Continental Tectonics (1980)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

IV CHARACTERIZATION OF CONTINENTAL CRUST

Seismic Exploration of the Continental Basement: Trends for the 1980's 10 INTRODUCTION JACK E. OLIVER Cornell University In scientific research it is helpful to view one's activities occasionally from a fresh and different perspective. Imagine, for example, an astronaut-scientist visiting the earth from an advanced civilization on another planet and having the task of reporting the sate of earthly science to his leaders. I think his report would be mixed in tone. He would report favorably about some of our efforts to inves- tigate our surroundings. On one scale, for example, space- craft leave the earth to explore the solar system and beyond in an effort that strains our technology. At another extreme of scale, huge sophisticated devices cause tiny subatomic particles to collide at high velocity in an effort to learn more and more about smaller and smaller entities. Once again our technology is strained. The astronaut would probably be favorably impressed by progress in certain methods for exploring the earth sophisticated laboratory devices, the techniques of the petroleum industry, perhaps deep-sea drilling vessels. But I think he would be surprised and dismayed to find that a society of four billion people confined to the surface ofthe earth has been content to know so liKle about the rocks a few ~7 hundreds or thousands of meters below it, and from which it derives much of its livelihood the rocks of the conti- nental basement. In making this statement, I do not mean to be critical of those scientists who work very capably and professionally on this topic but rather of the magni- tude ofthe total effort directed toward study ofthis part of the earth, an effort that seems too small in relation to the need for an inventory of its resources by an expanding and ravenous society. I hope new advances within the next decade will ease the mind of the astronaut on this point. It is not that methods and tools for exploration of the continental basement are unavailable. We have them; many of them will be discussed in this paper. The seismo- logical methods that I shall discuss are but a part of our overall capability. Others include further mapping, im- proved and extended field and laboratory studies of sur- face rocks, drilling for informational purposes, study of crustal xenoliths, and a variety of geophysical techniques. The problem is one of focusing scientific interest on the topic and of devoting an appropriate portion of our efforts to study of this region, for sound economic as well as scientific reasons.

118 Seismology has been, of course' a major producer of information on the earth's interior~the major producer of certain kinds of information including structure and cer- tain mechanical properties. Therefore, ~ wish to consider the potential of further seismic exploration of the conti- nental basement. By continental basement I mean the entire continental crust below the sediments and Me uppermost mantle. From the broadest perspective and for several reasons, it is clear that the potential of the seismic method for exploration of the earn, and particularly the continental basement, is by no means exhausted. First, to do so would require sources and receivers scattered over and through the earth at the Nyquist spacing for the shortest waves that can be detected after propagation through the deep region of interest. We are certainly far from achieving such a level of observations at this time and from predicting what we would observe if we did. Second, we lag in analysis; we are not able to understand and to make useful conclu- sions from all the information we obtain now. Third, the current rate of discovery of new features continues to be high. From these three points, one can deduce conf~- dently that a much improved understanding of the earth's interior remains to be revealed by seismology. The chal- lenge of seismology is to approach this ideal in an opti- mum manner given economic and other practical con- straints. In a sense, the various kinds of seismological studies represent venous routes toward this goal. Let us consider some ofthem, more or less in order of decreasing wavelength. SPECTRUM OF SEISMIC WAVELENGTHS Since 1960, seismologists have been observing and stucly- ing waves of very long periods (up to almost an hour) and hence very long wavelengths that may be thought of as corresponding to free oscillations of the earth. Most studies of free oscillations treat the earth as almost spheri- cally symmetrical (they include flattening and rotations. The resolution of differences between continents and oceans is very limited and the differences are averaged out in most cases. If some of the very-high-mode, i.e., sho~t-waveleng~, Dee oscillations can be observed, identified, and resolved in sufficient detail, information on gross structural differences between continents and oceans including associated mantle structure may be provided. Many such higher-mocle oscillations are alcin to their traveling counterparts, seismic surface waves. Traveling surface waves offer the opportunity for detennin~ng earn structure based on one pass of the waves, as opposed to many passes with consequent averaging with other effects in the case of free oscillations. Surface waves have been used regularly in recent years to.detennine cn~staI and upper-mantle structure, including regional variations of such structure. Many of the measurements of depth to the top of the low-velocity zone in the mantle, which is pre- sumably though not necessarily near or related to the base JACK E. OLIVER of the lithosphere. are derived from surface waves. Al- though the results are sometimes ambiguous and very dependent on the validity of certain assumptions, there is probably a good deal more information about the earth including the continental basement, to be obtained from surface waves. To utilize surface-wave data for conti- nental structure will require, as a minimum, more complete and more closely spaced observations of the phenomenon, farther attention to focusing and multi- pathing, integration of surface-wave observations with those of Other phenomena, and further development of techniques for using models that depart from flat-lying layered structures. In fact, the inadequacy of spherical or flat-layered models that portray the earth as lacking in lateral hetero- geneity is growing in importance and has probably reached the crisis stage in the case of the continental basement. Our level of understanding has reached the point where refinement of such models may be more mis- leading than informative. As a first approximation, geo- physicists have utilized simplified layered models, not only in the case of seismic surface waves but also in many other kinds of seismic studies and in other branches of geophysics as well. This approach is not without good reason. Gravity is an important factor in geology, and hence there is indeed a strong tendency for spherical layering. Furthermore, layered models or one-dimen- sional models are relatively easy to handle Mom the theo- retical point ofview. Such simple models are no longer an adequate representation ol the continental crust, where a bewildering variety of structures and rock types is found in almost any large outcrop (Figure 10.~. Figure 10.1, taken from a paper by Smithson et al. (1977), but in turn borrowed from Berthelsen (1960), shows pyroxene granu- lite layers surrounded by granitic gneiss. The important point in the present context has to do with the complex three-dimensional structure of this feature. Resolution of such complex buried contortions by seismic methods, or any methods, will be difficult, but on the other hand to apply flat-lying layered approximations to such structures is nonsense. We must develop models and observational and analytical techniques that will provide information an structures of complexity greater than that of simple layers and to the level of complexity shown in Figure 10.1, if possible. Such structures must be probe c] at depth within the crust, and perhaps the upper mantle. Seismologists have been moving away gradually from simple-layered models for some time. Plate tectonics was a step in this direction. Perhaps more than anything, study of lateral variations and complex structures will charac- terize the application of seismology to the study of the continental basement during the 1980's. Continuing through the spectrum to shorter wave- lengths, consider the body waves generated by earth- quakes, typically in the range of a few seconds per cycle to a few cycles per second. The body-wave travel-time method, which has produced so much of our knowledge of the earth's deep interior, has been applied widely on a crustal scale for study of the continents. However, the

Seismic Exploration of the Continental Basement FIGURE 10.1 A three-dimensional structure in the continental basement. To approximate such structures by [lat- layered models is of very limited value. From Smithson et al. ( 1977) (taken from Berthelsen, 1960). IBM 1~ traditional, so-called near-earthquake studies have not produced much new information recently, partly because superior precision and flexibility may be had by use of artificial sources and partly because of the limitations of the flat-layered models usually used in interpretation of near-earthquake studies. In recent years there have been some significant depar- tures from the well-wom path, however. Tightly spaced detection networks have produced more precise locations of sources in time and space and more reliable travel-time data. Models involving lateral variations of structure such as faults can be considered using powerful computer- based methods. Techniques involving differential travel times at networks of stations encompassing a particular feature have led to investigation of features of unusual shape' the work, for example, of Aki et al. (1977) and the group at the U.S. Geological Survey (USGS) flyer, 1973; Ellsworth and Koyangi, 1977~. On the basis of the clear signs of life in current study of body waves, and the renewed emphasis that is likely to result from new efforts to predict earthquakes and reduce the earthquake hazard and to detect and identify nuclear explosions, one can foresee new results, particularly those emphasizing lateral variations, over the next decade at least. The keys for application ofthree-dimensional inver- sion of travel times are tightly spaced, probably movable, networks and developments in methods for using not only travel-time differences but also wave character resulting from attenuation, focusing, conversion, or other effects. NATURAL SEISMIC SOURCES In the case of natural seismic sources, perhaps as much information can be obtained from the sources as from the wave propagation. Historically, with each increase in pre- cision of location of hypocenters of earthquakes, our un- ~9 ~,~ 1.1 . ~ ~ it' _ _ derstanding of tectonics has grown. For example, in the 1950's and earlier, hypocenters of earthquakes in the southern hemisphere were frequently mislocated by more than 100 km. At that stage, an earthquake could at best be associated with a major regional feature such as an island arc. In the 1960's, with the advent of the World Wide Standardized Seismic Network (WVItSSN) and other advances, it became possible to locate hypocenters with a precision of less than 10 or 20 km. Earthquakes could then be associated with tectonic features of smaller scale- a rift valley, a deep earthquake zone, the outer wall of a trench. Now, with the closely spaced observing networks that are available in a few areas, earthquake hypocenters can be located with a precision of less than a kilometer, much less in some cases. Hence we can now confidently asso- ciate earthquakes with particular faults, and also associate properties ofthe quake with properties ofthat fault. Fault- plane solutions tell us of the orientation of the fault and the movement; other focal mechanism data tell us of the change in stress and scale of the movement. Further improvements in precision of hypocentral loca- tion may tell us about movements along a particular part of a fault, progression in fault activity, and complexities such as asperities and slices. Figure 10.2 shows how an interpretation of tectonics may be strongly dependent on the precision of hypocentral data. This figure is taken from a paper by Barazangi and Isacks (in press). A cross section of the seismicity through central and northern Peru is shown twice. In the upper half of the figure only hypocenters of very high precision are plotted. In the lower half of the figure other hypocenters located during the same time period but with lower precision are shown. There is clearly a great contrast between the structure defined by the well-located hypocenters and what might be deduced from the less well-Iocated hypocenters. A similar effect may be anticipated at smaller scale. Other modem techniques are telling us of fault move-

120 meets that are slower than the abrupt displacements of typical earthquakes. Such slow movements may fail to generate short-period seismic waves, may generate only very long-period seismic waves, or may generate no detectable seismic wave and may be thought of as epi- sodes of fault creep. Such studies enhance greatly our understanding of fault motion. The study of seismicity and sources is an area of vast potential and one in which integration of seismology and geology is likely to be crucial. One may confidently state that interaction of seismol- ogy, or geophysics in general, and geology is likely to increase markedly in the next decade. A recent report by the NRc Committee on Seismology (1976) notes that much of the progress in unclerstanding the problem of earth- quake hazards in recent years has come from geological, not purely seismological, evidence. Plans to drill deeply into the San Andreas Fault are afoot. We can anticipate growing interaction between these disciplines, with mutual benefit as our understanding of the continental basement grows. ARTIFICIAL SEISMIC SOURCES In general, artificial sources have the a~lvantage of precise timing, simplicity and control of source Unction, and flexibility of location and the disadvantage of lower en- ergy except in the special case of nuclear explosions. Thus, nonnuclear artificial sources are currently of no value for study ofthe deep parts ofthe earth's interior, but for the continental crust ant! the upper mantle they can indeed provide information, ant! in fact the infonnation with the best resolution at the level of detail that we now are seeking. Seismic studies using artificial sources fall loosely and somewhat arbitrarily into three classes: (1) refraction, (2) deep seismic sounding (DSS) or refraction and wide-angle reflection, and (3) reflection profiling. The refraction rnetho~ is responsible for much of what is known about the structure of the deep crust; it provides the depth to the Mohorovicic discontinuity and typically a simple model of crustal structure consisting of layers of different velocities. Seismic refiaction studies have been carried out in the United States by venous university and private groups and the USGS. However, almost all of this work was clone in the 1960's and before. Figure 10.3, from Prodehl (1977), shows a summary of crustal models deduced from refraction data for venous parts of the United States. Nearly flat-lying layered models prevail for each area, and the differences from one area to another are illustrated at the bottom of the figure. There is considera- ble similarity in all the crustal models, and, in fact, the rather uniform depth of the mantle throughout the conti- nent is a rather striking feature. There are also substantial and measurable variations from one province to another on this gross scale. This figure is a summary of current knowledge of U.S. crustal structure based on refraction JACK E. OLIVER data. Surprisingly, for about the last decade there has been relatively little activity of this type directed toward study of the continental basement In the United States. Abroad, however, the stow Is different. The Soviet Union has operated a program of crustal exploration at a very high level of activity since World War II and has developed the DSS method, in essence the use of re- fracted and wide-angle reflected waves observed with closely spaced stations, to determine crystal structure with resolution better than that obtained by refraction measurements alone. Furthennore and after a somewhat slower start, groups of European seismologists have been using similar techniques and obtaining detailed and abundant results. Figure 10.4 shows a typical set of seis- mic refraction data from Europe arid a crustal model deduced Tom the data. Detailed velocity{lepth function and a great deal of other seismic information are obtained that is not explained by the simple model. The work in the Soviet Union and in Europe is typically characterized by much more thorough observation through very closely spaced detecting stations than is the case for the older studies of this type in the United States. The papers in Giese et al. (1976) provide a comprehensive survey of explosion seismology in central Europe. Further seismic retraction studies in the United States will surely provide new information on the crust The extent to which the new information will be coupled with other geological and geophysical information to provide enhanced under- standing of the continents will depend on He further cle~relopment of methods of interpretation to produce more realistic crustal models than the present simple layered configurations. The models should present a more realistic geological picture. The newcomer on the scene of seismic exploration of the continental basement is seismic reflection profiling. In venous countries, within the last decade, this tech- nique has been applied with some modification to study of basement to depths as much as 50 or so kilometers, primarily by the Consortium for Continental Reflection Profiling (COCORP) project in the United Sates (see Oliver et al., 1976, for a review). The method uses closely spaced vibratory or explosive sources, listening arrays of thou- sands of geophones, and sophisticated computer analysis of the data. Typically, tens of millions of rays penetrating the earth are sampled and utilized at each site. The potential resolution of structural features ofthe deep crust through reflection profiling is greater than that of any other method, but it is no small operation. Figure 10 ~ shows a field party in action. Of particular interest is that the sources of the seismic waves are not explosives but large truck-mounted vibrators. The V=ROESIS method (registered trademark of the Continental Oil Company) has been used in all the cOcORP studies to date. The re- sults Tom seismic reflection profiling so far clearly indi- cate vast potential for this method in fixture studies of the continental basement. Figure 10.6 shows a seismic reflection profile across the Rio Grande nR. The abscissa is distance in decameters;

Seismic Exploration of the Continental Basement 5 TRENCH COAST ANDES ~ AL 7~ ' 300 ' 'U: su 5 it, I ~ , , ~ IT ~ , ~ - 70C 100: 200L SECTIONS 2,3 AND 4 | CLASS a AND B I o o Loo ° deco ° - 7 OC o ~ o ~ ~ cat i ~~° °8 ~ 0 ~ ° ~ ~` 'en ~ °°C ~ — C ~ of oC ~ ~ ~ ' - - ., ! 1 BLOC 2, 3 4 200 400 600 ~ 00 ~ ~ ~ ~ o _ =~ `~ A i ~ ~ - ~ ~ A ~ ~ ~~ I ~ ~ -— ~ 00 ~~ ~ s~ - V amp ~ °-° ^° tOC 200 c SECTIONS 2 ,3 AND 4 o | CLASS C AND D I ~ I ! 200 300 ~ -- 00 121 ~ of NO 3 ~ o o o I o 400 DISTANCE, KM. 1100 [20C 6 1° - - 1 - 1300 FIGURE 10.2 A generalized section through Peru showing by hypocentral locations of high precision in upper view and other hypocentral locations of lower precision in lower view. Combining these two sets of data can result in an interpretation different from one based only on the better data. From Barazangi and Isacks (in press). FIGURE 10.3 A summary by region of seismic refraction studies of the U.S. continental crust. Note (1) overall simi- larity, (2) variations from one province to another, (3) use of layered models. From Prodehl (1977, copyrighted ' American Geophysical Union). - ~ ,/' T +~ - ~ interior Plains / ~ r Atlantic Plain 04~' O. ~°~ 0 ~ 2 3 ~ 5 _ ~~ _ lo Velocity tams) 6 ? 8 9 to

122 - -2 FIGURE 10.4 Typical record sec- tion, travel-time curves, and flat- layered model based on refraction study. From Mueller ( 1977, copy- righted by American Geophysical Union). 45 ~ ~ 5 B 7 the ordinate is two-way travel time in seconds. To convert time in seconds to depths in kilometers approximately, multiply by 3. In a gross sense such a section may be thought of as a section of the eat, but a detailed interpre- tation of the data requires knowledge of wave propagation and the data processing involved, so that it is invalid to assume that the details of a time section such as the one shown are necessarily or precisely features of the deep earn. This section, for example, has not yet been sum jected to migration to position the reflections in their proper spatial locations. Even so, certain gross features can be seen in the data that are illustrative of the capabil- ity of the method. The sedimentary sections and the sedi- mentary basement boundary are clear near the top of the section. There is an intragraben horst of substantial size and other more subtle evidence for faulting of the sedi- ments. At the time of about 7~ see there is a rather strong reflector that corresponds to the top of the magma body first proposed for this area by Sanford et al. (1977) on the FIGURE 105 Seismic reflection pro filing park in the field in Wyoming. JACK E. OLIVER ._ ~ .~ _ ' 101) tSO 200 250 ~(km) STEINBRUNN-SW 8 ~ Vpiltm/ - c) basis of microearthquake data. Although this particular section does not show much information at a time corre- sponding to the crust~nantle boundary, other sections do. Figure 10.7 shows a line drawing of the data from the section in Figure 10~6 and another section to the east. The hvo sections together span the rift valley in the vicinity of Belen, New Mexico. The top of the basement and various sedimentary features are clear, as are certain other fea- tures within the basement itself. The magma body shows as a strong reflector in parts of both sections. At a time of about 12 see in the easternmost section there is an arrival that may be associated with the crust~nantle boundary, although it is not continuous for a very long distance. Other profiles in the Rio Grande no area show a stratified pattern to the reflectors in the vicinity of this boundary. Much more detail can be found throughout the section in the original data. Figure 10.8 shows a line drawing of a short section talcen as a test of the method in the vicinity of the San am; ~,= - , ~ . _ ... , .. _, ., — ... 1 .% .-g -;~ _ - ILL ~ _

Seismic Exploration of the Continental Basement VP NO. 400 3SO 1 1 FIGURE 10.6 COCORP seismic reflec- tion profile (unmigrated) across west- em part of Rio Grande no north of Socorro. Abscissa Is vibration point number or distance in decameters; or- dinate is two-way travel time in sec- onds. Velocity structure from a nearby refraction study shown in column on left. Courtesy Larry Brown, Comell University. Andreas Fault. There is great contrast in the data from one side of the fault to the other. In the fault zone, to a depth of about 10 or 12 km, diffraction hyperbolas, associated with discontinuities marking the fault zone, are seen. Below Mat zone, however, there is a region of little infor- mation, suggesting that the zone was not penetrated by the seismic waves, which would be surprising in view of SOCORRO Ll N E lA w WHO SOCORRO Ll N E I A 123 2s0 200 1 ~ 50 100 1 1 so 10 1 1 o ~5 We information from much greater depths on both sides, or that the zone is so distorted through flow that no coherent reflected energy was obtained. The latter seems more probable. At a depth near the crust-mantle boundary, particularly in the western part of the section, there are many reflectors, suggesting a complex feature for this boundary. ABO PASS Ll NE I 400 5.S ~ - - A,-_ ~ ~ Ad. 5 _ 6. 7.5 _~ . N~ — ^ 200 150 100 so 10 10 5c '~,=~ ' -` ~ _ . ~ - f VP ~ 00 150 200 250 ~-~` Am- - . ,. ~ —10 —IS FIGURE 10.7 Line drawing based on data of Figure 10.6. See text for discussion. Column on leR shows velocity structure determined from a nearby seismic refraction study.

