+ model
ARTICLE IN PRESS
Engineering Geology xx (2005) xxx – xxx www.elsevier.com/locate/enggeo
Strug landslide in W Slovenia: A complex multi-process phenomenon Matjazˇ Mikosˇ a,*, Mitja Brilly a, Rok Fazarinc b, Mihael Ribicˇicˇ c a
c
University of Ljubljana, Faculty of Civil and Geodetic Engineering, Jamova 2, SI-1000 Ljubljana, Slovenia b Water Engineering Ltd., Teslova 30, SI-1000 Ljubljana, Slovenia University of Ljubljana, Faculty of Natural and Technical Sciences, Asˇkercˇeva 12, SI-1000 Ljubljana, Slovenia Accepted 24 June 2005
Abstract Large rock slides frequently cause secondary unstable phenomena. Many such instability processes happened after December 2001, when the Strug rockslide estimated at 95,000 m3 was triggered above the Kosecˇ village near Kobarid in the Julian Alps, W Slovenia. It was initiated at the contact between high permeable calcareous rocks (Cretaceous scaglia) thrusted over nearly impermeable clastic rocks (Cretaceous flysch). Soon after the rockslide initiation, a rock fall with a volume of 45,000 m3 was initiated within the rockslide. The kinetic push of the rock fall caused the movement of a translational soil landslide with a volume of 180,000 m3 that partially slipped into the torrential ravine of the Brusnik Stream. After a sudden drop of 15 m in December 2001, the rockslide average velocity exponentially slowed down to less than 10 m/year till the end of 2002, and came to a practical stillstand in 2003. After the rainfall in spring 2002, small debris flows made of clayey gravels with a volume of up to 1000 m3 started to flow from the zone of accumulation of the rock fall over the soil landslide to and along the channel of the Brusnik Stream. In 2002, more than 20 debris flow events were registered. The statistical analysis of the measured local rainfall intensities showed that debris flows were initiated at daily rainfall reaching from 20 to 30 mm, depending on the antecedent precipitation. This value may be taken as a specific hydrologic threshold for this site. Because in 2003 no more debris flows were registered, a conclusion was drawn that debris flow events were rainfall-induced but governed on the same time by the availability of rock fall debris in its zone of accumulation. D 2005 Elsevier B.V. All rights reserved. Keywords: Landslide; Rockslide; Rock fall; Debris flow; Flysch; Mitigation; Scaglia; Rainfall; Kosecˇ; Slovenia
1. Introduction Landslides may be of very different forms. Therefore, comprehensive terminology on these phenomena exists in the literature (Cruden, 1993). The EPOCH classification (Dikau et al., 1996) defines mass move* Corresponding author. Fax: +386 1 251 98 97. E-mail addresses:
[email protected] (M. Mikosˇ),
[email protected] (M. Brilly),
[email protected] (R. Fazarinc),
[email protected] (M. Ribicˇicˇ).
ment by a phrase using two words, one for the type of movement (fall, topple, slide, spreading, flow, complex) and the other for the type of material (rock, debris, soil). According to this classification, a complex landslide is one where one form of failure develops into a second form of movement (i.e., a change of behaviour downslope by the same material), and should be distinguished from a compound landslide where it consists of more than one type of movement. Several field case studies have been reported on complex and compound landslides in the literature
0013-7952/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.enggeo.2005.06.037 ENGEO-02462; No of Pages 14
ARTICLE IN PRESS 2
M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
(Santaloia and Cancelli, 1997; Toni and Rizzo, 2001; Ocakoglu et al., 2002; Guzzetti et al., 2004). Each of them is a case history for its own, and generalisation is hardly to be achieved. In many cases heavy precipitation is recognised as the main cause, and thresholds under different climatic conditions have been empirically evaluated (Caine, 1980; Canuti et al., 1985; Finlay et al., 1997; Crosta, 1998, Crozier, 1999; Glade, 2000; Alcantara-Ayala, 2004). Landslides may mobilise to form debris flows by three processes: (a) widespread Cuolomb failure within a sloping soil, rock, or sediment mass, (b) partial or complete liquefaction of the mass by high pore-fluid pressure, and (c) conversion of landslide translational energy to internal vibrational energy (Iverson et al., 1997). By observing larger landslides (Stozˇe, Strug, Slano blato, Macesnik, Fig. 1), initiated in the last several years in Slovenia, two of them (Stozˇe, Strug) were associated by debris flow events and should be considered complex landslides. Complex instability phenomena with consecutive debris flows were observed after the initiation of the Strug rock fall in December 2001. The rock fall happened above the Kosecˇ village near Kobarid in the Socˇa River basin, W Slovenia. These secondary processes are basically triggered by the original rock sliding phenomenon (Ribicˇicˇ and Kocˇevar, 2003).
