On capillary barrier effects and debris slide triggering in unsaturated layered covers
Highlights
► Stability of layered slope in relation to capillarity barrier are analyzed. ► Infiltration in layered slopes with capillarity barrier are modeled. ► Capillarity barrier effects in relation to different slope initial soil moisture. ► Influence of capillarity barrier in activation of 1998 Sarno debris flow.
Introduction
The present work is focused on investigating the potential role that capillary barriers may have on the stability of unsaturated layered soil deposits laying upon high permeability layer. Although the following discussion has a general character, some basic ideas and considerations have been drawn from a peculiar catastrophic event that shocked the Italian People on May 1998 and is still object of an open discussion (Esposito and Guadagno, 1998, Del Prete et al., 1998, Crosta and Dal Negro, 2003, Basile et al., 2003, Olivares and Picarelli, 2003, Fiorillo and Wilson, 2004, Guadagno et al., 2005, Cascini et al., 2008, among many others). The catastrophic landslides that occurred on May 1998, involved pyroclastic covers laying upon the slopes of Pizzo D'Alvano, just upslope of the towns of Bracigliano, Quindici, San Felice a Cancello, Sarno and Siano, in Campania (Southern Italy).
Following the classification given by Hungr et al. (2001), landslides took place in the form of debris flows and debris avalanches. They were triggered by sliding of superficial pyroclastic debris from the covering mantle, and were subsequently channelized as debris flows (Fiorillo et al., 2001, Crosta and Dal Negro, 2003, Revellino et al., 2004). Several millions of cubic meters of slope materials were displaced in the landslides and caused the death of 161 people.
It is useful to recall that these extreme instability phenomena took place after a period of several days of not intense, but quite continuous rainfall, characterized by a return time not very relevant if compared to the severity of landslides (Crosta and Dal Negro, 2003, Fiorillo and Wilson, 2004). This event can be considered somehow hydrologically exceptional for its duration and continuity.
The total amount of rainfall was significantly less than other major precipitation events recorded during wet periods, which, however, did not trigger slides (Fiorillo and Wilson, 2004). The recorded precipitation at Lauro amounts to 173 mm in the last 48 h, with a mean intensity of only 3.6 mm/h. Fiorillo and Wilson (2004) have computed a return time of 22 years for the 24 h rainfall at Lauro, 4 years for the 2 days rainfall at the Sarno gauge, and 5 years for the 2-days rainfall at the Castellammare gauge.
Besides the return time, that is not very considerable if compared to the severity of its consequences, the constancy and temporal extent are the most remarkable characteristics of the May 1998 precipitation. Rainfall lasted from 28 April to 6 May with very few and short interruptions.
The evaluation of thresholds of rainfall triggering landslides is a hot topic Guzzetti et al., 2008, Crosta and Frattini, 2002, Frattini et al., 2009, Picarelli et al., 2009. There are many approaches to this problem based on the integration of hydrologic models and slope stability analysis, either in the form of local or generalized areal models. For the studied area, many authors (Amanti et al., 1999, Calcaterra et al., 2000, Chirico et al., 2000, Crosta and Dal Negro, 2003, Fiorillo and Wilson, 2004) have tried to determine a reliable empirical relationship between duration and intensity of rainfall to identify the threshold for landslide occurrence in the Campanian pyroclastic deposits. Recently Greco et al., 2010 have approached the problem of early warning of rainfall-induced landslides in unsaturated granular soils by means of a laboratory flume tests. They show that, in consequence of soil volumetric collapse, suction values are smaller than expected from laboratory water retention curves favoring the triggering of instability phenomena. Pagano et al. (2010) have proposed a simplified approach able to predict the evolution of key factors affecting slope stability with special reference to granular pyroclastic soils. It is rather difficult and is still a research challenge, to identify a regional approach or a simplified local model that can capture the complexity of hydro-geomechanical behavior of these soils and be operational, that is being able to produce reliable predictions of landslide initiation. After the event occurred in May 1998, it could be observed that a dense concentration of landslide initiation zones was located at higher elevations and was characterized by a wide areal extent. Hundreds of initial failures, within a relatively short time span (few hours) could be counted (Guadagno et al., 2005) on angles steeper than 40° at higher elevations. All these almost simultaneous failure phenomena can be considered an indicator for widespread loss of soil strength. The area is characterized by the presence of pyroclastic mantles, deposited by the explosive eruptive activity of Phlegrean Fields (Campi Flegrei) and Summa-Vesuvium, on the carbonatic Mesozoic substratum of the Pizzo D'Alvano relief that is often very steep at higher elevation. Deposition took place under different conditions leading to an alternation of several pumiceous and ashy deposits (e.g. Rolandi et al., 2000, Crosta and Dal Negro, 2003). This sequence of layers have different grains size ranging from clay to gravel-sized lapilli. At a local scale, this character can reflect significantly on the hydraulic properties and behavior of soils, especially since layers with great contrast in hydraulic conductivity are arranged in alternating sequence (Basile et al., 2003, Crosta and Dal Negro, 2003, Mancarella and Simeone, 2008). This might widely influence water circulation and moisture redistribution. Moreover, measured friction angles in fine layers are lower than slope angles (Cascini et al., 2008). Finer soils are almost permanently unsaturated (Cascini et al., 2008), this can provide an ephemeral defense from instability due to suction, that can likely disappear if saturated conditions are reached in consequence of rainfall.
