Geomorphology 219 (2014) 27–41

Contents lists available at ScienceDirect

Geomorphology

j ourna l homepage: www.e lsev ie r .com/ locate /geomorph

The geomorphic evolution and sediment balance of the lower RhôneRiver (southern France) over the last 130 years: Hydropower damsversus other control factors

Mireille Provansal a,⁎, Simon Dufour b, François Sabatier a, Edward J. Anthony a,Guillaume Raccasi a, Sébastien Robresco a

a Centre Européen de Recherche et d'Enseignement des Géosciences et de l'Environnement, UMR 34, Europôle de l'Arbois, B.P.80, 13545 Aix-en-Provence Cedex 04, Franceb UMR CNRS 6554 LETG Rennes COSTEL/Université Rennes 2, place recteur Le Moal, 35000 Rennes, France

⁎ Corresponding author at: 12 rue des Saintes-MariTel.: +33 442268277, +33 620554806.

E-mail address: provansal@cerege.fr (M. Provansal).

http://dx.doi.org/10.1016/j.geomorph.2014.04.0330169-555X/© 2014 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 8 October 2013Received in revised form 21 April 2014Accepted 23 April 2014Available online 15 May 2014

Keywords:Fluvial geomorphologySecular sediment balanceRiver managementHydropower damsSource-to-sink sediment transportLower Rhône River

The geomorphic evolution of large European fluvial systems since the end of the nineteenth century has been de-termined, according to recent studies, by changes that have affected thedrainage basins, notably climatic changesat the end of the Little Ice Age (LIA), agricultural practices, and successive phases of channelmanagement involv-ing embankments and dams. However, the respective roles of these various control factors in modifying sedi-ment storage and channel stability are often difficult to identify. The aim of this paper is to determine the mainfactors that have driven morphological changes over the past 130 years on the 120-km-long downstreamreach of the Rhône River. A GIS-based analysis of maps and aerial photographs available since 1860, coupledwith chrono-stratigraphic data on the river's alluvial deposits, has enabled quantification of sediment storageand erosion in different morphological units. Changes measured include riverbed incision, narrowing of themain channel in relation with accumulation of fine sediments along the channel margins, and reduction of sed-iment input at the mouth. On the basis of local- and reach-scale sediment balances, we show that these changesare synchronous with engineering works carried out at the end of the nineteenth century and beginning of thetwentieth century, a period that coincided with a decline in river sediment inputs largely generated by humanactivities. The early stabilisation of mountain slopes resulting from a decline in rural agriculture, rural exodus, re-forestation and engineered torrent control has had a much greater impact in geomorphically transforming theRhône than dams and gravel mining, the major identified causes of river destabilisation in the world. This callsfor caution in the all-too-common and wholesale attribution of declining sediment budgets of European rivers,especially in the Mediterranean region, to dams.

© 2014 Elsevier B.V. All rights reserved.

1. Introduction: secular changes in river systems and theirdownstream impacts

The geomorphic evolution of large rivers in Europe and elsewheresince the end of the nineteenth century has been the object of severalstudies. Changes in natural and/or socioeconomic conditions havebeen identified as having had significant but variable geomorphic andsedimentological impacts. A reduction in the frequency of extreme riverdischarges at the end of the LIA (Gregory et al., 2006; Hohensinneret al., 2008) and agricultural decline in mountain catchments attendedby reforestation have been reported to have led to a decrease in sedimentsupply (Bravard, 2002; Wohl, 2006; Hoffmann et al., 2010; Hinderer,2012). Morphological changes induced in downstream reaches by these‘external’ control factors have included the disappearance of braided

es, F-13090, Aix-en-Provence.

channels to the benefit of simple channel configurations, channel-bedincision, and changes in longitudinal profiles, as well as sediment storageand afforestation of the riparian zone (Steiger et al., 2003; Rodrigues et al.,2006; Comiti et al, 2011).

In addition to these external, catchment-scale controls, direct engi-neering interventions in river channels and on banks, notably throughthe construction of embankments, groynes, andwalls have exacerbatedchannel instability and reinforced sediment storage in the fluvialmargins (Bravard and Peiry, 1993; Kiss et al., 2008; Siché and Arnaud-Fassetta, in press). Over the last 50–60 years, sediment mining hasfurther reduced the available bedload fluxes and modified channel-bedtopography, commonly engendering deep artificial pools, as has beendemonstrated by several studies in Western Europe (Petit et al., 1996),the USA (Kondolf, 1997), and Asia (Luo et al., 2007; Brunier et al., inpress). However, hydropower dams appear to stand out as the overarch-ing cause of fluvial sediment deficit through the interception and storageof much of the total sediment flux in reservoirs (e.g., Surian and Rinaldi,2003; Phillips et al., 2005; Chen et al., 2010; Wang et al., 2011; Räsänen

Fig. 1. Location, geological contexts, and relevant features of thefive studied reaches in thelower Rhône River. 1. limestone outcrops; 2. Pleistocene terraces; 3. Modern alluvial plainand deltaic wetlands; 4. Main cities; 5. Run-of-the-river dams. Reaches: Caderousse(1), Avignon (2), Vallabrègues (3), Beaucaire (4), Grand Rhône (5a), Mas Thibert (5b).

