Conceptual Model of Permafrost Degradation in an Inuit Archaeological Context (Dog Island, Labrador): A Geophysical Approach

The analysis of changes in the inverted electrical resistivity profiles attests to several tendencies, resulting in four units with different characteristics (Figure 5): 1. a relatively low-resistivity unit from the top of the profiles to a depth of about 1 m (U1—300 Ω·m to 1000 Ω·m); 2. a moderately resistive unit (U2—1000 Ω⋅m to 25k Ω·m); 3. a highly resistive unit (U3—>25 kΩ⋅m) usually found at a depth of about 1 m; and 4. a low-resistivity unit (more conductive) (U4—Figure 5A,D,E) had resistivities of up to 56 kΩ·m, 56 kΩ·m, and 72 kΩ·m k, respectively.

Unit U1 corresponds to the thawed active layer, composed of a peat layer overlying humid sand. This conductive zone’s thickness ranges from 50 cm to 1 m. An abrupt transition from a few hundreds of Ω⋅m to thousands of Ω·m was detected between U1 and U2 (Figure 5) on all profiles. This transition between U1 and U2 is interpreted as the limit between the active layer and underlying frozen ground, which was confirmed by at least one on-site test pit. The depth of this sharp increase in resistivity depends on the terrain slope and the proximity to thermokarst features at the surface. An approximative minimum electrical resistivity value for permafrost in each profile was determined and later used for the calibration of the GPR interfaces. U2 presents resistivities ranging from a few thousand Ω·m up to 25k Ω·m and is therefore interpreted as frozen sand, as was confirmed by direct observations in test pits [53]. An exception can be made for the test pit near Profile 2, where no permafrost was found even though a U2 zone was identified on the profile. This can be explained by the distance between the profile and the test pit, as the latter was 3 m closer to the coast. This underscores the high variability in permafrost distribution at the site, even on a small scale.

U3 is a highly resistive unit (U3—>25 kΩ⋅m) found beneath U1. The vertical and horizontal extent of unit U3 is heterogeneous from profile to profile and inside each profile. The very-high-resistivity zones in Profiles 1, 4, and 5 (Figure 5A,D,E) could indicate a high ground-ice content in the sediments, or massive ice. Massive ice has been known to have resistivity values ranging from 10k Ω·m to 40k Ω·m and more [53,54], but its presence could only be confirmed by drilling. Some profiles had no resolution beneath U3, which can be due to the umbrella effect of highly resistive permafrost. Indeed, some zones and depths under highly resistive structures cannot be assessed with electrical measurements from the surface [41,55]. In Profile 1 (Figure 5A), a thickness gradient of unit U3 from inland (north) to the coast (south) is present; residual highly resistive zones become thinner when located closer to the coast. A comparison of the thickness of U3 between Profile 2 (Figure 5B) and Profiles 4 and 5 (Figure 5D,E) also illustrates this gradient. U3 is absent in the central depression of house H3, and the lower-resistivity zones U1 and U2 expand downwards (Figure 6). This can be explained by the insulation from the cold being provided by snow accumulation in the depression during the winter. Profile 5 (Figure 5E) presented a U3 zone between 20 and 25 m, where the electrodes were planted in the midden of house H3. The high resistivity in this zone could be due to the presence of frozen sediments, frozen cultural material (faunal remains, charcoal, etc.) [29], or buried resistant materials.

Unit U4 has resistivity values similar to U1, indicating the absence of permafrost (Figure 5). The conductive zone of U4 is interpreted as being either taliks or the water table. Here, taliks refer to regions of perennially unfrozen ground surrounded by permafrost bodies, as well as regions in proximity to permafrost bodies (Figure 5C). Taliks can be due to a local anomaly in thermal, hydrological, or hydrochemical conditions [56]. Electrical resistivity is generally lower in a water-saturated substrate [14]. The electrical resistivities of U4 therefore are consistent with wet/unfrozen sediment. As highlighted earlier, caution must be used when interpreting this unit, as the highly resistive U3 unit may shield the actual electrical resistivity of those deeper zones. However, U4 zones not overlain by highly resistive permafrost, such as between 20 and 40 m of Profile 3 (Figure 5C), offer a better resolution and can be interpreted with greater confidence. These low-resistivity zones correlate with the presence of taller and denser surface vegetation like shrubs, as in the center of the semi-subterranean sod house H3 (Figure 5D,E) and the low-topography zones on the extremities of Profile 2 (Figure 5B). The U4 zone of Profile 3 is more extended than in the other profiles (Figure 5C). This could be explained by its proximity to a zone of thermokarst gullying, and therefore a thermal and hydrological anomaly.

