Crosta and di Prisco 1065Fig. 9. (a) Grain-size analyses for a series of samples from the Groppello and Sairano sites. Samples correspond to those located alongthe stratigraphic profile in Fig. 8 and listed in Table 1. (b) Microscope images of the sand involved in the 1993 Groppello event(sample 6, as in Table 1 and Fig. 8).Investigating the mechanismof soils and the dislodgement of fine particles (Terzaghi andPeck 1967; Skempton and Brogan 1994).To understand this <strong>slope</strong> <strong>instability</strong> the following observationsmust be introduced:(1) A concentrated discharge of muddy water was observedat the toe of the <strong>slope</strong> in the late evening on June 27,© 1999 NRC Canada
1066 Can. J. Geotech. Vol. 36, 1999Fig. 10. Curves that define the dependency of (a) constantpermeability k and (b) volumetric water content on suction forthe different materials in the <strong>slope</strong>.a few hours before the <strong>slope</strong> failure. This clearly points to an<strong>erosion</strong> process at the toe of the <strong>slope</strong> which caused the removalof a specific size range of particles or all the availablesizes, increasing the pore size in the <strong>slope</strong> and resulting inturbulent water flow.(2) The limit of the failed area corresponds to the boundarybetween the rice and corn fields (Figs. 3, 4).(3) The upper sealing layer of organic topsoil was cut <strong>by</strong>20–30 cm deep incisions (Fig. 3). These incisions were created<strong>by</strong> tractor edged-blade wheels (or disk-shaped wheels)which are commonly employed for rice cropping in Italy.(4) Sub-horizontal cylindrical cavities (35–40 cm diameter)(Fig. 3d) located in the lower half (between 0.7–1.4 m)of the upper fine sand layer were preserved (Figs. 3c, 3d).The axial distance between each pair of pipes was about 2 m(Fig. 4), which is the distance between tractor wheels. Thetractor wheels cut the impermeable superficial layer causinginfiltration. Three pairs of pipes were found (Fig. 4), one coaxialwith respect to the channel and two lateral. This followsthe competitiveness principle mentioned in theliterature (Baker et al. 1990).(5) Partially preserved subcylindrical cavities (30–40 cmof diameter) with vertical axes were observed starting justbelow the bottom of the upper silt layer (Figs. 3c, 3d) andaligned with the overlying sub-horizontal pipes. The ensembleof sub-horizontal and vertical pipes suggests the occurrenceof a well-developed piping phenomenon. These verticalpipes suggest the presence of vertical cracks within the stiffsilt layer which allowed for the migration of <strong>seepage</strong> in theunderlying sand layer. These vertical cracks were likely theresult of both previous wetting–drying cycles and stress releasesubsequent to the river valley <strong>erosion</strong>.(6) Continuity of recharge was guaranteed <strong>by</strong> the shallowinterconnected ponds. The continuous recharge, hydraulicconductivity properties of the materials, number of infiltrationpoints, and duration of the event controlled the quantityof water flowing through the alluvial terrace.(7) The finer material forming the more external <strong>slope</strong>sector retarded water exfiltration, allowing for the increaseof water content in the coarser sand and the increase in waterpressure.(8) Two sectors with different <strong>slope</strong> gradients were recognizedin the deposition area. The steeper sector was locatedclose to the channel outlet, as described <strong>by</strong> Hagerty (1991a).The other sector, with a lower <strong>slope</strong> gradient, developed upto 100 m from the outlet (Fig. 4). This geometry is connectedto flow which spreads at the pipe outlet, with a consequentdrop in velocity and particle deposition close to theoutflow zone. The larger cone could be referable to rapidmass wasting or to the continuous reworking <strong>by</strong> the shallowrunoff fed <strong>by</strong> the floodwaters.(9) No vegetation survived within the area 30–40 m fromthe outlet, whereas beyond this area long reed stems wereonly partially submerged. This suggests violent depositionclose to the outlet and more fluid deposition farther from theoutlet.As mentioned in the Introduction and suggested <strong>by</strong> theabove observations, four distinct but strictly linked phenomenamay be assumed to have taken place: <strong>seepage</strong> <strong>erosion</strong>and tunnel scouring in the superficial layers, saturation ofdeeper soil layers, <strong>seepage</strong> <strong>erosion</strong> at the <strong>slope</strong> toe, and liquefactionof large soil masses. For the sake of simplicity,these phenomena will be discussed separately.Shallow <strong>seepage</strong> <strong>erosion</strong> and tunnel scouringHydraulic concentration may take the shape of pipes,which are tubular structures (Fig. 1), often ephemeral, frequentlysubject to collapse of their roofs with consequentformation of gullies with steep, amphitheatre-shaped heads.In the example presented here, the formation of multiplepipes was observed (Figs. 3, 4). The most important or leadingpipe, in the simpler case, <strong>induced</strong> the development ofsecondary pipes <strong>by</strong> increasing convergence and consequentlythe contributing area and the local hydraulic gradient. Competitionamong adjacent pipes (observation 5) was likely sostrong as to allow for the development of some of them.Before a detailed analysis of the observed uphill tunnelscouring phenomenon, it is important to observe the following:(1) The presence of superficial incisions and tensioncracks within the unsaturated soil layers <strong>induced</strong> localized© 1999 NRC Canada