28 M. Provansal et al. / Geomorphology 219 (2014) 27–41

et al., 2012). Downstream of dams, incision of the longitudinal profileand/or armouring of the bed floor have been attributed to local deficitsin bedload, leading to channel readjustments and reactivation of bankmobility (Rinaldi, 2003; Petts and Gurnell, 2005; Rollet et al., 2013).These impacts propagate downstream at rates of 0.1 to 2 km/y, and ex-ceptionally 10 km/y (Bravard and Petit, 2000; Phillips et al., 2005;Rollet, 2007). The reduction of sediment input fromriver to coast is some-times investigated in a source-to-sink approach (Anthony and Julian,1999; Phillips et al., 2004; Syvitski et al., 2005; Milliman andFarnsworth, 2011;Wang et al., 2011), withmany deltas rendered vulner-able to erosion and sinking as a result of sediment sequestering behinddams (Ericson et al., 2006; Syvitski et al., 2009; Anthony, 2014).

Because of a lack of homogeneous data on sediment fluxes throughthe source-to-sink continuumover the long term(secular scale), thede-termination of sediment balances over these spatial and temporal scaleshas rarely been conducted. Similarly, the identification and evaluationof the role of individual variables impacting a single fluvial reach is achallenging task, given themultiplicity of these variables and their non-linear interactions (Surian and Rinaldi, 2003; Rollet et al., 2013). Thetheme of the intertwining of natural climatic changes and human im-pacts on river catchments through agricultural practices is even morepertinent in the well-watered circum-Mediterranean catchments ofwestern Europe, where geomorphic changes in the late nineteenth cen-tury, and afterwards, simply prolong a long history of such interactions(Poulos and Collins, 2002; Hooke, 2006; Kettner and Syvitski, 2009).Meanwhile, our capacity to propose efficient management and restora-tion strategies to river managers will increasingly depend on a fine un-derstanding of these interactions (Downs and Thorne, 2000; Jungwirthet al., 2002; Brierley and Fryirs, 2009; Dufour and Piégay, 2009).

The Rhône River catchment covers an area of 98,000 km2 and is thelargest circum-Mediterranean river system in Western Europe. Severalstudies have documented marked geomorphic changes, generated byexternal factors, that have affected the lower Rhônemain stem in south-ern France over the last century (Arnaud-Fassetta, 2003; Antonelli et al.,2004; Raccasi, 2008), and the upper Rhône (Bravard, 1986), as well asits main southern Alpine tributaries, notably the Drôme (Liébault andPiégay, 2002), the Ain (Rollet, 2007), and the Durance (Miramontet al., 1998; Chapuis, 2012). Amongst reported changes followingfluvialmanagement dating back to the end of the nineteenth century are chan-nel adjustments that have resulted in heterogeneous fluvial metamor-phosis from historical braiding into wandering, the latter stateinvolving simplification of channel morphology through disappearanceof the braided style, erosion of the bed, and reduction of channel width.Although the lower Rhône system may be considered convincingly asone of the best-studied fluvial systems in Europe, a global approach tothese changes and the disentangling of their underlying drivers hasnever been implemented.

The aim of this paper is to document the geomorphic changes in thelower Rhône River that have operated at various time and spatial scalesover the last 130 years in response to various control factors. Achievingthis objective requires an updated quantitative evaluation of the sedi-ment balance and geomorphic evolution over this period, and especiallyof themagnitude, thepace and the chronology of changes in the channeland the margins at this secular scale. Such an evaluation necessarily in-volves distinguishing the respective impacts of hydrological and socio-economic changes in the catchment area, and of river engineeringworks, mining activities and especially hydropower reservoir dams.This paper provides the first overall evaluation of the secular evolutionof the lower Rhône River enabling discrimination of human impactson the geomorphic evolution of this fluvial system, the secondmost im-portant in the Mediterranean after the Nile, as well as on discharge andstorage of bedload and fine sediments. In particular, the role played bydams relative to other control factors is analysed and quantified. Thestudy also throws light on the specificities of a system characterisedby a Mediterranean bioclimatic context and by mountainous relief inthe upper catchment sector.

2. Study site and environmental setting

2.1. Morphology and sedimentology

The lower reach of the Rhône River runs from the City of Orange tothe Mediterranean Sea in a wide, 120-km-long alluvial plain (Fig. 1).The main tributaries are the Ardèche River (65 m3/s) on the rightbank and especially the Durance River (190 m3/s) on the left bank, thecoarse load of which has built a large alluvial fan that modifies the lon-gitudinal slope of the Rhônemain stem (Bravard et al., 2008). Just northof the historic city of Arles (Fig. 1), the Rhône splits into two branches inits deltaic reach (the Camargue delta) to form the Grand Rhône (52 kmlong) and the Petit Rhône (70 km long), which convey, respectively, 90%and 10% of the water discharge. In this study, we will ignore the verynarrow (b100 m) and strongly engineered Petit Rhône branch, thechannel-width variations ofwhich are insignificant and difficult tomea-sure accurately. Our focus, as far as thedeltaic reach is concerned,will beon the Grand Rhône, largely dominant in the Rhône delta branchesrelative to the Petit Rhône, and expressing, with much better spatial

Fig. 2. Annual discharge of the Rhône River (1840–2007). 1 — Maximum annual discharge; 2 — 10-year mobile windows; 3 — 10-years return discharge.

