Navigation – Plan du site

Distribution and spatial analysis of rockslides failures in the Icelandic Westfjords: first results

Répartition spatiale des glissements de terrain dans les fjords du nord-ouest d’Islande : premiers résultats
Aurore Peras, Armelle Decaulne, Etienne Cossart, Julien Coquin et Denis Mercier
p. 25-35

Résumés

Les glissements de terrain sont nombreux dans les fjords de l’Ouest de l’Islande. Dans cette étude, les glissements de terrain visibles ont d’abord été inventoriés, par observation des images aériennes et par photo-interprétation. Au total, 186 glissements ont été répertoriés, puis leur répartition a été étudiée en utilisant les Systèmes d’Information Géographique et l’analyse statistique, notamment l’Estimation de Densité de Kernel. Le relief de la zone d’étude est largement modelé par les différentes périodes glaciaires et interglaciaires, ainsi que l’est la localisation des glissements : les mouvements de masse se concentrent principalement sur la bande littorale, où le relief est le plus marqué par le passage des glaciers qui ont contribué au façonnement de fortes pentes. Au contraire, les plateaux centraux ne présentent logiquement pas de glissements. Cependant, l’analyse statistique ne met pas en évidence un lien entre la pente et la concentration des glissements, l’hypothèse nulle n’étant pas rejetée. D’autres éléments déjà mentionnés par différents auteurs sont concordants avec la localisation des glissements, tels que le pendage favorable et la roche basaltique Tertiaire. Ces éléments ne suffisent pas à expliquer la répartition régionale des glissements de terrain. Même sans datation précise, la déglaciation peut être une piste pour expliquer cette répartition. Ce travail fournit un nouvel inventaire des glissements de terrain pour lesquels les facteurs de déclenchement habituellement identifiés ne permettent pas une explication satisfaisante de leur répartition dans les fjords du Nord-Ouest Islandais. 

Haut de page

Notes de la rédaction

Article soumis le 27 septembre 2015, accepté le 10 mars 2016.

Texte intégral

The authors thank the two anonymous reviewers for their constructive comments that help improving the quality of the manuscript.

1. Introduction

1Rockslides are common features in the Icelandic Westfjords, but they have not been widely studied at a local nor at a regional scale. Jónsson (1957, 1976) offered a first record at the scale of Iceland, showing a large number of rockslide bodies all over the country. Then, Whalley et al. (1983) reworked the first census, and tried to explain the location of clusters of rockslides, and to document a possible paraglacial hypothesis of settlement (i.e. high frequency and magnitude of rockslides during postglacial times). In the present study we list the large rockslides visible in the Westfjords, creating a new database: many of the rockslides we count are not specified in both works cited above; Jónsson (1957, 1976) doesn’t propose a map of the rockslides, but add a precise location of the landforms, and in some instance proposes a photograph or a sketch of the major ones, identifying 73 rockslides in the Westfjords (nr. 135 to 207 in Jónsson, 1976); Whalley et al. (1983) based their work on Jónsson (1976) inventory, showing about 95 rockslides in the Westfjords (some of those we have not observed; the location map is presented at the scale of Iceland, and a precise account of rockslides is difficult on the document; although the authors state that the rockslide collection is derived from Jónsson 1976, they do not explain the difference in number in between the two collections); the classification they used was based on the size; they also proposed a possible correlation between the size and the age of each rockslides, without any dating evidence (no radiocarbon or tephra dating). In our research, only rockslides visible from photo-interpretation have been recorded, i.e. some of the previous databases were disregarded in the new one. We attempt to understand the spatial distribution of the rockslides with the help of spatial analysis tools. Rockslides are here used as a generic term for falls, toppling, slides and rock avalanches, excluding snow avalanches and debris flows, following the classification proposed by Selby (1993).

2In the literature, several explanations regarding the location and triggering of rockslides have been proposed, suggesting that glacial valleys are particularly prone to landsliding (e.g. Thorarinsson et al., 1959; Rapp, 1960; Caine, 1982; Cruden and Eaton, 1987; Ballantyne, 2002; Jarman, 2006; McColl, 2012). Leith et al. (2010) proposed that different mechanisms cause landsliding according to their position on the slope, due to bottom-up erosional processes related to post-glacial debuttressing. Cossart et al. (2008) also suggested that the lower part of slopes of South French Alps is more sensitive to landsliding due to the debuttressing processes. From New-Zealand examples, Allen et al. (2011) explain that the occurrence of rockslides is either related to glacier retreat or to permafrost degradation. For other authors, the topography of slope is the sole cause for rockslides, favoured by a change in the profile of the slope, especially a break in a convex slope (e.g. Holm et al., 2004; Dadson and Church, 2005).

3In Iceland, most of rockslides have occurred in Tertiary basaltic terrains (Whalley et al. 1983; Mercier et al., 2013; Feuillet et al., 2014). From case studies in the Skagafjörður area, in northern Iceland, Cossart et al. (2014) demonstrated that paraglacial causes explain the distribution of rockslides rather than topographic or geologic parameters, with a trigger more particularly in relation to the glacio-isostatic rebound. 

2. Study area

4The study area is the whole Westfjords peninsula, in northwestern Iceland (fig. 1); it ignores the administrative boundary of the district in the south, as the southern margin of the study area runs from Búðardalur in the SW to Borðeyri in the SE. The western coast is fringed by the Denmark Strait, while the northern and eastern coasts are fringed by the Greenland Sea. The area is situated between 65°06-66°27’N and 21°16’-24°30’W.

5Successive glaciers and following deglaciation processes during the Quaternary have heavily shaped the study area. During the Weichselian, an ice sheet was entirely covering the study area (Ingólfsson, 1991; Ingólfsson and Norðdahl, 2001; Principato et al., 2006; Principato, 2008; Geirsdóttir et al, 2009). Severe ice thinning has occurred by 26.2 ka (Brynjólfsson et al., 2015b), and break-up of the shelf based ice sheet off the Westfjords peninsula was recorded about 15 ka, leading to rapid deglaciation of the whole area through the Bølling-Allerød interstadial (Syvitski et al., 1999; Andrews et al., 2000; Eiriksson et al., 2000; Jennings et al., 2000; Geirsdóttir et al., 2002; Andrews and Helgadóttir, 2003; Andrews, 2007; Geirsdóttir et al., 2007). Marine studies highlight a two-step glacier retreat after the Younger Dryas glacial re-advance, between 12.3-11.9 ka (Ólafsdóttir, 2004) and around 10.3 ka (Geirsdóttir et al, 2002).

Fig. 1Location map of the study area (A), typical landscape characteristics (B) and some views of rockslides in Svínadalur, Skálmarfjörður and Vatnadalur (C).
Fig. 1Carte de localisation du terrain étudié (A), paysages caractéristiques (B) et quelques glissements de terrain dans Svínadalur, Skálmarfjörður et Vatnadalur (C).

Fig. 1 – Location map of the study area (A), typical landscape characteristics (B) and some views of rockslides in Svínadalur, Skálmarfjörður and Vatnadalur (C).Fig. 1 – Carte de localisation du terrain étudié (A), paysages caractéristiques (B) et quelques glissements de terrain dans Svínadalur, Skálmarfjörður et Vatnadalur (C).

Photographs by Armelle Decaulne. Background map from SAR data source
1. Fjord Ísafjarðardjúp towards W; 2. The Gláma plateau towards N; 3. Fjord Arnarfjörður towards W; 4. One of the rockslides of the Svínadalur valley; 5. Rockslide in the fjord Skálmarfjörður; 6. Rockslides in Vatnadalur.
Photographies d’Armelle Decaulne. Image de fond de SAR data source.
1. Fjord Ísafjarðardjúp, vue vers l'W ; 2. Le plateau de Gláma, vue vers le N ; 3. Fjord Arnarfjörður, vue vers l'W ; 4. Un des mouvements de terrain dans la vallée de Svínadalur valley ; 5. Mouvement de terrain dans le fjord de Skálmarfjörður ; 6. Mouvements de terrain dans le Vatnadalur.

