-magnetic. Within the crust, the division into a felsic upper crust and a more mafic lower crust has led to granulite-facies rocks in the latter to be suggested as the predominant source for long-wavelength magnetic anomalies [1]. Multi-domain magnetite is commonly inferred to be the mineral responsible for this magnetism, although certain forms of the hematite–ilmenite series could also be contributors [2]. The magnitude of crustal magnetization at upper levels within the crust can be determined from surface samples and modelling of short-wavelength magnetic anomalies. For the deeper
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
parts, xenolith magnetic properties are useful, as is the interpretation of long-wavelength magnetic anomalies, such as those measured by satellite-altitude surveys [3]. Here, we report an example of almost a full thickness of non-magnetic continental crust suggested by joint interpretation of magnetic and reflection and refraction seismic data, where magnetic sources are restricted to a thin (< 5 km) surface layer, and discuss the consequences for the global description of crustal magnetization and the modelling of satellite magnetic data. 2. Magnetic field data The Canadian Cordillera is characterized by a subdued and mainly featureless long-wavelength (> 100 km) magnetic field (Fig. 1). This is in marked contrast to the numerous, extensive, high-amplitude magnetic anomalies that are associated with exposed and buried Precambrian basement to the east and north of the Cordillera. Indeed, the boundary between the two magnetically contrasting regions closely mimics the geologically determined extent of the Cordillera, i.e., the Cordilleran deformation front (CDF, Fig. 1). The only discrepancy between the two occurs north of 65°N where the CDF turns northward to the Arctic coast (e.g., [4]) but the boundary between subdued and higher-amplitude magnetic fields continues approximately westward to the Alaskan border. The data in Fig. 1 are derived from a series of highaltitude (∼5 km) aeromagnetic surveys flown in the
477
1970s [5]. Flightline spacing averages 37 km over the area shown and the data were gridded at a 10-km interval. A cosine-tapered, low-pass filter with center roll-off wavelength of 100 km was applied to the gridded data to reduce residual navigational error effects. This filtering also suppresses anomalies due to near-surface magnetic sources and emphasizes the effects of deeper bodies occurring in the middle and lower crust. Even in nonfiltered data, Haines et al. [5] and Coles et al. [6] noted the clear distinction in magnetic character between the Cordillera and its surroundings. They suggested that the subdued field over the Cordillera was due to the absence of Precambrian crystalline basement material in the region. On the basis of several long aeromagnetic profiles, Caner [7] suggested the smooth Cordilleran field was due to shallowing of the Curie isotherm or the presence of a more felsic crust. To the east of the CDF, crystalline basement rocks making up the North American craton produce numerous, high-amplitude (hundreds of nanoteslas), laterallyextensive (hundreds of kilometers) anomalies that persist even when covered by thick sedimentary basins (e.g., > 5 km in southern Alberta). These kinds of anomalies are not seen within the Cordillera. In the longwavelength field some higher-amplitude (> 200 nT) anomalies do occur, for example, one just to the east of Vancouver Island coincident with the coastline and another centered at 127.5°W, 57.7°N. The low-altitude (∼300 m), high-resolution (1-km grid interval) magnetic
Fig. 1. High-altitude (∼ 5 km) magnetic field of western Canada filtered to remove wavelength components <100 km. CDF, Cordilleran deformation front. Data collection details are given in Haines et al. [5].
478
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
Fig. 2. Low-altitude (∼300 m mean terrain clearance) magnetic field of western Canada. CDF, Cordilleran deformation front. Dashed numbered boxes outline areas where power spectra are calculated (see Fig. 4). Details on sources and processing of individual survey data can be found at http://gdcinfo.agg.nrcan.gc.ca.
field over the Cordillera (Fig. 2) demonstrates the difference in character (and source) between these two anomalies and those seen over buried or exposed Precambrian rocks. The former are distinguished by their isolated, short-wavelength character indicating a nearsurface origin. This is the general case for the Cordilleran field, where individual anomalies are well correlated with surface geology and rock property measurements (e.g., [8–11]). The Cordilleran anomalies often have high amplitudes comparable to Precambrian basement anomaly levels but do not persist laterally, hence they do not contribute so efficiently to the longer wavelength (> 100 km) magnetic field shown in Fig. 1. For the low-altitude data (Fig. 2), individual anomalies can often be related to intrusive and volcanic outcrops, although detailed analysis of magnetic properties indicates the relationship is far from simple, with complications from alteration and thermal effects [12]. The Cordilleran field exhibits a strong northwesterly trend that reflects tectonic strike (Fig. 3). The Tintina– Rocky Mountain trench fault system divides a smooth featureless magnetic field over the Foreland Belt (ancestral North America) from the more variable field over the numerous accreted and pericratonic terranes to the west (cf. Figs. 2 and 3). In contrast to the northwesterly Cordilleran fabric, Precambrian basement trends vary from northeasterly south of 55°N, to northerly up to 65°N, where the predominant strike becomes east–west.
Similar regional differences in the strike of gravity anomalies can also be observed [11]. The persistence of high-amplitude, long-wavelength anomalies west of the
Fig. 3. Simplified terrane map of the Canadian Cordillera, after Gabrielse et al. [4]. TF-RMT, Tintina–Rocky Mountain trench fault system; CDF, Cordilleran deformation front; PA, Purcell anticlinorium. Thick grey lines are LITHOPROBE seismic profiles.
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
479
Both low- and high-altitude magnetic field data suggest that the Cordilleran crust comprises a shallow, thin, magnetic upper crustal source layer corresponding to igneous, metamorphic and volcanic rocks of various accreted terranes underlain by a weakly or non-magnetic crust. This view is supported by interpretation of recently acquired seismic reflection and refraction data acquired along traverses crossing the Cordillera, from the west coast to the ancient North American craton. 3. Seismic data
Fig. 4. Magnetic field power spectra for data within the dashed boxes in Fig. 2. Numbers refer to box locations. Thick lines indicate predicted spectral slopes [14] produced by average source depths of 1 and 5 km.
CDF in both the high- and low-altitude magnetic data indicate that cratonic North America extends westwards beneath the Cordilleran boundary. Where imaged by seismic reflection data, the edge of the craton appears as a westward facing monocline with up to 20 km of relief over distances of up to 80 km [13]. The difference in wavelength character between the field over the Cordillera and that over the surrounding regions is quantified through the power spectrum. Fig. 4 shows example power spectra for data within the dashed boxes in Fig. 2. The slope of the logarithmic magnetic field power spectrum is proportional to the average depth to the causative magnetic sources [14], hence Cordilleran regions exhibit consistently small source depths (< 5 km) compared to the buried Precambrian basement areas. Fig. 4 also indicates that for wavelengths <40 km (0.025 cycles/km) how much stronger the Cordilleran anomalies are than the Precambrian ones and vice versa for > 40 km components.
