The author is indebted to M. Baker (University Paris 1) who edited the English manuscript. We thank two anonymous reviewers and J.-C. Thouret for his constructive comments on the original draft.
1Introduced by J. Ryder (1971), the term paraglacial literally means “beyond the glacier” (Slaymaker, 2004), and refers to “non-glacial processes conditioned by glaciation” (Church and Ryder, 1972). While this concept was first largely ignored outside North America, there is a significant increase of papers focusing on paraglacial processes since the 1990s (Mercier, 2007). Indeed, the use and significance of the term has evolved, and a working definition of paraglacial was proposed by C.K. Ballantyne (2002) in a most important review: “non-glacial processes, sediment accumulations, landforms, land systems and landscape that are directly conditioned by glaciation and deglaciation”. However, while the author attempted to define a typology of the geomorphic contexts in which the paraglacial activity may occur, many papers now use the term paraglacial “to cover a bewilderingly large variety of circumstances, almost making the word redundant” (Slaymaker, 2007). Yet a paraglacial influence was for instance evident both in the debuttressing of rock-slopes (Bovis, 1990; Shakesby and Matthews, 1996; Ballantyne and Stone, 2004; Cossart et al., 2008) and in the progradation of some deltas (Mercier and Laffly, 2005). As the paraglacial concept is now accepted and appears in various contexts (Ballantyne, 2002), it is necessary to better document what the limitations of the paraglacial activity are, both in space and time. The duration of the paraglacial period is indeed difficult to determine (Ritter and Ten-Brink, 1986; Orwin and Smart, 2004; Étienne et al., 2008). M. Church and J. Ryder (1972) defined the paraglacial period as the period following the deglaciation during which the sediment yield is higher than the “geological norm” (fig. 1A).
Fig. 1 – The paraglacial model.
Fig. 1 – Le modèle paraglaciaire.
A: paraglacial period defined by M. Church and J. Ryder (1972); B: application of the exhaustion model to assess the evolution of the volume of sediments within a paraglacial store (Ballantyne, 2003).
A : la période paraglaciaire, définie par M. Church et J. Ryder (1972) ; B : estimation, par le modèle de tarissement, de l’évolution du volume de sédiments stockés dans un réservoir paraglaciaire (Ballantyne, 2003).
2However the comparison of sediment yield with a “geological norm” is highly questionable, hampering an accurate identification of the end of the paraglacial period. More recently, C.K. Ballantyne (2002) suggests that the paraglacial period lasts as long as the paraglacial sediment sources and paraglacial sediment storage areas are not exhausted and that sediment yield declines following an exponential law, according to the volume of available sediments (fig. 1B). The main problem is that some paraglacial sediment sources may be disconnected from the sedimentary cascade and then are never exhausted, especially at the Holocene time-scale (Cossart and Fort, 2008a). According to this result, paraglacial processes may contribute intermittently to the paraglacial sedimentary cascade.
3This paper aims to focus on the intermittence of paraglacial sedimentary cascade during the paraglacial period, using the example of an alpine catchment. Our approach is based on the hypothesis that the intermittence of paraglacial processes is due to some negative feedbacks during the geomorphic adjustments to glacier shrinking, which reduce significantly the sediment export. The identification of these feedbacks is based upon an analysis of a sedimentary cascade through time and to an assessment of connectivity between the components of the system. This requires accurate chronological data, thus a study at a short timescale (Cossart, 2004; Étienne et al., 2008): the period following the Little Ice Age (LIA). We focus on the Upper Durance catchment, namely on the Tabuc subcatchment (Guisane valley) where chronological data were already obtained (Cossart et al., 2006), making the reconstruction of temporal sequences easier.
4The Upper Durance catchment (fig. 2A) is an area particularly prone to paraglacial processes, during sequences following both the Last Glacial Maximum (LGM) and the Little Ice Age (LIA) periods.
Fig. 2 – The Tabuc subcatchment.
Fig. 2 – Le bassin du Tabuc.
