Geophysical modelling of the Bjerkreim–Lobe, southern Norway

SUMMARY

The Bjerkreim–Sokndal (BKS) intrusion in southern Norway has been studied for decades due to the presence of magnetic remanence creating anomalies 12 000 nT below background as measured by airborne magnetic surveys. The strong magnetic remanence also makes the BKS intrusion a good Earth analogue for remote studies of planets that have prominent magnetic signatures, such as Martian geological environments. Although numerous geophysical surveys and samples have been collected in the area, there are limited 3-D geological interpretations of the subsurface. Here, we used existing geophysical data to conduct forward and inversion modelling of the Bjerkreim lobe to investigate the subsurface geometry of the BKS intrusion. An extensive petrophysical property compilation was used as input data for the models, in combination with airborne magnetics and digital elevation models. This petrophysical compilation was initially analysed using principal component analysis to understand which variables would have the greatest impact on the models. Forward and inversion modelling show that cross-cutting jotunite bodies, and small anorthosite blocks within the Bjerkreim lobe have a limited depth extent of 1 km. Massive and foliated anorthosites to the west of the Bjerkreim lobe extend to depths greater than 4 km indicating that the BKS intruded into these anorthosites. Complications in magnetic field fitting during the forward modelling of megacyclic units with strong magnetic remanence and the results from a new ground magnetic survey support the need to revisit mapped contacts of the cyclical units.

1 INTRODUCTION

The Proterozoic Rogaland Anorthosite Province hosts the Bjerkreim–Sokndal (BKS) layered intrusion. The BKS has been the centre of research for several decades due to the presence of strong magnetic remanence and complex geological structure (McEnroe et al. 2009; Brown et al. 2011; Biedermann et al. 2016, 2017; Biedermann & McEnroe 2017; Pastore et al. 2022). This area has been used as an Earth analogue for planetary studies (McEnroe et al. 2004a, b, c) and has been mapped for critical mineral potential due to numerous mineralized units (Schiellerup et al. 2003; Ihlen et al. 2014). The ilmenite Tellnes deposit located in the southern portion of the BKS (Korneliussen et al. 1999). There is limited understanding on the BKS subsurface despite extensive collection of data and related studies. Although public drill cores exist, these are limited to tens of metres below surface. The estimated overall thickness of the BKS is in the order of kilometres, with a minimum base thickness estimated at 4 km based on previous geophysical modelling (Smithson & Ramberg 1979). In this study, construction of subsurface geometry of the Bjerkreim lobe in the northern region of the BKS is conducted using existing geophysical data to characterize magnetically distinct units and complete forward and inverted geological modelling. The presented forward and inversion models indicate complex structures and physical property variations within MCUs, especially in the eastern margin, and warrants follow up with higher-resolution surveys.

2 GEOLOGICAL BACKGROUND

The BKS norite-quartz mangerite layered intrusion has an estimated areal extent of 230 km² (Fig. 1). The intrusion is folded in a double-plunging syncline composed of the Bjerkreim and Sokndal lobes (Paludan et al. 1994; Wilson et al. 1996; Robins & Wilson 2001; Bolle & Duchesne 2007). The Bjerkreim lobe is located at the north end and plunges southeast, while the Sokndal lobe is located at the south end and plunges north. Five Megacyclic Units (MCU IA, IB, II, III and IV) were produced due to fractional crystallization punctuated by the influx and mixing of primitive magmas (Wilson et al. 1996). Wilson et al. (1996) build on the work of Michot (1961) to subdivide the MCUs into zones (a–f) based on the presence or absence of the following index minerals: plagioclase, orthopyroxene, clinopyroxene, olivine, ilmenite, magnetite and apatite. The primitive layers of the BKS are dominated by plagioclase cumulates, the middle layers by plagioclase, orthopyroxene and hemo-ilmenite cumulates, and the evolved upper layers by plagioclase, orthopyroxene, pigeonite-augite-ilmenite, magnetite and apatite. Previous work on the oxides has shown that the more primitive magmas contain hemo-ilmenite (ilmenite with hematite exsolution) and only rare occurrences of magnetite, whereas the more evolved magmas contain nearly pure ilmenite, and magnetite with oxidation-exsolution lamellae of ilmenite (Duchesne 1970, 1972; McEnroe et al. 2000, 2000, 2001; Robinson et al. 2001, 2016). Stratigraphically above these units is the Transition Zone (TZ), which represents the most reduced magma in the BKS. The TZ contains abundant magnetite with ulvöspinel exsolution lamellae (100) creating a distinct dense cloth like microtexture. The ulvöspinel lamellae were later oxidized to ilmenite. McEnroe et al. (2009) proposed that MCU IVe should be further subdivided to include an additional subunit, IVe’. This subdivision is discussed later in the modelling section.

