Crustal conductivity anomalies in central Sweden and southwestern Finland

Summary

Lateral conductivity variations in the Earth's crust in central Sweden and southwestern Finland were mapped using magnetic field variations recorded simultaneously by 29 Gough-Reitzel-type magnetometers in summer 1985. Three events of 3 or 4 h were digitized and analysed to obtain single-station induction arrows and interstation horizontal field transfer functions. The 1985 magnetometer array covered part of the BEAR (Baltic Electromagnetic Array Research) array operated in 1998 on the Fennoscandian Shield. Results from eight BEAR sites are used to control and to supplement the 1985 array results. The conductivity of the crust was found to increase in the western part of the array in western central Sweden. Crustal conductors were found both in the western part of the array and in the Baltic Sea area in the eastern part of the array. 3G-D modelling based on the induction arrows revealed a major conductivity anomaly in the middle and lower crust under the archipelago of Southwest Finland.

crustal structure, electrical conductivity, Fennoscandian Shield, Finland, magnetometer array, Sweden

Introduction

Magnetometer arrays provide an efficient way to map lateral subsurface conductivity variations i.e. to delineate large-scale crustal conductors and identify crustal units having different conductivity properties. Knowledge of crustal conductivity structure (Pajunpää 1987; Korja & Hjelt 1993), in turn, has proved to be important in the tectono-geological modelling of the structure and evolution of the Fennoscandian Shield (Korja 1993; Korsman 1999). In the early 1980s, a set of 31 Gough-Reitzel-type magnetometers from the Münster University in Germany (Küppers & Post 1981) were deployed in Finland in several array measurements. The results exposed the major crustal conductivity structure in Finland (,,Pajunpää 1984, 1986, 1987) and helped in planning intense magnetotelluric (MT) soundings during the same decade (see e.g. Korja & Hjelt 1993; Korja & Koivukoski 1994). In 1984 the magnetometers were moved to Sweden and until 1986 the Geological Survey of Sweden operated six arrays in northern and central Sweden. Some preliminary results have been presented from arrays in northern Sweden (e.g. Korja & Hjelt 1993) but none from central Sweden. In this paper, we aim to present results from the magnetometer array in the latter region.

The array of 29 magnetometers in central Sweden recorded magnetic field variations for two months in summer 1985. The magnetometers were installed on three east-west-directed lines (C, D and E, see Figs 1 and 2) across Sweden covering an area of about 100 × 400 km². The northernmost line C extended with three magnetometer sites (C11-C13) to islands in Finland.

Figure 1

Map of northern Europe showing the location of the 1985 magnetometer array. BEAR MT-sites are marked with open circles and the EISCAT magnetometer cross in the north with black dots. Approximate boundaries of the major geological provinces are drawn after Gorbatschev & Bogdanova (1993). T.I.B denotes Transscandinavian Igneous Belt.

Figure 2

Sites (black dots) and station codes (C, D and E) of the 1985 magnetometer array together with BEAR sites (open circles; A- and B-sites) used in this study. The MT-sites of Rasmussen (1988) are shown with triangles and those of the FENNOLORA MT-line (Rasmussen 1987) with diamonds. The inverted triangles show the eight MT sites used by Rasmussen (1988) for 1-D and 2-D modelling.

The recent BEAR project (Baltic Electromagnetic Array Research; Korja & the BEAR Working Group 2000) has demonstrated the importance of the knowledge of subsurface conductivity structure obtained from spatially dense observations such as older magnetometer arrays in Finland. The BEAR project deployed 46 magnetotelluric instruments in addition to 20 permanent magnetic stations that recorded simultaneously for about one and a half months in summer 1998. The BEAR array covers the entire shield and, owing to the large station spacing of about 150 km, is not optimally powerful for locating lateral conductivity structures in the crust. Its main objective is to investigate structures deep in the crust and upper mantle. The older magnetometer arrays in Fennoscandia have much denser spatial sampling (20–50 km) than the BEAR array. The older arrays can therefore give information on lateral crustal conductivity variations for the modelling of the MT deep sounding data. Data from the denser arrays can also be used to fill the BEAR observational gaps in inversions.