A: regional setting; B: the Tabuc subcatchment in the Guisane catchment; 1: main ridges and summit; 2: stream; 3: town; 4: glacier; 5: LIA moraine; 6: Last Glacial Maximum trimline; 7: LIA equilibrium line; 8: current equilibrium line; C: oblique view of the upper part of the subcatchment. The regenerated glaciers are fed by avalanches from north-faced mountain slopes (photograph É. Cossart, June 2004). 1: debris-covered glacier; 2: frontal moraine; 3: breaches; 4: Pré les Fonds stream; 5: scree and avalanches cones; 6: Late Glacial moraine; D: oblique view of the valley-train named “Pré du Tabuc”. Vegetation covers two thirds of the plain, abandoned channels are suggested by dashed lines (photograph É. Cossart, June 2004).
A : contexte régional ; B : localisation du vallon du Tabuc, dans la vallée de la Guisane ; 1 : crêtes et sommet ; 2 : cours d’eau ; 3 : village ; 4 : glacier ; 5 : moraine du petit âge de glace ; 6 : niveau d’englacement lors du dernier maximum glaciaire ; 7 : ligne d’équilibre glaciaire lors du petit âge de glace ; 8 : ligne d’équilibre glaciaire actuelle ; C : partie amont du vallon du Tabuc. Les glaciers régénérés sont alimentés par les avalanches depuis les versants en ubac (cliché É. Cossart, juin 2004). 1 : glacier couvert ; 2 : moraine frontale ; 3 : brèches ; 4 : torrent de Pré les Fonds ; 5 : cône s ; 6 : moraine tardiglaciaire ; D : vue oblique du glarier à chenal tressé du Pré du Tabuc. Le couvert végétal s’étend sur les deux tiers de la plaine, les chenaux abandonnés sont suggérés en tiretés (cliché É. Cossart, juin 2004).
5The Durance glacier was the main glacial tongue of the southern French Alps during the LGM period; its retreat implied a significant release of meltwater and generated debuttressing forces on mountain slopes (Gidon and Montjuvent, 1969; Lahousse, 1994; Cossart, 2008). The geomorphic consequences are partially documented, highlighting that the sediment export through catchments are relatively low while high amounts of debris were supplied by slope instabilities. Landslides are indeed common features in this area: they probably occurred shortly after the disappearance of main glacial tongues (Cossart et al., 2008). Many landslides caused persistent river damming, producing large (1 to 5.106m3) sediment traps in subcatchments, because the landslide dams are often made up of large boulders that cannot be removed by rivers (Cossart and Fort, 2008a). Finally, the efficiency of the paraglacial cascade system is questioned in such an alpine headwater, and possible interruptions of the paraglacial sedimentary cascade should be documented.
6In this paper, the research is based on a short temporal scale, i.e. during the LIA deglaciation to make the reconstruction of the geomorphic sequences easier and more precise, for instance using archives or chronicles and lichenometry. The geomorphic setting of Tabuc subcatchment is suitable for this topic. Firstly, it is typical of an alpine glacierized subcatchment. Altitudes range from 1470 m.a.s.l. (at the outlet of the catchment at its confluence with the Guisane, next to Monêtier-les-Bains) to 3650 m.a.s.l. at the Montagne des Agneaux (fig. 2B). Second, a landslide occurred from the right bank of the river at the altitude of 1800 m.a.s.l., creating a persistent dam and isolating the “Pré du Tabuc” floodplain. This floodplain is densely covered by vegetation (larches), suggesting a current low rate of activity despite the vicinity of the glaciers and sediment sources (figs. 2C and 2D): it collects sediment fluxes from the Glacier des Dômes du Monêtier, from north-facing mountain slopes (sediment transfer by gravity and avalanches) and from south-facing mountain slopes (by reworking of glacigenic materials and scree taluses). In both mountain slopes the bedrock consists of massive rocks and provides large metric boulders: gneisses and granites build up the north-facing slopes, massive Triassic limestones and dolomites build up the south-facing slopes.
