Réf. Cossart & al. 2008

Référence bibliographique complète

COSSART, E., BRAUCHER, R., FORT, M., BOURLÈS, D.L., CARCAILLET, J. 2008. Slope instability in relation to glacial debutressing in alpine areas (upper Durance catchment, southeastern France): Evidence from field data and 10Be cosmic ray exposure ages. Geomorpholgy, 95, 3–26.

Abstract: The Upper Durance catchment is an area prone to rock-slope failures. Such failures reflect the combination of high relief, lithostructural controls and paraglacial stress release. The aim of this study is to determine the role of deglacial unloading and resulting paraglacial stress release in conditioning or triggering slope failure. Former dimensions of the Durance glacier are reconstructed, then combined with Digital Elevation Model data in a raster Geographic Information System to quantify the spatial pattern of stresses associated with glacial loading at the Last Glacial Maximum. Preliminary calculations suggest that major rock falls and rock avalanches are associated with areas subject to the highest decompression stresses. Focus on two case studies allows the consequences of paraglacial stress release on slope instability to be evaluated. Description of slope failure runout deposits allows reconstruction of the nature of slope failure. Surface exposure dates based on concentration of cosmogenic 10Be allows the timing of both deglaciation and that of post-glacial rock-slope failures to be established. It is shown that rock-slope failures are concentrated on lower valley-side slopes within the area occupied by ice at the Last Glacial Maximum, and that their locations coincide with zones of inferred high glacial loading stress, consistent with interpretation of both bedrock disruption and large-scale rock-slope failures as paraglacial phenomena induced by stress release following deglaciation. Timing of initial rock avalanche runout deposition at one site is consistent with this conclusion, though later instability episodes at the same site may have occurred independent of the influence of paraglacial stress release.

Paraglacial; Rock-slope failure; Glacial debuttressing; Stress release; Cosmogenic nuclides; Late Glacial; French Alps

Organismes / Contact
• Equipe DYNMIRIS, UMR PRODIG — 8586 CNRS, Universités Paris 1-7, 2 rue Valette, 75005 Paris, France — Corresponding author (E. Cossart): etienne.cossart@univ-paris1.fr
• CEREGE, UMR 6635 — CNRS, Université Aix-Marseille 3, Europôle Méditerranéen de l'Arbois, BP 80, 13545 Aix en Provence cedex 04, France

(1) - Paramètre(s) atmosphérique(s) modifié(s)
(2) - Elément(s) du milieu impacté(s)
(3) - Type(s) d'aléa impacté(s)
(3) - Sous-type(s) d'aléa
    Mass movements Rock-slope failures (rock avalanche)

Pays / Zone
Massif / Secteur
Site(s) d'étude
Période(s) d'observation
France / Southern Alps Upper Durance catchment Case studies:
- Post-glacial modification of roches moutonnées in the Upper Clarée valley
- Pré de Madame Carle rock-slope failure (upper Vallouise catchment)

(1) - Modifications des paramètres atmosphériques

Informations complémentaires (données utilisées, méthode, scénarios, etc.)

(2) - Effets du changement climatique sur le milieu naturel

Glacial extent and quantification of basal buttressing:
The Durance palaeo-glacier was one of the main alpine glaciers during the Last Glacial Maximum. It extended 200 km downvalley, and terminated close to Sisteron. [The authors] estimate that the minimal thickness of the glacier in the Upper Durance area ranged from 600 m to 900 m in the tributary valleys. The maximum estimated values occur in the Briançon basin and in the lower part of Vallouise and Guisane valleys, where glacier thickness locally reached at least 1000 m. The inferred gradient of the former glacier surface was steepest in the Vallouise valley, where it locally reached 15%. As a consequence, estimated debuttressing (stress-release) values are the highest in these western valleys. In both the Vallouise and Guisane valleys, estimated normal stresses (7000–8000 kPa) and longitudinal stresses (200–300 kPa) due to glacier overburden were greatest. In contrast, in the eastern part of the study area (Clarée, Cerveyrette and Guil valleys), the reconstructed ice thickness is less than 800 m and is associated with a relatively low surface gradient (less than 10%). For these valleys, the calculated normal stress is 5000–7000 kPa and the longitudinal stress 50–200 kPa. Thus, inferred debuttressing values are significantly lower in these areas than in the western valleys.