6Today Drangajökull is the only ice cap present in the Westfjords. Therefore, fjords and highland plateaux are the major features in the Westfjords; while the coastline is highly indented with several inland glacial valleys featuring past glacial outlets. The highest peak culminate at 998 m a.s.l. (Kaldbakur), west of the study area, but most plateaux have rolling surfaces ranging from 600 to 800 m a.s.l. Slopes along fjords show generally 300-600 m denivellation.

7The bedrock of most of the area corresponds to 3-17 millions years old sub-aerial tholeitic and porphyritic basalts, and some outcrops of olivine basalts. The Miocene and Pliocene flood basalts are interbedded with thin sedimentary layers and in some places with volcanoclastic sedimentary horizons. At very local places, acid intrusions exist; according to the geological map, the overall orientation of the regional dip is southeastwards; east of Drangajökull glacier the dip is eastwards, and southeast of the Westfjords peninsula the dip is southwards (Sæmundsson, 1979; Einarsson, 1991; Kristjánsson and Jóhannesson, 1994; Guðmundsson et al., 1996; Harðarson et al., 1997).

3. Methods

8This work has been carried out by compiling first a rockslide inventory, then analysing its spatial patterns. In a first stage, rockslides have been inventoried in two ways. A prior field prospection was permitted to admit that the Westfjords are a prosperous area for rockslides. The second way to inventory, providing much more complete outcomes, has been the observation of aerial images of the whole area and subsequent photo-interpretation. Four sources have been used (aerial and satellite pictures): GoogleEarth, allowing 3D views in addition to aerial views, and different online satellite image navigators available on different Icelandic websites, depending from the National Planning Agency (Skipulag rikisins - www.skipulagsstofnun.is, with pictures from Loftmyndir ehf.) or online map databases (national directory www.ja.is and map resources www.map.is, also with pictures from Loftmyndir ehf.). Compiling these different sources, at different scales, some issues regarding information availability due to dropping shadow, snow cover or cloud cover, have been avoided. Thanks to this inventory, a new database has been created, listing geographical coordinates and several morphometric and lithographic features. Departure zone and deposit elevations have been measured. The elevation of the departure zone of the rockslide corresponds to the highest point of the scarp and the lowest elevation corresponds to the furthest reach of the deposit. The runout has been measured in the main axis of each rockslide. Geographic Information Systems (GIS) SAGA GIS® and ArcGIS® are used to map all recognized rockslides at the regional scale. It represents the spatial distribution of rockslides, with the first geo-visualization exhibiting a number of recognized rockslide features that overpasses the previously known records.

9In a second stage, the study aims at identifying the spatial pattern of the rockslide distribution in the Icelandic Westfjords. Many geomorphologists have tried to explain the location of mass-movements from large inventories (Brabb, 1991; Dikau et al., 1996; Guzzetti et al., 2012). Nevertheless, such spatial approach remains difficult as inventories are not necessarily exhaustive and the occurrence of a rockslide also depends on the characteristics of the area located around the rockslide itself. Many geographers cope with these problems, and apply a smoothing method to identify specific spatial patterns (Fotheringham et al., 2000; Joyner and Rohli, 2010). A smoothing method calculates the density of a phenomenon in each place of an area, by averaging the number of patterns located within a specific bandwidth around the area. In this calculation, the closer events are characterized by a more important weight. Both the bandwidth and the rate of decrease of pattern weight following the distance are to be calibrated. Here a Kernel Density Estimator (KDE) is applied, it allows rough spatial delineation: discriminating areas where the studied patterns are particularly frequent or, to the opposite, where the frequency is close to zero. A specific bandwidth of 10 km is chosen. It corresponds to the mean distance between rockslides, plus standard-deviation of the distance to the closer neighbour. The weight decrease corresponds to a Gaussian law. Calculations are made through SagaGIS software from the rockslide inventory.

10The results of a KDE are interpreted as qualitative data (classes of occurrence probability) and here, we consider a binary result (occurrence vs. lack of rockslides) that must be explained. The predictors are both quantitative (topographical parameters: slope gradient, relief s.s.) and qualitative (elapsed time since the deglaciation). To reject (or not) the null hypothesis of an independence between occurrence of rockslides and topographical parameters, the Fischer test F-test is applied, while a chi² test is applied to reject (or not) the null hypothesis between rockslide location and deglaciation timing. The calculations are mostly implemented through GIS software SAGA GIS®. The results of KDE are combined with a 50 x 50 digital elevation model (DEM) data and its derivative (slope gradients, valley depth) to calculate the main statistics of areas affected and non-affected by rockslides: mean value, variance, standard deviation and range are calculated, and then integrated within the F-test procedure. The results of the KDE are combined to deglaciation data through an intersection procedure: KDE zones are cut by the polygons corresponding to former glacial extents. The number of cells within each intersected zone is finally calculated and compared with the number of cells expected from a theoretical chi² distribution.

4. Results and discussion

4.1. Description of the new rockslide database

11In this study 186 rockslides have been identified (fig. 2). Their geographical distribution clearly shows that slopes close to the coast are preferred even though some of them are free of slides, with very few rockslides found inland, where relief in the high plateau area is lacking to release rockslides. Some clusters also appear, other areas being clearly void. This complex spatial pattern requires to discuss which parameters can discriminate areas where rockslides are under- or overrepresented. 

12Mass movements are mostly concentrated in the outer northwesternmost areas (between Ísafjarðardjúp and Arnarfjörður fjords), part of the northeastern area (Strandir) and south area (within the narrow isthmus), and the fjords north of Breiðafjörður. The southwesternmost area counts a very small amount of rockslide features. In all the north area, rockslides are scattered, and found around the Jökulfirðir fjords and in Hornstranðir. The inner part of Ísafjarðardjúp fjord is bereft of rockslides.

Fig. 2Distribution of post-glacial rockslides, northwest Iceland.
Fig. 2Localisation des glissements de terrain post-glaciaires en Islande du Nord-Ouest.

Fig. 2 – Distribution of post-glacial rockslides, northwest Iceland.Fig. 2 – Localisation des glissements de terrain post-glaciaires en Islande du Nord-Ouest.

1: rockslides (from our own inventory).
1 : glissement de terrain (d’après notre propre inventaire).

4.2. Explanations for rockslide distribution

4.2.1. Geological parameters

13In Iceland, most of rock-slope failures are located in the Tertiary basalts (from the geological maps by Jóhannesson and Sæmundsson, 1989); this observation has already been made by Whalley at al. (1983) from previous observations by Jónsson (1957 and 1976) at the scale of the island (fig 3A) and by Mercier et al. (2013) and Feuillet et al. (2014) at the Skagajörður scale. More precisely, the quite homogeneous lithology appears however insufficient to explain why rock-slope failures are numerous in Hornstrandir and close to Flateyri, while Patreksfjörður area is non-affected by failures (fig. 3B). Dip (from the tectonic map by Jóhannesson and Sæmundsson, 1998) is also a favourable parameter to explain rockslide location (fig. 3C). But only 62 rock-slope failures are compliant with the local dip (which means 33% of all the rock-slope failures inventoried). 42% of rock-slope failures are perpendicular with the local dip, and 25% are non-compliant. Collectively, these regional geological data cannot help in discriminating over- and underrepresentation of rock-slope failures.

