Reviewed research article Paleomagnetic research on Icelandic lava ows Leó Kristjánsson Institute of Earth Sciences, University of Iceland, Sturlugata 7, 101 Reykjavík leo@raunvis.hi.is Abstract — Following a short review of the main efforts in paleomagnetic eld work on basement rocks in Iceland since 1950 and a description of their main magnetic characteristics, some examples of paleomagnetic research on these rocks are described. This research has had two major aims. One has concerned the stratigraphy of the lava pile, where reversals of the remanence polarity are useful in correlating between mapped proles of similar age some kilometers apart. The other aim has been to obtain information on various aspects of the geomagnetic eld conguration and its changes through the last 15 M.y. Additionally, knowledge of the magnetic properties of basement rocks is essential in the geological interpretation of magnetic anomalies. INTRODUCTION – HISTORICAL The area of research known as “Paleomagnetism” originated as the study of the Earth’s past magnetic eld, by means of measurements on stable components of remanent magnetization in rocks. For general properties of the geomagnetic eld, the reader is referred to books such as Lanza and Meloni (2006). The subject of paleomagnetism has also come to include other applications of the stable magnetic remanence directions e.g. in age estimates, regional tectonics, or rock deformation. Therefore, research which is concerned solely with actual aspects of geomagnetic elds has in recent years been referred to by the more appropriate name “Paleo-geomagnetism”. A complementary research area called “Rock magnetism” embraces the magnetic properties of rocks and minerals in general, including the mechanism of remanence acquisition. Although remanence measurements on samples of igneous rocks from Iceland (at 64–66◦N) were initiated before 1930, their potential for paleomagnetic purposes was rst appreciated through the work of J. Hospers in the early 1950’s (e.g. Hospers, 1953). A few years later, Sigurgeirsson (1957) and others improved the techniques of Hospers and demonstrated additional applications of lava remanence directions, both in local geological research and in a global context. Subsequently, a U.K.-Icelandic expedition in 1964–65 carried out major sampling in East Iceland (Dagley et al., 1967; Watkins and Walker, 1977). Their collection included over 850 distinct lava ows and additionally about 200 thin ow units within those lavas that were classied as compound ows. The samples were used for a variety of paleomagnetic, rock-magnetic, and dating studies. Among the accomplishments of the above pioneers we may mention: • Early applications of some key concepts, e.g. “virtual geomagnetic pole” (VGP) and “Fisher statistics” (Fisher, 1953). • Development of the alternating-eld (AF) technique for eliminating secondary viscous remanent magnetization (VRM) • Conrmation of the occurrence of geomagnetic reversals at variable intervals through the period 2–13 M.y. ago, at a rate of at least 5–6 per M.y. • Application of polarity reversals in stratigraphy across 10 km or more (Figure 1) • Demonstration of the long-term dependence of the (virtual) geomagnetic dipole moment upon VGP latitude JÖKULL No. 58, 2008 101 L. Kristjánsson Research on Icelandic rocks in the 1960’s also contributed to an improved understanding of the physico-chemical factors inuencing remanence stability (see Ade-Hall et al., 1971, and references therein), and helped in establishing the reality of short polarity-reversal events (Wensink, 1964). From the work in the 1950’s–1960’s it became evident that basalts in Iceland are characterized by highly stable primary remanence directions and constitute some of the best material available anywhere for paleo-geomagnetic research. This statement however, applies without qualication only to relatively fresh lavas, i.e., those that have not suffered secondary regional hydrothermal alteration to a degree characterized by the presence of the zeolites mesolite and scolecite in the lava pile. It may indicate a maximum temperature of around 150◦C although both higher and lower temperature estimates have appeared in the literature. Only limited studies have been carried out on more altered rocks (in deeply eroded localities, in drillholes, and in the vicinity of extinct or active central volcanoes). They indicate that in lava ows heated beyond some 200◦ C (characterized by laumontite) serious decay of the original remanence may have taken place (Watkins and Walker, 1977). In lavas suffering more than 250–300◦C reheating (epidote zone), secondary magnetite has appeared. Presumably, that magnetite has acquired a thermal remanence while buried within a large body cooling slowly from the maximum temperature reached, the geomagnetic eld even reversing one or more times during that time. These observations have a bearing on the signicance of some paleomagnetic directions reported from igneous rocks in much older localities abroad, where Figure 1. The south-facing side of mt. Kistufell, part of the Esja central volcano complex near Reykjavík, rises from an altitude of 100 m to 750 m. Above a hyaloclastite that reaches to about 300 m, 65 lava ows are found. They presumably date from the Matuyama geomagnetic chron (2.6–0.8 M.y. ago) and are reversely magnetized except for 13 ows at around 500 m altitude. This thin “N3” polarity zone (Sigurgeirsson, 1957; Kristjánsson et al., 1980; Goguitchaichvili et al., 1999) can be followed in outcrops for at least 20 km to the northeast and is also found 4–5 km to the south-southeast. – Suðurhlið Kistufells í Esju rís úr 100 m í 750 m hæð, og er að mestu samsett úr hraunlögum ofan við um 300 m. Sú syrpa 65 laga er með “öfuga” segulstefnu fyrir utan 13 lög um miðbikið sem má rekja bæði inn eftir Hvalrði og til suðurs. 