Ricardo Casale · Claudio Margottini (Eds.)
Floods and Landslides: Integrated Risk Assessment
With 150 Figures and 30 Tables
123
Chapter 24
Methodological Approach in the Analysis of Two Landslides in a Geologically Complex Area: The Case of Varenna Valley (Ligury) P. Brandolini · S. Nosengo · F. Pittaluga · A. Ramella · S. Razzore
24.1
Introduction This paper provides a brief study of Varenna basin (Ligury) through the analysis of its instability features and mass movement proneness; this area, during the five years from 1991–1995, has gone through frequent episodes of flood damages and landslides. The quick slide of topsoil debris and its displacement into the stream has dramatically increased the over-alluvion along the riverbed reaches with a strong dissipation of energy, and has made the river more liable to an esondation; besides, in correspondence of a restricted hydraulic section it reaches, several new bank erosions, being themselves responsible for the triggering mechanism in the slopes. For mountain basins, as well as the present one, it is necessary to define geomorphologic and geotechnical characteristics of the landslides, rainfall triggering threshold and thickness critical rate of the materials involved in the mass movements, in order to set out new activities for the reduction of the flood and landslides integrated hazards. These criteria have been used in the analysis of two mass-movements which occurred in Varenna valley during the autumnal rains in 1993–1994; they are still representative of the geological characteristics of most of the examined basin: the figures derived from the study show that these typical landslides always occur in presence of serpentine-schists strongly influenced by tectonic movements and widely spread along the whole basin. The geomorphologic and geotechnical data have been compared to those derived from the survey for the assessment of the slope stability, using the “back analysis” method suggested by Jambu. The results allow to define several degrees of proneness to the instability and, as well as, the degree of geomorphologic hazards in all the above-mentioned area. 24.2
A Geological and Geomorphological Outline of Varenna Valley Varenna valley, along which the homonymous stream runs, lies in the western part of Genoa, near Pegli, it extends for 23 km2 and it stretches for 9 km perpendicular to the coast line, with a maximum width of 4.5 km (Fig. 24.1). The highest points are those belonging to the Po valley-tirrenic watershed between the M. Pennello and Mt. Fasciallo ridge-line, slightly inferior to 1 000 m a.s.l. This area is geologically influenced by the overlapping of the two units belonging to the “Voltri Group” (metaophiolites) and “Sestri-Voltaggio group” (serpentinites, dolomite rocks, marly limestones, phyllite, slates). Both the structural order, formed by two main lines with approximate N-W orientation, and the local lithological groups have strongly modified the slope morphology and the hydrographical network; the two factors are responsible for an evi-
358
P. Brandolini · S. Nosengo · F. Pittaluga · A. Ramella · S. Razzore
1 2 3 4
N
T. B
i sa
gn
o
na
0
5 km
r vi T. N e
T. S t
ur l a
vagna
sa
T. Polc evera
T. C er u
T. Chiara
T. V a
o
Fiorino 44°26'00''
ren
ir T. Le
44°29'00''
Mad. della Guardia
Old Port of GENOA
8°49'00''
9°02'00''
Fig. 24.1. Location of studied area: 1 Varenna valley; 2 analysed landslides; 3 pluviometric stations; 4 watersheds
dent valley asymmetry (right slope being more developed than left one). The most frequent types of mass movements are “mud and debris flows”, of little thickness (up to 2 m) and generally composed by calc-schists and schists; landslides of higher complexity and volume have been found by tectonised serpentine-schists and calc-schists contact layers or inside the serpentinites. From their first triggering on they are characterised by a great regressing and widening activity. In autumn, owing to unfavourable meteorological and geomorphologic conditions, approximately 50% of the examined area was exposed to the integrated flood and landslides hazard. 24.3
Rainfall Analysis and Return Periods Estimation The valley climate is strongly influenced by its orography. A peculiar slope exposure, as well as a brief distance between the Po valley-tirrenic watershed and the coast-line provoke strong changes in the climate, also influenced by the altitude; the rainfall has therefore, a discontinuous behaviour, with higher monthly mean values recorded on tops, with an absolute maximum value in winter and one in springtime. The rainfall analysis has been carried out using data furnished by the weather stations in Fiorino, (slightly western to the Val Varenna, 236 m a.s.l.) and at Madonna della Guardia, eastern to the watershed next to the Val Polcevera, 809 m a.s.l. The monthly rainfall mean values in 1991, 1992 and 1993, measured near the station in Fiorino, show that they have dramatically increased in comparison with the long set
CHAPTER 24 · Methodological Approach in the Analysis of Two Landslides
359
Distribution per hour of rainfalls Madonna della Guardia Pluviometric station – 23th September 1994 120
rainfalls (mm)
100 80 60 40 20 0 0
2
4
6
8
10 12 14 16 18 20 hours
S1 22
24
Monthly distribution of rainfalls Fiorino pluviometric station – 1991/94 and average of 1921/50 1200 21–50 91 92 93 94
rainfalls (mm)
1000 800 600 400 200 0 J
F
M
A
M
J
J
A
S
O
N
D
months Annual distribution of rainfalls Fiorino pluviometric station – 1991/94 and average of 1921/50 3000
rainfalls (mm)
2500 2000 1500 1000 500 0
21–50
91
92 years
93
94
S1
Fig. 24.2. Distribution of rainfalls
of rainfall mean values recorded during the historical thirty years from 1921–1950 (Fig. 24.2): in particular, it should be remarked the high value recorded in September and October 1993 (623 mm and 754 mm compared to the thirty years rainfall average, i.e. 164/210 mm).
