Volume 13 Issue 4 - April 2, 2010 PDF
Internal Soil Moisture And Piezometric Responses To Rainfall-Induced Shallow Slope Failures
Department of Civil Engineering, College of Engineering, National Cheng Kung University
Font Enlarge
Better knowledge of internal soil moisture and piezometric responses in the process of rainfall-induced shallow slope failures is the key to an effective prediction of the landslide and / or debris flow initiation.  To this end, internal soil moisture and piezometric response of 0.7 m-deep, 1.5 m-wide, 1.7 m-high, and 3.94 m-long semi-infinite sandy slopes rested on a bi-linear impermeable bedrock were explored using a chute test facility with artificial rainfall applications.  The internal response time defined by the inflection point of the soil moisture and piezometric response curves obtained along the soil-bedrock interface were closely related to some critical failure states, such as the slope toe failure and extensive slope failures.  It was also found that the response times obtained at the point of abrupt bedrock slope decrease can be used as indicators for the initiation of rainfall-induced shallow slope failures.  The present study proposes an ‘internal response time’ criterion to substantiate the prediction of rainfall-induced shallow slope failures.  It is believed that the ‘internal response time’ reflects the overall characteristics of a slope under rainfall infiltration and can be as useful as the conventional meteorology-based threshold times.  The ‘internal response time’ theory can be generalized via numerical modeling of slope hydrology, slope geology and slope stability in the future.

Model slopes, 3.94 m-long, 0.7 m-deep, and 1.5 m-wide, were built at the upper portion of a 20 m - long and 6 m - high reinforced concrete chute with a rectangular cross section of 1.5 m - wide and 1.0 m – high and a slope angle of α = 29˚ (Fig. 1).  Model slopes were made using a locally available soil (Gs = 2.63, D50 = 0.21 mm; Gs : specific gravity; D50 : median particle size) classified as ‘SM’ based on the Unified Soil Classified System (USCS) proposed by the American Society of Testing and Materials (ASTM D2487).  100 mm-thick lift and a light vibratory compactor were used to construct the slope with a target moisture unit weight (γm) of 16-17 kN/m3 (see Fig. 1), corresponding to dry unit weight (γd) of 14.5-16.2 kN/m3, void ratios (e) of 0.6-0.78, and relative degrees of compaction of 73%-81% based on a standard laboratory compaction procedure.  The initial value of water content of the soil (ω = weight of water / weight of solids) was between 5% and 10%, equivalent to a degree of saturation (Sr) between 17% and 44%.  Figs. 2(a), 2(b), and 2(c) show the rainfall simulator, the base of flume and the soil slope at completion, respectively.  Figures 3(a) and 3(b) schematically show the plan and side views of the locations of moisture sensors used in this study.  Moisture sensor Nos. 1, 2, 5, 6, 9, and 10 (namely, M 1, M 2, M 5, M 6, M 9, and M 10) were placed at the soil-bedrock interface; M 3, M 4, M 7, M 8, M 11, and M 12 were placed 0.2 m below the slope surface, as shown in Figs. 3(a) and 3(b), respectively.  Piezometers (P1, P2, P3, P4, P5 and P6) were placed in three trenches immediately below moisture sensors M 1, M 2, M 5, M 6, M 9 and M 10, respectively, as shown in Fig. 3(b).  In the following, (x, y) is used to represent the location of sensors; x is the horizontal distance from the slope toe (= S); y is the vertical distance from the soil-bedrock interface.   
Fig. 2(a) A front view of the sprinkler and digital cameras.
Fig. 1 Test slope geometries and artificial rainfall simulator.

Fig. 2(c) A front view of the sand slope before the test.
Fig. 2(b) A front view of the concrete flume with three trenches at its base for porewater pressure measurement.

Fig. 3(b) A side view of the locations of piezometers and moisture sensors.
Fig. 3(a) A planar view of the locations of piezometers and moisture sensors.

Figures 4(a), 4(b) and 4(c) show the moisture response for the MS at S = 0.95 m, 1.7m and 2.6m, respectively.  These figures show that soil water content response curves are characterized by abrupt rising segments with clear inflection points.  In addition, times needed to reach the inflection point for M (0.95, 0.5), M (1.7, 0.5) and M (2.6, 0.5) are always smaller than those for M (0.95, 0), M (1.7, 0) and M (2.6, 0), reflecting time lags induced by the downward propagation of the ‘wetting front’. 
Fig. 4(b) Soil water content measured on the soil-bedrock interface at S = 1.7 m (Test No.4, I = 70 mm/hr ).
Fig. 4(a) Soil water content measured on the soil-bedrock interface at S = 0.95 m (Test No.4, I = 70 mm/hr ).