102 JÖKULL No. 58, 2008 Paleomagnetic research on Icelandic lava ows the rocks have been subjected to even more severe thermal and chemical inuences as well as mechanical deformation. Paleomagnetic eld work in Iceland in 1972–78, especially a large collection of lava samples acquired by N.D. Watkins and collaborators, aided progress in paleomagnetism in various ways. Their work included radiometric dating of some of the geomagnetic reversals occurring in the last several M.y. (McDougall et al., 1977), and statistical processing of remanence vectors from over 2100 stable lava ows (Kristjánsson and McDougall, 1982). ROCK TYPES PRESENT IN ICELAND, AND THEIR GENERAL MAGNETIC PROPERTIES Volcanism in Iceland (as exposed now above sea level) has been essentially continuous for the past 15 M.y., concurrent with steady east-west spreading of the crust away from active zones. Those rock formations which have not been erupted under Pleistocene glaciers or lakes consist mostly of basalt lava ows with thin intervening sediments. It is thought that the lavas were generated in eruptions which chiey took place in so-called central-volcano complexes. Each such complex may have had a lifetime of some 0.5 M.y. Their main features include a caldera structure of several km size accompanied by intrusions and a zone of relatively intense hydrothermal alteration, as well as a dike swarm of tens of km in length. The volcanic activity built up a plateau which was subsequently eroded and dissected by fjords, creating excellent exposures of the lava pile. Due to a continuous process of subsidence of the areas around the volcanic zones, the lava sequences tend to tilt towards these by a few to several degrees, but that situation is complicated by eastwards jumps of these zones which appear to have taken place a few times in the past 15 M.y. The dominant magnetic mineral in Icelandic basalts is titanomagnetite in various stages of oxidation. The titanium content is reected in the Curie point (Tc ) which in fresh rapidly cooled material such as pillow basalt may even be below 100◦ C. Most subaerial lava ows exhibit nal Curie points exceeding 500◦C when heated in air, but the thermomagnetic curves have a very varied appearance. In many of these cases, a homogeneous component with a Curie point of the order of 300◦ C or less is apparently being converted to a magnetite-ilmenite intergrowth as the heating proceeds. Lava samples that have been oxidized during their initial cooling, often have Curie points (or at least tails on the thermomagnetic curves) exceeding the 578◦ C value for pure magnetite, which points to the presence of a stable maghemite. In those having a reddish color, as well as in red interbasaltic sediments, Tc -values may reach 620◦ C. The thermomagnetic curves are irreversible in most cases, cf. typical examples illustrated in Goguitchaichvili et al. (1999) and Kristjánsson et al. (2003). An increase of susceptibility upon return to room temperature after heating to 300◦ C or more often occurs in connection with the conversion of a low-Tc component; the author has even observed a consistent increase (averaging 5%) in this property after heating 23 stable lava samples from Northwest Iceland to 160◦ C (L. Kristjánsson, unpublished data 1968, 2007). A decrease in susceptibility of a specimen with a high Curie point after heating, which is also often noted, is probably associated with the formation of minerals like hematite and pseudobrookite. No clear correlation between magnetic properties and silicate lithology has been noted. Typical values for the initial susceptibilities of lava samples are of the order of 0.025 SI units, indicating roughly a titanomagnetite + maghemite content of 1% by volume. Overall remanence intensities in large collections of Tertiary lavas after 10 mT AF treatment are around 3–4 A/m. Lavas with a weak and unstable remanence tend to have high room-temperature magnetic susceptibility and somewhat complex thermomagnetic curves. SAMPLING AND MEASUREMENT PROCEDURES, WITH EMPHASIS ON LAVA FLOWS Since 1980 paleomagnetic eld work has been carried out every summer, in various regions of Iceland. Usually, local geologists have been involved in the initial stages of these projects. Their task has consisted in JÖKULL No. 58, 2008 103 L. Kristjánsson choosing an interesting and favorable area, surveying it to nd proles that are suitable for sampling, mapping these proles in detail, establishing correlations between them, and numbering the lava ows. The sampling has generally included 4 core samples of 4–8 cm length per ow: an extra core is sometimes added e.g. in outcrops that are crumbly or are suspected of having been disturbed in some way. The standard procedure of measurement at the Science Institute, University of Iceland employs an “Institut Dr. Förster” four-probe static uxgate meter, calibrated at intervals with a 19-turn coil of the same shape as the standard core specimens. It has been checked against other instruments and found to give very satisfactory results. AF demagnetization is done with a 2-axis tumbler: a bulky system of current coils was replaced by a compact “Molspin” instrument in 1989. In the majority of the lava samples, a normal-polarity viscous magnetization has been completely eliminated at 10 mT AF treatment. In order to ascertain this, measurements are also made after 15 and 20 mT. In the infrequent cases when the directional change between these two is 2◦ or more (including alignment errors in replacing a specimen in its holder), the treatment is repeated at 20 mT and the results averaged. If the remanence direction appears to be changing systematically, the treatment is extended as necessary. Very few samples are rejected due to e.g. lightning strikes, suspected baking, or gross orientation errors. In some surveys there have been opportunities to resample problematic lavas, usually resulting in improved condence parameters of mean remanence directions. Rotational remanence (RRM) which apparently may be acquired by some types of igneous rocks during AF treatment at high eld intensities, is hardly ever noticed in Icelandic lavas. A problem which is fairly common but easily dealt with, is the rapid acquisition of new VRM while specimens are being transferred from the demagnetizing equipment to the measuring device. The high stability of primary remanence directions in Icelandic basalts, evident from the above description, is also reected in excellent agreement of directions within sampling sites after simple AF treatment. Typical values for the 95%-condence radii of 104 JÖKULL No. 58, 2008 average directions from four samples spread over a few to several meters, are 5◦ or less after demagnetization at 20 mT peak eld. The median destructive elds (MDF) at which the treatment has reduced the remanence intensity to half its initial value, lie commonly in the range 10 to 40 mT, as shown by examples in several papers including Kristjánsson (2002) and Kristjánsson et al. (2003). However, directions even from samples where the MDF may be of the order of 5 mT or less, usually agree with each other or with more stable samples