Geological setting and previous geoelectric studies in the research area

The research area is located in the southwestern part of the Fennoscandian (Baltic) Shield (Fig. 1). The Shield is bordered in the west by the Palaeozoic Caledonian Orogen (Caledonides) and in the east and south by the East European Platform. The basement beneath the Phanerozoic sedimentary cover, however, is Precambrian, and, together with the Fennoscandian Shield, comprises the East European Craton (Gorbatschev & Bogdanova 1993).

The 1985 magnetometer array extended from the Svecofennian Domain in the east to the Sveconorwegian Domain in the west. The crust in Finland and central Sweden (eastern and central part of the array) was formed primarily 1.90-1.88 Ga ago in a closure of a large oceanic basin and accretion of a pre-Svecofennian (>1.91 Ga old) island arc complex. This was followed by the intrusion of late and post-orogenic granitoids 1.85-1.80 Ga ago. At the western margin of the Svecofennian Domain (central part of the array), the granites and porphyries of the Transscandinavian Igneous Belt (TIB) were emplaced in several pulses between 1.85 and 1.65 Ga ago in an east-west extensional regime, probably as a result of eastward-directed subduction beneath a mature continental margin (Nironen 1997). The last major crustal reworking event in the Svecofennian Domain is related to the intrusion of rapakivi granites between 1.67 and 1.51 Ga. The extensional events resulted in locally considerable crustal thinning and the intrusion of several rapakivi batholiths in the easternmost part of the 1985 array beneath Åland and southwestern Finland (Haapala & Rämö 1992).

In the west, the 1985 array extends to the Sveconorwegian Domain that was formed in the Gothian orogeny 1.72-1.63 Ga ago in several phases of accretion and igneous activity. The crust was later reworked in the Sveconorwegian orogeny 1.1-0.9 Ga ago (Gorbatschev & Bogdanova 1993).

Previous knowledge on crustal conductivity in the research area stems from three studies, namely magnetotelluric studies in Sweden by Rasmussen (1987) and Rasmussen (1988) and a magnetometer array study in southwestern Finland (Pajunpää 1987). In the western part of the research area, Rasmussen (1988) presented MT results from 15 sites along an east-west-directed line coinciding with the D-line in our 1985 magnetometer array. Using data from the eight westernmost sites, Rasmussen attempted both 1-D and 2-D regional modelling for an area covering the easternmost part of the Sveconorwegian Domain and TIB granites (Fig. 2). The 2-D model yielded 10 000-20 000 Ω m with the regional strike of N10°E for crustal resistivity at the western part of the MT profile area, whereas the 1-D model gave an anisotropic layer in the middle and lower crust with 400 and 17 000 Ω m in the transverse directions, the most conducting direction being N75°W. Rasmussen (1987) performed MT measurements at 30 sites along a north-south-oriented line across the whole Sweden. Five of their sites were in or close to the area of the 1985 magnetometer array in the central part of the research area in eastern Sweden (Fig. 2). They completed 1-D inversion of the data, but concluded that the complexity of the transfer functions and the large spacing between the stations inhibits a quantitative interpretation of the conductivity structure in this area. In the eastern part of the research array in southwestern Finland, Pajunpää (1987) operated a magnetometer array in summer 1984. Data from this array revealed a SE-NW-directed conductor close to the southwestern coast of Finland but its width and conductance could not be determined.

Measurements and data

The 29 magnetometers were installed in May 1985 in three east-west-directed lines (C, D and E) north of Stockholm and lake Vänern, running from the border between Norway and Sweden in the west to the western coast of the Baltic Sea in the east (Fig. 2). The northernmost C-line continued across the Baltic Sea with two sites on Åland Island and one magnetometer in the archipelago close to the mainland Finland. The distance between the three lines is about 50 km and the magnetometer spacing on the lines varies from 30 and 50 km.

The analogue data on the films were digitized manually using a digitizing table. The main criteria in choosing the events was to avoid disturbed times with southward-expanding ionospheric currents and to get enough power in the variations to go above the estimated 2 nT digitization noise. The following three events were chosen: 

The research area is about 1000 km south of the auroral oval and therefore the source field problem is not as serious as in northern Fennoscandia, but does require some consideration. The EISCAT magnetometer cross (Lühr 1984) operated simultaneously with the 1985 array in the northern Fennoscandia and data from EISCAT cross were used to find the source field behaviour of the chosen events. The variations at the EISCAT cross were of the order of 100 nT, while at the magnetometer array, 1000 km southward, they were about 20 nT. As the largest field variations could be located at or north of the EISCAT cross, this showed that the field variations were slowly decaying southward from the auroral zone. At the 1985 magnetometer array the variations of the horizontal components were almost constant across the array and the vertical component (excluding anomalous sites) was only about 10 per cent of the horizontal components. Furthermore, the correlation between this ‘normal’ vertical field and the horizontal fields was usually low. Therefore, the conditions for single station induction arrows were valid.

