INTRODUCTION

Fennoscandia, or the Baltic Shield, is a protrusion of the Precambrian basement in the northwestern part of the East European Platform with relatively high seismic activity compared to the rest of the platform. The nature of this seismicity has been discussed in the literature over the last 40–50 years, but there has not yet been a conclusive finding on its origins. The specific feature of this seismicity is that it occurs on a territory with a repeated cycle of largest glaciations (the last one with a length of almost 20000 years).

There was no theory of “glacial periods” for Fennoscandia until the mid-19th century. Jamilson (1865) first proposed considering the present uplift of the Baltic Shield as evidence for the deformation of a non-rigid Earth under the ice cap (“glacial isostasy”). One of the pioneering studies that associated this deformation with seismicity (Kolderup, 1930) suggested that the seismicity of Fennoscandia is largely due to the postglacial uplift. This idea was developed in (Arvidsson, 1996; Bungum et al., 1991; Ekman, 1998; Gudmundsson, 1999; Mörner, 1979; Wahlstrom, 1989), which noted that seismicity was confined to the neighborhood of the postglacial dome. Gudmundsson (1999) used model calculations of the distribution of stresses around the uplifting dome to propose that the discontinuous tectonic movements are possibly confined to the marginal parts of the dome due to its postglacial uplift. At the same time, according to instrumental measurement data, weak seismicity (M ≤ 4) is nevertheless observed in the region of peak postglacial uplift of Fennoscandia (Keiding et al., 2015; Olesen et al., 2013a).

The rate of postglacial uplift is sufficiently high. At a not so distant historical age, the rate of uplift along the Fennoscandia coasts after the last glaciation cycle was so high that its effects were easily observed during the life of a single generation (Ekman, 1991; Fjeldskaar et al., 2000; Steffen and Wu, 2011). It is this uplift that has been often associated with remarkable earth tremors recorded in chronicles as well as observations of numerous paleoseismic dislocations with sizes pointing to the possibility of sufficiently strong seismic events in the distant past up to M = 7–8 (Mörner, 2005, 2017; Mörner et al., 2008).

With time, the uplift rate reduced and the number of remarkable earthquakes decreased. This was discussed in (Mörner, 1979), where it was suggested that the observed postglacial uplift consists of two (linear and exponential) components. The exponential component has a glacial isostatic origin and the linear component is of a tectonic origin. In addition, Mörner (1979) stated that the current uplift is most likely a tectonic process with a relatively low linear uplift rate determining the low level of present seismic activity of the given territory, rather than a postglacial isostatic phenomenon. If this assumption is accepted, it makes sense to admit that the high prehistoric seismic activity of Fennoscandia should be regarded only as a past episode and its effects on present activity should be extended very carefully.

Fjeldskaar et al. (2000) attempted to directly identify a possible neotectonic component in the postglacial uplift and obtain its quantitative estimate by considering seismotectonic deformation from data on the focal mechanisms of current weak earthquakes. The authors of that study concluded that the present deformation along the coastline of the middle part of Norway and along the east coast of Sweden, which is accompanied by relative increased seismicity, cannot be explained by glacial isostasy and most likely is associated with present tectonics.

The nature of seismicity of Fennoscandia and generating stresses has been discussed in the literature. The nature of the current stress field was widely discussed in terms of possible driving mechanisms (Fejerskov and Lindholm, 2000; Gregersen, 1992, 2006; Gregersen et al., 1991; Stephansson, 1988). Four force mechanisms affecting the seismic process in Fennoscandia are discussed to varying degrees.

The first is the northwest-to-southeast motion of the lithospheric plate under the Norwegian and Barents Seas due to the spreading of the Mid-Atlantic Ridge in the segment from Iceland to Spitsbergen. It is assumed that the sub-horizontal contraction stress caused by this motion can act as a source for the observed seismicity of the Baltic Shield (Arvidsson 1996; Bungum et al., 1991; Byrkjeland et al., 2000; Craig et al., 2016; Gregersen, 1992, 2006; Slunga, 1981, 1984, 1989; Usenko et al., 2018).

Secondly, the present postglacial isostatic uplift known locally is often taken as one of the most important sources for both paleoseismicity and present seismicity. In this case, it is assumed that the tensile stresses acting due to this uplift are confined to the periphery of the uplifting dome, and this should be seen in the spatial structure of seismicity (Arvidsson and Kulhanek, 1994; Ekman, 1998; Fjeldskaar et al., 2000; Gregersen, 2006; Muir-Wood, 1989; Wahlstrom, 1989; Wu et al., 1999).

Thirdly, a number of authors consider the observed local (with linear sizes in the first tens/first hundreds of kilometers) deviations from the postglacial uplift model as a possible manifestation of independent present neotectonic motions contributing to the manifestation of present seismicity (Fjeldskaar et al., 2000; Olesen et al., 2013a, 2013b; Solheim et al., 2005; Vestol, 2006).

Fourthly, a number of authors interpret the relatively high present seismic activity of the northern and western coasts of Norway as a manifestation of the gravitational effect caused by the accumulation of a thick sedimentary stratum in the coastal shelf due to the intense erosion from the rising crystalline domain of the Baltic Shield, which coupled with the isostatically rising land leads to bending (shear) stresses at the contact of the continental coast and the sea shelf, leading to the seismicity observed here (Arvidsson, 1966; Arvidsson and Kulhanek, 1994; Olesen et al., 2013a, 2013b).

This study aims to perform a combined analysis of the above-listed sources of seismicity of the Baltic Shield on the basis of literature data and choose potential priority areas.