The heterogeneity of thickness and resistivity values from the ERT profiles indicates that the characteristics of the subsurface may vary over small distances, even if only over a few meters. The electrical resistivity of frozen soil depends on unfrozen water content, soil temperature, soil type and texture, and soil salinity [14]. Therefore, the resistivity at which the ground is considered frozen may vary from profile to profile. This is to be expected, as the study site is in a zone of discontinuous and dispersed permafrost [57].

Four main units were identified and interpreted in the 160 MHz radargrams (Figure 7): 1. a unit from the top of the profiles to a depth of about 1 m, composed of many linear reflectors underlain by a clear interface (U1); 2. an unattenuated unit underneath the clear interface at the bottom of U1, or faint signal multitudes (U2); 3. a unit of strong signal multitudes (U3); and 4. a unit of highly attenuated signal (U4) at greater depths and near humid zones. These units are classified by their radargram characteristics and correspond to the units identified by ERT. The 160 MHz radargrams were used to identify permafrost characteristics, as this frequency allows more penetration depth beyond the top of the permafrost table.

The interface at the bottom of U1 (at a depth of 1 m) (Figure 8) has a shape close to the topography and could be an indicator of the bottom of the active layer [58]. Furthermore, this strong interface indicates a high contrast of relative dielectric permittivity, which could indicate the contrast between frozen and unfrozen sediments [53,59]. Accordingly, this interface was identified as the bottom of the active layer (top of the permafrost table).

In all the profiles, linear reflectors in the thawed active layer (U1) correspond to the sandy stratified sediments confirmed by test pit findings (Figure 8). Figure 8B presents a 670 MHz radargram from Profile 1, cropped at a distance of 30 m to visualize the many interfaces imaged in the active layer.

Linear dipping reflectors were found at depths between 50 and 80 m on Profile 3, in unit U4 (Figure 7C). As these reflectors follow the shape of surface topography, they were identified as layers of unfrozen sediment overlain by peat. GPR has been known to successfully detect the interface between peat and silt in the active layer [59]. Due to the high resistivity of U3, radar waves are reflected on this ice-rich layer and therefore detail the layers within the active layer. The many visible layers are consistent with the pedostratigraphy identified in the test pits (Figure 4), which showed sediments with variable grain size. The characteristic pattern of peat on a radargram is of planar bedding [60], which was observed at the east end of Profile 3 (Figure 7C).

U3 is a zone that shows strong multiples of the permafrost table reflector (Figure 7). These multiples result from the resonance of the linear reflector at the bottom of U1 (Figure 8), which suggests a strong contrast in the dielectric properties of the materials. According to previous studies [54,61], a zone where multiples of the reflector are visible may correspond to ice-rich frozen sediment or massive ice. Due to the relative homogeneity of ground ice, few reflections other than the resonance of the previous interface were visible [60]. Profiles 1, 3, and 6 (Figure 7A,C,D) show hyperbola reflectors that could be associated with ice wedges. Indeed, GPR is known to detect ice wedges [62,63,64], as ice wedges act as point sources that induce hyperbolic reflections [62] because of the contrast in electrical permittivity between the massive ice and surrounding permafrost. The head of the visible hyperbolas would correspond to the top of the ice wedges. They would therefore be located at the base of the active layer. Ice wedge depth would correspond to the hyperbolic reflections caused by the base of the wedge [65]. Those hyperbolic reflectors are not visible in the zones near water or thermokarst, as in Profile 2 and the last 30–80 m of Profile 3 (Figure 7). These findings reinforce the probability that these reflectors indicate ice wedges, as they seem to be present only in profiles with dry and cold conditions. Moreover, Foury [29] has noted the presence of ice wedges measuring about 20 cm long inside the midden of the semi-subterranean sod house H1 on the site. However, no polygonal network was visible at the surface. This may be because the ice wedges appear to be positioned beneath the active layer, rendering their pattern invisible at the surface.

For Profiles 2 and 6 (Figure 7B,D), signal attenuation at the beginning and end of the profiles is associated with proximity to humid zones, such as thermokarst ponds, that have high electrical conductivity. U4 zones can therefore be interpreted as taliks.

A maximum penetration with a good resolution of about 3 m in depth was attained for all profiles at 160 MHz (Figure 7). For the 670 MHz antenna, the signal penetrated to a maximum of about 2 m (Figure 8B). As the more electrically conductive the medium, the more the radar signal is attenuated, this attenuation of the signal may be caused by unfrozen wet sediment (U4) or the presence of more fine-grained sediment. In support of this interpretation, Roy et al. [27] showed the presence of silty sand at a depth of about 2 m through stratigraphic cross-sections. The attenuation of the signal may also be the result of the strong dielectric contrast between the active layer and resistive permafrost, preventing the radar wave from penetrating deeper [53,60,66]. Due to this signal attenuation, bedrock depth cannot be determined.