The Bjerkreim–Sokndal intruded into massive anorthosites of the Rogaland Anorthosite Province (Slagstad et al. 2022). The Egersund-Ogna and the Håland-Helleren anorthosites are located along the western margin of the Bjerkreim lobe. The country rock is composed of granulite-facies gneissic metamorphic rocks which are exposed along the northern and eastern contacts except for a small curtain, the Lakssvelefjeld migmatites, located along the western margin separating Egersund-Ogna and Håland-Helleren anorthosites (Marker 2013).

The BKS has significant economic resources, especially in the critical minerals ilmenite and vanadium-rich magnetite (Schiellerup et al. 2001). Three apatite-bearing sequences are found in the Bjerkreim lobe in MCU IBe, MCU IIIe and MCU IVe. Along the eastern perimeter of the Bjerkreim lobe, outcrops exhibit steeply dipping structures compared to the northern perimeter where the units have shallower dips. Exploration has been conducted throughout the Bjerkreim lobe through both cuts and drilling, with a focus in the Teksevatnet area on the eastern side of the lobe and new drillcores (>1 km) located in the western side of the BK lobe. The surrounding anorthosites host numerous historic ilmenite and magnetite deposits, including the Tellnes deposit (Korneliussen et al. 2000) which is one of the largest active ilmenite deposits in the world.

3 GEOPHYSICAL DATA SETS

Several airborne fixed-wing and helicopter magnetic surveys were flown over the Rogaland Province. Fugro collected an airborne survey in 2010 (Fugro 2010). The aeromagnetic survey was flown with a flight spacing of 250 m at an altitude of 60 m above ground. A helicopter survey compilation through Norges geologiske undersøkelse (NGU) is comprised of two surveys. First, a 1995 survey with flight line spacing of 100 and 30 m sensor altitude with the SCINTREX MP-3; and secondly, a 2014 survey with flight line spacing of 200 and 25 m sensor altitude with the SCINTREX CS-3. All aeromagnetic surveys were merged using Geosoft's Grid-knitting with microlevelling and a 5 × 5 symmetrical convolution filter to remove short-wavelength noise. Full discussion on aeromagnetic compilation can be found in Rønning (1995), Nasuti et al. (2015) and Olesen et al. (2015). The final aeromagnetic grid is shown in Fig. 2(a). The Mega Cyclic Units of the BKS are apparent as repetitive magnetic highs and lows. The most prominent magnetic anomaly is associated with MCU IVe, in the southeast region of the study area, near Heskestadvatnet (Fig. 2a, white box). This magnetic low (below background) is due to high magnetic remanence in the norite samples with steep negative inclinations and results in magnetic values up to 12 000 nT below background, as measured by airborne magnetic surveys (McEnroe et al. 2004a).

Gravity data (Fig. 2b) is available at variable measurement stations with spacing between metres and kilometres (Smithson & Ramberg 1979). Smithson & Ramberg (1979) estimate an initial base depth of 4 km based on block modelling of a 10–30 mGal gravity anomaly. Sparse sampling of gravity stations limits spatial resolution of density anomalies, however the Bjerkreim lobe exhibits higher density relative to the proximal anorthosites and country rock. A single seismic profile (Fig. 2b) was collected running west–east (Deemer & Hurich 1997) and identified the subsurface synformal structure to a depth of 4.5 km. Radiometric measurements were collected coincidently (Fig. 2c). Radiometric data provides surficial information, typically on the top 30 cm. The mangerite and quartz mangerite bodies are associated with high levels of potassium due to the abundance of k-feldspars. High levels of thorium are associated with MCU III and IV along the eastern edge of the Bjerkreim lobe, which is synonymous with the variation in magnetization between the western and eastern edges. Elevated levels of thorium are also present along faults and water bodies, which may be due to sediment accumulation along these topographic depressions (Fig. 2d). The country rocks to the east and the mangerite bodies are associated with topographic highs. Gabbronorites of the layered intrusion are more eroded, resulting in topographic lows and coinciding with local agricultural land usage.