The BEAR array had five MT sites plus three magnetometer stations in the area of or close to the 1985 magnetometer array (Fig. 2). Results of the eight sites are used here to supplement the array and also to control the validity of the induction arrows from the 1985 array data. As the BEAR sites recorded data digitally with a 2 s sampling interval for several weeks, the data could be processed with modern robust methods. The processing showed that magnetotelluric and geomagnetic transfer functions are free of source field bias up to the longest period used in this study, 2000 s. The BEAR data yield statistically more stable induction arrows, which could be compared with the results deduced from the 1985 magnetometer array.

Induction arrows and horizontal magnetic transfer functions

The small amount of data, altogether 10 h, limits the possibilities for statistical analysis of the data. Therefore, the correctness of the induction arrows depends mostly on the choice of proper events. The equation  

(1)

defines the complex transfer functions A(ω) and B(ω) at the frequency ω between the horizontal magnetic field components X(ω) and Y(ω) and the vertical component Z(ω). X is positive to the north, Y to the east and Z downwards. The real induction arrows are formed from the real parts of A and B by changing their signs. In this way the arrows are ‘reversed’ and they point towards the induced subsurface currents.

We have used mean values of three methods to determine the induction arrows. First, we calculated the standard least-squares solution (Everett & Hyndman 1967; Schmucker 1970) to get the transfer functions at five periods. Thereafter we used the robust method of Egbert & Booker (1986) to get both single station and remote reference transfer functions.

The three methods did not give any systematic differences in the results, although at short periods (<200 s) arrows at single sites have noticeable differences. Owing to the small number of data points the 95 per cent confidence limits were often larger than the arrows and we will not present them. The transfer functions of the eight BEAR sites have their 95 per cent confidence intervals typically below 0.01, revealing their high statistical accuracy compared with the 1985 array results.

The real reversed induction arrows are shown in Fig. 3 together with arrows of the eight BEAR sites and with six arrows of the 1984 magnetometer array in southwestern Finland (Pajunpää 1987). Data from the 1984 array are not available for all the periods used in Fig. 3. By comparing the BEAR sites with the neighbouring sites of the 1985 array, we can see that the agreement both in direction and magnitude is fairly good. This confirms that the arrows of the 1985 array are reliable for interpretation.

Figure 3

Reversed real induction arrows at five periods (a-e). Geological boundaries are as in Fig. 1. Reversed induction arrows point towards subsurface current concentrations. Estimated locations of local conductors are shown in (a).

Following the notation of Beamish (1982) and Fujiwara & Toh (1996) we can write the relation between a field site and a reference site. First, we assume that the horizontal field at the reference site ‘r’ is ‘normal’ in the sense that it is composed of an external source field and of a field arising from the currents induced by the source field in a horizontally layered normal earth. The field at the field site ‘s’ is assumed to have an equal normal field and an anomalous part that arises only from lateral discontinuities in the geoelectric earth structure:  

(2)

 

(3)

The interstation transfer functions C, D, E and F describe the anomalous part of the horizontal field at the field site. Their values were estimated by using the LS method. Real parts of the interstation horizontal field transfer functions were used to plot ellipses describing the anomalous horizontal field at the sites. We can replace the reference field with a unit circle (Fujiwara & Toh 1996),  

(4)

where the angle is positive from the north to the east. The anomalous fields are then 

(5)

 

(6)

The reference field was chosen from the site where the horizontal field is small compared with the horizontal fields at anomalous sites. However, some of the small ellipses may actually be ‘negative’ if the horizontal field at that site is smaller than the reference field. The ellipses describe the anomalous horizontal field that is normalized by the inducing source field. The short axis of an ellipse shows the main current direction under the site and the long axis is proportional to the magnitude of the anomalous current.