The present study aims to explain the causes and possible future development of the Strug landslide. As soon as the Strug landslide occurred, first field observations and geological mapping were made. Afterwards, additional field studies such as analysis of aerial photographs and terrestrial geodetic measurements in the landslide area were made in order to evaluate the immediate hazard for the settlement downslope the landslide. In addition to these, reports from the civil defence units as well as the observations of local eyewitnesses were evaluated and used together with scientifically gathered material. In 2002, it was too dangerous to perform any classical field investigations (i.e., bore holes, geophysics, etc.) in the landslide area in order to apply the conventional slope stability analysis methods. Combining all available data, the mechanisms of the Strug landslide were determined and adequate technical measures were proposed and carried out. The Strug landslide is a well-documented case study of a complex landslide and thus of scientific and practical value. It was initiated by an interplay of several initiating factors, described in the paper. Another interesting aspect are small debris flows of volumes up to 1000 m3 originating in the accumulation zone of the rock fall on the contact between high permeable calcareous rocks (Cretaceous scaglia) thrusted over nearly impermeable clastic rocks (Cre-
Fig. 1. Map of Slovenia with active large landslides in 2004 and the Socˇa River basin.
ARTICLE IN PRESS M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
taceous flysch). They occur during moderate or heavy local showers and are therefore rainfall-induced but, on the other hand, governed by the availability of rock fall debris, i.e., the activity of the rock fall. 2. Description of the event After minor movements being active for some days, the Strug rock fall on the south-west slopes of the Planica Mountain (1376 m asl) in the Krn Mountains above the Kosecˇ village (650 m asl) was initiated on December 22, 2001 (Fig. 2). The rock fall happened in the drainage basin of the Brusnik Stream (drainage basin area of 0.80 km2), which flows through Kosecˇ village and enters the Rocˇica Torrent (drainage basin area of 10.8 km2) flowing through the Ladra village. Kosecˇ is a small village with 69 inhabitants (in 2002), composed of three parts: Podbrdo, Vas, and Orehovlje. The village is surrounded by meadows, and spreads on the altitude between 560 and 600 m asl. On the very same day the village was evacuated for two days. On December 26, 2001, a
3
second larger rock fall happened in the same area. The longitudinal profile of the initial geological situation of the rock fall in December 2001 is given in Fig. 3. In December 2001, the weight of weathered rock material of scaglia overcame the shear resistance on the critical slide plane. Therefore, it came to a fast sliding of unstable masses in a form of a rockslide with a volume of approximately 95,000 m3 that moved almost 15 vertical metres (determined from tree height on a photograph taken immediately after the event). As a consequence of the initial fast displacement of the rockslide, nearly at the same time, a rock fall with a volume of approximately 45,000 m3 of blocks was triggered within the rockslide. Therefore, the latter has now a volume of 50,000 m3. The kinetic energy of the fall of rock masses (boulders, blocks, and rock pieces) of scaglia mixed with ripped slope weathered material pushed the soil material and formed a large soil landslide with a volume of approximately 180,000 m3. The landslide mass filled the ravine of the Brusnik Stream that shifted its channel to the left-hand side of the landslide, and
Fig. 2. The Strug landslide area (photo taken on December 31, 2001).
ARTICLE IN PRESS 4
M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
Fig. 3. The geological longitudinal profile of the Strug landslide with the rockslide and the soil landslide features.
locally eroded it by incision. The landslide mass is composed of debris of scaglia rocks in a clayey–silty– sandy matrix, and is very likely a remnant of old landslides and rock falls. Surficial signs of such kind of mass wasting phenomena in the area can be easily observed on the neighbouring slopes of the Planica mountain (1376 m asl). During the slip of the landslide for some 50 vertical metres, an up to 8-m high scarp was formed due to side relaxation on its right-hand side (looking downslide). This is how the side support of the unstable masses to the right of the landslide decreased and a new smaller landslide commenced. This slide is called the brighthand side relaxation landslideQ and has a volume of approximately 35,000 m3. It is composed of silty–sandy coarse scaglia debris and is in composition similar to the main landslide. In early 2002, within the soil landslide in its topmost part a surficial landslide with a volume of approximately 30,000 m3 formed, made of up to 4 m thick layer of clayey scree material. A plan view of different slope instability processes in the Strug landslide area is given in Fig. 4. The total volume of actively falling, sliding and labile masses was estimated at 310,000 m3 (95.000 m3 of initial rockslide and 215.000 m3 of two soil landslides).
These rough volume estimations were made by using the results of a comparison of aerial photographs made before and after the event in December 2001. Until April 2002, larger quantities of rock debris were released, which then started to flow as occasional debris flows from the rock fall source area across the soil landslide towards the Brusnik Stream channel. The main reason for such debris flow events was larger
Fig. 4. The second debris flow event on June 7, 2002, at around 11 am (snap shot from a video filmed by a local TV Primorka television station). The head of the debris flow was around 3 m high.