Fig. 3Rockslides and the geologic setting highlighting their occurrence within the Tertiary basalts (A – modified from Whalley et al., 1983), and at lithological contacts (B – modified from the geological maps from Jóhannesson and Sæmundsson, 1989) and local dip compliance (C – modified from the geological maps from Jóhannesson and Sæmundsson, 1998).
Fig. 3Glissements de terrain et contexte géologique soulignant leur occurrence dans les basaltes tertiaires (A – modifié de Whalley et al., 1983), et aux contacts lithologiques (B – modifié de la carte géologique de Jóhannesson and Sæmundsson, 1989) et leur conformité au pendage (C – modifié de la carte géologique de Jóhannesson and Sæmundsson, 1998).

Fig. 3 – Rockslides and the geologic setting highlighting their occurrence within the Tertiary basalts (A – modified from Whalley et al., 1983), and at lithological contacts (B – modified from the geological maps from Jóhannesson and Sæmundsson, 1989) and local dip compliance (C – modified from the geological maps from Jóhannesson and Sæmundsson, 1998).Fig. 3 – Glissements de terrain et contexte géologique soulignant leur occurrence dans les basaltes tertiaires (A – modifié de Whalley et al., 1983), et aux contacts lithologiques (B – modifié de la carte géologique de Jóhannesson and Sæmundsson, 1989) et leur conformité au pendage (C – modifié de la carte géologique de Jóhannesson and Sæmundsson, 1998).

1: rockslide (from Whalley et al., 1983 in A, and our inventory in B and C); 2: dip (from the geological map from Jóhannesson and Sæmundsson, 1998); 3: active zone; 4: Quaternary formations; 5: Tertiary formations; 6: basic and intermediate extrusive rocks with intercalated sediments (Lower to Upper Miocene); 7: acid extrusives (middle Miocene); 8: basic and intermediate intrusions; 9: acid intrusions; 10: extinct central volcanoes; 11: major sedimentary horizon with lignite; 12: compliant dip rockslides; 13: perpendicular dip rockslides; 14: non-compliant dip rockslides.
1 : glissement de terrain (d’après Whalley et al., 1983, en A ; notre inventaire en B et C) ; 2 : pendage (d’après la carte géologique) ; 3 : zone active ; 4 : formations quaternaires ; 5 : formations tertiaires ; 6 : roches extrusives basiques et intermédiaires avec sédiments intercalés (Miocène inférieur et supérieur) ; 7 : extrusions acides (Miocène moyen) ; 8 : intrusions basiques et intermédiaires ; 9 : intrusions acides ; 10 : volcans centraux éteints ; 11 : horizons sédimentaires à lignite majeurs ; 12 : glissements de terrain conformes au pendage ; 13 : glissements de terrain perpendiculaires au pendage ; 14 : glissements de terrain non conformes au pendage.

14These results nevertheless cope with the spatial scale at which the locational analysis was led. At such a regional scale, local variations in mechanical properties (shattering, even faulting, dip variability) are necessarily smoothed. The influence of such classical preconditionning factors cannot be exhibited here without an extensive fieldwork. A regional examination highlights that some rockslides are present either at locations where central volcanoes have favored lithological contacts between basic and acidic bedrocks (e.g. Strandir and Hornstrandir) or along the weak major sedimentary horizons (e.g. between Ísafjarðardjúp and Arnarfjörður). For instance, in Arnarfjörður 7 rockslides are located along the contact zone and next to Reykhólar 8 rockslides are observed in the same settings. Contacts between basalts and intrusive rocks generally correspond to weaken materials affected: schistosity planes, shattering patterns are superimposed to former structural joints. A buffer area around the contact delineation may thus be prone to water infiltration, rock weathering and weakening. Regarding the presence of lignite rich sedimentary horizon, widespread in the Westfjords, some clusters of rockslides seem to be strongly associated with it, like in the NW tip of the Westfjords; others parts of this lignite horizon are totally bereft of rockslides; so this geological factor can not be the only trigger. This hypothesis should be taken into account in further fieldwork strategy, but a systematic correlation cannot be exhibited at a regional scale: several rockslides have occurred outside these more fragile areas (fig. 3B).

4.2.2. Deglaciation history

15As they are visible today, rockslides that are accounted in this database occurred after the last glacial maximum (LGM) and have not encountered any subsequent glaciation that would have eroded and heavily reworked the deposits. It is therefore possible to derive the maximum potential age of these rockslides by examining the deglaciation history of the area. The deglaciation of the Westfjords has been inferred mainly from sediment cores and ice rafted debris records carried out on the northern, northwestern and western Iceland shelves (Andrews et al., 2000; Andrews et al., 2002; Geirsdóttir et al., 2002; Geirsdóttir, 2009) and from few terrestrial dating obtained from shells, sea level changes and lacustrine records in northern and western Iceland (Rundgren et al., 1997; Geirsdóttir et al., 2013), offering only very few local dating and patterns of ice disappearance in the Westfjords. Recently, cosmogenic exposure results from moraines, erratics and roches moutonnées features from the northernmost part of the Westfjords, in the vicinity of glacier Drangajökull (Principato et al., 2006; Brynjólfsson et al., 2015b), offer local new insights in the deglaciation history of the region. The now accepted scheme is that the whole Westfjords were totally covered with ice during the LGM; a progressive deglaciation of the highlands and plateaux started during the late Weichselian, around 26 ka, during which cold-based glaciers melted, leading to a collapse of the shelf based ice-sheet by 14 ka, then a deglaciation of some valleys with the shrinking of the warm-based valley glaciers; most of coastal areas being then free of ice during the Bølling-Allerød interglacial. Rockslides closest to the coast might have occurred at this time, or more precisely occurred from this time as most coastal areas have not been englaciated during the Younger Dryas glacier re-advance (fig. 4).

Fig. 4 – Potential age maximum for rockslides based on Late Weichselian ice extent reconstruction (from Geirsdóttir et al., 2009).
Fig. 4 – Age maximum potentiel pour les mouvements de masse sur la base de la reconstitution de l’extension glaciaire au cours du Weichselien Tardif (d’après Geirsdóttir et al., 2009).

Fig. 4 – Potential age maximum for rockslides based on Late Weichselian ice extent reconstruction (from Geirsdóttir et al., 2009).Fig. 4 – Age maximum potentiel pour les mouvements de masse sur la base de la reconstitution de l’extension glaciaire au cours du Weichselien Tardif (d’après Geirsdóttir et al., 2009).

A: Maximal extent of glacier readvance during the Younger Dryas cold event; B: Maximal extent of glaciers during the Preboreal period; C: Maximal extent of glaciers during the Holocene; 1: glacial extent; 2: rockslides which can’t have potentially occurred before the Bølling-Allerød interstadial and the Younger Dryas; 3: rockslides which can’t have potentially occurred before the glacial retreat following the maximal ice extent of the Younger Dryas; 4: rockslides which can’t have potentially occurred before the glacier retreat following the maximal ice extent of the Preboreal.
A : Extension maximum de la réavancée glaciaire au cours de l’épisode froid du Dryas Récent ; B : Extension maximum des glaciers au cours de l’épisode froid du Préboréal ; C : Extension maximum des glaciers au cours de l’Holocène. 1 : extension glaciaire ; 2 : les glissements de terrain qui ne peuvent être potentiellement survenus avant l’interstade Bølling-Allerød et le Dryas récent ; 3 : les glissements de terrain qui ne peuvent être potentiellement survenus avant le retrait glaciaire qui a suivi l’extension maximale de la glace au Dryas récent ; 4 : les glissements de terrain qui ne peuvent être potentiellement survenus avant le retrait des glaciers qui a suivi l’extension maximale de la glace au Préboréal.