We used averages of the time-series from sites C5-C7 as the reference field to calculate the interstation horizontal transfer functions for the 1985 array. These three sites have small induction arrows, their magnetograms reveal no anomalous induced currents and they are located favourably almost in the middle of the 1985 array. The largest distance between the reference sites and the field sites was about 300 km, which is much less than the estimated spatial wavelengths of the time variations of the events. The correlation of the horizontal components of the reference field and the field of each site was more than 0.9 at most sites and periods. At the shortest period 100 s and occasionally at 215 s the correlation was between 0.7 and 0.9. From the eight BEAR sites the site B07 was chosen as the reference. The real parts of the anomalous horizontal fields are presented as ellipses in Fig. 4. The radius of the circle with the period label corresponds to a value of 0.5 of the anomalous field with respect to the reference field. Note that the site B07 has no ellipse, as it was a reference site for BEAR data. The average of the sites C5-C7 were used as the reference for the 1985 array and therefore small ellipses are found for each of the reference sites.

Figure 4

Anomalous horizontal field ellipses expressing the anomalous horizontal field normalized by the inducing horizontal field. The average of the sites C5, C6 and C7 was used as the reference for the 1985 array and the site B07 for the BEAR sites. The circles with the periods have a radius of 0.5. Geological boundaries are as in Fig. 1. Estimated locations of local conductors are shown in (a).

Conductivity structures

The induction arrows (Fig. 3) and the anomalous horizontal field ellipses (Fig. 4) reveal many large- and small-scale features within the magnetometer array. A dominant large-scale feature nearly at all periods is the smooth rotation of induction arrows from dominantly SW orientation in the western part of the array to S orientation in the central part and to SE orientation in the eastern part of the array. This phenomenon is associated with a slight increase of the arrow magnitude towards sea areas, i.e. to SW in the western part of the array and to SE in the eastern part of the array (Fig. 3). Results from recent multisheet modelling (Engels 2002) of the Fennoscandian S-map model including realistic conductances for sea water and sea bottom sediment layers and having a fairly homogeneous crustal conductivity structure in southern Sweden (Korja 2002) can be used to investigate the effects of the surrounding sea areas (coast effect) in southern Sweden. Multisheet modelling shows that the conducting sea areas that surround southern Sweden produce a smooth and regular pattern of induction arrows that resembles the pattern of arrows in the 1985 array: arrows in the west point to SW, in the centre to S and in the east to SE, and the magnitude is increasing towards sea areas. There are, however, many important differences both in magnitude and direction of arrows between the model (S-map) and observations, which require the presence of subsurface conductors in the study area. These differences will be described and discussed below separately for each subregion of the research area.

Based on the behaviour and properties of transfer functions, the study area can be divided into four units: western part, central part, eastern coast of Sweden and Åland, and islands and southwestern Finland. In the following, we will interpret and discuss the calculated arrows and ellipses. Locations and orientations of crustal conductors, estimated from the observations of this study, are shown in Figs 3(a), 4(a) and 5. Since the new data together with the data from the older array in SW Finland (Pajunpää 1987) suggest the presence of a major crustal conductor between Åland and the western coast of Finland, a 3-D model is presented (Fig. 5) for the easternmost unit.

Figure 5

3-D model for the Åland Sea area. Upper panels: the plan view of the central part of the model (total model area is 800 × 800 km²) as integrated conductance from surface to the depth of 60 km. Black dashed lines outline rapakivi intrusions in Åland archipelago and southwestern Finland. Lower panel: model resistivity cross-section along an EW-directed profile AB (white dashed line in upper panel). Åland RK =Åland rapakivi intrusion. VE = vertical exaggeration.

Western part

Arrows in the western part of the 1985 array in the Sveconorwegian Domain point to the southwest or west at all the periods exhibiting an anticlockwise rotation from predominantly westerly orientation to southwesterly orientation with increasing periods. At the longest period of 2150 s, however, the arrows rotate back to a more westerly orientation. At long periods the magnitude of observed arrows increases towards WSW. The ellipses at D1 are large and fairly circular at all the periods. The ellipses at D2 are similar but smaller. Similarly, in the Transscandinavian Igneous Belt, induction arrows at longest periods (1000–2150 s) point towards the west or southwest. At shorter periods, however, the behaviour of the azimuth and magnitude of arrows is more irregular, suggesting more heterogeneous small-scale conductivity structure at shallower depths. The anomalous horizontal fields at D2 and D1 indicate anomalous induced currents under the sites (Fig. 4a). Also the arrows at 100 and 215 s periods indicate that there are NNW-striking conductors in the crust north of and under lake Vänern (Fig. 3a). At long periods the arrows reveal high conductivity further to the WSW and probably at greater depths.

