Delineation of areas with different temporal behavior of soil properties at a landslide affected Alpine hillside using time-lapse electromagnetic data

Landslide activity is largely controlled by changes in soil properties, particularly soil moisture and the corresponding changes in pore pressure within the vadose zone. While knowledge of changes in soil conditions is of utmost importance for the prediction of landslides, it is difficult to obtain reliable information on the field scale. A possibility of filling that information gap is the monitoring of changes in soil properties by time-lapse electromagnetic induction (EMI) data. Given the relative stability of soil properties, changes in apparent electric conductivity (ECa) are mainly related to changes in soil water content and its mineralization. Thus, we use time-lapse ECa data over a nine-month period from different investigation depths (0.75, 1.5, 3, and 6 m) to separate areas with different temporal behavior of soil properties. However, working with time-lapse EMI data raised the comparability problem since the recoded ECa is also affected by several day-specific survey conditions (e.g., instrument temperature, operator). Consequently, the reproducibility of accurate ECa measurements is difficult due to potential dynamic shifts which hinders a direct comparing. We introduce in this study a straightforward method for comparability of ECa values from different time steps by normalization of data ranges assuming that the majority of shifts of measured data originate from field calibration. We identify the intensity of spatial changes by means of the standard deviation (SD) as an indication for the intensity of soil properties variability. To obtain the temporal changes and its progression over time, we separate the dynamic signal from the background. A two-layer system could be identified: one shallow more dynamic layer with an east–west-oriented structure and a deeper, more stationary layer with a north–south-oriented structure. The ECa dynamics of the shallow layer is related to the altitude (R 2 = 0.84) while the deeper dynamics follow a different regime. The decreasing of ECa dynamics with depth was consistent with the decreasing of SWC dynamics observed by previous studies.