ARTICLE IN PRESS M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
local precipitation. The same development was observed with many other rock falls in the Julian Alps, initiated during an earthquake on April 12, 1998. In those cases too, the rock fall debris was occasionally flowing as debris flows from slopes towards torrential channels during local precipitation (Mikosˇ and Fazarinc, 2000). Basically only the debris flow events that reached Kosecˇ and happened during daylight were recorded. One may assume that at least some debris flows did not reach the village, and thus were not registered. The recorded debris-flow events had one or occasionally several pulses with a volume of up to 1000 m3. The volume of released material for all the events was more or less the same. These debris flows started in the rock fall area during heavy rainfalls, reaching the Brusnik Stream channel and proceeding through Kosecˇ toward the channel of the Rocˇica Torrent, where they turned into over-saturated sediment laden flows and where their front stopped. The inflowing coarse material accumulated and, after several floods, reached the Ladra village in the lower course of the Rocˇica Torrent, before it enters the Socˇa River. An example of a debris flow surge observed in the Brusnik Stream is shown on Fig. 4. Each debris flow event has an estimated volume of the order of 1000 m3. A combined field geodetic, geologic and hydrologic analysis was made in order to investigate the reasons for these events and to help forecast possible future developments in the area as a part of the mitigation plan of the Strug landslide. 3. Analysis of the complex landslide The understanding of the initiation mechanisms of the slope instabilities was important in order to be able to forecast future development. After the reambulation of all known geologic data for the area, a detailed engineering geologic map (Fig. 5), a longitudinal profile of the Strug landslide (Fig. 3) and a plan view of different instability processes (Fig. 6) was prepared on the basis of a detailed field engineering geologic mapping. In the investigated area, the following rock types were determined (Fig. 5): 1. Lower and Upper Cretaceous flysch (bFQ): alternation of claystone, siltstone and sandstone; 2. Cretaceous scaglia (bSQ): red, marly limestone; 3. Cretaceous limestone with chert (bAQ): gray, weathered on surface; 4. Quaternary rock fall blocks and scree slope deposits originated from scaglia (bQSQ);
5
5. Quaternary silty and clayey slope scree and minor flysch blocks with inclusions of slope scree and scaglia blocks (bQGGQ); 6. Quaternary thinly colluvial deposits of scree fans and smaller rocks along minor local stream channels, unstable slopes (bQVQ); 7. Quaternary mixed rock fall and slope scree sediments of the Brusnik Stream: angular rock pieces, local moraine material (bQPQ). The geological longitudinal profile (Fig. 3) shows the relations between the aforementioned rock types, the position of the initial rockslide and the slide plane, and the thickness of different sliding phenomena. The fieldwork confirmed the existence of a fossil landslide on the NW boundary of the soil landslide. Many data on sliding processes were obtained in 2002 during regular (in some wet periods even daily) geologic, hydrologic, and geodetic inspections. All noticeable changes in the field were registered in the form of field diaries, by taking photos using digital cameras, and were later analysed. Such regular field observations provided an insight into changes in the Brusnik and Rocˇica channels. Landslides in the area were inactive except of the surficial landslide (Fig. 6), as confirmed by field inspections. In order to estimate the depth of sliding masses, we used old topographic maps and results from boreholes. Altogether four boreholes were situated in the area (Fig. 6). Because of the high rock-fall activity in early 2002, no boreholes could be situated on the active landslide, where we estimated its depth from the available topographic maps. Three boreholes were drilled on the fossil landslide (K-1 through K-3, Fig. 6), and were equipped with piezometers and inclinometers. In 2002, five measurements were performed. The level of ground water in boreholes on the fossil landslide was between 13 and 16 m below the surface, and their changes measured in 2002 were of the order less than 1 m, confirmed with measurements in 2004. The displacements in boreholes were of the order of few millimetres, and the diagram did not have characteristic shape for a clear sliding plane. The borehole in Kosecˇ (K-4, not on Fig. 6) was done on the Brusnik alluvial fan and it confirmed the first assumptions that this fan was formed by past debris flows. This fact stressed the importance of present debris flows. The laboratory tests on materials from boreholes were done in 2002. The natural water content of the drilling core was between 10% and 15%. The shear
6
ARTICLE IN PRESS
M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
Fig. 5. Engineering geologic map of the Strug landslide (threatened houses in Kosecˇ are given in red colour).
ARTICLE IN PRESS M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
7
Fig. 6. A plan view of different slope instability processes in the Strug landslide area with measuring sites.
strength uV = 37.58 was determined using triaxial tests on undrained consolidated cores. Debris flow deposits were sampled in the Brusnik channel below the toe of the landslide and analysed in different ways (Majes et al., 2002). It was classified as GC (clayey gravel) with the percentage of 35% of material b 0.063 mm. For fines (b 0.080 mm) from debris flow samples the liquid limit x L = 50%, the plastic limit x P = 19% and the plasticity index PI = 31% were estimated. The natural water content of the samples was above 20%. The shear strength uV = 308 was determined using triaxial tests on undrained consolidated samples b4.75 mm. In 2002, terrestrial geodetic measurements of a few selected fixed points (fissures, boreholes’ tops, unstable blocks etc.) were performed in the area. In 2003, more precise measurements of the rock fall surface in 21 fixed points and over 1000 surface points were performed using a laser distance meter (LaserAce 3 with angle encoder). Its precision was 1 cm at a maximum measurable distance of 300 m. The results were later put into a GIS environment, for further analysis of volume changes or displacements of single measuring points (blocks, fissures, etc.). This technique was used in order to follow the intensity of the rock fall processes and morphological changes caused by them. Using the analysis of both the first photographs of falling processes and the later terrestrial geodetic observations, a diagram of the rockslide displacements in time and of the rockslide displacement velocities was constructed (Fig. 7).
