Réf. Frayssines & Hantz 2006 - A

Référence bibliographique complète

FRAYSSINES, M., HANTZ, D. 2006. Failure mechanisms and triggering factors in calcareous cliffs of the Subalpine Ranges (French Alps). Engineering Geology 86, 256–270. [Etude en ligne]

Abstract: In order to enhance the detection of prospective rock falls in calcareous cliffs, 25 rock falls have been described in a more detailed way than for an inventory. They are representative of middle size rock falls (10 to 100,000 m3) occurring in the French Subalpine Ranges, at an elevation between 200 m and 2000 m. Structural conditions of the rock masses, morphology of the initial cliff surface and the scar, possible failure mechanisms and processes have been studied. Typical failure configurations have been identified, based on the attitude of the failure surface, in relation to the bedding planes and the cliff surface. Irregular cliff morphology appears to be another important susceptibility factor. In most cases, the classical comparison of the average planes of the main joint sets with the average plane of the slope could not define the potentially unstable masses. Rather, those ones are due to joint planes that deviate from their mean set plane or to irregularities of the cliff surface. The proposed investigation method to detect prospective rock falls mainly consists in observing stereoscopic aerial photographs in order to look for critical configurations. Once a critical mass has been detected, its failure probability for a period of the order of one century must be evaluated (or its life expectancy). The main factor to consider for this purpose appears to be the proportion of rock bridges in the potential failure surface.

The triggering factors of rock falls in our study area have been investigated, by analysing an inventory of 46 rock falls. Statistical tests have been carried out to study the relation between rock falls and daily rainfall, freeze–thaw cycles or earthquakes. A good correlation has been obtained with freeze–thaw cycles, a slight correlation with rainfall and no correlation with earthquakes. This suggests that ice jacking could the main physical process leading to failure by causing microcrack propagation.



Rock fall - Failure mechanism - Rock bridge - Triggering factor


Organismes / Contact

• Université Joseph Fourier, Grenoble, 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

Temperature (freeze–thaw cycles)



Mass movements



Pays / Zone

Massif / Secteur

Site(s) d'étude



Période(s) d'observation


Subalpine Ranges (Vercors and Chartreuse massifs)

46 rockfall sites


200-1600 m asl



(1) - Modifications des paramètres atmosphériques




In the Chartreuse and Vercors massives, a mountain climate prevails, where temperatures and precipitations are mainly influenced by the elevation. For example, at the Autrans station (1060m) located in the Vercors massif, the mean annual temperature is 6.1°C and the mean annual rainfall is 1459 mm. During the 1984– 1985 cold season, freeze period (without thaw) represents 47% of the total period at 2500m, 11% at 1350m and 5% at 800m (Rovera,1990).






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



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










Sensibilité du milieu à des paramètres climatiques

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




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




Rock fall frequency : A rock fall inventory for the Grenoble area has been made by a forest service (RTM), which has recorded rock falls occurring before and during the 20th century, which have left physical or historical traces (Hantz et al., 2003). Using a power law distribution of the rock fall volumes, the mean failure frequencies for the investigated 150 km of cliff located in the study area have been estimated as 65, 18 and 5 events per century for the volume ranges of 102–103, 103–104, and 104–105 m3.

Triggering factors and failure processes (database 2)

Climatic factors: The annual distribution of the rock falls is analysed. It appears that rock falls are more frequent during December, January and February. Moreover, the 7 biggest rock falls occurred from January to April. Contrary to rock falls, monthly rainfalls are the most important in September, October and November. The potential influence of daily rainfall has been also studied by comparing the distribution of all the daily rainfalls in the 1970–2004 period (12,389 days), by determining which and how much daily rainfall occurred the same day as the rock falls (46 days), in the same period. The mean value of the second distribution (3.8 mm) is higher than the one of the first distribution (2.7 mm). At first sight, this suggests that daily rainfall influences rock falls. But the Kolmogorov–Smirnov test (Cheeney, 1983) shows that this difference is not significant, according to the small size of the second population (46 days with a rock fall) compared to the size of the first one (12,389 days without rock fall): The maximum discrepancy D between the two cumulative distribution functions is 0.163; this value has a probability greater than 0.05 to be reached, should both populations be identical. Thus, the hypothesis of a null influence of rainfall can not be rejected. The same conclusion is reached when considering 2-day cumulative rainfalls.

Our data don't prove the influence of rainfall on rock falls, contrary to the results of Chau et al. (2003) concerning rock falls in Hong Kong , where a good correlation was found between daily rainfall and rock falls. But there are two important differences between the two data sets: first, the rock fall volumes in the Hong Kong data are smaller than those in Grenoble (50% are less than 1 m3 in Hong Kong, they are all greater than 10 m3 in Grenoble); second, the daily rainfall in Hong Kong can exceed 300 mm, whereas it does not reach 100 mm at the Saint-Martin d'Hères station (212 m) and is probably less than 200 mm at the higher rock fall scars.

