Seawater Intrusion on the Arctic Coast (Svalbard): The Concept of Onshore-Permafrost Wedge

: Numerous hydrogeological studies on the coastal zone describe the intrusion of sea water inland, salting underground aquifers. The phenomenon is commonly observed in the coasts outside polar areas. However, the impact of sea water has so far not been an object of detailed investigation in a periglacial environment devoid of subsea permafrost. Geophysical measurements at the west coast of the Wedel-Jarlsberg Land in Svalbard indicate that the border between the unfrozen seabed and the frozen ground onshore is not delimited by the shoreline. A zone of coastal unfrozen ground is located under a thin layer of permafrost reaching toward the sea. This state was observed with the use of electrical resistivity tomography under rocky headlands and capes, uplifted marine terraces located at the foot of mountain massifs and valley mouths as well as in the marginal zone of the Werenskiold Glacier. This short article presents the results of such a measurement, supplemented with electromagnetic detection. The measurements are unique in that they were conducted not only on the land surface, but also at the ﬂoor of the sea bay during the low water spring tide. The author proposes name structures detected in the coastal zone as a “permafrost wedge”, extending an identiﬁcation of the permafrost base between the coast and the glaciers of Svalbard. However, in the absence of boreholes that would allow determining the thermal state of the ground in the study sites, the concept is based only on the interpretation of the geophysical imaging. Therefore, further discussion is required on whether the identiﬁed contrasts in electrical resistivity indeed result from thermal di ﬀ erences between the rocks or if they only indicate the cryotic state of the ground (saline cryopeg) within the range of seawater intrusion.


Introduction
Hydrogeological investigations performed in a number of locations worldwide confirm the strong impact of the sea water percolating deep in the ground toward inland due to the difference in density between salt and fresh groundwater [1][2][3][4][5][6][7][8][9][10]. In the coastal zone, sea water forms an intrusion propagating under the non-saline rock layers just below the ground surface. This type of contact zone was been described among others on carbonate islands and coasts [11,12]. Saline water is found in boreholes and may be well recognized with the use of geophysical methods [12][13][14][15][16][17][18][19][20].
These brief descriptions allow for an assumption that the impact of saline water should be analogical along the coast in the periglacial zone, where permafrost is developed onshore. An example can be found on the west coast of Spitsbergen, the largest island in the Svalbard archipelago. Despite a good understanding of the characteristics of the local permafrost obtained from numerous boreholes and measurement points [21,22], the question of its evolution in the contact zone with the sea remains open. The problem has not been solved by hydrogeological observations either, and examinations of water flows under the permafrost have not focused directly on the coastline [23][24][25]. However, the possibility of saline water intrusions and heat wave from the sea was indicated by the geophysical measurements performed by the author in the years 2012-2017 in Wedel-Jarsberg Land between Skoddebukta (bay) in the north and Hornsund (fjord) in the south [26][27][28]. The sea water impact was detected in the grounds of each of the investigated coastline sections with various lithologies (schists, quartzites, marbles) and morphologies (under rocky headlands and capes, uplifted marine terraces at the feet of mountain massifs and valley mouths as well as in the marginal zone of the Werenskiold Glacier, see sites 1-3 in Figure 1). This has also been confirmed in recent research conducted in Svalbard by other authors, who observed water flow within the permafrost [29] and thawing of deeper ground layers in contact with saline water [30]. Similarly, changes in the geophysical properties of deeper ground are known from the coast of West and North Greenland [31] or the Canadian territory of Nunavut [32].
This article aims at both proposing a model of seawater impact on the coasts of the High Arctic and initiating a discussion. The model applies to regions with no subsea permafrost presence or where it lies deeper under the seabed, which would limit the percolation of saline water into the ground [33][34][35]. Although the assumptions for the model have already been developed [26] and are based on the above-mentioned research, this article presents the results of an additional, representative geophysical imaging. It is unique in that it jointly shows the cross-section of the part under the floor of a marine embayment and the adjacent onshore part. The measurement was possible when the sea floor was exposed during the low water spring tide in Nottinghambukta ( Figure 1). Finally, it is possible to show the shape of the permafrost from the shoreline to the fronts of glaciers filling the Svalbard valleys, completing the known schemes [36,37].  Figures 2 and 3, ERT-electrical resistivity tomography section presented in this study, 1-location of near-surface geophysical imaging of schist bedrock published in [28], 2-location of near surface geophysical imaging of quartzite and marble bedrock published in [26], 3-location of near surface geophysical imaging of marble and schists bedrock published in [27]. The figure is based on Norwegian Polar Institute digital data.
