main scarpminor scarpmud flowscrees

15 km

Isère riv.

Dracriv.

Ebronriv.

Clelles

Sinard

Drac riv.

Vizille

IsèreGlacier

GrenobleVercorsmassif Belledonne

massif

Corps

La Mure

Varved clays depositsin the Trièves areaExtension of the Isère glacierat the end of the Würm period

Dévoluymassif

Studiedarea

Paris

Lyon

Marseille

Grenoble

200 km

Study area

Chartreusemassif

I- INTRODUCTION / GEOLOGICAL CONTEXT

Figure 1: Location of the study area, extension of the varvedclays and of the Isère glacier; after Monjuvent (1973).

(1) Laboratoire de Géophysique Interne et de Tectonophysique, Université Joseph Fourier, Grenoble, France (gregory.bievre@ujf-grenoble.fr)(2) Centre d’Études Techniques de l’Équipement de Lyon, Laboratoire Régional d’Autun, Autun, France

Grégory BIÈVRE (1, 2), Ulrich KNIESS (1), Denis JONGMANS (1), Stéphane SCHWARTZ (1) & Vilma ZUMBO (2)

The Trièves area is a large depression (800 m asl) located within theFrench alpine foreland 40 km south of the town of Grenoble. It coincideswith the position of a 300 km2 palaeolake created by the damming of theDrac River by the Isère glacier during the last maximum glacial extension(Würm period, 45 000 yr BP, Fig. 1). The lake was progressively filled byvarved clays resulting from the erosion of the surrounding crystallineand limestone massifs. These clayey formations which can reach 250 mthick overlay an irregular paleo-topography made of compact alluviallayers and mesozoic marly limestones. After the glacier retreat (10,000 to15,000 years BP) the rivers cut deeply into the geological formations, trig-gering landslides in the clay deposits over at least 15% of the Trièves area.Some of these slides might affect surfaces as large as 500,000 m2 with thedeepest slip surface estimated or measured at about 40 m depth.To the north of the Trièves area, the left bank of the NS oriented artificialMonteynard lake (Fig. 2) is affected by several large imbricated land-slides. Our study is focused on two of these landslides (location on Fig. 2).

The translational Avignonet slide affects a surface of about 1.106 m² witha global eastward downslope motion towards the lake. Just south of thisslow moving slide, a quick mudslide (L‘Harmalière) occurred in March1981, creating a head scarp of 30 m high and affecting a surface of about450,000 m² in the same material. Between 1981 and 2004, the head scarphas continuously regressed with an average of 10 m/year in a north-eastward direction and now intersects the southern limit of the Avignonetlandslide. The global direction of the landslide motion is SE, making anangle of about 45° with the topography slope.

This differential motion between the two slides, although they develop inthe same material and exhibit comparable settings, remains difficult tounderstand without taking into account a possible role of thepalaeotopography upon which settled the glaciolacustrine clays. The aimof this work is to present the results of a geomorphological mapping,built from geodetic (LiDAR) and geophysical (ambient vibration) mea-surements, that proposes a map of the base of the varved clays thatallows to understand the 2 slides differential motion.

Figure 7: DEM & orthophotography of the study area. Red dots standfor measurement points. Coordinates are in meters, according to French

system Lambert-93.

HV1T1

HV21

Vertical exaggeration: x 1.3

6431500

6431000

6430500

6430000

6429500

6429000

6428500

20 80 140 200 260 320

910000 910500 911000 911500

Computed clay thickness (m)

Measured resonance frequency variogram

Direction: 0.00Tolerance: 90.0

0 200 400 600 800 1000 1200 1400Lag distance (m)

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Vario

gram

Calculating the fundamental ellipticity curves of Rayleigh waves, the resonance frequency below each station was comput-ed from the 1D geometry and the properties of the geological layers (moraine, clay, alluvium and bedrock). Data calibration(thickness of each layer) was obtained from cross-section ZZ’ of Figure 3 where a detailed H/V profile was conducted(location on Fig. 6). The considered characteristics (compressional wave velocity Vp, shear wave velocity Vs and density) ofthe layers are given in Table 1. Vp and Vs values were obtained from our study in the shallow layers and from data foundin the literature. Theoretical and experimental values of the resonance frequency f0 are compared in Figure 10a and show avery good agreement. Sensitivity tests have shown that the main factor controlling f0 is the total clay thickness. On the con-trary, no change in the f0 value is found when passing the landslide top. Figure 10b presents the comparison of our results(base of the varved clays) with previous ones (Blanchet, 1988) depicting the top of the Jurassic bedrock. A good agreementis observed between the 2 sections, the difference being explainable by the alluvial layers.