from the same ow after careful AF treatment. Tests (Kristjánsson and Auðunsson, 2007; Kristjánsson and Jónsson, 2007) on widely distributed sampling sites in several lava ows or owunit series have conrmed that the sites yield uniform directions (within a few degrees) even when spread over distances of 1 km or more. All these successful demonstrations of consistency have enabled the acquisition of reliable directions from a large number of lavas, without need for spending time on excessive sampling, extended demagnetization treatments, or sophisticated statistical processing. The circumstances have also resulted in an emphasis on magnetic studies of lava ows; much less information has been gathered on the paleomagnetism of Icelandic dikes and other intrusions, subglacial extrusives, or sediments, where consistency checks cannot be made as thoroughly as in lava ows. STRATIGRAPHIC APPLICATIONS OF PALEOMAGNETISM IN ICELAND Stratigraphic mapping of the lava pile and central volcanoes in East and Southeast Iceland was initiated by G.P.L. Walker in the mid-1950’s (e.g. Walker, 1959). For correlation he employed occasional extensive clastic sediments as well as fairly distinct groups of feldspar-porphyritic lavas and of olivine tholeiites which are found within the predominantly tholeiitic lava pile. The mapping efforts of Walker, his students and other successors have formed a very useful foundation for paleomagnetic studies in Iceland. These studies have concentrated on the regularly layered Tertiary to Early Pleistocene lava pile where a large number of well-exposed mountainside or stream Paleomagnetic research on Icelandic lava ows proles are accessible. Much less has been done in the late Pleistocene to Recent active volcanic zones (cf. the section “Conclusions and Discussion” below), although paleomagnetic work has e.g. been demonstrated to provide quite useful results on the extent of certain contemporaneous sequences in these areas (Levi et al., 1990). A unique aspect of the lava sequences in Iceland is that they cover in a fairly continuous fashion the time interval from 0 to 15 M.y. ago, during which the geomagnetic eld reversed its polarity at least 120 times (Kristjánsson and McDougall, 1982) including shortlived events. Along with the easy accessibility, stability and reliability of the lavas as already mentioned, this allows the charting of some average properties of the geomagnetic eld in the above period. Due to the unknown and variable but usually rather long time interval (thousands of years, cf. below) inferred to have passed between eruption of any two successive lavas in the Icelandic lava pile, details such as paths of the VGP during individual geomagnetic polarity transitions and excursions are very rarely recorded. It should be noted here that the average thickness of a lava ow in the older parts of Iceland is of the order of 10 m. The duration between two successive ows appears to vary widely, with an overall mean of the order of 5–10000 years. That estimate does not include the above-mentioned ow units, which (judging e.g. from the similarity of their remanence directions) may be separated in time by only years to centuries. Several sampling proles, usually composed of 20–80 lavas each, some km distant from each other in the down-dip direction, and partly overlapping in age, may be connected together into a composite section. In the last few decades, several sections were studied in this way (Figure 2). Each such collection usually contains over 300 lava ows (including prole overlaps), covering an interval of millions of years (e.g. Figure 3). The number of lava ows in a zone of uniform polarity within such a stratigraphic column can vary between 1 and 70, but an average value is 15–20 lava ows. Paleomagnetic polarity results on the lava ows have therefore often been helpful for geologists in their stratigraphic work in Iceland. The polarity of the natural remanence (NRM) in hand samples can generally be measured in the eld with a portable uxgate magnetometer. However, caution is needed in such measurements and their interpretation because the total magnetization of a hand sample in this situation may include signicant induced and viscous (VRM) components in addition to its primary thermal (TRM) or thermo-chemical remanence. SOME PROPERTIES OF THE PALEO-GEOMAGNETIC FIELD IN ICELAND A number of long-term properties of the paleomagnetic eld can be studied when large collections of data from basalt lava sequences spanning many geomagnetic reversals are available. The results from such studies in a relatively small area like Iceland can provide important constraints on models of the general conguration of the eld and theories of its generation in the Earth’s core. A few simple studies of this type will be addressed in separate sub-sections below. Other properties which have long been discussed among paleomagnetists and which have been treated in recent papers on Icelandic lava collections, include: • Normal/reverse asymmetry as regards remanence intensity, chron lengths, secular variation parameters and so on (see Kristjánsson, 1999, 2002) • The frequency distribution of virtual geomagnetic poles as a function of geographic latitude, and the relative frequency of occurrence of excursions and transitions (see Kristjánsson, 1999; Kristjánsson and Jónsson, 2007). These distributions tend to have a long tail towards the Equator, compared to e.g. a Fisher’s (1953) distribution peaking at similar VGP latitude. • The overall frequency of reversals and major eld excursions (Kristjánsson and McDougall, 1982). Progress regarding estimates of this kind has been hampered by the fact that few reliable radiometric dates on Icelandic Tertiary lavas have been published for the last two decades. JÖKULL No. 58, 2008 105 L. Kristjánsson It would be valuable to have access to another set of lava ows of comparable age range, size and reliability from a site at a different latitude, but such sets do not seem to be existing. Possibly, some of the above-mentioned results from Iceland can be compared to carefully compiled results from many small surveys on magnetically stable lavas around the globe. However, statistical results from the latter as presented in publications, have often been unnecessarily modied by procedures such as arbitrary rejection of low- and mid-latitude VGPs. Figure 2. Sampling of lava ows in Iceland, 1964–2007. Circles refer to surveys where two samples were collected per ow (mostly in 1964–65), triangles three (mostly in 1972–78) and squares four (from 1979). Each symbol stands for 60–100 lava ows. Crosses indicate smaller isolated studies on 35–60 lavas each. The approximate positions of surface exposures of 0.8, 3.3 and 10 M.y. age are shown. Some sampling expeditions from which direction results have not been published in detail, are not included. The map is revised and simplied from Kristjánsson and Jónsson (2007) where the relevant publications are also listed apart from the recent ones by Kristjánsson et al. (2006, two valleys in N-Iceland) and Kristjánsson (in preparation 2008, southwest part of the Northwestern peninsula). – Yrlitskort af mestallri sýnasöfnun til bergsegulmælinga á Íslandi frá 1964, ef niðurstöður um segulstefnur hraunanna hafa birst eða verða það jótlega. Yrleitt hafa verið tekin fyrir fáein hundruð lög á hverju svæði. Hvert tákn á við 60–100 hraunlög nema krossar, sem sýna smærri verkefni. 