The cause for the enhanced conductivity to the southwest of the research area is not clear, but obvious candidates are the conducting oceanic water and sea bottom sediments. The total conductance of the seawater and sediments in the North Sea and Oslo Graben ranges from c. 2000 to 4000 S. All of these features are included in the Fennoscandian S-map model used for multisheet modelling (Engels 2002). A comparison of model arrows (not illustrated) with the observed arrows show, however, that all observed large-scale features cannot be explained by the surrounding seas. First, the direction of arrows differ c. 20°-40°—observed arrows point to S45°W-S65°W, whereas model arrows point at 512 s to S10°W-S20°W. Secondly, the magnitudes of the observed and the model arrows in the western part of the array are c. 0.3 and 0.15, respectively. These differences suggest that the western part of the array is more conducting than indicated in the S-map model (6 S for 60 km, i.e. 10 000 Ω m on average) and that there is conducting material in Norway to the west of the research area, which is not incorporated into the S-map model. One possibility could be the presence of conducting alum shales beneath the Caledonides. There are no geoelectric data from the Caledonides in the research area and its vicinity, but magnetotelluric data from the Jämtland region to the north of the research show that the Caledonian metasedimentary lithologies and, in particular, alum shales are highly conducting metasedimentary rocks (Gharibi 2000). The partial contribution of alum shales and other conducting lithologies in Caldonides is further corroborated by long induction arrows at site A23 close to the Caledonian Front near Oslo (Fig. 2) not predicted by the S-map model.

Rasmussen (1988) presented a regional 2-D and an anisotropic 1-D model based on magnetotelluric measurements at eight sites in the western part of his MT-line (roughly between sites D2 and B02; Fig. 2). Both models fitted the long-period data (>10 s) reasonably well and suggested that conductivity in the lowermost crust and upper mantle is isotropic and resistive (>6000 Ω m). Strike analysis of magnetotelluric data suggested N10°E as the dominant strike at long periods. A 1-D anisotropic model, on the other hand, was used to explain the phase split at shorter periods (0.1-10 s), i.e. at upper and middle crustal depths. The conductivity increase towards WSW from our study supports the model of isotropic lower crust and upper mantle. Yet it should be noted that the strike of N10°E obtained from magnetotelluric data at longer periods (>10 s) differs from the strike estimates obtained from induction arrows. Reversed induction arrows point towards current concentrations in the Earth and therefore the strike would rotate anticlockwise roughly from NS at the period of 100 s to NW at 2150 s.

Our data do not allow us to distinguish between isotropy and anisotropy at upper and middle crustal depths owing to the limited period range. However, the behaviour of induction arrows at the shortest periods (100 and 215 s) in the TIB region, in particular, suggests the presence of a conductor to the south of the MT profile. Similarly, the horizontal magnetic field transfer functions (Fig. 4) suggest the presence of considerable lateral conductivity variations in the research area. These conductors together may partly contribute to the observed phase split at short-period magnetotelluric data. Yet it is clear that to conclusively distinguish between isotropic and anisotropic conductivity in central Sweden requires a separate detailed 3-D modelling of all observed data.

Central part

The arrows are shortest in the central part of the array in the Svecofennian Domain. At 100 s the arrows point mainly to SE but rotate clockwise to S or SW with increasing periods. The ellipses are small except at 100 s. There are no large-scale conductivity anomalies in this area but the behaviour of the transfer functions at 100 s indicates that there are local conductive structures in the crust. Rasmussen (1988) wrote that towards the eastern part of his MT-profile (Fig. 2), the response functions display considerably more variation from station to station, and apparent resistivity values are generally lower in the east than in the west. In contrast, based on the magnitude of the arrows and the ellipses we can say that the crust in this central part is more resistive than in the western part of the 1985 array. Modelling of magnetotelluric data (Rasmussen 1988) suggested that resistivity of the crust beneath the Sveconorwegian Domain and TIB is around 10 000 Ω m. Consequently, the magnetometer array data shows that the resistive region beneath TIB extends further to the east into the Svecofennian Domain and becomes even more resistive (>10 000 Ω m). One possible interpretation of this feature is that the resistive rocks of the TIB are dipping eastwards beneath the Svecofennian rocks.

