A systematic evaluation of electrical resistivity tomography for permafrost interface detection using forward modeling

• Published in: Permafrost and Periglacial Processes
• Main Authors: Herring, Teddi, Lewkowicz, Antoni G.
• Other Authors: Natural Sciences and Engineering Research Council of Canada
• Format: Article in Journal/Newspaper
• Language: English
• Published: Wiley 2022
• Subjects: Ice; permafrost; Permafrost and Periglacial Processes
• Online Access: http://dx.doi.org/10.1002/ppp.2141; https://onlinelibrary.wiley.com/doi/pdf/10.1002/ppp.2141; https://onlinelibrary.wiley.com/doi/full-xml/10.1002/ppp.2141

Abstract The accuracy of electrical resistivity tomography (ERT) as a method for locating frozen‐to‐unfrozen interfaces in permafrost environments was investigated systematically for simplified scenarios using forward modeling. The impacts of varying the resistivity, thickness, and lateral continuity of the frozen region, altering the thickness of the surface thaw layer, and of differing array types were evaluated in relation to the detection and positioning of frozen–unfrozen interfaces. The results from these simple scenarios show that boundaries between frozen and unfrozen ground are more accurately indicated by maximum gradients rather than a fixed threshold value based on the resistivity at the base of the surface thawed layer. The resistivity of the frozen region plays a significant role in interpreted boundary locations, with high resistivity values causing a decrease in model sensitivity at depth and increased uncertainty in the interpreted base of the frozen zone, particularly in laterally continuous permafrost. Error in the interpreted base of the frozen zone also increases for thicker permafrost bodies, while thaw layer thickness plays a less significant role. In laterally discontinuous permafrost, wider frozen bodies cause the boundary at the base of the frozen region to become less distinct. Array type affected the appearance of the inverted resistivity models and the frozen–unfrozen boundaries located using the threshold method, but boundary locations were comparable among array types when the maximum gradient method was used. This synthetic modeling showed that the boundaries between unfrozen and frozen regions in ERT images should be interpreted with caution, particularly in ice‐rich, laterally continuous permafrost where sensitivity at depth is low. We conclude that forward modeling is a useful tool for permafrost investigations, both for assessing the likelihood of achieving ERT survey goals prior to fieldwork, and as an interpretive aid after field data have been acquired.