Cascade Landslide Complex
Detection of slow or limited landslide movement within broad areas of forested terrain has long been problematic. My work combined continuous GPS data of high temporal resolution with spaceborne InSAR measurements of high spatial resolution to study the complexity of movement over a ~4 km2 seasonally active translational Crescent Lake landslide, a part of the Cascade Landslide Complex, in the Columbia River Gorge (Washington State).
I applied novel time-series InSAR processing strategies to characterize the spatial distribution and temporal behavior of the rainfall-triggered landslide movement. Time-series measurements reveal the seasonal deformation of a landslide lobe, with a much larger magnitude compared to the lower elevated riverbank, suggesting an amplified hydrological loading effect associated with a thick unconsolidated zone. The landslide moves mainly during the wet season, much of it at an average rate of 15–20 cm/yr. From the time-series amplitude information on the terrain upslope of the headscarp, we also re-evaluate the incipient motion related to the 2008 Greenleaf Basin rock avalanche, not previously recognized by traditional SAR/InSAR attempts [Hu et al., 2016, RSE].
Figure 1. Location maps of study area: (a) SAR data coverage; and (b) shaded relief map of the Cascade landslide complex [Pierson et al., 2016], with dashed outline of Crescent Lake landslide determined by geomorphic evidence and early InSAR interferograms. Solid black lines show other landslides within the complex. The shaded area (covered by the array of circles) encompasses the area of detected movements [Hu et al., 2016], downsampled to 100 m by 100 m grids for smoothness preservation; it is the area used for thickness inversion. The red dot marks the location of the continuous GPS station that provided the data for this study.
Threshold rainfall intensities and durations wet seasons have been associated with observed movement upon shearing: antecedent rainfall triggered precursory slope-normal subsidence, and the consequent increase in pore pressure at the basal surface reduces friction and instigates downslope slip over the course of less than one month [Hu et al., 2018, GRL].
Figure 2. GPS displacements at the semipermanent GPS station (red dot in Figure 1), projected onto the LOS direction in comparison with the Sentinel-1A measurements of (a) ascending track P137 and (b) descending track P115, along with pre-30-day precipitation total (30-day cumulative precipitation before the assigned date) from a nearby weather station (Cascade Locks, Oregon). Light green shading under the pre-30-day precipitation curves shows the antecedent rainfall period, and the dark green shading corresponds to the period of slope-normal subsidence when the precipitation is more intense. The inset diagram shows how ground displacement is sensed by a right-looking satellite on a descending track.
A quasi-three-dimensional deformation field is created using Sentinel-1A, ALOS-1, and ALOS-2 InSAR observations constrained by the topographical slope, and is further used to invert the complex geometry of landslide basal surface [Hu et al., 2018, GRL].
Figure 3. Thickness variation of the Crescent Lake landslide, obtained by inversion of the quasi-3D displacement field.Quasi-3D displacement velocity maps in unit of mm/yr during (a) 2007-2011, (b) 2014-2016 and (c) the average of 2007-2011 and 2014-2016 of the shaded zone shown in Figure 1. Horizontal velocity vectors indicate direction (effectively imposed by topographic slope) and magnitude of horizontal displacements, and color shows magnitude of vertical displacements. Thickness is shown in meters when the rheological parameter (d) f=1/2, (e) f=2/3, and (f) f=1, respectively. The solid line shows the mapped boundary of the Crescent Lake landslide [Pierson et al., 2016], and the cross-hatched zone marks the area outside of the southwest margin of the landslide. (g) Geometry of the smoothed landslide top surface and basal surface when f=1. Note that the boundaries of the topographic ground surface and the basal surface are superimposed, but the topographic ground surface is raised to better reveal the basal surface variations. (h) Elevation contours of the smoothed slope surface (gray lines and underlined digits) and the basal surface (colored lines and digits). (i) Profiles of surface elevation and landslide basal surface elevations along the dashed-line transect.