As said before, rock falls are more frequent in December, January and February. These months are the coldest ones in the year, but the mean daily maximal temperature is still positive in most of the failure sites. This means that variations of temperature around the freezing point are frequent in this period. The influence of these variations has been analysed from a contingency Table. The first column shows the number of days with a rock fall and a freeze–thaw cycle, against the number with a rock fall and without freeze and thaw. The second column shows the number of days without rock fall and with freeze and thaw, against the number without rock fall and freeze and thaw. A chi-squared test (Cheeney, 1983) has been performed to test the independence between rock falls and freeze and thaw. The obtained χ2 value, which expresses the deviation from the hypothesis of independence, is 9.85. The probability to obtain such a high value, should the factors be independent, is less than 0.01. Our data show a significant correlation between rock falls and freeze–thaw cycles. Note that a good correlation (deviation equal to 9.33) is also obtained when considering freeze (not necessarily freeze and thaw). Our conclusion is in agreement with the result of a statistical analysis of rock falls in Norway (Sandersen et al., 1996). The distribution of rock falls along the year shows two maxima, in early spring and late autumn, which coincide, in Norway , with the periods of frequent variations of temperature around the freezing point. The first maximum also coincides with the time of highest rate of snowmelt, the other with the months of highest precipitation. Our distribution shows a secondary maximum in April, which is the month when snowmelt is the most intense between 1000 and 2000 m elevations. An analysis of the rock fall activity in the Hosozawa Cirque, Japan (Matsuoka and Sakai, 1999), concluded that the intense activity does not reflect precipitation events. Nevertheless, the activity reaches its maximum in May–June, i.e. 5–15 days after the melt out of the cirque wall. A thermal conduction model suggests that a lag of 5–15 days would represent thaw penetration to a depth of about 1 m. The authors concluded that the primary factor controlling rock falls is seasonal frost weathering. Note that, contrary to our study area, Hosozawa Cirque underwent a deep frost penetration in winter.


Failure processes: The analysis of the triggering factors allows proposing some possible failure processes for the observed rock falls. The good correlation between rock falls and frost suggests that ice jacking is the main failure process for the investigated volume range. Although the limestone in our study area is not very sensitive to freezing, ice jacking probably occurs in microcracks, which have formed near the limits of rock bridges. If the area of a rock bridge is critical, then ice jacking triggers failure. Otherwise, it induces microcrack propagation and rock weakening. Matsuoka and Sakai (1999) calculated the nocturnal and seasonal frost penetration due to heat conduction in a continuous rock mass and applied their result to the Hosozawa Cirque wall, which is made of sandstone and shale. In our study area, cliffs are made of limestone, in which joints have been enlarged by dissolution, permitting heat transfer by air flow in the open joints. It ensues that frost can penetrate more rapidly and deeper than in a relatively continuous rock mass. This suggests that nocturnal frost can produce rock falls, whose thickness reaches up to 10 m. Ice jacking needs frost and water. The morphological conditions in our study area are favourable to water seepage during thawing periods, because the plateau on the top of the cliffs is sufficiently flat to make snow accumulation possible.

Although the correlation between rock falls and rainfall appears very weak, the influence of water is suggested by the occurrence of numerous rock falls in April, when snowmelt is intense. Moreover, the good correlation between rock falls and freeze–thaw cycles may show not only the influence of freeze, but also the influence of thaw. Water seeping in the rock mass, while the outlet of natural drains is blocked up by ice, can be a triggering factor. A decrease of the rock strength due to increasing water content (Serratrice and Durville, 1997 and Frayssines, 2005) can cause the failure if the stability of the rock mass is critical. In the climatic and seismic context of our study area, and during the observation period (1970–2004), freeze and water seepage appear to be more active triggering factors than earthquakes. But, if no significant short-term influence has been shown, earthquakes could weaken the rock mass and make it more susceptible to rock falls from other triggers (Keefer, 1984).






Paramètre de l'aléa

Sensibilité des paramètres de l'aléa à des paramètres climatiques

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

Triggering factors and failure processes

Freeze–thaw cycles represent the main triggering factor of the observed rock falls. This suggests that ice jacking could cause microcrack propagation leading to failure. A slight influence of rainfall has been observed and no direct influence on earthquakes.