50
1
Tr = Time thresholds (a)
5
380
1.1
Fig. 24.3. Diagram “FrequenceRainfall” (Gumbel Pluviogram)
100
P. Brandolini · S. Nosengo · F. Pittaluga · A. Ramella · S. Razzore
1
360
360 24 hours 340 320 300 12
280 260
6
240
P = Rainfall (mm)
220
3
200 180 160 140 1
120 100 80 60 40
0.999
0.997
0.99
0.995
F = Frequence
0.98
0.95
0.9
0.8
0.5
0.1
0
0.001 0.01
20
Annual rainfall peaks related to the 24 h, recorded by the raingauge at Madonna della Guardia, underwent a statistic elaboration with the aid of the Gumbel Method. Return periods (Tr) derived from the diagram “Frequence-Rainfall” (Fig. 24.3) using maximum rainfall (P) and the corresponding duration (D) data in terms of hourly rainfall intensity (I). 24.4
Landslides Analysis The meteorological events of high intensity have the strongest effects on bedrock topsoils, in which first-time landslides are triggered; the high amount of quick-filtering water brings to saturation the most superficial layers, being also the reason for their collapse. First-time landslides made from “soil slip”, and “mud and debris flow” have been detected on serpentine slopes; they generally become active along altered talc-schist terraced lands, on calc-schists and prasinites slopes, on limestone rocks.
CHAPTER 24 · Methodological Approach in the Analysis of Two Landslides
361
A correlation between superficial landslides and terraced slopes has also been found. Translational and roto-translational ground and debris displacements are found in great number and localised all along the valley, even in proximities of ridges; they are brought about either by bank erosions and the consequent widespread landsliding and rilling, and an increased load on slopes, owing to local groves (mainly chestnut trees). 24.4.1 Chiesino Landslide This landslide is next to Chiesino settlement, on the right bank of Varenna Torrent, (Fig. 24.4). The mass movement event occurred suddenly, on the 4th November 1994, following intense rainfall and without any previous evidence of dangerous instability. The raingauge data show that rainfall values referring to the days immediately antecedent the landslide event were about 290 mm, 72% of which refer to the 4th November (more or less 210 mm). The landslide reactivated an old debris accumulation, which underwent a secular stabilisation in a terrace leaning on a serpentine-schist layer belonging to the “Voltri Group” (see Table 24.1).
Fig. 24.4. Chiesino landslide
362
P. Brandolini · S. Nosengo · F. Pittaluga · A. Ramella · S. Razzore
Table 24.1. Typical parameters of material in the landslide body –1
γ
γ sat
Wn (%)
Sr (%)
Wl
Wp
Ip
n
neff
k (m s )
1.97
2.02
22.8
0.54
38
28
10
0.61
0.2
5.8 × 10
–7
These are the most relevant morphometric features of this mass movement: § § § § §
Longitudinal mass development: 140 m; Mass movement body medium width: 60–70 m; Mass thickness: 5–8 m; Landslide global surface area, measured on plan, of about 8 000 m2; Landslide body estimated volume: 40 000 m3.