The laser distance meter was tested in the field on October 30, 2002, and used four times in 2003. It proved useful in finding out morphological changes of the rock fall surface caused by consecutive debris flows (Fig. 8). For a more in-depth analysis additional data are needed. Nevertheless, the measurements showed that the majority of the fixed points of the rock fall did not move from the end of 2002 to late 2003, showing the dormant phase of the rock fall (Mikosˇ et al., 2005). Firstly, daily total precipitation data as measured in a rain gauge in Kobarid (263 m asl) were used. Secondly, additional field hydrologic measurements and observations were performed during 2002 at selected measuring sites (see Fig. 6). For this purpose, an automatic weather station (VAISSALA MAWS-201 model) was erected on February 6, 2002, in Kosecˇ on altitude of 607 m. The station in Kosecˇ was 3.5 km far from the rain gauge in Kobarid. The parameters measured were air temperature, air pressure, air humidity, wind direction and velocity, and 5-min rainfall intensities. An automatic downloading of data via GSM was possible since February 15, 2002. The measured data were put to public access on the web server of the Faculty of Civil and Geodetic Engineering of the University of Ljubljana. Soil infiltration was measured at four sites (II, III, and IV, Fig. 6, I in the Kosecˇ village is not shown on Fig. 6) on March 19 and 29, 2002, using the Kopecky infiltrometer. The measuring sites were as follows: I (607 m asl) on meadow soils near the automatic weather station in Kosecˇ, II (806 m asl) on forest soils, III
ARTICLE IN PRESS 8
M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
Fig. 7. Measured displacements and the velocity of the Strug rockslide.
(893 m asl) on meadow soils, and IV (887 m asl) on meadow soils next to the right-hand side relaxational landslide. The yearly sums of daily totals measured in years 2001, 2002, and 2003 in the rain gauge in Kobarid are given in Fig. 9. A correlation between daily totals measured in 2002 in Kobarid and Kosecˇ is given in Fig. 10. Measured infiltration rates at 4 sites in the Strug landslide area are given in Fig. 11. The annual total rainfall of 2637 mm, measured in Kobarid in 2001 (Fig. 9), is slightly below the average annual total of 2699 mm, as determined for the period
1961–1990 (Kolbezen and Pristov, 1998). Differently, the annual total rainfall of 2307 and 1708 mm, measured in Kobarid in 2002 and 2003 (Fig. 9), was 15% and 37% below the average, respectively. The linear correlation between measured daily totals in rain gauges in Kobarid and Kosecˇ in 2002 (Fig. 10) gives a good agreement (R 2 = 0.804). In 2002, an average of 11% more precipitation was measured in Kosecˇ on a daily basis when compared to Kobarid. The rain gauge in Kobarid is situated at 263 m asl, and in Kosecˇ at 607 m asl. Normally, at higher elevations higher precipitation is to be expected. The difference is more
Fig. 8. Debris flow channel eroded into the rock fall debris (taken in September, 2002).
ARTICLE IN PRESS M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
9
Fig. 9. Yearly sum of daily precipitation measured in the rain gauge in Kobarid (263 m asl) for three years under investigation.
pronounced in the summer time, when the majority of precipitation occurs. The diagram on Fig. 10 does not show substantially higher precipitation in Kosecˇ for all rainfall events. For the highest registered rainfalls, the measured values for Kobarid were almost the same. For rainfall events with daily precipitation between 20 and 50 mm, Kosecˇ has a noticeably higher precipitation. The sum of measured precipitation for all days, i.e., when the weather station in Kosecˇ was operating, was 1397.4 mm in Kosecˇ and 1118.5 mm in Kobarid, respectively. This comparison indicates more precipitation for Kosecˇ (25%) than for Kobarid. Considering this proportion, more than 3000 mm would be the total rainfall for the landslide area in 2001, compared to 2637 mm measured in 2001 in Kobarid.
Fig. 10. Correlation between daily total rainfall in mm measured in the rain gauges in Kobarid (263 m asl) and Kosecˇ (607 m asl) in 2002 (only n = 33 daily totals were used, when in both rain gauges more than 10 mm were registered; R 2 = 0.804).
4. Discussion The field engineering geological mapping showed that similar slope instability events had already happened in the geological past. Areas of rock fall deposits (mainly scaglia) were determined. Using field estimation of terrain stability, in addition to the previously described slope instabilities in 2002, a large fossil landslide (see Fig. 6) was outlined. This was confirmed by the results from three boreholes, situated on it. The average thickness of the fossil landslide was estimated at 25 m. The borehole K-4 in Kosecˇ also confirmed the expectations that the part of Kosecˇ along the Brusnik Stream lies on old debris flow deposits with a local thickness of over 10 m. The rockslide in 2001 was the first event in a chain of unstable processes that resulted from it. In the hinterland of the Planica Mountain (1376 m asl) a slide
Fig. 11. Results of the infiltration tests (I – undisturbed meadow soils on 607 m, II – undisturbed forest soils on 806 m, III – undisturbed meadow soils on 893 m, and IV – disturbed meadow soils on 887 m, few meters away from the scar of the right-hand side relaxational landslide).