The study area is a typical example of a rocky coast in Svalbard, free of glacier ice and subjected to the influence of relatively warm sea water. The investigated bedrock is formed of mica schists and quartzites [38]. The coastline is in the form of a rock cliff, with uplifted marine terraces of Kvartsittsletta above. In the north, the cliff ends at the mouth of the Brattegg Valley, in the vicinity of the Stanisław Baranowski Polar Station. Near-surface parts of the ground are thermally monitored in that area [39].
The measurements revealed a deep (1.5-4 m) permafrost active layer, with the maximum thawing reached in mid-August.

Methods
The properties of the bedrock in the coastal zone were identified with the use of electrical resistivity tomography (ERT). This method is commonly used to detect the presence of permafrost and various types of ground ice (e.g., [40][41][42][43][44][45][46][47]). ERT, which is basically a combination of electrical resistivity sounding and profiling, consists of the measurements of resistivity (R) in a number of four-electrode arrays, in which electric current (I) is transmitted to earth by two electrodes (C 1 , C 2 ), and voltage-potential difference (V) is measured by the second pair of electrodes (P 1 , P 2 ). As the rock medium is not a uniform body, the measured resistivity, expressed as the ratio between voltage and current with allowance for coefficient (k), which depends on the electrode array and the distances between the electrodes, is apparent resistivity (ρ app ). Good vertical resolution and depth penetration were simultaneously obtained with the use of the Wenner-Schlumberger electrode array [48][49][50]. The device used was ARES (GF Instruments, Brno, Czech Republic) with a set of 88 electrodes. The results of the ground resistivity measurements (expressed in Ω·m) were subjected to standard geophysical interpretation (inversion) in RES2DINV software (Geotomo, Malaysia). The procedure was based on default smoothness-constrained inversion formulation [51].
The measurement was performed on 31 July 2017 during the low water spring tide when the sea water exposed the floor of a tidal flat in Nottinghambukta ( Figure 2). The measurement section was 995 m in length, obtained using the roll-along technique including the sea floor (the first 200 m) and continued on the surface of the uplifted marine terraces toward the mouth of the Brattegg Valley, perpendicular to the shoreline. The total number of obtained data points was 5931 and the inversion model had 16 layers and 2757 blocks.
Due to the 5 m spaces between the electrodes, the inversion model could not include the superficial layer, which was therefore additionally examined with the use of electromagnetic induction (EMI) [52,53]. This method allowed one to obtain images of the permafrost active layer [54]. The measurements were carried out with an electromagnetic conductivity meter CMD-Explorer (GF Instruments, Brno). The device is a portable tube that contains both transmitter and receiver coils inside at dipole center distances of 1.48, 2.82, and 4.49 m, which allow for measurements of apparent electrical conductivity/resistivity with the maximum sampling rate of 10 Hz (the effective depth ranges were at 1.1, 2.2, and 3.4 m simultaneously). The measurement results were presented graphically with the use of digital data from the Norwegian Polar Institute, processed in Metashape v 1. The analysis of the measurement results was based on the fact that rocks cooled to temperatures below 0 • C drastically reduce their ability to conduct electric current [50]. The published interpretations (e.g., [42,43,54,55]) support an assumption that resistivity approximating or exceeding 1 kΩ·m may indicate frozen ground (Figure 3), although the literature gives very different threshold values depending on local ground features including groundwater salinity.

