Mountain permafrost, a key component of the cryosphere, is highly sensitive to climate change, particularly in high mountain areas such as the European Alps, where permafrost temperatures are close to the melting point. Over recent decades, permafrost warming and degradation have led to increased active layer thicknesses and rising ground temperatures, trends projected to continue with very high confidence throughout this century. These changes strongly affect mountain ecosystems and communities, with implications for slope stability, human safety, infrastructure, and hydrological processes. Characterizing permafrost is therefore essential for understanding its thermal state, hydrological regime, and stability with key parameters including ground temperature, active layer thickness, soil moisture, snow cover, and ground ice content.

While borehole temperature data provide valuable one-dimensional records, they are sparse and cannot capture the strong spatial heterogeneity of mountain permafrost. Ground ice content, in particular, is difficult to quantify and monitor due to its high spatial variability, restricted accessibility, and the logistical constraints of core drillings or nuclear well logging. Geophysical methods have therefore become indispensable for imaging subsurface properties over large areas and depths, and for distinguishing frozen from unfrozen ground. Recent studies quantified ice and water contents by combining or jointly inverting electrical, electromagnetic, seismic and/or gravimetric methods and linking them through petrophysical models. Despite these advances, applications in the European Alps remain sparse and uncertainties persist.

This addresses the applicability and development of spectral induced polarization (SIP), an electrical method rarely applied in permafrost research, for investigating spatial and temporal variations of ground ice in different mountain permafrost landforms. While electrical resistivity data alone yield ambiguous interpretations when distinguishing between ice, air and rock, SIP measures not only conductive but also capacitive properties of the subsurface. These are linked to polarization processes at electrical double layers along grain-fluid and ice-fluid interfaces, which are strongly influenced by the presence of ice. Three key challenges are addressed within this dissertation: (i) optimizing SIP survey design for reliable data quality under alpine conditions, (ii) investigating SIP responses across different mountain permafrost landforms with varying ice contents and their seasonal to interannual variability, and (iii) identifying proxy parameters for estimating ground ice content.

First SIP measurements at the well-characterized Lapires talus slope in Switzerland evaluated the frequency dependence (0.1-225 Hz) of the polarization response for distinguishing frozen from unfrozen substrate. The aim was to (a) establish a field protocol that provides SIP imaging data sets less affected by electromagnetic coupling and deployable in rough terrain, (b) cover the spatial extent of the local permafrost distribution, and (c) evaluate the potential of the spectral data to resolve variations in substrates and ice content. Data uncertainty was assessed by analysing misfits between normal and reciprocal measurements across different profiles and frequencies. A comparison between different cable setups showed lowest misfits for coaxial cables and enabled high-quality SIP data acquisition in the 0.1-75 Hz range. Results revealed a clear contrast between the ice-rich talus material and unfrozen surroundings, with polarization increasing with frequency in ice-rich zones. A raster of different SIP profiles allowed the characterization of the entire landform, showing the ice-rich permafrost body to be smaller than previously assumed. Furthermore, the spectral behaviour helped to improve the discrimination between ice-rich permafrost and unfrozen bedrock in ambiguous cases.

In 2019, the first permanent SIP monitoring profile in mountain permafrost was installed at the bedrock permafrost site Cervinia Cime Bianche (Italy), covering a monitoring period of 6 years. Temporal variations in polarization were closely linked to freeze–thaw processes in the active layer, absolute phase values increasing during freezing in autumn, peaking in winter, and decreasing during spring thaw. Field data were validated through freeze–thaw laboratory experiments on site-specific rock samples, showing consistent spectral behaviour and confirming the temperature dependence of the SIP response. A key contribution of this work was the development and application of the phase frequency effect (φFE) as a proxy for ice content. The φFE followed clear freeze-thaw cycles and reflected differences in textural properties and ice-water ratios. Limitations were identified in winter, where high resistances caused increased phase errors and data uncertainty due to parasitic capacitive coupling effects. Laboratory electric circuit tests reproduced these effects and quantified the lower bound of the error level of the phase measurements expected in field conditions.

Expanding the scope across different mountain permafrost landforms, SIP responses were evaluated at ten permafrost sites in the European Alps, covering bedrock, talus slopes, and rock glaciers including pure ice and an unfrozen site. φFE systematically varied with landform and ice content, showing highest values in rock glaciers, intermediate values in frozen talus slopes, and lowest values in bedrock permafrost and unfrozen sites. A strong correlation between φFE and validation ice contents (r² = 0.94), confirmed φFE as a potential parameter for future ground ice estimations. The role of surface conduction was investigated across all sites, a parameter often neglected in petrophysical models. SIP results showed that surface conductivity varied both spatially and temporally, and thus needs to be considered in petrophysical models.

The findings of this thesis clearly demonstrate that SIP is a sensitive and promising method for detecting and monitoring ground ice in permafrost environments. Temporal, spatial and multi-site analyses show its clear potential, though further work is needed to establish widely applicable and reliable ice content estimates with SIP in alpine and polar permafrost systems.