ARTICLE IN PRESS 10
M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
plane formed. The initiation of this slide plane was caused by several interrelated unfavourable factors. The first factor was tectonics. The rockslide formed above the regional thrust of the cretaceous scaglia on calcareous flysch (Fig. 3). The thrust is positioned in the toe of the rockslide and forms the lower part of the slide plane, where the sliding rock material is pushed out of the slope. Due to tectonic stresses, the initially highly bedded rocks are additionally fractured. Tectonics also caused joints that are nearly parallel to the surface. Some of them became part of the slide plane. The next influential factor for destabilising the natural slope stability was surficial rock weathering. The slide happened in the uppermost 10 to 20 m thick weathered rock layers. The weathered part of rocks, as determined in the field by change in colour and more opened cracks, is still visible on the main scarp of the rockslide. The contacts between adjunct rock formations became less tight, and the friction on contact planes decreased. The slide plane parallel to the slope formed first at its upper part, and then prolonged to the thrust zone at its toe. This became clear by regular field observations which revealed how rock blocks were coming out at the tectonic thrust zone at the toe of the slide. Another releasing factor that caused local displacements along the potential slide plane, which led to a decreasing shear resistance along it, was presumably the earthquake on Easter Sunday on April 12, 1998 (Ribicˇicˇ and Vidrih, 1998). During the earthquake, which is believed to be the strongest earthquake with the epicentre in Slovenia in the 20th century (Vidrih et al., 2001), vertical fractures that increased water infiltration into the rockslide body opened very likely on the top of the slope. After the earthquake, local farmers observed similar cracks on neighbouring slopes. Even though no stability analysis of the Strug rockslide was performed, one may conclude on the basis of small measured displacements of the rock slide (b 10 m/ year in October 2002 decreased to 0.1 m/year in October 2003, Fig. 7) that it went into dormant phase. Its present mass of 50,000 m3 is low enough to assume no immediate hazard for Kosecˇ, which is situated nearly 1000 m away with a height difference of 400 m. This situation gives a shadow angle of around 228, which is less than minimum shadow angles found in the field (Evans and Hungr, 1993). When it will come to a possible reactivation of the rockslide, this may partially stay in the present zone of accumulation of the rock fall. The foot of the soil landslide is wedged into the narrow Brusnik valley, and will eventually proceed, but is not an immediate hazard for Kosecˇ. The slope of the soil landslide surface is around 308 and is less than the
estimated shear strength of soil material uV = 37.58, which is also unfavourable to mobilise and form a debris flow. Under such conditions, debris flows were considered as the most hazardous phenomenon for the Kosecˇ village. Thus, they will be described in detail. The first registered debris flow event occurred on April 12, 2002, after two days of heavy rainfall (128 mm, Fig. 12) following a relatively dry spring period with only 155 mm rainfall from the beginning of March until the event (Fig. 12). This debris flow stopped in the rock fall source area. The first debris flow that reached Kosecˇ was released on May 4, 2002, and had three pulses. The daily rainfall measured in Kosecˇ was 46 mm (Fig. 12). The last debris flow in 2002 occurred on November 17, during a 52-mm rainfall event (Fig. 12). In 2002, nearly 20 debris flow events were registered, the majority of them occurred in daylight. The rainfall analysis took into account hourly and 10-min rainfall data, as shown for three debris flow surges on June 7, 2002 (Fig. 13). As a result of this statistical analysis, the daily rainfall accumulation of 30 mm was determined as a threshold value for the initiation of debris flows in 2002 under wet conditions (with enough antecedent precipitation to keep rock fall debris in wet conditions). This value is a site specific threshold for triggering debris flows. It is below the lower bound threshold of around 2 mm/ h for 24 h duration, given in the rainfall intensityduration diagram of empirical thresholds, summarised from worldwide data by Crosta (1998). It is obviously the antecedent soil moisture that governs the hydrological thresholds for rainfall-induced debris flows, clearly shown for pyroclastic deposits by Fiorillo and Wilson (2004). For a refinement of the proposed threshold value of 30 mm/day, additional analysis of rainfall intensities of local showers with the duration of several hours measured in the landslide area will be needed. There were some exceptions to this rule. Despite more than 30 mm in Kosecˇ, there were no debris flow events recorded in Kosecˇ on the following days: May 13, June 25, and July 4, 2002 (Fig. 12). Among them, the first rainfall event was a very short (1 h) and intensive thunderstorm, the other two events were similar events with intensive downpours of short duration. Normally, debris flow events were recorded at the bridge in Kosecˇ, in most cases some 3 to 4 h after the rainfall accumulation reached 30 mm. An exception was the first event on May 4, 2002, which happened some 6 to 7 h after, and on June 7, 2002, when the event happened without any delay (Fig. 13). With the
ARTICLE IN PRESS M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
11
Fig. 12. The daily totals measured in the rain gauge in Kosecˇ. For the days with technical troubles with the weather station, correlation with measured rainfall in Kobarid was used instead.