Apart from local variations as described above, the induction arrows in this region point southwards at longer periods but turn to SW at the longest period. This behaviour is compatible with multisheet modelling (Engels 2002). The rotation of induction arrows to SW at longest periods (2150 s in our study, 4096 s in multisheet modelling) is understandable because the Baltic Sea in the west is much shallower and less conducting than the North Sea in the west. Yet, as in the western part of the array, the observed arrows are twice as long as model arrows, indicating the presence of more conducting crust to the south of the 1985 array.

Eastern coast of Sweden and Åland

From C9, D8 and E6 to the coast of the Baltic Sea the arrows point to ESE. Their length has a maximum at a few hundred seconds and a clear reduction in their magnitude occurs after 1000 s. The ellipses remain small except at D10 where they have almost linear polarization in the direction of the arrows. The latter suggest the presence of a SW-NE-directed local conductor under the coast and roughly parallel to it (Figs 3a and 4a). We call this conductor the ‘Stockholm Conductor’. Another local conductor can be identified beneath Åland. The arrows in Åland point to the east, except for 100 s where they point to NE, and have almost a constant direction and magnitude at all the periods. The arrows indicate anomalous currents east of Åland as will be shown later by modelling. The ellipses at sites C11 and C12 in Åland are almost linearly polarized at 215 s and the largest ellipses are at the shortest periods, revealing approximately N20°E-directed currents under the island. At 2150 s the ellipses almost disappear. At the BEAR site B12 the ellipses have the same main features although different reference sites cause differences in the size of ellipses. The behaviour of the ellipses is very similar at site D10 on the coast of Sweden and on Åland but it is unclear whether they are caused by the same conductivity anomaly. Therefore, we call the conductor under the island the ‘Åland Conductor’. A graphite-bearing black schist in the upper part of the crust is most probably the reason for both the Stockholm and the Åland anomalies. Note also that the arrows at C10 indicate a similar conductivity structure NE of the site C10. Seawater and sediments do not explain the anomalous features, which will be shown by modelling in the next section (see Fig. 6a to compare observed data at site C10 at 100 s and the corresponding model data).

Figure 6

Reversed real induction arrows. The black arrows are from Fig. 3. The arrows with open heads are calculated from the models. (a) and (b) S-map model from Korja (2002), (c) and (d) the conductor is added to the S-map model of (a) and (b). The dark area shows the 10 Ω m conductor at a depth of 10–30 km and the black lines the boundaries of the conductor at a depth of 5–10 km.

Islands and southwestern Finland

The easternmost sites C13 and B18 have a very anomalous behaviour. The ellipses are large and fairly circular at short periods and become linear at 2150 s. The maximum is at 1000 s. The arrows on the mainland Finland point mostly to SW and almost opposite to the direction of the arrows on Åland. The ellipses as well as the arrows reveal an induction anomaly between Åland and the southwestern coast of Finland. We will call it the ‘Southwestern Finland Conductivity Anomaly’. The arrows at C13 and at the 1984 array site just north of C13 have directions perpendicular to the mean direction and opposite to each other (see Fig. 3b). A local conductor at a shallow depth and perpendicular to the main strike strongly disturbs the arrows at the two sites, causing their deviation from the regional direction.

A 3-D conductivity model was constructed to study the structure of the Southwestern Finland Conductivity Anomaly (Fig. 5). The 3-D forward algorithm of Mackie (1994) was used to calculate induction arrows from the model. The model consists of 80 × 80× 30 cells where horizontal dimensions of the cells were 10 × 10 km², i.e. the model covers an area of 800 × 800 km². To incorporate effects of seawater, sediments and known crustal conductors, the crustal conductivity model (S-map) of Korja (2002) was applied as a background model. Note that only the central part of the model is shown in Fig. 5.

Fig. 6 shows the induction arrows both from the measurements and from the initial model. Figs 6(a) and (b) clearly show that the background model does not explain the anomalous behaviour of the arrows. Therefore, we added a NE-dipping conductor describing the SW Finland Anomaly into the original S-map model (Fig. 5). The conductor has a resistivity of 10 Ω m. Its upper boundary is located at a depth of 5 km and the lower boundary at 30 km, respectively. The conductance of the conductor is 2500 S, whereas an integrated conductance (0-30 km) of the background model varies from 150 to 600 S in the study area. The conductor is NW-SE-directed but it turns northwards between Åland and southwestern coast of Finland. The width of the conductor in the model is 80 km at the depths of 5-10 km, whereas it is narrower (60 km) at depths of 10-30 km.