124 JACK E. OLIvER Sw 10 0.0- 2 n- 2 - wAr 3 0- TI~E 4.0- 5.0- 6.+ 7.0- °ARKFIEL3 CAL~F SAF SAF 1 1 HE ~ so ~ ~ 60 ~ 8° ~ `= 110 i~ '= ." i~ '" '70 ' - t~ 2= 2~0 220 2= 24c 2x 2" I ~ ! ' I I I I ~ I I I ~ ~ I ~ I ~ ~ I ~ 1 1 ~ 10.0- 1 1.0- 12.0- FIGURE 10.8 Line drawing of short test section crossing San Andreas Fault near Parkfield. Distance in decameters. Tw~way travel time in seconds. S.A.F. indicates boundary of San Andreas Fault zone. See text for discussion. Figure 10.9 shows a map of the southeasternmost por- tion of the Wind River range and indicates the line of a seismic reflection profile made in this area. Line drawings of parts ofthis section are indicated in Figure 10.10. There is a great clear of over information of considerable variety. The most prominent feature is the eastward extension of We thrust fault on We western boundary of the Wind Rivers Mountains. This Cult extends to a depth of at least 25 km and more likely to 40 km without much change in dip. These data, then, seem to resolve the long-standing geological problem of the origin of the Wind Rivers in We sense that compressional forces seem to predominate as opposed to those producing vertical uplift. Substantial shortening of the crust is clear. CONCLUS ION With We possible exception of study of the lower modes of free oscillations, all the seismic methods described above have something substantial to contribute to under- standing of the continental basement. Furthermore, there are other useful methods and techniques that I will not discuss here, and in any case the dividing lines between the various methods are somewhat arbitrary. My purposes are (l ~ to draw attention to the large unrealized potential for understanding the continental basement through application of present-day seismic methods and ¢2J to point to means for improving that potential through con- tinuing evolution and improvement of seismological methods. Suppose, for example, that a method could be found for artificially generating shear waves of sufficient amplitude so that the deep crust and upper mantle could be explored using such waves, or that movable, tightly spaced seismic arrays could be deployed so as eventually to cover the continent. Suppose the dee~sea floor, the continental shelves, and venous remote areas were no longerareas of essentially no seismic data on earthquakes. Suppose that seismic profiling were conducted along long, closely spaced lines spanning the continents. A1- though this might seem like an ambitious undertaking to some, one should remember that it is only within about the last 30 years that there has been significant seismic exploration of the deposed floor and only about 20 years since reflection profiling ofthe seafloor began. The ocean basins, which, of course, occupy a much greater portion of the earths surface than do Me continents, have in that short time been crisscrossed innumerable times by seis- mic reflection profiling. A comparable achievement in the form of seismic profiling of the continents is within the grasp of the present generation of geophysicists. Suppose microprocessors and other electronic developments result in more sensitive and selective seismographs and new ways of managing large complex sets of data. Suppose the seismic reflection method is generalized so that three- dimensional images rather than tw~dimensional sections are produced. Suppose comparable advances are made concurrently in related branches of geophysics and geol- ogy. Every one of these suppositions is technically within our reach today. If even some of these suppositions are fulfilled, I think we can look forward to a new understanding of Me crust, one in which deep subsurface structural features become as familiar to continental scientists as the midocean edges, the trenches, the seamounts, and the submarine canyons have become to ocean scientists over the last few decades. When they are, surely our understanding of the evolution and the genesis of continents will become more

Seismic Exploration of the Continental Basement 30' /5' _ I09°00' 45' W Do' 15' 108°00' 45' . .~ ~ ~ / ]43°00' .' .,'.' ~ .. ~ \ . 1 ~ ,. . to,; got Jon Toteraccle Butte \ f / Aikoli Otter i5 elf P~ p,450 l -~ ~ ~ =~ ~ ~ ::: :- : I; 'T' "I ~ Who r'~ ' "''2""."~ \~N \ ~~\~' :/~250 /~00 ~50 6' :~OC~,>~ Oregon Buttes OContlnenla Peal _ ,45' N 30' _ /5' 42-00' 0 10 FIGURE 10.9 Clap of southeastern Wind River Mountains in Wyoming showing location of COCORP seismic reflection line. COCORP wrOM`NG I iNE ~ wtOMiNG L'NE iA G~_ R`~ ~~.e S. sr~rro~ ~vu-ffeS 50 `00 Is0 20o 250 100 ...a Q,~e, 84,n, sour~ PaSS C/TY | sr~rloa au_tes 20 ·m 125 WYOM`NG LINE 2 a.~ ei~—.,. ~ c,'. ~E s0 It00 150 2nt~ ~—~ ·~ ~ ,~ ~ ~ ,~ 5rA7~— ~RS / 1 \ _' ~ r? -_~ ~_-- ~ -___ '~ \ _ \ \ \,,,- 1s 0 1 . i 20.0 FIGURE 10.10 Line drawing of section along line of Figure 10.9. Note especially major thrust fault bounding the Wind River uplift on west. See text for discussion. Courtesy Jonathan Brewer and Scott Smithson, University of Wyoming.

126 profound, just as our understanding of Me origin of the ocean basins became more thorough. Are these Noughts idle dreams never to be fulfilled? Somehow I feel Me visiting astronaut would not see them that way but instead would view them as imminent and inevitable steps in man's progress toward a better scientific understanding of his planets a better life for its inhabitants. ACKNOWLEDGM ENT S The Outlook wishes to thank Muawia Barazangi, Bryan Isacks' Larry Brown, Jonathan Brewer, and Scott Smith- son for permission to use figures from papers in prepara- tion and venous colleagues for constructive comments. The COCORP research is supported by National Science Foundation grants EAR 77-13653 and EAR 77-14674. BE FERENCE S AIci, K, A. Christofferssen, and E. S. Husbye (1977). Detennina- tion of the three~imensional seismic structure of the litho- sphere, J. Geophys. Res. 82, 277~95. Barazangi, M., and B. Isacks (in press). Subduction of the Nazca plate beneath Pent: evidence from spatial distribution of ear~- quakes, Geophys. J. JACK E. OLIVER Berthelsen, A. (1960). Structural studies In the Precambrian at western Greenland. Weds. Croenl. 123, 1-~!22. Ellsworth, W. L., and R. Y. Koyanagi (1977) Three-dimensional crust and mantle structure of KiIauea Volcano, Hawaii, J. Geop)'ys. Res. 82, 5379~394. Giese, P., C. Prodehl, and A. Stein, eds. ( 1976) Expl`~:on Sei.s- mology in Central Europe, Data and Results, Springer-Verlag, New York, 429 pp. Iyer, H. M. (1973). Anomalous delays of teleseismic P waves in Yellowstone National Park, Nature 253, 425 427. Mueller, S. (1977). A new model of the continental crust, in The Earth's Crust, J. G. Heacock, ea., Am. Geophys. Anion Geophys. Monogr. 20, pp. 289 317. ARC Committee on Seismology (1976). Predicting Earthquakes, National Academy of Sciences, Washington, D.C., 62 pp. Oliver, J., M. Dobrin, S. Kaufman, R. Meyer, and B. Phinney (1976). Continuous seismic reflection profiling of the deep basement, [Iardeman County, Texas, 13~11. Geol. Soc. Am. 87, 1537-1546. Prodehl, C. (1977). The structure of the crust~nantle boundary beneath North America and Europe as derived from explosion seismology, in The Earth's Crust, J. G. Heacock, ea., Am. Geophys. Union Geophys. Monogr. 20, pp. 349~69. Sanford, A. R., R. P. Mott, Ir., P. I. Shuleski, E. J. Rinehart, F. J. Caravella, R. M. Ward, and T. C. Wallace (1977~. Geophysical evidence for a magma body in the crust in the vicinity of Sm corro, New Mexico, in The Earth's Cmst, J. G. Heacock, ea., Am. Geophys. Union Geophys. Monogr. 20, pp. 385~3. Smithson, S. B., P. N. Shive, and S. K. Brown (1977). Seismic velocity, reflections, and structure of the crystalline crust, in The Earth's Crust, J. G. Heacock, ea., Am. Geophys. Union Geophys. Monogr. 20, pp. 254~70.

Exploration of the Continental Crust Using Aeromagnetic Data INTROD UCTION ISIDORE ZIETZ U.S. Geological Survey It is significant that many countries have undertaken aero- magnetic mapping programs to obtain complete coverage of their nations in a reasonable period of time. The Soviet Union, for example (Solov'yeva, 1968), had complete coverage by 1967 using a flight-spacing of 2 km. The Canadian Geological Survey had initiated a program to map the entire Canadian Shield at a flight separation of 1/2 mile (804 m). Almost three quarters of the shield has now been flown, and the maps have been made available to the public (Hood, 1974~. Yet' at this writing, there is no federal program to map the United States aeromagnetic- ally. To date, published aeromagnetic maps are available for only 20 percent of the conte~`inous United States. Although at least half of the country has been adequately surveyed by private industry, the results hare not been made available to the public. Most of the available cover- age in the United States has been provided by the U.S. Geological Survey, the funds oRen coming from other organizations, such as state geological surveys. For the past few years, the U.S. Geological Survey has adopted a 127 policy of promptly releasing aeromagnetic data to the public via the open-f~le route. These recent publications are mostly at a scale of 1:250,000. The obvious advantages of the airborne method over ground measurements are the speed and economy of the survey, the coverage of areas that are inaccessible on the ground, and the nature of the data recorded continu- ously along a profile rather than as discrete points. An advantage of the magnetic method is that it permits the mapping ofthe basement crystalline rocks underneath a cover of sedimentary rocks, for the latter are nonmag- netic. The specifications of an airborne magnetic survey are dictated by the nature of the problem. For detailed surveys, the detector should be close to the ground and the separation between flight lines small. For regional and crustal geological studies, the flight elevation should be higher, and the flight spacing larger. Aeromagnetic data may reflect rock distributions any- where from the surface to depths at which the Cune point of the magnetic rocks is reached. In the continental crust, this may vary anywhere from 15 50 km, depending on heat flow and the thermal properties of the rocks. The

128 ; '~0~°,'J' )- W~ 4~ .- , ~ i_ ~ . --,'. 2 - .) . 'by ~ At' . , Frogs- ~~ ,/,i ~pS~' _ Add/ U ~~ ' ~ :t . . C _ Z A — _ _ >` _ ~ .- C, ;J '— C ~ . ~ - tC :5 _ _ _ - _ ~ — — <e ~ I} ._ _ ~ ~ :) — aC ~ _ _ ~ 3 ~ .,~ ~ _ :n _ _ J — ;) C) ~ ~ _, ~ _ ~ _ :n

Exploration of the Continental Crust Using Aeromagnetic Data depth of penetration is not so great as those reached by other geophysical methods, such as seismic, gravity, or magnetotellunc methocls. However, the magnetic method, which "sees" the shallow rocks, in combination with the aforementioned geophysical methods, which "see" the deep rocks, provides a powerful tool in evaluat- ing the total crust. In this paper, I will discuss selected aeromagnetic sur- veys and their geological interpretations for areas in the western United States, the midcontinent, and the eastern United States. I selected these three provinces because they reflect three distinctly different phases of crusta1 evolution the tectonically active area in the west' the old and stable crust of the midcontinent, and the Appalachian province in the east. WESTERN UNITED STATES The work of Blake et al. (1978) is particularly significant in that it attempts, by examining available aeromagnetic data, to correlate a number of ophioIite belts in California and to relate these belts to fanner plate boundaries. This is possible because the serpentinized ultramafic rock, which makes up a large volume of the California ophio- lites, has unusually high magnetic susceptibilities. Blake et al. (1978) suggest that the ophiolite belts For the entire western margin of both North and South America could be mapped, provided adequate aeromagnetic coverage were available. In addition, the authors suggest that the Great Valley anomaly may be related to the emplacement of ultramafic and magic rock in a marginal basin behind a Late Jurassic island arc. Gnscom (1973) deduces some important tectonic rela- tionships from aeromagnetic data at the junction of the San Andreas Fault and Mendocino fracture zone. "Straight magnetic lineaments on the continental shelf west of the San Andreas fault between Point Arena and Cape Mendocino have northwest trends, and are inter- preted to be caused by ophiolite belts. An east-west magnetic anomaly associated with the Mendocino frac- ture zone extends from the deep ocean inland at Punta Gorda, a distance of 20 km. During the time this anomaly has existed, the San Andreas fault cannot have extended north of Cape Mendocino, and lateral movement between the oceanic and continental plates cannot have taken place near Cape Mendocino." A compilation of all available aeromagnetic data in Nevada and Utah (Stewart et al., 1977) clearly shows east-west trends (Figure 11.1, A-A', B-B', C-C'), in spite of the fact that the basin and range structures trend north- south. Geological investigations subsequent to this dis- covery by aeromagnetic mapping showed that these east- west trends are correlatable with Cenozoic igneous rocks and that significant mineral deposits are aligned along these same belts in eastern Nevada and western Utah. In a regional sense, the aeromagnetic map over the Basin and Range province in Nevada and Utah is espe- 129 cialIy "quiet" when compared with the magnetic data elsewhere over the continental United States. In eastern Utah, the amplitude ofthe anomalies is an order of magni- tude greater than those to the west. If these anomalies are related to Precambrian rocks, the magnetic data might indicate their western limit (Figure 11.1, D-D'). The major volcanic areas of the Columbia Plateau, Snake River Plain' Yellowstone Park, High Cascades, and the Tertiary basalt areas of coastal Oregon and \Vash- ington all show characteristic magnetic patterns and anomalies. A high-altitude (15,000-R barometric) aeromagnetic survey with widely spaced flight lines (5 miles) for a strip across the northwestern United States (Zietz et al., 1971) proved useful in studying the gross features of the earth's crust. The strip is bounded approximately by latitudes 45°30' N and 47°00' N. and extends from the Pacific Ocean approximately 92 km from west of the coast, eastward to and including the Great Plains in Montana. The results are spectacular in that the magnetic map is marked by conspicuous northeast and northwest anomaly trends, lineaments, and breaks in the anomaly pattern. Their regional distribution, overall magnetic character, and geo- logical evidence suggest that they are major structural features in the basement rocks. The close correspondence of structural and geological features in younger rocks with these basement magnetic and structural trends suggests that basement trends controlled, or at least greatly ~n- fluenced, intrusion, deposition, and structural history of younger rocks. In some cases, evidence suggests that basement structures have been reactivated during later tectonic activity. Perhaps even more striking than the northeast- and northwest-trending features are large east-west magnetic discontinuities that, in some cases, extend completely across the strip to the edge of the shelf and that, in some cases, can be correlated with large-scale discontinuities dating back to the Precambrian. MIDWESTERN UNITED STATES A compilation of magnetic data (Plate 11.1) for a large area in the northern part of the Midwestern United States shows the characteristic magnetic patterns typical of the Precambrian stable craton of North America. The map was prepared by reducing large-scale aeromagnetic maps and removing the earth's main magnetic field (International Geomagnetic Reference Field). The residual map was colored at 200-gamma intervals, the magnetic intensities varying in a rough way with the colors of the optical spec- trum. Black-and-white presentations of aeromagnetic maps are important in structural geological interpre~- tions but are less helpful in evaluating lithologies. The colored map adds another dimension to the original data and is indispensable in the discrimination of lithologic units. In addition, the colored map acts as a filter as it shows Inroad regional anomalies that would be indiscern- ible in black-and-white presentations. The amplitudes of