Deep seismic reflection profiles collected as part of the LITHOPROBE program form three major transects across the Canadian Cordillera (Fig. 3): one in the Yukon, and one each in northern and southern British Columbia [15,16]. The two northerly transects map a thick wedge of reflective layers that stretches from the western edge of the North American craton to accreted terranes hundreds of kilometers further west (Fig. 5). The strong continuity of reflectors within this unit has allowed the surface correlation of individual reflective layers to be mapped down to depths of ∼ 30 km [15,17]. Stratified Proterozoic units make up most of the reflective wedge, including the 0.8–0.54 Ga Windemere, 1.0–0.92 Ga MacKenzie Mountain and 1.84–1.71 Ga Wernecke Supergroups, plus the 1.81–1.5 Ga Muskwa Assemblage. Where exposed in the Foreland Belt, combined thicknesses of 7–21 km are indicated, and projections based on seismic data suggest a maximum thickness of 25–30 km. Although many reflections within these Proterozoic sedimentary groups can be tied from surface to depth, some reflective features within the lower crust could be caused by tectonically-interleaved, older crystalline continental crust. Refraction seismic data do, however, support the possibility of sedimentary rocks making up most of the crustal column, particularly in the northern Cordillera. Coincident with the proposed reflective wedge are low compressional to shear wave velocity ratios (Vp/Vs) which, based on comparison with laboratory measurements, suggest the presence of highly quartzose rocks [18]. In the southern Cordillera, similar
Fig. 5. Cartoon cross section of Cordilleran crust based on seismic reflection profiling, after Snyder et al. [17]. Pz, Paleozoic cover. Scale is omitted since this is a relationship diagram only.
480
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
thicknesses of stratified crust are observed from reflection profiles over the southern transect with over 20 km of (meta-) sedimentary rocks and associated sills making up the Purcell anticlinorium [19]. Such large thicknesses of sedimentary rocks are interpreted to have formed in a rift margin along the western edge of the North American craton that may have persisted from 1.85 to 0.54 Ga [17]. They overlie pre-1.8 Ga crystalline basement making up the edge of the North American craton in the east and form a base upon which the numerous accreted terranes were grounded to the west (Fig. 5). Importantly, both the reflection [15] and refraction seismic [18] data reveal that the northeasternmost terranes are just thin flakes thrust up onto the sedimentary wedge, while more southwesterly terranes such as Stikinia constitute most of the crust. The easternmost terranes, such as Cache Creek and Quesnelia (Fig. 3), were detached from their roots and thrust 200–400 km onto the wedge. Since stacking of terranes is unlikely, the order of their accretion is suggested by their current location. 4. Magnetic character of Cordilleran crust Volumetrically, the Cordilleran crust is dominated by the reflective Proterozoic sedimentary sequence. The remainder consists of westward-thinning North American craton (deep crust), thin accreted terranes (shallow crust) and thick attached terranes (whole crust; Fig. 5).
Magnetically, the sedimentary package is not significant because of the weak magnetizations (commonly <0.01 A/m) of these rock types. Where these strata are exposed, subtle (< 10 nT) anomalies can be detected by high-resolution, low-altitude (i.e., 300 m or smaller) aeromagnetic surveys, but these effects are quickly attenuated at higher elevations and have a negligible contribution at wavelengths greater than a few kilometers. Based on the seismic reflection data, it is possible that more magnetic intrusive rocks, in the form of sills, may be present within the Proterozoic strata. Nevertheless, because of the gentle dip (∼ 10°) and small thickness (<< 1 km) of any sills present, they are not expected to produce significant magnetic anomalies. For example, a 300-m thick sill dipping at 10° with a 1 A/m magnetization only produces a 10 nT anomaly even when exposed. Beneath the Proterozoic sedimentary strata is the westerly-tapering edge of the North American craton, probably consisting of pre-1.8 Ga crystalline basement. This appears as a poorly reflective zone at depths of >25 km along the Yukon and northern British Columbia seismic lines [15]. Further south, the basement is also less reflective than overlying strata, and appears to taper down from 15 km depth at the Rocky Mountain trench to > 25 km depth beneath the Purcell anticlinorium [16]. Measurements of heat flow in the Cordillera are consistently high and indicate Moho temperatures of 800– 900 °C [20]. This puts the Curie isotherm (580 °C) in the
Fig. 6. Satellite magnetic data for western Canada from the CHAMP mission [29]. Altitude average is 400 km. MR = Mackenzie River anomaly. ABC = Alberta–British Columbia anomaly. NWP = North Wyoming Province anomaly.
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
20–25 km depth range, based on an estimated 32– 35 km crustal thickness [21]. Consequently, the majority of the North American cratonic material within the Cordillera lies beneath the Curie isotherm and is nonmagnetic. The negligible effect of the buried Canadian shield beneath the Cordillera is also expected from the truncation of south-westerly trending basement-sourced anomalies around the position of the Rocky Mountain trench in southern British Columbia (Fig. 2). Both thick, attached and thin, accreted terranes contribute significantly only to the short-wavelength (< 100 km) magnetic field (cf. Figs. 2 and 4). Their longwavelength effect is minor and restricted to the coastal anomaly near Vancouver Island (Fig. 1) related to the Cascadia subduction zone [8]. This character is quantified by the example power spectra for these terranes which suggest shallow sources confined to upper crustal levels, i.e., less than ∼5 km deep. Thus the remainder of the crust appears essentially non-magnetic. Such a large lateral extent (600 × 2000 km) of weakly magnetized crust would be expected to produce a detectable signature in satellite-altitude magnetic data. Fig. 6 shows that this is the case, with a low-amplitude field characterizing the continent west of the CDF. This contrasts with higher amplitudes to the north and east, presumably associated with greater thicknesses of more magnetized crust occurring in cratonic North America. Prominent positive anomalies occur over the Mackenzie River region (MR, Fig. 6), northern Alberta and British Columbia (ABC) and to the north of the buried Wyoming Province in the United States (NWP). The contribution of Cordilleran crust to the satellite field is significantly modified by the effects of the measurement altitude; coalescence of the effects of magnetized sources in the crust stretches over distances equivalent to the satellite altitude, 400 km. As a result, even highly magnetic regions such as those within the craton are not always associated with large (positive) anomalies at satellite altitudes. Only those areas where large amplitude fields (> 400 nT in the data of Fig. 1) persist over distances comparable to satellite elevations produce detectable anomalies. For example, the Great Bear magmatic arc (striking north–south at 117°W between 62 and 67°N), which produces a significant response in the low-altitude data sets (Figs. 1 and 3), has almost no effect in the satellite field (Fig. 6). Only the Mackenzie River and the combined N.W. Alberta and N.E. British Columbia group of anomalies (Fig. 1) have large enough amplitudes and extents to contribute to the satellite field. Therefore the weakly magnetic crust within the Cordillera is not so clearly manifest in satellite data compared to the lower altitude data sets due to the
481
large difference in source-sensor distances, i.e., 400 km compared to ∼ 5 km and 0.3 km. 5. Discussion Excepting the thin shallow source layer of the exotic terranes, the crust within the Canadian Cordillera is weakly magnetic. This observation provides an independent constraint on regional-scale models of crustal magnetization, either those constructed for forward modelling purposes (e.g., [22,23]) or those resulting from inversion of magnetic field (commonly satellitealtitude) data (e.g., [24,25]). The construction of global crustal magnetization models has been simplified and improved through available global databases, e.g., Moho depth, sedimentary basin thickness, regional geology. These commonly have resolutions of 2–5° (200–500 km) which are comparable to that of satellite magnetic data. At such scales, crustal divisions based on tectonic setting represent necessarily broad averages. Nonetheless, anomalously low magnetization levels for practically the whole of the Canadian Cordilleran crust, with an extent of 600 × 2000 km, are not expected from knowledge of its surface geology. Consequently, forward models should not be based solely on the expected magnetic properties of a given tectonic unit, e.g., craton, mobile belt, assigned from surface geologic mapping, but should also be compatible with crustal magnetization values inferred from lower-altitude magnetic data and other geophysical data sets. Hemant and Maus [23] have constructed a global magnetization model based on worldwide compilations of geology, seismically-derived crustal thickness, and magnetic properties of rocks. Contrary to the forward models of Hahn et al. [22] and Purucker et al. [25] which assign a constant susceptibility to a limited number of geological classes, the Hemant and Maus [23] model contains a unique susceptibility value for each geological region. They assume that the Moho separates a magnetic crust from a non-magnetic upper mantle and that the continental crust is dominated by induced magnetization. Based on compilations of rock properties, magnetizations are assigned to surface rock types which may then be averaged within a given tectonic province. These values are assumed to apply to the upper crust while a multiple of these (greater than unity) based on crustal age is assigned to the lower crust. Finally, a verticallyintegrated susceptibility (VIS) value is determined from the combined crustal thickness and susceptibility information for each region (Fig. 7). One advantage of forward models based on purely geologic information is the possibility of specifying a high
482
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
Fig. 7. Vertically-integrated susceptibility (VIS) map for western Canada after Hemant and Maus [23].