POSTGLACIAL ISOSTASY

The history of glaciations in Fennoscandia is at least 600000 years old (Olesen et al., 2013a). During this time period, the territory has undergone at least 10 cycles of glaciation and deglaciation. Here, each phase of deglaciation in the mainland of Fennoscandia was accompanied by a surge in seismic activity, while powerful ice sheets almost completely suppressed it (Johnston, 1987).

The isostatic response to the last phase of deglaciation in Fennoscandia was theoretically calculated, for example, in (Fjeldskaar et al., 2000) using a model of the Earth with a layered mantle viscosity of less than 7.0 × 1019 Pa s in the asthenosphere with a thickness of around 150 kilometers, overlain by an elastic lithosphere, while the viscosity of the mantle below the asthenosphere was assumed to be 1021 Pa s. The modeling is based on the mapped deglaciation with a spatial resolution of approximately 20 × 20 km (Fjeldskaar et al., 2000). The model of the glacier thickness gives the flexural rigidity of 1023 N m at the Norwegian coast, increasing to above 1024 N m in central parts of Fennoscandia. The rate of present uplift of Fennoscandia was calculated using tide-gauge data, high-precision leveling data, and gravity and GPS measurements.

Using all available data, Fjeldskaar et al. (2000) developed a model of the total coverage and thickness of ice cover during the deglaciation of Fennoscandia for different time periods from 20 000 to 9300 BP (the simplified version of this model is shown in Fig. 1). These data allow one to estimate the rate of deglaciation. It can be seen that the ice sheet of Fennoscandia almost disappears by around 9000 BP, which can be clearly seen by comparing Figs. 1a and 1b. Of course, the ice sheet is absent at present.

Fig. 1.
figure 1

Total area and thickness of ice cover (m) during the deglaciation of Fennoscandia (Fig. 5 adapted from (Fjeldskaar et al. (2000)). (a) 20 000 BP; the interval between contours is 500 m and (b) 9300 BP; the interval between contours is 200 m.

Full size image

The model of present rates of postglacial isostatic uplift on the basis of the above-mentioned glacier melting model was used in (Agren and Svensson, 2007; Fjeldskaar et al., 2000; Vestol, 2006). Figure 2 shows its version taken from (Agren and Svensson, 2007). According to this model, the present rate of postglacial uplift is close to vanishing along the Norwegian coast, reaches small negative values in northwestern Europe, and increases to 7–8 mm/year in the central part of the Gulf of Bothnia of the Baltic Sea. These values are close to the estimates of present vertical shifts according to GPS measurements.

Fig. 2.
figure 2

Model NKG2005LU of modern rates of the postglacial isostatic uplift of Fennoscandia (adapted from (Agren and Svensson, 2007)).

Full size image

It is this postglacial isostasy that is considered to be the main cause of the present uplift of Fennoscandia, which, however, does not contradict the assumption that the observed uplifts also have a neotectonic (i.e. nonglacial–isostatic) component. It is assumed that the local areas with significant differences between observed and calculated uplifts are presently subject to vertical deformations of clearly nonglacial–isostatic origin (Agren and Svensson, 2007; Fjeldskaar et al., 2000; Vestol, 2006). Similar significant deviations are shown in Fig. 3.

Fig. 3.
figure 3

Domains of significant deviations of observed uplift rates from calculated rates of glacial isostatic uplift (Figs. 10 and 11 adapted from (Fjeldskaar et al., 2000)).

Full size image

The most significant positive deviations of the observed uplift rates (1.0 mm/year or more) relative to the calculated rates of the glacial isostatic uplift occur in the mountainous regions of coastal Norway. The most significant negative deviations of the observed uplift rates relative to calculated values (1.0 mm/year or more) are observed along the coastline of the northern part of the Baltic Sea in the Gulf of Bothnia and less clearly south of the east coast of Sweden. It should be noted that these anomalous regions are characterized by a significant role of neotectonic processes in the manifestations of the present topology, while the remaining mainland of Fennoscandia is characterized by a significant role of postglacial isostasy (Fjeldskaar et al., 2000).

The simulation also predicts the emergence of a transition zone between the uplift of Fennoscandia as a single crystalline domain and the lowering of the bottom of the shelf part of Norway. This zone should be located at some distance away from the coast, and the bending stresses in the crust peak in this zone. Further, we will show that it is here within the Norwegian shelf that the highest seismic activity is observed in the given territory; by contrast, there is almost no activity in the central mainland of Fennoscandia.

SEISMIC ACTIVITY

The earliest historical records of earthquakes in Fennoscandia are found in documents intended for other purposes (church books, parish documents, etc.) and have limited scientific value (Fjeldskaar et al., 2000). Although the first surviving reports on seismic tremors date back to 1073, the assessment of macroseismic observations of the distant past in Fennoscandia is complicated by the fact that remarkable earthquakes were poorly documented due to the rather high tremor-resistance of wooden houses, which were the predominant type of local housing in the past (Mantyniemi et al., 2004).

More complete earthquake surveys in this region began in the first decades of the 19th century, while the systematic use of macroseismic questionnaires in surveying remarkable earthquakes began late at the 19th century (Mantyniemi et al., 2004). It was only in the 1980s that the first sensitive electromagnetic seismographs were installed in Norway, capable of detecting earthquakes that are insensible to humans; similar installations appeared almost concurrently in Sweden, Finland, and Denmark (Fjeldskaar et al., 2000). In Russia, a seismic network was installed during the late 1980s and early 1990s with scales sufficient for recording weak (with M > 1) seismic events at reasonable distances away from recording stations.