Results
The inversion model resulting from the ERT measurements indicates highly contrasting ground resistivity (Figure 4a). The similarity to the results obtained in the geoelectric measurements cited above allowed for a tentative assumption that this contrast was due to the negative temperatures of the rocks and ground ice presence onshore. The ground under the floor of Nottinghambukta showed low resistivity values (ρ from 1 to 500 Ω·m), which practically eliminates the possibility of ground ice presence in this zone. The rocks under the terrestrial part showed extremely high resistivity values (ρ > 10,000 Ω·m), which are characteristic of permafrost. These values indicate practically no electric conductivity and mask all local lithological differences. In this zone, the near-surface ground layer also showed lower resistivity values, which were related to the permafrost active layer and penetration of fresh groundwater. This phenomenon is particularly well visible at about 820-890 m of the ERT section. The area is located at the foot of a mountain slope, which supplies meltwater. Here, the ground thawing in summer, identified with the use of the above method, can be assumed to reach at least 10-20 m under the surface. Due to the electrode spacing, the inversion model does not cover the uppermost~2.5 m of the ground. This gap is partially filled by the results of the EMI measurement (Figure 4b). Based on the obtained data, the permafrost active layer on land can be assumed to reach a typical depth of at least 2-3 m. No high electrical resistivity was observed below the bottom of the bay, so was therefore interpreted as free of ground ice.
The most interesting element of the geophysical imaging is the transition zone between the extreme values obtained for offshore and onshore parts of the coastal zone. This was not only seen on the ERT inversion model, but also very well reflected in the ERT measured apparent resistivity pseudosection (Figure 4). At the ground surface, it coincides with the shoreline, but at greater depth, it is shifted inland by approximately 60 m. The high resistivity body thus assumes a wedge shape, resembling a clip-point type knife blade. In the inversion model, the transition zone below the wedge has the shape of a near-vertical wall (resembling a cliff line), although the analysis of the apparent resistivity suggests that it may just have the shape of a cavity retreated toward the land.

Discussion
The combined results of the ERT and EMI measurements allowed for better geophysical imaging of the coastal bedrock than the previous geophysical measurements in the Wedel-Jarlsberg Land [26][27][28]. Extending the measurement section along the exposed sea floor of the bay proved to be of key importance. As was the case in the previous geophysical images, the ERT inversion model demonstrated high electrical resistivity contrast of the ground between the land and the sea. Low resistivity obtained for the latter area, typically not exceeding 100 Ω·m, suggests that the investigated parts of the seabed are not frozen. The extremely high values, which continue regardless of the varying lithology, indicate the presence of permafrost, although the literature review clearly refutes the possibility of finding a limit value for the presence of permafrost with spatially varying environmental factors. There are also examples of resistivity much lower than 1 kΩ·m found for permafrost (e.g., [47]).
This article indicates that the border between high and low electrical resistivity values of rocks in the coastal zone has a specific shape. At greater depths, the low-resistivity body penetrates inland from the sea side and shows as seawater intrusion ( Figure 5). The high-resistivity body at the surface forms a wedge, which is attached to the land side and diminishes in thickness toward the sea. Unlike saline intrusions observed at lower latitude coasts, the landward impact of sea water is limited by the presence of permafrost. An assumption can thus made be that in the west coast of Svalbard, and probably also in other sections of polar coasts subjected to neotectonic uplift and lacking subsea permafrost, or where subsea permafrost is found at greater depths below the seabed [58,59], the near-sea terrestrial part of permafrost assumes a characteristic shape that will be referred to here as a "permafrost wedge". Below it, the contact between rocks having different thermal properties most likely assumes the shape of a cavity. The described observation resembles ground-temperature distribution and a tongue of developing permafrost in front of the retreating glacier [36,37]. Such an analogy makes it possible to draw a simple diagram suitable for the coasts of Svalbard, in which the permafrost, bounded on one side by the glacier and on the other side by the sea, take a similar shape ( Figure 6). More complicated is the form of the interface between the frozen and unfrozen ground. This must be set by the diffusion of heat landward from the sea-covered ground and have an impact on the properties of permafrost far inland, which can be indirectly seen in the ERT measured apparent resistivity pseudosection (Figure 4).   [36,37], (b) based on geophysical imaging; P-B AL-permafrost-base active layer (?); vertical dashed line drawn in permafrost to distinguish ground with different apparent resistivity (reference to the ERT measured apparent resistivity pseudosection in Figure 4).