6431500

6431000

6430500

6430000

6429500

6429000

6428500910000 910500 911000 911500

0.4 1 1.6 2.2 2.8 3.4 4Measured resonance frequency (Hz)

HV1T1

HV21

Detailedprofile P01for signalcalibration

0 200 400 600 800 1000 1200 1400Lag distance (m)

0

2000

4000

6000

8000

10000

Vario

gram

Computed clay thickness variogram

Direction: 0.00Tolerance: 90.0

Figure 11: Clay thickness map.Figure 8: Example of ambient vibration measurement.

HV1T1

0

4

8

12

16

0.2 0.4 0.8 2 4 6 10 20 40 60Frequency (Hz)

H/V

ampl

itude

0.58 Hz

HV21

0

4

8

12

16

0.2 0.4 0.8 2 4 6 10 20 40 60Frequency (Hz)

H/V

ampl

itude

3.63 Hz

Frequency peak Average frequency (Hz) Standard deviation

Measured resonance frequency (Hz)0 1 2 3 4 5

Com

pute

dre

sona

nce

frequ

ency

(Hz)

0

1

2

3

4

5

Vp (m/s) Vs (m/s) DensityMorainic deposits 500 250 1.9

Slipped clays 1850 300 2Varved clays 1850 600/800 2

Compact alluvium 2350 1250 2Jurassic bedrock 3000 2000 2.6

Table 1: Considered parameters for the computationof the resonance frequency

Figure 9: Frequency map computation.

These results indicate that the H/V method is a robust tool for determin-ing soft deposit thickness, as already observed by several authors (e.g.Delgado et al. (2000)). In the Trièves area, this method can then be usedfor mapping the clay thickness variations, which is one of the majorinformation for understanding landslide distribution.

III- AMBIENT VIBRATION MEASUREMENTS

II.1- Resonance frequency mapping

II.1- Clay thickness determination

Combination of geological, geomorphological and geophysical approaches show to be an efficient way to understand the differential behaviour between the two adjacentslides. All the acquired data suggest that the palaeotopography of the bedrock is the main parameter controlling the geometry and the dynamics of the two studied land-slides.The Avignonet landslide moves slowly perpendicular to a NS oriented bedrock ridge, while the Harmalière mudslide developed over a former talweg which has allowedand still allows the slide and flow of a large volume of clayey material into the lake.Complementary work, especially on the clay vertical seismic velocities, as well as on a better cover of the area with ambient vibration measurements will lead to gain moreconfident results and produce a more precise geomorphological map of the base of the varved clays formation.

Bard P.-Y. (1998) - Microtremor measurements: a tool for site effect estimation ? - Proceeding of the Second International Symposium on the Effects of Surface Geology on Seismic Motion, Yokohama, Japan, 3, 1251-1279.

Blanchet F. (1988) - Etude géomécanique de glissements de terrain dans les argiles glaciolacustres de la vallée du Drac. - Ph. D. thesis, Grenoble university, 157 p.Bonnefoy-Claudet S., Cornou C., Bard P.-Y., Cotton F., Moczo P., Kristek J. & Fäh, D. (2006) - H/V ratio: a tool for site effects evaluation. Results from 1-D noise simulation. - Geophys. J. Inter., 167, 827-837.Delgado J., Lopez Casado C., Estevez A., Giner J., Cuenca A., & Molina S. (2000) - Mapping soft soils in the Segura River valley (SE Spain): a case of study of microtremors as an exploration tool. - J. Appl.

Geophys., 45, 19–32.Monjuvent G. (1973) - La transfluence Durance-Isère. Essais de synthèse du Quaternaire du bassin du Drac. - Géologie Alpine, 49, 57-118.Vallet, J. & Skaloud J. (2004) - Development and experiences with a fully-digital handheld mapping system operated from a helicopter. - The International Archives of the Photogrammetry, Remote Sensing

and Spatial Information Sciences, Istanbul, Vol. XXXV, Part B, Commission 5.Wathelet M., Jongmans D. & Ohrnberger M. (2004) - Surface wave inversion using a direct earch algorithm and its application to ambient vibration measurements. - Near surface Geophysics, 2, 211-221.

REFERENCES

Figure 4: Flowchart for LiDAR data processing.

Geomatica®

scop++®

Matlab®

Helimap system® mounted on helicopter

xyz point cloud

GPS receiver intertial measurementunitlaser scanner

Filter helicopterreflections, sorting

classificationbare earth, trees, buildings

rasterize

DEM analysis

shaded relief slope map profiles

Raster DEM

manualfiltering

test areapoint cloud

rasterize

difference

Quality Check

Light detection and ranging (LIDAR) data come from the helicopter-borne mapping Helimap system® (Vallet& Skaloud 2004), that acquired simultaneously numerical photos. The height of flight was about 500 m abovethe ground, allowing a density measurement of 5 points by square meter in average, with a positioning accura-cy of ~10 cm both in horizontal and vertical. After filtering houses and trees from the LiDAR point cloud, aDigital Elevation Model (DEM) of 2 m resolution was produced. The quality of the automatic filter process waschecked by testsites, where the point-cloud was classified manually.