16During the Preboreal period, glaciers occupied only the highlands (fig. 4B), and their shrinking towards almost total disappearance during the Holocene Thermal Maximum leaves place for the other rockslides to occur. Figure 4C represents the largest englaciated areas, Drangajökull ice cap being larger than today, with outlet glaciers flowing down the closest valleys (Brynjólfsson et al., 2015a) while the Gláma highlands had only presented firns during the whole historical time of Iceland (post 871 AD; Sigurðsson, 2004). With such a deglaciation scheme, 137 rockslides on a total of 186 (73%) might have been triggered as early as the Bølling-Allerød – Younger Dryas transition, between 14 to 11.7 ka (tab. 1). If such results are supported by precise dating of some of the rockslide bodies, the paraglacial crisis that may led to such numerous rock-slope failures occurred earlier than recent findings revealed in northern Iceland (Mercier et al., 2013; Decaulne et al., 2016). Our results are in contradiction with Whalley et al. (1983) study. They considered, without any precise date, that all rockslides in Iceland are younger than 10 ka.

Tab. 1Potential maximal age periods of occurrence of recorded rockslides.
Tab. 1 – Age maximum potentiel des séquences d’apparition des glissements de terrain inventoriés.

Tab. 1 – Potential maximal age periods of occurrence of recorded rockslides.Tab. 1 – Age maximum potentiel des séquences d’apparition des glissements de terrain inventoriés.

4.2.3. Regional patterns

17Results of the KDE procedure highlight a concentration of rockslides at the margins of the peninsula (excepting the south-western part), and also along the isthmus. The visual observation of the rockslide locations is therefore in perfect agreement. Nevertheless, while those maps (fig. 2 and fig. 5) are useful to identify this spatial pattern, the interpretation is complicated by the rugged topography and the delineation of relief and coastlines. A statistical examination is thus required.

Fig. 5Density of rockslides, estimated by a Kernel Density Estimator (Background DEM from SAR data source).
Fig. 5Densité de glissements de terrain, estimée à partir d’un lissage de Kernel (MNT de fond de SAR data source).

Fig. 5 – Density of rockslides, estimated by a Kernel Density Estimator (Background DEM from SAR data source).Fig. 5 – Densité de glissements de terrain, estimée à partir d’un lissage de Kernel (MNT de fond de SAR data source).

18On the one hand, the areas affected and non-affected by rockslides are not significantly different regarding the topography: the null hypothesis is not rejected according to the F-test both for the slope-gradient and the valley-depth (fig. 6). This result is not in accordance with classical models, highlighting the preponderance of slope gradient to generate slope instability. Nevertheless, the physical setting is quite particular, as slope gradient is high everywhere within the peninsula fjords. Thus, the slope gradient can be here seen as a factor prone to landsliding everywhere but is not a parameter that can discriminate areas affected by rockslides from areas non-affected by rockslides.

19On the other hand, the location of areas affected by rockslides is significantly associated with the spatial pattern of deglaciation. More precisely, rockslides are significantly over-represented in the area deglaciated between the LGM and the Younger Dryas. Per contra, they are clearly under-represented in more recently deglaciated areas (fig. 6: the fjords between Ísafjarðardjúp and Arnarfjörður gather two distinct rockslide clusters; the southern isthmus also gather two clusters; the northeastern part in Strandir also represents one cluster; other secondary clusters are scattered in the south and north fjords. While it suggests that rockslides probably occurred between the LGM and the Younger Dryas, we consider cautiously this chronological hypothesis. We should also consider the deformations imposed to the lithosphere by the deglaciation. In Iceland the lithosphere is hot and thin, particularly prone to high-rate glacio-isostatic rebound (Biessy et al., 2008; Le Breton et al., 2010).

Fig. 6Main statistical characteristics discriminating (or not) areas affected by rockslides and areas non-affected by rockslides.
Fig. 6Principales caractéristiques statistiques visant à discriminer (ou pas) les zones affectées par des mouvements de masse de celles non-affectées par des mouvements de masse.

Fig. 6 – Main statistical characteristics discriminating (or not) areas affected by rockslides and areas non-affected by rockslides.Fig. 6 – Principales caractéristiques statistiques visant à discriminer (ou pas) les zones affectées par des mouvements de masse de celles non-affectées par des mouvements de masse.

A: F-test comparing topographical parameters (slope gradient, valley-depth) in stable and unstable areas, null hypothesis is not rejected. B: chi²-test showing a significant overrepresentation of rockslides in early-deglaciated areas.
A : Test de Fisher comparant des indicateurs topographiques (pente, profondeur des vallées) entre les zones stables et instables. B : Test du chi² montrant la recrudescence significative dans les zones précocement déglacées.

20Series of raised beaches have indeed been identified at different locations in the Westfjords: from 5 to 48 m a.s.l. around the Drangajökull ice cap (John and Sugden, 1962; Principato, 2008; Brynjólfsson et al., 2014); in the southern part of the Westfjords, raised beaches have been modelled up to 100-225 m a.s.l (Le Breton et al., 2010). The rebound is particularly significant on the margin of the decaying ice-sheet that also corresponds to the areas affected by rockslides (Stewart et al., 2000). It has furthermore been demonstrated that post-glacial rebound can significantly predispose the occurrence of landsliding in Iceland (Cossart et al., 2014; Feuillet et al., 2014), but also in Scotland (Ballantyne et al., 2014). Even though, a clear explanation from spatial analysis is missing, as none of the selected parameters are able to highlight the potential main causes for rockslides in the Westfjords. 

5. Conclusions

21From field survey and aerial photographs analysis, 186 rockslides have been identified in the Westfjords, in NW Iceland. This record completes substantially the existing database.

22The spatial distribution of rockslides shows a clear preference for slopes close to the coastal area, where slope gradient and valley depth are suitable for landsliding. However, these classical topographical parameters are unable to discriminate within these prone areas. No statistical difference can be exhibited between hillslopes affected by rockslides and hillslopes non-affected by rockslides: they are very similar in terms of topography, as slopes affected or non-affected by rockslides present the very same characteristics. Most of the rockslides are confined within the area deglaciated during the LGM to Younger Dryas transition (14 to 11.7 ka): the overrepresentation of rockslides in this area is highly significant. This result does not mean that all rockslides have been triggered during this period, it does not provide any dating pieces of evidence. 

23These conclusions correspond to first pattern, drawn at a regional scale, at which many parameters are smoothed: local variability regarding geology (bedrock contacts, periphery of old central volcanoes, presence of a weak sedimentary layer), topography, are here not examined at a fine scale. Thus, these results ask for further fieldwork in these areas to discuss the possible mechanisms of hillslope deformations and rockslides. Such fieldwork might highlight the role of paraglacial conditions as a main cause for rockslide triggering, associated to slope debuttressing and isostatic rebound that precondition rockslope to failure (Mercier, 2011; Coquin et al., 2015), and help to discriminate between preconditioning and preparatory factors for rockslide occurrence (McColl, 2012).

24Further dating are more particularly necessary on individual rockslides, as well as an accurate morphometric analysis of the rockslides and its specific geologic setting to get a better picture of rockslide conditions in the area and its association with the deglaciation patterns. 