Two databases of occurred rock falls have been elaborated for two different purposes. The first database (1) is aimed to study the intrinsic ground conditions that can favour rock falls, and the failure mechanisms of middle size (25–50,000 m3) rock falls occurring in the calcareous cliffs. The objective of the second database (2) is to identify the triggering factors of rock falls and then to better understand their failure processes. Contrary to the RTM inventory, these databases are not supposed to be exhaustive. They are supposed to be representative samples of the rock falls occurring in the Subalpine Ranges . Most of them took place in the Vercors and Chartreuse Massifs. The studied rockfalls are located between the elevations of 200 and 2000m.

Database (1) contains geometrical, structural and mechanical information concerning 25 rock falls. The rock falls have been found from either the RTM inventory (for the older ones), the media (local radios and newspapers), or from continuous observation of the cliffs in the Grenoble area. Most of them have a volume greater than 100 m3 and have been easily identified thanks to the bright scar they left (in contrast with the patina of the cliff), and the damage they produced in the forest. The scars have been surveyed, sometimes using rope, in order to collect geometrical and structural information: dimensions, intact rock fracture, discontinuities making up the failure surface (orientation, extension, roughness, cover).

Database (2) contains 46 rock falls that occurred between 1970 and 2004, whose day of occurrence is known. Their volume varies from 10 m3 to 30,000 m3. Most of them belong to the RTM inventory, the other being posterior to the inventory, and 12 belong also to database (1). They all occur in the same area than rock falls of database (1).


(4) - Remarques générales

The main objective of the study presented in this paper is to enhance the detection of potentially unstable rock masses in calcareous cliffs, thanks to a better knowledge of the intrinsic conditions and causal processes of the rock falls, according to the principle that slope failures in the future will be more likely to occur under the conditions which led to past instability (Guzzetti et al., 1999). For this purpose, a first database has been constructed, in which 25 rock falls have been described in a more detailed way than for a basic inventory. Structural conditions of the rock masses, morphology of the initial cliff and the scar, possible failure mechanisms and processes have been studied. A second database, including 46 rock falls, has been elaborated to analyse the relation between rock falls and climatic and seismic factors.


(5) - Syntèses et préconisations

The following conclusions concerning the calcareous cliffs in the Subalpine Ranges can be drawn from our observations.


·        The observation of 25 rock falls made it possible to identify typical failure configurations, which must be looked for in rock fall hazard detection.


·        In the A and B configurations, the mean cliff surface is defined by one (or two) of the main joint sets and rock falls are due to morphologic irregularities of the cliff surface or to the dispersion of the joints around their average plane. Most of the failure surfaces could be seen before the rock fall occurs.


·        The recommended investigation method for the detection of prospective rock falls consists in observing aerial photographs.


·        The rock falls have been initiated by intact rock failure in rock bridges. Thus, the persistence of the joints appears to be the main factor to consider in the evaluation of the failure probability. Mechanical back analysis of the observed rock falls will be described in a further paper.


·        Freeze–thaw cycles represent the main triggering factor of the observed rock falls. This suggests that ice jacking could cause microcrack propagation leading to failure. A slight influence of rainfall has been observed and no direct influence on earthquakes.

Références citées :

Chau, K.T., Wong, R.H.C., Liu, J., Lee, C.F., 2003. Rockfall hazard analysis for Hong Kong based on rockfall inventory. Rock Mechanics and Rock Engineering 36 (5), 383–408.

Cheeney, R.F., 1983. Statistical Methods in Geology for Field and Lab Decisions. George Allen & Unwin, London. 169 pp.

Frayssines, M., 2005. Contribution à l'évaluation de l'aléa éboulement rocheux. PhD thesis, Grenoble University.

Guzzetti, F., Carrara, A., Cardinali, M., Reichenbach, P., 1999. Landslides hazard evaluation: a review of current techniques and their application in a multi-scale study, Central Italy . Geomorphology 31, 181–216.

Hantz, D., Vengeon, J.M., Dussauge Peisser, C., 2003. An historical, geomechanical and probabilistic approach to rock fall hazard assessment. Natural Hazard and Earth System Sciences 3, 693–701.

Keefer, D.K., 1984. Landslides caused by earthquakes. Geological Society of America Bulletin 95 (4), 406–421.

Matsuoka, N., Sakai , H., 1999. Rockfall activity from an alpine cliff during thawing periods. Geomorphology 28, 309–328.

Sandersen, F., Bakkehøi, S., Lied, K., 1996. The influence of meteorological factors on the initiation of debris flows, rockfalls, rockslides and rockmass stability. In: Senneset (Ed.), Landslides. Balkema, Rotterdam , pp. 97–114.

Serratrice, J.F., Durville, J.L., 1997. Description des roches et des massifs rocheux, exploitation de deux bases de données. Bulletin des Laboratoire des Ponts et Chaussées 211, 73–87.