degree of complexity of their parameterization. Areas smaller than those resolvable from satellite-altitude magnetic data can be parameterized and the boundaries between adjacent regions can be, and are usually, abrupt, compared to heavily smoothed transitions derived from inversions of satellite data. Fig. 8 shows the calculated satellite-altitude field based on the initial VIS model (Fig. 7) of Hemant and
Maus [23]. The match between the field predicted from surface geology and the observed field is reasonable in terms of spatial location of the major anomalous features and their amplitudes. Two of the three major positive anomalies within the study area are well-defined in the predicted map, the ABC anomaly (Fig. 6), caused by buried magmatic arc complexes, being more subdued.
Fig. 8. Predicted magnetic field at 400 km altitude based on the VIS data in Fig. 7.
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
483
Fig. 9. Modified, vertically-integrated susceptibility (VIS) map for western Canada.
The prominent low extending from 49°N to 65°N over most of the Cordillera is also reproduced in the calculated field, but its amplitude is smaller than observed in Fig. 6. Hence the assigned VIS values are too high. VIS values specified for most of the Cordillera (Fig. 7) were based on assuming a Phanerozoic-age crust giving a mean value of 0.434 SI km. This results in a small but still detectable
contrast with the adjacent oceanic crust, which based on its young age, has been assigned a similar value of 0.385 SI km [26]. The similarity in crustal magnetization levels between the Cordillera and oceanic regions is demonstrated by the lack of level change in the longwavelength magnetic field measured at lower altitudes over these areas (Fig. 1) even though the short-wavelength
Fig. 10. Predicted magnetic field at 400 km altitude based on the modified VIS map in Fig. 9.
484
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
character is different (Fig. 2). This similarity is not restricted to comparable VIS values, but is also a result of the comparable depth variation in magnetization for Cordilleran and oceanic regions, both consisting of thin (∼5 km) magnetic surface layers. As a first-order adjustment to the forward model to incorporate inferences from the seismic data, we reduced the VIS values over the whole extent of Cordilleran crust (Fig. 9). The entire Cordillera was assigned the same VIS value simply to group together those regions expected, from seismic information, of having a non-magnetic middle and lower crust. A value of 0.350 SI km, determined by trial and error, resulted in the best fit to the observations. The field produced by this adjusted model is shown in Fig. 10. The dominant magnetic low over the Cordillera in the satellite field is now better matched, although the boundary with the North American craton now produces a stronger effect. Clearly, adjustment of VIS values at scales smaller than the Cordillera is needed to improve the fit, particularly for the area of northern Alberta and British Columbia (ABC anomaly, Fig. 6). Our aim, however, is not to reach a final model, but to show that forward models of this type can be improved with guidance from geophysical information. The value of 0.350 SI km is within, but at the lower end of, the range of assigned VIS values for Phanerozoic crust worldwide [26]. In the Canadian Cordillera, the thin (<5 km) magnetic surface layer underlain by non-magnetic sedimentary rocks, shows the possible complexity of the depth variation of magnetization within continental crust which may persist over large lateral extents (100's km). It emphasizes the limitations of interpretations based solely on satellite magnetic data inversions, specifically the lack of depth discrimination for derived source distributions. This leads to estimates of only the depthintegrated magnetization of the crust and hence ambiguity regarding rock types within the crustal column and even information on the crustal type. Forward geological modelling, however, allows much greater detail in crustal models, but is limited by our lack of knowledge of crustal structure, composition and magnetic properties in many areas. Any successful approach to modelling crustal magnetization will, at the least, have to satisfy constraints from a priori geologic information plus those from magnetic inversions (measured at satellite altitude or lower). The dominance of shallow, upper crustal magnetization in the Canadian Cordillera also contrasts with the commonly-held view (e.g., [1,27]) that the lower continental crust is the primary source for long-wavelength magnetic anomalies. Sources for these longwavelength anomalies are often assumed to be located
in the lower crust when surface magnetizations are insufficient to produce such anomalies or no correlation exists between the magnetic field and the mapped surface geology. The Cordillera is an example of a non-magnetic lower crust overlain by a more magnetic upper crust that is, however, not magnetized strongly enough to produce significant long-wavelength magnetic anomalies. Predominant upper crustal magnetization is also present within the Ungava peninsula in eastern Canada, but here the magnetization levels are high enough to produce detectable long-wavelength anomalies measured both in aeromagnetic and satellite data [28]. Geologically, this area differs greatly from the Cordillera, being part of the North American craton and comprising mainly Archean plutonic rocks with some high-grade gneisses and small greenstone belts. Clearly, various scenarios must be considered possible when making even a simple division of continental crust into an upper and lower unit with different magnetization levels. Acknowledgments This is