Infiltration-induced seasonally reactivated instability of a highway embankment near the Eisenhower Tunnel, Colorado, USA

Abstract

Infiltration-induced landslides are major natural hazards. When they occur along highways they can impede traffic, damage infrastructure, and threaten public safety. This paper presents a case study of an active landslide on an embankment of Interstate-70 west of the Eisenhower Tunnel in central Colorado, USA. Records indicate that the hillslope under I-70 has moved episodically over the previous 40years. In the previous two decades the road surface has been displaced vertically by more than 60cm. The objective of this work is to develop a conceptual model capable of quantifying the seasonally reactivated landslide movement at the site. Inclinometer data of subsurface deformation and geologic and hydrologic mapping are used to develop the conceptual model as well as to constrain a numerical model. A two-dimensional hydro-mechanical numerical model is used to test the conceptual model under three different infiltration rates during the period of snowmelt in the spring. The framework used in the numerical model accounts for the major physical processes driving instability of the slope: time-dependent variably saturated flow, the resultant changes in stress, and induced deformation. When the snowmelt water infiltrates into the slope, the soil water content and the water-table level vary accordingly. These time-dependent variations result in changes in soil matric suction, effective stress, and consequently change in slope stability. The model calculates pore-water pressures, suction stress, and the distribution of effective stress in the embankment slope at different times. Global factors of safety as a function of time are calculated along the predetermined sliding surface using effective stresses calculated with finite elements. The slope stability assessment quantitatively confirms the conceptual model and is consistent with the displacements monitored at the site during the years of 2007–2009. It is shown that annual snowmelt infiltration of 60–100cm of water can reduce the factor of safety by 6%, enough to sustain landslide movement for as long as 8months each year.