130 the anomalies within the area are very high, in marked contrast to the smaller amplitudes of anomalies shown in the aeromagnetic map of Nevada and Utah. One ofthe more prominent features on the map extends from Flue Superior, passing through the entire states of Minnesota and Iowa and put of Nebraska. It corresponds to We midcontinent gravity high (Figure 11.2) originally described by Woollard (1943), probably the most outstanding feature on the gravity map of the United States. A more detailed aeromagnetic map of this feature, FIGURE 11.2 Midcontinent gravity high (~m Zietz, 1969, copyrighted by American Geophysical Union). ISIDORE ZIETZ using a contour interval of 100 gammas, is shown in Figure 11.3. The combined geophysical data, surface out- crops in the Lake Superior district, and drilling data sug- gest the existence clef a thick sequence of basalts failing a trough approximately 40 miles wide and 4 miles deep extending from the Lake Superior district (where exten- sive areas of basalt of Keweenawan age are known to crop out) to Kansas. Basalt crops out in only a very small area in the northeast part of the map in northwestern Wiscon- sin. The rest of the area is covered by a sedimentary se- ll N I IdINNESOTA , _ , , _ _ _ ~ NEBRASKA ~ WISCONSIN

Exploration of the Continental Crust Using Aeromagnetic Data lows FIGURE 11.3 Aeromagnetic map of midcontinent gravity high (from Zietz, 1969, copyrighted by the American Geophysical Union). quence that graclually thickens in a southerly direction to a maximum thickness of approximately 5000 feet in Nebraska. The presence of basalt in the Precambrian is verified by 11 drill holes in Minnesota and Iowa, all of which bottom in basalt. The flows are bounded by high- angle faults on both sides and are flanked by a series of elongated basins containing red elastic sandstones 1 or 2 miles thick (King and Zietz, 1971~. The geophysical data suggest that the basalt flows were subsequently faulted and folded. The trough probably 131 represents a major riR in the older Precambrian crust and appears to be discordant with tile older structures. This midcontinent rift may well have been part of a Keweena- wan global rip system, with initial offsets consisting of transform faults along pre-existing fractures, but appar- ently it never fully developed laterally into an ocean basin (see Chapter 4), and the upwelling mafic material was localized along a relatively narrow belt. Much geological information can be extracted by the combined use of the aeromagnetic and gravity data to

132 I . _ I i ~\ I i ~ rl 1.1~ ~ ~

- Exploration of the Continental Crust Using Aeromagnetic Data make a geological map of the buried Precambrian surface and to construct geological cross sections at right angles to the strike of the flows. In northern Minnesota and eastern South Dakota, nor~east-trending broad-wavelength magnetic anoma- lies, on Me order of 30~0 km, are clearly discernible (Plate 11.1~. These alternating bands of highs and lows can be traced to the northeast part of the state, where they can be correlated with rocks that crop out at the surface. Surprisingly, the positive magnetic anomalies reflect granite gneisses, and the negative anomalies are iden- tified with a greenstone terrane. This correlation is consis- tent with geophysical interpretations published for the Canadian Shield but inconsistent with the commonly ac- cepted generalization that felsic rocks produce magnetic lows, and mafic rocks magnetic highs. A detailed gravity survey of the entire State of Minnesota corroborated our predicted results. The outlines and shapes of the gravity and magnetic anomalies match splendidly but have an inverse correlation, i.e., gravity high with magnetic low and gravity low with magnetic high. We do not yet under- stand why certain major metamorphic belts are magnetic, but the answer is probably a combination of original com- position and the chemical potential of oxygen (i.e., the chemical conditions in effect during metamorphism). We do know, however, that the granitic gneisses, being less dense than the host rocks, should give gravity lows, whereas the greenstones, which are more dense than the host rocks, should produce gravity highs. The use of this geophysical approach (combined magnetics and gravity) has broad ramifications in other areas of the midwestern United States. It should be possible to produce lithologic and possibly structural maps of tlie Precambrian base- ment surface by the proper analysis of the geophysical data when used in conjunction with all available drill-hole data on basement cores. In the southwestern part of the map (PIate 11.1), a broad-wavelength magnetic trough, indicated by the deep blue area, covers all of Iowa. I suspect that it is caused by anomalous conditions deep in the crust. It may be that the Curie point is elevated in this area. Additional geo- physical investigations to probe into the deep crust, such as deep resistivity or magnetotellunc soundings, would be helpful. Since the compilation of this map, a detailed aero- magnetic survey was conducted ofthe entire Precambrian shield area in northern Wisconsin. The {light spacing was 1/2 mile, and the flight elevation was 1000 feet baro- metric. The aeromagnetic maps (1:250,000 scale), both in black and white and in color, were placed on open file by the U.S. Geological Survey (USGS>, and thus made availa- ble to the public. At the time of this survey, geological mapping of this area of Wisconsin was sketchy, except for local areas (see, for example, the lithologic map of Pre- ca~nbrian rocks in Wisconsin produced by Dutton and Bradley, 1970~. Through the use of the aeromagnetic map and a gravity map based on widely spaced observation points, a new geological map (1:250,000 scale) was pro- . _ . . 133 duced by Sims et al. (19781. It is significant that, to pro- duce the geological map, no additional field mapping was necessary. This new map is an order of magnitude more detailed than the previous one and is summarized by the abstracts of Sims and Petennan (1978) and Cannon (1978), presented at the 24th annual meeting of the Institute on Lake Superior Geology in May 1978. EASTERN UNITED STATES Except for the State of Maine, the entire exposed Appa- lachian crystalline belt from Alabama to Newfoundland has been flown aeromagnetically, with a line spacing of one mile or closer. In addition, most of the Coastal Plain of the eastern United States has been surveyed aeromagnetically at one-mile spacings. The survey covers the area from east of the Fall Line to the shore line and from the Potomac River to the approximate latitude of Ocala, Florida. The magnetic anomalies over the crystalline rocks are distinctly linear, probably reflecting tectonic trends. These linear trends may be caused by belts of magnetic plutons (either mafic or granite), belts of certain amphi- bolites and magnetite-beanng schists, or belts of mafic and ultrarnafic rocks that may represent rifts, sutures, or former subduction zones. Because of the linearity of the aeromagnetic trends, aeromagnetic maps have proven usefi~1 as an aid to field mapping over the Appalachian crystalline rocks. They are especially useful in the southeast United States, where outcrops are sparse and much of the area is covered by sapprolite. The USGS has initiated a program to publish aeromag- netic maps over the entire crystalline belt in black and white and in color, at a scale of 1:250,000. Each of these maps will be accompanied by a geological interpretation. South of New York City, new geological maps will be based primarily on the aeromagnetic data but will obviously include all existing geological information. Radioactivity maps, when used in conjunction with the aeromagnetic data, are particularly useful in surf~cial mapping. Whereas the felsic rocks are generally less magnetic than the mafic rocks, the reverse is true for radioactivity, the granitic rocks being more radioactive. For the Appalachian orogen, the literature is becoming inundated with papers dealing with plate tectonics, that is, those related to the opening and closing of the Atlantic Ocean. Any such model would have to be continental in scope. Unfortunately, reliable field investigations are limited to small areas, on the order of a few 15' quad- rangles, and each author can usually accommodate his mapping to fit the plate-tectonic model of his choice. These 1 :250,000-scale aeromagnetic maps and the accom- panying regional geology covering very large areas, should fill an important gap in arriving at a more meaning- ful model. An important paper by Hatcher et al. (1977) is typical of

134 the use one can make of aeromagnetic data in the south- east. To quote from the abstract: Geologic mapping, interpretation, and field checking of recent aeromagnetic data suggest the existence of a closely associated series of faults and splays extending from Alabama to Virginia, herein termed the Eastern Piedmont fault system. Charactenstic magnetic anomalies were found to be associated with known faults, and were used to trace them through covered intervals. SOUTH CAROLINA 34° r 'it 7W `-r_ ~ A vie) I GEORGIA ISIDORE ZIETZ The fault system extends northeastward from the Goat Rock faul of Alabama and west-central Georgia. crossing the lower Pied- mont of South Carolina, passes beneath a segment of the Coastal Plains in the Carolinas, and then flanks the Raleigh belt in North Carolina and continues into Virginia. From east~entral Georgia to Virginia, cataclastic rocks along the faults of the system are bounded to the northwest and southeast by rocks of the Carolina slate belt, conning perhaps the most extensive fault system in eastern North America. 1 ~~ f1 - ' Of' ~ _ ~ r . ~ `~: -/ _ . . be,,' Hi.% ~ ' ~N', -~, ? ~ . ,, TV'.' `. . ~~ Bruns - CHAR ~ .. ~ - ~81. — ~ A' NORTH CAROIINA _ ~ DIG hi, ~ ~ 1~ 1~ KILOS I ~ , . · , , , FIGURE 11.5 Generalized aeromagnetic map of southeastern South Carolina and eastern Georgia. Contour interval is 100 gammas (from Popenoe and Zietz, 1977).

~ : ~ - - - Y75 o Ct C) . Cat o £ o C) 135 - . - c . o U: - C~ . TIC o - o C' - o o Cal -

136 The Eastern Piedmont fault system is shown in Figure 11.4. The aeromagnetic dam on which the fault system is based shows anomalies of unusual linearity and small amplitude (tens of gammas). Aeromagnetic maps for a small area of Me system are shown in Figure 11.5 (A-A') (Popenoe and Zietz, 1977~. One of the more significant magnetic features offshore is the East Coast anomaly, which is a prominent linear magnetic high that follows the scarp at We eastern Norm American Continental Shelf edge. Off the Georgia coast, the anomaly turns westward, crosses the Shelf, and comes ashore near Brunswick, Georgia. This segment is called the Brunswick anomaly. The magnetic data combined with gravity, heat flow, and drill-hole data show quite clearly that the Brunswick anomaly is axial to the Southeast Georgia Embayment, one of the largest Atlantic Coastal Plain sedimentary basins, with a history of subsidence that was active throughout the Cenozoic and is still taking place. Seismic profiles and computed profiles and models from the aeromagnetic and gravity maps indicate that the anomaly~ausing body is a mafic intrusion at or just below the top of the basement beneath the Coastal Plain. The intrusive body and the volcanic rocks fonn what is prob- ably an anomalous crust separating the continental rocks of Georgia from those of noncom Flonda The intrusive body probably marks an old riR zone, and Eternal contrac- tion of the anomalous crust has caused subsidence of the Southeast Georgia Embayment. The Brunswick and East Coast anomalies probably mark the old edge of the North American continent. One of the longest magnetic lineaments in the North American continent (Figure 11.6) is prominently dis- played along the length of the Appalachian basin (King and Zietz, 1978~. It doubtless reflects a discontinuity in the basement, trending in a northeasterly direction for 1600 km from the Mississippi Embayment to the Green Mountains. King and Zietz have called this magnetic fea- ture the New York-AIabama lineament It is marked by a series of linear magnetic gradients that bound areas of magnetic rocks in the basement, implying strike-slip dis- placement along a profound crystal break. The lineament tends to coincide with the west side of the regional Appa- lachian gravity low (Figure 11.7) and separates a province of dominantly north-trending gravity anomalies on the northwest from a province of prevailing northeast trends on the southeast. The linearnent seems to mark the south- east edge of a stable crustal block that acted as a buttress for the strong deformation of the Appalachian fold belt, so that the arcuate salients in Pennsylvania and Tennessee are tangential to it. Present-day seismic activity shows a correlation of the location of the lineament (Figure 11.8) with an active area to the southeast in Me eastern part of the Appalachian basin and a seismically inactive block along the northwest side of the lineament. A parallel ac- tive zone to the west extends Tom New Madrid, Missouri (see Chapter 7) to the valley of the St Lawrence River. I S IDORE Z I ETZ FIGURE 11.7 Bouguer gravity map of area near lineament (from King and Zeitz, 1978, courtesy of the Geological Society of America). Contour interval 10 meal.

Exploration of the Continental Crust Using Aeromagnetic Data it. ~ FIGURE 11.8 Seismotectonic map of eastern United States (from King and Zeitz, 1978, courtesy of the Geological Society of America). The lineament is in the subsurface extension of the Gren- ville province, which may mark a region of a former con- tinental collision analogous to the India-Asia region. Here, strike-~lip faults of great length and linearity and large horizontal displacement, such as the .Altyn Tagh Fault of Tibet, typify the region north of India and may have their counterpart in the New York-Alabama linea- ment, which records probable strike-slip displacement of great magnitude in the basement rocks under the Appala- chian basin. 137 CONCLUSIONS With just a few selected examples in the United States this paper atterr~ts to show the significance of the aero- magnetic method in evaluating the upper part of the earth's crust. Because crystalline rocks usually have a magnetic signature associated with them, the magnetic method can be used in geological mapping for both lith- ology and structure. In order to integrate on a regional scale detailed mapping of local areas, oRen done by dif- ferent people, using vastly different approaches, aero- magnetic maps provide a common denominator by relat- ing s~aat~graphic units defined using different criteria and tracing units through unmapped areas. BE FE RE NC E S Blake, M. C., Jr., I. Zietz, and L. Daniels (1978). Aeromagnetic and generalized geologic map of parts of central Califomia, U.S. Geol. Surv. Inv. Map GP-918. Cannon, W. F. (1978). A middle and late Precambrian fault system in northern Wisconsin and northern Michigan (abstr.), Proceedings of the 24th Annual Meeting, Institute on Lake Supenor Geology, May 1978. Dutton, C. E., and R. E. Bradley (1970). Lithologic, geophysical and mineral commodity maps of Precambrian rocks in Wiscon- sin, U.S. Geol. Surv. Misc. Field Inv. Map I 631. Gnscom, A. (1973). Tectonics at the junction of the San Andreas Fault and Mendocino fracture zone from gravity and magnetic data, in Conference on Tectonic Problems of the San Andreas Fault System, Proceedings, Stanford U. Publ. Geol. Sci. 13, 38~390. ' Hatcher, R. D., Jr., D. E. Howell, and P. Talwani (1977). Eastern Piedmont fault system: speculations on its extent, Geology 5, 636 640- Hood, P. ( 1974). Mineral exploration, trends and development in 1973, Can. Min. J. 95 (3), 1-20. King, E. R., and I. Zietz (1971). Aeromagnetic study of Me mid- continent gravity high of central United States, Geol. Soc. Am. Bull. 82, 2187-2208. King, E. R., and I. Zietz (1978). The New York-Alabama linea- ment: geophysical evidence for a major costar break in the basement beneath the Appalachian basin, Geology 6, 312 318. Popenoe, P., and I. Zietz (1977). The nature of the geophysical basement beneath the Coastal Plain of South Carolina and northeastern Georgia, in Studies Related to the Charleston, South Carolina, Earthquake of 1886 A Preliminary Report, W. Rankin, ea., US. Geol. Sur?;. Prof. Paper 1028, 11~137. Sims, P. K., and Z. E. Peterman (1978). Precambrian geologic framework of nor~em Wisconsin (abstr.), Preceedings of the 24th Annual Meeting, Institute on Lake Superior Geology, May 1978. Sims, P. K., W. F. Cannon, and M. G. Mudrey, Jr. ( 1978). Prelimi- nary geologic map of Precambrian rocks in part of northern Wisconsin, U.S. Geol. Surv. Open-File Rep. 78~18. Solov'yeva, N. M. (1968). Karta anomal' nago magnitnogo poly temtorii, S.S.S.R 1968 (Map of Me anomalous magnetic field of the territory of the U.S.S.~.), Moscow: Bsesoyuznyy Nauchno-Issled-Ovatel'skiy Geologisheskiy Institue (Vsye- gyei) (1:10,000,000).

138 Stewart, J. lI., W. J. Moore, and I. Zietz (1977). East-west pat- terns of Cenozoic igneous rocks, aeromagnetic anomalies, and mineral deposits, Nevada and Utah, Geol. Soc. Am. Bull. 88, 67-77. Woollard, G. P. (1943~. Transcontinental gravitational and magnetic profile of Norm America and its relation to geologic structure, Geol. Soc. Am. Bull. 54, 747-790. ISIDORE ZIETZ Zietz, I. (1969~. Aeromagnetic investigations of the earths crust in the United States, in The Earth's Crust and Upper Mantle, P. J. Hart, ea., Am. Geophys. Union Geophys. Monogr. 13, pp. 404~15. Zietz, I., B. Hearn, Jr., M. W. Higgins, G. D. Robinson, and D. A. Swanson (1971). Interpretation of an aeromagnetic StTip across the northwestern United States, Geol. Soc. am. Bull. 82 3347 3372.