Geological Survey of Canada contribution 2005010. Mike Thomas and Carmel Lowe are thanked for providing useful reviews. References [1] P.N. Shive, R.J. Blakely, B.R. Frost, D.M. Fountain, Magnetic properties of the lower continental crust, in: D.M. Fountain, R. Arculus, R.W. Kay (Eds.), Continental Lower Crust, Elsevier, New York, 1992, pp. 145–177. [2] S.A. McEnroe, F. Langenhorst, P. Robinson, G.D. Bromiley, C.S. J. Shaw, What is magnetic in the lower crust? Earth Planet. Sci. Lett. 226 (2004) 175–192. [3] R.A. Langel, W.J. Hinze, The Magnetic Field of the Earth's Lithosphere: The Satellite Perspective, Cambridge Univ. Press, New York, NY, 1998. [4] H. Gabrielse, J.W.H. Monger, J.O. Wheeler, C.J. Yorath, Morphogeological belts, tectonic assemblages and terranes [chapter 2: tectonic framework], in: H. Gabrielse, C.J. Yorath (Eds.), Geology of the Cordilleran Orogen in Canada, Geology of Canada Series, vol. 4, Geol. Surv. of Can., Ottawa, 1991, pp. 15–28. [5] G.V. Haines, W. Hannaford, R.P. Riddihough, Magnetic anomalies over British Columbia and the adjacent Pacific Ocean, Can. J. Earth Sci. 8 (1971) 387–391. [6] R.L. Coles, G.V. Haines, W. Hannaford, Large scale magnetic anomalies over western Canada and the Arctic: a discussion, Can. J. Earth Sci. 13 (1976) 790–802. [7] B. Caner, Long aeromagnetic profiles and crustal structure in Western Canada, Earth Planet Sci. Lett. 7 (1969) 3–11. [8] R.L. Coles, R.G. Currie, Magnetic anomalies and rock magnetizations in the Southern Coast Mountains, British Columbia: Possible relation to subduction, Can. J. Earth Sci. 14 (1977) 1753–1770. [9] F.A. Cook, J.L. Varsek, J.B. Thurston, Tectonic significance of gravity and magnetic variations along the Lithoprobe Southern
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
[10]
[11]
[12]
[13] [14] [15]
[16]
[17]
[18]
[19]
Canadian Cordillera transect, Can. J. Earth Sci. 32 (1995) 1584–1610. C. Lowe, R.B. Horner, J.K. Mortensen, S.T. Johnston, C.F. Roots, New geophysical data from the northern Cordillera: preliminary interpretations and implications for the tectonics and deep geology, Can. J. Earth Sci. 31 (1994) 891–904. C.E. Lynn, F.A. Cook, K.W. Hall, Tectonic significance of potentialfield anomalies in western Canada: results from the Lithoprobe SNORCLE transect, Can. J. Earth Sci. 42 (2005) 239–1255. C. Lowe, R.J. Enkin, L.C. Struik, Tertiary extension in the central British Columbia Intermontane Belt: magnetic and paleomagnetic evidence from the Endako region, Can. J. Earth Sci. 38 (2001) 657–678. F.A. Cook, K.W. Hall, C.E. Lynn, The edge of northwestern North America at ∼ 1.8 Ga, Can. J. Earth Sci. 42 (2005) 983–997. A. Spector, F.S. Grant, Statistical methods for interpreting aeromagnetic data, Geophysics 35 (1970) 293–302. F.A. Cook, R.M. Clowes, D.B. Snyder, A.J. van der Velden, K.W. Hall, P. Erdmer, C.A. Evenchick, Precambrian crust beneath the Mesozoic northern Canadian Cordillera discovered by Lithoprobe seismic reflection profiling, Tectonics 23 (2004), doi:10.1029/ 2002TC001412. F.A. Cook, Lithospheric processes and products in the Southern Canadian Cordillera: a Lithoprobe perspective, Can. J. Earth Sci. 32 (1995) 1803–1824. D.B. Snyder, R.M. Clowes, F.A. Cook, P. Erdmer, C.A. Evenchick, A.J. van der Velden, K.W. Hall, Proterozoic prism arrests suspect terranes: insights into the ancient Cordilleran margin from seismic reflection data, GSA Today 12 (10) (2002) 4–10. G. Fernández-Viejo, R.M. Clowes, J.K. Welford, Constraints on the composition of the crust and uppermost mantle in northwestern Canada: Vp/Vs variations along Lithoprobe's SNorCLE transect, Can. J. Earth Sci. 42 (2005) 1205–1222. F.A. Cook, A.J. van der Velden, Three-dimensional crustal structure of the Purcell anticlinorium in the Cordillera of southwestern Canada, Geol. Soc. Amer. Bull. 107 (1995) 642–664.
485
[20] T.J. Lewis, R.D. Hyndman, P. Fluck, Heat flow, heat generation, and crustal temperatures in the northern Canadian Cordillera: thermal controls of tectonics, J. Geophys. Res. 108 (2003), doi:10.1029/2002JB002090. [21] R.D. Hyndman, C.A. Currie, S.P. Mazzotti, Subduction zone backarcs, mobile belts, and orogenic heat, GSA Today 15 (2) (2005) 4–10. [22] A. Hahn, H. Ahrendt, J. Meyer, J.-H. Hufen, A model of magnetic sources within the Earth's crust compatible with the field measured by the satellite Magsat, Geol. Jahrb., Reihe A Allg. reg. Geol. BR Dtschl. Nachbargeb. Tekton. Stratigr. Palaontol. 75 (1984) 125–156. [23] K. Hemant, S. Maus, Geological modeling of the new CHAMP magnetic anomaly maps using a geographical information system technique, J. Geophys. Res. 110 (2005) B12103, doi:10.1029/ 2005JB003837. [24] J. Arkani-Hamed, J. Dyment, Magnetic potential and magnetization contrasts of Earth's lithosphere, J. Geophys. Res. 101 (1996) 11401–11426. [25] M.E. Purucker, R.A. Langel, M. Rajaram, C. Raymond, Global magnetization models with a priori information, J. Geophys. Res. 103 (1998) 2563–2584. [26] K. Hemant, S. Maus, Why no anomaly is visible over most of the continent–ocean boundary in the global crustal magnetic field, Phys. Earth Planet. Inter. 149 (2005) 321–333. [27] A. Hahn, H.A. Roeser, The magnetization of the lower crust, in: R.F. Mereu, S. Mueller, D.M. Fountain (Eds.), Properties and Processes of Earth's Lower Crust, Geophys. Monogr. Ser., vol. 51, AGU, Washington, 1989, pp. 247–253. [28] M. Pilkington, J.A. Percival, Relating crustal magnetization and satellite-altitude magnetic anomalies in the Ungava peninsula, northern Quebec, Canada, Earth Planet. Sci. Lett. 194 (2001) 127–133. [29] S. Maus, M. Rother, R. Holme, H. Lühr, N. Olsen, V. Haak, First scalar magnetic anomaly map from CHAMP satellite data indicates weak lithospheric field, Geophys. Res. Lett. 29 (2002), doi:10.1029/2001GL013685.
