Disaster Mitigation of Debris Flows, Slope Failures and Landslides 571

Seismic Monitoring of Deep-Seated Mass Movements

Ewald Bruckl and Stefan Mertl1) Institute of Geodesy and Geophysics, Vienna University of Technology, Gusshausstrasse 27-29, 1040 Vienna,Austria, (ebrueckl@luna.tuwien.ac.at)

Abstract

Deep-seated mass movements in crystalline rock start frequently with internal deformation comprisingthe whole rock mass. In an advanced state deformation concentrates to one or several shear zones. The surfacevelocities remain still in the range of 0.01–1 m/year and the deformation process can be classified as creep.Most of these mass movements stabilize after a displacement of the center of gravity in the range of 100–300 m.However, a few examples exist that transformed from creep to rapid sliding. This transition can be equivalentto a catastrophic event. State and velocity dependent friction laws can explain this behavior. A fully developedbasal shear plane is prerequisite for a transition from creep to rapid sliding. The only observational indicationof such an event is approximately exponential increasing velocity. Additional observational quantities relatedto the development of deep-seated mass movements would be very helpful to support prediction. The massmovements we consider take place in brittle rock and are expected to produce fractures or to display stick-slip movement on existing shear planes. Both processes generate seismic energy. Information that can begained from monitoring and analyzing this seismic activity is manifold. Focal coordinates tell where brittledeformation takes place and other seismic parameters may be related to the state of the mass movement.

Since 2001 several passive seismic monitoring campaigns were carried out on three deep-seated massmovements (saggings) in the Eastern Alps in Austria: Gradenbach (Schober Range, Carinthia), Hochmais-Atemskopf (Otztaler Alpen, Tyrol), and Niedergallmig-Matekopf (Samnaun Range, Tyrol). Different recordingsystems have been used, mostly with 3C-seismometers, natural frequency 4.5 Hz. Up to 11 stations operatedsimultaneously on one mass movement. Since September 2004 also one broadband seismometer has beenoperating continuously on Gradenbach. The seismic events produced by the mass movements and recordedby our monitoring system have frequency contents (above noise level) up to 30 Hz, durations of 5s to 20 s,and magnitudes MW of −2 to 0. Several automated event detection and localization methods are underdevelopment. We will present the results we achieved so far. A characterization of the observed seismicactivity related to status and velocity of each mass movement will be attempted.

Keywords: deep-seated mass movement, creep, seismic activity, instability, prediction

Introduction

Deep-seated gravitational creep is a phenomenon frequently observed on alpine rock slopes. The termssackung (sagging) and rock-flow are also in use for this type of mass movement (Zischinsky, 1969; Hutchinson,1988; Dikau et al., 1996). Crystalline rocks of a wide range of metamorphic grades may undergo deep-seatedgravitational creep. These rocks deform by the mechanisms of brittle rheology (fracturing, sliding, and rolling).Structural characteristics are cracks and scarps in the upper part of slopes (“Bergzerreissung”) and bulging(“Talzuschub”) in the lower part (Ampferer, 1939; Stini, 1941; Savage and Varnes, 1987; Bisci et al., 1996).The volume of the rock mass of one sagging is in the range of 0.1–1 km3.

The development of deep-seated gravitational creep is thought to be in an analogy to the three stagesof creep in material testing: primary, secondary, and tertiary creep. In the primary stage the creeping rockmass develops by internal deformations comprising the whole rock mass. Glacial retreat at about 15 000 aB.P. (Zischinsky, 1969; Husen, 1987) is assumed to have initiated the processes of slope destabilization in theEuropean Alps. In an advanced stage deformation concentrates to one or several shear zones. The surfacevelocities remain still in the range of 0.01–1 m/year. Many saggings remain in this quasi-stationary, secondaryphase or stabilize after a downslope displacement of the center of gravity by a few hundred meters. However,there are also examples that transformed from an observed creep phase to rapid sliding (Muller, 1964), orwhere an initial creep phase can be assumed before the transition to rapid sliding took place (Bruckl et al.,2001). This transition is frequently equivalent to a catastrophic event, and its prediction, both on short andlong term is a major issue of disaster prevention.