Chemistry of the Lower Crust: Inferences from Magmas and Xenoliths 12 INTRODUCTION ROBERT W. KAY an] SUZANNE MAHLBURG KAY Cornell University c In any discussion of the lower crust, deep crustal xeno- liths and crustal-derived or crustal-contaminated magmas should receive an important share of attention. This ratio- nale is directly analogous to the interest in peridotites and basalts as probes of the upper mantle. Studies of crustal- derived magmas and xenoliths from different regions in- dicate regional variations in the composition, age, and thermal history of lower crustal regions. Coordination of xenolith and geophysical studies has the potential of defining subsurface map units of regional extent. More comprehensive study of continental magmas and xeno- Iiths should result in the development of increasingly re- alistic crustal models (e.g., Smithson and Brown, 1977~. The following is a review of some approaches toward models of the lower crust using magmas and xenoliths. It will become apparent that there are more questions than answers in much of this work. The last section will for- mulate some fundamental questions and suggest ways to get answers to them. 139 CONCEPTUAL FRAMEWORK Questions about the lower crust must be asked within a spatial and temporal framework. Foremost is the basic observation that continental fission, fusion, and strike-slip motion implicit in the plate-tectonic development of the earth's surface have resulted in continents composed of crustal fragments of diverse ages and histories, separated by cryptic suture zones. Because of these processes, later events may be superimposed on juxtaposed crustal frag- ments having quite different early histories. A common later event may be intrusion and attendant heating of the crust by magmas in continental rift zones or convergent plate margins [e.g., Andean arc magmas intruding Pre- carnbrian cmst in Peru (Dalmayrac et al., 197711. In addi- tion, we recognize that at present, new continental crust is being formed largely by magmatic activity along arcuate zones at convergent plate margins and within broad intra- plate tensional zones, including rift zones. We expect that zero-age crust will have distinctive differences at these two localities. Lastly, the mechanisms of formation of Ar-

140 chean crust, particularly the crust under greenstone belts and anorthosites, may not be operating today and are an intrinsically more difficult problem (Hargraves, 1976~. CONTINENTAL MAGMATIC STUDIES The study of melts derived from the crust is an indirect way of studying the crust itself. Some magmas that rise to shallow crustal levels have either formed in the crust or have been extensively modified by addition of a "gra- nitic'' crustal low melting ~action. The magmas occur in two general regions~convergent plate margins and intra- plate regions with high thermal gradients. In both cases, intrusion of mantle-derived basaltic magmas into the crest has increased Thea gradients to crustal melting condi- tions as indicated by surface heat flow (Lachenbruch and Sass, 1977~. Figure 12.1 shows the distribution of high- heat-flow areas in the United States. Figure 12.2 shows the relationship between thermal gradients and melting curves of crystal rocks. A good way to identify crustal-derived components in magmas is to match chemical and isotopic signatures in the magmas with those of known sedimentary rock types or basement age provinces (e.g., Zartman, 1974~. The work of Annstrong et al. (1977) shows that high 87Sr~6Sr ratios in young igneous rocks in the western United States are restricted to regions underlain (or inferred to be underlain) by Precambrian basement (Figure 12.3~. Simi- larly, Sr and Pb isotope tracers have been used by Lipman et al. (1978) in a comprehensive study of volcanic rocks from a small region in southern Colorado. They recognize temporal as well as spatial differences in isotopic compo- s~tion and find that both upper and lower crustal cont~ni- nation can be identified. The implication of these data is FIGURE 12.1 A generalized heat- flow contour map of the conterrninous United States. The areas of highest heat flow coincide with the areas of Plio-Pleistocene volcanism. The Battle Mountain high in Idaho and Nevada is especially prominent and is an interior area with high convective heat loss. Differences in heat flow in different areas probably influence the mineral- ogy and composition ofthe lower crust, creating provinces that may mimic the heat-flow map, particularly where heat flow is exceptionally high. Figure is from Lachenbruch and Sass (1977, copyrighted by the American Geophys- ical Union). Also shown are several xenolith localities referred to in the text: SF (San Francisco Volcanic Field), FC (Four Corners region, Colorado Plateau), AH (Lucite Hills, Wyoming), (Kilboume Hole, New Mexico), ST (Stockdale, Kansas). ROBERT W. KAY and SUZANNE MAHLB URG KAY that basement provinces can be outlined by isotopic anal- yses of recent shallow magmatic rocks that have inter- acted with the deep crust in an intraplate environment. As a further illustration, Turi and Taylor (1976) and Taylor and Turi (1976) found a correlation of high 87Sr/"Sr and ~80 values in Tuscan (Italy) silicic magmas (Figure 12.4) that demonstrates a major contamination by crustal melt- ing. The effect continues into the Roman province to the south, where mixing of the silicic magmas with basic mantle-derived mag~nas apparently has occurred. Chemi- cal trends in aluminous S-type (sedimentary-type) grani- toids analogous to Turi and Taylor's (1976) Tuscan mag- mas have been used by White and Chappell (1976) to argue for a sedimentary crystal source, which they con- ~ast to igneous-type, or 1-type, granitoids that have an igneous (basaltic) crustal source. Probably not all isotopic variations in continental mag- mas can be attributed to crustal contamination. Brooks et al. (1976) and Whitford et al. (1977) have identified iso- topic variations in magrnas that reflect subcontinental lithospheric mantle and subducted oceanic sediments. However, evidence looks persuasive for interaction of magmas with deep continental crust in Italy and in the western United States, reinforcing the hypothesis that magmas can provide infonnation on the lower crust. CONSEQUENCES OF CRUSTAL MELTING: TWO TYPES OF LOWER CRUST RESIDUAL CRUST An important consequence of the contamination of mag- mas by a crusted low melting Faction is that a refractory residue complementary to the melt exists at depth This pleat Flow Contours @~2.5HFO 1.5-2.5 O 0.75-1 5 1.8-15 75 ~ 1.0 ; it, - _,,'-t'~ ~ ,_.~. _/3;' Hi,. , ~ ,~ I /' ..2...L ~ 2,\ ''my i :..,' I..,\ ~ ; '~ ~r~ ,. .

Chemistry of the Lower Crust ~0 - ~: 20 30 40a TEMPERATURE ~ ° C 400 600 800 1000 , . , . , ~ ,  (Cat on- )~ HEAT ~ O.9HFU 1.3 FLOW 1200 - 1 Boselt ~ \=dus l ~ Liquidus| V ill \~2~5l 17 Z1 FIGURE 12.2 Generalized conductive temperature profiles (from left to right) for the Sierra Nevada crust, a stable reference crust, the characteristic range for the Basin and Range, and the lower limiting and typical conditions in the Battle Mountain high. All curves are drawn for a surface heat production (Ao) of 5 HGU (heat generation units) and thennal conductivity (K) of 6 meal cm~~ sect ICE. Corresponding surface heat flow is shown at the bottom of each curve. Melting relations (Wyllie, 1971) are shown for intermediate crustal roclc by the curves GWS (g!ranodionte water saturated solidus, dashed line), and GDS (granodionte "dry" solidus, heavy line), and for basalt by the dry basalt solidus, and dry basalt liquidus. Figure and caption modified from Lachenbruch and Sass (1977). Sr / Sr .c.704 +~.706 ~_, ;12~°_~_ 117° 1!3° 44o- 4~'\< '` r- i . \, ! ~ + · 1,,$ !+ 46°- 1--.~-~-.-~ ~ :-. / . - . ) . 7 ^++` .' ~ *+ I _+. (+ 42° \~ +-~-M , . . O 100 200 300 / Km FIGURE 12.3 Map of the northwestern United States showing locations of Mesozoic and Cenozoic igneous rocks with 87Srl86Sr ratios of less than 0.704 (circles) and greater than 0.706 (crosses). The heavy line is the approximate location of the 0.704 contour, and is drawn to join its trace in California defined by Kistler and Peterman (1973). References for data points are given in Arrn- strong et at. (1977). The strontium isotope differences may be controlled by the age and nature of the lower crust. (Figure is from Annstrong et al., 1977, repented Tom Bulletin of the Geo- logical Society of America, with mission.) 12 10t o - o of A_ O 8 GO OF VESUVIUS_ I. EL kIULSINI ~ ~ ~ R - OCEANIC BASALTS T. I Province Roman Region N: Naples Area . . . . . . . . . 702 .706 .710 .714 .718 S / FIGURE 12.4 Plot of the generalized range of whole rock 6'80 versus 87Sr/"Sr for venous volcanic rocks in the Tuscan and Roman provinces of Italy. The correlation illustrates major con- tamination by cmstal melting. Figure is from Tun and Taylor (1976). residue (restite) is probably at equilibrium with the melt at the depth and temperature of magma segregation with respect to mineralogy; partitioning of major, minor, and trace elements; and isotopic ratios. For example, restites in equilibrium with granitoids should be aluminous gran- ulites in the case of politic starting material (Green, 1976) or amphibolite to pyroxene granulite in the case of ba- saltic starting material (Wyllie et al., 1976~. White and Chappell (1976) believe that many of these restites are now found as mafic clots and inclusions in S-type and I-type granitoids, but the restite mineralogy does not gen- erally indicate a high-pressure origin because of re- equilibratior~ at low pressure. Several studies have used a residual crust concept to predict the composition of the lower crust. Models of Zie- linski and Lipman (1976), Taylor (1977), DeLong (1974), and Arth and Hanson (1975) illustrate the usefulness of Iithophile elements, particularly rare-earth elements in modeling mineral melt equilibria. Three points might be made in connection with these models: 1. Melts derived from continental crust are diverse, and different magmas leave different residual crusts. Thus, the residual crust concept is easily reconciled with lower crustal heterogeneity. 2. Predictive characteristics of the residual crust may have shortcomings because the trace-element partition coefficients that are the key to the calculations are not well known. This is particularly true for potentially im- portant accessory phases (apatite, scapolite, and zircon)

142 and for the heat-producing elements U. Th, and K. Acidic compositions in the lower crust are difficult to model because accessory minerals may incorporate large propor- tions of the total trace-element content of the rock (Gromet and Silver, 1978~. A further problem concerns whether minerals and melt are at equilibrium. 3. The above residual-crust modeIs assumed that ma- terial was transferred by a melt. Alternative methods of material transport, possibly involving a fluid phase, must also be considered. For exarnpIe, the role of aqueous fluid phase in upward concentration of U. Th, and K in the crust (Tarney, 1976) will have to be evaluated quantitatively. Diffusive transport (Orville, 1963) or adjective transport in a hydrous fluid are alternative mechanisms. Aqueous fluids are also important in the problem of phase transpor- tation kinetics (Ahrens and Schubert, 1975~. SOLIDIFIED BASIC MELTS A second consequence of generation of abundant silicic crustal melts is that a large part of the heat of fusion must come from intrusion and solidification of higher-tem- perature mantle~erived basic melts. Large regions of lower crust, equivalent to the volume of crustal~enved silicic magma, must be crystallized basic intrusions. This gives us a rationale for another class of lower crustal rocks that are represented in xenolith population~umulate rocks of basic composition. EXPOSED LOWER CRUSTAL SECTIONS Possible lower crustal sections, now exposed at the sur- face by tectonic activity, that have been investigated by geophysical techniques indicate what we may expect in some regions of the lower crust. The most cited example of an exposed lower crustal section is the Ivrea zone in northern Italy (Berckhemer, 1969), where both restites of sedimentary parentage and metamorphosed equivalents of basaltic magrnas have been found (Mehnert, 1975; Schmid and Wood, 1976~. The lower crust in this region appears to have originally been composed of a sedimen- tary sequence of sandstone, shales, and limestones with mafic lavas intruded into the section particularly at the base. The grade of metamorphism varies from the upper amphibolite to the granulite facies (Schmid and Wood, 1976~. XENOLITH STUDIES In the absence of drill holes to the relevant depths and because of questions in the interpretation of surface gran- ulite facies rocks in crustal sections of normal thickness, xenolith data must be regarded as a primary source of infonnation about the lower crust. The anticipated scale of lateral inhomogeneities of the lower crust far exceeds the spacing of xenolith localities. Because of the possibil- BOBERT w. ICAY and SUZANNE MAHLBURG KAY ity that xenolith localities may represent unique or unusual conditions in the crust, a better approach might be to group the localities based on tectonic similarities: convergent plate margins, riPr valleys, areas of intraplate volcanism, and Precambrian cratonic regions. Xenoliths have not been reported from all tectonic regimes, e.g., continental collision zones and greenstone belts. Prior to a discussion of specific xenolith areas, two im- portant points should be noted: 1. Xenoliths represent the mantle and crustal section only at the time of eruption of the conveying medium (e.g., lava or kimberlite). The section is not necessarily the same as the present one, because of possible subsequent tectonic activity or textural re-equilibration. 2. Pressure~emperature (P-T) conditions in the crust- mantle column may or may not be reflected by P-T es- timates based on coexisting minerals from the xenoliths. Equilibration of xenoliths from the lower crust during periods of kimberlite eruption may not occur because of slow reaction rates at low crustal temperatures. In this case, the geothermal gradient determined from xenolith studies is probably a fossil gradient. In contrast, during volcanic episodes on a regional scale (e.g., western United States), temperatures may reach 800 900°C in the lower crust (see Figure 12.2) causing textural re~quil- ibration. This statement follows from the observation that diffusion over i cm will occur within 10 million years (m.y.) for diffusion coefficients greater than 3 x 1~5 cm2/sec (which occur in the range 800 900°C in silicates, see HoEmann and Hart, 1978; Buckley, 1973~. Textural equilibration will probably occur in a shorter time, per- haps 105 years, if a melt phase or a fluid phase is present, because diffusion coefficients are several orders of magni- tude higher in liquids than in solids (Aherns and Schu- bert, 1975~. However, intergranular transport may not al- ways be the rate-controlling process (Loomis, 1976) because the rate of divariant reactions can depend on diffusion rates within reactants even if intergranular transport is fast. XENOLITHS ASSOCIATED WITH CONVERGENT PLATE BOUNDARIES In the Japanese arc, where the tectonic setting is well known, there are two contrasting xenolith localities in young volcanic rocks. One (Ichinomegata) has a lower equilibration temperature and is more hydrous than the other (Oki-Dogo). 1. Oki-Dogo alkali basalt has abundant xenoliths rang- ing from granite (underlying crust) to peridotite. Taka- hashi (1978) has constructed a crustal column based on mineralogically derived P-T estimates (Figure 12.5~. An important observation is that the transition from lower crust to upper mantle is within a series of layered cum- ulate rocks ranging Mom gabbros to peridotites. The xeno- liths display no deformation effects, yet they are not sim-

Chemistry of the Lower Crust 0, 10 _ ~ 20 - I 0~ 30 40 50 /+ Granite + \ I+ + + + + +\ I \ v <~°C v v _ ~ ~ ~ ,— A Oli ne a ~ a close 900-1100° O Czmulate 000_1100°~_~ 0 0 0 0 0 1100-1200° Spinei ~herzolite Zone 0 0 0 0 0 0 ?- > 1 200°C ~ ~ /: Melt //.~/ _ v Pyroxene Gabbro v ~ ~ LOWER CRUST UPPER MANTLE FIGURE 12.5 Petrological model of Oki-Dogo Island, Japan, based on xenoliths found in alkali basalt flows. Figure is from Takahashi (1978). ply crustal cumulates of the Oki-Dogo magna itself; their Ar release pattern indicates ages of perhaps 30 m.y. to 40 m.y. and their 87Sr/86Sr ratios are generally higher than the host lava. Other volcanic rocks of southwest Honshu have similar assemblages. The geothermal gradients denved from the xenoliths for the crust resemble those of the Battle Mountain high of Figure 12.1. 2. Ichinomegata andesite also has abundant xenoliths ranging from granite to peridotite. However, equilibra- tion temperatures of the xenoliths are consistently lower than those of the Oki-Dogo xenoliths for equivalent pres- sures. The mafic and ultramafic assemblages contain hornblende and are deformed (sheared) rather than cum- ulate in texture. The regional extent of Ichinomegata-type lower crust is indicated by similar xenolith assemblages found in high-alumina basalts firom the Honshu and Kurile arcs (Katsui et al., 1978; Shimazu et al., 1978~. In summary for Japan, the lower crustal xenolith suite includes abundant types of basaltic composition and ap- pears to lack metasedimentary and refractory types. The xenolith suite indicates that crystallized basic melts, mod- ified by crystal settling and shearing, form extensive regions of the lower crust in Japan. Considering the im- portance of island arcs in many models of lower-crustal development (see Chapter 3 as an example), the Japanese data seem especiaIly important. XENOLITHS ASSOCIATED WITH RIFT VALLEY ENVIRONMENTS AND INTRAPLATE VO L CANI C A REAS Rio Grande Rift Two xenolith localities (Kilboume Hole and the Elephant Butte Reservoir area) are young volcanic maars and flows associated with the Rio Grande rif;r in New Mexico. Both 143 metasedimentary and meta-igneous rocks of granulite metamorphic grade have been recognized among the abundant xenoliths at Kilbourne Hole (Padovani and Carter, 1977~. Metasedimentary rock types include sillimanite-bearing garnet granulites, which have been interpreted as high-temperature residues of pelitic sedi- ments. Meta-igneous rock types include two-pyroxene granulites that are compositionally basalts and are per- haps solidified basaltic magmas. Anorthosites and char- nockites also occur in the meta-igneous suite and have led Padovani and Carter (1977) to suggest that the assem- blages in the lower crust at Kilboume Hole are similar to some high-grade metamorphic terranes exposed at the earth's surface. Two-pyroxene granulities of basaltic com- position and charnockites have also been recognized in the Elephant Butte Reservoir area farther north by A. M. Kudo (University of New Mexico, personal communica- tion, 1978). The principal techniques used to s~dy these xenolith suites include pe~ographic analyses and electron- microprobe analyses of coexisting minerals. Such studies are indispensable for any quantitative evaluation of pres- sure and temperature. A major result of the Kilboume Hole study was that the geothe~`al gradient of 30°C/km calculated from mineral P-T conditions (Figure 12.6) is roughly equivalent to the infetTed gradient for the Battle TEMPERATURE, °C 0 200 400 600 800 1000 1200 101 . . y - I 20 _ ~ . C} 30 40 S (. ~ P-T colculated from coexisting garnet- plagiociase - sillimanite- quortz / Bosi :, RGnge ~ ~nde FIGURE 12.6 Temperature-depth plot showing distribution of values (plotted symbols) determined by Padovani and Carter (1977) from coexisting games and plagioclase in Kilboume Hole gan~et granulites. The aluminum silicate stability field from Hol- daway (1971) is plotted for reference (K: Kyanite; A: Andalusite; S: Sillimanite). The vertically lined field represents the range of temperature with depth for surface heat fluxes of 2.4 RFU (heat flow units) in the southern Rio Grande riR (Decker and Smithson, 1975). The horizontally lined f~eld represents the range of tem- perature wi~ depth for surface heat fluxes of 2.0 HFU for the Basin and Range (Decker and Smithson, 1975). Figure is mod- ified from Padovani and Carter (1977).

