A linear regression framework for predicting subsurface geometries and displacement rates in deep-seated, slow-moving landslides

摘要

A new numerical integration/linear regression tool is used to investigate kinematic behavior in deep-seated(sliding surface deeper than 3 m) mass movements. The technique is developed in the context of realinclinometer records and numerical integration using nine inclinometer case histories from four welldocumented, large landslides: Carrot River, Montebestia, Karya village, and Pietrapertosa. Axial metric,subsurface geometric deformation rates are predicted and compared for four different case studies from theliterature: a reactivated, composite, extremely slow earth slide-earthflow; a reactivated composite,extremely slow debris slide-rock fall; an active, composite, very slow earth slide-earth flow; and areactivated complex, slow earth slide-earth spread. Sensitivity analysis, based on sliding surface depth-to-length (D/L) ratios, show that the mobility of these slow-moving masses is closely dependent on the mode ofsliding. The results also show that short term and long term dynamics of slow moving landslides can becaptured by geometrical patterns. Because the parameter determined is a geometric property, the techniqueused in the investigation can be applied in new landslide problems independent of local conditions andtriggering mechanisms, within the confines of the stipulated boundary conditions. Hence with thisframework also, unrelated landslide problems can be analyzed and compared, as demonstrated herein.Additionally, this approach is useful for two dimensional reconstructions of subsurface displacement andvelocity profiles, and thus may act as a precursor to detailed field investigation programs, warning systemsand mitigation projects at minimal costs.