Additional details in the magnetic data are shown in Figs 2(e)–(h) through image enhancement products. Upward continuation (e) to 1 km displays long wavelengths and deeper features. The anorthosites, Bjerkreim lobe and country rock retain magnetization to much deeper crustal levels. At 1 km the jotunite dikes are no longer visible, therefore these are assumed to have a shallow depth extent and limited thickness. The map of the analytic signal (f) displays areas of increased magnetization. The magnetization associated with IIId, IVe, foliated anorthosites and jotunites are apparent. The first vertical derivative map (g) displays high frequency, shallow features, especially in the jotunites, faults and cyclic units within the Bjerkreim lobe. The tilt derivative map (h) displays unit contacts and the fold axis. All grids are represented with histogram equalization colour scale and minor artefacts are exhibited in proximity of Heskestadvatnet where data underwent grid knitting.

An initial petrophysics database is compiled of the samples collected by NTNU and NGU (n = 3009) from the Bjerkreim–Sokndal intrusion (NGU Geoscience Data Portal 2023). This database includes petrophysical measurements conducted on-site and in the lab. The petrophysical parameters include Density (D), Susceptibility (K), Induced Magnetization (Ji), Natural Remanent Magnetization (NRM), Inclination (I), Declination (D) and Koenignsberger Ratio (Q), which is the ratio between NRM and Ji and is calculated by:

The induced magnetization is calculated as the product of the magnetic susceptibility of the sample and the value of the geomagnetic field intensity, which for this locality is B = 50 265 nT or H = 40 A m⁻¹.

This database is pared down to samples collected within the Bjerkreim lobe, proximal anorthosites to the west, and country rock to the east. Fig. 1 shows the location of sample localities in the Bjerkreim lobe study area (n = 1377). Some sample locations on the map represent areas with multiple measurements (5–10) commonly collected over an area of at least 10–15 m. These are from sites originally collected for palaeomagnetic studies. There is also an unequal distribution of unit sampling across the study area. This is due to a high density of samples collected within MCU IVe based on previous research into the source of magnetic remanence and mineral exploration.

4 ROCK MAGNETIC AND PHYSICAL PROPERTIES

Scatterplots are presented in Fig. 3 and boxplots are shown in Fig. 4. Averages of each MCU for magnetic susceptibility (K), natural remanent magnetization (NRM) and density (D) are listed in Table 1. All averages are scalar except for the remanence direction, indicated by vector averaged inclination and declination angles. The analysis of this master database guides the forward and inversion modelling.

4.1. Density (D, kg m⁻³)

The samples across the study area have an overall density of 2926 kg m⁻³. The plagioclase rich hemo-ilmenite leuconorites of zone a and c, have the lowest average densities of 2806 and 2842 kg m⁻³, respectively. Leucotroctolite and troctolites of zone b have average density value of 3016 kg m⁻³; Magnetite–ilmenite norite/leuconorite of zone d have an average of 3072 kg m⁻³; Magnetite–hemo-ilmenite gabbronorite with cumulus apatite have an average of 3004 kg m⁻³ and magnetite–ilmenite gabbronorite with inverted pigeonite of zone f has density average of 3040 kg m⁻³.

4.2. Magnetic susceptibility (K, SI)

Magnetic susceptibility varies across the units. The mean value for the entire study area is 5.0 × 10⁻² SI. The highest induced magnetizations are found in upper sections of MCU III and IV (d, e, e’, f), where these magmas are more evolved. The highest site susceptibility value is 13.4 × 10⁻² SI for MCU IIId. This was also shown by McEnroe et al. (2009), where zone d has a magnetic susceptibility an order of magnitude higher than most of the other units and is due to cumulus magnetite with abundant oxy-exsolved lamellae of ilmenite. The lower sections of MCU I, II and III (a, b, c) are more primitive magmas and exhibit the lowest magnetic susceptibility values. These low magnetic susceptibilities are due to an absence, or only very minor amounts of magnetite coexisting with hemo-ilmenite which is the major oxide phase in these units.

4.3. Magnetic remanence (NRM, A m⁻¹)

Zone b has the lowest overall remanence of 0.39 A m⁻¹, but still has a Koenigsberger ratio greater than 1. Zones c, f, a which have remanence averages of 1.15, 1.56 and 1.84 A m⁻¹, respectively. While zones d and e have the greatest remanence of 2.0 and 3.6. Considering most samples have Q ratios >1, remanence should be considered in modelling, especially in MCUs IIIe and IVe with Q ratios >10.