Weakly magnetic crust in the Canadian Cordillera Mark Pilkington a,⁎, David B. Snyder a , Kumar Hemant b,1 a
Geological Survey of Canada, 615 Booth Street, Ottawa, ON, Canada K1A 0E9 b GeoForschungsZentrum, Telegrafenberg, 14473 Potsdam, Germany
Received 11 April 2006; received in revised form 6 June 2006; accepted 8 June 2006 Available online 11 July 2006 Editor: S. King
Abstract Current models of continental crust favor an increase in magnetization with depth. Here we report a counter example from the Canadian Cordillera where almost a full thickness of non-magnetic continental crust is suggested by joint interpretation of magnetic and seismic data. The magnetic field over the Cordillera is characterized by complex, short-wavelength (< 100 km) anomalies associated with intrusive, metamorphic and volcanic rocks that occur at shallow depths (< 5 km) within accreted terranes. The longwavelength (>100 km) portion of the Cordilleran field is subdued and mainly featureless, and suggests a lack of magnetic sources at greater depths. Seismic reflection and refraction data from three major transects in the Yukon and British Columbia, Canada support this interpretation and indicate that sedimentary-like formations make up the majority of the crust. The dominance of shallow, upper crustal magnetization in the Canadian Cordillera contrasts with the generally-held view that the lower continental crust is the primary source for long-wavelength magnetic anomalies. Sources for these anomalies are often assumed to be located in the lower crust when surface magnetizations are insufficient to produce such anomalies or no correlation exists between the magnetic field and the mapped surface geology. The Canadian Cordillera appears to be an example of a non-magnetic lower crust overlain by a more magnetic upper crust that is, however, not magnetized strongly enough to produce significant long-wavelength magnetic anomalies. © 2006 Elsevier B.V. All rights reserved. Keywords: continental crust; magnetic field; susceptibility
1. Introduction The magnetic properties of continental crust are dependent on its composition from the surface down to the Curie isotherm, specifically on the distribution of magnetite, the most commonly occurring ferrimagnetic mineral. Generally, crystalline igneous and metamorphic rocks are the main contributors to anomalous magnetic ⁎ Corresponding author. Fax: +1 613 952 8987. E-mail address: [email protected] (M. Pilkington). 1 Present address: University of Leeds, Leeds, U.K. 0012-821X/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2006.06.010
fields, whereas sedimentary rocks can be considered essentially non-magnetic. Within the crust, the division into a felsic upper crust and a more mafic lower crust has led to granulite-facies rocks in the latter to be suggested as the predominant source for long-wavelength magnetic anomalies [1]. Multi-domain magnetite is commonly inferred to be the mineral responsible for this magnetism, although certain forms of the hematite–ilmenite series could also be contributors [2]. The magnitude of crustal magnetization at upper levels within the crust can be determined from surface samples and modelling of short-wavelength magnetic anomalies. For the deeper
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
parts, xenolith magnetic properties are useful, as is the interpretation of long-wavelength magnetic anomalies, such as those measured by satellite-altitude surveys [3]. Here, we report an example of almost a full thickness of non-magnetic continental crust suggested by joint interpretation of magnetic and reflection and refraction seismic data, where magnetic sources are restricted to a thin (< 5 km) surface layer, and discuss the consequences for the global description of crustal magnetization and the modelling of satellite magnetic data. 2. Magnetic field data The Canadian Cordillera is characterized by a subdued and mainly featureless long-wavelength (> 100 km) magnetic field (Fig. 1). This is in marked contrast to the numerous, extensive, high-amplitude magnetic anomalies that are associated with exposed and buried Precambrian basement to the east and north of the Cordillera. Indeed, the boundary between the two magnetically contrasting regions closely mimics the geologically determined extent of the Cordillera, i.e., the Cordilleran deformation front (CDF, Fig. 1). The only discrepancy between the two occurs north of 65°N where the CDF turns northward to the Arctic coast (e.g., [4]) but the boundary between subdued and higher-amplitude magnetic fields continues approximately westward to the Alaskan border. The data in Fig. 1 are derived from a series of highaltitude (∼5 km) aeromagnetic surveys flown in the
477
1970s [5]. Flightline spacing averages 37 km over the area shown and the data were gridded at a 10-km interval. A cosine-tapered, low-pass filter with center roll-off wavelength of 100 km was applied to the gridded data to reduce residual navigational error effects. This filtering also suppresses anomalies due to near-surface magnetic sources and emphasizes the effects of deeper bodies occurring in the middle and lower crust. Even in nonfiltered data, Haines et al. [5] and Coles et al. [6] noted the clear distinction in magnetic character between the Cordillera and its surroundings. They suggested that the subdued field over the Cordillera was due to the absence of Precambrian crystalline basement material in the region. On the basis of several long aeromagnetic profiles, Caner [7] suggested the smooth Cordilleran field was due to shallowing of the Curie isotherm or the presence of a more felsic crust. To the east of the CDF, crystalline basement rocks making up the North American craton produce numerous, high-amplitude (hundreds of nanoteslas), laterallyextensive (hundreds of kilometers) anomalies that persist even when covered by thick sedimentary basins (e.g., > 5 km in southern Alberta). These kinds of anomalies are not seen within the Cordillera. In the longwavelength field some higher-amplitude (> 200 nT) anomalies do occur, for example, one just to the east of Vancouver Island coincident with the coastline and another centered at 127.5°W, 57.7°N. The low-altitude (∼300 m), high-resolution (1-km grid interval) magnetic
Fig. 1. High-altitude (∼ 5 km) magnetic field of western Canada filtered to remove wavelength components <100 km. CDF, Cordilleran deformation front. Data collection details are given in Haines et al. [5].
478
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
Fig. 2. Low-altitude (∼300 m mean terrain clearance) magnetic field of western Canada. CDF, Cordilleran deformation front. Dashed numbered boxes outline areas where power spectra are calculated (see Fig. 4). Details on sources and processing of individual survey data can be found at http://gdcinfo.agg.nrcan.gc.ca.
field over the Cordillera (Fig. 2) demonstrates the difference in character (and source) between these two anomalies and those seen over buried or exposed Precambrian rocks. The former are distinguished by their isolated, short-wavelength character indicating a nearsurface origin. This is the general case for the Cordilleran field, where individual anomalies are well correlated with surface geology and rock property measurements (e.g., [8–11]). The Cordilleran anomalies often have high amplitudes comparable to Precambrian basement anomaly levels but do not persist laterally, hence they do not contribute so efficiently to the longer wavelength (> 100 km) magnetic field shown in Fig. 1. For the low-altitude data (Fig. 2), individual anomalies can often be related to intrusive and volcanic outcrops, although detailed analysis of magnetic properties indicates the relationship is far from simple, with complications from alteration and thermal effects [12]. The Cordilleran field exhibits a strong northwesterly trend that reflects tectonic strike (Fig. 3). The Tintina– Rocky Mountain trench fault system divides a smooth featureless magnetic field over the Foreland Belt (ancestral North America) from the more variable field over the numerous accreted and pericratonic terranes to the west (cf. Figs. 2 and 3). In contrast to the northwesterly Cordilleran fabric, Precambrian basement trends vary from northeasterly south of 55°N, to northerly up to 65°N, where the predominant strike becomes east–west.
Similar regional differences in the strike of gravity anomalies can also be observed [11]. The persistence of high-amplitude, long-wavelength anomalies west of the
Fig. 3. Simplified terrane map of the Canadian Cordillera, after Gabrielse et al. [4]. TF-RMT, Tintina–Rocky Mountain trench fault system; CDF, Cordilleran deformation front; PA, Purcell anticlinorium. Thick grey lines are LITHOPROBE seismic profiles.
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
479
Both low- and high-altitude magnetic field data suggest that the Cordilleran crust comprises a shallow, thin, magnetic upper crustal source layer corresponding to igneous, metamorphic and volcanic rocks of various accreted terranes underlain by a weakly or non-magnetic crust. This view is supported by interpretation of recently acquired seismic reflection and refraction data acquired along traverses crossing the Cordillera, from the west coast to the ancient North American craton. 3. Seismic data
Fig. 4. Magnetic field power spectra for data within the dashed boxes in Fig. 2. Numbers refer to box locations. Thick lines indicate predicted spectral slopes [14] produced by average source depths of 1 and 5 km.