Fig. 5 Measured water pressure heads along the soil-bedrock interface (Test No. 4, I = 70 mm/hr)
Fig. 4(c) Soil water content measured on the soil-bedrock interface at S = 2.6 m (Test No.4, I = 70 mm/hr ).

Figure 5 shows an example of porewater pressure head (hp) response in the same test discussed in Figs. 4(a) - 4(c).  Similar to the soil moisture response, abrupt increasing porewater pressure heads can also be seen.  Comparing the response of M (0.95, 0) in Fig. 4(a) and the response of P (0.95, 0) in Fig. 5, it is noted that the response times for these two types of measurements are similar.  The above observations suggest that both soil moisture sensors and piezometers are effective in detecting the arrival of the wetting front. 

To describe the infiltration process and the slope failure regarding the water content distribution inside the slope, inflection points for the ω vs. time (t) curves were examined.  The time at the first inflection point of the ‘ω vs. t’ curve measured along the soil-bedrock interface is defined as ‘Tm1’ and the time for the maximum value of ω in a ω vs. t curve measured along the soil-bedrock interface is defined as ‘Tmp’.  The time of first inflection point and peak value of ω obtained from the ‘ω vs. t’ curve obtained at 0.2 m-below slope surface location are denoted by Tm1(soil) and Tmp(soil), respectively.  This ‘internal response time’ analysis focuses on critical moments regarding slope failures and soil moisture (or porewater pressure) response.

Figs. 6(a) and 6(b) schematically shows the key response times for soil moisture and porewater pressure, respectively, associated with rainfall infiltration.  Response times defined by Tm1(soil) and Tmp(soil) are typical for the soil moisture sensors at 0.2 m-below slope surface; Tm1, Tmp, Tw1 and Twp are typical for those at soil-bedrock interface.  Figure 7 shows typical examples of response times as a function of distance from the slope toe (S) obtained in the case of rainfall intensity, I = 70 mm/hr.  Response times are in the following sequence :

(1) Tm1(soil) → (2) Tm1 → (3) Tmp

The scenario of wetting front progress described above may suggest that :
(1) Downward wetting front reaches 0.2 m-depth → (2) Downward wetting front reaches impermeable base → (3) Peak moisture state (and / or saturated state) attained at the impermeable base.
The rainfall-induced soil moisture response scenario discussed above is similar to the experimental evidences described by previous researchers. 
Fig. 6(b) Schematic pressure head response curves and the threshold times.
Fig. 6(a) Schematic soil water content response curves and responding times.

Fig. 7 A typical example of responding times based on soil moisture measurements obtained at various locations of the slope subjected to I = 70 mm/hr.

Figures 8(a)-8(d) show some critical moments of retrogressive failures of the slope subjected to a rainfall intensity I = 70mm/hr.  Fig. 8(a) is the slope immediately before the rainfall test.  Fig.8(b) shows that the toe of the slope is slightly washed away by the seepage at the toe of the slope.  This condition is called ‘minor toe wash-out’ at which only a very small amount of slope mass is washed away.  Starting from about this moment, a notable seepage outflow from the toe of the slope can be observed.  No surface runoff was observed throughout the tests performed in the present study, and the seepage outflow observed at the slope toe was considered a result of interflow along the soil-bedrock interface.  Fig. 8(c) shows a notable failure at the toe associated with the fluidization of the toe.  This failure condition is called ‘toe failure’ at which a scarp appears at S ≒ 0.24 m and a ratio (Pv) between the volume of slumped mass and the total slope volume, Pv is about 0.5%.  The term ‘scarp’ is used in the following to describe the steep cliff found at the uppermost location of the slumped soil mass.  The debris induced by the slope failure was washed away mainly by the seepage outflow from the slope toe in a short period of time.  Fig. 8(d) shows a typical moment of failure with an uppermost scarp reaching the second trench (S ≒ 1.7 m, S : horizontal distance from the toe of the slope), at which the volume of slumped mass of Pv ≒ 35% was prone to wash-away.  The slumped soil mass being prone to flow due to the synergistic effects of gravity, high soil water contents, stream flow, and other water sources has also been discussed by previous researchers.  Retrogressive failures presented in Figs. 8(a)-8(d) are summarized in Fig. 9, using the distance of the uppermost scarp from the toe (S) vs. t curves.  This figure shows that rates of retrogressive mass slump (in terms of ΔS/Δt, ΔS : changes in the location of the scarp, Δt : time interval) at the initial stage of collapsing (e.g., t = 50 -170min.) are related to the rainfall intensity.  However, this may not be true when the retrogressive failure proceeds further. 
Fig. 8(b) A close view of the slope with a minor wash-out of the toe (Test No.4, I = 70 mm/hr, t = 70 min )
Fig. 8(a) A close view of the near-toe of the slope before rainfall (Test No.4, I = 70 mm/hr, t = 0 min)
Fig. 8(d) A close view of the slope with a uppermost scarp around the second trench (S≒1.7m), showing a large extent of mass wasting ( Test No.4, I = 70 mm/hr, t = 104 min)
Fig. 8(c) A close view of the slope with a clear sign of toe failure(Test No.4, I = 70 mm/hr, t = 84 min )