5 MODELLING METHOD

5.1. Principal component analysis

Principal component analysis (PCA) is conducted to understand how different variables in the master data set are related to each other and how they influence the overall distribution of our master data set. In this case, the variables are measured physical properties, such as density, magnetic susceptibility, or NRM. The variance ratios and loading matrix are computed, where the number of principal components (PC) are equal to the number of input variables (m). The variance ratios show the statistical dispersion, so the higher the variance ratio, the greater impact that principal component has on the data distribution. A loading matrix is produced and includes the scores for each variable, where the scores indicate the weight on calculating the principal component. PC1 will be the most important and PCm will be the least important. Therefore, we look to PC1 for the variables most important to magnetic modelling and can be used as initial model parameters.

Eight petrophysical measurements and derivatives from the master database will be used as variables in PCA: density (D), susceptibility (K), induced magnetization (Ji), natural remanent magnetization (NRM), inclination (I), declination (D) and Koenigsberg ratio (Q). The residual magnetic field (IGRF-corrected) is also included (ResTF) using the interpolated grid (Fig. 2a) at each database sample coordinate.

5.2. Forward modelling

Forward modelling is conducted in profile format using Geosoft's GM-SYS2D. Each profile is separated by 2 km and oriented along southwest to northeast so that they are orthogonal to the syncline's hinge axis. The aeromagnetic data set had a 250 m flight line spacing, and therefore the interpolated grid (Fig. 1a) had a grid cell size of 50 m (1/5 flight line spacing). As such, the forward models have an average sampling rate of 50 m along the profiles. Intermediate profiles between the original 2 km spaced profiles are created in areas with complex geology. Considering previous minimum base depth estimates of 4 km (Wilson et al. 1996), the profiles are modelled to a maximum depth of 4.5 km. The local area magnetic field is H = 40 A m⁻¹, I = 72° and D = 2° based on the IGRF at the acquisition date of the aeromagnetic survey. A sensor elevation above ground of 150 m is used to estimate the calculated magnetic anomaly produced by the subsurface geological units in the presence of ambient local geomagnetic field. The input magnetic anomaly profile (observed) is sampled from the NGU aeromagnetic grid (Fig. 2a), and the terrain used is the 10 m digital elevation model (Fig. 2d).

The limited structural information in the study area is taken from the NGU 1:75 000 geology map that includes strike and dip measurements (Marker et al. 2001). In general, the units have a shallow dip (∼60°) along the western ridge compared to the eastern ridge (∼85°). The dip on both sides of the fold steepens moving south.

Initial geological units are constructed using the surface contacts and dip angles from the 1:50 000 geological shapefile (NGU Data Repository). The preliminary magnetic properties are used from the master petrophysical database analysis in Table 1. The 1:75 000 map and 1:50 000 shapefile do not include a subunit IVe’, and therefore the initial forward models only include MCU IVe. Because the subsurface distribution is not well understood, the depth of the vertices of the geological units and the magnetic parameters are modified until the resultant magnetic anomaly generated by the forward model has a reasonable fit (<500 nT) with the observed, while still geologically feasible. 2-D units from the forward modelled profiles are exported for visualization in 3-D. The subsurface vertices of similar 2-D units are then wireframed to create volumes and provide a full 3-D product of the Bjerkreim lobe

5.3. Inversion modelling

There are more studies implementing full magnetization vector (MacLeod & Ellis 2013, 2016; Ellis et al. 2016; Liu et al. 2017; Li et al. 2021; Jackish et al. 2022). Considering the Bjerkreim lobe is a layered intrusion with magnetic units having ratios both greater and less than 1 (Fig. 3c), we apply a two-pronged approach of both magnetic scalar and vector inversions. The scalar inversion process uses only magnetic susceptibility. This will result in reliable modelling for MCU with Q ratios less than 1 because these are dominated by induced magnetization. Modelling results on MCU with Q ratios greater than 1, will result in negative inversion susceptibility values. Conversely, the vector inversion process uses magnetic remanence and therefore is most dependable for MCU dominated by remanence. The inversions are conducted using Geosoft Oasis Montaj's VOXI inversion process that is run through an online network (Seequent Ltd). The inversion method is described in Ellis et al. (2016).