CDF in both the high- and low-altitude magnetic data indicate that cratonic North America extends westwards beneath the Cordilleran boundary. Where imaged by seismic reflection data, the edge of the craton appears as a westward facing monocline with up to 20 km of relief over distances of up to 80 km [13]. The difference in wavelength character between the field over the Cordillera and that over the surrounding regions is quantified through the power spectrum. Fig. 4 shows example power spectra for data within the dashed boxes in Fig. 2. The slope of the logarithmic magnetic field power spectrum is proportional to the average depth to the causative magnetic sources [14], hence Cordilleran regions exhibit consistently small source depths (< 5 km) compared to the buried Precambrian basement areas. Fig. 4 also indicates that for wavelengths <40 km (0.025 cycles/km) how much stronger the Cordilleran anomalies are than the Precambrian ones and vice versa for > 40 km components.
Deep seismic reflection profiles collected as part of the LITHOPROBE program form three major transects across the Canadian Cordillera (Fig. 3): one in the Yukon, and one each in northern and southern British Columbia [15,16]. The two northerly transects map a thick wedge of reflective layers that stretches from the western edge of the North American craton to accreted terranes hundreds of kilometers further west (Fig. 5). The strong continuity of reflectors within this unit has allowed the surface correlation of individual reflective layers to be mapped down to depths of ∼ 30 km [15,17]. Stratified Proterozoic units make up most of the reflective wedge, including the 0.8–0.54 Ga Windemere, 1.0–0.92 Ga MacKenzie Mountain and 1.84–1.71 Ga Wernecke Supergroups, plus the 1.81–1.5 Ga Muskwa Assemblage. Where exposed in the Foreland Belt, combined thicknesses of 7–21 km are indicated, and projections based on seismic data suggest a maximum thickness of 25–30 km. Although many reflections within these Proterozoic sedimentary groups can be tied from surface to depth, some reflective features within the lower crust could be caused by tectonically-interleaved, older crystalline continental crust. Refraction seismic data do, however, support the possibility of sedimentary rocks making up most of the crustal column, particularly in the northern Cordillera. Coincident with the proposed reflective wedge are low compressional to shear wave velocity ratios (Vp/Vs) which, based on comparison with laboratory measurements, suggest the presence of highly quartzose rocks [18]. In the southern Cordillera, similar
Fig. 5. Cartoon cross section of Cordilleran crust based on seismic reflection profiling, after Snyder et al. [17]. Pz, Paleozoic cover. Scale is omitted since this is a relationship diagram only.
480
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
thicknesses of stratified crust are observed from reflection profiles over the southern transect with over 20 km of (meta-) sedimentary rocks and associated sills making up the Purcell anticlinorium [19]. Such large thicknesses of sedimentary rocks are interpreted to have formed in a rift margin along the western edge of the North American craton that may have persisted from 1.85 to 0.54 Ga [17]. They overlie pre-1.8 Ga crystalline basement making up the edge of the North American craton in the east and form a base upon which the numerous accreted terranes were grounded to the west (Fig. 5). Importantly, both the reflection [15] and refraction seismic [18] data reveal that the northeasternmost terranes are just thin flakes thrust up onto the sedimentary wedge, while more southwesterly terranes such as Stikinia constitute most of the crust. The easternmost terranes, such as Cache Creek and Quesnelia (Fig. 3), were detached from their roots and thrust 200–400 km onto the wedge. Since stacking of terranes is unlikely, the order of their accretion is suggested by their current location. 4. Magnetic character of Cordilleran crust Volumetrically, the Cordilleran crust is dominated by the reflective Proterozoic sedimentary sequence. The remainder consists of westward-thinning North American craton (deep crust), thin accreted terranes (shallow crust) and thick attached terranes (whole crust; Fig. 5).
Magnetically, the sedimentary package is not significant because of the weak magnetizations (commonly <0.01 A/m) of these rock types. Where these strata are exposed, subtle (< 10 nT) anomalies can be detected by high-resolution, low-altitude (i.e., 300 m or smaller) aeromagnetic surveys, but these effects are quickly attenuated at higher elevations and have a negligible contribution at wavelengths greater than a few kilometers. Based on the seismic reflection data, it is possible that more magnetic intrusive rocks, in the form of sills, may be present within the Proterozoic strata. Nevertheless, because of the gentle dip (∼ 10°) and small thickness (<< 1 km) of any sills present, they are not expected to produce significant magnetic anomalies. For example, a 300-m thick sill dipping at 10° with a 1 A/m magnetization only produces a 10 nT anomaly even when exposed. Beneath the Proterozoic sedimentary strata is the westerly-tapering edge of the North American craton, probably consisting of pre-1.8 Ga crystalline basement. This appears as a poorly reflective zone at depths of >25 km along the Yukon and northern British Columbia seismic lines [15]. Further south, the basement is also less reflective than overlying strata, and appears to taper down from 15 km depth at the Rocky Mountain trench to > 25 km depth beneath the Purcell anticlinorium [16]. Measurements of heat flow in the Cordillera are consistently high and indicate Moho temperatures of 800– 900 °C [20]. This puts the Curie isotherm (580 °C) in the
Fig. 6. Satellite magnetic data for western Canada from the CHAMP mission [29]. Altitude average is 400 km. MR = Mackenzie River anomaly. ABC = Alberta–British Columbia anomaly. NWP = North Wyoming Province anomaly.