Chronological benchmarks:
In the Upper Clarée Valley, the approximate thickness of ice cover at the Last Glacial Maximum is constrained by the difference between the 10Be age obtained for the upper part (2170 m; 28.8±5.6 10Be ka) and the lower part (1880m; 10.5±1.3 10Be ka) of the valley slope. Although former ice limits cannot be determined, cosmic ray exposure dates from polished, roche-moutonnée surfaces in the Upper Clarée Valley imply that the higher parts of the catchment became progressively ice free during the Early Holocene, between 10.5±1.3 10Be ka and 7.8±0.9 10Be ka. In the Vallouise valley, the last glacial advance prior to the Little Ice Age advance is recorded by the Ailefroide moraines (Cossart, 2005). A well-preserved polished rock surface that had been deglaciated before this stage yielded an exposure age of 7.4±1.1 10Be ka, consistent with identification of a major stage of deglaciation in the upper valleys during the Early Holocene. Collectively, the dating evidence indicates that the upper valleys were occupied by glacier ice until the Late Glacial, and that glacier termini retreated to altitudes above 2000 m in the Early Holocene. [...]


Sensibilité du milieu à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)
  [See below...]

(3) - Effets du changement climatique sur l'aléa

Slope failure at a regional scale:
Slope failure sites are widespread in the Upper Durance catchment. In this study, 81 slope failures with volumes >100 m3 have been identified, mapped and classified according to the three types defined [in the paper]. The resulting map hows that different types of failure are associated with specific areas: rock avalanches are found around the Massif des Ecrins, earthflows are frequent in the Guil, Cerveyrette and Guisane valleys, and rock topples mainly occur in the upper Clarée catchment. The distribution of major rock falls is strongly associated with areas affected by the highest decompression values (σ=5000–9000 kPa and τ=200–300 kPa) and the steepest slope gradients (100 to 150%). In the Vallouise area, 34 slope failures were identified and mapped: 6 major rock falls and/or rock avalanches, 9 earthflows and 19 translational rockslides. Comparison of the altitudinal ranges of these slope failures with the reconstructed trimline altitude shows that the slope failures mainly occurred in formerly glaciated areas. The largest slope failures are located where the former longitudinal stress was high (up to 250 kPa), such as in the Gyr valley, the Pelvoux basin, and the lower part of the Onde Valley. Conversely, the sites of rock topples and earthflows tend to be associated to moderate former basal stresses (σ=000–5000 kPa and τ=100–200 kPa), but these failures occur in specific lithological settings [...]. Collectively, these data show that lithology has conditioned the type of slope failure in most cases. However, the spatial coincidence of most rock-slope failures with areas where glacial loading imposed high normal and longitudinal stresses buttressing forces, together with the preponderance of slope failures located below the trimline, clearly supports the notion that paraglacial stress release played a major role in rockslope weakening and failure. [...]

Case study - Post-glacial modification of roches moutonnées in the Upper Clarée Valley:
[...] The observations can be encapsulated in a typology of rock topples and block displacement that permits discrimination between structural features (type A) that occur across a wide range of altitudes, and paraglacial stress-release features (type B), that are restricted to sites where glacial loading and hence stress release was at a maximum (i.e. on lower valley-side slopes and on the stoss faces of roche moutonnées). Attempts were made to date both types of feature. Exposure dating of a type A topple failed because of the high concentration of boron in the sample, but an exposure age obtained from the failure plane of a type B rock topple (10.1±2.1 10Be ka) is consistent with block displacement very soon after deglaciation of the site. This result is consistent with the hypothesis that bedrock displacement and associated block toppling occurred as a result of paraglacial stress release in the immediate aftermath of glacial unloading.

