Early warning of shallow landslides: monitoring of pre-failure suction-induced deformation

Most of the Landslides Early Warning Systems (LEWS) in operation are based uniquely on monitoring rainfall data, limiting their performance due to false alarms generated by rainfall thresholds. The accuracy of LEWS may be remarkably improved by monitoring soil-based variables and the stress-strain response of the ground during intense rainfall events. This paper investigates whether slope pre-failure deformation can be used as an additional precursor of landslide triggering. This further precursor would substantially improve LEWS accuracy, especially if pre-failure deformation is combined with suction monitoring. Some tests were conducted using a small-scale physical model of a slope built with unsaturated volcanic silt subjected to artificial rainfall. A new device named tensio-Inclinometer was developed to monitor simultaneously suction and suction-induced deformation. It combines a conventional tensiometer and an accelerometer installed at the top of the tensiometer shaft. It is shown that pre-failure deformation detected by the tensiometer shaft tilting is a suitable landslide precursor. Moreover, it can provide reliable soil-based thresholds for early warning systems if combined with suction.