Our understanding of the low seismic activity of Fennoscandia is largely based on 35-year instrumental microseismic data. However, this low (or even average) seismic activity of Fennoscandia is the highest in Northern Europe. During the 19th–21st centuries, Norway (the most seismically active part of Fennoscandia) and the adjacent marine regions had seven earthquakes with M = 5.3–5.8 (Olesen et al., 2013a). The earthquake occurrence frequency according to all available historical data is described by the relation log(N) = 4.32 – 1.05 МW (Bungum et al., 2000; Olesen et al., 2013a). This means one earthquake with M = 5 every eight to nine years and one with M = 6 every 90–100 years. It follows from here that such a strong earthquake may occur relatively soon in one of the most seismically active regions of Norway—most likely in the Oslo Fjord region or in the coastal regions of Western and Northern Norway because there have been no earthquakes with M = 5 for more than 20 years; according to calculations, such earthquakes should occur once in 10 years (Olesen et al., 2013a). It should be noted in this text that there had been no such earthquakes at the time of writing. There were only three earthquakes (in 2015, 2016, and 2017) with magnitudes MB = 4.8–4.9 (according to the ISC catalog); however, all of them occurred offshore, in the Norwegian Sea.

The literature contains a large number of maps for different time intervals and with different minimum magnitude levels, demonstrating the spatial structure of the Baltic Shield seismicity. The main elements of the spatial structure of seismicity persist in various schemes. Here, we consider one of these schemes constructed in (Keiding et al., 2015) according to data of the Helsinki Bulletin (FENCAT) of earthquakes of magnitude ML ≥ 2 for 1980–2011 (Fig. 4). It is likely that this scheme most reliably describes the details of the spatial structure of seismicity because the seismic network in the given time period was sufficiently complete to record seismic events.

Fig. 4.
figure 4

Map of epicenters of earthquakes in Fennoscandia according to the Helsinki bulletin of earthquakes (FENCAT) of magnitude ML ≥ 2 for 1980–2011 (a fragment of Fig. 1 adapted from (Keiding et al., 2015)). (1) Postglacial thrust faults and (2) shelf margin.

Full size image

It should be noted that the earthquake foci in the Baltic Shield are concentrated mainly within the upper crust at depths between 0 and 20 kilometers. The highest seismic activity is observed in a highly fragmented rift zone in the North Sea at the coast of Southwestern Norway (Hicks et al., 2000; Lindholm et al., 2000, 2005; Olesen et al., 2013a, 2013b). In this case, the oceanic crust of the Norwegian and Barents seas is predominantly aseismic, with the exception of the eastern part of the Lofoten sedimentary basin in the Norwegian Sea, possibly due to the local lithospheric bending associated with the rapid sedimentation of glacial deposits in the last phase of postglacial isostasy (Byrkjeland et al., 2000).

In continental Fennoscandia, the most seismically active zone is southwestern Norway, where the crust is highly fragmented by post-Caledonian faults and shear zones (Boe et al., 2010). Let us compare with it the seismicity of the Nordland region represented by the coast of the middle part of Norway, where there have been two clearly expressed (see Fig. 4) near-surface swarms of earthquakes in recent years (Atakan et al., 1994). This swarm activity is characterized by magnitudes rarely reaching M = 4. At the same time, it was here that the strongest earthquake (in 1819, with MS = 5.8) known historically in Fennoscandia occurred (Bungum and Olesen, 2005).

It should be noted that Nordland is a region where, according to Fig. 3, significant positive deviations of observed uplift rates relative to calculated postglacial isostatic uplift rates are concentrated. In addition, this region is characterized by gravitational bending deformations at the contact of the strong 50-kilometer crust of the crystalline domain of Fennoscandia with a relatively thin crust of the sea shelf due to intense erosion from the rising crystalline domain of the Baltic Shield.

More diffuse seismicity occurs along the coast of Sweden in the Gulf of Bothnia and goes further through Northern Finland towards the Barents Sea as a dispersed cloud. Here, the seismicity is matched by a clearly expressed region of negative deviations of observed uplift rates relative to calculated glacial–isostatic uplift rates (see Fig. 3). These deviations are sometimes interpreted as a manifestation of present tectonic movements as a lowering of the bottom of the Gulf of Bothnia. Similar diffuse seismicity is observed in southern Sweden and Oslo-Graben, where, as shown below, a subhorizontal strain in the northeast–southwest direction prevails.

The central part of the crystalline domain of Fennoscandia is almost aseismic. It should be noted that it is here, according to Fig. 2, that a maximum rise of the domain surface is observed due to its postglacial isostatic uplift. Therefore, it should be recognized that this rise is not the driving force of seismicity.

FOCAL EARTHQUAKE MECHANISMS AND STRESS STATE OF THE EARTH’S CRUST OF FENNOSCANDIA

Information about the mechanisms of earthquake foci allows one to set the main directions of stress and deformation mode. Earthquakes are generated from the accumulation of stresses and their subsequent release; if the stress accumulation mode is given by some dominant force mechanism, the information about it should manifest itself in focal mechanisms.

Hicks et al. (2000) performed a rather comprehensive compilation of all the 112 options known at that time for the mechanism of earthquake foci in Norway and adjacent regions. The results of the analysis reveal a prevailing tendency to thrust formation in the northwest–southeast direction in the open sea and in the northeast and southwest of Norway. This is consistent with the supposed regional field of compressive stresses oriented in the northwest–southeast direction, which is presumably associated with the movement of the oceanic lithospheric plate in the indicated direction due to the expansion of the Mid-Atlantic Ridge in the course of spreading.