Haut de page

Bibliographie

Allen S., Cox S., Owens I. (2011) – Rock avalanches and other landslides in the central Southern Alps of New Zealand: a regional study considering possible climate change impacts. Landslides, 8, 33-48.
DOI : 10.1007/s10346-010-0222-z

Andrews J.T. (2007) – Holocene denudation of the northwest sector of Iceland as determined from accumulation of sediments on the continental margin. Boreas, 36, 240-252.
DOI : 10.1111/j.1502-3885.2007.tb01248.x

Andrews J.T., Harðardóttir J., Helgadóttir G., Jennings A.E., Geirsdóttir Á., Sveinbjörnsdóttir Á.E., Schoolfield S., Kristjánsdóttir G.B., Smith L.M., Thors K., Syvitski J. (2000) – The N and W Iceland Shelf: insights into Last Glacial Maximum ice extent and deglaciation based on acoustic stratigraphy and basal radiocarbon AMS dates. Quaternary Science Reviews, 19, 619-631.
DOI : 10.1016/S0277-3791(99)00036-0

Andrews J.T., Harðardóttir J., Geirsdóttir Á., Helgadóttir G. (2002) – Late Quaternary ice extent and glacial history from the Djúpáll trough, off Vestfirðir peninsula, north-west Iceland: a stacked 36 cal. Ky environmental record. Polar Research, 21, 211-226.
DOI : 10.1111/j.1751-8369.2002.tb00074.x

Andrews J.T., Helgadóttir G. (2003) – Late Quaternary ice extent and deglaciation of Hunafloaall, north Iceland: evidence from marine cores. Arctic, Antarctic and Alpine Research, 35, 218,232.
DOI : http://dx.doi.org/10.1657/1523-0430(2003)035[0218:LQICEA]2.0.CO;2

Ballantyne C.K. (2002) – Paraglacial geomorphology, Quaternary Science Reviews, 21 (18-19), 1935-2017.
DOI : 10.1016/S0277-3791(02)00005-7

Ballantyne C.K., Wilson P., Gheorghiu D., Rodés G. (2014) – Enhanced rock-slope failure following ice-sheet deglaciation: timing and causes. Earth Surface Processes and Landforms 39 (7), 900-913.
DOI : 10.1002/esp.3495

Biessy G., Dauteuil O., Van Vliet-Lanoë B., Wayolle A. (2008) – Fast and partitioned postglacial rebound of southwestern Iceland. Tectonics, 27, TC3002, 18 p.
DOI : 10.1029/2007TC002177

Brabb E.E. (1991) – The world landslide problem. Episodes, 14 (1), 52-61.

Brynjólfsson S., Schomacker A., Guðmundsdóttir E.R., Ingólfsson Ó. (2014) – Geomorphology and the Little Ice Age extent of the Drangajökull ice cap, NW Iceland, with focus on its three surge-type glaciers. Geomorphology, 213, 292-304.
DOI : 10.1016/j.geomorph.2014.01.019

Brynjólfsson S., Schomacker A., Ingólfsson Ó. (2015a) – A 300-year surge history of the Drangajökull ice cap, northwest Iceland, and its maximum during the ‘Little Ice Age’. The Holocene, 27, 1076-1092.
DOI : 10.1177/0959683615576232

Brynjólfsson S., Schomacker A., Ingólfsson Ó., Keiding J.K. (2015b) – Cosmogenic 36Cl exposure ages reveal a 9.3 ka BP glacier advance and the Late Weichselian-Early Holocene glacial history of the Drangajökull region, northwest Iceland. Quaternary Science Reviews, 126, 140-157.
DOI : 10.1016/j.quascirev.2015.09.001

Caine N. (1982) – Toppling failures from alpine cliffs on Ben Lomond, TasmaniaEarth Surface Processes and Landforms, 7, 133-152.
DOI : 10.1002/esp.3290070207

Coquin J., Mercier D., Bourgeois O., Cossart E., Decaulne A. (2015) – Gravitational spreading of mountain ridges coeval with Late Weichselian deglaciation: impact on glacial landscapes in Tröllaskagi, northern Iceland, Quaternary Science Reviews, 107 (1), 97-213.
DOI : 10.1016/j.quascirev.2014.10.023

Cossart E., Braucher R., Fort M., Bourlés D.L., Carcaillet J. (2008) – Slope instability in relation to glacial debuttressing in alpine areas (Upper Durance catchment, southeastern France): evidence from field data and 10Be cosmic ray exposure ages. Geomorphology, 95, 3-26.
DOI : 10.1016/j.geomorph.2006.12.022

Cossart E., Mercier D., Decaulne A., Feuillet T., Jónsson H.P., Sæmundsson Þ. (2014) – Impacts of post-glacial rebound on landslide spatial distribution at a regional scale in northern Iceland (Skagafjörður). Earth surface processes and landforms, 39 (3), 336-350.
DOI : 10.1002/esp.3450

Cruden D.M., Eaton T.M. (1987) – Reconnaissance of rockslide hazards in Kananaskis Country. Alberta Canadian Geotechnical Journal, 24, 414-429.
DOI : 10.1139/t87-052

Dadson S.J., Church M. (2005) – Postglacial topographic evolution of glaciated valleys: a stochastic landscape evolution model. Earth Surface Processes and Landforms, 30, 1387-1403.
DOI : 10.1002/esp.1199

Decaulne A., Cossart E., Mercier D., Coquin J., Feuillet T., Jónsson H.P., (2016) – Early Holocene dating of the Vatn landside (Skagafjörður, central north Iceland) and Holocene slope development, The Holocene (in press).

Dikau R., Brunsden D., Schrott L, Ibsen M.L. (1996) – Landslide recognition, Wiley, Chichester, 274 p.

Einarsson Þ. (1991) – Geology of Iceland. Mál og Menning, Reykjavík, 309 p.

Eiriksson J., Knudsen K.L., Haflidason H., Henriksen, P. (2000) – Late-glacial and Holocene palaeoceanography of the North Icelandic shelf. Journal of Quaternary Science, 15, 23-42.
DOI : 10.1002/(SICI)1099-1417(200001)15:1<23::AID-JQS476>3.0.CO;2-8

Feuillet T., Coquin J., Mercier D., Cossart E., Decaulne A., Jónsson H.P., Sæmundsson Þ. (2014) – Focusing on the spatial non-stationarity of landslide predisposing factors in northern Iceland: Do paraglacial factors vary over space?, Progress in Physical Geography, 38 (3), 354-377.
DOI : 10.1177/0309133314528944

Fotheringham S.A., Brunsdon C., Charlton M. (2000) – Quantitative Geography: Perspectives on Spatial Data Analysis. Sage, London, 269 p.

Geirsdóttir Á., Andrews J.T., Ólafsdóttir S., Helgadóttir G., Harðardóttir J. (2002) – A 36 Ky record of iceberg rafting and sedimentation from north-west Iceland. Polar Research, 21, 291-298.
DOI : 10.1111/j.1751-8369.2002.tb00083.x

Geirsdóttir A., Miller G.H., Andrews J.T. (2007) – Glaciation, erosion, and landscape evolution of Iceland. Journal of Geodynamics, 43, 170-186.
DOI : 10.1016/j.jog.2006.09.017

Geirsdóttir Á., Miller G.H., Axford Y., Ólafsdóttir S. (2009) – Holocene and latest Pleistocene climate and glacier fluctuations in Iceland, Quaternary Science Reviews, 28, 2107-2118.
DOI : 10.1016/j.quascirev.2009.03.013

Geirsdóttir Á., Miller G.H., Larsen D.J., Ólafsdóttir S. (2013) – Abrupt Holocene climate transitions in the northern North Atlantic region recorded by synchronized lacustrine records in Iceland, Quaternary Science Reviews, 70, 48-62.
DOI : 10.1016/j.quascirev.2013.03.010

Guðmundsson A., Bergerat F., Angelier J. (1996) – Off-rift and rift-zone palaeostresses in Northwest Iceland. Tectonophysics, 255 (3-4), 211-228.
DOI : 10.1016/0040-1951(95)00138-7

Guzzetti F., Mondini A.C., Cardinali M., Fiorucci F., Santangelo M., Chang K.T. (2012) – Landslide inventory maps: new tools for and old problem. Earth-Science Reviews, 112, 42-66.
DOI :
10.1016/j.earscirev.2012.02.001

Harðarson B.S., Fitton J.G., Ellam R.M., Pringle M.S. (1997) – Rift relocation: a geochemical and geochronological investigation of a paleo-rift in Northwest Iceland. Earth and Planetary Science Letters, 153, 181-196.
DOI : 10.1016/S0012-821X(97)00145-3

Holm K., Bovis M., Jakob J. (2004) – The landslide response of alpine basins to post-Little Ice Age glacial thinning and retreat. Geomorphology, 57, 201-216.
DOI : 10.1016/S0169-555X(03)00103-X

Ingólfsson Ó. (1991) – A review of Late Weichselian and Early Holocene glacial and environmental history in Iceland. In Caseldine C., Russel A., Harðardóttir J and Knudsen O. (Eds.): Iceland – Modern Processes and Past Environments. Elsevier, Amsterdam, 13-29.