106 JÖKULL No. 58, 2008 Paleomagnetic research on Icelandic lava ows Figure 3. Left: The positions of sampling proles in one typical local survey of paleomagnetic directions in the lava pile of Iceland. This survey encompassed two valleys in the Skagafjörður district of North Iceland. As the tectonic tilt here is generally to the south, ages are expected to decrease southwards, from some 9 M.y. at the base of TD to around 5 M.y. at the top of PO. Center: Altitudes of the proles, and some suggested correlations between them. Right: A tentative polarity column in the composite section derived from the proles (180 lavas, after elimination of prole overlaps). Black: normal polarity. Dotted: sedimentary beds. Ages are somewhat uncertain, as radiometric dates are only available from locations west of TD and east of PO, and unconformities have been noted within the area. Simplied from Kristjánsson et al. (2006). – T.v.: Kort af sýnasöfnun í Skagafjarðardölum, í nýlegri dæmigerðri jarðlaga-kortlagningu hraunastaa. Í miðið: Hæðir sniðanna sem safnað var úr, og nokkrar tengingar milli þeirra. T.h.: Segulstefnurnar í samsettu sniði, að skörun slepptri. Mean geomagnetic pole positions It is generally considered by paleomagnetists that the geomagnetic eld direction when sampled many times at a single location during a long time interval and averaged, approximates that of an axial central dipole (cf. Hospers, 1953). This has been a foundation for common applications of paleomagnetic studies worldwide, in the estimation of geological ages, continental drift, and apparent polar wandering. The number of independent “spot readings” of the eld required for obtaining an average direction with less than 5◦ (say) condence radius, has often been given in the literature as 20 or less, which however (even in the absence of uncertainties in tilt corrections, etc.) is in fact far too small, see Section 5 of Kristjánsson (2002). It is also obvious that a set of remanence directions to be averaged should preferably span a few to several polarity reversals. Figure 4 shows mean VGPs from twelve such large (> 300) collections of lava ows in Iceland. Using Fisher’s (1953) statistical parameters the 95% condence radii (α95 ) are around 3◦ for each mean pole, which would indicate that at least four of those in the Figure are signicantly different from the geographic pole. However, the α95 -values derived in this way are probably underestimated, both because the VGPs do not in fact follow Fisher’s distriJÖKULL No. 58, 2008 107 L. Kristjánsson bution and because of serial correlation, i.e. clustering of VGPs which often is seen in two or more successive lavas. See a discussion of the latter point in particular cases by Kristjánsson et al. (1980, 2003, 2006). Those poles which are most “right-handed” (i.e. nos. 7/7a, 10, 12/12a and 14) include some of the oldest rocks in the country, but otherwise their positions do not show a systematic variation with age. Divergence of these means from the geographic pole may in some cases be in part due to e.g. incorrect estimates of the local tectonic tilt. Tectonic rotations about a vertical axis have not been considered here, although it is for instance possible that they occur through “bookshelf faulting” in some tectonic situations in Iceland (see LaFemina et al., 2005). Paleo-secular variation: eld and dipole strength as function of VGP latitude Sigurgeirsson (1957) noted that the remanence intensity in hand samples from lavas with intermediate (transitional) virtual geomagnetic poles was in general lower than in lavas with high-latitude poles. These intensities are of course inuenced by a large number of rock properties which vary greatly both within and between lava ows, as well as by the ambient eld intensity. As the other properties vary irregularly with time in an independent fashion, it may be assumed that effects of the eld dominate in suitable comparisons between sufciently large collections of results. Long-term trends in the eld intensity with time cannot be extracted from such collections due to the possible effects of long-term variations in the composition of erupted magma, and alteration effects including viscous decay of the primary remanence. On the other hand, variations in the eld intensity with for instance VGP latitude should show up in large collections. Dagley and Wilson (1971) made the rst attempt at quantifying this relationship by plotting average remanence intensities (after a selected AF demagnetization step) against VGP latitude in two data sets from Icelandic lava ows. Similar attempts involving considerably larger sample collections were published by Kristjánsson and McDougall (1982) and Kristjánsson (1995, 1999). In the lastnamed publication, data from 3514 lava ows were processed. 