144 Mountain high (Figure 12.2) and the Rio Grande rid (Cook et al., 1978; Reiter et al., 19781. Midcontinent Rift The kimberlite at Stoclc~lale, Kansas (Meyer and Brookins, 1976) is located on the midcontinent gravity high and yields only a few granulite and pyroxenite xenol~ths of probable lower crustal or upper manmade origin. Most are altered, but a few have flesh cores and contain minor sillimanite and sapphirine. One chemically analyzed sam- ple has the whole-rock composition of an olivine tholeiite and could represent magma crystallized at depth asso- ciated with riding 1100 m.y. ago. No xenoliths of possible sedimentary origin have been reported. Granites appar- ently overlie the gabbros and other igneous-derived rocks of the lower crust. If the granites are a low-tempera~re crustal melt fraction, the residues in the lower crust have not been found or apparently are not present in the lower crust beneath Stockdale. Dating of the xenoliths by Rb/Sr mineral isochron tech- niques (Brooking and Woods, 1970) suggests a complex thermal history for the region. The postulated riPc environ- ment suggests very high heat flow in the Precambrian. The question of the timing of the alteration becomes significant. Has the lower crust in this region undergone retrograde metamorphism, and does this retrogression correspond to present heat flow values? Or was alteration associated with the later intrusion of Me kimberlite, and has a fossil high-temperature geothenn beer recorded by the xenoliths? Brooking and Meyer (1974) believe that the alteration accompanied kimberlite intrusion. Massif Central, France Xenoliths Bom Boumac, a Plio-Pleistocene volcanic pipe in the Massif Central of southern France, were studied by Leyreloup et al. (1977) and Bilal (1976~. Numerous hag- ments of granulite facies metamorphic rocks derived from both sedunen~ry and igneous parents have been found. The igneous suite is made up principally of basic granu- lites but includes rocks with compositions ranging from ultrabasic to acidic. Leyreloup et al. (1977) interpret the meta-igneous suite as low-pressure crystal fractionates of a tholeiitic magma that recrystallized in the granulite metamorphic facies. Granulites of sedimentary composi- tion include metamorphosed shales, graywackes, and ar- koses. Sparse calcareous granulites have been found in other xenolith localities in the Massif Central. Intercala- tion of the igneous and sedimentary components is pres- ent in composite hand specimens, where it is presumed that the igneous component intruded Me sedimentary component. Age dating by whole-rock Rb~r (Hamet et al., 1978) indicates an age of 1300 m.y. for the xenoliths; zircon dating indicates heating events at 300 m.y. and 600 m.y. Outcrops of old granulitic terrace in the nearby Variscan (late Paleozoic) metamorphic rocks may be ROBERT W. DAY and SUZANNE MAHLBURG KAY equivalent in age (Harnet and Allegre, 1976), implying that the polymetamorphism is also recorded in surface exposures in this region. Much remains to be learned of He thermal history of the deep crust from surface expo- sures. Leyreloup et al. (1977) conclude from extrapolation of Boumac xenolith abundances and types that the regional composition of the lower crust is andesitic and that the metasedimentary rocks are not depleted in large-ion litho- phile elements as in the classic depleted granulite terrane of Heier (1973~. lntraplate Volcanic Areas Other volcanic areas may also have a lower crust similar to that in rift valleys. For example, an assemblage of co- existing charnockites and two-pyroxene granulites similar to the Kilbourne Hole suite has been recognized by Stoesser (1973) in the recently active San Francisco vol- canic field in Arizona. Stoesser (1973) interprets these xenoliths as wall rock into which an ult~nafic layered sequence associated with the volcanic activity has been intruded. This suggests that one component of the louver crust in volcanic areas that include rift valleys would be layered mafic intrusions formed by cumulate processes. Francis (1976) reaches the same conclusion for granulites from Nunivak, Alaska, as have Kay et al. (1978) for the Leucite Hills, Wyoming. An intraplate xenolith locality that has received consid- erable attention is the Australian Delegate pipe (Lovenng and White, 1968; Irving, 1974), where possible lower crustal xenoliths include a charnockite and various granu- lites of basaltic composition. Similar xenolith types have been found in alkali basalt flows in an intraplate area to the north in Queensland (Stephenson and Canon, 1976~. A summary of xenolith types in Australian localities has been compiled by Wass and Irving (1976~. The lower crust in volcanic regions, particularly in rift valleys, appears to be in the granulite metamorphic facies and appears to have two-pyroxene granulites of basaltic composition as a component. These basaltic composition rocks may represent the metamorphic equivalent nonbasic intrusions responsible for crustal dilation in riR valleys and uplift ofthe flanking crust. Low-melting granitic com- ponents present in the crust prior to intrusion may rise to upper crustal levels as contaminants of basaltic lavas or as silicic intrusives or extrusives. Other assemblages, such as residual sillimanite~amet granulites and anorthosites may be present, but if so, they are probably related to the history of the crust prior to He high-heat-flow episode. XENOLITHS FROM POLYMETAMORPHOSED PRECAMBRIAN LOWER CRUST Two of the most highly studied suites of probable lower crustal xenoliths come from kimberlites in regions where polymetarnorphism of Precambrian lower crust can be

Chemistry of the Lower Crust demonstrated. Similar complex histories for other lower crustal suites may be developed with furler investiga- tion. Southern Africa Kimberlites with abundant mantle and crustal xenoliths occur within the Precambrian Kaapvaal Craton and its surrounding younger mobile belt in Lesotho. Regional difference in the lower crust may be indicated by contrast- ing xenolith populations in the kimberlites (Griffin et al., 1979~. High-temperature anhydrous granulites are con- fined to the mobile belts (Figure 12.7), while the intercra- tonic crustal xenoliths are typically Iower-temperature (amphibolite facies) gneisses that may or may not repre- sent deep crustal material. The mobile belts are inter- preted by Kroner (1977) to have been formed by rework- ing of older rocks in the Precambrian with little addition of new material. Work has concentrated on granulite xeno- liths from the mobile belt (Griffin et al., 1979; Rogers, 1977), while xenoliths from the craton have received little attention. At one locality in the mobile belt, 50-70 percent of the meta-igneous granulite xenoliths are basaltic in composi- tion and 30 50 percent are intermediate to acidic in com- position (Bloomer and Nixon, 1973~. Whole-rock chemical trends (i.e., Mg/Fe increases with SiO2) do not resemble those resulting from crystal fractionation. Variable con- centrations of incompatible elements (such as K and Rb) may reflect modification by high-grade metamorphic pro- cesses, perhaps including partial melting (Gnffin et al., 1979; Rogers, 1977~. The rare-earth analyses for magic gar- net g~ranulites indicate that amphibole-bearing gamet- free mineral assemblages were present during the solid- liquid fractionation. The rocks are at present garnet bearing. The Lesotho lower crust appears to have undergone a multistage heating history zircon ages are Proterozoic, but the last heating episode probably coincided with the widespread Karoo volcanism in Cretaceous time. Geo- thermal gradients were probably fairly high at the time of kimberlite intrusion as the time period of their intrusion is similar to that of the Karoo volcanism. These xenoliths may have resided in abnormally warm lower crust, al- though probably not so hot as the Rio Grande riPc. Colorado Plateau Tertiary kimberlite diatremes in the Four Corners region of the Colorado Plateau contain abundant xenoliths and xenocrysts from both the crust and upper mantle. McGetchin and Silver (1972) examined xenoliths from the Mule Ear diatreme and based a lower crustal and mantle model of the Colorado Plateau on the shape, relative size, and abundance of the crystalline fragments. The lower crust in the model consists of amphibolite and granulite facies metamorphic rocks, all of which have been hy- 145 J O GRANULITES \ + NO GRANULITES - ~ + l) ~ LESOTHO FIGURE 12.7 Distribution of granulites in crustal xenoliths from South AfFican kimberlites. The crosses represent pipes where no granuIites have been found; all lie within the Kaspvaa1 Craton (area surrounded by the solid line). The open circles represent pipes in the surrounding mobile belt where granulites have been found. Figure modified from Griffin et al. (1979). crated during retrograde metamorphism in the amphibole facies. The most common lower crustal rock type is meta- gabbroic granulite gneiss containing garnet. Metasedi- mentary gneisses of upper amphibolite grade (E. Pado- vani, Massachusetts Institute of Technology, personal communication, 1978), eclogite, chlorite schist, serpen- tine schist, and pyroxenite also occur. The ubiquitous retrograde metamorphism, even in mantle xenoliths (Smith and Levy, 1976), stands in sharp contrast to xenoliths discussed from other localities. Geo- thermal gradients implied by the mineral assemblages are low, approximating the stable reference crust geotherrn of Figure 12.2. Helmstaedt and Schulze (1977) advocate an even lower geothermal gradient, on the basis of a low-T, high-P chlorite-eclogite retrograde mineralogy of an orig- inally high-T eclogite. They argue that subcontinental hy- dration accompanied retrograde metamorphism, which was caused by underthrusting of hydrated oceanic litho- sphere. In contrast, Smith (1977) has suggested that hy- dration is localized along zones of lithospheric weakness that coincide with the monoclines along which the kim- berlites occur. Xenoliths from South Africa and the Colorado Plateau are compositionally diverse and have complex thermal and Reformational histories. Yet several generalizations can be made. First, metasedimentary rocks are subordi- nate to meta-igneous rocks. Second, fossil metamorphic gradients seem to be represented in Africa, since regional heat-flow values are low and relatively uniform across the mobile belt~raton boundary. Third, despite the retro- grade metamorphic and hydration effects found in the

146 Colorado Plateau xenol~ths, the rock types are not signifi- cantly different from those found in other xenolith locali- ties of the western United States. This finding implies that differences in the lower crust may be due to regional variations in water content and metamorphic grade as well as to heterogeneity in chemical composition. OUTSTANDING QUESTIONS AND FUTURE DIRECTIONS Research on continental magmas and xenoliths has a potentially unique role in crustal studies. Several studies have resulted in important observations, but the most basic questions about the lower crust remain unanswered. Continental crystal studies are probably at a stage of de- velopment equivalent to that reached in the study of oce- anic crust 20 years ago. It may seem presumptuous to ask questions about a region so complex as the continental crust, but it seemed equally presumptuous then to ask equivalent questions about the oceanic crust. We will ask three questions about the crust and outline what the role of xenolith and magma studies might be in formulating answers. WHAT IS THE ORIGIN OF THE ROCK SUITES AT SITES OF CONTINENTAL CRUSTAL FORMATION? What is the origin of rocks that make up new crust? Are they sedimentary, or are they derived Dom sedimentary rocks by melting or fluid-phase transport (metasomatism), or are they igneous? Are they hydrous or anhydrous? There are a number of chemical and isotopic criteria that have a wide applicability in answering these questions, particularly the use of trace elements (lithophile ele- ments, transition metals) and radiogenic isotopes (Sr, Pb, Nd). Interpretation of xenoliths and magrnas will benefit Tom comparison with a large number of existing analyses on a wide range of igneous and sedimentary rock types. Some types of analyses that have not been widely done, in particular, rare gas, volatile element, and oxygen isotope analyses, may also be valuable. For example, Phinney et al. (1978) found positive Axe anomalies in CO2-rich well gas, indicating Mat the gas was not atmospheric. In addi- tion, trapped gases may prove to be good monitors of diffusive material transfer. The possible use of oxygen isotope analyses may be demonstrated by We unexpected and only partially understood variability found in eclogites (Vogel and Garlick, 1970~. Shieh and Schwarcz (1974) have shown that oxygen isotope analyses of ex- posed granulite facies rocks are similar to basalts. The fi~ctionation of volatile elements (such as hydrogen, the halogens, and sulkier) in deep crustal xenoliths is impor- tent in view of the role of volatiles in melting, diffusive transport, electrical conductivity, and over properties of We lower crust (Goldsmith, 1976; Necut et al., 19771. ROBERT w. KAY and SUZANNE MAHLBURG KAY The interpretation of trace-element and isotope data will remain uncertain until basic questions about the par- tition of trace elements between granitic~ioritic melt and residual crystals and the kinetics of equilibration are resolved. The application of radiogenic isotopic criteria (Sr, Nd, Pb) for characterization of melts from crustal res- ervoirs is likewise hampered by lack of knowledge of equilibration times and trace-element distributions (Ben Othman et al., 1978~. Studies of both experimental and natural mineral assemblages are necessary to resolve these questions. The preliminary conclusion drawn from several xeno- lith studies is that many deep crustal rocks are igneous in origin and basaltic to andesitic in bulk chemistry. How- ever, sampling bias may exist because studied suites are from volcanic and kimberlite areas, where the lithosphere may be structurally and Metrologically abnormal. A sec- ond possible bias is the ability of some xenolith types to survive transportation better than others. Additional bias results in the collection of xenoliths, in which small or altered types may be underrepresented. Field studies must take into account these biases. Improved represen- tation oftectonic environments should be a major focus of additional sampling (e.g., greenstone belts and adjacent gneiss terranes of the Precambrian and continental colli- sion zones). Field studies on xenoliths and magmas could also bene- fit greatly from geophysical studies in the same region. Thus it seems desirable to search for localities also in- vestigated using seismic reflection techniques (the Leu- cite Hills and the cocoaP wind River line, Smithson et al., 1978, for example). In the future, seismic reflection lines might be planned across regions with interesting con- trasts in xenolith populations (e.g., the Kaapvaal shield and mobile belt or the Four Corners region). Field work within areas of exposed probable lower crusted rocks (e.g., the Ivrea zone, Berckhemer, 1969) is a valuable guide to the relationship of rock types and struc- tures in xenolith suites. However, some types of lower crust that are present in xenolith suites may not be repre- sented in surface exposures. WHAT ARE THE AGE RELATIONSHIPS WITHIN VERTICAL AS WELL AS HORIZONTAL CRUSTAL SECTIONS? At present there is a lack of knowledge ofthe mechanisms of continental crustal thickening. It is possible that under- plating and horizontal under~msting continually add young crust under continents, while volcanism and shal- low plutonism add material to the upper crust. The dating of polymetamorphic rocks (and examination of initial iso- tope ratios) may be the only way to resolve these ques- tions. There are many ages to be sorted out, i.e., thermal events (including heating by igneous intrusions), metaso- matic events, and melting events. Rb~r, K-Ar, U-Pb, and Nd~m dating methods will be useful. The relatively new dating method, Nd~m, will have a

Chemistry of the Louder Crust broad application in the granulate facies rocks, especially those that are low in K and Rb. The fractionation of Nd and Sm between garnet (low NdlSm) and pyroxene (high Nd/Sm) make them ideal for dating crystallization ages of eclogites. The solubilitv of rare^earth elements in metaso- matic fluids is probably less than the solubility of K and Rb, so that the Nd~m isotopic system may be relatively immune to metasomatic events. WHAT IS THE TEMPERATURE—STRAIN HISTORY OF LOWER CRUSTAL REGIONS? Prerequisite to any statements about the temperature his- tory of the lower crust inferred from xenolith minerals is the P-T calibration of mineral assemblages and coexisting mineral compositions found in xenoIiths (Whitney and Stormer, 1977~. Laboratory experiments and theoretical advances are needed to clarify the calibration problems. Some rocks yield satisfactory P-T estimates. For other assemblages, especially high-variance ones, estimates are not very good, mainly because of uncertainties of distribu- tion coefficients with P. T. and composition. Eclogites (garnet-clinopyroxene rocks) are a good example. Griffin et al. (1979) noted a gap in Lesotho xenolith depth es- timates between a lower crustal group (depths to 35 km) and a high-pressure mantle group (depths greater than 85 km). They suspect that xenoliths in the depth range of 25~0 km are present and may include eclogites. Intercrystalline equilibnum, involving the distribution of elements between sites in a crystal lattice, is a poten- tially powerful method for determining P and T. Crystallo- graphic methods are required to determine site occupan- cies, but distinctive solid-solution behavior, calculated from chemical analyses, is a reflection ofthe site behavior. For example, Figure 12.8 is a plot of two solid-solution components dadeite and Tsche~l~'ak's molecule) in clino- pyroxenes from granulite xenoliths. The amount of alumi- num in fourfold coordination is thought to increase with temperature, while the amount of aluminum In sixfold coordination is thought to increase with pressure. AIumi- num in fourfold coordination is calculated by the amount of Tscherrnak's molecule, while the amount of aluminum in sixfold coordination coupled with sodium is calculated by the amount of jadeite (White, 1964~. The grouping of analyses from one locality is distinctive in Figure 12.8, but interpretation hinges on further P and T calibration. Griffin et at. (1979) attribute the high jadeite component in the Lesotho granulites to pressures in the lower crust. Since similar Nanette components are not found in surface granulites, they postulate that aluminum has re~quili- brated upon uplift if surface granulite outcrops were for- merly in the deep crust. Determination of a lower crustal geothermal gradient has been a major concern of several xenolith studies. Some granulites yield temperatures that are compatible with present heat flow, and the geothermal gradients derived from these xenoliths may reflect the present gra- dients (see Padovani and Carter, 19771. Dating of textur- 147 2s 20 15 . . . . — ECLOGIT~ FIELD - ~ ~ / C lo - ~ a/ / GRANULITE FIELD · ^! V . . · O ·/ - O. / ~0 it. / · ~ 0 5 How ton 1 5 20 % TSCHERMAK S MOLECULE GRANULITE CLI NOPYROX£N~S ~ LESOTHO O LEUCITE H I LLS V DELEGATE QUEENSLAND Kl L80URNE HOLE D sTocKDAL~ o NUNIYAK FIGURE 12.8 Plot of mole percent jadeite against Tschennak's molecule for granulite clinopyroxenes from xenoliths of probable lower crustal origin. Solid symbols represent ga~net-bearing as- semblages; open symbols represent gan~et-free assemblages. When Fe2O3 was not reported in the analyses, it was calculated using the method of Papike et al. (1974). Jadeite and Tschermak contents were calculated using the method of Yoder and Tilley (1962) and White (1964), Lesotho analyses from Griffin et al. (1979), Delegate analyses from Lovenng and White (1968) and Irving ( 1974)' Kilbourne Hole analyses from Padovani and Carter (1977), Stockdale analyses from Meyer and Brookins (1976), and Leucite Hills and Queensland analyses by authors. ally defined mineral assemblages would be a valuable constraint on the timing of the P-T conditions recorded by the mineral assemblage and the relation to present heat cow. In several xenolith studies, retrograde metamorphic ef- fects have been noted, indicating partial re-equilibration of lower crustal mineral assemblages in response to changing P-T conditions. A major question is: How fast do the assemblages respond to changes in geothermal gradient? At the earth's surface, high-grade metamorphic rocks react too slowly to equilibrate. Reactions at mantle (asthenospheric) depths should be complete within a hand specimen. The depth to the zone of equilibration is un- doubtedly variable. In some areas this zone may reach into the mantle (Frazer and Lawless, 1978; Herzberg and Chapman, 1976) but in areas of crustal melting and a high geothermal gradient, the partially re~equilibrated zone should be in the lower crust. Mineral zoning and corona textures of rock in the zone may be used to determine temperature and time of reaction (e.g., Loomis, 19761. In some cases (e.g., Colorado Plateau) partly equilibrated xenoliths indicate both a high paleogeothermal gradient and a low present geothermal gradient, corresponding to a hydrated retrograde mineral assemblage. Exsolved phases in plagioclases and pyroxenes from deep crustal xenoliths may also yield important information on the kinetics of mineral equilibration and residence time of the xenoliths in the lower crust.