Figure 5: Shaded LiDAR DEM map with GPS velocity mea-surements.

II- LiDAR DATA

Base of the varved clays (elevation, m NGF)

6431500

6431000

6430500

6430000

6429500

6429000

6428500910000 910500 911000 911500

370 430 490 550 610

A

B

C

Base of the varved clays (elevation, m NGF)

Cross sectionsAvignonet & Harmalière slides limitsAmbient vibration measurements

800

600

400

0 500 1000 1500 2000

500 1000 15000

A B

A C

800

600

400

Distance (m NGF)

Ele

vatio

n(m

NG

F)E

leva

tion

(mN

GF)

LiDAr topography

Base of varved claysafter H/V measurements

Varved clays

Mon

teyn

ard

lake

Monteynard dam

Z Z’

Dra

criv

.

6433000

6432000

6431000

6430000

6429000

6428000

909000 910000 911000 912000

A A’

B B’Harmalière

landslide

Avig

none

tla

ndsl

ide

T0

T2T1

600

700

800Elevation, m

0 200 400 600 Distance, m

?

W Morainic depositsVarved claysAlluvial depositsSlip surfaceGPS measurements41

41444748

Planimetric cumulated displacements (m)0 0.2 0.4 0.6 0.8 1.0 1.2

0

-0.1

-0.2

-0.3

-0.4

1.4c

Alti

met

riccu

mul

ated

disp

lace

men

ts(m

)

1991/12/101992/09/241993/04/26

Cumulated displacements (cm)0 2 4 6 8 10 12 14

Dep

th(m

)

-18-16-14-12-10-8-6-4-20

b

41 (14.1)

44 (4.6)

47 (2.7)

48 (2.15)

46 (2.4)

42 (8.4)

45 (2.9)

Z

Ea

Z’

1 2 3Distance, km

W E

520m610m

A A’

500

1000

Z(m

,NG

F)

1 2 3Distance, km

W E

ADRGT (1988)

Morainic deposits

Varved clays

Alluvium(Sinard Palaeodrac)Alluvium(Cros Palaeodrac)

Jurassicbedrcok

B B’

500

1000

Z(m

,NG

F)

Figure 3: a) Geotechnical cross section ZZ’ (location on Fig. 2) with the positionof GPS points along the slope (displacement rate in cm/yr are indicated betweenparentheses) and the slip planes deduced from inclinometer data in boreholes T0,T1 and T2. The vertical scale is exaggerated. b) Cumulated displacements curveat inclinometer T2; reference curve was measured on 1990/04/10. c) Displace-ments in the horizontal and vertical planes at GPS points 41, 44, 47 and 48(between 1995/10/01 and 2005/10/26).

Figure 2: Geological map (moraines not represented) and geological cross-sections of the study area. Coordinates are in meters, according toFrench system Lambert-93.

Non-perennial

PerennialStream

Varved clays

Compact alluvial layers

Jurassic bedrock

IV- DISCUSSION

V- CONCLUSIONS & PERSPECTIVES

Influence of the geological structure on the geometry and dynamics of two large landslideswithin glaciolacustrine clays (Trièves area, french Alps)

Figure 6: Morphological interpretation of the LiDAR DEM.3 mm/y

10 mm/y

100 mm/yGPS

500 m

Geomorpholgy methodTwo with different lightning azimuth angles shaded Lidar DEMs, the2003 aerial orthophotos, stereoscopic aerial photos and field observa-tions were used to map general geo-morphological features of the twolandslides. Indicated are also average GPS displacements of biannualmeasurements between 1995 and 2006 (Fig. 5).

Geomorphology interpretationThe Avignonet slide has moved slowly since that the first disturbanceswere observed at the end of the seventies. The GPS measurements haveshown up surface displacement rates between 1 to 13cm/year (Fig. 5). Incontrast, the Harmalière slide evolved into quick mudflow in March1981, creating a head scarp of 30 m and affecting a surface of about450,000 m2 in the same clay material. Parts of higher surface displace-ment velocities seem to be correlated to areas with higher roughness oftopography.With the Lidar DEM numerous scarps previously unmapped have beenidentified in particular in forested areas (Fig. 6). The western part ofAvignonet shows steeper slopes with screes, which indicate the pres-ence of the harder alluvial layer and bedrocks which are slowly erodedfrom the west.