Planar and rotational slider block models in combination with state and velocity dependent friction

pp. 571–580 c©2006 by Universal Academy Press, Inc. / Tokyo, Japan

572

laws (Dieterich, 1992; Ruina, 1983) can explain this behavior and can be used for prediction (Helmstetter etal., 2004; German and Bruckl, 2006). A fully developed basal shear plane is prerequisite for a transition fromcreep to rapid sliding (Bruckl and Parotidis, 2005). The existence of a basal sliding surface can be derived fromborehole measurements, but also from geodetic data (surface velocity and changes of elevation) in combinationwith a structural model (extent of creeping rock mass) based on geophysical data (Bruckl et al., in press).The most important observation for the prediction of slope instability is velocity. A velocity increase followingthe inverse Omori law (Voight, 1988) can principally be used to predict the time of failure. However, for ashort term prediction the problem exists to separate the influence of pore water pressure on velocity from statedependent friction. Representative observations of pore pressure on the sliding surface are hardly to achieve.For long term predictions we need more information about the time-displacement relation than is availablefrom geological observations, considerations, and dating.

Additional observational quantities related to the development of deep-seated mass movements wouldbe very helpful to support prediction. The mass movements under consideration take place in brittle rockand we expect generation of new fractures and stick-slip movement on existing shear planes. Both processesgenerate seismic energy. Information that can be gained from monitoring and analyzing this seismic activityis manifold. Focal coordinates tell where brittle deformation takes place and seismic efficiency (ratio betweenseismic energy and total potential energy release) is related to the state of the mass movement. Seismicmonitoring of deep-seated gravitational creep on slopes in the Austrian Alps started in 2001. Currently threemass movements are temporally under observation. We will describe the monitoring and present the resultsachieved so far. A characterization of the observed seismic activity related to status and velocity of each massmovement will be attempted. Seismic events on other types of mass movements than deep-seated creep aredescribed in Roth et al. (2005) and Spillmann et al. (2006).

Monitoring sites

Three examples of deep-seated gravitational creep in the Austrian part of the Eastern Alps were se-lected for seismic monitoring: Gradenbach (GB), Hochmais-Atemskopf (HA), and Niedergallmigg-Matekopf(NG). GB is located in Schober Range, Carinthia, HA in Otztaler Alpen, Tyrol, and NG in Samnaun Range,Tyrol (Figure 1). They lie in crystalline complexes of the Austrian Alps. The petrology at GB is characterizedby phyllites and calcareous mica shists (Moser and Kiefer, 1988; Gottschling, 1999), at HA by paragneisses,micaschists, orthogneisses, and amphibolites (Tentschert, 1998). At NG pelitic gneisses, phyllites, and amphi-bolites cover the lower part of the slope, the upper part is built up by paragneisses and two-micashists of theSilvretta crystalline basement (Kirschner et al., 2004). The mass movement at NG comprises a western part,starting from the scar on top of the slope, and an eastern part, which affects only the middle and lower partof the slope. In our study we concentrate on the main, western part.

3D-models of the P-wave velocity of the mass movements were derived from refraction seismic measure-

Fig. 1. Location of monitoring sites.

573

Fig. 2. Topography, present extend/boundary of the mass movement (solid line), escarpment area(dashed line), and thickness (H) of the disintegrated and creeping rock mass (grey code); (a) GB,(b) HA, (c) NG.

ments and structural models were constructed by the additional consideration of borehole, photogrammetric,geodetic, geological, and geomorphological data (Bruckl and Bruckl, 2006; Bruckl et al., 2004; Chwatal et al.,2005). Figures 2a–c show the topography, the present extent/boundary of the mass movements, the escarp-ment area, and the thickness of the disintegrated and creeping rock mass. Values for area, volume and averagediscontinuous displacement along the basal sliding surface are compiled in Table 1. The latter parameter hasbeen derived from the extent of the scarps at the uphill end of the mass-movements and from the estimatedoverthrusting at the fronts (toes).