148 Imm llJ cleft, 100 IOkb Ikb 1 3 lOOb lOb lOmm ~ , I ~ 1 ' ~ _ 101` l ll + !+ m.g.d. "C ~-1.6 - 1 1, 1, 1, 1, 1, 1, 10 9 8 7 6 ~ 0 10 10 10 10 10 JO cm~ FIGURE 12.9 Mean grain size versus dislocation density and differential stress for olivine. Analogous relations can be defined for plagioclase and clinopyroxene and should be useful in defin- ing differential stress for defonned granulites. Figure is from Koldstedt et al. (1976). An important question that follows from study of the retrograde hydration in some xenoli~s is the source ofthe water. Stable-isotope studies (hydrogen and oxygen) may be able to distinguish alternatives such as hydrothermal circulation or dehydration of underlying subducted oce- anic lithosphere. Microcrack studies may also be impor- tant in defining mechanisms for migration of fluid phases in the deep crust (Simmons and Richter, 1978~. Substantial progress on the dynamics of upper-mantle flow has been made by studies of deformation mech- anisms of olivine in the laboratory and application to peridotite xenoli~s (e.g., Kohlstedt et al., 1976~. A paral- led study using lower crustal granulites deserves atten- tion. The density of unannealed defects can be examined using the transmission electron microscope. Experi- mental work needs to be done to understand the deforrna- tion mechanisms of feldspar and pyroxene, probably the most abundant lower crustal minerals. In addition, the simple relationship between grain size and stress found in olivine (Figure 12.9) encourages We search for a simi- lar relationship for feldspars and pyroxenes. One ob- serves Mat grain size in basic granulites is variable from locality to locality; could this reflect a variable stress in ROBERT W. I[AY and SUZANNE MAHLBURG KAY Me lower crust? If stress and temperature can be deter- mined for a xenolith, Den strain rate can be calculated, and questions of lower crustal dynamics can be addressed. SUMMARY AND CONCLUSIONS Xenoliths and magmas can help to answer such questions as (a) what is the parentage of the lower crust, i.e., per- centage of original igneous versus sedimentary material, percentage derived directly from the mantle versus ma- terial derived from crustal processes; (b) is the lower crust hydrous or anhydrous; (c) what are the temperature- pressure regimes in the lower crust; (d) are rocks in the lower crust at equilibrium with present or past tempera- ture~ressure conditions; (e) what are the age relations between various units of the lower crust; and (f) what is the deformation history and what is the present state of stress in the lower crust? When knowledge derived from xenoliths and magmas is combined with drill holes and geophysical information, a three~imensional picture of the crust can be constructed. Crustal studies have barely begun: a great expansion of our ability to answer all these questions can be expected in the near future. ACKNOWLE DGM E NTS We thank D. Smith and E. Padovani for reviews and members of the Continental Tectonics Panel for discus- sion and acknowledge Me National Science Foundation for providing financial support under Grant EAR 77- 13656. BE FERENCE S Ahrens, T., and G. Schubert (1975). Gabbro~clogite reaction rate and its geophysical significance, Rev. Geophys. Space Phys. 13, 383~00. Armstrong, R. L., W. Taubeneck, and P. Hales (1977). Rb~r and K-Ar geochronometry of Mesozoic granitic rocks and their Sr isotopic composition, Oregon, Washington and Idaho, Geol. Soc. Am. Bull. 88, 397411. Arth, J. G., arid G. N. Hanson (1975). Geochemistry and origin of the early Precambrian crust of northeastern Minnesota, Geo- ch~m. Cosmochim. Acta 39, 325~62. Ben Othman, D., A. Juery, and C. J. Allegre (1978). Limitations on granite connation inferred by Nd, Sr and Pb isotope sys- tematics (abstr.), Eos Trans. Am. Geophys. Union 59, 392. Berckhemer, H. (1969). Direct evidence for the composition of the lower crust and the Moho, Tectonophysics 8, 97-105. Bilal, A. (1976). Les fluides carboniques dans les enclaves char- nockitiques de Boumac (Masif Central firangais). Implications sur les structures de Ia croute continentals, These Be cycle, Nancy. Bloomer, A. G., and P. J. Nixon (1973). The geology of the Letseng-la-terae kimberlite pipes, in Lesotho K:mberl~tes, P. H. Nixon, ea., Lesotho National Development Corporation, Masern, Lesotho, pp. 20~2.

Chemistry of the Louver Crust Brookins. D. G., and H. O. A. Meyer (1974). Crustal and upper mantle stratigraphy beneath eastern Kansas, Geophys. Res. Lett. 1, 969-272. Brookins, D. G., and M. J. Woods (1970). Rb~r geochronologic investigation of basic and ultrabasic xenoliths from the Stock- dale kimberlite, Riley County, Kansas, Kansas Geol. Sure. Bull. 199, 3-12. Brooks, C., D. E. James, and S. R. Hart (1976). Ancient litho- sphere: its role in young continental volcanism, Science 193, 1086-1094. Buckley, G. (1973). The effect of diffusion on garnet zoning, Ph.D. thesis, U. of Illinois, 112 pp. Cook, F. A., E. R. Decker, and S. B. Smithson (1978). Preliminary transient heat flow model of the Rio Grande rift in southern New Mexico, Earth Planet. Sci. Lett. 40, 316~26. Dalmayrac, B., J. Lancelot, and A. Leyreloup (1977). Two billion year granulites in the late Precambrian metamorphic basement along the southern Peruvian coast, Science 198, 49~1. Decker, E. R., and S. B. Smithson (197~). Heat flow and gravity interpretation across the Rio Grande rip in southern New Mex- ico and west Texas, J. Geophys. Res. 80, 2S42-2552. DeLong, S. E. (1974). Distribution of Rb, Sr, and Ni in igneous rocks, central and western Aleutian Islands, Alaska, Geochzm. Cosmochzm. Acta 38, 245-266. Francis, D. M. (1976). Corona-bearing pyroxene granulite xeno- liths and the lower crust beneath Nunivak Island, Alaska, Can. Mineral. 41, 291-298. Frazer, D., and P. Lawless (1978). Paleogeotherms: implications of disequilibrium in games Iherzolite xenoliths, Nature 273, 220-222. Goldsmith, J. R. (1976). Scapolites, granulites, and volatiles in the lower crust, Geol. Soc. Am. Bull. 87, 161-168. Green, T. H. (1976). Experimental generation of cordierite or games-bearing granitic liquids from a politic composition, Geology 4, 85~8. Griffin, W. L., D. A. Carswell, and P. H. Nixon (1979). Lower crustal granulites and eclogites from Lesotho, Southem Africa, in The Mantle Sample: Inclusions in Kimberlites and Other Volcanics, F. R. Boyd and H. O. A. Meyer, eds., American Geophysical Union, Washington, O.C., pp. 09-86. Gromet, L. P., and L. T. Silver (1978). Implications of rare earth distribution among minerals in a granodiorite. Peninsular Ranges Batholith, Southern California (abstr.), Eos Trans. Am. Geophys. Union 59, 399~4C)O. Hamet, J., and C. J. Allegre (1976). Hercynian orogeny in the Montagne Noire (France): application of 87Rb-~7Sr systematics, Geol. Soc. Am. Bull. 87, 1429-1442. Hamet, J., A. Vitrac, and C. J. Allegre (1978). 87Rb-~7Sr and U-Pb systematics in granulitic xenoliths in volcanic rocks and limita- tions on the formation of the lower crust (abstr.), Eos Trans. Am. Geophys. Union 59, 392. Hargraves, R. (1976). Precambrian geologic history, Science 193, 363~71. Heier, K. S. (1973). Geochemistry of granulite f:acies and prob- lems of their origin, Phil. Trans. Roy. Soc. London S273, 429Js42. Helmstacdt, H., and D. J. Schulze ( 1977). Type A type C eclogite transition in a xenolith from the Moses Rock diatreme. Further evidence for the presence of metamorphosed ophiolites beneath the Colorado Plateau, in Extended Abstracts of the Second International K~mberl~te Conference, 1977. Herzberg, C., and N. Chapman ( 1976). Clinopyroxene geo- thermometry of spinel-lherzolites, Am. Mineral. 61, 696~37. i49 Hofmann, A. W., and S. R. Hart (1978). An assessment of local and regional equilibrium in the mantle, Earth Planet. Sci. butt. 38. 4442. Holdaway, PI. J. (1971). Stability of andalusite and the aluminum silicate phase diagram, Am. J. Sci. 271. 97-131. Irving, A. J. (1974). Geochemical and high pressure experimental studies of games pyroxenite and pyroxene granulite xenoIiths Mom the Delegate basaltic pipes, Australia,l. Petrol. 1~. 1~0. Katsui, Y. K. Niida, M. Yamonoto, and S. Nemoto (1978). Genesis of calc-alkalic andesite from Oshima-Oshima and Ichinome- gata volcanoes, North Japan (abstr.), in International Geody- namics Conference "Western Pacific" awl "magma Genesis," Tokyo, March 13-17, 1978, Inter-Union Commission on Geodynamics, and Science Council of Japan, p. 966. Kay, S. NJ., R. W. Kay, J. Hangas, and T. Snedden (1978). Crustal xenoliths from potassic lavas, Leucite Hills, Wyoming, Geol. Soc. Am. Abstr. Programs 10, 432. Kistler, R. W., and Z. Peterman (1973). Variations of Sr, Rb, K, Na and initial 87Sr/86Sr in Mesozoic granitic rocks and intruded wall rocks in central California, Geol. Soc. Am. Bull. 84, 3489~512. Kohlstedt, D., C. Goetze, and W. Durham (1976). Experimental deformation of olivine single crystals with applications to flow in the mantle, in Petrophys~cs: The Physics and Chemistry of Minerals and Rocks, R. Strens, ea., John Wiley, New York, pp. 35 Js9. Kroner, A. (1977). The Precambrian geotectonic evolution of Africa: plate accretion versus plate destruction, Precambrian Res. 4, 163-213. Lachenbruch, A., and J. Sass (1977). Heat flow in the United States and the thermal regime of the crust, in The Earth's Crust, J. Heacock, ea., Am. Geophys. Union Geophys. Monogr. 20, pp. 626~75. Leyreloup, A., C. Dupoy, and R. Andriambololona (1977). Cata- zonal xenoliths in French Neogene volcanic rocks: constitution of the lower crust, Contrib. Mineral. Petrol. 62, 283~00. Lipman, P., 13. Doe, C. Hedge, and T. Steven (1978). Petrologic evolution of the San Juan volcanic field, southwestern Colo- rado: Pb and Sr isotope evidence, Geol. Soc. Am. Bull. 89, 59~2. Loomis, T. P. ( 1976). Kinetics of a gannet granulite reaction, Con- tnb. Mineral. Petrol. 62, 1-22. Lovering, J. F., and A. J. R. White (19t;8). Granulitic and eclogitic inclusions from basic pipes at Delegate, Australia, Contrib. Mineral. Petrol. 21, 9~. McGetchin, T. R., and L. T. Silver (1972). A crustal - upper mantle model for the Colorado Plateau based on observations of crystalline rock fragments in the Loses Rock Dike,J. Geophys. Res. 77, 7029-7037 Mehnert, K. R. (1975). lithe Inea Zone a model of the deep crust, Neues Jahrb. Mineral. Abh. 125. part 2, 156-199. Meyer, H. O. A., and D. G. Brookins (1976). Sapphirine, silli- manite and garnet in granulite xenoliths from Stockdale kim- berlite, Kansas, Am. Mineral. 61, 1194-1202. Necut, A., J. Connerney, and A. F. Kuckes (1977). Deep crustal electrical conductivity: evidence of water in the lower crust, Geophys. Res. Lett. 4. _39-~42. Ore ille, P. ~\I. ( 1963). Alkali ion exchange between ~ apor and feldspar phases, Am. J. Sci. 261, 678~83. Padovani, E., and J. Carter (1977). Aspects of the deep crystal evolution beneath south central New Mexico, in TI2e Earth's Crust, J. Heacock, ea., Am. Geophys. Union Geophys. Nlonogr. 20, pp. 19~.

150 Papike, J. J., K. L. Cameron, and K. Baldwin (1974). Amphiboles and pyroxenes: characterization of other than quadrilateral components and estimates of femc iron from microprobe data, Geol. Soc. Am. Abstr. Programs 6, 1053-1054. Phinney, D., I. Tennyson, and U. Trick (1978). Xenon in CO2 well gas revisited, J. Geophys. Res. 83, 2319-2319. Reiter, M., C. Shearer, and C. L. Edwards (1978). Geothermal anomalies along the Rio Grande riPc in New Mexico, Geology 6, 85-105. Rogers, N. W (1977). Granulite xenoliths from Lesotho kimber- l~tes and the composition of the lower crust, Nature 270, 681~84. Schmid, R., and B. J. Wood (1976). Phase relationships in granu- litic metapelites from the Irvea-Verbano zone (northern Italy), Contnb. Mineral. Petrol. 54, 255-279. Shieh, Y.-N, and H. P. Schwarcz (1974). Oxygen isotope studies of granite and migmatite, Grenville province of Ontario, Canada, Geochim. Cosmoch~m. Acta 38, 21~5. Shimazu, M., T. Yano, and M. Tazima (1978). Gabbroic inclu- sions in calc-alkalic rocks of the Fossa Magna, central Japan (abstr.), in International Geodynamics Conference "Western Pacific" and "Magma Genesis," ToJcgo, March 13-17, 1978, Inter-Union Commission on Geodynamics, and Science Coun- cil of Japan, pp. 324~25. Simmons, G., and D. Richter (1978). Microcracks in rocks, in The Physics and Chemistry of Minerals and Rocks, R. Strens, ea., Wiley-Interscience, New York, pp. 105-137. Smith, D. (1977). Hydrous minerals and carbonates in peridotite inclusions from the Green Knobs and Buell Park kimberlitic diatremes on the Colorado Plateau, in Extended Abstracts of the Second International Kimberlite Conference, 1977. Smith, D., and S. Levy (1976). Petrology of the Green Knobs diatreme and impIications for the upper mantle below the con- tinental plateau, Earth Planet. Sc:. Lett. 29, 107-125. Smithson, S., and S. Brown (1977). A model for lower continental crust, Earth Planet. Sci. Lett. 35, 132-144. Smithson, S., J. Brewer, S. Kauf~nan, J. Oliver, and C. Hunch (1978). Nature ofthe Wind River thrust, Wyoming, from COCORP deep-reflection data and from gravity data, Geology 6, 648~52. Stephenson, P. J., and T. J. Griffin (1976). Cainozoic Volcanicity North Queensland, Guidebook Excursion No. 7A, 25th Inter- national Geologic Congress, 39 pp. Stoesser, D. B. (1973). Mafic and ultramafic xenoIiths of cumulus origin, San Francisco volcanic f~eld, Arizona, Ph.D. disserta- tion, U. of Oregon. Takahashi, E. (1978). Petrological model of the upper mantle and the lower crust of the island arc: petrology of mafic and ul~a- mafic xenoliths in Cenozoic all~ali basalts of the Oki-Dogo Is- land in the Japan Sea (abstr.), in International Geodynamics Conference "Western Pacific" and "Magma Genesis," Tokyo, March 13-17, 1978, Inter-Union Commission on Geodyna- mics, and Science Council of Japan, pp. 334~35. ROBERT W. KAY and SUZANNE MAHLBURG KAY Tarney, J. (1976). Geochemistry of Archean high-grade gneisses, with implications as to the origin and evol~tion of the Precam- brian crust' in The Early History of the Earth, B. Windley, ea., Wiley-Interscience, New York' pp. 405~17. Taylor, S. R. (1977). Island arc models and the composition of the continental crust, in Deep Drilling Results in the Atlantic Ocean, Am. Geophys. Union Maurice Ewing Senes, Vol. 1, pp. 229-242. Taylor, H. P., and B. Ttlri (1976). FIigh ~xo igneous rocks from the Tuscan magrnatic province, Italy, Contrzb. Mineral. Petrol. 55, 33~5. Turi, B., and H. P. Taylor (1976). Oxygen isotope studies of potassic voIcanic rocks of the Roman province, central Italy, Cont~b. Mineral. Petrol. 55, 1~1. Vogel, D. E., and G. D. Garlick (1970). Oxygen-isotope ratios in metamorphic eclogites, Cont~b. Mineral. Petrol. 28, 188-191. Wass, S., and A. J. Irving (1976). Xenmeg: A Catalogue of Occur- rences of Xenoliths and Megacrysts in Basic Volcanic Rocks of Eastern Australia, Australian Museum, Sydney, 441 pp. White, A. J. R. (1964). Clinopyroxenes from eclogites and basic granulites, Am. Mineral. 49, 883~89. White, A. J. R., and B. W. Chappell (1976). Ultrametamorphism and granitoid genesis, 25th Int. Geol. Congr. Abstr. 3, 674~75. Whitford, D. J., W. Compston, I. Nicholls, and M. Abbott (1977). Geochemistry of late Cenozoic lavas firom eastern Indonesia: role of subducted sediments in petrogenesis, Geology 5. 571~75. Whitney, J. \, and J. C. Stonner, Jr. (1977). Geothennometry and geobarometry of epizonal granitic int~usions: a comparison of iron~itanium oxides and coexisting feldspars, Am. Mineral. 61, 751-761. Wyllie, P. J. (1971). Experimental limits for melting in the earth's crust and upper mantle, in The Structure and Physical Proper- ties of the Earth's Crust, J. Heacock, ea., Am. Geophys. Union Geophys. Monogr. 14, pp. 279~01. Wyllie, P. J., W. -L. Huang, C. R. Stem, and S. Maal~e (1976). Granitic magmas: possible and impossible sources, water con- tents and crystallization sequences, Can. 1. Earth Sci. 13, 1007-1019. Yoder, H. S., and C. E. Tilley (1'-~2). Origin of basalt magrnas: an experimental study of natural and synthetic rock systems, J. Petrol. 3, 342~32. Zartman, R. (1974). Lead isotopic provinces in the Cordillera of the westem United States and their geologic signif~cance, Econ. Geol. 69, 792~05. Zielinski, R. A., and P. W. Lipman (1976). Trace-element varia- tions at Summer Coon volcano, San Juan Mountains, Colorado. and ~e origin of continental-interior andesites, Geol. Soc. Am. Bull. 87, 1477-1485.