At the same time, there are areas with predominant fault movements in the southern part of the North Sea, in the Oslo fjord area, and in Nordland. Specifically, for a relatively large number of weak earthquakes in coastal areas in the middle part of Norway (Nordland), tensile stresses perpendicular to the coastline are dominant. These tensile stresses can be caused both by local tectonic uplifts recorded as positive deviations from the calculated values of this postglacial uplift (see Fig. 3) and by bending deformations at the contact of strong continental coastal crust with a relatively thin crust of the sea shelf due to intense erosion from the rising crystalline domain of the Baltic Shield. Similar but more emphasized results based on almost the same number of focal mechanisms were obtained in (Fjeldskaar et al., 2000).

The most comprehensive analysis of the focal mechanisms of Fennoscandia was performed recently in (Keiding et al., 2015). The authors of that study had a sample of 224 focal mechanisms, which is almost twice as large as in (Hicks et al., 2000); 126 mechanisms had a high quality of resolving (A or B). A number of earthquake epicenter clusters with certain focal mechanisms were identified; for them, the mean moment tensors weighted by the sum of earthquake seismic moments were calculated. The form of these tensors and the projection of the major axes P and T of the tensor on the horizontal plane are shown in Fig. 5.

Fig. 5.
figure 5

Focal mechanisms of Fennoscandia averaged over seismic clusters with the projections of major axes of compression P (1) and tension T (2) to the horizontal plane and their relative values of axes (a fragment of Fig. 3 adapted from (Keiding et al., 2015)).

Full size image

Based on the distribution of major axes of compression and tension of focal mechanisms shown in Fig. 5, we describe the type of seismotectonic deformation in the horizontal plane. In the region corresponding to the Norwegian Sea shelf, the focal mechanisms demonstrate consistent compression in the west–northwest to east–southeast direction. The focal mechanisms are almost the same in the continental part of Fennoscandia in the extreme northwest in northern regions of Sweden and Finland as well as in southern Finland. For the remaining mainland of Fennoscandia, the mechanisms show large variations both along the major axes and in the deformation tensor type.

This is especially typical for regions surrounding the central dome of the postglacial isostatic uplift. Specifically, Nordland in Norway and the coast of the Gulf of Bothnia in Sweden are characterized by the prevalence of subhorizontal strain in the northwest–southeast direction almost across the coasts. In regions northeast and southwest of the central dome, this subhorizontal strain sharply changes its orientation, acquiring a northeastern–southwestern direction. Unlike the above-considered normalized subhorizontal compression in the northwest–southeast direction, this predominance of subhorizontal tensile strain is the most clearly expressed feature of the stress-strain state of Fennoscandia.

This finding is sufficiently reliable because it was confirmed by additional analysis of focal mechanisms for larger seismic events (Keiding et al., 2015). Thus, it can be assumed that the observed distribution of tension axes in the focal mechanisms (quasi-radial with respect to the glacial uplift dome) may be associated with the effect of postglacial isostasy.

PRESENT DEFORMATIONS ACCORDING TO GPS-MEASUREMENT DATA FOR FENNOSCANDIA

The BIFROST project was launched in Fennoscandia in 1990 to deploy a network of GPS stations to estimate the field of present postglacial isostatic uplift rates in Fennoscandia with a typical distance of 100–200 kilometers between neighboring GPS stations. The first field of GPS movements for the entire Fennoscandia was analyzed in (Kierulf et al., 2014), which combined data from the earlier existing spatially dense Norwegian network with data from the BIFROST network stations and other high-quality GPS stations. According to the resulting field of displacement rates, the rise in the central part of Fennoscandia near the northern part of the Gulf of Bothnia is up to 10 mm/year with a decrease to 1–2 mm/year along the periphery of Fennoscandia. The distribution of the horizontal component of the GPS displacement rates was also controlled by this rise: the displacement field is quasi-radial with a rate of 1–2 mm/yr relative to the dome in the center of the rise (Kierulf et al., 2014).

A more comprehensive analysis of present deformations according to GPS measurements was performed in (Keiding et al., 2015). Figure 6 shows the vertical displacements of the daytime surface as well as the directions and magnitudes of horizontal displacements. It can be clearly seen that the GPS stations with a rise of up to 8–10 mm/year (gradation 5–6) are localized within an ellipsoid region with a major axis of around 500 kilometers, elongated in the northeast–southwest direction and almost exactly coinciding with the region of maximum postglacial isostatic uplift shown in Fig. 2

Fig. 6.
figure 6

Velocity field of daytime surface movements in Fennoscandia according to GPS measurements (project BIFROST) with respect to stable Eurasia in the reference system ITRF2008 (a fragment of Fig. 1 adapted from (Keiding et al., 2015)). Vertical movements: (1) minimum and (6) maximum. The arrows show the direction and magnitude (mm/year) of horizontal shifts. The size of the ellipse at the arrow ends corresponds to the variance of the estimate in azimuth and magnitude.

Full size image

As they spread in different directions from this central part, the uplift rate decrease to 1 mm/year and, in some places (on the northern, northwestern, and southern coasts of Scandinavia), even below that value, which also agrees with the calculated present postglacial isostatic uplift rates shown in Fig. 2. The length of the horizontal displacement velocity vectors for GPS stations with maximum vertical motion velocities of 8–10 mm/year (gradation 5–6) turns out to be insignificant (less than 0.5 mm/year). This may serve as evidence of the rise of the upper part of the dome of the postglacial uplift almost as a whole.