The landslide can be classified as a rock and debris flow, for which both an undermined slope toe and a saturated topsoil are equally responsible. The material collapsed is mostly made from incoherent terrain (slime gravels) deriving from an alteration in the serpentinite-schist rocks. The granulometric curves result well graded, with a high amount of fine material, varying from 12–30%. On the above mentioned materials several shear tests have worked out a friction angle ϕ‘ ≈ 35° and a cohesion c’ = 0 24.4.2 Carpenara Landslide The landslide occurred near Carpenara, on the left bank of Torrente Varenna, at the confluence with a rill (Fig. 24.5) and affects a part of serpentine-schists altered soils belonging to the “Voltri Group”, of about 7 m of maximum thickness. The slope shows a very wide alteration band in the serpentinte-schists, covered by a coarse debris topsoil. The mass movement event occurred on the 24th October 1993, one month after the intense showers of the last days of September, followed by a period of downpours (about 25 mm per day) for a total amount of 1 350 mm. It should be noted that 350 mm of rain fallen on the 24th September, more or less one month before the landslide event, did cause a severe flood without any landslide trigger sign. These are the relevant morphometric data referred to the landslide: § § § § §
Maximum longitudinal development: 100 m; Mass movement body approximate width: 60–80 m; Maximum thickness: 5–7 m; Maximum areal extension of the landslides, measured on plan: about 4 500 m; Mass movement body estimated volume: 10 000 m.
The landslide can be classified as a rock and debris translational and rotational flow, while the entire slope shows some evidence of instability under the appearance of little landslide scarps and small traces of debris flow.
CHAPTER 24 · Methodological Approach in the Analysis of Two Landslides
363
Fig. 24.5. Carpenara landslide
Table 24.2. Typical parameters of lime material and in the landslide body –1
γ
γ sat
Wn (%)
Sr (%)
Wl
Wp
Ip
n
neff
k (m s )
2.04
2.17
18.1
0.84
29
20
9
0.4
3
2.0 × 10
–7
Materials mostly consist of sands and lime gravels, well graded and showing great affinity with the Chiesino samples (see Table 24.2). Also, in the landslide body, it has been found the presence of thin levels (10–30 cm) mostly made from lime material and with a very low degree of stress resistance. The recorded parameters are resumed in the following table. The results of dimensional analysis using the “thin section” method show that the material is very altered, it ranges uninterruptedly from “very fine” (1/1 000) to “coarse” and features a very low stress resistance. The fragments are mostly made from highly oxidised serpentinites, sometimes in lamella and sometimes in fibroid shape. A certain amount of amphibole is also present.
364
P. Brandolini · S. Nosengo · F. Pittaluga · A. Ramella · S. Razzore
On the serpentinites, particles of some dimensions also show an oxidation rim more compact and more resistant than the original material, hint of an advanced pedogenesis. 24.5
Soils Stability and “Back Analysis” The available data enabled the back analysis on stability, exploiting the “equilibrium limit” method proposed by Jambu, with imposed slide-surfaces. Chiesino landslide tests (y = 1.97 t/m3, ϕ‘ = 35°, c’ = o) show that in absence of the water table the slope parameters maintain themselves within the stability threshold (Fs = 1.4); on the contrary, the presence of the water table is responsible for the steep decrease of the security factor below unity (Fig. 24.6). With reference to the circumstances preparing the landslide event (after four days of intense rains) and to the pre-existing natural conditions (steep acclivity of 33°, untended terraced lands no longer tilled) the landslide event may be ascribed to both an increase in the water table level and the lack of intervention for the surface-waters drainage. The analysis carried out on the Carpenara landslide have been more difficult for the presence, in the serpentinites topsoil, of strongly altered fine material, (mostly lime) with very poor mechanic qualities, (ϕ‘ = 14°, c’ = 0). The reconstruction of the geometric disposition of the cited material inside the topsoil has not been possible, for the lack of previous in-situ surveys. With the help of the back analysis method it has been possible to verify that the mentioned material can trigger a landslide only when placed in a rather continuous level, being therefore, able to influence in a remarkable way the slope stability. 24.6
Landslide Trigger Studies on thickness and critical rainfall rates, with reference to return periods, using the “Pradel and Raad Model” have been achieved in order to complete the present survey. This model finds application only when the rainfall is considered the unique responsible for the seepage triggering mechanism, with no consideration to the uphill recharge. The following necessary conditions are needed for the application of the model: § § § §
Constantly wet soil surface; Constant permeability in the saturated terrain; Linear saturation front; Constant negative pressure above saturation limits. Parameters valid for both the events are the following ones:
§ § § §
D10 = 0.1 mm; Pris = 0.18; Nsat = 0.2; K = 5.8 10–7 m/s.