Harmalière & Avignonetslides extension

N

500 m

6431500

6431000

6430500

6430000

6429500

6429000

6428500910000 910500 911000 911500

370 430 490 550 610Base of the varved clays (elevation, m NGF)

N

500 m

N

500 m

N

1 km

N

1 km

TopographyJurassic bedrock top(Blanchet, 1988)Base of varved claysafter H/V measurements

800

700

600

500

400

Ele

vatio

n(m

NG

F)

0 200 400 600 800Distance (m)

W E

Figure 10: a) Parameters used for thickness computation; b) measured versuscomputed theoretical frequencies for the whole points; c) comparison of H/V Pro-

file P01 (location on Fig. 6) and bedrock position after Blanchet (1988).

370 430 490 550 610

Vertical exaggeration: x 1.3

Palaeovalley acting as an outlet for theHarmalière landslide

Geomatica®

Point CloudMapper

Figure 12: Base of the varved clays. a) 3D representation of the varved clays basal surface; b) Contour map showing the position of the two slides.c) Geomorphological interpretation of the map. Coordinates are in m, according to French system Lambert-93.

Figure 13: Cross-sections representing the base of the varved clays after H/V measurements. Cross-sections ABand AC are located on Fig. 12b, and AA’ on Fig. 2. Croos-section AA’ after ADRGT (1988).

A A’B B’

Z Z’

Cross sections above

Cross section of Fig. 3

N

Extensive geotechnical investigation was performed on the Avignonetlandslide and showed the existence of several slip surfaces, from verysuperficial ones (a few m) to deep ruptures at 40 m (Figs. 3a & 3b). Theslide velocity at the surface, measured by GPS during more than 10 years,increases downhill, varying from 0 to 2 cm/y at the top to more than 14cm/y at the toe (Fig. 3c). These geodetic data are consistent withgeomorphic observations (scarps, fissures, swampy areas, bulges), whichshow an increase of the landslide activity downstream. Piezometric datahave pointed out a shallow water table (a few m to 1 m deep) which fluc-tuates with rainfalls. If some data show a relation between water leveland displacement rate, the highest slide velocity values do not seem todepend on rainfalls. In case of heavy rainfalls or quick snow melt, thiskind of slide might evolve into a mudflow, as shown by the events ofL’Harmalière (1981) and of La Salle en Beaumont (1994), as well as bynumerous solifluction flows.

N500 m

N

Seismic noise measurements have been increasingly used during these last fifteen years in earthquake engineering for determiningthe geometry and shear wave velocity values of the soil layers overlying the bedrock (Bonnefoy 2006; Wathelet 2004). The singlestation method (also called the H/V technique) was used at 78 locations (Fig. 7). It consists in calculating the horizontal to verticalspectral ratio of the noise records and allows the resonance frequency of the soft layer to be determined (Nakamura 1989; 2 exam-ples are presented in Fig. 8). For a single homogeneous soft layer, this fundamental frequency is given by f0 = Vs/4H where Vs isthe soft layer shear wave velocity and H is the layer thickness.

After individual signals quality check (Fig. 8), a frequency map canbe produced using gridding and geostatistical tools (Fig.9)

a

b

The two adjacent landslides, which exhibit very different geometrical and dynamical characteristics, occur on identical slopes made of the same material. The combination ofmorphological, geological and geophysical investigations allows the understanding of this differential behaviour. The DEM points out the boundaries and the internalgeometry of the slides. The Avignonet landslide exhibit a crescent shape with a longitudinal convex profile at the toe. On the contrary, the Harmalière slide is more elongat-ed with a funnel shaped track zone through which the material flows to the lake with a concave regular slope (Figs. 5, 6 & 13). Geological mapping below the Avignonetlandslide showed that the bedrock and compact alluvial layers outcrop along the lake and constitute a buttress which prevents deep slip surfaces to daylight on the slope.On the contrary, the Harmalière landslide developed at a site where compact alluvial layers do not outcrop along the lake. This might correspond to a former NW-SE orient-ed talweg of the Drac (Fig. 2). The base of the varved clays surface, computed from the substraction of the clay thickness map (Fig. 11) from the DEM. (Fig. 5) is concordantwith the geological evidence (Figs. 12 and 13).

Varved clays

Jurassic bedrock

Alluvial compact layers

Distance (m NGF)

Undiferrenciated alluvial layersand jurassic bedrock

Undiferrenciated alluvial layers and jurassic bedrock

1 2 3Distance, km

W E

A A’

500

1000

Z(m

,NG

F)

Morainic deposits

Varved claysAlluvium(Sinard Palaeodrac)Alluvium(Cros Palaeodrac)JurassicbedrcokBase of varvedclays (this work)

Comparison of our results with the cross-section of Figure 2 indicates a relativelygood agreement. Nevertheless, between abscissa 1 and 1.5 km, a more than 100 m dif-ference is observed. It may come either from an underestimation of the clay thicknesscalculation after H/V measurements or from an overestimation of clay thickness fromADRGT (1988). Complementary measurements and more detailed clay thickness cal-culation parameters may help to overcome this point.

Exponential modelExponential model

Higher mode orsuperficial layer

BU

TRES

S