29M. Provansal et al. / Geomorphology 219 (2014) 27–41

accuracy, geomorphic change related to external and human-inducedcontrol factors.

The channel width varies between 250m at the upstream end of thestudy area in Caderousse (Fig. 1), and 900 m at the mouth of the GrandRhône; and the depth ranges from 5 to 20 m. The longitudinal bedprofile decreases from 0.07 m/km upstream, from Caderousse toVallabrègues, to 0.06 m/km at Beaucaire (Fig. 1), and then drasticallyin the delta down to 0.003 m/km, contributing to a drop in specificstream power and increasing sedimentation along the alluvial margins(Bravard, 2010). The generally large overbank accumulation since theend of the nineteenth century has varied significantly in terms of loca-tion, chronology, thickness, and grain size. These aspects have beenthoroughly documented by Provansal et al. (2010, 2012), and only asummary description is given here. Deposits are coarse at the base(sands or gravels inherited from the braiding period) and overlain by2 to 5 m of fine stratified sandy or clayey silts. In the various formerbranches, currents slowed down by dense riparian forests favour fine-grained sedimentation. In contrast, incision in the active riverbed rangesfrom 3 to more than 10 m, depending on the sector. The present mouthof the Grand Rhône displays a classic morphology typical of a wave-dominatedmicrotidal river delta and is the latest of a series of changingriver mouths and their locations in the course of the last 130 years,under the strong influence of river engineering (Sabatier et al., 2006).Much of the accumulation at the mouth concerns shallow depthsdown to −20 m. Sands are dominant down to a depth of −8 m,below which silts dominate (Maillet et al., 2011).

1 PK (for kilometric point).

2.2. River liquid and solid discharge

The mean annual discharge of the Rhônemeasured at the Beaucairegauging station (1920–2007) is 1850 m3/s (100-year return interval (RI)flood N 10,300 m3/s). Following a long period of frequent large floodsduring the LIA until 1850–1860 (Pichard, 1995; HistRhône Databank,2013), high discharge ranging from 7000 to 8000 m3/s (i.e., close to the10-year RI N 8000m3/s) prevailed between 1880 and 1940–1945, follow-ed by irregular discharge up to 1990, a period characterised by low flowand moderate floods. In contrast, several relatively important (10-yearRI) floods have occurred over the last 15 years (Fig. 2).

Total suspended sediment (TSS) has been systematically measuredin Arles since the end of the nineteenth century. An automatic gaugingstation, SORA, installed in Arles, has been in operation over the last10 years. Total suspended sediment has undergone a reduction ofabout 60%, from about 18.8 Mm3/y at the end of the nineteenth century(Pardé, 1925) to 6.1 Mm3/y at present (Pont et al., 2002; Eyrolle et al.,2010), with an intermediate value of 9.1 Mm3/y in 1956–1958

calculated from archive data and corresponding to the transition be-tween the pre- and post-dam periods (IRS, 2000). The overall reductionin suspended load has, thus, been of the order of 12Mm3/y over the last130 years. The present bedload discharge has been estimated at be-tween 0.4 and 1.4 Mm3/y from dredging in Arles (Dugas, 1989), butalso from a comparison of transverse profiles in the delta (Antonelli,2002; Antonelli et al., 2004; Maillet et al., 2007) and from empiricalcalculations between Beaucaire and Arles (IRS, 2000). This representsa massive reduction in bedload from about 10–15 Mm3/y at the begin-ning of the twentieth century (Pardé, 1925; IRS, 2000). Pebbles, gravelsand sand compose the bedload from Orange, at the upstream limit ofthe study sector, to kilometric point (PK)1 292, downstream of Arles;whereas sand is present throughout the deltaic reach (Arnaud-Fassetta et al., 2003).

In addition to these efforts at establishing partial and/or grain-sizebased secular sediment budgets in the Rhône River, Sabatier et al.(2006) and Maillet et al. (2007) have attempted to derive sedimentbudgets of the delta lobe off the mouth of the Grand Rhône 3 km off-shore and down to a depth of 20 m. In agreement with the significantdecline in river sediment input highlighted in the foregoing paragraph,these authors identified, from bathymetric chart differencing, a de-crease in lobe sedimentation by a factor of 4 between 1841 and 1974.The morphology of the mouth of the Rhône and the dynamics of lobesedimentation have been described in detail by Maillet et al. (2007,2011) and Sabatier et al. (2009), respectively.

3. Secular changes in factors controlling morphogenesis of theRhône catchment

Several geomorphic, stratigraphic, andGIS-based studies have docu-mented the metamorphosis of the Rhône River and tributaries in thecourse of the last 130 years. This secular transformation is attributedto a succession of, and interactions amongst, multiple control factorsat the catchment and local scales. The main factors and their impactsare briefly described here.