7The Equilibirum Line Altitude (ELA) currently ranges from 3000 m.a.s.l. on the north-facing slopes to 3200 m.a.s.l on the southern slopes, allowing the development of small glaciers in the order of 4 km. Indeed, only small north-facing regenerated glaciers still reach the valley floor; on the south-facing slopes glaciers are now very limited, perched a few hundred meters above the valley floor. However, the glacier area was twice during the LIA as evidenced by many moraines. The timing of the deglaciation has already been established by lichenometry (Cossart et al., 2006). The oldest glacial remnant is a left-flank moraine that is probably a Late-Glacial landform. During the recent LIA period, there was a 2-km-long valley glacier, whose maximal advance occurred around AD 1860. The front then reached the altitude of 2250 m.a.s.l. During the Late-LIA, the front stayed at the altitude of 2300 m.a.s.l. until AD 1918-1928 due to regenerated glaciers that merged from the north-facing slope (fig. 3). Only subtle variations of the glacier occurred between 1923 and the 1950s, when significant retreat began. As the glacier front stalled for many decades around 2300 m.a.s.l. a large moraine ridge was formed.
Fig. 3 – Geomorphic map of the study area. Coordinates are in UTM (32 N).
Fig. 3 – Cartographie géomorphologique de la zone étudiée (Coordonnées en UTM, zone 32N).
1: ridge and summit; 2: glacier (in 2004); 3: glacigenic deposits and moraine ridges (date mentioned in the map); 4: debris cone; 5: sagging; 6: aggradational plain; 7: rock-slope; 8: stream. A: see fig. 5, B: see fig. 6, C: see fig. 7.
1 : crête et sommet ; 2 : glacier (en 2004) ; 3 : dépôts glaciaires et crêtes morainiques (date précisée sur la carte) ; 4 : cône de débris ; 5 : sackung ; 6 : plaine d’aggradation; 7 : versant rocheux; 8 : cours d’eau. A : voir fig. 5, B : voir fig. 6, C : voir fig. 7.
8The objective of the work was to document the sequences of the paraglacial evolution pattern in the Tabuc subcatchment to eventually discriminate periods at which evacuation of glacigenic sediment is efficient and periods at which the paraglacial sedimentary cascade is interrupted. In alpine areas, sedimentary cascade aims at describing how sediment is routed from its sources, through various storage reservoirs, to the basin outlet (Jordan and Slaymaker, 1991), then documenting the landscape evolution patterns (Schrott and Adams, 2002) (fig. 4). Yet, major problems are related to the highly variable and changing residence times of stored sediments (Caine, 1986; Slaymaker, 2003).
Fig. 4 – Mesoscale sedimentary cascade components of the Tabuc Subcatchment.
Fig. 4 – Éléments de la cascade sédimentaire du vallon du Tabuc, à méso-échelle.
1: subsystem; 2: storage; 3: flux; 4: regulator.
1 : sous-système ; 2 : stockage ; 3 : flux ; 4 : régulateur.
9The paraglacial landsystem is considered as a cascade system through which the impacts of deglaciation are propagating downward, affecting sediment transfers (Church and Slaymaker, 1989). The spatial and temporal continuum of processes that stems from a decaying glacier must be reconstructed to identify what are the components of the cascade sedimentary system that are really under the influence of paraglacial activity. This reconstruction requires examining the following links in the Upper Tabuc catchment: mountain slope-valley floor, channel-floodplain, upstream-downstream (Chorley and Kennedy, 1971; Warburton, 1990; Brierley et al., 2006). Furthermore, it is assumed that a decrease in sediment yield could be due to some negative feedbacks during the paraglacial evolution. The negative feedbacks that may have an impact on such sediment fluxes are examined to assess the temporal variability of the rate of sediment export and to document the duration of the paraglacial period.