GB has been under geodetic and geotechnical control for more than 30 years. In 1965, during aperiod of high precipitation and severe floods strong movements resulted in catastrophic debris flows. Highdisplacement rates (on average about 6 m/year) were also observed during 1974–1975 (Kronfellner-Kraus,1980). Evaluation of aerial photographs from the years 1962 and 1996 (Bruckl et al., in press) gave an averagehorizontal displacement of 0.6 m/year for this period. The displacements were remarkably uniform for thewhole moving rock mass. GPS measurements at 4 stations distributed over the whole slope started in 1999and still continue (Brunner et al., 2003; Brunner, personal communication). The average velocity observedduring this period is 0.2 m/year. The velocity undergoes strong temporal variations (from 0.01 m/year–0.02m/month), but is synchronous on all 4 stations. An analysis of the kinematics showed, that GB exhibits ablock movement at present and deformation concentrates to the basal sliding surface.

The slope of the HA mass movement ends in a hydroelectric power reservoir constructed about 40years ago. Therefore, HA has been under geodetic and geotechnical control since that time (Tentschert,1998). Several scarps define individual sliding slabs; the velocity of the most active slab is 0.03–0.04 m/yearat the surface and about 0.025 m/year at its base. In the upper part of the slope, the creep rates drop to< 0.006 m/year.

At NG, observation of up to 30 landmarks by GPS and terrestrial geodetic methods started in October2003. The creep rates are 0.07–0.08 m/year for the main western part of the mass movement, and 0.02–0.03m/year for the less active eastern part. The displacement vectors of the western part fit well a rotational sliderblock model and indicate that most of the shear deformations concentrate to the basal sliding surface derived

574

Table 1. Present area of moving mass (A), volume (V),escarpment area (A0) and average discontinuous dis-placement (s) on the basal plane of the mass move-ments GB, HA, and NG.

Table 2. Monitoring campaigns with short period seismometers on GB, HA, and NG.

Fig. 3. Location of seismic stations during different monitoring campaigns (a)GB, circles:GB 200208, triangles: GB 2005ß6; (b)HA, circles: HA 200108, triangles: HA 200509; (c)NG,triangles: NG 200506

from the geophysical data.

Data acquisition

Passive seismic monitoring started on GB and HA in 2001 and has been extended to NG in 2005.Different recording systems have been used, mostly with 3C-seismometers, natural frequency 4.5 Hz. Up to 11stations recorded simultaneously on one mass movement. Detailed information on the data acquisition duringthe different monitoring campaigns is compiled in Table 2. Sample interval was 0.008s or 0.010s. The locationsof the seismic stations are shown in Figure 3a–c.

The periods of the earlier monitoring campaigns were mainly determined by the availability of instru-ments. Now, equipment is available for more regular observations. It comprises 10 3C-geophones with 4.5 Hznatural frequency, 5 3-channel RefTEK-Texan 125 recording systems with 256 MB memory for each channel,and 5 3-channel RefTEC-Texan 130 recording systems with 4 GB memory.

Since September 2004 also one broadband seismometer (GURALP CMG-6TD, 0.033–50 Hz, 4 GB

575

Fig. 4. Localization of an event. (a) seismic data; (b) location of seismic stations (triangles) andprobability density function of epicenter.

memory) has been operating continuously on GB. The goal of broadband monitoring is two-fold: detection oflow frequency seismic events on the mass movement and combination of the seismic signal and a parallel GPSrecording to one broadband displacement output. The second goal will not be further considered in this paper.

Event identification, localization, and seismic moment determination

The seismic data recorded by the passive monitoring stations is processed by eliminating known sig-nals like human induced noise and global-, regional-, and local earthquakes from the data. The remainingevents, which cannot be allocated to any known type of seismic signal are categorized as “interesting events”and undergo further investigations to extract the “candidates” of events being excited by the moving mass.Earthquake catalogues and nearby seismic stations from permanent networks are used to discriminate fromglobal and local earthquakes, and spectro- and sonogram (Joswig, 1990) are used to detect weak signals andvisualize the frequency content.

Determination of focal parameters (source location and focal time) is attempted for seismic eventsclassified as candidates. For candidates with sufficient signal to noise ratio this is done on the basis of firstarrival travel times and the existing 3D P-wave velocity models. Travel time calculations by solving the eikonalequation, grid-search and probabilistic methods are used for this purpose (Podvin and Lecomte, 1991; Lomaxet al., 2005). The methods were successfully tested by the localization of controlled seismic sources (dynamiteshots) recorded during one monitoring campaign on HA. An example of localization together with the seismicdata is shown in Figure 4. Generally, the epicenter, referenced to a plane fitting the slope, is much betterconstraint than the focal depth. Many candidates for earthquakes caused by the mass movements have such alow signal to noise ratio that accurate travel time picking is not possible. For these events a method is underdevelopment that attempts an approximate localization by the use of amplitudes. Seismic moments havebeen roughly estimated from the recordings using a relationship with maximum ground velocity and distance(McGarr and Bicknell, 1990) and preliminary determination of the focal parameters. The relation of Hanksand Kanamori (1979) has been used to convert M0 to moment magnitudes (MW).