Case study - Pré de Madame Carle rock-slope failure in the upper Vallouise catchment (next to Mont Pelvoux):
[...] Surface exposure dating of boulders at this site suggests that there were at least two stages of runout deposition [deglaciation of this site at 7420±1079 10Be yr ; emplacement of the lower part of the runout deposit at ±6.5 10Be ka ; and a second failure event several millennia after the first, at ~1.5 10Be ka]. Collectively, the sedimentological, morphological and dating evidence suggests three stages in the evolution of the Pré de Madame Carle site. At the Last Glacial Maximum, the Vallouise glacier was at its maximum thickness (stage 0), imparting maximum overburden stresses on the underlying bedrock. During stage 1, immediately after deglaciation, the initial failure appears to have occurred on lower slopes that had experienced maximum glacial loading, and hence were subject to the greatest stress release. Debris runout associated with this event accumulated on the valley floor, damming the river and causing the aggradation of the flat 'Pré' (meadow) immediately upvalley. Both the location of the failure plane in the zone of maximum glacial loading and the timing of failure imply that this initial failure was, at least in part, influenced by paraglacial stress release and joint development. At stage 2 several millennia after deglaciation, destabilization and failure propagated upslope into a zone of steeper slope gradients. This later failure event was largely sourced from above the maximum altitude of former glacier cover, and hence appears unlikely to reflect the direct influence of paraglacial stress release.

Results of the spatial approach:
Comparison within a GIS framework of the distribution of rock-slope failures with zones of inferred high stresses imposed by glacial loading provides a new method for integrating the influence of paraglacial stress release into models of slope instability. It has been shown that the main unstable areas are consistently located in areas formerly occupied by glacier ice. This pattern conforms with observations in other alpine mountain areas (e.g. Shreve, 1966; Porter and Orombelli, 1981; Evans and Clague, 1994; Augustinus, 1995). The influence of paraglacial stress release on rock-slope instability is here supported by two sets of data. At a regional scale, the location of landslides is associated with high normal and longitudinal basal stresses imposed by glacial loading; the location of rock avalanches is particularly strongly associated with sites of inferred high stress release. At a more localised scale, the hypothesis of a paraglacial stress-release origin for post-glacial deformation of roches moutonnées is supported by both the geometry and location of stress release joints. [...] [see Clarée valley case study...]

Results of the chronological approach:
Surface exposure dating of deglaciated rock surfaces using cosmogenic 10Be has allowed the timing of deglaciation in the study area to be refined, and the deglaciation history to be related to that previously established in the northwestern part of Massif des Ecrins. Exposure dating of rock-slope failure events suggests that initial failure occurred fairly soon after deglaciation, consistent with a relationship between deglacial unloading, the rapid development of paraglacial stress-release joints, and rock-slope failure. [...]. The Pré de Madame Carle case study suggests that alpine-scale slopes may continue to exist in a state of critical conditional stability for several millennia after the initial triggering event, with progressive upslope or retrogressive movement of the zone of potential rock-slope failure. This inferred pattern appears consistent with the results of other studies in European mountains that indicate two distinct stages of mass-movement activity: an initial period of slope instability during the Early Holocene, associated with numerous post-glacial landslide events; and a later period of slope instability during Subboreal and Subatlantic chronozones, during which slope failure has often been explained by non-paraglacial triggering involving, for example, increased precipitation (e.g. Alexandrowicz, 1993; Starkel, 1997; Corsini et al., 2001; Soldati et al., 2004). Further investigations and dating are needed to document a correlation between [the present] results and such studies.


Paramètre de l'aléa
Sensibilité des paramètres de l'aléa à des paramètres climatiques
Informations complémentaires (données utilisées, méthode, scénarios, etc.)

Geomorphic evidence for a paraglacial (debuttressing and stress release) origin of rock-slope failure is indicated by spatial coincidence between the location of failures and areas where glacial debuttressing was at its maximum (Panizza, 1973; Holm et al., 2004). The use of GIS makes this comparison easier, but such an approach must be based upon an evaluation of glacial stress release, estimated from the basal stresses applied by a former glacier on adjacent bedrock. This latter parameter varies at three scales (Benn and Evans, 1998). First, at a regional scale it is a function of ice thickness and former ice surface gradient; second at the scale of individual slopes, basal stress is at its maximum close to the valley floor; and third, at a more local scale, basal stress is higher on the stoss face of glacial rock-bars. Thus, coincidences between glacial debuttressing and areas of rock displacement must be verified at all three scales.