108 JÖKULL No. 58, 2008 Figure 4. Mean VGPs from 12 collections of lava ows, varying in size from 310 to 580 ows. Those nos. 1–12 are taken directly from the compilation in Kristjánsson and Jóhannesson (1989) as extended by Kristjánsson (1995). Their mean poles no. 2, 4 and 8 which were derived from collections with mostly two samples per lava, have been omitted here. The rejection criterion for within-ow consistency was α95 = 20.5◦ . In three more recent collections where four samples were collected per ow, very few withinow α95 values exceed 15◦ . Of these, nos. 13 and 14 are from Kristjánsson et al. (2003, 2004); the mean pole no. 15 combines results from Kristjánsson and Guðmundsson (2001, 2005) and Kristjánsson et al. (2006). The total number of ows is 5029; note that low-latitude VGPs are not excluded from these averages. One typical α95 -circle of radius 3◦ is shown. Due to the non-linear relation between virtual geomagnetic paleolatitude and eld inclination, poles corresponding to the mean eld direction in each collection would lie some 2◦ farther away from Iceland. – Meðalstaðsetning sýndar-jarðsegulskauts í 12 syrpum hraunlaga á Íslandi sem bergsegulmælingar hafa verið gerðar á; í hverri eru yr 300 lög. Paleomagnetic research on Icelandic lava ows Figure 5 shows the results from a new analysis based on the data set of Kristjánsson (1999) and a large number of data from Iceland (almost all with 4 samples per lava) obtained subsequently. Figure 5. Bottom points: results from remanence intensity measurements (after 10 mT AF treatment) in 4970 Icelandic lavas of 1–15 M.y. age; geometric averages at 10◦ intervals in VGP latitude. In the upper two sets of points, the intensity values have been converted for each lava (i.e. multiplied by a factor between 1 and 2 depending on paleo-geomagnetic latitude, assuming a central dipole eld) to the value that would have been observed if that lava had been erupted at its VGP. Middle points: geometric averages, with a curve showing one possible simple interpretation. Upper points: arithmetic averages, with standard errors indicated in two cases. – Meðalstyrkur segulmögnunar í íslenskum hraunlögum; efri tveir punktahóparnir eru úr gildum reiknuðum fyrir hvert hraunlag um sig ef það hefði verið statt á viðkomandi sýndarsegulskauti. Þeim hefur verið skipt í hópa á 10◦ - bili í breiddargráðu sýndarsegulskautsins. Neðri tveir punktahóparnir eru geometrisk meðaltöl, sá efsti eru venjuleg meðaltöl fyrir sömu hraunlögin. The acceptance criterion has been tightened considerably from the within-lava directional α95 value of 20.5◦ used by Kristjánsson (1995, 1999). In the present case we reject those lavas which have α95 values greater than 16◦ if the VGP latitude is less than 40◦ N or S, greater than 14◦ if the VGP latitude is between 40◦ and 50◦ , and greater than 12◦ if the VGP latitude is greater than 50◦ . This sliding cutoff criterion is an attempt to allow for the fact that remanence direction measurements on lavas yielding low-latitude VGPs tend to be less accurate than others. The results are shown in Figure 5 for averaged remanence intensities (after 10 mT AF treatment) in 10◦ - intervals of VGP latitude. The total number of acceptable lavas is 4970, of which 284 are in the three groups with VGP latitudes between 0◦ and 30◦ and have mean α95 -values of 7◦ , while the average α95 -value in the remaining lavas is about 5◦ . The bottom set of points shows the geometric average remanence intensities by themselves. The central set shows how the geometric average intensity of remanence would vary if each lava ow had been erupted at its VGP instead of in Iceland. These values are clearly proportional to the virtual geomagnetic dipole moment (VDM). The standard deviation of logarithm values around each point in this set corresponds approximately to a factor of 2.0 up or down. The top set shows arithmetic averages of the intensities as they would have been recorded at the VGPs. The standard error of each of these 10◦ -means is about 0.2 A/m for the three points on the right, 0.1– 0.15 A/m for the others. These results form a useful constraint on theoretical predictions and models of the long-term variations of the eld. Kristjánsson and McDougall (1982) and Kristjánsson (1995) assumed that the VDM might decrease in a linear fashion all the way as the VGP moves from the geographic poles to the Equator. This rst-order approximation indicated a reduction in the VDM by a factor of the order of 4. The results from the improved data sets point to the possibility, already hinted at in the previous publications, that the curve of inferred VDM vs. VGP latitude is fairly at above 75–80◦ VGP latitude (in part probably due to effects of non-dipole elds and various error sources) as well as below 25–30◦ VGP latitude. Consequently, the above factor is around 2.5 rather than 4. Possibly, the geomagnetic dipole eld is becoming weak in comparison with irregular non-dipole elds at around 30◦ VGP latitude. Furthermore, the VGPs in Figure 5 may consist of two populations, one from times of “ordiJÖKULL No. 58, 2008 109 L. Kristjánsson nary” secular variation of the VGP in latitudes above 30◦ and another one (comprising of order 10% of the total) of considerably weaker poles distributed at random over the globe (Harrison, 1980). This is supported by the observation of a fairly at tail of the frequency distribution of Icelandic VGPs in latitude, below 25–30◦ (Kristjánsson and Jónsson, 2007). It should be noted that a few published studies have described absolute paleointensity determinations on individual Icelandic lava samples, usually from heating experiments. These include Levi et al. (1990), Tanaka et al. (1995), and most recently Goguitchaichvili et al. (1999) yielding results from a single reversal that are qualitatively similar to our Figure 5. However, as Icelandic rocks do often alter during laboratory heating (cf. above) and some of their original remanence may also have decayed with time, especially in lavas with low MDFs (Kristjánsson, 1999, Sect. 4), further research in this eld is needed. There is still no consensus on the optimum procedures for paleointensity measurements on igneous materials, and internally inconsistent results continue to be obtained from lava units abroad where the samples selected for the intensity determinations have passed conventional quality tests (cf. Biggin et al., 2007). sions. Kristjánsson (1999; 2002, Sect. 8.2) concluded that such preferences are not evident in the distribution of low-latitude VGPs from Icelandic lavas or in the corresponding mean dipole moments. Lowand mid-latitude VGPs have now been regrouped in 36◦ - longitude intervals, using the same data set and acceptance criteria as in the section on secular variation above. The results are indicated in Table 1, where 525 lava ows with VGP latitudes between 40◦ S and 40◦ N, 419 lavas with VGP latitudes 41– 50◦ and 425 lavas with latitudes 51–56◦S or N are grouped separately. The rms value of within-ow α95 radii in the ows in the Table is 7◦ . Iceland itself lies around the middle of the last longitude interval. It is seen from Table 1 that the preference is at any rate not very strong. For poles at 50◦ latitudes or less, the highest numbers are at 36–71◦E