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
20–25 km depth range, based on an estimated 32– 35 km crustal thickness [21]. Consequently, the majority of the North American cratonic material within the Cordillera lies beneath the Curie isotherm and is nonmagnetic. The negligible effect of the buried Canadian shield beneath the Cordillera is also expected from the truncation of south-westerly trending basement-sourced anomalies around the position of the Rocky Mountain trench in southern British Columbia (Fig. 2). Both thick, attached and thin, accreted terranes contribute significantly only to the short-wavelength (< 100 km) magnetic field (cf. Figs. 2 and 4). Their longwavelength effect is minor and restricted to the coastal anomaly near Vancouver Island (Fig. 1) related to the Cascadia subduction zone [8]. This character is quantified by the example power spectra for these terranes which suggest shallow sources confined to upper crustal levels, i.e., less than ∼5 km deep. Thus the remainder of the crust appears essentially non-magnetic. Such a large lateral extent (600 × 2000 km) of weakly magnetized crust would be expected to produce a detectable signature in satellite-altitude magnetic data. Fig. 6 shows that this is the case, with a low-amplitude field characterizing the continent west of the CDF. This contrasts with higher amplitudes to the north and east, presumably associated with greater thicknesses of more magnetized crust occurring in cratonic North America. Prominent positive anomalies occur over the Mackenzie River region (MR, Fig. 6), northern Alberta and British Columbia (ABC) and to the north of the buried Wyoming Province in the United States (NWP). The contribution of Cordilleran crust to the satellite field is significantly modified by the effects of the measurement altitude; coalescence of the effects of magnetized sources in the crust stretches over distances equivalent to the satellite altitude, 400 km. As a result, even highly magnetic regions such as those within the craton are not always associated with large (positive) anomalies at satellite altitudes. Only those areas where large amplitude fields (> 400 nT in the data of Fig. 1) persist over distances comparable to satellite elevations produce detectable anomalies. For example, the Great Bear magmatic arc (striking north–south at 117°W between 62 and 67°N), which produces a significant response in the low-altitude data sets (Figs. 1 and 3), has almost no effect in the satellite field (Fig. 6). Only the Mackenzie River and the combined N.W. Alberta and N.E. British Columbia group of anomalies (Fig. 1) have large enough amplitudes and extents to contribute to the satellite field. Therefore the weakly magnetic crust within the Cordillera is not so clearly manifest in satellite data compared to the lower altitude data sets due to the
481
large difference in source-sensor distances, i.e., 400 km compared to ∼ 5 km and 0.3 km. 5. Discussion Excepting the thin shallow source layer of the exotic terranes, the crust within the Canadian Cordillera is weakly magnetic. This observation provides an independent constraint on regional-scale models of crustal magnetization, either those constructed for forward modelling purposes (e.g., [22,23]) or those resulting from inversion of magnetic field (commonly satellitealtitude) data (e.g., [24,25]). The construction of global crustal magnetization models has been simplified and improved through available global databases, e.g., Moho depth, sedimentary basin thickness, regional geology. These commonly have resolutions of 2–5° (200–500 km) which are comparable to that of satellite magnetic data. At such scales, crustal divisions based on tectonic setting represent necessarily broad averages. Nonetheless, anomalously low magnetization levels for practically the whole of the Canadian Cordilleran crust, with an extent of 600 × 2000 km, are not expected from knowledge of its surface geology. Consequently, forward models should not be based solely on the expected magnetic properties of a given tectonic unit, e.g., craton, mobile belt, assigned from surface geologic mapping, but should also be compatible with crustal magnetization values inferred from lower-altitude magnetic data and other geophysical data sets. Hemant and Maus [23] have constructed a global magnetization model based on worldwide compilations of geology, seismically-derived crustal thickness, and magnetic properties of rocks. Contrary to the forward models of Hahn et al. [22] and Purucker et al. [25] which assign a constant susceptibility to a limited number of geological classes, the Hemant and Maus [23] model contains a unique susceptibility value for each geological region. They assume that the Moho separates a magnetic crust from a non-magnetic upper mantle and that the continental crust is dominated by induced magnetization. Based on compilations of rock properties, magnetizations are assigned to surface rock types which may then be averaged within a given tectonic province. These values are assumed to apply to the upper crust while a multiple of these (greater than unity) based on crustal age is assigned to the lower crust. Finally, a verticallyintegrated susceptibility (VIS) value is determined from the combined crustal thickness and susceptibility information for each region (Fig. 7). One advantage of forward models based on purely geologic information is the possibility of specifying a high
482
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
Fig. 7. Vertically-integrated susceptibility (VIS) map for western Canada after Hemant and Maus [23].
degree of complexity of their parameterization. Areas smaller than those resolvable from satellite-altitude magnetic data can be parameterized and the boundaries between adjacent regions can be, and are usually, abrupt, compared to heavily smoothed transitions derived from inversions of satellite data. Fig. 8 shows the calculated satellite-altitude field based on the initial VIS model (Fig. 7) of Hemant and
Maus [23]. The match between the field predicted from surface geology and the observed field is reasonable in terms of spatial location of the major anomalous features and their amplitudes. Two of the three major positive anomalies within the study area are well-defined in the predicted map, the ABC anomaly (Fig. 6), caused by buried magmatic arc complexes, being more subdued.
Fig. 8. Predicted magnetic field at 400 km altitude based on the VIS data in Fig. 7.
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
483
Fig. 9. Modified, vertically-integrated susceptibility (VIS) map for western Canada.
The prominent low extending from 49°N to 65°N over most of the Cordillera is also reproduced in the calculated field, but its amplitude is smaller than observed in Fig. 6. Hence the assigned VIS values are too high. VIS values specified for most of the Cordillera (Fig. 7) were based on assuming a Phanerozoic-age crust giving a mean value of 0.434 SI km. This results in a small but still detectable
contrast with the adjacent oceanic crust, which based on its young age, has been assigned a similar value of 0.385 SI km [26]. The similarity in crustal magnetization levels between the Cordillera and oceanic regions is demonstrated by the lack of level change in the longwavelength magnetic field measured at lower altitudes over these areas (Fig. 1) even though the short-wavelength
Fig. 10. Predicted magnetic field at 400 km altitude based on the modified VIS map in Fig. 9.
484
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