3.1. Climatic and socioeconomic factors

At the end of thenineteenth century and in the course of thefirst halfof the twentieth, changes affecting the Rhône catchment led to a signif-icant reduction in sediment flux in the tributaries and the Rhône mainstem. This reduction is jointly attributed to a natural decrease in the fre-quency of major floods at the end of the LIA (Pichard, 1995; HistRhône

30 M. Provansal et al. / Geomorphology 219 (2014) 27–41

Databank, 2013), rural exodus associatedwith a decline inmountain ag-riculture, catchment reforestation, and engineered stabilisation of feed-er torrents (Sclafert, 1959; Miramont et al., 1998; Bravard, 2002). TheMediterranean–Alpine tributaries of the Rhône were important sedi-ment sources, especially the large Durance River basin (14,000 km2).A French government programme aimed at reforesting the Rhônecatchment and controlling torrents (known as Restoration des Terrainsen Montagne or RTM) has resulted in a considerable reduction insediment delivery to the Rhône. Engineered torrent control wasprogressively implemented during the second half of the nineteenth cen-tury in the southern Alps (Jorda, 1985; Miramont et al., 1998), and from1940 to 1960 in the central and northern Alps (Liébault and Piégay,2002; Piégay et al., 2004). On the Durance River, TSS, estimated at2.5 Mm3/y at the end of the nineteenth century, is presently 1.3 Mm3/y(SOGREAH et al., 2001), whereas the gravel supply decreased from 0.3to 0.02 Mm3/y over the twentieth century (Chapuis, 2012). The activechannel width in this tributary was reduced by 50% between 1898 and1936. At the same time, the decrease in the frequency of major floodsand the embankments built at the end of the nineteenth century resultedin a slow-down in bedload transfer and in storage of suspended loadunder the stabilising effect of riparian forest growth in the margins(Miramont et al., 1998; Piégay et al., 2004; Warner, 2006).

3.2. Engineering practices to improve fluvial navigation

Following the 100-year flood of 1876, continuous embankmentswere built in 1860 all along the alluvial margins of the Rhône to protectinhabitants. Between 1870 and 1920, secondary channels were cut offand a series of floodable embankments, compartments2 and groyneswere built in the main channel from the Swiss border to the sea inorder to improve navigation. These engineering works progressivelygenerated a change in channel morphology from braiding to a moresimple single-thread sinuous pattern associated with erosion of thebed and reduction of width as fine sediments became increasinglytrapped along the channel by the riparian forest (Bravard et al., 1999;Arnaud-Fassetta, 2003; Piégay et al., 2008). Since the end of the nine-teenth century, an estimated 20 to 200 Mm3 of sediments have thusbeen stored in the margins of the Rhône between Lyon (Fig. 1) andArles (Bravard and Clemens, 2008).

3.3. Hydropower dams

During the second half of the twentieth century, two types of hydro-power damswere constructed in the Rhône catchment (Fig. 3). The firsttype consists of the reservoir dams of Genissiat on the Rhônemain stemupstream, Vouglans on the Ain River, and mainly Roselend, Tignes, andSerre-Ponçon on the Alpine tributaries, constructed between 1950 and1970 by Electricité de France (EDF). Since 1950–1960, the total amountof sediment retained by the main Alpine tributary dams has been esti-mated at about 200 Mm3, i.e., 4 Mm3/y (EDF, unpublished data), ofwhich 25 Mm3 (0.5 Mm3/y) is behind the large Serre-Ponçon Damand smaller dams on the Durance River. Channel incision downstreamof these reservoir dams has been attributed to a local deficit in bedload,generating channel readjustment and armouring (Rollet et al., 2013). Inseveral dams on the Ain, Isère, and Durance tributaries, fine-grained de-posits are periodically evacuated downstream through artificially gen-erated flash flows. The EDF claims (unpublished data) that their damsthus let through suspended sediments. The last artificial flash flows inMay–June 2008 reportedly led to the evacuation of about 4 Mm3 fromthe Isère and the Durance Rivers, of which 30 to 50% was stored up-stream of Caderousse or Beaucaire (Eyrolle et al., 2012).

2 Compartment: the usual technical French word is (casier), corresponding to crossedlongitudinal and transverse dykes, which divide the borders of themain channel into sev-eral compartments. The system is different from the simple transverse dykes in the sec-ondary barred arms.

The second type consists of run-of-the-river dams, 19 of whichwereconstructed between 1947 and 1986, most achieved by 1975, on theRhône main stem by the Compagnie Nationale du Rhône (CNR),which still runs these dams. Caderousse (in operation since 1975),Avignon (since 1973), and Vallabrègue (since 1970) are the last down-streamdams of this type on the lower Rhône. A diversion canal bypassesthe natural channel and directs the flow to a hydropower plant. In thebypassed channel, the usual managed reduced discharges (50 to 400m3/s) are insufficient to transport the coarser sediments, and this hasresulted in local armouring of the bed (IRS, 2000). However, given thelimited flow capacity (2000 to 2500 m3/s) of this canal, much of thewater and sediments (both suspended and bedload) are conveyedthrough the bypassed channel during large floods. Between Lyon andArles, about 11 Mm3 of gravel bedload have been eroded over thelast 60 years (0.18 Mm3/y) in this bypassed channel (Bravard andClemens, 2008). However, this load is lower than that prior to river reg-ulation (0.4 Mm3/y, IRS, 2000). Notably hydropower dams do notmodify the liquid discharge. The CNR dams on the Rhône let throughflow at discharges exceeding 2500m3/s, whereas the EDF dams erectedon the Alpine catchments (especially the Serre-Ponçon Dam on theDurance River) only affect low to moderate floods but not major floods(SMAVD, 2003).