Ingólfsson Ó., Norðdahl H. (2001) – High relative sea level during the Bolling interstadial in western Iceland: a reflection of ice sheet collapse and extremely rapid glacial unloading. Arctic, Antarctic and Alpine Research, 33, 231-243.
DOI : 10.2307/1552224

Jarman D. (2006) – Large rock slope failures in the Highlands of Scotland: characterisation, causes and spatial distribution. Engineering Geology, 83, 161-182.
DOI : 10.1016/j.enggeo.2005.06.030

Jennings A., Syvitski J., Gerson L., Grönvold K., Geirsdóttir Á., Harðardóttir J., Andrews J.T., Hagen S. (2000) – Chronology and paleoenvironments during the late Weichselian deglaciation of the South-West Iceland shelf. Boreas, 29, 167-183.
DOI : 10.1111/j.1502-3885.2000.tb00976.x

Jóhannesson H., Sæmundsson K. (1989) – Geological map of Iceland, 1:500000, Bedrock geology. Icelandic Institute of Natural History, Reykjavík.

Jóhannesson H., Sæmundsson K. (1998) – Geological map of Iceland, 1:500000, Tectonics. Icelandic Institute of Natural History, Reykjavík.

John B.S., Sugden D.E. (1962) – The morphology of Kaldalón, a recently deglaciated valley in Iceland. Geografiska Annaler, 3-4, 347-365.

Joyner T.A., Rohli V.A. (2010) – Kernel Density Estimation of Tropical Cyclone Frequencies in the North Atlantic Basin. International Journal of Geosciences, 1 (3), 121-129.
DOI : 10.4236/ijg.2010.13016

Jónsson Ó. (1957) – Skriðuföll og snjóflóð. I og II. Bókaútgáfan Norðri, Akureyri. 586 p.

Jónsson Ó. (1976) – Berghlaup. Ræktunarfélag Norðurlands, Akureyri, 623 p.

Kristjánsson L., Jóhannesson H. (1994) – Stratigraphy and paleomagnetism of the lava pile south of Ísafjarðardjúp, NW Iceland. Jökull, 44, 3-16.
DOI : 10.1007/BF00240570

Le Breton E., Dauteuil O., Biessy G. (2010) – Post-glacial rebound of Iceland during the Holocene. Journal of the Geological Society, London, 167, 417-432.
DOI : 10.1144/0016-76492008-126

Leith K., Amann F., Moore J.R., Kos A., Loew S. (2010) – Conceptual modelling of near-surface extensional fracture in the Matter and Saas Valleys, Switzerland. Delegate Papers, Geologically Active, 11th Congress of the International Association for Engineering Geology and the Environment, Auckland, Aotearoa, 5–10 September 2010, Auckland, New Zealand, 363-371.

McColl S.T. (2012) – Paraglacial rock-slope stability. Geomorphology, 153-154, 1-16.
DOI : 10.1016/j.geomorph.2012.02.015

Mercier D. (2011) – La géomorphologie paraglaciaire. Changements climatiques, fonte des glaciers et crises érosives associées, Editions universitaires européennes, 256 p.

Mercier D., Cossart E., Decaulne A., Feuillet T., Jónsson H.P., Sæmundsson Þ. (2013) – The Höfðahólar rock avalanche (sturzström): Chronological constraint of paraglacial landsliding on an Icelandic hillslope, The Holocene, 23 (3), 431-445.
DOI : 10.1177/0959683612463104

Ólafsdóttir S. (2004) – Currents and climate on the northwest shelf of Iceland during the deglaciation: high-resolution foraminiferal research. M.Sc. thesis, University of Iceland, Reykjavik, 117 p.

Principato S.M., Geirsdóttir Á., Jóhannsdóttir G.E., Andrews J.T. (2006) – Late Quaternary glacial and deglacial history of eastern Vestfirðir, Iceland using cosmogenic isotope (36Cl) exposure ages and marine cores. Journal of Quaternary Science, 21, 271-285.
DOI : 10.1002/jqs.978

Principato S.M. (2008) – Geomorphic evidence for Holocene glacial advances and sea level fluctuations on eastern Vestfirðir, northwest Iceland. Boreas, 37, 132-145.
DOI : 1502-3885.2007.00003.x

Rapp A. (1960) – Recent development of mountain slopes in Kärkevagge and surroundings, Northern Scandinavia. Geografiska Annaler, 42, 65-200.

Rundgren M., Ingólfsson Ó., Björck S., Jiang H., Hafliðason H. (1997) – Dynamic sea-level change during the last deglaciation of northern Iceland. Boreas, 26, 201-215.
DOI : 10.1111/j.1502-3885.1997.tb00852.x

Selby M.J. (1993) – Hillslope materials and processes. Oxford University Press, Oxford, 451 p.

Sigurðsson O. (2004) – Gláma, að vera eða vera ekki – jökull. Náttúrufræðingurinn, 72, 47-61.

Stewart I.S., Sauber J., Rose J. (2000) – Glacio-seismotectonics: ice sheets, crustal deformation and seismicity. Quaternary Science Reviews, 19 (14-15), 1367-1389.
DOI : 10.1016/S0277-3791(00)00094-9

Syvitski J., Jennings A.E., Andrews J.T. (1999) – High-resolution seismic evidence for multiple glaciation across the southwest Iceland shelf. Arctic, Antarctic, and Alpine Research, 31, 50-57.
DOI : 10.2307/1552622

Sæmundsson K. (1979) – Outlines of the geology of Iceland. Jökull, 29, 7-28.

Thorarinsson S., Einarsson T., Kjartansson G. (1959) – On the geology and geomorphology of Iceland. Geografiska Annaler, 41, 135-169.

Whalley B., Douglas G.R., Jónsson Æ. (1983) – The magnitude and frequency of large rockslides in Iceland in the postglacial. Geografiska Annaler, Series A, Physical Geography, 65, 99-100.
DOI : 10.2307/520724

Haut de page

Annexe

Version française abrégée

Si les glissements de terrain sont nombreux en Islande du Nord-Ouest, ils sont peu étudiés à l’échelle locale ou régionale (Jónsson, 1957, 1976 ; Whalley et al., 1983). Ici, une nouvelle base de données des glissements de terrains visibles dans les fjords du Nord-Ouest est créée, et avec l’aide d’outils d’analyse spatiale, nous essayons de comprendre leur répartition. Jusqu’à présent, les principales explications apportées par les scientifiques sont liées au relief, notamment à la localisation dans les vallées glaciaires, à la position sur la pente ou à la nature de la roche. Pour Cossart et al. (2008, 2014), les facteurs déclenchants seraient des causes paraglaciaires (le rebond isostatique, la décompression post-glaciaire). 

