Surveyed monitoring of the retaining wall began in 2011 and is conducted with a total station and prisms installed on 19 of the piles of the retaining wall. The prisms were located at the top of the piles, and the monitoring station was located northeast of the landslide site, about 300 m from the wall. A plain view with the location of the piles within the wall is shown in Fig. 10. Monitoring was carried out by a third party survey company. The station was considered stable relatively to the landslide displacement rates, and it was regularly checked as part of the surveyors QA/QC. For the distances between prisms and the total station, the expected single measurement accuracy was ± 2 − 3 mm. This was considered adequate for the large displacement rates observed in the field. Initially, monitoring was performed with quarterly measurements (in average) and was then progressively enhanced up to weekly measurements by July–August 2015, reflecting the acceleration of the landslide and the perceived hazard levels. No significant pile displacements were recorded until early 2015, when displacement rates started to increase. Monitoring at Pile 17 had technical difficulties and was stopped prematurely; thus, no information was obtained for this pile.
Of the 19 piles, 10 of them (piles 0–9, or “central piles”) are definitely identified within the boundaries of the landslide movement, whereas the others (piles 10–18, or “lateral piles”) are located on or just outside such boundaries (Fig. 10).
The characteristics of the movements of the central piles are markedly different from those of the lateral piles. Conversely, the characteristics of the movements are extremely similar among piles of the same group. In the following sections, in cases where for illustration purposes it is not convenient to show data for all piles, piles 0 and 6 represent the characteristics of the movements of the central piles, while piles 13 and 14 represent the characteristics of the movements of the lateral piles.
Cumulative displacement and displacement velocity
The surveyed data consist of the 3-dimensional coordinates of each pile prism for each monitoring date. Incremental displacements \( {\widehat{d}}_{i(X)} \), \( {\widehat{d}}_{i(Y)} \), and \( {\widehat{d}}_{i(Z)} \) for the period between time t
i − 1 and time t
i
correspond to movements in the North, East, and elevation directions, respectively. These are calculated as the difference in position between t
i − 1 and t
i
. The horizontal component of the incremental displacement at time i is calculated according to:
$$ {\overset{\hat{\mkern6mu} }{d}}_{i(h)}=\sqrt{{\left({\overset{\hat{\mkern6mu} }{d}}_{i(X)}\right)}^2+{\left({\widehat{d}}_{i(Y)}\right)}^2} $$
(3)
while the vertical component of incremental displacement at measurement i is directly derived from \( {\widehat{d}}_{i(v)}={\widehat{d}}_{i(Z)} \).
Consequently, cumulative displacements at time i are calculated as:
$$ {D}_{i(h)}={\sum}_{j=1}^i{\overset{\hat{\mkern6mu} }{d}}_{j(h)} $$
(4)
$$ {D}_{i(v)}={\sum}_{j=1}^i{\overset{\hat{\mkern6mu} }{d}}_{j(v)} $$
(5)
Figure 11 depicts the measured cumulative horizontal and vertical displacements of selected piles between February 2011 and September 2016. The central piles show persistent displacements from January 2015 up to September 2016, with total displacements ranging between 2.5 and 4.5 m horizontally and between 1.5 and 2.1 m vertically. Cumulative displacement acceleration of these central piles was observed in January 2015 and June 2016 (red dashed lines in Fig. 11a). While the first acceleration phase reflected the natural dynamics of the landslide, the second acceleration phase was associated with the installation of the new shear piles (see “Retaining wall” Section); this in fact required drilling and consequently determined a temporary increase of pore water pressure, which prompted acceleration of the landslide. On the other hand, movement of the lateral piles is about one order of magnitude smaller and in most instances does not appear to be consistent over time (Fig. 11b—note the scale is amplified by a factor of 10 with respect to Fig. 11a).
Figure 12 illustrates the corresponding displacement velocities of the piles in Fig. 11, in millimeters per day. The discrepancy between the behavior of central and lateral piles results evident: the central piles show the two aforementioned distinct phases of acceleration, separated by a prolonged phase of mostly constant velocity which is characterized by values ranging from approximately 3 to 9 mm/day horizontally and from 2 to 4 mm/day vertically. A peak velocity of 15.9 mm/day in the horizontal direction is identified on 25 July 2016 for pile 6. Afterwards, velocities started to decrease and ceased to be significant at the end of September 2016, when the stabilization works were being finalized. Conversely, the lateral piles never showed a consistent phase of acceleration in the dataset. Starting from August 2015 velocities were typically scattered between 0 and 3 mm/day in the horizontal direction and between 0 and 1 mm/day in the vertical direction (i.e., within the interval of measurement accuracy).