3.4. Gravel mining

The amount of dredged gravel after World War II, mainly between1970 and 1990, has been estimated at about 84 Mm3 (0.5.6 Mm3/y)for the entire Rhône basin (IRS, 2000; Bravard and Clemens, 2008). Be-tween Lyon and Arles, 13.5Mm3 (0.83Mm3/y) were extracted over thisperiod in the Rhônemain-stem channel, 2.2Mm3 (0.3Mm3/y) of whichconcerned the lower Rhône downstream of Caderousse. In the tribu-taries, mining activities were five times more important, mainly in thesouthern part of the catchment: 60 Mm3 (2 Mm3/y) between 1960and 1990 in the Durance, 3 Mm3 (0.06 Mm3/y) in the Cèze and theGardon (SOGREAH et al., 2001). These extractions resulted in an overalldecrease in available bedload flux (Petit et al., 1996) and contributed tobed incision and channel compartmenting, a process that restricts sed-iment through-flow (Warner, 2006).

To summarise, the sediment supply in the Rhône River has beendrastically reduced over the last 130 years by a combination of severalfactors, startingwith climatic and socioeconomic changes. The sedimenttransport system has been particularly affected over the last 50–60 years by dams and mining. The respective roles of these factors,especially dams, are a core part of our study. We need to emphasise,however, that the volume estimates quoted above, and derived fromthe grey and the published literature, incorporate large uncertainties.

4. Materials and methods

To analyse the impact of the various control factors that have influ-enced the lower Rhône system and to obtain an accurate evaluation ofthe river's sediment balance over the past 130 years, morphologicalchanges and their time frame were determined based on GIS analysisofmaps and aerial photographs. Fine-tuning of the chronological frame-work has been further obtained from sediment cores and their isotopeanalysis.

4.1. Reach delimitation

Using geological criteria and dams to delimit themorphosedimentaryreach units analysed in this study, we identified five reaches covering atotal length of 97 km (Fig. 1, Table 1). The first three contiguous reaches,Caderousse (reach 1), Avignon (reach 2), and Vallabrègues (reach 3),correspond to a variably wide alluvial plain bounded upstream anddownstream by gorges cut in outcrops of hard Cretaceous limestone.The three reaches correspond to segments of the lower Rhône separated

Fig. 3. Location of the main hydropower dams and the Marcoule nuclear plant in the Rhône River catchment. 1. Run-of-the-river dams (CNR); 2. Alpine reservoir dams (EDF); 3. Maincities; 4. Marcoule nuclear plant.

31M. Provansal et al. / Geomorphology 219 (2014) 27–41

fromeach other by CNRhydropower dams. The rocky gorge segments aredevoid of alluvial deposits and are not further considered in this study,nor the large alluvial fan of the Durance River confluence downstreamof the Avignon reach (about 8 km long). Downstream of the Vallabrèguesdam, the Beaucaire-Arles (4) and Grand-Rhône (5) reaches correspond,respectively, to the wide coastal alluvial plain downstream of the lime-stone hinterland and the delta. Within reach 5, the focus is essentiallyon theMas Thibert section (identified as reach 5a), for which a large cor-pus of unpublished data exists. We also propose a sediment balance for

Table 1Characterisation of the studied reaches.

Reaches PK Length (km) Bed Slope

1. Caderousse 210–220 10 0.07%2. Avignon 230–249 19 0.07%3. Vallabrègues 260–269 9 0.07%4. Beaucaire–Arles 272–279 7 0.06%5. Grand Rhône 279–334 52 0.003%5a. Mas Thibert subreach 302–308 6

the active lobe off the mouth of the Grand Rhône, enabling an overallsource-to-sink view of the lower Rhône.

4.2. Evolution of the channel and fluvial margins

The two-dimensional changes in the channel and the riverbanks be-tween the nonsubmersible embankments built in 1860were quantifiedusing topographic maps and orthophotographs covering the period1860 to 2008 (Table 2). The GIS processing of maps and aerial

Date of dam building Cores Isotopic data Drills (PANDA)

1975 2 21973 61970 2no 4 3 11no 1no 1 1 8

Table 2Two- and three-dimensional (topographic and depth) data.

Reaches Planimetric data (2D) Topographic data (3D) CNR dam construction

1. Caderousse 1860, 1910, 1955, 1973, 2006 1902, 1975, 2008 19752. Avignon 1860, 1905, 1953, 1973, 2004 1902, 1973, 2004 19733. Vallabrègues 1905, 1947, 1962, 1971, 2006 1876, 1970, 2006 19704. Beaucaire–Arles 1876, 1905, 1950, 2008 1876, 1950, 20085. Grand Rhône 1855, 1944, 1999 1908, 1965, 19995.1 Mas Thibert subreach 1883, 1907, 1950, 1970, 2006 1869, 1907, 1999, 2006Rhône mouth 1872, 1975, 2006

Two-dimensional data are based on archival sources: (i) the Ponts et Chaussees Atlas (1860 to 1876), the Branciard Map (1905 to 1910); (ii) aerial photographs from 1944 to 2006 pro-vided by the Institut Géographique National (IGN).Three-dimensional data are based on archival sources: (i) topographic and depthmaps (Ponts and Chaussees, 1869 to 1876), Service Spécial du Rhône (1902), and CNR three-dimensionalmaps from 1976 to 2008; (ii) IGN orthophotographs, photogrammetric analysis, and Lidar surveys (2004–06 to 2008). The CNR acquired heterogeneous and discontinuous bathymetricdata only in the channel in 1950 and 2008 in reach 4, and in 1965 and 1999 in reach 5 (Antonelli et al., 2004).