L’aire d’étude correspond à la péninsule des fjords de l’Ouest, dans le Nord-Ouest de l’Islande (fig. 1). Elle était totalement couverte par une calotte glaciaire pendant le Weichselien, puis a subi une rapide déglaciation à partir de 15 ka. Aujourd’hui, Drangajökull est le seul glacier restant dans la péninsule, qui se caractérise principalement pas un relief de vallées découpées et de hauts plateaux, façonnés au cours des périodes glaciaires et interglaciaires. La plupart des plateaux s’élèvent entre 600 et 800 m et la dénivellation le long des fjords est généralement comprise entre 300 et 600 m. La majeure partie de l’aire d’étude se compose de roches basaltiques anciennes de 3 à 17 millions d’années. Le pendage général est orienté Sud-Est, excepté à l’Est du glacier Drangajökull (pendage Est) et au Sud-Est de la péninsule (pendage Sud).

Suite à une première prospection sur le terrain, un important travail basé sur l’observation d’images aériennes et la photo-interprétation a permis de réaliser un inventaire quasi exhaustif des glissements de terrain. Quatre sources ont été utilisées : GoogleEarth, le site de l’agence nationale de planification islandais (www.skipulasstofnun.is) et deux sites de bases de données cartographiques islandais (www.ja.is et www.map.is). Une nouvelle base de données a ainsi été créée, localisant les différents glissements observés sur l’aire d’étude, et listant les différents facteurs morphométriques et lithographiques propres à chaque entité. Ces mouvements de masse ont été cartographiés en utilisant les Systèmes d’Information Géographiques (suite ArcGis).

La seconde étape est l’étude de la répartition des glissements de terrain répertoriés. La méthode statistique utilisée est le calcul de densité d’un phénomène dans chaque zone d’un espace étudié. Plus les évènements sont proches, plus leur poids est important. L’application de l’estimation de densité de Kernel (KDE) permet de discriminer les espaces où le motif étudié est particulièrement fréquent ou au contraire quasi inexistant. Les calculs sont effectués avec le logiciel SagaGIS. Les résultats de la KDE sont ici interprétés qualitativement de façon binaire : présence ou absence de glissements de terrain. Pour rejeter ou non l’hypothèse nulle, un test de Fisher F-test est appliqué pour les paramètres topographiques, et un test de chi2 est appliqué pour la relation entre la localisation des glissements et le stade spatial de la déglaciation. Les résultats de la KDE sont associés au modèle numérique de terrain et ses dérivés pour calculer les principales statistiques des espaces concernés ou non pas les mouvements de terrain. Les résultats sont également intersectés avec les données de déglaciation et comparés avec les résultats de la répartition théorique obtenus par le test de chi2

La répartition des 186 glissements identifiés (fig. 2) montre une concentration des mouvements sur les pentes à proximité du littoral alors que les hauts plateaux de l’intérieur de la péninsule en sont dépourvus, faute de relief adéquat. Les plus fortes concentrations se trouvent dans l’extrême Nord-Ouest (entre les fjords d’Ísafjarðardjúp et Arnarfjörður, fig. 2A) dans une partie du Nord-Est (Strandir, fig. 2B) et du Sud (au niveau de l’isthme, fig. 2C) ainsi que dans les fjords du Nord de Breiðafjörður. 

Comme l’avaient observé Whalley et al. (1983, fig. 3A) à l’échelle de l’Islande, la plupart des glissements se trouvent sur des roches basaltiques Tertiaire. Cependant, leur localisation montre également que les zones de contact avec les roches intrusives sont des zones favorables aux glissements (fig. 3B). Le pendage est également un facteur favorable à la localisation des glissements de terrain (fig. 3C). Cependant, même combinés, ces facteurs géologiques ne suffisent pas à expliquer la répartition des glissements.

Les glissements visibles aujourd’hui et répertoriés ici se sont mis en place après le dernier maximum glaciaire et n’ont pas subi une autre glaciation significative. Il est possible de déduire l’âge maximum potentiel des glissements de terrain grâce aux calages chronologiques offerts par la littérature sur la déglaciation (Geirsdóttir et al., 2002, 2009, 2013). L’ensemble de la péninsule des fjords du Nord-Ouest était totalement couverte par la calotte glaciaire avant que la déglaciation ne commence à la fin du Weichselien, libérant alors la majorité des espaces côtiers durant l’interglaciaire du Bølling-Allerød. Les glissements côtiers se sont mis en place à partir de cette période puisque ces espaces côtiers n’ont pas été réenglacés lors du refroidissement du Dryas récent (fig. 4). Pendant le Préboréal, seuls les hauts plateaux centraux étaient englacés (fig. 4B). Avec un tel schéma de déglaciation, 73 % des glissements répertoriés pourraient avoir eu lieu dès la période de déglaciation située entre le Bølling-Allerød et le Dryas récent (tab. 1).

Les résultats de la KDE confirment l’observation préalable de la répartition des glissements, mais nécessite une analyse statistique approfondie. D’une part, contrairement aux modèles classiques, la topographie n’est pas significativement discriminante dans la répartition des glissements de terrain (fig. 5). D’autre part, les espaces affectés par les glissements sont largement associés au schéma de déglaciation, avec une surreprésentation dans les espaces désenglacés entre le dernier maximum glaciaire et le Dryas récent et au contraire une sous-représentation dans les espaces les plus récemment libérés des glaces (fig. 6). Il faut également prendre en considération la prédisposition de l’aire d’étude au rebond isostatique et l’observation des plages soulevées à différents points de l’aire d’étude.

Ainsi, les 186 glissements de tous types identifiés dans les fjords du Nord-Ouest viennent compléter significativement les données existantes. Si la répartition des glissements semble montrer une concentration sur les pentes proches du littoral, la topographie ne peut pas expliquer à elle seule cette répartition. Sans fournir d’éléments de datation définitif, la concentration des glissements dans les zones desenglacées entre le dernier Maximum Glaciaire et le Dryas récent, associée à la décompression post-glaciaire et au rebond isostatique, met en évidence le rôle des conditions paraglaciaires comme facteur déclenchant, mais nécessiterait une analyse statistique plus approfondie, ainsi que des éléments de datation fermes. Les causes des glissements nécessitent également plus de recherches pour être identifiées de façon absolue.