character is different (Fig. 2). This similarity is not restricted to comparable VIS values, but is also a result of the comparable depth variation in magnetization for Cordilleran and oceanic regions, both consisting of thin (∼5 km) magnetic surface layers. As a first-order adjustment to the forward model to incorporate inferences from the seismic data, we reduced the VIS values over the whole extent of Cordilleran crust (Fig. 9). The entire Cordillera was assigned the same VIS value simply to group together those regions expected, from seismic information, of having a non-magnetic middle and lower crust. A value of 0.350 SI km, determined by trial and error, resulted in the best fit to the observations. The field produced by this adjusted model is shown in Fig. 10. The dominant magnetic low over the Cordillera in the satellite field is now better matched, although the boundary with the North American craton now produces a stronger effect. Clearly, adjustment of VIS values at scales smaller than the Cordillera is needed to improve the fit, particularly for the area of northern Alberta and British Columbia (ABC anomaly, Fig. 6). Our aim, however, is not to reach a final model, but to show that forward models of this type can be improved with guidance from geophysical information. The value of 0.350 SI km is within, but at the lower end of, the range of assigned VIS values for Phanerozoic crust worldwide [26]. In the Canadian Cordillera, the thin (<5 km) magnetic surface layer underlain by non-magnetic sedimentary rocks, shows the possible complexity of the depth variation of magnetization within continental crust which may persist over large lateral extents (100's km). It emphasizes the limitations of interpretations based solely on satellite magnetic data inversions, specifically the lack of depth discrimination for derived source distributions. This leads to estimates of only the depthintegrated magnetization of the crust and hence ambiguity regarding rock types within the crustal column and even information on the crustal type. Forward geological modelling, however, allows much greater detail in crustal models, but is limited by our lack of knowledge of crustal structure, composition and magnetic properties in many areas. Any successful approach to modelling crustal magnetization will, at the least, have to satisfy constraints from a priori geologic information plus those from magnetic inversions (measured at satellite altitude or lower). The dominance of shallow, upper crustal magnetization in the Canadian Cordillera also contrasts with the commonly-held view (e.g., [1,27]) that the lower continental crust is the primary source for long-wavelength magnetic anomalies. Sources for these longwavelength anomalies are often assumed to be located
in the lower crust when surface magnetizations are insufficient to produce such anomalies or no correlation exists between the magnetic field and the mapped surface geology. The Cordillera is an example of a non-magnetic lower crust overlain by a more magnetic upper crust that is, however, not magnetized strongly enough to produce significant long-wavelength magnetic anomalies. Predominant upper crustal magnetization is also present within the Ungava peninsula in eastern Canada, but here the magnetization levels are high enough to produce detectable long-wavelength anomalies measured both in aeromagnetic and satellite data [28]. Geologically, this area differs greatly from the Cordillera, being part of the North American craton and comprising mainly Archean plutonic rocks with some high-grade gneisses and small greenstone belts. Clearly, various scenarios must be considered possible when making even a simple division of continental crust into an upper and lower unit with different magnetization levels. Acknowledgments This is Geological Survey of Canada contribution 2005010. Mike Thomas and Carmel Lowe are thanked for providing useful reviews. References [1] P.N. Shive, R.J. Blakely, B.R. Frost, D.M. Fountain, Magnetic properties of the lower continental crust, in: D.M. Fountain, R. Arculus, R.W. Kay (Eds.), Continental Lower Crust, Elsevier, New York, 1992, pp. 145–177. [2] S.A. McEnroe, F. Langenhorst, P. Robinson, G.D. Bromiley, C.S. J. Shaw, What is magnetic in the lower crust? Earth Planet. Sci. Lett. 226 (2004) 175–192. [3] R.A. Langel, W.J. Hinze, The Magnetic Field of the Earth's Lithosphere: The Satellite Perspective, Cambridge Univ. Press, New York, NY, 1998. [4] H. Gabrielse, J.W.H. Monger, J.O. Wheeler, C.J. Yorath, Morphogeological belts, tectonic assemblages and terranes [chapter 2: tectonic framework], in: H. Gabrielse, C.J. Yorath (Eds.), Geology of the Cordilleran Orogen in Canada, Geology of Canada Series, vol. 4, Geol. Surv. of Can., Ottawa, 1991, pp. 15–28. [5] G.V. Haines, W. Hannaford, R.P. Riddihough, Magnetic anomalies over British Columbia and the adjacent Pacific Ocean, Can. J. Earth Sci. 8 (1971) 387–391. [6] R.L. Coles, G.V. Haines, W. Hannaford, Large scale magnetic anomalies over western Canada and the Arctic: a discussion, Can. J. Earth Sci. 13 (1976) 790–802. [7] B. Caner, Long aeromagnetic profiles and crustal structure in Western Canada, Earth Planet Sci. Lett. 7 (1969) 3–11. [8] R.L. Coles, R.G. Currie, Magnetic anomalies and rock magnetizations in the Southern Coast Mountains, British Columbia: Possible relation to subduction, Can. J. Earth Sci. 14 (1977) 1753–1770. [9] F.A. Cook, J.L. Varsek, J.B. Thurston, Tectonic significance of gravity and magnetic variations along the Lithoprobe Southern
M. Pilkington et al. / Earth and Planetary Science Letters 248 (2006) 476–485
[10]
[11]
[12]
[13] [14] [15]
[16]
[17]
[18]
[19]
Canadian Cordillera transect, Can. J. Earth Sci. 32 (1995) 1584–1610. C. Lowe, R.B. Horner, J.K. Mortensen, S.T. Johnston, C.F. Roots, New geophysical data from the northern Cordillera: preliminary interpretations and implications for the tectonics and deep geology, Can. J. Earth Sci. 31 (1994) 891–904. C.E. Lynn, F.A. Cook, K.W. Hall, Tectonic significance of potentialfield anomalies in western Canada: results from the Lithoprobe SNORCLE transect, Can. J. Earth Sci. 42 (2005) 239–1255. C. Lowe, R.J. Enkin, L.C. Struik, Tertiary extension in the central British Columbia Intermontane Belt: magnetic and paleomagnetic evidence from the Endako region, Can. J. Earth Sci. 38 (2001) 657–678. F.A. Cook, K.W. Hall, C.E. Lynn, The edge of northwestern North America at ∼ 1.8 Ga, Can. J. Earth Sci. 42 (2005) 983–997. A. Spector, F.S. Grant, Statistical methods for interpreting aeromagnetic data, Geophysics 35 (1970) 293–302. F.A. Cook, R.M. Clowes, D.B. Snyder, A.J. van der Velden, K.W. Hall, P. Erdmer, C.A. Evenchick, Precambrian crust beneath the Mesozoic northern Canadian Cordillera discovered by Lithoprobe seismic reflection profiling, Tectonics 23 (2004), doi:10.1029/ 2002TC001412. F.A. Cook, Lithospheric processes and products in the Southern Canadian Cordillera: a Lithoprobe perspective, Can. J. Earth Sci. 32 (1995) 1803–1824. D.B. Snyder, R.M. Clowes, F.A. Cook, P. Erdmer, C.A. Evenchick, A.J. van der Velden, K.W. Hall, Proterozoic prism arrests suspect terranes: insights into the ancient Cordilleran margin from seismic reflection data, GSA Today 12 (10) (2002) 4–10. G. Fernández-Viejo, R.M. Clowes, J.K. Welford, Constraints on the composition of the crust and uppermost mantle in northwestern Canada: Vp/Vs variations along Lithoprobe's SNorCLE transect, Can. J. Earth Sci. 42 (2005) 1205–1222. F.A. Cook, A.J. van der Velden, Three-dimensional crustal structure of the Purcell anticlinorium in the Cordillera of southwestern Canada, Geol. Soc. Amer. Bull. 107 (1995) 642–664.
485
[20] T.J. Lewis, R.D. Hyndman, P. Fluck, Heat flow, heat generation, and crustal temperatures in the northern Canadian Cordillera: thermal controls of tectonics, J. Geophys. Res. 108 (2003), doi:10.1029/2002JB002090. [21] R.D. Hyndman, C.A. Currie, S.P. Mazzotti, Subduction zone backarcs, mobile belts, and orogenic heat, GSA Today 15 (2) (2005) 4–10. [22] A. Hahn, H. Ahrendt, J. Meyer, J.-H. Hufen, A model of magnetic sources within the Earth's crust compatible with the field measured by the satellite Magsat, Geol. Jahrb., Reihe A Allg. reg. Geol. BR Dtschl. Nachbargeb. Tekton. Stratigr. Palaontol. 75 (1984) 125–156. [23] K. Hemant, S. Maus, Geological modeling of the new CHAMP magnetic anomaly maps using a geographical information system technique, J. Geophys. Res. 110 (2005) B12103, doi:10.1029/ 2005JB003837. [24] J. Arkani-Hamed, J. Dyment, Magnetic potential and magnetization contrasts of Earth's lithosphere, J. Geophys. Res. 101 (1996) 11401–11426. [25] M.E. Purucker, R.A. Langel, M. Rajaram, C. Raymond, Global magnetization models with a priori information, J. Geophys. Res. 103 (1998) 2563–2584. [26] K. Hemant, S. Maus, Why no anomaly is visible over most of the continent–ocean boundary in the global crustal magnetic field, Phys. Earth Planet. Inter. 149 (2005) 321–333. [27] A. Hahn, H.A. Roeser, The magnetization of the lower crust, in: R.F. Mereu, S. Mueller, D.M. Fountain (Eds.), Properties and Processes of Earth's Lower Crust, Geophys. Monogr. Ser., vol. 51, AGU, Washington, 1989, pp. 247–253. [28] M. Pilkington, J.A. Percival, Relating crustal magnetization and satellite-altitude magnetic anomalies in the Ungava peninsula, northern Quebec, Canada, Earth Planet. Sci. Lett. 194 (2001) 127–133. [29] S. Maus, M. Rother, R. Holme, H. Lühr, N. Olsen, V. Haak, First scalar magnetic anomaly map from CHAMP satellite data indicates weak lithospheric field, Geophys. Res. Lett. 29 (2002), doi:10.1029/2001GL013685.