and 216–251◦E, passing through the Middle East countries and the Western U.S. respectively. These two antipodal longitude zones lie west of those generally suggested in the literature to be preferred by transitional VGPs, i.e. through Western Australia and Eastern U.S. Paleo-secular variation: changes in its overall amplitude with time Kristjánsson and Jóhannesson (1989) studied the observed scatter of VGPs around their mean positions in the various large collections of lava ows then published from Iceland. They pointed out that this scatter (as measured for instance by the angular standard deviation of poles) appeared to have been decreasing signicantly since 15 M.y. ago. This observation was followed up by Kristjánsson (1995, 2002) The distribution of low- and mid-latitude VGPs in longitude A question which has been debated since suggested by Laj et al. (1991) concerns the possibility that the geomagnetic pole travels preferentially along certain longitude intervals in polarity transitions and excur- Table 1: Number of low- and mid-latitude geomagnetic virtual geomagnetic poles in various longitude intervals on the globe. – Dreing sýndarsegulskauta á lengdargráður. Long. deg. E 0–35 36–71 72–107 108–143 144–179 180–215 216–251 252–287 288–323 324–359 40◦ S to 40◦ N 54 69 47 39 42 52 60 50 60 52 51–56 S and N 45 45 49 44 30 38 43 59 38 34 41–50◦ S and N ◦ 39 55 110 JÖKULL No. 58, 2008 44 43 38 30 51 41 38 40 Paleomagnetic research on Icelandic lava ows and Kristjánsson et al. (2003) with augmented data sets, conrming the original conclusion. One manifestation of this change is a signicant reduction in the proportion of low-latitude VGPs (Kristjánsson and Jónsson, 2007). As emphasized in these papers, the observed variations are very unlikely to be due to artifacts such as effects of alteration or tectonics. Figure 6 is based on recomputation of the lava data set of Kristjánsson (2002) after inclusion of most of the comparable data from Iceland published subsequently and some unpublished results from Northwest Iceland (L. Kristjánsson, in preparation 2008). This is the same set as in the previous section, plus 56 lavas for which intensity information was not available; the same acceptance criteria have been applied. An age estimate between 1.0 and 15.0 M.y. was assigned to each lava, to the nearest 0.5 M.y., although it must be kept in mind that some may be uncertain by 1 M.y. or more due to the scarcity of radiometric dates already mentioned. These lavas have been split into seven groups, each covering three to ve such age slots. The angular standard deviation of VGPs in each group is then computed. Results are plotted in the top graph of Figure 6. The number of accepted lava ows in each group is between 510 and 950, with a total of 5025 compared to 4230 in Figure 4 of Kristjánsson (2002). In the bottom graph of Figure 6, all lavas with VGPs less than 40◦ have been rejected, leaving 4523 lavas. In the middle graph, only those 4392 which have a within-ow α95 less than or equal to 8◦ are included. The graphs are very similar to those already presented by Kristjánsson (2002). Although they may only be considered as semi-quantitative, they indicate that the a.s.d. of all reliably recorded VGPs in Iceland was around 34◦ 12–15 M.y. ago, around 31◦ 6–12 M.y. ago and 27◦ or less 1–5 M.y. ago. As most paleomagnetists seem prepared to accept the occurrence of long-term variations in the rate of geomagnetic reversals and in the average geomagnetic dipole moment, there is no reason to doubt that the secular variation can also undergo long-term changes. The current state of knowledge from about all these global properties is, however, still quite insufcient to allow speculations about possible relationships between them. Figure 6. The angular standard deviation (a.s.d.) of VGPs in 1–15 M.y. old Icelandic lava ows, approximated as cos−1 (R/N) for a vector sum R of N unit vectors. The age of each ow was estimated to the nearest 0.5 M.y., and then they were split into seven non-overlapping age groups. Each group covers 3, 4 or 5 of the 0.5 M.y.- slots. The groups in the top curve contain N=510 to 950 lavas. Corrections for within-ow scatter (which are of the order of 0.2◦ , similar for all the points of each graph) have not been applied. The vertical bar indicates 95% condence limits for 500 Fisher-distributed poles (Cox, 1971). – Staðalfrávik horna milli sýndar-segulskauta og meðalstaðsetningar þeirra á jörðinni, í sjö stórum hópum íslenskra hrauna. Hver hópur er áætlað að nái yr 1.5–2.5 milljón ára tímabil í aldri. AEROMAGNETIC ANOMALIES A subject closely related to the high intensity of magnetization in oceanic basalts in general, is the appearance of magnetic anomalies in their vicinity. The major effort in aeromagnetic surveying of Iceland was carried out by T. Sigurgeirsson in 1968–80, mostly at 900–1200 m altitude above sea level and 3 km spacing. Additional measurements were made by the present author in 1972–92 both over Iceland and parts of the surrounding shelf, partly in collaboration with G. Jónsson. All the maps were then converted to a digJÖKULL No. 58, 2008 111 L. Kristjánsson Figure 7. A composite total-eld magnetic map of Iceland and parts of the surrounding area, mostly acquired from aircraft at around 1000 m altitude in 1968–92. Yellow and red: positive anomalies, green and blue: negative anomalies. – Kort af niðurstöðum segulsviðsmælinga yr Íslandi, ásamt mælingum frá ugvélum og skipum yr ýmsum svæðum kringum landið. 112 JÖKULL No. 58, 2008 Paleomagnetic research on Icelandic lava ows ital format suitable for viewing larger areas, cf. Figure 7 (G. Jónsson and L. Kristjánsson, unpublished data). Sigurgeirsson’s (1970–85) prole maps in scale 1:250,000 on nine sheets are available from the author, as well as copies of reports on these surveys, other maps and reprints of papers (e.g. Jónsson et al., 1991). No further surveys have been carried out after 1993, when a detailed aeromagnetic map of the Reykjavík area was acquired (Jónsson and Kristjánsson, 2002). The information available from such magnetic maps is of many kinds but it is often of a qualitative nature only. The very distinct magnetic lineations which are seen to lie parallel to many oceanic spreading ridges, are not much in evidence here. Among interesting features that can be studied in the anomaly structure over Iceland and its surroundings are the following (in order of decreasing dimensions): • Major tectonic trends, especially at the