Geochemical Evolution of the Continental Crust l2 INTRODUCTION GILBERT N. HANSON State University of New York at Stony Brook To place limits on possible origins of the earth's conti- nental crust it is necessary to understand how the earth's crust has evolved and how the various processes acting have modified the geochemistry of the pre-existing crust. Prior to about 3900 million years (m.y.) ago the earth as well as the moon must have undergone significant infall of very large extraterrestrial bodies (Smith, 1976~. This bombardment must have played a significant role in crustal evolution. However, on the earth the record of this event has yet to be found. Thus speculations on the geo- chemical evolution of the continental crust based on the lithological record must start from 3800 m.y. ago, the age of the oldest terrestrial rocks found so far. The main purpose of this chapter is to suggest isotopic and trace-element approaches useful for studies leading to a better understanding of the geochemical evolution of the earth's continental crust. There are a number of recent papers pertinent to this topic, for example, Lowman (1976~; Tugarinov and Bilikova (19161; Smithson and Decker (19741; Smithson and Brown (1977~; Hargraves 151 (1976~; Taylor (in press); Tarney and Windley (1977~; Armstrong and Hein (19731; Jahn and Nyquist (19761; Moorbath (1977~; Heier (1973~; Tarney (1976~; Collerson and Fryer (1978~; Green (1972~; Pankhurst (19771; Brooks et al. (1976b); O'Nions and Pankhurst (19781; Oversby (19781; Engel et al. (19741; Shaw (1976), and O'Nions et al. (1979). Models for the evolution of the crust can be placed between two extreme schools ofthought (also see Chapter 15). One is that the continental crust formed early in the history of the earth (during the Archean) and that only small fractions of material have been added since then. The other model is that the continental culst has grown substantially since the Archean. Both models acknowl- edge the more or less continuous addition of igneous rocks into or upon the upper continental crust. There are, however, two possible sources for this material, the man- tle or the lower crust. Material added from the mantle will, of course, increase the mass of the continents, whereas material derived from the lower cn~st will not change the mass of the continental crust but only redis- tribute matter within it.

152 The continental crust makes up only 0.3 percent of the mass of the earth, but it is strongly enriched in elements such as K, U. Th, Rb, Ba, and Sr (Gast, 19601. Based on heat-flow data and the abundances of K, U. and Th, Heier (1973) suggests that the lower crust has lower abundances of these elements than the upper crust and that granulite- grade rocks of intermediate composition are reasonable candidates for the lower-crust composition. A~ more exten- sive argument for this model is presented by Smithson and Brown (1977~. One of the most important factors in any interpretation of crustal evolution is how mantle convection has changed with time. In the plate tectonic model, the crust is a pas- sive feature riding on lithospheric plates, the motions of which are determined by convection within the astheno- sphere. The igneous as well as tectonic activity within the crust is directly or indirectly related to activity in the asthenosphere. Thus, to understand crustal evolution it is essential also to understand the present convection re- gimes ofthe mantle, how these regimes may have evolved with time, and the possible interactions that various parts of the mantle may have had with the continental crust. As a first approximation, the upper mantle may be divided into two parts: the suboceanic mantle and the subcontinental mantle. Based on isotope and trace- element ratios for basalts, there are two principal sources of magma in the suboceanic mantle: one is the source of ocean-ridge basalts, the other the source of the ocean- island basalts. Radiogenic isotope data would suggest that the sources are separate and have been isolated for some 2000 m.y. (Church and Tatsumoto, 1975; Brooks et al., 1976a; Sun and Hanson, 1975~. The ocean-ridge types of basalts appear to be restricted to zones of spreading either at ocean edges or in marginal basins; basalts ofthe ocean- island type occur in nearly every tectonic environment in the oceans and continents (Schwarzer and Rogers, 1974~. Where the ocean-island-type basalts occur on continents, there may be little reaction with the continental crust (e.g., Ross Island, Sun and Hanson, 1976~. The mantle source for continental basalts (a large and geochemicaIly variable group of basalts), however, may in some cases have a history associated with the continents (Peterman et al., 1970; Leeman, 197~;; Brooks et al., 1976b), and the source may have interacted or mixed with crustal compo- nents (Faure et al., 1972; 1974~. RADIOGENIC ISOTOPES Some of the key data for understanding the evolutionary history of sources for igneous rocks are the initial isotope ratios of Pb, Sr, and Nd. It must be emphasized that the initial ratios alone cannot be used to tell whether the immediate source of a rock is the mantle or the crust. The isotopic ratios only allow an estimation of the U/204Pb, Rb/Sr, and Sm/Nd ratios of the source and a determination of the time these ratios existed. If continental evolution involves input of significant quantities of igneous rocks | BUSHVELD In ID ran Or =°~ G PEAT D I K E ~ =:=— _ 4.6 4.0 3.5 3.0 2.5 2.0 I.S 1.0 0.5 0.0 109 YEAR S GILBERT N. HANSON 40 707 0.706 0.705 ,ISLAND O704 0.703 BRIDGE 0702 . 0701 . 0.700 0699 FIGURE 13.1 Strontium evolution diagram for mantle with data for basic and ultrarnaf~c rocks modified from Jahn and Ny- quist ( 1976), with data for the Great Dyke and Bushveld Complex from Hamilton (1977). "Island" designates field for ocean-island basalts. "Ridge" designates field for ocean-ridge basalts. derived from the mantle, it is important to understand how the subcontinental and suboceanic mantle regimes have evolved. Figure 13.1 shows some schematic mantle evolution curves for Sr. The large variation in 87Sr/86Sr in modern oceanic basalts indicates that the suboceanic mantle has considerable heterogeneity. This heterogeneity may have also existed in the Precambrian, but the limited number of basaltic rocks analyzed may not adequately sample the Precambrian mantle. Hamilton (1977) suggests that the initial 87Sr/86Sr ratios for the 2100-m.y.-old Bushveld Com- plex, which vary Tom 0.7056 to 0.7086, may reflect a het- erogeneous mantle source variably enriched in Rb/Sr and is not a result of mixing with crustal components. If he is correct, prior to 2100 m.y. ago the subcontinental mantle in the vicinity of the Bushveld Complex had been vari- ably enriched in Rb/Sr for a significant period of time. Veizer and Compston (1976) have determined initial Sr isotope ratios on sedimentary carbonates throughout the geological record. If these values represent carbonates from oceanic environments, they should indicate the average Sr isotope ratios of the rocks supplying Sr to the oceanic environment. It can be seen in Figure 13.2 that the Sr isotope ratios in the Archean are low, typical of values assumed for the mantle. This may indicate that if the continents were extensive in the Archean, either they had low 87Sr/86Sr ratios and low Rb/Sr ratios or, if the continents had higher 87Sr/86Sr ratios, the strontium in the oceanic environment was predominantly derived from volcanic regimes and thus reflected a mantle source. After the Archean, the 87Sr/86Sr ratio of the carbonates in- creases significantly. This would suggest that the conti- nental source is more exposed and volcanics are less of a source or that there is significant growth ofthe continental crust at the end of the Archean. The same evolutionary

Geochemical Evolution of the Continental Crust 41~ K2O N a2o 0.71OI 87s r 86sr _ SEDIMENTS I GNEOU S ROC K S SEA WATER _ . 0.700 _ 0 1 109 YEARS FIGURE 13.2 87Sr/86Sr ratios in sedunenta~y carbonates (Veizer and Compston, 1976) and K2O/Na2O in sediments and volcanics (Engel et al., 1974) as a function of We age of the rocks. relationship can be seen in the K/Na ratio of sedimentary and volcanic rocks (Engel et al., 1974) and in the rare- ear~ elements content of sediments (Taylor, in press). Figure 13.3 is a single-stage mantle growth curve for Pb on a 207pbl204pb versus 206Pb/204Pb plot. The data from mod- ern oceanic basalts indicate that there is not a simple growth curve for the recent mantle. The oceanic basalt data lie along lines with slopes the ages of which are approximately 2000 m.y., suggesting that some 2000 m.y. ago their sources were variably enriched in the 238UI204Pb ratio relative to the growth curve. Some basalts from pre- sumed subcontinental mantle show a quite different rela- tionship. For example, the Pb isotope data for Absaroka volcanics from Wyoming (Peterman et al., 1970) lie about a line with a slope of 2800 m.y. These rocks, whether derived from the mantle or the lower crust, indicate a source that has had a low 238U/204Pb ratio with respect to the mantle growth curve since 2800 m.y. ago. This age is approximately that of the basement rock in this region. Leeman (1975) found similar results for basalts from the Snake River Plain as well as from Yellowstone National Park. He suggests that the trace- and major-element com- position of the basalts require their derivation from the mantle. In both studies, the lead and strontium isotopes are not correlated and cannot be explained by a simple mixing relation between crust- and mantle-derived end 153 members. These studies suggest that in these regions the subcontinental mantle has been attached to the conti- nental crust as a mantle keel since at least 2700 m.y. ago. The volcanics from these areas have 87Sr/86Sr ratios of 0.704~.709, on the higher end of the oceanic basalts, suggesting that although their source was depleted in U relative to Pb it was not depleted in Rb relative to Sr. If anything, it was enriched. Based on initial Sr, Pb, and Nd ratios, many granitic and basaltic rocks would appear to have either a source in the mantle or a source with only a short history in the crust (e.g., Moorbath, 1977; McCullogh and Wasserburg, 1978; DePaolo and Wasserburg, 1976; and O'Nions et al., 19791. This suggests an episodic and continuous addition of ma- terial to the crust through time. Further geochemical study of rocks derived from crustal sources, but with es- sentially mantle ratios, may make it possible to place limits on how the crust evolved and the times involved. Likewise, further geochemical study of rocks derived from the mantle may allow a characterization of the scale of heterogeneities in the mantle, show how they are evolving through time, and help to distinguish parts ofthe mantle interacting with the continental crust at a given time. As convection models for the mantle improve, this inforTnation should allow a direct correlation between an- cient tectonic regimes and convection in the mantle. PETROGENESIS OF IGNEOUS ROCKS Petrogenetic studies emphasizing isotope and trace- element analyses of a suite of igneous rocks are particu- larly pertinent for placing limits on the geodynamic fac- tors in the mantle responsible for tectonic activity in an area at the time of formation of the suite of rocks. The purpose of a petrogenetic study of an igneous suite is to determine the chemical and mineralogical composition of the source rocks at the time of melting; the history of the sources prior to melting; the extent of melting; the tem- perature (T) and pressure (P) or depth conditions during ~6 t5 - 1.0~ U OCEAN R ~ DO E A, 14 lo Cal D 1 3 Cal 1 2 11 ABSAROKA VOLC. _~ _ , , ~ _e~CEAN ISLAND BASALT S ~ ~ BASALTS TOO GNEISS. ISUA /4.o 0/ , ty 4.5,7 xl~y 10 9 10 11 12 13 14 15 t6 17 IS 19 20 2! 206 pb/20 P b FIGURE 13.3 Mantle growth curve for Pb with selected rock types plotted (modified from Tatsumato, 1978). Data for Absaroka volcanics are from Petennan et al. (1970). Data for Amitsoq gneisses, Isua, W. Greenland, are from Moorbath et al. (1975).

154 melting; and modification of the primary melts by dif- ferentiation, assimilation, metasomatism, or late-stage fluids. Although a petrogenetic study relies heavily on major-j minor-, and trace-element analysis and isotopic ratios, it must be based on rocks for which the field, geochronological, and petrologic relations are well un- derstood. The major-element analyses when combined with modal mineral data allow a comparison with experi- mental studies for estimating T. P. and volatiles during melting or differentiation. The isotopic data for Sr, Pb, and Nd allow an estimate to be made of the history of the U/Pb, Rb/Sr, and Sm/Nd ratios of the source. Modeling of Pace elements allows an estimate of the trace-element composition of the source, the mineral composition of the residue at the time of removal of the melt, sequences of fractional crystallization, and the extent of these pro- cesses. Along with the initial radiogenic isotope ratios, the trace elements allow an estimate ofthe extent of mixing or reactions with other melts or rocks (Vollmer, 1976; Lang- muir et al., 1978~. To obtain the maximum information, each of the dif- ferent types of analyses must be made on the same sam- ples. There are few places where a complete study can be made in one laboratory, and it may be questioned as to how many suites of rocks require such detailed analysis. The extensive data, however, are warranted for selected suites, because they can lead to a more quantitative in- sight into crustal evolution. Once the data are available, the best petrogenetic interpretation may not be immedi- ately obvious but will probably lead to new approaches and models. As an example of tectonic application, the petrogenesis of granitic rocks in two tectonic settings will be compared. The two settings are (1) an intrusive granite~reenstone belt in northeastern Minnesota in which all the rocks dated give ages of 2700 m.y. (Arth and Hanson, 1975) and (2) a high-grade Weiss terrane in southwestern Minnesota with ages as old as 3600 m.y. (G. N. Hanson, State Uni- versity of New York at Stony Brook, in preparation). In the northeastern Minnesota greenstone belt the initial 87Sr/86Sr ratios of basic, as well as granitic, rock are all between 0.700 and 0.701, suggesting that they were derived from a mantle source or sources with high Rb/Sr ratios that existed for only a short period of time prior to melting. Dacitic and tonalitic rocks have KlRb and Rb/Sr ratios similar to those of Archean tholeiite and strongly depleted heavy rare-earth element patterns. The model Mat best fits the data is that the dacites and tonalites are derived by partial melting of a tholeiitic parent, probably derived from an oceanic mantle, leaving a residue of gar- net and clinopyroxene. The quark monzonites *om the greenstone belt have lower K/Rb ratios and higher Rb/Sr ratios than the to- nalites and dacites, and rare-ear element patterns simi- lar to that of the tonalites and dacites but with higher abundances and negative Eu anomalies. The best model for the origin of the quartz monzonites is partial melting (upper amphibolite grade) of short-lived (<~;0 m.y.) grey- wacke. The greywacke consists of dacitic and tholeiitic GILBERT N. HANSON detritus derived from within the greenstone belt that has been enriched in K and Rb by sedimentary processes. In this greenstone belt all the components are thought to be derived from either the mantle or from rocks with short histories outside the mantle. The belt probably developed on an oceanic crust. If there were a continental crust underlying the greenstone belt, it was apparently not a major source for the volcanic or intrusive rocks analyzed. In the high-grade gneiss terrane in southwestern Min- nesota, the 3600-m.y.-old Morton and Montevideo gneisses were intruded by granitic rocks at 3100, 2600, and 1800 m.y. (Goldich et al., 1970; S. S. Goldich, North- ern Illinois University, and J. Wooden, Lockheed E:lec- tronics Company, in preparation; S. S. Goldich and C. E. Hedge, U.S. Geological Survey, in preparation). The gneisses vary from quartz diorite through quartz mon- zonite, and the intruding granitic rocks are granodiorite to quartz monzonite. The rare-earth element patterns for the gneisses and the later intruding granites are all very simi- lar to one another, suggesting that they have similar sources. These patterns are quite distinctive from those of the tonalites but similar to those of the quartz monzonites from northeastern Minnesota. Based ore the trace-element abundances and the geological relations, the best model is that the gneisses and the later granites are derived from melting of similar sources, presumably the lower conti- nental crust. This model is supported by Pb isotope data (Doe and Delevaux, in press), which suggest that the later granites are derived from related sources with a signifi- cantly long history in the crust. The K content of the gneisses, mainly tonalites, is generally lower than that of the later granites, mainly granodiorites to quartz mon- zonites. Lower K content for high-grade metamorphic rocks as compared with those of lower grade is not unusual (Heier, 1973~. This might imply that the gneisses originally formed under conditions that led to melts of lower K content or that the gneisses have lost K since the time of their origin. These two examples of petrogenesis would indicate that although the major-element compositions of quartz monzonite and quartz diorite are similar in both the intru- sive granite~reenstone belt and the ancient gneiss ter- rane, a more careful study oftheir chemistry and relations shows that the similarity is superficial and that the origins are probably quite different. The greenstone belt devel- oped in a short period of time and consists of rocks derived principally from the mantle or from rocks with a short history outside the mantle; whereas the gneiss ter- rane developed over a longer period of time, and the prin- cipal source for the granite rocks appears to be the melt- ing of pre-existing crustal sources. CHEMISTRY OF THE LOWER CRUST Based on heat-flow data, Heier (1973) suggested that the lower crust has lower abundances of K, U. and Th than the upper crust. If the lower crust is made up of granulite-