latter case, one should take into account the antecedent precipitation (41 mm for the two days before in Kosecˇ). In 2003, the first half of the year was very dry (Fig. 9), and the rock fall was dormant. The first heavy rainfall event was registered between July 2 and July 5, 2003, with 113 mm of rainfall in Kobarid, producing no debris flow. The rock fall reactivated after heavy rainfall of 201 mm from October 30, 2003 to November
2, 2003. In this period there were several small debris flows, which stopped at the toe of the landslide and did not reach Kosecˇ. In the first months after the rock fall in December 2001, a slow movement of the rockslide along the slide plane proceeded (Fig. 7), accompanied by intensive falling of boulders, blocks and single rocks from the rock fall surface. In the area below the rock fall and on the top of the landslide, a large talus was formed. The
Fig. 13. Timing of three debris flow events (black columns) observed on June 7, 2002, and measured cumulative rainfall in the Kosecˇ rain gauge from 9 pm on June 6, 2002 till 5 pm on June 8, 2002.
ARTICLE IN PRESS 12
M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
piling of scaglia material, which accompanied the movements of the rockslide, proceeded during the dry spring period in 2002. In mid-April 2002, occasional rain began (Fig. 12). On the tectonic contact between scaglia and flysch several springs collecting waters from above arose on the left side and in its middle part. These springs and water from overland flow above the thrust during rainfall started watering marl that prevails in the flysch formation below the thrust and they slowly started to erode its finer particles. Fine sediment fraction from marl infiltrates into coarse rock material. Coarse rock scaglia material started to creep on marl and later, when it was wet enough, it started to roll on it and mix with it. When wet enough, this heterogeneous mixture started to flow as a debris flow. Hence, in order to flow, the rock fall material must be wet enough. For such conditions also antecedent precipitation is important. Because the thrust is dipping into the slope, the springs are activated only when precipitation infiltrates and when groundwater raises and starts to overflow the thrust. When long duration precipitation in a certain period is low and comparable to evapotranspiration losses, the groundwater level is too low to activate the springs. The estimated value of annual evapotranspiration losses for the region around the Strug landslide is around 500 mm, as determined by using data from the Socˇa River basin (Kolbezen and Pristov, 1998). Results of the infiltration tests (Fig. 11) show that all precipitation of less than 200 mm a day may well infiltrate into the disturbed soils on the landslide without generating substantial overland flow. This is also true for the lower part of the rock fall source area, covered by rock fall coarse debris, but not for the steep slope above the thrust. The contributing area of the debris flow amounts to 2.63 ha. Rainfall of at least 10 mm may produce overland flow, which triggers a debris flow with a volume of several 100 up to 1000 m3. During more intense precipitation several events were generated on the same day. Due to relatively small contributing area of the debris flow, only limited amounts of water were collected, even during intense local downpours. Also the production of fine particles in the marl formation was limited. Consequently, the debris flows were rather bdryQ and thus composed of coarse rock fragments or rock fall debris that rolled and pushed larger boulders, and of fine grained clay material saturated with water (Fig. 4). From the debris flow deposits found in the Brusnik channel, the porosity of debris flows was estimated to be around 50%. This may be the main reason why the
average debris flow velocity of observed events was found to be 1 to 3 m/s only in the Brusnik Stream channel with the 20% longitudinal slope. In some cases, the rather dry and coarse front of the debris flow even stopped for some time (even up to 10 s). The debris flow proceeded to advance, when its rather wet tail flowed to the backside of the front and pushed it further. The main reason for no debris flows in 2003 when compared to 2002 was the dormant phase of the rockslide (Fig. 7) and low activity of the rock fall. Much less rock fall debris was available for generating debris flows in 2003. The additional reasons were low precipitation (Fig. 9), high summer temperatures and thus corresponding higher evapotranspiration losses. This conclusion was also confirmed by field geodetic measurements in 2003, using the laser scanner technique. They showed low rock fall activity and some minor morphological changes restricted to the debris flow source area. One may conclude that, in the case of the Strug landslide, debris flows are Jsediment supply limitedV events. This implicates that future development in the Strug landslide area will largely depend on the rockslide stability and rock fall activity. One may expect debris flow events to be generated more frequently only when enough fresh but coarse-grained rock fall material will be available to mix with marl finegrained material. One may also expect that the volume of an individual debris flow event will remain not far away from values experienced so far. The two-year field observations of the Strug landslide have shown that after the first initial intensive phase in December 2001 and the first few months in 2002 the landslide temporarily found a labile balance. This may be ruined by more intense events such as larger local earthquake or severe precipitation. Such severe conditions may destabilise the rockslide as well as the soil landslide. Under such new conditions the already performed mitigation measures should be redefined and a new hazard assessment under the new situation should be done. 5. Mitigation of the Strug landslide Shortly after finishing the field engineering geological investigations and knowing what was happening in the landslide area, it became clear that easy remediation using technical measures in the source area was out of the question. Any classical remediation measures using geotechnical means were impossible due to difficult accessibility of the source area (it can be reached only
ARTICLE IN PRESS M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