32 M. Provansal et al. / Geomorphology 219 (2014) 27–41

photographs have been based on standard methods using ArcGis 9.3software (Gurnell et al., 2003). All local and historical geodetic systemswere converted to the common and official French coordinate system(Lambert 93). The horizontal error margin of the maps following theirgeoreferencing and digitisation is estimated at ±10 m for the nine-teenth century maps and ±5 m for the more recent ones (Raccasi,2008). Artificial structures and the active channel limits were digitisedfrom all the documents. The active channel width corresponds to thechannel and the mobile bars visible during low-flow conditions, islandsand vegetated bars being excluded. In order to compare the geomorphicevolution between the different reaches and periods, the changes inpercentage of the initial surface for each period constitutes a good crite-rion for evaluating variations in width of the channel. Given that thebankmorphology is very commonly high (N2 m) and vertical, the com-parison of surface area between two dates is not affected by moderatevariations in flow discharge as all the maps and aerial photographsdepict conditions during low discharges.

The three-dimensional diachronic evolution of the channel bed andchannel margins was quantified by differencing successive digital ter-rain models (DTM) constructed from topographic and depth surveys(Table 2). Depth profiles in the channel were measured every 10 or 20m in 1876 and 1902. Between 1950–1970 and 2008, the high densityof points on profiles spaced every 100 m (CNR surveys) has enabledthe drawing of depth contours. Vertical errors associated with depthmeasurements are estimated to be between ±0.6 and ±0.2 m for theolder maps (1876–1902) and ±0.1 m for the more recent CNR-derived depths (Raccasi, 2008). Horizontal errors are similar to thosefor the two-dimensional evolution, i.e., around 10 m for earlier maps(1876–1902) and 5 m for the more recent ones.

In order to have a complete picture of source-to-sink sediment budgetchanges, accumulation at the mouth of the Rhône River was quantifiedfrom bathymetric surveys carried out by the Service Hydrographique etOcéanographique de la Marine (SHOM)3 in 1872 and 1975 and by theCoast GuardOffice in 2006, complementing the previouswork conductedby Sabatier et al. (2006) and Maillet et al. (2007). Only the 1872 data re-quired digitisation, the rest being available in digital format and directlyintegrated in a GIS. The nineteenth century SHOM data have a verticalerror range of between +0.7 and −0.2 m and ±10 m for position(Sabatier et al., 2006); the twentieth century data are more precise,with ±0.2/0.1 m for the vertical error range and ±10/0.3 m for positionin 1975 and 2006, respectively.

Once soundings and elevation points were digitised, DTMs wereconstructed from triangulated irregular networks (TINs) in ArcGis. Thediachronic evolution of the channel bed, channel margins, and deltalobe were quantified by differencing successive DTMs.

The temporal resolutions are not the same for the available two- andthree-dimensional data. Three time slices can be distinguished for theformer, each corresponding to specific engineering control factors:(i) 1880–1920, works on the channel to improve navigation along the

3 SHOM Service hydrographique et océanographique de la marine.

entire Rhône River; (ii) 1945–1970, construction of reservoir dams intheAlps andmost of the run-of-the-river dams on the upper andmiddleRhône; and (iii) 1970–1975, construction of the three downstream run-of-the-river dams on the lower Rhône. This third period concerns onlythe bypassed sectors in reaches 1, 2, and 3, as well as reaches 4 and 5where (natural) fluvial processes persist. The three-dimensional datacannot be individualised for the first two periods. Nevertheless we putforward the hypothesis that the two- and three-dimensional sedimentbalances follow the same trends in the five studied reaches. Given theheterogeneity of the data sets concerning the five reaches, two periodsof analysis are defined herewith: before and after the construction ofthe EDF and CNR dams around 1950/1970–1980, hereafter referred toas BD and AD, respectively.

4.3. Alluvial chronostratigraphy

The fine sediments (silt and sand) overlying gravel bars and banksfrom the formerly braided channel (a pattern that prevailed until theend of the nineteenth century) were analysed from drill cores at 25locations using a PANDA penetrometer and from ten 2–5 m-longcores extracted with a semiautomatic drill core (Fig. 4, Table 1). Sam-pling locations were based on diachronic GIS results (Raccasi, 2008),field observations and geomorphic classification (Antonelli, 2002;Arnaud-Fassetta et al., 2003), and recent geophysical surveys (electricresistivity tomography— ERT) in the Avignon and Beaucaire–Arles allu-vial plain (Vella et al., 2013). Grain size measurements were made bylaser diffractometry using a Beckmann Coulter LS 13 320. Textural anal-ysis was used to describe the major flood periods in conjunction withflood histories. These data are simply evoked here to enhance the dia-chronic GIS analysis but will not be discussed in detail. Details on thegeomorphic location and stratigraphic and sedimentological descrip-tions of the drill and core data are reported, respectively, in Provansalet al. (2010), Provansal et al. (2012), and Ferrand et al. (2012) for the10 cores. The five core sites selected in reaches 1, 4, 5, and 5a (Fig. 4)are representative of the various environments that could influencethe sedimentation rate and the grain size of the deposits: sandy–siltyhigh banks (cores CA 1, PI, and MT), silty low banks (FO), and compart-ments (BA).