CHAPTER 24 · Methodological Approach in the Analysis of Two Landslides
365
170
a
Label COLTRE
Soil Type No 1
Total Saturated UnitWt. UnitWt. 3 3 kN/m kN/m 17
Cohesion Friction Pore Intercept Angle Pressure kPa (deg) Param.
17
0
35
Pressure Constant kPa
Piez. Surface No
0
W1
0
150
1
W1
1 130 1 Elev. (m) 1 110
1
90
70 W1 0 170
b
# 1 2 3 4 5 6 7 150 8 9 10
20
FS 0.98 0.98 1.01 1.01 1.02 1.02 1.03 1.04 1.04 1.05
Label COLTRE
40 Soil Type No 1
60
Total Saturated UnitWt. UnitWt. 3 3 kN/m kN/m 21.7
80
Cohesion Friction Pore Intercept Angle Pressure kPa (deg) Param.
21.7
0
25
100
Pressure Constant kPa
Piez. Surface No
0
W1
0
120
1 10 7 8 6 3 1
130 Elev. (m)
1 W1
1 W1 110 1 W1
W1
90
W1
70 W1 0
20
40
60
80
100
120
Fig. 24.6. Back analysis a Chiesino landslide: hypothesis of a slope with a water table; b Carpenara landslide: hypothesis of a slope with a water table
By varying the terrain critical thickness rates, and by keeping the permeability rate (k) constant, it has been possible to calculate the intensity and duration minimum values necessary to the saturation, with reference to the different thickness rates, and afterwards to deduce the saturation conditions line. The envelop lines of maximum rainfall intensities, referred to the various return periods, have been derived from the
366
P. Brandolini · S. Nosengo · F. Pittaluga · A. Ramella · S. Razzore
Fig. 24.7. Diagram “Intensity/Duration of rainfall” with minimum intensity (Imin) and duration (Tmin) values for different topsoil thickness and correlation with the several return (Tr). In this diagram are indicated the graphic areas representing the relative critical rainfall able to saturate topsoil of equivalent thickness (0.5; 1; 2)
“Frequence-Rainfall” graphic, (Gumbel Pluviogram). By transferring the obtained figures on the bilogarithmic scale diagram, it has been possible to build up the “Intensity-Duration” graphic (Fig. 24.7) from which a first indication on the geological haz-
CHAPTER 24 · Methodological Approach in the Analysis of Two Landslides
367
ard for different topsoil thickness rates with all those geotechnical characteristics is obtainable. The line showing duration and minimum intensity values necessary for the saturation (I, T, K) intercepts those related to several return periods, in points S1, S2, S3 and S4 (Time thresholds) and singling out thickness rates for which minimum saturation conditions and return periods are strictly related. Figure 24.7 indicates, for different thickness rates, (0.5 m, 1 m, 2 m) the graphic areas representing the relative critical rainfall able to saturate topsoil of equivalent thickness. It should be noted how highly intensive but brief rainfalls are capable of saturating only thin topsoils (B = 0.5 m) while lesser intense but long-lasting rains saturate thicker soils. For instance, terrains with a thickness ranging from 2–4 m may be saturated only by rains with a return period varying from 10–20 years, while soils thicker than 5 m need rain with a return period longer than 100 years. Graphic analysis has made possible the detection, for both the examined landslides, of critical thickness rates related to the precipitation of the days foregoing the events. In fact, 208.6 mm in the 24 h (Tr 5–10 years), equivalent to the rain values recorded on the 4th November 1994, may bring to saturation a stratum with the thickness of 0.5 m but considering also the four days antecedent the landslide event (Pmax = 287.8 mm; T = 96 h) the thickness rate increases up to 1.5 m The rainfall data related to Carpenara landslide show a high intensity average figure rainfall event at the end of September (Pmax = 350 mm in four days and a peak equal to 350.8 mm in the 24 h). The graphic clearly shows how these figure fall in the area representing the critical precipitation for the return period within 20 and 30 years, able to saturate soil with a thickness between 1 and 1.5 m. The low intensity rains of the following days maintained the soil saturated by adding weight and enabling a deeper water infiltration. Unlike the Chiesino landslide for which even the immediately previous rainfall of high intensity is held responsible, the Carpenara landslide has been essentially triggered by long lasting previous precipitation. 24.7
Conclusions The figures obtained with the use of Pradel and Raad Model show that only small portions of topsoil are liable to be saturated (ranging from 0.5–1.5 m). An empirical assessment is also possible, since the severe downpours cause damages only to the colluvial cover, which is quickly brought to saturation by reason of an insufficient drainage of the underlying strata (almost impermeable and prone to behave as preferential slide surfaces). A comparison between the former values and the ones obtained through in-situ surveys shows that the two examined landslides displaced a larger amount of material (4–7 m in Chiesino, 6–7 m in Carpenara). A possible explanation lies in the fact that high intensity rainfalls trigger first-time landslides, above all when in presence of two different strata with a sharp contrast in the permeability coefficient, so that a seepage water flow may run down parallel to the slope.