Fig. 9 Progress of the scarp of failure mass observed in three tests.

Fig.10(b) Comparisons of internal response times based on the piezometric measurements at S = 0.95 m and critical times of slope failures.
Fig. 10(a) Comparisons of internal response times based on the soil moisture measurement at S = 0.95 m and critical times of slope failures.

Threshold times in terms of intensity and / or cumulative rainfall have been widely used in warning systems for mitigating disasters caused by slope failures.  These ‘rainfall thresholds’ criteria are often region-specific and are subject to high uncertainties.  The response time based on internal moisture and piezometric measurements (called ‘internal response time criteria’ here) may increase the accuracy of hydrology-based predictions of shallow slope failures (and / or debris flow initiation).  This work also offers a means of identifying the representative ‘time’ for the onset of retrogressive slope failures using ‘response time’ analyses.  Results of the analyses suggest that the response times (Tm1 and Tw1), which represent the arrival of a downward wetting front at the soil-bedrock interface, may serve as a precursor for the onset of slope failure (namely, a failure state between a minor toe wash-out and a toe failure).  Figs. 10(a) and 10(b) show typical examples of response times obtained at S = 0.95 m of the soil-bedrock interface as a function of rainfall intensity.  To show the relationship between the response times and the extent of retrogressive failures, critical times for various degrees of failures (such as : the toe failure, a scarp at S = 0.95 m, and a scarp at S = 1.7 m) are also presented as a function of rainfall intensity.  Fig. 10(a) shows the case of response times based on soil moisture measurements.  It is clear that the time of first inflection point on the soil moisture curve or the time of a wetting front arriving at the bedrock (Tm1, denoted by ‘×1’) consistently serve as an early indicator (or precursor) for minor toe wash-out; the time of peak soil moisture (Tmp denoted by ‘×2’) accurately simulates the time of slope toe failure at which Pv ≒ 0.5 %.  This observation is independent of the rainfall intensity, as shown in Fig. 10(a).  Fig. 10(b) shows a similar plot to that shown in Fig. 10(a), except that this figure is for the response time based on porewater pressure response.  In this figure, ‘×1’ denotes the time of the first inflection point on the porewater pressure vs. time curve, namely, Tw1, and ‘×2’ denotes the time of peak porewater pressure, namely, Twp, on the porewater pressure vs. time curve.  It can be seen that Tw1 approximates a critical moment between the minor toe wash-out and the toe failure; Twp simulates various degrees of slope failure between a toe failure and a proceeded slope failure with Pv > 35%.  Findings from Figs. 10(a) and 10(b) suggest that both soil-moisture-based and porewater-pressure-based internal response times, Tm1 and Tw1, obtained at the locus of the bedrock slope transition (S = 0.95 m in this study) are effective as indicators for the toe failure (or the initiation of shallow slope failures with a small value of Pv ≒ 0.05%).  This suggests that the internal-response-based criteria proposed here can be established, based on a similar framework as that used for conventional rainfall threshold criteria, to significantly improve the accuracy of slope failure predictions.

When quantitatively evaluating the stability of the slope, piezometers can be used in conjunction with soil moisture sensors in order to describe the full process of slope failure.  The deflection of subsurface flow lines around the transitional point of the bi-linear bedrock may cause a ‘mounding’ of subsurface flow and porewater pressure around that point (indicated by the high water content and porewater pressure at S = 0.95 m).  This in-turn reduces the effective stress around the point of bedrock deflection (or around the toe of the slope), causing toe instability and subsequent wash-away.
< Previous
Next >
Copyright National Cheng Kung University