Besides the inclusion of a new crustal conductor (Southwestern Finland Conductor) into the 3-D model, two other modifications were made in the original S-map model. First, the conductivity of the crust in the S-map model was increased to the northeast of the Southwestern Finland Conductor. Korja & Koivukoski (1994) modelled a low-resistive middle crust extending c. 50 km to the south of the Southern Finland Conductor (Pajunpää 1987) (Fig. 1). Our 3-D modelling suggests that this low-resistive layer (conductance 120 S) continues as far to the southwest as the Southwestern Finland Conductor. Secondly, the total conductance of the Southern Finland Conductor had to be reduced to c. 2000 S from the original value of c. 20 000 S in the background model (NE corner in Fig. 5).

The model responses of the final 3-D model (Fig. 5) were calculated at periods of 100 and 1000 s and are shown in Figs 6(c) and (d). The comparison of observed and modelled induction arrows, in particular at a period of 1000 s, clearly shows that the inclusion of the Southwestern Finland Conductor improved the reproduction of induction arrows. It should be emphasized that we did not attempt any detailed modelling but tried to fit the main features in the observed data. This is evident at the shortest period (100 s) where many local features are not reproduced by the large-scale model (e.g. sites C10 and C13).

Geologically the eastern part of the array is characterized by a number of rapakivi granite batholiths. The largest of them, c. 70 km in diameter, is located beneath Åland. Other smaller batholiths can be found in the western coast of Finland. According to Korsman (1999), the thickness of the Åland rapakivi batholith is about 5-10 km and the batholith is underlain by a 10 km thick layer of gabro-anortosites related to the intrusion of rapakivis. These rocks represent melting of the lower crust caused by mantle upwelling. Although the actual depth of the conductor is difficult to resolve from magnetometer data, it is probable that the Southwestern Finland Conductor is beneath the resistive rapakivi granites within a sequence composed of pelitic sediments and anatectic granites (Korsman 1999) and hence a plausible cause for enhanced conductivity is the presence of graphite- and sulphide-bearing lithologies. The conductivity may have further been enhanced in shearing as the conductor coincide with the recently identified shear zone (‘Estonia line’) striking from Estonia to NW between Åland and western coast of Finland (Nironen 2000).

Conclusions

Data from the 1985 magnetometer array in central Sweden and southwestern Finland were analysed to obtain information concerning lateral subsurface conductivity variations. Induction arrows and anomalous horizontal field ellipses were calculated from the digitized 10 h data at 27 magnetometer sites. The 1985 array was supplemented with eight BEAR sites operated in 1998 in the same region and with six sites of the 1984 magnetometer array operated in SW Finland. A 3-D modelling was carried out to study the major conductor in SW Finland.

Seawater and sediments contribute to the observed induction arrows but do not explain the main parts of the arrows and ellipses. The westernmost part of the Svecofennian Domain in the central part of the research area is the most resistive region. There are no large-scale conductors but probably many local ones in the crust. The crust in the Sveconorwegian Domain is more conducting. There are NNW-striking conductors in the crust and probably a good conducting layer in the crust extending to the Oslo region in Norway. On the eastern coast of Sweden and in Åland Island there are conductors probably in the upper crust. The NE striking Stockholm Conductor runs along the coast. The Åland conductor is closer to NS direction.

The Southwestern Finland Anomaly having SE-NW orientation along the southwestern coast of Finland and under the islands is one of the major anomalies in the Fennoscandian Shield. It is dipping northeastward and extends from the middle to the lower crust.

Acknowledgments

The BEAR is a subproject of EUROPROBE/SVEKALAPKO studies. Data processing and interpretation of the BEAR project is supported by INTAS (97-1162). The work of IL and TK has been funded through the contracts 39222 and 73249 from the Academy of Finland. The EISCAT magnetometer cross was a common German-Finnish project headed by the Technical University of Braunschweig. We thank Lasse Häkkinen, Pertti Kaikkonen, David Milling and Ari Viljanen for useful comments and advice. We thank Ivan Varentsov for sending BEAR results for this study.