10A geomorphic map has been carried out to identify the components of the cascade sedimentary system within the upper watershed. Following several authors (Chorley and Kennedy, 1971; Caine, 1986; Schrott and Adams, 2002), such components can be subdivided into three categories or subsystems: sources, where only degradation may occur; storage areas, where both aggradation and degradation may occur; and sinks where only aggradation may occur. The components are identified and mapped: we particularly focused on glacigenic deposits (till, lateral moraines), slope deposits (scree, avalanches, and debris flows deposits), and valley-floor deposits (fig. 3). Such components can be connected to each other through processes: the connectivity is identified by the geomorphic active areas (Schrott et al., 2003; Otto and Dikau, 2004) and their extent in the cascade system (fig. 4).
11In alpine environments, an area is considered to be subject to geomorphic reworking if the vegetation cover is less than 20% (Schrott et al., 2003). Active surface areas can therefore be identified at various dates based on aerial photographs or old maps (tab. 1).
Table 1 – Inventory of archive documents. Aerial photograph acquired in 1971 provides an incomplete view of the study area.
Tableau 1 – Inventaire des documents d’archives. La photographie aérienne de 1971 ne fournit qu’une vue partielle de l’aire étudiée.
Year
|
ID
|
Type of document
|
Scale
|
2003
|
FD 05
|
Aerial photograph
|
1:25000
|
1993
|
FD 05
|
Aerial photograph
|
1:20000
|
1981
|
IFN 05 P
|
Aerial photograph
|
1:17000
|
1971
|
FR 2116
|
Aerial photograph
|
1:15000
|
1967
|
F 3435 - F3436
|
Aerial photograph
|
1:25000
|
1960
|
F 3435 – F 3436
|
Aerial photograph
|
1:25000
|
1952
|
F 3435 – F 3436
|
Aerial photograph
|
1:25000
|
1928
|
3436
|
Topographic map
|
1:20000
|
12The active areas were mapped within a GIS at various dates: 1928, 1952, 1960, 1971, 1981, 1993, and 2003. In the field, lichens such as Rhizocarpon geographicum are used for depicting active areas because they cannot grow on unstable ground (Benedict, 1967): the date at which a surface becomes stable can then be revealed by the age of the lichens. The mapped active areas were merged with sediment storage areas in order to calculate a ratio of active to inactive areas for each component of the cascade sedimentary budget (Schrott et al., 2006) and to identify the type of connectivity between the components. Four types of sediment fluxes can be identified, then interpreted in terms of connectivity: (a) sediment input and output occurs (total connectivity); (b) sediment input but no output occurs (disconnectivity); (c) sediment output occurs but without input (sediment export downstream); (d) neither input nor output occurs (area decoupled with the cascading system). Some snapshots of connectivity are drawn from the database from 1923 to 2003.
13Calculations of sediment yield are based upon a sediment budget approach, following the statement that sediment is routed through storage reservoirs: sediment fluxes are inferred from the variation of the volume of the reservoirs. In details, the rate of sediment export is assessed from the volume of degradational features that affect a reservoir: they were modeled as simple geometric landforms (Campbell and Church, 2005; Cossart and Fort, 2008). For instance, gullies or trench shapes are modeled as a triangular prism. The pre-degradation landform is assumed to correspond to the geometry of the ‘time-zero’ of landform evolution. The calculation requires basic measurements (height, width, etc.) provided by field measurements (Leica laser telemeter).
14Since 1975, the Dômes du Monêtier glacier significantly retreated: in 2004 its front was located 260 meters above and 810 meters away upstream from the 1975 frontal moraine. However, the north-facing slope is still glacierized due to the influence of regenerated glaciers, while the south-facing mountain slope is ice free, where the 1923 lateral moraine is completely disconnected from current glaciers and subject to reworking (figs. 3 and 5A). On the valley-floor, the proglacial channels are reworking the till that was exposed since 1975 (fig. 5B).
Fig. 5 – Geomorphic description of the area that was deglaciated during the 1950s.
Fig. 5 – Description géomorphologique de la zone déglacée depuis la décennie 1950.
A: transverse profile. The debris cones are decoupled from the area reworked by proglacial streams; B: proglacial stream reworking the till deposits. The largest debris of the glacigenic deposits cannot be carried out by streams, armouring the channel bed (Photograph É. Cossart, July 2003).