Both inversions use an initial master model with the dimensions of 20 km × 20 km × 5 km. The master model is discretized using 400 m cubic voxels. This voxel size was selected due to uncertainty in the subsurface, especially with increasing depth. Unfortunately, no subsurface geological information is available for the study area. Publicly available drillcore from Tekse, Mjåsund and Ollestad only extend to a maximum depth of 35 m below surface and are therefore not applicable as a constraint for this scale of modelling. The only available constraints are surficial measurements from the master petrophysical database, specifically the NRM and magnetic susceptibility measurements. A 3-D grid is created using a spherical model for kriging and 400 m cubed voxels. This 3-D grid defines the surface layer, or top 400 m of the model. The remaining voxels in the master model below this surface grid all have the same average NRM or susceptibility values for the vector and scalar inversions, respectively. This master model is used in conjunction with the aeromagnetic survey data (Fig. 2a) and digital elevation model for surface topography (Fig. 2d). An error of 5 nT is assumed for all data points. VOXI allows implementation of iterative reweighting inversion (IRI). This is a constraint that utilizes previous inversion results as the weight constraint for the current inversion iteration. A negative IRI focus will emphasize sources that are assumed to be lower than the background. Conversely, a positive IRI focus will emphasize sources that are higher than the background. The default value is 1, but the user can select a value between 0 and 3. Regarding the expected model, units of interest will have a large amplitude difference (susceptibility–0.003 SI and remanence–2 A m⁻¹) relative to the background and therefore the maximum value of 3 is used for IRI focus in the inversion modelling.

6 RESULTS

6.1. Principal component analysis

PCA results are shown as variance ratios in Table 2 and loading matrix in Fig. 5. PC1 has a variance ratio of 39 per cent and is based on the loading matrix, magnetic susceptibility (K), natural remanent magnetization (NRM) and density (D) have the greatest impact on the data distribution. This works effectively because these variables can be modified directly in GM-SYS during the forward modelling process.

With respect to remanence vectors, inclination and declination both have low scores for PC1 (Fig. 5). However, inclination has a higher score than declination for PC2. This is reasonable considering that vectors with steep inclinations, which are common in this study, have large variability in declination, but this variability has minimal impact on the amplitude of the magnetic anomaly. It is verified in the forward modelling that the change in the modelled magnetic anomaly is negligible when the declination is modified.

6.2. Forward modelling

Eighteen 2-D forward models are created running NE–SW (50°) over the Bjerkreim lobe (Fig. 6). It is important to note that a magnetic anomaly may represent an infinite number of subsurface configurations, which is why physical property and geometry (e.g. dip) constraints must be used to limit the number of subsurface solutions. The initial physical properties for the units are selected using the values from Table 1. The geological units in the subsurface are first modelled manually until an RMS error below 500 nT is achieved. GM-SYS2D allows for an inversion to be applied to vertices and units’ properties. The depth of the vertices and magnetic susceptibility are fine-tuned using this internal GM-SYS inversion tool to achieve a closer fit and reduce RMS error, which was typically only tens of nT. The profiles are on average 8 km long. Units at either end of the profiles, typically the anorthosites to the west and country rock to east, are extended beyond the profile to 1 km to minimize edge effects. Profiles 1, 7 and 15 are assessed in further discussion (Fig. 6).

6.2.1. Profile 1

Profile 1 lies at the north end of the Bjerkreim lobe. This profile intersects the early part of the intrusion and includes MCU I and II. In situ measured dip angles (Marker et al. 2001) are shallow, varying between 60° at the margins and 30° near the middle of the profile. The forward model of profile 1 is shown in Fig. 7(a). There are few in situ samples from this area, but the geology is well mapped and is simpler than the more evolved parts of the BKS, therefore we consider the averages from Table 1 used in this profile to be appropriate to achieve a dependable forward model. The general fit has an error of 128 nT.

6.2.2. Profile 7

Profile 7 lies near the centre of the Bjerkreim lobe, crossing just north of Teksevatnet. This profile intersects most units of MCU II, III and IV, as well as the Transition Zone and mangerite and quartz mangerite bodies. The units become quite condensed at the eastern margin, where the Bjerkreim lobe is pinched against the country rock. In situ measured dip angles are around 85° along the western and eastern boundary with the anorthosites and country rock, respectively, but then shallow towards the axis of the fold, closer to 60°. This area is interesting because samples from MCU IVe have a large variability in NRM values from east to west with remanence dominating the magnetic response on the western side of the intrusion. This unit has a large potential for an apatite, ilmenite, and magnetite deposit with up to 12 wt per cent apatite (Schiellerup et al. 2001). Many historic ilmenite and magnetite mines are in the surrounding anorthosites. The Tellnes norite, which is one of the largest actively mined ilmenite deposits in the world is in the Åna Sira Anorthosite, located just south of the Bjerkreim lobe.

The forward model of profile 7 is shown in Fig. 7(b). The units are modelled through a combination of known mapped dips, and in situ measurements, as there are many samples in proximity to this profile. The general fit has an error of 368 nT, which is high due to the mismatch along the western margin, specifically in MCU IVe and IIId. This mismatch is due to properties used for MCUs IVe and IIId on the eastern margin, which are not suitable for the western margin, and need to be modelled separately. The quartz mangerite is also modelled to a depth of 2 km, which aligns with the block models presented in Smithson & Ramberg (1979).