on foot). Therefore, it became obvious that, as a first step, preliminary caution and safety measures for life and property of Kosecˇ village had to be undertaken. The first measure immediately after the rock fall event on December 22, 2001, was the evacuation of 18 inhabitants from the 6 most threatened houses in Orehovlje. After a few days, the inhabitants came back. But ever since the overnight stop for the selected threatened houses near the channel of the Brusnik Stream has been declared (see Fig. 5). Regular inspections (geodetic, geologic, hydrologic, civil defence) and rainfall gauging were used for determining the immediate danger to life and property in the villages of Kosecˇ and Ladra. In spring 2002, dry weather made it possible that, apart from the alarm systems, no other rigorous safety measures had to be met, such as another full evacuation of the inhabitants. On the other hand, two important structural measures were realised in early 2002 in order to prevent damages due to expected debris flows: a) clearing of channel vegetation and widening of the narrow channel of the Brusnik Stream from approximately 10 m2 by a factor of five from upstream of Kosecˇ and through it down to the channel inlet to a very steep reach (ravine) upstream of the confluence with the Rocˇica Torrent (wider channel should be able to convey debris flows without channel overtopping and their spreading in the village), and b) replacing an old bridge with a cross-section area of only 3 m2 on the local road over the Brusnik Stream in Kosecˇ, as the most obvious debris flow obstacle, with a larger bridge with a cross-section of 25 m2. According to all inspections made in early 2002, a further advance of the landslide towards the village was expected during or immediately after heavier rainfalls. The landslide toe did not move but first debris flows were observed in May 2002 after a wet period. The riprap withstood all debris flows in 2002 without any serious damage. Additional levees were later placed on the right side of the channel to better protect the village in the case of larger or faster flowing debris flows. In the same time, an alarm system was erected just below the toe of the landslide and put into function. It will be released in the case of landslide advancement and not in the case of a debris flow. A similar system was erected on a check dam on the Rocˇica Torrent upstream of the Ladra village. Since it has been decided that a technical remediation of interrelated unstable phenomena is not feasible, a hazard map of the area was prepared. For that pur-
13
pose, one-dimensional and two-dimensional mathematical models of debris flows, used for the determinantion of the hazard map of the village of Log pod Mangartom for the case of debris flow from the Stozˇe landslide (Cˇetina and Krzyk, 2003; Mikosˇ et al., 2004) were used and validated using raw data on registered debris flows in Kosecˇ. On the basis of this hazard map, additional widening of the Brusnik channel in Kosecˇ in the reach of the bridge has been planned. 6. Conclusions The Strug landslide case study has once more shown that handling complex landslides of large dimensions in a combination of different sliding phenomena is not a simple task, and that there is often no final remediation but only mitigation possible. Under such conditions, a responsible and proper hazard assessment is possible only by regular landslide inspections of landslide changes and by field measurements, which both help identify the possible consequences of different slope instability phenomena. From the geodetic, geological and hydrological studies performed in the Strug landslide area, the following conclusions can be drawn: (1) A combination of high permeable calcareous rocks thrusted over nearly impermeable clastic rocks are potential sources of rock falls on steep slopes and of rockslides on more gentle slopes. Not all these sources have the same hazard level. The highest level can be found where rocks are tectonically highly fractured and therefore the weathering effects can penetrate deeply into the rocks. (2) The rockslide first initiated the rock fall, and then other slope instabilities. On one hand, two soil landslides were triggered. On the other hand, the rock fall debris under described geological conditions, turned out to be the source of occasional debris flows. (3) Debris flows were initiated in the rock fall source area. They were generated by mixing coarsegrained scaglia blocks and stones with finegrained marl particles, with enough water. If one of these three components was missing, no debris flow would be generated. (4) Only when enough rock fall material was available, as was the case for the Strug landslide in 2002, a good correlation between debris flow occurrence and precipitation was found. The debris flows were initiated with daily total rainfall larger
ARTICLE IN PRESS 14
M. Mikosˇ et al. / Engineering Geology xx (2005) xxx–xxx
than 30 mm after a day or two without precipitation. In a wet period (higher soil moisture) a debris flow was initiated with daily total rainfall of 20 mm. Both values may be taken as site specific hydrologic thresholds for triggering debris flows. (5) From all the slope instability phenomena presented in the case of the Strug landslide, under present conditions the debris flows are considered as the most hazardous phenomenon for the Kosecˇ village. Acknowledgements The authors would like to thank the State Rehabilitation Commission of the Republic of Slovenia for the financial support of the project. The Environmental Agency of the Republic of Slovenia made available the rainfall data from Kobarid. The assistance in fieldwork by Marko Kocˇevar, Dusˇan Kosmacˇ, Urban Umek, and Andrej Vidmar is also greatly acknowledged. Furthermore, the insight comments of two anonymous reviewers greatly improved the paper. References Alcantara-Ayala, I., 2004. Hazard assessment of rainfall-induced landsliding in Mexico. Geomorphology 61, 19 – 40. Caine, N., 1980. The rainfall intensity-duration control of shallow landslides and debris flows. Geogr. Anal. A 62, 23 – 27. Canuti, P., Focardi, P., Garzonio, C.A., 1985. Correlation between rainfall and landslides. Bull. Int. Assoc. Eng. Geol. 32, 49 – 54. Cˇetina, M., Krzyk, M., 2003. Two-dimensional modelling of debrisflow movement in log pod Mangartom as an example of a nonNewtonian fluid. Stroj. vestn. — J. Mech. Eng. 49 (3), 161 – 172. Crosta, G., 1998. Regionalization of rainfall thresholds: an aid to landslide hazard evaluation. Environ. Geol. 35, 131 – 145. Crozier, M.J., 1999. Prediction of rainfall triggered landslides: a test of the antecedent water status model. Earth Surf. Proces. Landf. 24 (9), 825 – 833. Cruden, D.M., 1993. The Multilingual Landslide Glossary. Bitech Publishers, Richmond, British Columbia. Dikau, R., Brunsden, D., Schrott, L., Ibsen, M.-L., 1996. Introduction. In: Dikau, R., Brunsden, D., Schrott, L., Ibsen, M.-L. (Eds.), Landslide Recognition: Identification, Movement and Causes. Wiley, Chichester, pp. 1 – 12. Evans, S.G., Hungr, O., 1993. The assessment of rockfall hazard at the base of talus slopes. Can. Geotech. J. 30, 620 – 636. Finlay, P.J., Fell, R., Maguire, P.K., 1997. The relationship between the probability of landslide occurrence and rainfall. Can. Geotech. J. 34, 811 – 824.