Characteristics of micro-earthquakes caused by gravitational creep

Many events classified as micro-earthquakes caused by deep-seated gravitational creep have a compositeor multiple-event character. Typical recordings (Z-component) of multiple-event micro-earthquakes observed atGB, HA, and NG are shown in Figures 5–7. For comparison, the recording of a dynamite shot on HA is shownin Figure 8. The total duration of the micro-earthquakes is 5–20 s, the main frequency content is < 30 Hz.The later events of one composite micro-earthquake have little lower frequency content than the first event.The multi-event or composite micro-earthquakes are not the only type, identified on the mass movements. Wefound also narrow frequency-band and low frequency events. An example is shown in Figure 9.

576

Fig. 5. Typical recording of multiple-event micro-earthquakes at GB (MW = −1.0); (left) seismictraces with spectrogram; (right) locations of seismic stations and epicenter;

Fig. 6. Typical recording of multiple-event micro-earthquakes at HA (MW = 0.7); (left) seismictraces with spectrogram; (right) locations of seismic stations and epicenter;

Work in progress

So far, event identification, classification, localization, and seismic moment estimation has not beencompleted and work on these topics is in progress. At present seismic moment (M0) and magnitude (MW)estimation for one monitoring campaign on GB (2002/197–206) and one on HA (2005/257–265) has beencarried out. During these campaigns 20 events were identified on GB and 7 events on HA. The magnitudesrange from MW ∼ −2 to MW ∼ 0. Empirical magnitude — frequency diagrams (Gutenberg-Richter) willbe prepared after completion of evaluation. Seismic stations currently available for monitoring GB, HA, andNG will make longer observation periods possible. This will bring more insight into the spatial and temporalcharacteristics of the seismic events and their relation to kinematics and dynamics of the mass movement. Anext step is also the calculation of moment tensors for events with high signal to noise ratio. The total seismicmoment release (SM0) and energy release (SE) during the monitoring periods has been estimated from thisdata and the use of standard scaling laws (Table 3). Stress drop was assumed 0.1 MPa and shear modulus was

577

Fig. 7. Typical recording of multiple-event micro-earthquakes at NG (no magnitude determinationbecause of uncertain epicenter); (left) seismic traces with spectrogram; (right) locations of seismicstations, epicenter estimated near station 2;

Fig. 8. Recording of dynamite shot at HA (MW = 1.3); (left) seismic traces with spectrogram;(right) locations of seismic stations and shot;

taken 2 GPa for both mass movements.An interesting seismological aspect of monitoring micro-earthquakes caused by mass movements is the

possibility to estimate total seismic moment release (MM0) and total mechanical energy release (ME) fromgeodetically observed displacement (s), moving mass, shear modulus, and an estimate of the seismic fault area(SA). We assume SA = 79 m2 for GB and SA = 645 m2 for HA as average seismic fault areas for the twomonitoring campaigns under consideration in order to equalize SM0 and MM0. The ratio SE/ME is seismicefficiency. Number of seismic events (N), displacement, average seismic fault area, total mechanical energyrelease, and seismic efficiency are also compiled in Table 3. The ratio of the total seismic fault area (N × SA)to the area of the sliding plane (see A0 in Table 1) is about 10−3. We may conclude from this low value, thatmost internal deformation or sliding occured silently or with seismic emissions below our detection limit. Thisconclusion may be justified even the estimates of the parameters compiled in Table 3 are only tentative.