6.2.3. Profile 15

Profile 15 lies near the south end of the Bjerkreim lobe, crossing just south of Bilstadvatnet. This profile intersects most units of MCU II, III and IV, as well as the Transition Zone, mangerite, and quartz mangerite. In situ measured dip angles are around 75° along the western and eastern boundary with the anorthosites and country rock, respectively. Samples in proximity of this profile have a high magnetic remanence (>5 A m⁻¹)

The forward model of profile 15 is shown in Fig. 7(c). The units are modelled using a combination of mapped dips and the property averages from Table 1, since the only nearby measurements available are for IVe. The best fit achieved has an error of 636 nT, which is due to the fit differences on the western margin, specifically IVe and IIId. Again, IVe and IIId need to be modelled differently on the western and eastern margins.

The 2-D units determined from the magnetic profiles are exported from the profiles and wireframed to create 3-D volumes. These 3-D volumes are brought into the same space to provide a full 3-D product of the Bjerkreim lobe, as shown in Figs 7(d) and (e). Here the overall shape of the syncline is well delineated and shows how the units to the southeast taper off, whereas the units along the southwest are truncated.

A forward model using the gravity data is conducted along one profile aligned with the seismic profile previously obtained along a roadway (Deemer & Hurich 1997; Fig. 8a). The gravity profile crosses the Bjerkreim lobe from west to east. This single profile is selected based on a higher number of gravity stations in this area, typically less than 1 km apart and provides the highest resolution of gravity anomalies across the Bjerkreim lobe. Although the sampling resolution is much lower than the aeromagnetic data set used for forward models 1 through 18 (Fig. 7), it is still sufficient to calculate the base depth of the Bjerkreim lobe.

The forward model suggests a depth of 4.5 km, which aligns with depths in Smithson & Ramberg (1979) and Deemer & Hurich (1997). The average densities used in the model of the Bjerkreim lobe, anorthosite and country rocks are 2931, 2644 and 2802 kg m⁻³, respectively.

6.3. Inversion modelling

A preliminary inversion is conducted for comparison with the forward model. Here the 20 km × 20 km × 5 km study area is comprised of 400 m × 400 m × 400 m voxels. The base model uses the average magnetic susceptibility and magnetic remanence, 0.03 SI and 0.04 A m⁻¹, respectively, for the entire inverted volume except for the top layer of voxels where the 3-D kriging grid of susceptibility and NRM are used for the scalar induced and vector inversions, respectively (Fig. 9).

The induced magnetization model uses the aeromagnetic grid, DEM and surface susceptibility grid (Fig. 9a) in combination with an IRI focus of 3 for positive anomalies, to drive the model in the direction of units with high magnetic susceptibility which contain high magnetite content such as IIId, magnetite-ilmenite norite/leuconorite and IVe, magnetite–ilmenite gabbronite with cumulus apatite. These units can have up to 10 per cent magnetite.

The magnetic vector intensity model uses the aeromagnetic grid, DEM and surface remanence grid (Fig. 9b) in combination with an IRI focus of 3 for the negative anomalies (below background), because this emphasizes the modelling in the direction of units with high remanence with a reversed inclination, such as IVe´. The results are shown in Figs 10 and 12.

Fig. 10 is a series of four depth slices from the scalar inversion, showing how the Bjerkreim lobe, the anorthosites to the west, and country rock to the east vary at 1 km depth increments. Fig. 10(a) is the inversion result at a 1 km depth slice with the surface geological contacts from the 1:50 000 geology map overlain. The megacyclic layered structure (MCUs) of the Bjerkreim lobe is still evident at 1 km. Here, the foliated anorthosites of the Egersund—Ogna (EO—pink) west of the Bjerkreim lobe are modelled with a magnetic susceptibility higher relative to the massive Egersund—Ogna anorthosites (EO—blue) and Håland—Helleren anorthosites. The country rock, specifically the weakly foliated granitic gneiss, is also modelled again with a comparatively high magnetic susceptibility. It is important to note these are relative evaluations. The anorthosites and country rock do not have characteristically high magnetic susceptibilities (Brown & McEnroe 2008). In addition, there are many historic ore deposits in the anorthosites (Duchesne 1970, 1999; Krause & Pedall 1980; Robinson et al. 2013) with unknown depth extents which could locally affect the magnetic properties. These foliated anorthosite units may also be modelled with elevated susceptibility. To model better the Egersund-Ogna foliated anorthosites, in situ magnetic susceptibility measurements from this area are required. At depths below 3 km (Figs 10c and d) the layered structure of the Bjerkreim lobe is no longer apparent, but rather a summation of the magnetic sub-units (d, e, f) especially towards the west and may be due to the compression and faulting along the eastern perimeter of the Bjerkreim lobe. The foliated anorthosites and weakly foliated granitic gneiss are still apparent at greater depths (Figs 10c and d). We have also identified two faults along the eastern margin in the Teksevatnet area (F1, F2 in Fig. 10a). These two are sketched at surface (0 km) and projected onto each depth slice. Perspectives (Fig. 11) show changes in geometry both laterally and with depth. These faults are both steeply dipping, which correspond to a cross-section in proximity of another fault from Marker & Slagstad 2018. Furthermore, F1 although extensive at the surface, shortens with depth and is no longer apparent at 4 km. Meanwhile, F2 becomes more extensive at depths, up to 4 km.