Geochemical Evolution of the Continental Crust grade rocks of intermediate composition, this could fit a model of a depleted lower crust and an enriched upper crust, because most granulite rocks are relatively de- pleted in K, U. and Th with normal abundances of Sr and Ba compared with similar rocks of lower grade. This re- sults in higher K/Rb (commonly 500 or greater) and lower Sr/Ba (~10), Rb/Sr (~0.02), and U/Pb ratios in granulite rocks (Tarney and \Vindley, 19771. The depletion in U is reflected in the low U/Pb ratios found in some granulite- grade rocks, leading to whole-rock leads that plot along isochrons below the mantle growth curve and to the left of the geochron in Figure 13.3. Is the relative depletion of these elements inherent in the origin of the types of rocks found in a granulite ter- rane, or have the rocks in a granulite tenant preferentially lost some of these elements? If the granulites have lost these elements, there are two means of transport: as melts or in aqueous or other solutions. An important difference between granulite and lower grades of metamorphism is the lower water content in the granulite facies rocks. Some of the loss of elements may thus be associated with the loss of water. One of the more surprising discoveries was that whereas fluid inclusions in rocks of amphibolite grade are rich in H2O, fluid inclusion in granulite-grade rocks have high proportions of CO2 (Touret, 1974), sug- gesting that the fluids with which they were in contact during high-grade metamorphism were CO2 rich. Gold- smith (1976) reminded us that a very important mineral in the lower crust is scapolite and that scapolite is a mineral into which substantial fractions of C03, S04, and C1 can be placed. He suggests that much of the carbonate is em- placed in the granulite terrane directly from the mantle. Lloyd and Bailey (1975) in studying peridotite nodules from the subcontinental mantle have found metasomatic textures, suggesting that normal lherzolite has been meta- somatized, resulting in the growth of titaniferous phlogo- pite, amphibole, diopside-salite, ferroaugite, titanomag- netite, sphere, perovskite, apatite, and calcite in what was originally lherzolite. It thus appears that many elements may be mobile in the mantle and are being added to the subcontinental mantle in carbonic or aqueous solutions. Wendlandt and Harrison (1978), for example, found that under mantle conditions CO2 vapor is 3 orders of magni- tude more enriched in rare-earth elements than is aqueous vapor and is also enriched in rare-earth elements relative to silicate melts. Shieh and Schwarcz (1974) have shown that the oxygen isotope ratios in rocks of the amphibolite grade in the Grenville province are characteristic ofthe~r unmetamor- phosed equivalents, whereas rocks of the highest meta- morphic grade have oxygen isotope ratios more indicative of the mantle. A similar relation has been found for Ar- chean rocks in the Superior province (Longstaffe and Schwarcz, 1977~. Although transporting material in the form of siliceous melts from the mantle to the crust or from the lower crust to the upper crust is undoubtedly important in terms of quantities of material moved, the effects of aqueous or carbonic vapors or fluids in transporting material within 155 the mantle, from the mantle to the crust, or within the crust may be significant. Particularly, they may play an important role in separating elements that behave simi- larly during magmatic processes. The solubilities of ele- ments in these fluids and mineral-fluid distribution coef- ficients must be determined experimentally under a variety of conditions so that a proper evaluation of these processes may be made. MANTLE-CRUST INTERACTION To understand the evolution of the continental crust it is necessary to understand how the mantle interacts with the continental crust. This requires characterizing variations within the mantle, determining their dimensions, and as- sessing how the variations are affected by mantle convec- tion. Sun and FIanson (1975) suggested that Rb~r and Pb-Pb isochron ages for ocean-island basalts of about 9000 m.y. reflect a real time of separation and isolation of the mantle sources for ocean-ridge and ocean-island ba- salts and that they are not the result of simple mixing between a large-ion-lithophile-element- (~IL) depleted ocean-ridge source and a -enriched ocean-island source. This is best shown in a plot of 87Sr/86Sr versus 206Pb/204Pb, in which the ocearl-ridge basalt plots away from the main trend of the data for the ocean islands and not at either end of a potential mixing curve. Although ocean-ridge basalts are only known to occur in spreading centers, whether at ocean ridges or in marginal basins, these environments encircle the globe. The ocean-island basalts are found in continental, island-arc, and oceanic terranes seemingly unrestricted in their geographic oc- currence. Thus both sources appear to be ubiquitous but separated. Until we have better information regarding convection in the mantle, the simplest model to explain these observations is a stratified mantle in which the source for the ocean-ridge basalts is a convecting mantle, below which is the source for the ocean-island basalts. This lower source may also be convecting id. Richter, University of Chicago, personal communication, 1978~. Applying this model to a continental environment, there may be a continental mantle keel attached to the conti- nental crust for hundreds to thousands of millions of years (Figure 13.41. In this model, starting from the leR side of Me figure and using the numbers in Figure 13.4: (1) Per- turbations in the convecting mantle produce upwellir~g, rifting, and melting of the continental mantle with the formation of continental basalts. The wide variety ofthese melts may or may not be a result of reaction with or melt- ing of the continental crust. (2) Carbonatites or lcimber- lites may result from melting or instability a: ;r the low- velocity zone. (3) Ocean-island-type basalts found on the continents are associated with deep-mantle plumes. (4) The addition of CON to the lower crust may be a result of continued production of CO2 over wide areas in the man- tle that reacts with the granulite-grade rocks in the lower crust, or it may be episodic, associated with tectonic dis- turbance.

156 FIGURE 13.4 Diagrammatic repre- sentation of present day mantle. Figure 13.4 also depicts a subduction zone (far right) on the continental margin in which there is extensive tec- tonic activity, (5) the connation of a marginal basin, and (6) in the arc, volcanism and the intrusion of gabbroic through granitic plutons. Below the arc there may be melting of: the subducted plate to produce tonalities; the subcontinental mantle or the ocean-ridge-type mantle to produce basalts; magic rocks near the base of the crust to produce a`nor~ositic or gabbroic plutons and possibly an- desites; and the intennediate-composit~on continental crust to produce granitic intrusions. The melting is prob- ably enhanced by the dehydration of the subducting plate. In the marginal basin the first volcanics would be derived by melting of the subcontinental mantle. As rift- ing proceeds and the marginal basin widens, ocean-ridge- type mantle becomes the dominant source of basalt. Detailed petrogenetic studies of suites of modern igne- ous rocks should allow testing of this and other models. Similar studies on other geological time spans should al- low an evaluation of the evolution of mantle regimes, mantle convection, and the interaction of the mantle with the continental crust. ACKNOWLE DGM E NTS S. R. Hart and I. Wooden reviewed tl~e manuscript. This report was supported by NSF Grant No. EAR 76-13354 AO1 (Geochemistry). REFERENCES Annstrong, R. L., and S. M. Hein (1973). Computer simulation of Ph and Sr isotope evaluation of the earth's crust and upper mantle, Geochim. Cosmochim. Acta 37, 1-18. GILBERT N. HANSON 1 2 3 4 5 6 ~ :— ~ ~ ~~ ~3~ ~ ~ = ~~;~ 1 {T^HOSPHERE~ .~ .~. . ~~ ~ +~N ~ 0. . ~ ~~ ~ ~ ~ ~ U ^, ~ ~ ~ U ~~ '4 ~ t~ ~ ~ * ,, ~ ~t- ^ _ 4 . - ~ ~ ~ n ~ ,~ .-, i a/ ,,. ~ At; At, ,,-,,, OCEAN RIDGE SOU RCE ~ ~ ~ ~ .~;~.*f r~1 B~> ~ / '1 ~ Iv ~ ~ A 7>1 1 ~1 V ~ ~ ~ L ~ ~ ~ < ~ ~ ~ ~ ~ ~ L ~1 ~ ~ A i, ~ ^~= ~ hi 1~1 ^~^ - ~1 - Arth, J. G., and G. N. Hanson (1975). Geochemistry and origin of the early Precambrian crust of northern Minnesota, Geochim. Cosmochim. Acta 39, 325~62. Brooks, C. R., S. R. Hart, A. Hoffman, and O. E. James (1976a). Rb~r mantle isochrons from oceanic regions, Earth Planet. Sci. Lett. 32, 51~1. Brooks, C., D. E. James, and S. R. Hart (1976b). Ancient lithosphere: its role in young continental volcanism, Science 193, 1086-1094. Church, S. E., and M. Tatsumoto (1975). Lead isotopic relations in oceanic ridge basalts from the Juan de Fuca~orda Ridge area N.~:. Pacific Ocean, Contrib. Mineral. Petrol. 53, 253-279. Collerson, K. D., and B. J. Fryer (1978). The role of fluids in the formation and subsequent development of the early con- tinental crust, Comets. Mineral. Petrol. 67, 151-167. DePaolo, D. J., and G. J. Wasserburg (1976). Nd isotope variations and petrogenetic models. Geophys. Res. Lett. 3, 249-252. Doe, B. R., and M. H. Delevaux (in press). Lead isotope investigations in the Minnesota River Valley I. Late- and post tectonic granites, Geol. Soc. Am. Afem. Engel, A. E. J., S. P. Itson, C. G. Engel, O. M. Stickney, and E. J. Gray, Jr. (1974). Crustal evolution and global tectonics: a petrogenetic view, Geol. Soc. Am. Bull. 85, 843-858. Faure, G., B. L. Hill, L. M. Jones, and D. H. Elliot (1972). Isotope composition of strontium and silica content of mesozoic basalt and dolerite from Antarctica, in Antarctica Geology and Geophysics, R. J. Adie, ea., Universitetaforloget, Oslo, pp. 413~18. Faure, G., J. Bowman, D. Elliot, and L. Jones (1974). Strontium isotope composition and petrogenesis ofthe Kirlcpatrick basalt, Queen Alexandria Range, Antarctica, Contrib. Mineral. Petrol. 48, 153-169. Cast, P. W. (1960). Limitations on the composition of the upper mantle,J. Geophys. Res. 65, 4. Goldich, S. S., C. E. Hedge, and T. W. Stern (1970). Age of the Morton and Montevideo gneisses and related rocks, southwestern Minnesota, Geol. Soc. Am. Bull. 81, 3671~696. Goldsmith, J. R. (1976). Scapolites, granulites and volatiles in the lower crust, Geol. Soc. Am. Bull. 87, 161-168.

Geochemical Evolution of the Continental Crust Green, D. H. (1972). Magrnatic activity as the major process in the chemical evolution of the earth's crust and mantle, Tec- tonophyszcs 13, 47-71. Hamilton, J. (1977). Sr isotope and trace element studies of the Great Dyke and Bushveld mafic phase and their relation to early Proterozoic magma genesis in southern Africa, J. Petrol. 18, 24~. Hargraves, R. B. (1976). Precambrian geologic history: growth and emergence of continents from beneath the primordial sea, Science 195, 363~71. Heier, K. S. (1973). A model for the composition of the deep continental crust, Fortschr. Mineral. 50. 174-187. Jahn, B. M., and L. E. Nyquist (1976). Crustal evolution in the early earth~noon system: constraints from Rb-Sr studies, in The Early History of the Earth, B. F. Windley, ea., John Wiley, New York, pp. 55-76. Langmnir, C. H., R. D. Vocke, Jr., G. N. Hanson, and S. R. Hart (1978). A general mixing equation with applications to Ice- landic basalts, Earth Planet. Sci. Lett. 37, 380~92. Leeman, W. P. (1975). Radiogenic tracers applied to basalt gen- esis in the Snake River Plain-Yellowstone National Park region—evidence for a 2.7 b.y. old upper mantle keel, Geol. Soc. Am. Abstr. Programs 7, 165. Lloyd, F. E., and D. K. Bailey (1975). Light element meta- somatism of the continental mantle: evidence and conse- quences, Phys. Chem. Earth 9, 389~16. Longstaffe, F. J., and H. P. Schwarcz (1977). '8Ol'6O of Archean elastic metasedimentary rocks: a petrogenetic indicator for Ar- chean gneisses?, Geochim. Cosmochim. Acta 41, 1303-1312. Lowman, P. D., Jr. (1976). Crustal evolution in silicate planets: implications for the origins of continents, J. Ceol. 84, 1-26. McCullogh, M. T., and G. J. Wasserburg (1978). Smoked and Rb~r chronology of continental crest formation, Science 200, 1003-101 1. Moorbath, S. (1977). Ages, isotopes end evolution of Precambrian continental crust, Chem. Geol. 20, 151-187. Moorbath, S., R. K. O'Nions, and B. J. Pankhurst (1975). The evolution of early Precambrian crustal rocks at Isua' West Greenland~eochemical and isotopic evidence, Earth Planet. Sci. Lett. 27, 229_2:39. O'Nions, R. K., and R. J. Pankhurst (1978). Early Archean rocks and geochemical evolution of the ear~'s crust, Earth Planet. Sci. Lett. 38, 211-236. O'Nions, R. K., S. B. Carter, N. M. Evensen, and P. J. Hamilton (1979). Geochemical and cosmochemical application of Nd- isotopes, Ann. Rev. Earth Planet. Sci. 7, 11~8. Oversby, V. M. (1978). Lead isotopes in Archean platonic rocks, Earth Planet. Sci. Lett. 38, 237-248. Pankhurst, R. J. (1977). Strontium isotope evidence for mantle events in the continental lithosphere, J. Geol. Soc. 134, 255. Peterman, Z. E., B. R. Doe, and H. J. Prostka (1970). Lead and strontium isotopes in rocks of He Absaroka volcanic field, Wyoming, Cont~b. Mineral. Petrol. 27, 121-130. 157 Schwarzer, B. R. and J. J. W. Rogers (1974). A world-wide com- panson of alkali olivine basalts and their differentiation trends, Earth Planet. Sci. Lett. 23, 286-296. Shaw, D. M. (1976). Development of the early continental crust, Part 2: Prearchean, Protoarchean, and later eras, in The Early History of the Earth, B. F. Windley, ea., John Wiley, New York, pp. 33-53. Shieh, Y. N., and H. P. Schwarcz (1974). Oxygen isotope studies of granite and migmatite, Grenville province of Ontario, Canada, Geochzm. Cosmochim. Acta 38, 21~5. Smith, J. V. (1976). Development of the earth-moon system with implications for the geology of the early earth, in The Early History of the Earth, B. F. Windley, ea., John Wiley, New York, pp. 3 '19. Smithson, S. B., and S. K. Brown (1977). A model for lower conti- nental crust, Earth Planet. Sci. Lett. 35, 134-144. Smithson, S. B., and E. R. Decker (1974). A continental crustal model and its geothermal implications, Earth Planet. Sci. Lett. 22, 215-~5. Sun, S. S., and G. N. Hanson (1975). Evolution of the mantle: geochemical evidence from alkali basalts, Geology 3, 297 302. Sun, S. S., and G. N. Hanson (1976). Differentiation of McMurdo volcanics, Ross Island, Antarctica: evidence from BEE, Contnb. Mineral. Petrol. 54, 139-155. Tarney, J. (1976). Geochemistry of Archaean high-grade gneisses, with implications as to the origin and evolution of the Precam- bnan crust, in The Early History of the Earth, B. F. Windley, ea., John Wiley, New York, pp. 405~17. Tarney, J., and B. F. Windley (1977). Chemistry, thermal gra- dients and evolution of the lower continental crust, J. Geol. Soc. London 134,153-172. Tatsumoto, M. (1978). Isotopic composition of lead in oceanic basalts and its implications to mantle evolution, Earth Planet. Sci. Lett. 38, 63~7. Taylor, S. R. (in press). Chemical composition and evolution of He continental crust: the rare earth element evidence, in The Earth: Its Origin, Structure arut Evolution, M. W. McElhinny, ea., Academic Press, New York. Touret, J. (1974). Facies granulite et inclusions carboniques, Ann. Soc. Geol. Belgique, Vol. Spec. P. Michot, pp. 267~!87. Tugarinov, A. T., and Ye. V. Bilikova (1976). Evolution of the chemical composition of the earth's crust, GeokEimiya 8, 1151-1 159. Veizer, J., and W. Compston (1976). 87Sr/86Sr in Precambrian car- bonates as an index of crustal evolution, Geochim. Cosmochim. Acta 40, 905~14. Vollmer, R. (1976). Rb~r and U-Th-Pb systematics of alkaline rocks: the alkaline rocks from Haly, Geochim. Cosmochim. Acta 40, 283-295. Wendlandt, R. F., and W. J. Harrison (1978). Rare earth element partitioning between coexisting immiscible carbonate and sili- cate liquids and CO2 in the system K2o-Al2orsio~co2' Geol. Soc. Am. Abstr. Programs 10, 514.