Artificial radionuclides stored in riverbank deposits offer the possi-bility of evaluating sedimentation rates, notably 210Pbxs and 137Csisotopes, as shown by Walling and He (1997), Piégay et al. (2008),Detriché et al. (2010), amongst others. The history of nuclear releasesfrom theMarcoule nuclear plant (Fig. 3), themain source of artificial ra-dioactivity in the Rhône catchment, provides a reliable chronology overthe last 100 years (Eyrolle et al., 2005; Ferrand et al., 2012). The 137Csflux arrived with liquid releases from Marcoule in the lower Rhonefrom 1961 to the early 1990s. The 210Pbxs activity profile is widelyused to date deposits younger than 100 years. The five selected coreshighlight the generally large accumulation of fine deposits over thelast century (2 to 5 m), the thickest deposits being in reach 4, as wellas their large chronological variability.

Fig. 4. Chronostratigraphy of the studied reaches. (A) Location of cores: 1. Modern alluvial plain with channel; 2. Substratum (limestone and Pleistocene deposits); 3. Reaches; 4. Cores(symbols correspond to locations of the cores,five ofwhich are described in the text). (B) Stratigraphy, grain size, and radionuclide activity at the six selected sites (adapted from Provansalet al., 2012). Grain size (in %): 1. Clay (b2 μm); 2. Silt (2–63 μm); 3. Sand (63–600 μm). Radionuclide activity (in Bq/kg, with error bars): 6. 137Cs; 7. 210Pbxs.

33M. Provansal et al. / Geomorphology 219 (2014) 27–41

5. Results

Given the disparities in the available data, the active-channel width(two-dimensional changes) and sediment balance in the channel andits margins (three-dimensional changes) are analysed separately forthe five reaches, followed by the sediment balance produced for theactive lobe of the Rhône.

5.1. Two-dimensional evolution of the width of the active channel

5.1.1. Reach 1 (Caderousse), river PK 210–220The former braided channel disappears in part between 1860

and 1910 in conjunction with navigation engineering works (bankstabilisation and channel narrowing). The main secondary branch be-came an abandoned channel by 1955. The total active channel width re-duction has been 71m since 1860. Thewidth (Fig. 5) initially decreasedrapidly (−0.34%/y, 1860–1910), and then slowly (−0.11%/y, 1910–1955). Since 1975, following the construction of the hydropowerdams, the width of the bypassed channel (5 km long, manageddischarge 10 m3/s) has been fairly stable (−0.05%/y, 1973–2006).

5.1.2. Reach 2 (Avignon), river PK 230–249In 1860 the fluvial system was characterised by an anabranching

style with forested islands and several connected branches. The total re-duction in active channel width has been 94m since 1860. Narrowing ofthe active channel (Fig. 5) was initially rapid (−0.43%/y from 1860 to1905) in conjunction with navigation works, and then slowed down(−0.12%/y from 1905 to 1953). Since dam construction (1973), the

width of the bypassed channel (13 km long, managed discharge 400m3/s) has remained stable (−0.05%/y from 1973 to 2004).

5.1.3. Reach 3 (Vallabregues), river PK 260–269The braided morphology, still persistent in 1905, disappeared

progressively up to 1962. The total active channel-width reductionhas been 370 m since 1905. As in the previous reaches, navigationworks initially led to rapid narrowing of the active channel (Fig. 5)(−0.71%/y, 1905–1947), followed by a decrease in the rate of narrowing(−0.12%/y, 1947–1962). Since the Vallabrègues Dam was completed(1970), the width of the bypassed channel (6.5 km long, managed dis-charge 10 m3/s) has remained stable (−0.05%/y, 1971–2006).

5.1.4. Reach 4 (Beaucaire–Arles), river PK 272–279The main channel was characterised by several secondary branches

andmobile bars in 1876 and evolved to a sinuous single-thread channelin 1906 (Provansal et al., 2010). The total active channel-width reduc-tion in this reach has been 230 m since 1876. In 1950, all the banksand former branches were forested (Fig. 6A). Active channel narrowing(Fig. 5) was rapid between 1876 and 1905 (−0.85%/y) as a result ofnavigation works carried out between 1876 and 1920, and sloweddown between 1905 and 1950 (−0.06%/y). Narrowing remained anactive process throughout the second half of the twentieth century(−0.26%/y between 1950 and 2008).

5.1.5. Reach 5 (Grand Rhône downstream of Arles), river PK 280 to 332At the end of the nineteenth century, this reach was braided with

mobile sandy bars (Antonelli, 2002), and it has undergone a width

Fig. 5. Secular evolution of the active channel width, expressed as a percentage of the initial surface for each period.