The horizontal displacement velocity vectors su-rrounding this uplift area, reaching 1 mm/year or more, are directed predominantly radially with respect to the central uplift in accordance with the expected movements of the postglacial uplift dome surface. This region of radial horizontal displacements has an elongated form in the northeast–sou-thwest direction and is characterized by the s-ignificant tensile deformations directed predominantly across, with a value of up to 4 × 10−9 years–1 (Keiding et al., 2015).

Thus, the distribution of vertical and horizontal displacements of the daytime surface observed during long-term GPS measurements within Fennoscandia (Fig. 6) is sufficiently well described by the computational model of present postglacial isostatic uplift rates (Fig. 2).

DISCUSSION

It is of interest to find force mechanisms governing the combined stress-strain state of Fennoscandia and their role in the formation of seismic activity. We already noted in the Introduction that according to modern concepts, the seismic process in Fennoscandia is affected by at least four force mechanisms: (1) movement of the lithospheric plate under the Norwegian and Barents seas from northwest to southeast due to the spreading of the Mid-Atlantic Ridge in the area from Iceland to Svalbard; (2) postglacial isostatic uplift; (3) present local neotectonic movements; and (4) gravitational bending deformations on the continent–sea shelf contact along the coast of Norway due to intense erosion from the rising crystalline domain of the Baltic Shield.

Many authors prefer the first mechanism (Arvidsson 1996; Bungum et al., 1991; Byrkjeland et al., 2000; Craig et al., 2016; Gregersen, 1992, 2006; Hicks et al, 2000; Slunga, 1981, 1984, 1989; Wu et al., 1999). This is justified by observations of the prevailing subhorizontal compression in the northwest–southeast direction according to the mechanisms of earthquake foci within the Norwegian Sea shelf as well as on the continental part of Fennoscandia in the extreme northwest—in northern regions of Sweden and Finland and in southern Finland (see Fig. 5). At the same time, some features disagree with these notions (see Fig. 5). In addition, the fact that the well-known strong paleoseismic events (unparalleled in the historical and present seismicity of Fennoscandia) are confined to the final phases of postglacial uplift raises some doubts about the ubiquity of the action of the first mechanism in Fennoscandia.

Nevertheless, many authors associate the well-known system of Holocene thrust faults in northeastern Fennoscandia with a total length of several hundred kilometers and propagating from northeast to southwest (see Fig. 4) to this effect of compression stresses caused by the movement of the lithospheric plate under the Norwegian and Barents seas from northwest to southeast (Craig et al., 2016; Gregersen, 1992, 2006; Olesen et al., 2013a; Wu et al., 1999). This observed discrepancy in the link between seismicity and the proposed force mechanism of the effect of the moving oceanic plate on the stable Fennoscandia massif requires an explanation. We will do this later.

The peak of present seismic activity of Fennoscandia is observed in southwestern Norway (in the North Sea). It is primarily associated with the big shear zone in a decayed structure of the graben (Song Graben) stretching from the Norway coast towards the North Sea in the southwest direction at a distance of almost 600 kilometers and in a long graben-like structure (Norwegian Channel) branching off from it in the south along the southwestern coast of Norway (Hicks et al., 2000; Olesen et al., 2013a, 2013b). At present, relatively strong earthquakes with M = 4–5 occur here regularly.

Unlike the above-mentioned prevalence of subhorizontal compression on the Norwegian Sea shelf, there is a region of stresses of subhorizontal expansion in the near-meridian direction according to the focal mechanism of earthquakes (Fjeldskaar et al., 2000; Hicks et al., 2000; Keiding et al., 2015) (see also Fig. 5). It is difficult to explain what causes this expansion, but it cannot be associated with regional subhorizontal compression; this expansion most likely has a local tectonic origin. It should also be noted that the Earth’s crust is significantly reduced here (Hyvonen, 2008), which normally can contribute to the rift formation as a manifestation of independent tectonics. At the same time, a number of researchers (for example, Hyvonen (2008); Olesen et al. (2013a); Solheim et al. (2005)) assume that this region emerges from a rapid postglacial uplift that led to the “burst” of strong earthquakes actuating this giant underwater fault structure.

The relatively high seismic activity along the Norwegian middle coast in Nordland (see Fig. 5) is interpreted by some authors as a vivid manifestation of the dominant role of postglacial isostasy in the formation of seismicity in this region (Fjeldskaar et al., 2000; Keiding et al., 2015; Olesen et al., 2013a) due to the high gradient of the postglacial uplift (see Fig. 2). This gradient is aggravated by the fact that this region seems to have an independent neotectonic uplift, which can be seen from the observed significant positive deviations from the computational model of postglacial uplift (Fjeldskaar et al., 2000) (see Fig. 3). According to the focal mechanism of earthquakes, subhorizontal strain across the coast also prevails here, leading to the prevalence of fault rupture in the coastal zone of Nordland (see Fig. 5). The relative increase in seismicity on the offshore shelf was explained by supposing that this increase may be associated with bending deformations caused by the present redistribution of sediments due to Pliocene erosion of the actively rising coast of Nordland (Olesen et al., 2013a).

These two reasons (namely, postglacial isostasy and bending deformations) can be regarded as the driving mechanism of seismicity observed in the Nordland region. In addition, this region is characterized by deep fjords and very steep slopes, which makes gravitational stresses probable due to the high contrast of the topography and increases the level of seismicity (Olesen et al., 2013a).