Haut de page

Table des illustrations

Titre Fig. 1 – Location map of the study area (A), typical landscape characteristics (B) and some views of rockslides in Svínadalur, Skálmarfjörður and Vatnadalur (C).Fig. 1Carte de localisation du terrain étudié (A), paysages caractéristiques (B) et quelques glissements de terrain dans Svínadalur, Skálmarfjörður et Vatnadalur (C).
Légende Photographs by Armelle Decaulne. Background map from SAR data source1. Fjord Ísafjarðardjúp towards W; 2. The Gláma plateau towards N; 3. Fjord Arnarfjörður towards W; 4. One of the rockslides of the Svínadalur valley; 5. Rockslide in the fjord Skálmarfjörður; 6. Rockslides in Vatnadalur.Photographies d’Armelle Decaulne. Image de fond de SAR data source.1. Fjord Ísafjarðardjúp, vue vers l'W ; 2. Le plateau de Gláma, vue vers le N ; 3. Fjord Arnarfjörður, vue vers l'W ; 4. Un des mouvements de terrain dans la vallée de Svínadalur valley ; 5. Mouvement de terrain dans le fjord de Skálmarfjörður ; 6. Mouvements de terrain dans le Vatnadalur.
URL http://geomorphologie.revues.org/docannexe/image/11303/img-1.jpg
Fichier image/jpeg, 468k
Titre Fig. 2 – Distribution of post-glacial rockslides, northwest Iceland.Fig. 2Localisation des glissements de terrain post-glaciaires en Islande du Nord-Ouest.
Légende 1: rockslides (from our own inventory).1 : glissement de terrain (d’après notre propre inventaire).
URL http://geomorphologie.revues.org/docannexe/image/11303/img-2.jpg
Fichier image/jpeg, 124k
Titre Fig. 3 – Rockslides and the geologic setting highlighting their occurrence within the Tertiary basalts (A – modified from Whalley et al., 1983), and at lithological contacts (B – modified from the geological maps from Jóhannesson and Sæmundsson, 1989) and local dip compliance (C – modified from the geological maps from Jóhannesson and Sæmundsson, 1998).Fig. 3Glissements de terrain et contexte géologique soulignant leur occurrence dans les basaltes tertiaires (A – modifié de Whalley et al., 1983), et aux contacts lithologiques (B – modifié de la carte géologique de Jóhannesson and Sæmundsson, 1989) et leur conformité au pendage (C – modifié de la carte géologique de Jóhannesson and Sæmundsson, 1998).
Légende 1: rockslide (from Whalley et al., 1983 in A, and our inventory in B and C); 2: dip (from the geological map from Jóhannesson and Sæmundsson, 1998); 3: active zone; 4: Quaternary formations; 5: Tertiary formations; 6: basic and intermediate extrusive rocks with intercalated sediments (Lower to Upper Miocene); 7: acid extrusives (middle Miocene); 8: basic and intermediate intrusions; 9: acid intrusions; 10: extinct central volcanoes; 11: major sedimentary horizon with lignite; 12: compliant dip rockslides; 13: perpendicular dip rockslides; 14: non-compliant dip rockslides.1 : glissement de terrain (d’après Whalley et al., 1983, en A ; notre inventaire en B et C) ; 2 : pendage (d’après la carte géologique) ; 3 : zone active ; 4 : formations quaternaires ; 5 : formations tertiaires ; 6 : roches extrusives basiques et intermédiaires avec sédiments intercalés (Miocène inférieur et supérieur) ; 7 : extrusions acides (Miocène moyen) ; 8 : intrusions basiques et intermédiaires ; 9 : intrusions acides ; 10 : volcans centraux éteints ; 11 : horizons sédimentaires à lignite majeurs ; 12 : glissements de terrain conformes au pendage ; 13 : glissements de terrain perpendiculaires au pendage ; 14 : glissements de terrain non conformes au pendage.
URL http://geomorphologie.revues.org/docannexe/image/11303/img-3.jpg
Fichier image/jpeg, 200k
Titre Fig. 4 – Potential age maximum for rockslides based on Late Weichselian ice extent reconstruction (from Geirsdóttir et al., 2009).Fig. 4 – Age maximum potentiel pour les mouvements de masse sur la base de la reconstitution de l’extension glaciaire au cours du Weichselien Tardif (d’après Geirsdóttir et al., 2009).
Légende A: Maximal extent of glacier readvance during the Younger Dryas cold event; B: Maximal extent of glaciers during the Preboreal period; C: Maximal extent of glaciers during the Holocene; 1: glacial extent; 2: rockslides which can’t have potentially occurred before the Bølling-Allerød interstadial and the Younger Dryas; 3: rockslides which can’t have potentially occurred before the glacial retreat following the maximal ice extent of the Younger Dryas; 4: rockslides which can’t have potentially occurred before the glacier retreat following the maximal ice extent of the Preboreal.A : Extension maximum de la réavancée glaciaire au cours de l’épisode froid du Dryas Récent ; B : Extension maximum des glaciers au cours de l’épisode froid du Préboréal ; C : Extension maximum des glaciers au cours de l’Holocène. 1 : extension glaciaire ; 2 : les glissements de terrain qui ne peuvent être potentiellement survenus avant l’interstade Bølling-Allerød et le Dryas récent ; 3 : les glissements de terrain qui ne peuvent être potentiellement survenus avant le retrait glaciaire qui a suivi l’extension maximale de la glace au Dryas récent ; 4 : les glissements de terrain qui ne peuvent être potentiellement survenus avant le retrait des glaciers qui a suivi l’extension maximale de la glace au Préboréal.
URL http://geomorphologie.revues.org/docannexe/image/11303/img-4.jpg
Fichier image/jpeg, 796k
Titre Tab. 1 – Potential maximal age periods of occurrence of recorded rockslides.Tab. 1 – Age maximum potentiel des séquences d’apparition des glissements de terrain inventoriés.
URL http://geomorphologie.revues.org/docannexe/image/11303/img-5.png
Fichier image/png, 15k
Titre Fig. 5 – Density of rockslides, estimated by a Kernel Density Estimator (Background DEM from SAR data source).Fig. 5Densité de glissements de terrain, estimée à partir d’un lissage de Kernel (MNT de fond de SAR data source).
URL http://geomorphologie.revues.org/docannexe/image/11303/img-6.jpg
Fichier image/jpeg, 92k
Titre Fig. 6 – Main statistical characteristics discriminating (or not) areas affected by rockslides and areas non-affected by rockslides.Fig. 6Principales caractéristiques statistiques visant à discriminer (ou pas) les zones affectées par des mouvements de masse de celles non-affectées par des mouvements de masse.
Légende A: F-test comparing topographical parameters (slope gradient, valley-depth) in stable and unstable areas, null hypothesis is not rejected. B: chi²-test showing a significant overrepresentation of rockslides in early-deglaciated areas.A : Test de Fisher comparant des indicateurs topographiques (pente, profondeur des vallées) entre les zones stables et instables. B : Test du chi² montrant la recrudescence significative dans les zones précocement déglacées.
URL http://geomorphologie.revues.org/docannexe/image/11303/img-7.jpg
Fichier image/jpeg, 42k
Haut de page

Pour citer cet article

Référence papier

Aurore Peras, Armelle Decaulne, Etienne Cossart, Julien Coquin et Denis Mercier, « Distribution and spatial analysis of rockslides failures in the Icelandic Westfjords: first results », Géomorphologie : relief, processus, environnement, vol. 22 – n° 1 | 2016, 25-35.

Référence électronique

Aurore Peras, Armelle Decaulne, Etienne Cossart, Julien Coquin et Denis Mercier, « Distribution and spatial analysis of rockslides failures in the Icelandic Westfjords: first results », Géomorphologie : relief, processus, environnement [En ligne], vol. 22 – n° 1 | 2016, mis en ligne le 16 mars 2016, consulté le 26 mars 2017. URL : http://geomorphologie.revues.org/11303 ; DOI : 10.4000/geomorphologie.11303

Haut de page

Auteurs

Aurore Peras

Université de Nantes, Campus du Tertre – BP 81227, 44312 Nantes cedex 3, France (a.peras@laposte.net).

Armelle Decaulne

CNRS-UMR 6554 LETG – Campus du Tertre – BP 81227, 44312 Nantes cedex 3, France (armelle.decaulne@univ-nantes.fr). Tél : +33 2 53 48 76 58.

Articles du même auteur

Etienne Cossart

Université Lyon 3 et CNRS-UMR EVS 5600 – 18 rue Chevreul, 69007 Lyon cedex 07, France (etienne.cossart@univ-lyon3.fr).

Articles du même auteur

Julien Coquin

CNRS-UMR 6554 LETG – Campus du Tertre – BP 81227, 44312 Nantes cedex 3, France (julien.coquin@wanadoo.fr).

Articles du même auteur

Denis Mercier

Université Paris Sorbonne, CNRS-UMR 8185 ENeC – 191 rue Saint Jacques, 75005 Paris, France (denis.mercier@paris-sorbonne.fr).

Articles du même auteur

Haut de page

Droits d’auteur

© Groupe français de géomorphologie

Haut de page
  • Logo CNRS - Institut des sciences humaines et sociales
  • Logo Groupe français de géomorphologie
  • Revues.org