At present we cannot decide, which seismic parameter is the most suited to quantify the developmentof the whole mass movement and especially the state of friction at the sliding surface. This is not only because

578

Fig. 9. Typical recording of a narrow frequency-band event at GB; (left) seismic traces with spec-trogram; (right) locations of seismic stations;

Table 3. Total seismic moment (SM0), total mechanical moment (MM0) release, total seismicenergy release (SE), number of seismic events (N), displacement of centre of gravity (s), averageseismic fault area (SA), total mechanical energy release (ME), and seismic efficiency (SE/ME)for two monitoring campaigns on GB and HA.

of the uncertainties in the seismologic data we acquired so far. Our monitoring campaigns with 5 or moreseismic stations lasted only 8 days. This is a too short time span to determine displacement geodetically withsufficient accuracy (especially at HA with an average displacement of about 10−4 m over 8 days). Therefore,much longer observation periods with more than 5 stations will be carried out in near future.

As stated in the introduction the main objective of seismic monitoring on slopes deformed by deep-seated gravitational creep is acquisition of additional information on the mechanics of deformation and the stateof friction on the sliding surface. The work done so far solved several methodological problems and broughtvaluable insight into seismic activity generated by the mass movements. We think, our preliminary resultspromise to substantially supplement traditional geodetic, geotechnical, and hydrological monitoring methods.

Acknowledgements

The investigations on Gradenbach and Hochmais-Atemskopf have been funded by the Austrian Academyof Sciences within the scope of IDNDR and ISDR. The work on Niedergallmigg-Matekopf is part of the projectA2.4 of alpS — Centre of Natural Hazard Management, Innsbruck, Austria. We gratefully acknowledge thesupport of W. Chwatal during the integration of the seismic velocity models.

References

Ampferer, O. (1939) Uber einige Formen der Bergzerreißungen, Sitz. Ber. Akad. Wiss. Wien, m.-n.Kl., 148,1–14.

Bisci, C., Dramis, F., and Sorriso-Valvo, M. (1996) Rock Flow (Sackung). In Landslide Recognition, (Ed.Dikau, R., Brunsden, D., Schrott, L., Ibsen, M.), John Wiley & Sons, New York, 150–160.

Bruckl, E., Bruckl, J., and Heuberger, H. (2001) Present structure and pre-failure topography of the giant

579

rockslide of Kofels. Zeitschrift fur Gletscherkunde und Glazialgeologie, 37, 1, 49–79.Bruckl, E., Zangerl, Ch., and Tentschert, E. (2004) Geometry and deformation mechanisms of a deep-seated

gravitational creep in crystalline rocks. In Proceedings of the ISRM Regional Symposium EUROCK 2004& 53rd Geomechanics Colloquy, editor W. Schubert, 229–232.

Bruckl, E. and Parotidis, M., (2005) Prediction of slope instabilities due to deep-seated gravitational creep.Natural Hazards and Earth System Sciences 5:155–172 SRef-ID: 1684–9981/nhess/2005-5-155.

Bruckl, E., Brunner, F. K., and Kraus, K. (200X) Kinematics of a deep-seated mass movement derived fromphotogrammetric, GPS and geophysical data. Engineering Geology, in press.

Brunner, F. K., Zobl, F., Gassner, G., 2003. On the capability of GPS for landslide monitoring. Felsbau (Rockand Soil Engineering), 21, 2, 51–54.

Chwatal, W., Kirschner, H., Bruckl, E., and Zanhgerl, C. (2005) Geology and 3D seismic structure of theNiedergallmigg-Matekopf mass-movement, Tyrol, EGU, General Assembly, Vienna, Austria, 24–29 April2005. Geophysical Research Abstracts, Vol. 7, 02566, 2005. SRef-ID: 1607-7962/gra/EGU05-A-02566.

Dieterich, J. (1992) Earthquake nucleation on faults with rate- and state-dependent strength, Tectonophysics,211, 115–134.

Dikau, R., Brunsden, D., Schrott, L., and Ibsen, M.L.: Introduction, in Landslide Recognition, editor R. Dikau,D. Brunsden, L. Schrott, M.L. Ibsen, John Wiley & Sons, England, 1–12, 1996.

German, V. and Bruckl, E. (2006) Hazard estimate of rockslides based on a velocity-displacement weakeningrotational slider block model. EGU, General Assembly, Vienna, Austria, 02–07 April 2006. GeophysicalResearch Abstracts, Vol. 8, 07597, 2006. SRef-ID: 1607–7962/gra/EGU06-A-07597.

Gottschling, H. (1999) Der Talzuschub Gradenbach — ein geologisch-geotechnischer Uberblick, Jb. Geol. B.-A.,Wien, 141/4, 345–367.