A number of studies discuss the existence of postglacial Holocene tectonics in Fennoscandia and its possible link with seismicity. The most significant manifestation of present tectonics can be the domains of large compact deviations from the computational model of postglacial isostasy (see Fig. 3). The domain of positive deviations has been considered above. The domain of negative deviations falls on the Swedish coast of the Gulf of Bothnia of the Baltic Sea, where, unlike the Norwegian shelf, there is no Pliocene erosion, but there is still a slight increase in seismic activity (see Fig. 3). This increase in seismicity, which is not present in the zones of maximum glaciation in Arctic Canada, where seismic activity is almost completely absent in the central part of the postglacial dome, was noted by G. P. Avetisov (1995). In this regard, he suggests that either in one of these cases or in both cases, glacioisostasy does not play a decisive role in the formation of present high seismicity.

According to the focal mechanism of earthquakes (see Fig. 5), tensile stresses across the coastline of the Gulf of Bothnia prevail. A considerable number of weak seismic events are observed here, predominantly in the upper 10–15 kilometers of the Earth’s crust with a fault movement in the source (Hicks, 1996; Hyvonen, 2008; Slunga, 1981, 1984, 1989). This territory can therefore be considered prone to present fault tectonics.

Fjeldskaar et al. (2000) suggested that there should be other local regions (tens/first hundreds of kilometers in size) with significant differences between the observed and calculated rises that are now subject to vertical deformations of neotectonic rather than glacial isostatic origin. These regions can be associated with local manifestations of dispersed microseismicity.

In the central part of Fennoscandia, which is characterized by the lowest level of seismic activity or even its complete lack locally, it is difficult to identify a prevailing mechanism for the formation of seismicity. This is also evidenced by the large uncertainty in the orientation of the stress axes in the few focal mechanisms. The analysis performed in (Keiding et al., 2015; Muir-Wood, 2000) shows that the tensile stresses formed in the northwest–southeast direction due to postglacial uplift reduce the effect of compressive spreading stress in the same direction. This leads to a low differential stress and, as a consequence, to low seismicity and very different focal mechanisms in the central mainland of Fennoscandia (Hyvonen, 2008).

At the same time, it should be noted that the seismicity is also affected by conditions other than the stress field, such as the lateral inhomogeneities in the Earth’s crust and the existence of weak zones and faults with orientation to the direction of the prevailing stress field in their vicinity.

Nevertheless, the most clearly expressed feature of seismicity in Fennoscandia is the presence of many postglacial paleoseismic dislocations that are well manifested in the topography. They are most reliably identified in northern Fennoscandia: in northern Finland (Kujansuu, 1972), in Sweden (Mörner, 2005), in Norway (Olesen et al., 2013b), and on the Kola Peninsula (the Russian Federation) (Nikolaeva, 2013; Nikolaeva and Evzerov, 2018; Nikolaeva et al., 2018; Nikonov and Zykov, 2017; Nikonov et al., 2018); however, they can also be found (though not so abundant nor of such gigantic sizes) in the central and southern parts of all the Scandinavian countries listed above as well as in the eastern part of Fennoscandia: in Northern (Gorbatov et al., 2017) and Middle (Nikonov et al., 2017; Shvarev and Rodkin, 2017) Karelia and in the southern border of the Russian part of the Fennoscandian shield in the Leningrad region (Nikonov et al., 2014; Poleshchuk et al., 2018; Shvarev et al., 2018a, 2018b; Subetto et al., 2018). A.A. Nikonov and S.V. Shvarev (2015) identified 12 seismo-lineaments in the Russian part of Fennoscandia associated with paleoearthquakes; a largely extended version of the map of seismo-lineaments supplemented by the results of field studies in 2018 can be found in (Erteleva et al., 2018). The related problems of seismic hazard assessment taking into account rare events are discussed in (Sidorin, 2018).

The palaeoseismic dislocations discussed here are induced by postglacial thrust faults (see Fig. 4) normally directed from north–northeast to south–southwest, which were formed by 9000–11 000 BP (Arvidsson 1996; Gregersen, 2006; Gregersen et al., 1991; Johnston, 1989; Keiding et al., 2015; Muir-Wood, 2000; Wu et al., 1999). These faults have a length of up to 150–160 kilometers and vertical shifts up to 15 meters with a secondary shear component and can possibly break the entire upper crust (Arvidsson, 1996; Wahlstrom, 1989). According to estimates, the formation of these faults could cause a strong earthquake with a magnitude of M = 7–8 (Arvidsson 1996; Mörner, 2017; Muir-Wood, 1989).

On the territory of Sweden, there were a total of 56 paleoearthquakes, 26 of which with an estimated magnitude of M = 6–7, 17 events with M = 7–8, and six events with a magnitude even exceeding 8 (Mörner, 2005, 2017; Mörner et al., 2008). Using the method of seasonal stratification of soil sediments to estimate the age of these paleoseismic dislocations, which often makes it possible to find the time of their formation accurate to a year, it was shown that 41 paleoearthquakes occurred 7000–13000 BP—during the most active phase of deglaciation. These include the strongest events (Mörner, 2017; Mörner et al., 2008). Based on this finding, Mörner (2005) suggested that the estimated magnitude of earthquakes in Fennoscandia dramatically decreases from eight or more for paleoearthquakes in the given time period to six for tectonic earthquakes in the historical time period.