Isosurfaces are created to visualize the structure in 3-D and permit for understanding better the dip structures below surface (Fig. 12). High magnetization (>0.1 SI) and low magnetization (<0.003 SI) are shown as red and blue, respectively. Based on Fig. 12, the fold structure of the Bjerkreim lobe extends to at least a depth of 4 km. The Bjerkreim lobe is pressed against the massive anorthosites to the west and the porphyritic metagranite to the east.

Total magnetization vectors are shown as elongated cones, with length and colour representing amplitude of remanent magnetization (Figs 12d–f). The Heskestad anomaly is identified at insets Figs 12(e) and (f) and has the highest remanent intensity (pink and red). The inclination of these vectors is near vertical and positive. The magnetization vectors associated with the Heskestad anomaly are reversed (negative) based on measurements of local in situ samples, however here due to the steep inclination of the remanence, the inversion has computed these as positive. High remanence values are apparent on the western boundary of the Bjerkreim lobe in the foliated anorthosite with vector clusters at depths greater than 1 km. This high magnetization is associated with the foliated anorthosites of the Egersund-Ogna and northern part of the Håland-Helleren. The reported inclinations are very steep (−71 to −79) and are documented by Brown & McEnroe (2004, 2008, 2015) and McEnroe et al. (2009). Here the northern part of the Håland-Helleren anorthosite borders the southwestern contact with BKS. The reported NRM intensities for the Håland-Helleren anorthosite are high with an average intensity of 6 A m⁻¹. In contrast the induced magnetization is only 1 A m⁻¹ (Brown & McEnroe 2008). This also shows that >95 per cent of Rogaland anorthosite samples are dominated by remanence. This distribution of induced and remanence values is displayed in Fig. 3(c). To ensure that the inversion results within the Bjerkreim lobe are not impacted by near-surface sediments, such as those in proximity to rivers, a regional-residual seperation is conducted (Fig 13). Here the aeromagnetic data is upward continued to 1 km (Fig. 2e, regional) and then the original aeromagnetic data (Fig. 2a, residual) is subtracted. In Fig 13, all magnetic anomalies within the Bjerkreim lobe are associated with MCUs or other geological units, such as jotunites.

7 DISCUSSION

Unit IVe has the highest n samples and the greatest spatial coverage of the study area, although the exposure is partially limited by large lakes in the area, including Heskestadvatnet in the south, and Teksevatnet and Bilstadvatnet in east. The coverage and high sample number permits spatial analysis by gridding the induced (Ji) and remanent (NRM) magnetizations measurements from the master database (Fig. 14). The highest induced magnetization is in the northwest and conversely, the northwest has the lowest remanent magnetization. The lowest induced magnetization is in the southeast in proximity to Heskestadvatnet, and conversely the highest remanent magnetization is measured in the southeast. The continuous distribution of IVe’ along the southwest is also reflected in Figs 2(f)–(h), unit IVe’.

Plotting the remanence directions of MCU IVe from the sample database (Fig. 15), as magnetization increases, the inclination becomes steeper. Where IVe changes from low remanence (<5 A m⁻¹) and shallow inclination (<50°) to high remanence (>20 A m⁻¹) and steep inclination (60–80°) which could indicate a change in mineralogy. This supports the consideration of the subunit, IVe’, which should be distinct from MCU IVe. This has previously been suggested by McEnroe et al. (2009) and Brown & McEnroe (2015).