volcanic zones • Approximate location of boundaries between large areas with a single remanence polarity dominating in each (e.g. the BrunhesMatuyama boundary at 0.8 M.y.) • Extensive topographic or bathymetric features, like the shelf edge south and southeast of Iceland • Central volcano complexes, both active and extinct • Strongly magnetized Quaternary volcanoes, probably formed under water or glacier ice • Geothermal areas, several of which were surveyed by T. Sigurgeirsson (unpublished contract reports and maps) in connection with exploration efforts. For a recent review of various aspects of the aeromagnetic anomalies over Iceland and the shelf, see Kristjánsson and Jónsson (2007). CONCLUSIONS AND DISCUSSION; FUTURE TASKS The geomagnetic reversals and excursions recorded in the Tertiary and early Quaternary lava pile in Iceland have rst and foremost served as a tool in local stratigraphic mapping. While we may expect that these and other variations of the eld will continue to be of use for correlation in such work, it has become increasingly clear that the distances between neighboring mapped proles must be reduced from 10 km or more as often happened in the past, to something like 2 km when feasible. This is because the rate of reversals is higher than previously anticipated, and the rate of eruptions has also been more variable in time and space than expected. The well-known correlation methods based largely on lava lithology and extended sediment layers will have to be expanded to include e.g. lava chemistry. Furthermore, an increased number of radiometric age determinations is needed, in particular in order to date major sediment horizons and other locations where unconformities are suspected. Accurate dating of the boundaries of thick polarity zones (McDougall et al., 1977, 1984) is also important for long-distance and possibly global correlations. As is evident from Figure 2, many parts in Iceland have still not been sampled for paleomagnetic studies. The reasons for this incomplete coverage are manifold: • In some highland areas: lack of exposures, difculty of access • In the Quaternary regions: dominating presence of unsuitable formations such as tuffs and breccias; correlation difculties due to the landscapes formed by glaciation • In the vicinity of central volcanoes and in areas where these are most numerous (e.g. Southeast and Central West Iceland): alteration, local tectonic disturbances • Elsewhere: limited nancial resources, lack of interest among geologists, etc. It is evident that just the basic mapping of paleomagnetic reversals in the Tertiary formations will take several decades yet at the current rate. Concurrent with the mapping of new areas, it would also be in some cases desirable to perform further mapping within the already sampled ones and to carry out sampling in additional proles there, in order to make correlations JÖKULL No. 58, 2008 113 L. Kristjánsson more certain. Paleomagnetic studies on basement rocks abroad have often concentrated on the acquisition of evidence about major episodes of alteration and tectonic upheavals to which these rocks have been subjected. These studies depend on knowledge of the overall properties of the geomagnetic eld, but due to the many factors inuencing the remanence directions in complex ways, they are not in a position to contribute much to that knowledge. In Iceland on the other hand, most of the formations that have been selected for paleomagnetic research have suffered only minor disturbances (< 150◦ C heating, < 8◦ tilt) and can therefore deliver much information about overall properties of the geomagnetic eld in the last 15 M.y. In this paper, examples have been presented regarding, for instance: • Mean VGP positions, for which perhaps as many as 300 lavas have to be averaged in order to bring 95% condence limits below 5◦ . • A decrease of the long-term average virtual dipole strength as a function of VGP latitude, from a value applying to VGP latitudes above 75–80◦ N or S, by a factor of about 2.5 to a fairly constant value for VGP latitudes less than 25–30◦. • A signicant decrease with time in the longterm scatter of VGPs around their long-term (1– 2 M.y.) mean positions. This decrease may have been most rapid near 12 M.y. ago and at 5 to 6 M.y. ago. Aeromagnetic surveys have since the late 1960’s been helpful in studying the tectonics and ages of various areas in Iceland and on the shelf. They are of especially great value where the bedrock is covered by water, ice caps, tuffs or detritus. Due to the large spacing between survey lines, many structures such as central volcanoes are only intersected by one or two such lines. Very little other geological/geophysical mapping of the central volcanoes has taken place thus far. In the future, detailed low-altitude surveys of volcanic centers may be expected to become an important part of integrated studies on these volcanoes. Extensive bibliographical lists of publications on paleomagnetism in Iceland, as well as on related 114 JÖKULL No. 58, 2008 stratigraphic and rock-magnetic research, are accessible at www.raunvis.hi.is/∼leo/vef−rit.html Acknowledgements Rósa Ólafsdóttir and Geirnnur Jónsson drafted Figures 2-7. ÁGRIP Mælingar á styrk og stefnu varanlegrar segulmögnunar í íslensku gosbergi voru fyrst gerðar um 1930. Á sjötta áratugnum varð ljóst, að slíkar athuganir gátu veitt jarðvísindamönnum ýmsar mjög gagnlegar upplýsingar, meðal annars vegna þess að segulmögnunarstefnan endurspeglar allnákvæmlega stefnu jarðsegulsviðsins þegar bergið myndaðist. Mælingar hafa nú verið gerðar hérlendis og erlendis á hátt í átta þúsund hraunlögum, að miklu leyti í þykkum syrpum (um og yr 300 lög) þar sem nokkur snið sem skarast í aldri, eru tengd saman. Segulstefnurnar hafa nýtst vel við slíka kortlagningu, einkum þeir umsnúningar jarðsegulsviðsins sem orðið hafa með óreglulegu millibili mörgum sinnum á hverri ármilljón. Ljóst er þó orðið nú, að til þess að tengja umsnúningana með nokkurri vissu milli sniða (með hjálp annarra atriða s.s. setlaga eða efnasamsetningar bergsins), þurfa fjarlægðir milli sniðanna yrleitt að vera minni en áður var talið duga, helst af stærðargráðu 2 km fremur en t.d. 10 km. Kortlagningu á uppruna, byggingu og aldri hraunlagastaans í landshlutum utan gosbeltanna hefur miðað hægt síðustu áratugina, og eru mörg svæði enn alveg ókönnuð að þessu leyti. Einnig hefur lítið verið gert af segulmælingum t.d. á setlögum hér, innskotum, gosbergi mynduðu í vatni, og hraunlögum ummynduðum