The presented interpretation is tentative, as the results of the geoelectric measurements can only be treated as indirect evidence of rock temperatures. At this stage of research, prior to thermal measurements in boreholes penetrating through the "permafrost wedge", the possibility still exists that the seawater intrusion only causes a change in the salinity of waters in the rock medium of the land part of the coast. With higher salinity, the ground can be unfrozen even at temperatures of −2 to −3 • C [60,61]. Measurements of temperature and chemical/structural properties of saline soils [62] demonstrated a shift of the phase transition of groundwater on average from −0.4 to −1.6 • C (frozen) and from −0.6 to −3.5 • C (thawed) for soils having a low degree of salinity as well as −1.2 to −21.8 • C (frozen) and −4.9 to −41.0 • C (thawed) at a high degree of salinity. In the bays of the Greenland Sea, in the Hornsund zone, at depths greater than 50 m, seawater salinity is in the range of 34.7-34.9 PSU (practical salinity unit, based on the properties of seawater conductivity, PSU salinity % ) [63,64]. At the bottom of Skoddebukta (bay), which neighbors the study area, the measured salinity was 33.5-34.5 PSU [65,66]. For comparison, the mineralization of water flowing from the springs located in front of the Werenskiold Glacier typically does not exceed 75 mg·L −1 and may exceed 200 mg·L −1 only during periods of intensive thawing processes [67,68]. In this hypothetical situation, lower freezing point would cause the underwaters within the marine intrusion to still have a sub-zero temperature, while the ground would be in a cryotic condition as in marine cryopegs.
The interpretation presented here, however, is supported by the measurements of seawater temperature. In the near-surface zone in the winter, the water temperature decreases to approx. −2 • C, but in the summer, it warms to +3 to +6 • C [64]. An even higher seawater temperature was measured by the author with the use of a submerged HOBO temperature datalogger in Hyttevika (bay), which is protected by skerries from the open sea. On 11 August 2017, the temperature was 7.8 • C, while the temperature of the permafrost in Svalbard, as given in older studies [69] and measured from boreholes at a depth of 30 m, was −3.0 • C and −2.4 • C at the mineralization of aquifers in the range of 34-44 mg·L −1 . Seawater temperature variations of this scale lead to the conclusion that the bottom part of the "permafrost wedge", together with the rest of the permafrost contact zone subjected to annual seawater temperature variations, forms a permafrost-base active layer. This would be consistent with observations of seasonal active layer associated with subsea permafrost found in sediments of nearshore shallow waters in Prudhoe Bay in the Beaufort Sea in Alaska [70].

Conclusions
The summary of the data presented in this article can be listed as three main points. Unique measurements using the electrical resistivity tomography performed on the floor of the sea bay and at the same time on the land in the area of Spitsbergen west coast confirmed the results of earlier geophysical imaging in this zone and the presence of seawater intrusion reaching inland.
Seawater intrusion inland causes the formation of structures referred here to as a "permafrost wedge". The "permafrost wedge" is the near-surface part of the land-side permafrost. It can be 60 m in length and 20 m in thickness, decreasing toward the shoreline. These values, however, depend on the morphology of the coast. The "permafrost wedge" resembles ground-temperature distribution and a tongue of developing permafrost in front of the retreating glacier.
No conclusive evidence has been found on whether the seawater intrusion causes the increase in the temperature of the deeper zones of the ground on the coast or just a phase change related to the shifting of the freezing point due to salinity. The proposal that the zone actually thaws is, however, supported by high seawater temperatures in the summer, which temporarily exceed 6-7 • C. In addition, annual seawater temperature variations suggest that the bottom part of the "permafrost wedge" and the entire contact zone below are annually subjected to changes in the thermal state, forming a permafrost-base active layer. This hypothesis will need to be verified by temperature measurements in boreholes penetrating the "permafrost wedge".