A : profil en travers. Les cônes de débris ne sont pas remaniés par les cours d’eau proglaciaires ; B : torrent proglaciaire remaniant les dépôts de till. Les débris plus grossiers, qui ne sont pas remaniés par le cours d’eau, pavent le lit (cliché É. Cossart, juillet 2003).
15Glacigenic deposits represent large sediment sources that are subject to reworking: fourteen hectares are completely free of vegetation cover (even lichens), suggesting a high rate of activity. However, sediment export does not seem efficient. On the south-facing slope, the lateral moraine is gullied; some debris cones are built at the contact with the valley floor. Such debris cones are still growing: their height is estimated as 20 meters in 1993 (based on aerial orthophotos draped on a DEM in a GIS), and was about 30 meters in 2003. Moreover no degradational scar is identified on aerial photographs (1993, 2003) and in the field (2004). It suggests there is no coupling with the proglacial outlet and the sediment storage areas located downward (fig. 5A).
16On the valley floor, the streams are reworking the disorganized till deposits. Examination of the aerial photographs acquired since 1981 indicates there is no persistent degradational feature due to river incision. Furthermore, the lack of valley-train terrace and the multi-thread channel suggest the stream power is too low to allow the removal of the large boulders from the till deposit (fig. 5B). A comparison of grain-sizes in the till deposit and inside the channel suggests that sediment export only consists of finer debris. An armouring pattern of the channel hampers the incision and degradation of the sediment storage area.
17Three ridges have formed a large frontal edifice made up of large boulders (up to 10 m3). They recorded the minor variations that characterized the front of the glacier during the Late LIA. On the left flank of the valley the moraine ridge is 20 to 30 m-high and 50 to 80 m-wide ridge. This lateral edifice is connected to a frontal one, which also constitutes a 300 m-large and 80 m-high landform (fig. 3). The outer ridge of both edifices corresponds to the LIA advance during the 1850s, whereas the inner was disconnected from the glacier during the 1950s. Between both, a ridge corresponds to a Late LIA stage dating back to the 1920s based on lichenometry. The deposit of this moraine was provided by rock falls and avalanches from the northern slope, which were deposited at the surface, then at the front of a regenerated glacier.
18Glacial deposits are major sediment sources that could be reworked to export paraglacial sediments but only small areas of such sources are currently active: an area of only 0.9 of the 5.6 hectares of the lateral glacigenic deposits is active, and only 1.8 of the 7.2 hectares of the frontal edifice. However, old maps and aerial photographs suggest that such areas were geomorphically active both in 1928 and in 1952. More precisely, according to aerial photographs, the shrinkage of active areas occurred between 1952 and 1960 (fig. 6). Since 1960, the values seem quite constant.
19During the 1950s, the geomorphic evolution was highly linked with a significant shrinking of the glaciers: the regenerated glacier and the main glacial tongue became disconnected. The spatial extent of the cone-shaped regenerated glacier decreased. As a consequence, the meltwater streams that washed the frontal moraines therefore merged into two channels. Between the 1920s and 1952, nine streams impinged and gullied the frontal deposits. Still visible in the field, gullies are 3 to 8 m-wide and 1 to 2 m-deep. According to their geometry, sediment export from the edifice was at least 8 x 103 m3 during 32 years. During the 1950s, the river source became lower while the frontal moraine still constituted a local base level (fig. 6A), so that only two channels reworked the deposits in 1960. Such channels were still active in 2005. They are slightly more deeply incised (3 m) than the former active gullies and very narrow (10 m-wide). The breaches incised across the frontal moraine imply the abandonment of the gullies and a significant reduction of the area subject to sediment reworking. According to the geometry of breaches, the sediment export has been about 100 m3 since 1960.
Fig. 6 – Geomorphic evolution pattern in the area close to the frontal moraine.
Fig. 6 – Évolution géomorphologique dans le secteur de la moraine frontale.
