The Differential Slow Moving Dynamic of a Complex Landslide: Multi-sensor Monitoring

Abstract

Monitoring is essential to understand the mechanics of landslides, and predict their behavior in time and space. In this work we discuss the performance of multi-sensor monitoring techniques applied to measure the kinematics and the landslide hydrology of Portalet landslide complex, which is located in the SW-facing slopes of Petrasos peak at the border between Spain and France. In the summer 2004, the excavation of a parking lot at the foot of the slides triggered a secondary failure in the lower part of the slope, accelerating the dynamic of the landslide complex. The deployed hydro-meteorological network has been useful to understand that the greatest infiltration in the moving mass is produced in spring due to the combination of snow melt and seasonal rainfall. Landslide surface kinematics measured with differential GPS (D-GPS) were useful to measure the slower (< 10 cm/year) and faster (20-30 cm/year) dynamic of the landslide complex. Advanced DInSAR was useful to monitor the slower ground displacements from long datasets of SAR images, providing a wider spatial coverage and measurement point density than the D-GPS. In addition, the NL-InSAR processing strategy was applied to monitor the faster motion using short datasets of TerraSAR-X images excluding the snow cover period. The installed horizontal extensometers were useful to study the extension of the head scarp and its relationship with landslide hydrology, which is affected by the retrogressive effect of the landslide due to the loss of lateral confining pressure. Finally, an inclinometric robot system (AIS) was the only technique capable of detecting 5-6 time faster motion after the snow melt, since it provides daily measurements with high accuracy even during the snow cover period. These data, even if expensive to gather, are necessary to improve the hydro-mechanical modeling of large slow landslides, such as those already proposed for Portalet landslide complex.