Thus, Fennoscandia seems to be significantly more seismically active (with seismic events with a magnitude of up to 8) in the late glacial age (Arvidsson 1996; Mörner, 2005, 2017; Mörner et al., 2008; Muir-Wood, 1989; Wu et al., 1999). High erosion activity on the surface of the rising crystalline Precambrian basement of the Fennoscandian domain led to a solid exposure of these paleoseismic dislocations, which contributes to their reliable identification (Arvidsson, 1996; Erteleva et al., 2018; Mörner, 2005, 2017; Mörner et al., 2008; Nikonov and Shvarev, 2015; Nikonov et al., 2014; Wahlstrom, 1989). Thus, there is no doubt about their existence. The time of their occurrence and the method of destruction indicate that postglacial stresses can be important in the initiation of faults and relevant earthquakes in the late-glacial and early postglacial times. However, this does not resolve the question of whether postglacial stresses can cause seismicity in Fennoscandia at present as well. In this regard, it should be noted that weak seismicity on the Karelian Isthmus has become more active in the past few years, which reflects the current significant change in the stress state of the region (Assinovskaya et al., 2018).

In our opinion, the above analytical review of various points of view on the nature of seismicity of Fennoscandia is rather convincing to make one believe that glacioisostatic movements cannot be a significant seismogenic factor in the given region at present. G. P. Avetisov (1995, 1996a) argues in favor of this point of view by the fact that, according to current understanding, the deglaciation of Fennoscandia, which was responsible for a high level of seismicity at that time, terminated no later than 9000–9500 BP. According to these views (Avetisov, 1996b) the peak rates of postglacial uplift, which is assumed to be associated with the observed paleoseismic dislocations, occur generally no later than 1000 years after a complete deglaciation and decrease exponentially with time; here, a sharp decrease in the rates occurs in 1000–1500 years after the peak. The first 1000 years involved at least one third of the total uplift. It is during this time period that glacioisostasy can play a leading role in seismogenesis. A similar point of view is also expressed by the authors of (Bungum and Olesen, 2005; Craig et al., 2016; Gregersen, 2006; Quinlan, 1984; Steffen and Wu, 2011; Steffen et al., 2014a, 2014b).

It seems that the seismic activity in the postglacial cycle is drastically reduced compared to the seismicity caused by common tectonic processes in the seismic cycle. This is consistent with the assumption that the seismic activity of Fennoscandia is associated with two processes—postglacial uplift and common tectonics (Fjeldskaar et al., 2000; Mörner, 1979). In this case, the intensity of postglacial uplift decreases exponentially with time from one glaciation to another, while the intensity of common tectonics varies linearly in the seismic cycle. Since the postglacial cycles cover the first tens of thousands of years, the consequences of their seismicity at a sufficient distance from the beginning of the cycle in Fennoscandia (i.e., at present) are little observed. In view of this, there is no reason to significantly increase the seismic hazard in the Fennoscandian regions that are characterized by large paleoseismic dislocations. Seismic events similar to them are likely to be expected only after a new glaciation at the stage of postglacial uplift.

Another argument in support of the above point of view is the fact that the seismic activity in the rigid crust during the peak phase of glaciation is largely reduced or even completely disappears and, conversely, it is sharply activated during strong melting of glaciers. Seismicity is almost completely absent at present under the powerful ice sheets of Greenland and Antarctica (Gregersen, 2006; Johnston, 1987; Munier and Fenton, 2004; Wu and Hasegawa, 1996). At the same time, when strong melting of ice occurs in these regions, seismicity is activated along margins of the isostatic dome, rather than in its the central part (Dahl-Jensen et al., 2010; Gregersen, 2006; Larsen et al., 2006; Sauber and Molnia, 2004; Voss et al., 2007).

This can be caused by an increase in the vertical component of stresses (σmin for tectonic regions with a predominance of thrust formation) due to ice load, which leads to a decrease in differential stresses (σmax–σmin), which is responsible for the occurrence of earthquakes. This load itself is apparently insufficient to overcome the strength of solid rocks under the ice sheet, which, in turn, prevents the occurrence of earthquakes with fault movements in the source (Quinlan, 1984).

Both of these factors determine the almost complete absence of earthquakes under ice sheets. In this case, the continued growth in the horizontal component of tectonic stress σmax most likely lags behind the growth of the vertical component σmin due to slower growth rates of tectonic stresses in comparison with the ice-load growth rate, which is determined by secular cycles of solar activity rather than by geological times. Still, this contributes to the reduction of differential stresses (σmax–σmin), which prevents the activation of seismicity.

At the final stage of deglaciation, which proceeds rather quickly in geological terms (no longer than 2000 years according to (Steffen et al., 2014a)), the value of σmax decreases sharply due to the ice load removal, and combined with a slightly increased value of σmax during the ice cycle, this factor leads to a significant increase in differential stresses (σmax–σmin) and, as a result, to a sharp activation of seismicity as thrust movements observed in Fennoscandia as large paleoseismic dislocations, which possibly renew them at earlier postglacial stages. The calculations of the effect of deglaciation on seismic activity showed that deglaciation is capable of causing earthquakes in already pre-stressed regions with a type of movement in accordance with these stresses, but cannot dictate an independent type of movements in the foci of these earthquakes (Quinlan, 1984; Steffen et al., 2014a, 2014b). Thus, the observed thrust type of movements in paleoseismic dislocations (see Fig. 4) should be regarded as a “legacy” of the long-term accumulation of horizontal stresses σmax due to the movement of lithospheric plates.