A: role of the glacial retreat on the abandonment of gullies reworking the frontal moraine; B: transverse profile. The lateral glacigenic deposit is decoupled from the proglacial streams. Note the impoundment of the area located between the lateral moraine and the slope; C: active vs. inactive areas on the glacigenic deposits; 1: active area; 2: inactive area.
A : rôle du retrait du front glaciaire dans l’abandon des ravines remaniant la moraine frontale ; B : profil transversal. Les dépôts glaciaires latéraux ne sont pas remaniés par les cours d’eau proglaciaires. Le matériel sédimentaire s’accumule entre la moraine et le versant ; C : extension des zones actives et inactives sur les dépôts glaciaires, dans le secteur de la moraine frontale ; 1 : zone active ; 2 : zone inactive.
20Another consequence of glacier shrinking can be evidenced between the 1920s and the 1950s. The meltwater streams were laterally confined by the glacier against the left flank moraine (fig. 6B). The inner face of the ridge was constantly rejuvenated, as shown by lack of vegetation in the 1952 aerial photograph. This suggests that debris were removed down-valley. Since 1960, the streams are not confined any more and do not flow against the moraine. Some debris cones are growing, immunizing the sediment source. Furthermore, the lateral moraine impacts on the geomorphic evolution of the south-facing slope. A stream, confined between the external flank of the lateral moraine and the slope deposits, flows from the margin of a small hanging glacier. As the lateral moraine generates a persistent local base level, glacio-fluvial deposition is forced and the stream cannot remove debris from the scree cones. Thus no degradational trend can occur because the base level is constant: a complex impoundment is evidenced, implying that the south-facing slope is decoupled from the cascade sedimentary system of the watershed.
21Downstream of the large moraine ridges, the 0.3 km-wide floodplain is called “Pré du Tabuc”. The morphology of the floodplain has changed significantly since the 1920s (fig. 7).
Fig. 7 – Geomorphic evolution of the Pré du Tabuc.
Fig. 7 – Évolution géomorphologique du Pré du Tabuc.
A: evolution of the aggradational plain; 1: active channel; 2: areas covered by shrubs; 3: areas covered by trees; 4: perennial stream; 5: temporary stream; B: evolution of the surface of the active area.
A : évolution de la plaine d’aggradation. 1 : bande active ; 2 : aire couverte par des arbustes ; 3 : aire couverte par des arbres ; 4 : cours d’eau pérenne ; 5 : cours d’eau temporaire ; B : évolution de la surface de la bande active.
22The 1928 map shows no trees, and no lichen is older than about 40 years: the whole valley floor was an active area during the 1920s. This high activity lasted for decades: in 1952, only 0.9 of the 31 hectares were covered by trees. Through time, the surface of the vegetated areas has quickly spread out: 8.3 ha are identified in 1960 and 16.9 in 1981. Since 1981, most of the surface is vegetated and can be considered as inactive (fig. 7b): 16.7 hectares were covered by vegetation in 2003. The 2003 area approximately equals the 1981 area, despite the trend to more vegetation. A thunderstorm in August 2002 partially reactivated a previously abandoned channel. However, the pluri-decadal trend reveals a significant decrease in the extent of the active channel that is directly linked with the degradation of the large frontal moraine. Former channels that reworked the entire surface of the floodplain were directly connected to the gullies that incised the frontal moraine as it can be observed today (fig. 8).
Fig. 8 – Decoupling of the mountain slopes due to the contraction of the active area on the valley floor.
Fig. 8 – Découplage entre les versants et le fond de vallée, à la suite de la contraction de la bande active.
A: view of the contact between the northern slope and the valley floor. The former channels are now inactive while the trees are bent by avalanches (photograph É. Cossart, May 2002). B: transverse profile. The area subject to avalanches is decoupled from the current active channel.
A : vue du contact entre les dépôts de bas de versants et le fond de vallée. Les anciens chenaux sont à présent inactifs tandis que les arbres sont courbés sous l’effet des avalanches (cliché E. Cossart, mai 2002). B : profil transversal. La zone d’extension des avalanches est découplée de la bande active actuelle.
