Spring trends in slow-moving landslide displacement: is it a reliable way to predict their movements? Two case studies in the Susa Valley (NW Italy)

Mountainous regions are highly susceptible to ground instabilities due to their geomorphological features and the climate events. Slow-moving landslides are influenced by multiple interacting predisposing factors, complicating the prediction of acceleration patterns. Rising groundwater levels, closely linked to net precipitation and snowmelt, are the main drivers of slope movements. Thus, reconstructing ground instability scenarios requires analysing groundwater dynamics. In this context, the present study focuses on the methodological development and validation of an alternative approach for landslide monitoring, rather than on the direct cause-effect relationship between groundwater variation and slope movements. The proposed method aims to assess the reliability of using spring water levels as a proxy for predicting slope displacements, and to compare this approach with the use of piezometric data, which are typically employed in conventional monitoring system. This is particularly relevant in mountainous environments, where the number of in situ instruments is often limited. The methodology combines statistical tools and Fourier spectral analysis to investigate the coherence between hydrogeological and kinematic time series. The analysis was applied to two large slow-moving landslides in the Western Italian Alps (Champlas du Col and Thures). Results show that when springs and inclinometers are in proximity and belong to the same landslide dynamics, as in Thures landslide, the spring trend can effectively predict conditions triggering movement. Conversely, at Champlas du Col landslide, discrepancy between spring and displacement trends suggests the presence of a sub-landslide within the main body. This is likely related to local geological settings, as confirmed by the strong correlation between piezometric data and displacement rates. Overall, this research highlights the methodological potential of integrating spring monitoring into landslide observation frameworks, providing a cost-effective and scalable tool for areas with limited instrumentation. This approach lays the groundwork for the development of generalised criteria to identify suitable springs and to support the forecasting of slope accelerations in similar geological contexts.