The presence of natural scarps and artificial road cuts interrupts the lateral stratigraphic continuity of the coverings and allows for kinematic freedom in the downslope direction (Crosta and Dal Negro, 2003, Guadagno et al., 2005). It has been observed often that the initial failures are located just above gully heads and upslope of mantle discontinuities (Guadagno et al., 2005).
The discussion on the triggering mechanisms of this catastrophic event is still open and a brief review will follow.
The failure surface is generally located within the pyroclastic mantle and not at its base. This would seem to disprove, as noted by Crosta and Dal Negro (2003), the possibility of failure mechanism due to hydraulic pressurizing of permeable layers. In that case, the slip surface should be located at the contact between the bedrock and the pyroclastics and preferably in the lower part of the slope, where springs are numerous. That is not usually the case. Moreover, the carbonate bedrock is evidently karstified and it appears unlikely that important under-pressures may develop diffusely and at the highest altitudes. Elsewhere in the area, this mechanism has, however, been observed and reported as responsible for landslide triggering, as evidenced by the appearance of springs (Cascini et al., 2008). The study of Crosta and Dal Negro (2003) excludes the need for groundwater flow from the underlying bedrock for instabilities to occur: the formation of pore pressure pulses is possible in pumice layers in correspondence of stratigraphic discontinuities as response to the rainfall event of 28 April and 5 May. They hypothesize the formation of perched water tables within pumice layers. Such a model would suggest detachment to be localized in the areas where a discontinuity is relevant in hydraulic properties of layers. The authors of the present work agree that the reason for the original detachment resides in the hydraulic contrast between pyroclastic layers that often can be very sharp (Mancarella and Simeone, 2008). However, it is noteworthy that saturated conductivity and trasmissivity in the mantle's coarser pumices are hundreds of times higher than the average rainfall intensity during the event (3–5 mm/h). Therefore it is possible but not likely that pore pressure in coarser layers builds up to such a value to uplift upper soil layers. Even if this mechanism may occur, due to specific site conditions, in the authors’ opinion, it cannot fully explain the quasi-contemporaneity, the number and the areal extent of the landslides which occurred in May 1998. Guadagno et al. (2005) relates the initial failures to the stratigraphic discontinuity due to road cuts and hypothesize that, because of the fertility of the pyroclastic material, a new soil would form on the steep along the cuts, creating a relatively impervious cover, able to block water outflows and allow for greater saturation of the soils located upslope of the cuts. As ruptures are often located at 2–4 m upslope of the discontinuities (Guadagno et al., 2005), this could be a useful explanation of the numerous slopes that are located upslope the cuts but it would require widespread impermeable covers to impede the outflow along the cuts.