af jarðhitavirkni. Vegna áreiðanleika og stöðugleika segulstefna í íslensku blágrýti, og hins mikla fjölda mældra hraunlaga, hafa segulmælingar á þeim jafnframt skipt verulegu máli í almennri öun þekkingar á alþjóðavettvangi um sögu jarðsegulsviðsins síðustu 15 milljón ár. Þar má nefna m.a. þróun tækni við mælingar, niðurstöður úr einstökum kortlagningarverkefnum, athuganir á venslum seguleiginleika bergsins við samsetningu steinda í því, og úrvinnslu mikils fjölda stefnu- og styrkleikamælinga Paleomagnetic research on Icelandic lava ows til að kanna langtímahegðun jarðsegulsviðsins. Fáein dæmi um slíka úrvinnslu eru sýnd og rædd í greininni. Að síðustu er þess getið, að bergsegulmælingar séu nauðsynlegar við jarðfræðilega túlkun á niðurstöðum nákvæmrar kortlagningar á frávikum jarðsegulsviðsins yr landinu. REFERENCES Ade-Hall, J. M., H. C. Palmer and T. P. Hubbard 1971. The magnetic and opaque petrological response of basalts to regional hydrothermal alteration. Geophys. J. Royal Astron. Soc. 24, 137–174. Biggin, A. J., M. Perrin and M. J. Dekkers 2007. A reliable absolute paleointensity determination obtained from a non-ideal recorder. Earth Planet. Sci. Lett. 257, 545– 563. Cox, A. 1971. Condence limits for the precision parameter k. Geophys. J. Royal Astron. Soc. 18, 545–549. Dagley, P. and R. L. Wilson 1971. Geomagnetic eld reversals - a link between strength and orientation of a dipole source. Nature 232, 16–18. Dagley, P., R. L. Wilson, J. M. Ade-Hall, G. P. L. Walker, S. E. Haggerty, T. Sigurgeirsson, N. D. Watkins, P. J. Smith, J. Edwards and R. L. Grasty 1967. Geomagnetic polarity zones for Icelandic lavas. Nature 216, 25–29. Fisher, R. A. 1953. Dispersion on a sphere. Proc. Royal Soc. A217, 295–305. Goguitchaichvili, A. T., M. Prévot and P. Camps 1999. No evidence for strong elds during the R3-N3 Icelandic geomagnetic reversal. Earth Planet. Sci. Lett. 167, 15– 34. Harrison, C. G. A. 1980. Secular variation and excursions of the Earth’s magnetic eld. J. Geophys. Res. 85, 3511–3522. Hospers, J. 1953. Reversals of the main geomagnetic eld, I-II. Proc. Royal Acad. Netherl. B56, 467–491. Jónsson, G. and L. Kristjánsson 2002. Þéttriðnar segulsviðsmælingar yr Reykjavík og nágrenni (Aeromagnetic survey over the Reykjavík area). Náttúrufræðingurinn 71, 42–49 (in Icelandic). Jónsson, G., L. Kristjánsson and M. Sverrisson 1991. Magnetic surveys of Iceland. Tectonophysics 189, 229–247. Kristjánsson, L. 1995. New palaeomagnetic results from Icelandic Neogene lavas. Geophys. J. Internat. 121, 435–443. Kristjánsson, L. 1999. On low-latitude virtual geomagnetic poles in Icelandic basalt lava sequences. Phys. Earth Planet. Inter. 115, 137–145. Kristjánsson, L. 2002. Estimating properties of the paleomagnetic eld from Icelandic lavas. Phys. Chem. Earth 27, 1205–1213. Kristjánsson, L. and H. Auðunsson 2007. Um segulstefnu í hraunlögum og óvissu í túlkun hennar (On paleomagnetic directions in lava ows and uncertainties in their interpretation). Raust 4, 17–25 (in Icelandic). Kristjánsson, L. and Á. Guðmundsson 2001. Paleomagnetic studies in Skarðsheiði, South-Western Iceland. Jökull 50, 33–48. Kristjánsson, L. and Á. Guðmundsson 2005. Stratigraphy and paleomagnetism of lava sequences in the Suðurdalur area, Fljótsdalur, Eastern Iceland. Jökull 55, 17–32. Kristjánsson, L. and H. Jóhannesson 1989. Variable dispersion of Neogene geomagnetic eld directions in Iceland. Phys. Earth Planet. Inter. 56, 124–132. Kristjánsson, L. and G. Jónsson 2007. Paleomagnetism and magnetic anomalies in Iceland. J. Geodyn. 43, 30– 54. Kristjánsson, L. and I. McDougall 1982. Some aspects of the late Tertiary geomagnetic eld in Iceland. Geophys. J. Royal Astron. Soc. 68, 273–294. Kristjánsson, L., I. B. Friðleifsson and N. D. Watkins 1980. Stratigraphy and paleomagnetism of the Esja, Eyrarfjall and Akrafjall mountains, SW-Iceland. J. Geophys. 47, 31–42. Kristjánsson, L., B. S. Harðarson and H. Auðunsson 2003. A detailed palaeomagnetic study of the oldest (c. 15 Myr) lava sequences in Northwest Iceland. Geophys. J. Internat. 155, 991–1005. Kristjánsson, L., Á. Guðmundsson and B. S. Harðarson 2004. Stratigraphy and paleomagnetism of a 2.9-km composite lava section in Eyjafjördur, Northern Iceland: a reconnaissance study. Internat. J. Earth Sci. 93, 582–595. Kristjánsson, L., Á. Guðmundsson, Á. Hjartarson and H. Hallsteinsson 2006. A paleomagnetic study of stratigraphic relations in the lava pile of Norðurárdalur and Austurdalur, Skagafjörður, North Iceland. Jökull 56, 39–55. LaFemina, P. C., T. H. Dixon, R. Malservisi, T. Árnadóttir, E. Sturkell, F. Sigmundsson, and P. Einarsson 2005. Geodetic GPS measurements in south Iceland: strain accumulation and partitioning in a prop- JÖKULL No. 58, 2008 115 L. Kristjánsson agating ridge system. J. Geophys. Res. 110, doi 10.1029/2005JB003675. Laj, C., A. Mazaud, R. Weeks, M. Fuller and E. HerreroBervera 1991. Geomagnetic reversal paths. Nature 351, 447. Lanza, R. and A. Meloni 2006. The Earth’s Magnetism: an Introduction for Geologists. Springer, New York, 278 p. Levi, S., H. Auðunsson, R. A. Duncan, L. Kristjánsson, P.Y. Gillot and S. P. Jakobsson 1990. Late Pleistocene geomagnetic excursion in Icelandic lavas: conrmation of the Laschamp excursion. Earth Planet. Sci. Lett. 96, 443–457. McDougall, I., K. Sæmundsson, H. Jóhannesson, N. D. Watkins and L. Kristjánsson 1977. Extension of the geomagnetic polarity time scale to 6.5 m.y.: K-Ar dating, geological and paleomagnetic study of a 3,500 m lava succession in Western Iceland. Bull. Geol. Soc. Am. 88, 1–15. McDougall I., L. Kristjánsson and K. Sæmundsson 1984. 116 JÖKULL No. 58, 2008 Magnetostratigraphy and geochronology of Northwest Iceland. J. Geophys. Res. 89, 7029–7060. Sigurgeirsson, T. 1957. Direction of magnetization in Icelandic basalts. Adv. Phys. 6, 240–246. Sigurgeirsson, T. 1970–85. Aeromagnetic survey of Iceland. Sheets 1–9 in scale 1:250,000. Science Institute, University of Iceland. Tanaka, H., M. Kono and S. Kaneko 1995. Paleosecular variation of direction and intensity from two PliocenePleistocene lava sections in Southwestern Iceland. J. Geomag. Geoel. 47, 89–102. Walker, G. P. L. 1959. Geology of the Reydarfjördur area, eastern Iceland. Quart. J. Geol. Soc. London 114, 367– 391. Watkins, N. D. and G. P. L. Walker 1977. Magnetostratigraphy of Eastern Iceland. Am. J. Sci. 277, 513–584. Wensink, H. 1964. Paleomagnetic stratigraphy of younger basalts and intercalated Plio-Pleistocene tillites in Iceland. Geol. Rundschau 54, 364–384.