Similar reasoning can be found in (Craig et al., 2016) and in more detail in (Steffen and Wu, 2011; Steffen et al., 2014a, 2014b) with a demonstration of changes in differential stresses using Coulomb–Mohr circular diagrams. It is thus recognized that a tectonically stable continental lithosphere can retain elastic deformation for a long time, while its implementation may be associated with rapid local transient changes in stresses, which are caused, in particular, by erosion, fluid migration, or ice unloading (Craig et al., 2016). This should lead to periodic intraplate and primarily postglacial seismicity on continental shields prone to cyclical glaciations, where the rates of tectonic loading are relatively low for short-term transient processes (in our case, this is a compression load due to a northwest-to-southeast movement of the lithospheric plate under the Norwegian Sea).

The analysis performed here allows us to recognize that the abnormally strong burst of seismicity in Fennoscandia and in its eastern surroundings, which occurred in the final phase of deglaciation at around 10 000 BP and involved a series of strong earthquakes with M = 7–8 on scarp faults of up to 155 kilometers in length with similar phenomena of landslides, soil liquefaction, shaking structures, etc., as markers of their seismicity (Gregersen, 2006; Mörner, 2005, 2017; Olesen et al., 2013a) was most likely caused by a combination of the following three factors: (1) a decrease in the crust weight due to melting of the glacier and, accordingly, a decrease in the vertical component of stresses σmin; (2) high pore pressure due to the penetration of melt water into the rock; and (3) long-term tectonic compression (σmax) due to the lithospheric plate subduction from the northwest caused by spreading in the Mid-Atlantic Ridge.

Similar views were formulated earlier in (Keiding et al., 2015; Lund et al., 2009; Muir-Wood, 1989). In this case, the simulated behavior of seismicity associated with glaciations showed that seismicity activation is related to the end of deglaciation and quickly decays after one or two strong earthquakes with a thrust type of the focal mechanism (Steffen et al., 2014a, 2014b). Here, the magnitude of these earthquakes is primarily affected by the thickness of the crust and lithosphere, while the ice cover size is determined only by the time of seismicity activation (Steffen et al., 2014a, 2014b).

Most likely, this seismic activation accompanied each phase of deglaciation after several glaciation cycles in mainland Fennoscandia during the last 600 000 years. There is still no clear understanding of the transition from high postglacial seismicity of 10 000–6000 BP to low present seismicity. A number of researchers (for example, Avetisov, 1995, 1996a, 1996b; Bungum and Olesen, 2005; Craig et al., 2016; Gregersen, 2006) believe that the present seismicity is largely associated with present tectonic processes, rather than with the effect of glacioisostatic changes. However, a different point of view can also be expressed.

The relatively low level of present seismicity, primarily in the central part of Fennoscandia, testifies that, unlike the situation that was prevalent immediately after the deglaciation, the stresses of present postglacial uplift play a rather inhibitory role in the occurrence of strong earthquakes. Indeed, the analysis performed by Keiding et al. (2015) for the range of horizontal deformation velocities of the daytime surface (up to 4 × 10–9 years–1) over a wide range of GPS measurements showed that the present postglacial uplift leads to the formation of a field of horizontal strain across the dome line. In turn, this affects the present state of stresses in Fennoscandia by reducing the tectonic compression stress (in our case, σmax) due to the northwest-to-southeast movement of the lithospheric plate of the Norwegian and Barents seas and leads to a decrease in the differential stress (σmax–σmin) and, consequently, to low seismicity in the central mainland of Fennoscandia.

One way or another, the complete deglaciation of Fennoscandia occurred no later than 9000–9500 BP which makes it possible to conclude that the glacioisostatic movements in most of its territory is insignificant at present in the formation of seismogenic stresses.

CONCLUSIONS

During seismic observations in Fennoscandia, not a single earthquake with M > 6.5 occurred on its territory. Although the first surviving reports on seismic tremors date back to 1073 (Mantyniemi et al., 2004), only one seismic event occurred in the Kandalaksha graben in the White Sea in 1627 and is assumed currently to have had a magnitude M > 6.5 (Nikonov, 2004). The frequency of seismic events in Fennoscandia constructed on the basis of all documented historical earthquakes (here, this strongest earthquake is assigned a magnitude of 6) indicates that an earthquake of magnitude ~6 should be expected here approximately once per 100 years (Bungum et al., 2000; Olesen et al., 2013a).

At the same time, a significant number of Pleistocene and even Holocene age paleo-seismic dislocations have been reliably found in the given region; due to their sizes, these dislocations can be associated with the occurrence of strong earthquakes of magnitude 7 or even 8 in the past. In view of this information, one has to be careful in dealing with the underestimated seismic hazard of this territory only on the basis of some instrumental seismic data. This factor can be taken into account by increasing the value of Mmax in the vicinity of well-documented paleoseismic dislocations in accordance with their sizes.

If one accepts the hypothesis that these paleo events occur only at the final stage of the glacial age during the intense postglacial isostatic upwelling of the Fennoscandia domain and their activity exponentially decays in time, the frequency of these paleo events can be estimated to be tens of thousands of years based on the time interval between adjacent glaciations. Therefore, it is hardly reasonable at present to expect a strong earthquake caused by the discharge of stresses in the Earth’s crust during its isostatic uplift.

In other words, it makes sense to recognize that the nature of present seismicity of Fennoscandia is determined by tectonic stresses caused by the global impact of the northwestern subduction of the lithospheric plate under the Norwegian Sea (a constant source for the accumulation of tectonic stresses), by local tectonic uplifts (in Nordland) or subsidences (on the Swedish coast of the Gulf of Bothnia), and by local processes of rift formation in the southern part of the North Sea, rather than by postglacial isostatic stresses It is this tectonic regime that determines the character of the regional seismicity, including the peak magnitude of possible earthquakes and their recurrence.