IVe’ is then included in the forward models, where updated profiles 7 and 15 are presented in Fig. 16. The overall error reduces to 174 and 163 nT, respectively, by including the IVe/IVe’ contact. For initial modelling, the contact between IVe and IVe’ is presented as abrupt but is gradational in reality. This transition between IVe and IVe’ is refined better through higher resolution magnetic surveys. Accordingly, ground magnetic data were acquired in 2023 with a Geometrics G-859AP cesium vapour magnetometer at a sampling rate of 5 Hz, approximately 2 m above ground and with 10 m line spacing along NE strike. Grids were collected across key contacts and within units of interest, which included IVe. Preliminary data is IGRF-corrected to exhibit magnetic anomaly below the ambient magnetic field. Fig. 17 presents two small magnetic surveys collected within IVe, south of Bilstadvatnet. The data show variability in magnetic anomaly on the order of 12 000 nT below background.

8 CONCLUSION

The BKS is a double-plunging layered intrusion in southern Norway. It exhibits complex geology, coupled with high magnetic remanence and potential for large critical mineral occurrences. Despite decades of research, there are limited studies presenting the distribution of the layered series in the subsurface.

Here, the different magmatic units (MCU) were characterized using a master database of magnetic measurements collected from the study area. The master database is used to define the units’ magnetic properties for creating 2-D forward modelled profiles. These forward models were combined to create a preliminary 3-D interpretation of the Bjerkreim lobe. A two-pronged approach was then conducted with scalar and vector inversion modelling with the aim of imaging units dominated by induced or remanent magnetization. Because there were no subsurface constraints due to limited availability of shallow drillcores (<35 m), the only constraints used are voxel grids of the magnetic susceptibility and magnetic remanence from the master database of petrophysics samples. The 3-D interpretation from forward modelling is then compared with inversion outputs to create an initial assessment.

The forward and inversion modelling indicates the Bjerkreim lobe dips south and reaches a depth of 5 km at the southern part of the Bjerkreim lobe. Near-surface features, less than 1 km deep, include the jotunite intrusions and anorthosite blocks within western and northern part of IVe. The Egersund -Ogna and Håland-Helleren anorthosites extend below the Bjerkreim lobe, at depths greater than 4 km, supporting the concept that the BKS intruded into these anorthosites. The complexity of fitting the observed magnetic field in the forward models containing IVe demonstrates that this unit is more complex than mapped. The spatial distribution of induced and remanent magnetization, and associated remanence directions of IVe within the Bjerkreim lobe support previous published work that the contact for IVe should be reconsidered, with the addition of subunit IVe’.

Based on the preliminary forward and inversion models, new insight on potential subsurface configurations and magnetic properties has been presented. However, the biggest hurdle encountered in the modelling was lack of subsurface constraints. Therefore, it is possible that there is subsurface topography we have not adequately accounted for and the details of the structure of the Bjerkreim lobe are more complex than presented here. This is especially true along the eastern margin of the Bjerkreim lobe in the Teksevatnet area that undergoes faulting and the MCUs present large physical property variations. Therefore, it is beneficial to acquire additional high-resolution data, in the form of core samples and high-resolution surveys over complex areas. Future work will incorporate more local variations in the geology interpreted from high-resolution magnetic data collected over the study area by an uncrewed aerial vehicle in 2021 (Lee et al. 2023) and 2023, along with new ground magnetic data. Secondly, deep drill cores measuring over 1000 m will hopefully be made available for scientific research.

Acknowledgement

NTNU and Research Council of Norway project 222666 to S.A. McEnroe supported numerous field campaigns. Peter Robinson's scientific contributions on the magnetic mineralogy and petrology of BKS provided a solid foundation for this work. Contributions from Alex Michels, Andrea Biedermann, Geertje ter Matt, Alexandra McEnroe, Laurie Brown and Peter Robinson for fieldwork and the support of the Helleland community are gratefully acknowledged. Nathan Church, Christine Fichler, Karl Fabian and Pedro Acosta are thanked for their scientific discussion. NGU is gratefully acknowledged for providing the aeromagnetic data which were acquired in the Coop project (Crustal Onshore-Offshore Project) of the Geological Survey of Norway. Finally, the authors thank Aziz Nasuti and an anonymous reviewer whose constructive suggestions helped improve and clarify the paper.

DATA AVAILABILITY

The NGU geophysics data sets used here (petrophysics, aeromagnetics, radiometrics, gravity, digital elevation model and shapefiles) are publicly available through NGU Geoscience Data Service, https://geo.ngu.no/geoscienceportalopen/Search.

References

Author notes