The authors of the present work, like others (see Section 2), believe that landslide initiation can be attributed to a generalized increase in saturation, but relate triggering conditions to the peculiar infiltration patterns induced by the rise of capillary barrier effects within the least permeable soil layers (Fig. 1). Such phenomena can take place in a sequence of unsaturated layered soils, characterized by contrasting hydraulic properties (Shackelford et al., 1994). The presence of a capillary barrier can highly disturb infiltration processes and cause moisture accumulation and diversion as the capillary tension at the interface must be equivalent in the two layers to allow proceeding of the downward flux. An infiltrating front in the upper finer layer might trigger a landslide in the finer layer by a decrease in suction, and a capillary barrier might thus enhance this phenomenon since the consequent widespread moisture accumulation in the finer soil can be responsible for a total loss of cohesion. Thus, a crucial contribution to stability can be lost.
These phenomena might be responsible for widespread effects on slope saturation or cause localized increase of water content because of water diversion along contact planes.
Section snippets
Capillary barrier effects and their possible role on infiltration processes and slope stability
At the light of the proposed mechanism for slope instability activation, the aim of this study is to point out the retention mechanisms of infiltrating water due to the build-up of capillary barriers (Shackelford et al., 1994) at the interface between fine and coarser pyroclastic layers.
The authors believe that capillary barrier effects may play a relevant role in the triggering of the aforementioned widespread landslide phenomena as the hydraulic contrast between pyroclastic layers is often
Experimental tests and capillary barrier effects in pyroclastic soils
A series of in situ and laboratory tests has been performed to characterize the grain size distribution along the soil profile, the permeability of the different layers, and the possibility to have capillarity barriers acting as hydraulic impedance in the soil sequence. A detailed description of in situ and laboratory test is reported in Mancarella and Simeone (in review). In situ permeability measurements were carried out by means of a Guelph permeameter on two sites of Pizzo D'Alvano just
Calibration of Van Genuchthen parameters
Data from experimental tests were first used to calibrate a numerical column infiltration model in order to tune characteristic curve parameters that could be subsequently used in fully 2-D simulations. The numerical model used was VSD2, developed by Lappala et al. (1987), and based on integration of Richards’ equations to simulate infiltration processes in variably saturated soils. The interested reader is referred to Lappala et al., 1987, Healy, 1990, Hsieh et al., 2000, Halford, 1997, McCord
Summary and conclusions
An experimental and numerical analysis on capillary barrier effects has been presented in this paper with a specific reference to the case history of Sarno slope in Italy that was affected by the catastrophic landslide of 1998. The general idea proposed here is that whenever there are fine unsaturated soils covering more permeable ones, i.e. a layered sequence of materials showing sharp contrast in hydraulic properties, capillary barrier phenomena may favor the increase in saturation within the
Acknowledgements
Special thanks must be addressed by the authors to Prof. Oldrich Hungr, from University of British Columbia (Vancouver, Canada) for the precious advice provided in preparation of the paper and to Dr Francesco Fiorillo (University of Sannio, Italy) for the support received during in situ tests. The authors wish to thank also Prof. Giovanni B. Crosta (University of Milano Bicocca, Italy) and Prof. Francesco Maria Guadagno (University of Sannio, Italy) for the useful discussion on the present
References (49)
- et al.
Soil hydraulic behaviour of a selected benchmark soil involved in the landslide of Sarno 1998
Geoderma
(2003) - et al.
Typical source areas of May 1998 flow-like mass movements in the Campania region, Southern Italy
Engineering Geology
(2008) - et al.
Rainfall induced debris flows in pyroclastic deposits, Campania (southern Italy)
Engineering Geology
(2004) - et al.
Soil water content and suction monitoring in model slopes for shallow flowslides early warning applications
Physics and Chemistry of the Earth, Parts A/B/C
(2010) - et al.
Some aspects of the initiation of debris avalanches in the Campania Region: the role of morphological slope discontinuities and the development of failures
Geomorphology
(2005) - et al.
Parameter estimation for unsaturated flow and transport models—a review
Journal of Hydrology
(1987) - et al.
Visualization experiment to investigate capillary barrier performance in the context of a Yucca Mountain emplacement drift
Journal of Contaminant Hydrology
(2003) - et al.
Relazione tra gli eventi franosi del maggio 1998 e la piovosità dell'area di Sarno
- et al.
Geotechnical characterization of pyroclastic soils involved in huge flowslides
Geotechnical and Geological Engineering
(2005) - et al.
The influence of meteoric events in triggering shallow landslides in pyroclastic deposits of Campania, Italy