368
P. Brandolini · S. Nosengo · F. Pittaluga · A. Ramella · S. Razzore
Therefore, other causes must be held responsible for deeper landslides triggering. The two events suggest the following observations: § Chiesino landslide: there is no evidence of poor quality materials along the slide surface. The landslide mechanism is therefore, to be put down to the intense rainfall occurred the previous day, which brought about a double increase in the water table and in the Varenna Torrent level, the latter responsible for the slope toe undermining. § Carpenara landslide may have been triggered by the action of two contributory causes, the former consisting in a long-lasting rainfall (during the whole month of October) and the latter being the presence of an altered and poor-quality material which may have acted as impermeable stratum and responsible for the saturationfront deepening. In the examined basins, geologic and geomorphologic studies lead to assert that the most remarkable events always originate from tectonic lines in which serpentinites come in contact with other lithologies. A detailed geotechnical survey on the tectonised terraces areas, together with a hydrologic verification, allow the definition of both the propensity to landslide for all the areas with homogenous geologic and geomorphologic features, and the landslide trigger threshold values for each individual rainfall rate, to a detailed scale. This methodology is particularly useful when researches need to define the criteria aiming at the reduction of flood and landslide integrated hazard.
References Benini G (1990) Sistemazioni idraulico-forestali. UTET, Torino Brandolini P, Ramella A (1994) Eventi alluvionali e dissesti idrogeologici: il caso della Val Varenna (Liguria). Studi Geografici in onore di D. Ruocco, Loffredo Editore, Napoli Brandolini P, Ramella A (1997) Processi erosivi e fenomeni di dissesto nei versanti terrazzati delle valli costiere genovesi, Atti Convegno Geografico Internazionale “I valori dell’agricoltura nel tempo e nello spazio”, Rieti 1–4 nov. 1995, vol II, pp 339–355 Bromhead EN (1986) La stabilità dei pendii. Dario Flaccovio (ed), Blackie and Son LTA Carrara A, D'elia B, Semenza E (1985) Classificazione e nomenclatura dei fenomeni franosi, Università di Bari, Ist. Geologia Applicata e Geotecnica – Geologia Applicata e Idrogeologia, vol XX., parte II Castagny G (1985) Idrogeologia, principi e metodi., Libreria Dario Floccovio (ed) Gostelow TP (1991) Rainfall and landslides, Preventio and control of landslides and other mass moviments, Almeida-Texeia ME et al. (ed) Ministero dei Lavori Pubblici (1953–1985) Annali Idrografici. Servizio Idrografico, Sez. Autonoma di Genova Pradel D, Raad G (1993) Effect of permeability on surficial stability of homogeneus slopes. Journal Geotech. Enging., ASCE Raviolo PL (1993) Il laboratorio Geologico. Procedure di prova: Elaborazione, Acquisizione dati, Editrice Controls, Milano Renard KG, Freimund JR (1994) Using monthly precipitation data to estimate the R-factor in revised USLE, Journal of Hydrology Skempton AW, DE Lory FA (1957) Stability of natura slopes in London Clay. Proc 4th Int. Conf. on Soil Mech. and Found. Eng Tropeano D et al. (1993) Gli eventi alluvionali del 22–27 Settembre 1992 in Liguria. Studio idrologico e geomorfologico. Supplemento a GEAM – Geoingegneria Ambientale e Mineraria, Anno XXX, No 4