This paper couples reconstruction of stresses at all three scales with a chronological framework for deglaciation and rock-slope failure for the Upper Durance catchment (Southern French Alps). [The authors] first document the debuttressing forces implied by former glacial extent and the application of glaciological laws within a GIS frame. Secondly, at a regional scale, [they] inventory landslides and compare their locations with areas where both normal and longitudinal stresses (applied by the former glacier) were at maximum. Thirdly, [they] select sites (i) to consider whether slope failures have been preferentially triggered on the lower part of the slope, where glacially-imposed stresses were highest, and (ii) to compare the jointing patterns of bedrock on the stoss faces of rock-bars (high imposed stress) with those on lee faces (low imposed stress). Finally, some chronological benchmarks provide the opportunity to test whether the timing of slope failure is consistent with proposed paraglacial models of slope evolution.

Methods [see the study]:
- Reconstruction of former glacial extent
- Identification of post-glacial debuttressing impact: (1) Slope instability survey at a regional scale and (2) Smaller-scale effects of post-glacial stress release
- Cosmic-ray exposure dating [10Be]

(4) - Remarques générales

(5) - Syntèses et préconisations

Rock-slope failures are widespread in the Upper Durance catchment: a total of 81 failures have been identified, 62 of which are located in the Vallouise or Clarée valleys. The concentration of failures at different spatial scales within areas where stresses imposed by former glacier loading were high implies that glacial debuttressing and associated stress release have played an important role in triggering rock-slope failures. More specifically, it explains the numerous occurrences of failure sites on lower valley-side slopes (i.e. below the trimline marking the upper limit of glacial erosion) and on the stoss faces of rock-bars, where inferred subglacial stresses were particularly high.
Cosmic ray exposure dating methods are particularly well suited to a chronological framework in such a paraglacial context as they can be applied to establish both the deglaciation age of glacially-polished bedrock surfaces as the age of rockslide runout debris, and thus the timing of rock-slope failure. The dates obtained suggest that the first rock-slope failures occurred soon after deglaciation of the failure site. Subsequent episodes of slope failure, however, may occur independently of the influence of paraglacial stress release. Systematic exposure dating of rockslope failures provides the opportunity to assess the response time of paraglacial slope failures after deglaciation. In the future, it may prove possible to couple such chronological data with GPS measurements and geophysical prospecting to reconstruct sediment budgets in a paraglacial context.

Références citées :

Alexandrowicz, S.W., 1993. Late Quaternary landslides at the eastern periphery of the national park of the Pieniny Mountains, Carpatians, Poland. Studia Geologica Polonica 102, 209–225.

Augustinus, P.C., 1995. Rock mass strength and stability of some glacial valley slopes. Zeitschrift für Geomorphologie 39, 55–68.

Benn, D.I., Evans, D.J.A., 1998. Glaciers and Glaciation. Arnold, London. 734 p.

Corsini, A., Marchetti, M., Soldati, M., 2001. Holocene slope dynamics in the area of Corvara in Badia (Dolomites, Italy): chronology and paleoclimatic significance of some landslides. Geografia Fisica e Quaternaria 24, 127–139.

Cossart, E., 2005. Evolution géomorphologique du haut bassin durancien depuis la dernière glaciation—contribution à la compréhension du fonctionnement du système paraglaciaire. Unpublished PhD Thesis, Paris 7 University, 380 p.

Evans, S.G., Clague, J.J., 1994. Recent climatic change and catastrophic geomorphic processes in mountain environments. Geomorphology 10, 107–128.

Holm, K., Bovis, M.J., Jakob, M., 2004. The landslide response of alpine basins to post-Little Ice Age glacial thinning and retreat in southwestern British Columbia. Geomorphology 57, 201–216.

Panizza, M., 1973. Glacio pressure implications in the production of landslides in the dolomitic area. Geologia Applicata e Idrogeologia 8 (1), 289–297.

Porter, S.C., Orombelli, G., 1981. Alpine rockfall hazards. American Scientist 69, 67–75.

Shreve, R.L., 1966. Sherman landslide, Alaska. Science 154, 1639–1643.

Soldati, M., Corsini, A., Pasuto, A., 2004. Landslides and climate change in the Italian Dolomites since the Late Glacial. Catena 55, 141–161.

Starkel, L., 1997. Mass movements during the Holocene: the Carpatian example and the European example. Paläoklimaforschung-Palaeoclimate Research 19, 385–400.