23Between 1928 and 1952 at least five channels were active. They incised from one to two meters into the plain. According to the geometric extent of 1.45 km-long; 5 to 10 m-wide channels, we can estimate the volume of exported sediments due to their geomorphic work to 80 x 103 to 130 x 103 m3, corresponding to an average rate of 3 x 103 to 5 x 103 m3 per year.
24The canalization of the sediment fluxes that have occurred in the moraine area during the 1950s also impacted downstream, so that only one channel remains active in the Pré du Tabuc. Other channels that crossed the plain became inactive, implying an abrupt contraction of the active plain between 1952 and 1960 (fig. 8). The width of the active channel has been remarkably constant since 1960. The incision trend of the active channel is rather low because abandoned channels are only perched at two to three meters above the current valley floor. Since 1960, sediments export from the floodplain has been due to the incision of the channel and may be estimated at between 45 and 90 x 103 m3 (according to the estimated deepening and the geometric extent of the channels). It corresponds to an average rate of about 1 x 103 m3 to 2 x 103 m3 per year so that paraglacial geomorphic activity has significantly decreased since the middle of the 20th century. This modification of the average rate of erosion occurred abruptly: a threshold effect is evident since the canalization of fluxes in the moraine area.
25Concomitantly, the avalanche cones that developed on the northern slope have been decoupled from the fluvial streams. The avalanches do not reach the active channel, as evidenced by bent trees and lichen distribution on blocks. However, the debris provided by the avalanches can bury abandoned channels. According to aerial photographs, this burying pattern of channels has begun during the 1950s; avalanches contributed to the sedimentary cascade during the first half of the 20th century because debris could be removed by active streams. Moreover debris supply by avalanches occurred at a higher rate than today during the LIA and at the beginning of the 20th century, in the area (Jomelli and Francou, 2000; Jomelli and Pech, 2004). Such supply of debris no longer contributes to the cascade sedimentary system.
26In this case study, two major interruptions of the sedimentary cascade are caused by the evolution of the large frontal moraine. Firstly, sediment export is impeded upstream of the frontal moraine. While glacigenic deposits are subject to reworking since the glacier retreat, two negative feedbacks affect the organization of the sediment cascade (fig. 9).
Fig. 9 – Negative feedback during the paraglacial period in the area close to the frontal moraine.
Fig. 9 – Boucle de rétroaction négative durant la phase paraglaciaire dans la zone située à proximité de la moraine frontale.
27(1) The formation of debris cone by lateral moraine gullying renders the moraine deposit immune; glacigenic sediments are progressively decoupled from the cascade sedimentary system: the active area on the moraine deposits is decreasing as the debris cones are growing. (2) The river was able to carry out only the matrix from the glacigenic deposit. The armouring of the bed hampers incision, thus the sediment export. Moreover, as the local base level (i.e. the river) is constant in this part of the catchment, lateral debris cones cannot be degraded. The position of the study area in the headwater of the fluvial catchment favors the trapping of coarse debris and the blockage of paraglacial sedimentary cascade.
28Secondly, downstream of the frontal moraine, a negative feedback occurred during the 1950s. More precisely, a decrease in sediment export from the valley-train deposit is due to three factors (fig. 10):
Fig. 10 – Negative feedback during the paraglacial adjustment in the Pré du Tabuc area.
Fig. 10 – Boucle de rétroaction négative durant la phase d’ajustement paraglaciaire dans le secteur du Pré du Tabuc.
29(1) a canalization of the active channels that impinged the frontal moraine provoked a disconnection of the main areas of the valley-train floor from the sedimentary cascade; (2) once all water fluxes were confined and the active surface area was reduced, the incision of the moraine was blocked by channel bed armouring, hampering the continuation of the incision of the valley-train floor, and (3) the debris supplied by avalanching from the north-facing slopes is no longer exported and is stored within the inactive part of the valley-train such as abandoned channels.
