Broadband Spectral Induced Polarization in Permafrost Peatlands of Northern Sweden

Permafrost peatlands, located in the Arctic and high mountain regions, are typically known to be ice-rich. This is primarily linked to significant water content and often oversaturation, a characteristic property of peat soil. The current understanding of the effects of human-induced climate warming suggests that these regions are approaching a climatic tipping point with substantial permafrost thaw expected in the coming decades. Ice content is an important parameter for modelling permafrost evolution and at present limited studies exist that determine its in-situ spatial distribution in such areas.

The geophysical method known as high-frequency induced polarisation (HFIP) is advantageous for cryohydrological research in these environments. This method can capture the frequency-dependent polarisation of ice (also termed dielectric relaxation peak), which occurs within the range of 100 Hz to 100 kHz and is expressed by complex resistivity. Therefore, by analysing the spectral behaviour of this complex resistivity within the target frequency range the distribution and quantity of ice can be estimated.

The results from the latest field campaign conducted at Storflaket mire and Stordalen mire in Abisko, Sweden, are presented. Two-dimensional HFIP profiles were measured to resolve the near-surface unfrozen layer (no-ice) and the underlying frozen layer (ice-bearing). The measurements were performed in late summer when the depth of the unfrozen layer was at its maximum. Field data are inverted as independent frequencies to obtain the spectral variation of complex resistivity. No-ice and ice-bearing regions are classified by the presence of the relaxation peak. Subsequently, a two-component mixture model, with one component as ice and the second as the surrounding matrix, is applied to determine ice content distribution. Boundary constraints and starting parameters are chosen using the spectral analysis of the inverted complex resistivity. The model accuracy is evaluated using unfrozen layer probing and a permafrost core extracted along the HFIP profile. The HFIP-derived ice content distribution is consistent with unfrozen layer probing, i.e., the classification of no-ice and ice-bearing regions is successful. The model tends to underestimate ice content percentages compared to permafrost core laboratory measurements. This discrepancy can be explained since laboratory measurements are based on gravimetric water content and assumes all pore-water is frozen. However, it is known that residual pore-water is present in these soils even below 0°C. Additionally, it is observed that the model performs well when the ice content percentage is 10% or greater and its applicability might be limited in scenarios where the ice content is less than 10%.

The latest results are discussed in comparison with previous findings from Heliport, a permafrost mire also located in Abisko. In the Heliport study, HFIP successfully resolved the complex resistivity and ice content distribution on a larger scale. Building on the field knowledge gained at Heliport, this study incorporates improvements in electrode configuration setup, data acquisition speed, and minimising cable-earth coupling effects. The findings contribute to the understanding of the induced polarisation of permafrost peatlands, which is an underexplored area from a geophysical perspective.