Evolution of displacement trends
Survey of the piles coordinates allowed also calculation of the direction of horizontal and vertical movement, expressed in terms of azimuth (α) and dip angle (β). The azimuth angle (°) of the horizontal component of movement at measurement i is calculated according to:
$$ {\alpha}_i=\arctan \left({\overset{\hat{\mkern6mu} }{d}}_{i(X)}/{\overset{\hat{\mkern6mu} }{d}}_{i(Y)}\right) $$
(6)
Similarly, the dip angle (°) of the displacement vector at measurement i is:
$$ {\beta}_i=\arctan \left({\overset{\hat{\mkern6mu} }{d}}_{i(h)}/{\overset{\hat{\mkern6mu} }{d}}_{i(v)}\right) $$
(7)
Differently, from its typical use for mapping planar features in engineering geology applications, the dip angle is here considered to vary between 0° and 360°, with a value of 0° indicating a perfectly horizontal movement out of the slope, and a value of 180° a perfectly horizontal movement toward the slope; consequently, a 90° dip angle indicates a perfectly vertical downward movement, while a 270° dip angle a perfectly vertical upward movement.
The geometrical characteristics of the displacements of the piles and their evolution can then be analyzed. Figures 13 and 14 show the direction of the increments of horizontal movement (azimuth) of selected central and lateral piles from January 2015 (start of the landslide deformation) to September 2016 (end of the landslide deformation due to completion of the stabilization works). In the mentioned figures, such time interval is divided into two segments for illustration purposes. The length of the vector increments is proportional to the average daily horizontal displacement for the relative interval of monitoring (different scales in Figs. 13 and 14 were also used for illustration purposes). Figure 13 indicates that the direction of horizontal movement of the central piles was mostly constant throughout the entire period of landslide deformation. The azimuth is generally sub-parallel to the aspect of the slope; however, in the eastern sector of the landslide (see piles 0 and 3), this is slightly shifted toward the central part of the landslide (see pile 6). This can be explained by the dragging action of the central section of the wall, which has suffered the largest displacements and thus pulls the other piles toward the center.
Conversely, the lateral piles (Fig. 14), and in particular those located outside the western boundary of the landslide (see piles 12 and 13), were characterized by a much more variable-measured direction of horizontal movement. In the last part of the monitoring period, pile 15 appears to assume more consistent azimuth values, similar to those of pile 0 (Fig. 13). However, the intensity of the displacements was not as significant. The scale in Fig. 14 is different than in Fig. 13 in order to magnify the increments of piles displacement; moreover, the scale of piles 12–14 is further amplified with respect to that of pile 15 to better appreciate the changes in movement direction.
Similar observations can be made by analyzing the trend of cumulative horizontal vs. vertical displacement of the piles for the period January 2015 to September 2016 (Fig. 15). At the central piles, a mainly constant relation between horizontal and vertical movements can be appreciated. The inclination of the line plots does not vary significantly with time and is at an angle close to 23°, consistent with the estimated inclination of the basal sliding surface. This was expected for a landslide sliding over a basal, planar surface. On the other hand, the erratic behavior shown by the lateral piles correspond to processes of landslide retrogression combined with the dragging action of the wall. As the landslide moves downslope, it drags the center of the pile wall, which in turn drags the lateral extents. Since movements of the lateral piles are much more limited with respect to the central piles, the mentioned variability may in part also be a consequence of measurement error constituting a larger percentage of the measured data.
Additional details on the geometry of the wall movements can be determined in Figs. 16 and 17, where the azimuth and dip angle of each increment of displacement of every pile since January 2015 to September 2016 are reported. Except for a few spikes, the characteristics of the movement of the central piles (Fig. 16a–c and Fig. 17a–c) remained consistent throughout the entire landslide deformation phase, with values of azimuth typically ranging between − 40° and − 70° with respect to the North (i.e., approximately in the NW-NNW direction) and values of dip angle ranging between 15° and 30°. The measured azimuth and dip angles of movement of the lateral piles (Fig. 16d–f and Fig. 17d–f) were instead extremely variable, with displacements ranging between − 90° and 90° with respect to the North (i.e., from West to East) and in both downward and upward direction. As previously mentioned, this marked variability may be explained with the dragging action of the wall and with the higher impact of measurement error at piles characterized by low displacements.
As a result, the piles of the railway retaining wall at the 10-mile Slide can be classified according to two types of deformation behavior:
-
Type 1 (central piles): characterized by large overall displacements (several meters in horizontal direction), consistent phases of progressive acceleration, and low variability of the azimuth and dip angles of movement.
-
Type 2 (lateral piles): characterized by lower overall displacements (< 1 m in horizontal direction), lack of consistent phases of progressive acceleration, and high variability of the azimuth and dip angles of movement.
In relation to the ongoing maintenance of the wall, in the latest part of the monitoring period, a change from Type 2 to Type 1 deformation behavior was observed concerning some of the piles located in proximity of the eastern boundary of the landslide. As mentioned preliminarily in Fig. 14, this resulted evident in particular in the case of pile 15, which displayed a progressive increase in velocity in July 2016 (values of up to 9 mm/day in horizontal velocity) and a stabilization of the measured azimuth of movement starting from April 2016 (Fig. 18). This suggests that the boundaries of the 10-mile Slide are subject to further phases of active retrogression.