30The size of the boulders of the large frontal moraine produced a persistent dam at the post-LIA time scale. Recently, several papers (Meigs et al., 2006; Cossart and Fort, 2008b; Étienne et al., 2008) demonstrated that both lateral and frontal moraines may temporarily interrupt the sediment fluxes. However, such disconnections generally last for one or two decades and the dismantlement of the moraines may subsequently produce a rapid and massive export of sediment (Boulton, 1986; Cossart, 2004; Meigs et al., 2006). In the present study, the blockage is significantly longer: a persistent damming by a frontal moraine is still efficient, due to the large grain-size and the volume of the frontal moraine.
31This pattern reinforces the idea that glacial deposits and paraglacial features can produce blockages of sediment fluxes in alpine areas because they are located in headwaters where rivers cannot remove some large boulders. The complete exhaustion of glacigenic sediments can thus be delayed and many disconnections within the paraglacial sedimentary cascade may occur during the paraglacial period. Consequently, the main question revolves around the duration and the reversibility of the blockage. Two main patterns of future evolution can be drawn. On the one hand, J.-J. Clague (1986) illustrated that reversibility may occur during paraglacial period without modifying the long-term trend at Holocene timescale. Following this observation, the Tabuc subcatchment can be considered as currently characterized by an ongoing paraglacial phase within which adjustments are intermittent. Sediment export would probably increase if the stream again impinges on the frontal moraine. This hypothesis seems to be highly probable but, on the other hand, some long-term bockages were highlighted when paraglacial landslides induced persistent dams during the Holocene (Slaymaker and Owens, 2004; Cossart and Fort, 2008a). As in these studies, the geomorphic evolution of the Tabuc subcatchment may refer to a “disturbance cycle” (Slaymaker and Owens, 2004; Hewitt, 2006), characterized by an impoundment stage, forcing local aggradation in the headwater and reducing the sediment export downstream. Following this statement, the valley floor would be considered as decoupled from most sources of paraglacial sediments. Sediment export from the Pré du Tabuc would then be due to both avalanching and particular climatic events such as thunderstorms, which can temporarily reactivate abandoned channels. We cannot currently decipher whether the interruption of the paraglacial cascading system will cease and, if so, when it will cease. Patterns of coupling between sources, storage and sink areas in deglaciated areas are highly variable, both in space and time (Warburton, 1990, 1992; Stott and Grove, 2001; Knudsen et al., 2007), requiring further studies focusing on sediment budget approaches to document the intermittence of sediment export during the paraglacial period.
32The original paraglacial evolution of the Tabuc subcatchment is due to the strong influence of a massive frontal moraine, made up of large metric boulders provided by avalanches and rock falls. Its subdued dismantlement generated three negative feedbacks on the paraglacial activity that are still sensitive today: the river cannot degrade the sediment storage areas close to the glacier margins, most of the glacigenic deposits are now considered as ‘inactive areas’, the valley-train deposit is decoupled from deglaciation perturbation. Finally the continuum of paraglacial sediment fluxes that commonly exist from the (de)glaciated areas to the catchment sink is broken. Interruptions of the sedimentary cascade have previously been defined at a larger time-scale in the case of river damming by paraglacial landslides during the Holocene. However, such a long interruption due to a moraine is an original scenario: blockages due to moraines generally last for quite a short time (one or two decades), as documented in other alpine areas. The impeding influence of glacial deposits and paraglacial features (landslide mass, etc.) on sediment fluxes is highlighted: it is due to their location in headwaters, i.e. where rivers cannot remove large boulders. The efficiency of the post-LIA paraglacial reworking in alpine areas is thus questioned, because most of glacial remnants are located in such headwaters.
33More generally, paraglacial processes can be intermittent during the paraglacial period, so that the decrease in sediment yield during the paraglacial period is not only a function of the amount of available glacigenic sediments. Studies focusing on the connectivity of system elements are now required to document the different patterns of blockage of paraglacial sedimentary cascade, then the reversibility of such blockages.