Hanks, T.C., and Kanamori, H. (1979) A moment magnitude scale, J. Geophys. Res., 84, 2348–2350.Helmstetter, A., Sornette, D., Grasso, J.-R., Andersen, J.V., Gluzman, S., and Pisarenko, V. (2004) Slider-

block friction model for landslides: Application to Vaiont and La Clapiere landslides, J. Geophys. Res. 109,B02409, doi:10.1029/2002JB002160.

Husen van, D. (1987) Die Ostalpen und ihr Vorland in der letzten Eiszeit (Wurm). Map 1: 500 000, GeologischeBundesanstalt, Wien.

Hutchinson, J.N. (1988) General Report: Morphological and geotechnical parameters of landslides in relationto geology and hydrology. In: Proc. 5th International Symposium on Landslides (Ed. Bonnard, C.), A.Balkelma, Rotterdam, Netherlands, 1, 3–35

Joswig, M. (1990) Pattern Recognition for earthquake detection, Bulletin of the Seismological Society of Amer-ica, 80, 1, 170–186.

Kirschner, H.; Zangerl, Ch., and Brandner, R. (2004) Structural influence on deformation mechanisms of slopeinstabilities in crystalline rock, Tyrol, Austria, 32nd International Geological Congress, Florence, Italy,20–28 August 2004.

Kronfellner- Kraus, G. (1980) Neue Untersuchungsergebnisse in Wildbachen - Der Talzuschub in Abhangigkeitvon Niederschlagen. Int. Symp. IINTERPRAEVENR Bad Ischl, 1, 179–192.

Lomax, A., J. Virieux, Volant P., and Berge, C. (2000. Probabilistic earthquake location in 3D and layeredmodels: Introduction of a Metropolis-Gibbs method and comparison with linear locations, in Advances inSeismic Event Location Thurber, C.H., and N. Rabinowitz (eds.), Kluwer, Amsterdam, 101–134.

McGarr, A., and Bicknell, J. (1990) Estimation of near-fault ground motion of mining-induced tremors from lo-cally recorded seismograms in South Africa. Rockbursts and seismicity in mines, editor Fairhurst, Balkema,Rotterdam, ISBN 906191 1 45 1, 245–248.

Moser, M. & Kiefer J. (1988) Die hydrologischen Verhaltnisse und ihre Beziehungen zur Kinematik im Bereichder Talzuschubsmasse Gradenbach / Karnten. Steir. Beitr. z. Hydrologie, Graz., 39, 95–115.

Muller, L. (1964) The rock slide in the Vajont valley, Felsmech. Ingenieurgeol., 6, 1–91.Podvin, P. and Lecomte, I., (1991) Finite difference computation of traveltimes in very contrasted velocity

models: a massively parallel approach and its associated tools., Geophys. J. Int., 105, 271–284.Roth, M., Dietrich, M., Blikra, L. H., Lecomte, I. (2005) Seismic monitoring of the unstable rock slope site

at Aknes, Norway. 19th annual Symposium on the Application of Geophysics to Engineering and Environ-mental Problems (SAGEEP), Seattle, Washington

Ruina, A. (1983), Slip instability and state variable friction laws, J. Geophys. Res., 88, 10,359–10,370.Savage, W.Z., and Varnes, D.J. (1987) Mechanics of gravitational spreading of steep-sided ridges (“sackung”),

International Assoc. of Engineering Geologists Bulletin, 35, 31–36.Spillmann, T., Heincke, B., Willenberg, H., Maurer, H.R., Green, A.G., Loew, S. (2006) Application of ac-

tive and passive seismic methods for characterizing fracturing processes in an unstable rockmass (Randa,Switzerland). EGU General Assembly 2006, Vienna, Austria.

Stiny, J. (1941) Unsere Taler wachsen zu, Geologie und Bauwesen, 13, 3, 71–79.

580

Tentschert, E. (1988) Das Langzeitverhalten der Sackungshange im Speicher Gepatsch (Tirol, Osterreich),Felsbau, 16/3, 194–200, 1998.

Voight, B. (1988) A method for prediction of volcanic eruptions, Nature, 332, 125–130.Zischinsky, U. (1969) Uber Sackungen, Rock Mechanics 1, 30–52.