Fiorillo, F., Wilson, R.C., 2004. Rainfall induced debris flows in pyroclastic deposits Campania (southern Italy). Eng. Geol. 75, 263 – 289. Glade, T., 2000. Modelling landslides: triggering rainfall thresholds at a range of complexities. Proceedings of the 8th int. symp. on landslides held in Cardiff on 26–30 June, Landslides in Research, Theory and Practice, vol. 2, pp. 633 – 640. Guzzetti, F., Cardinali, M., Reichenbach, P., Cipolla, F., Sebastiani, C., Galli, M., Salvati, P., 2004. Landslides triggered by the 23 November 2000 rainfall event in the Imperia Province, Western Liguria, Italy. Eng. Geol. 73, 229 – 245. Iverson, R.M., Reid, M.E., LaHusen, R.G., 1997. Debris-flow mobilization from landslides. Annu. Rev. Planet Sci. 25, 85 – 138. Kolbezen, M., Pristov, J., 1998. Povrsˇinski vodotoki in vodna bilanca Slovenije (Surface Streams and Water Balance of Slovenia). Hydrometeorological Office of the Republic of Slovenia. 98 pp. (in Slovene with English abstract). Majes, B., Petkovsˇek, A., Logar, J., 2002. The comparison of material properties of debris flows from Stozˇe, Slano blato and Strug landslides. Geologija 45/2, 457 – 463 (in Slovene with English abstract). Mikosˇ, M., Fazarinc, R., 2000. Earthquake-induced erosion processes in two alpine valleys in Slovenia. In: Zollinger, F., Fiebiger, G. (Eds.), Schutz des Lebensraumes vor Hochwasser, Muren und Lawinen. Klagenfurt: Internationale Forschungsgesellschaft INTERPRAEVENT, 2000, pp. 143 – 154. Mikosˇ, M., Cˇetina, M., Brilly, M., 2004. Hydrologic conditions responsible for triggering the Stozˇe landslide. Slovenia. Eng. Geol. 73, 193 – 213. Mikosˇ, M., Vidmar, A., Brilly, M., 2005. Using a laser measurement system for monitoring morphological changes on the Strug rock fall, Slovenia. Nat. Hazards Earth Syst. Sci. 5, 143 – 153. Ocakoglu, F., Gokceoglu, C., Ercanoglu, M., 2002. Dynamics of a complex mass movement triggered by heavy rainfall: a case study from NW Turkey. Geomorphology 42, 329 – 341. Ribicˇicˇ, M., Vidrih, R., 1998. Plazovi in podori kot posledica potresov (Earthquake-triggered landslides and rockfalls). Ujma 12, 95 – 105 (in Slovene with English abstract). Ribicˇicˇ, M., Kocˇevar, M., 2003. Mehanizmi plazenja nad Kosecˇem v obcˇini Kobarid in s tem povezani sanacijski ukrepi (Mechanisms of sliding above the village of Kosecˇ and with them linking remediation measures). Geologija 46 (1), 159 – 166 (in Slovene with English abstract). Santaloia, F., Cancelli, A., 1997. Landslide evolution around Mt Campastrino (Northern Apennines, Italy): a complex and composite gravitational movement. Eng. Geol. 47, 217 – 232. Toni, G., Rizzo, V., 2001. Kinematic evolution and shear strength in the complex landslides of the Maratea valley (PZ, Italy). Phys. Chem. Earth (C) 26 (9), 677 – 681. Vidrih, R., Ribicˇicˇ, M., Suhadolc, P., 2001. Seismogeological effects on rocks during the 12 April 1998 upper Socˇa Territory earthquake (NW Slovenia). Tectonophysics 330 (3/4), 153 – 175.