34 M. Provansal et al. / Geomorphology 219 (2014) 27–41

reduction of 133m since 1855, the year the bars started disappearing. Adrastic reduction in the active channel width (Fig. 5) occurred between1855 and 1944 (−0.24%/y), followed by a slow-down and stabilisationafter 1944 (−0.03%/y).

5.1.6. Subreach 5a (Mas Thibert), river PK 302–308In 1869, this subreach was predominantly braided with several mo-

bile sandy bars. These bars disappeared progressively starting in 1876,attended by an overall channel-width reduction of 200m since. Naviga-tion works in 1905 led to narrowing of the channel and its change inplanform from straight to moderately sinuous. Changes have beenminor since 1905 (Fig. 6B). The reduction in channel width (Fig. 5), ini-tially rapid (−0.93%/y, 1883–1907), slowed (−0.52%/y, 1907–1950),and finally stabilised in 1970–2006.

5.2. Three-dimensional evolution and sediment balance

The GIS analysis enables assessment of the three-dimensional evolu-tion of the channel and themargins (Fig. 7). The volumes offine-grainedsediments stored in the margins on the top of the inherited gravel barsand those of the coarse bedload eroded on the channel floor areanalysed separately for the two periods BD and AD. The values areexpressed in Mm3 to yield the overall sediment balance, and inm3/y/km in order to compare sediment budgets of reaches at differentspatial and temporal scales.

5.2.1. Reach 1 (Caderousse)In the channel, BD corresponds to a period of mild erosion

(−0.37 Mm3, −500 m3/y/km), the process ceasing after 1975. In thefluvial margins, fine-grained deposition attained a positive balanceof +10 Mm3 (+13,200 m3/y/km) in 1902–1975, then +0.01 Mm3

(+0.6 m3/y/km) in 1975–2008 for the AD period. Core CA1 extractedon a high bank close to the channel shows 3.5 m of sandy silts overlyingan inherited pebble bar. Radioisotope dating indicates deposition of

these sandy silts over the last 100 years, excepting the last decimetre,which accumulated after 1960 (Fig. 4).

5.2.2. Reach 2 (Avignon)In the channel (Fig. 8A), erosion, prevalent during the BD period

(−1.9 Mm3, −1420 m3/y/km), slowed down between 1973 and 2004(−0.37 Mm3,−700 m3/y/km). In contrast, the BD balance of fine sedi-ments trapped along the fluvial margins in 1902–1976was very largelypositive at +30 Mm3 (+22,500 m3/y/km), but only slightly so in the1976–2004 AD period with 0.22 Mm3 (+416 m3/y/km). Six drill cores(undated) located close to the channel revealed a 0.3–1 m-thick layerof fine sediments at the upstream end of the reach (PK 232–233) andanother layer more than 5 m thick at the downstream end (PK 239).

5.2.3. Reach 3 (Vallabrègues)In themain channel, BD erosion, initially strong (−7.8Mm3,−9000

m3/y/km), decreased in the 1970–2006 AD period to −0.77 Mm3

(−3030 m3/y/km). In the margins, the fine sediment balance was ofthe order of +3 Mm3 (+5830 m3/y/km) in the 1876–1970 BD period,then +0.75 Mm3 (+3630 m3/y/km) in 1970–2006. Two (undated)cores confirm the variable thickness of these fine deposits: 1.5 m atthe top of a former gravel bar to over 6.5 m in a former channel.

5.2.4. Reach 4 (Beaucaire–Arles)In the channel, erosion occurred to the tune of −3.3 Mm3 (−6300

m3/y/km) during the BD period, increasing strongly after 1950(−9.4 Mm3, −18,800 m3/y/km). The margins had a positive balanceof +6.7 Mm3 (+12,900 m3/y/km) for the BD period of 1876–1950that nearly doubled to +12.8 Mm3 (+31,500 m3/y/km) in 1950–2008 (Fig. 8B). The sediment succession in two cores (PI and FO,Fig. 4) shows accumulation of 4 to 5 m of fine sediments since the endof the nineteenth century, deposition being greater on the higher bank(PI). Radioisotope dating shows an increase in the rate of depositionsince 1960 (1.7 to 5m in cores FOand PI respectively) and thus confirmsthe results of the GIS analysis.

Fig. 6. Two-dimensional geomorphic evolution of the channel and fluvial margins during the BD and AD periods in reaches 4 and 5a. (A) Reach 4 (Beaucaire–Arles); (B) Subreach 5a (MasThibert). 1. Channel; 2. Afforested banks; 3. Mobile silt-sandy or gravelly banks and bars; 4. Alluvial plain (farmland); 5. Built-up area; 6. Unfloodable embankments (erected in 1860); 7.Submersible longitudinal embankments (erected between 1870 and 1880); 9. Groynes. Compartments consist of submersibles embankments associated with groynes.

35M. Provansal et al. / Geomorphology 219 (2014) 27–41

5.2.5. Reach 5 (Grand Rhône)Because of the absence of data on bank sedimentation, determining

the balance of fine sediments in this reach has not been possible.Instead, core BA extracted in a compartment located south of Arlesshowsmore than 3.8mof fine sediments, the top 1.6mhaving been de-posited after 1960 (Fig. 4). Channel erosion volumes covering the twen-tieth century, andmeasured in a riffle and pool sequence (river PK 292–296 and 306–314 respectively; Antonelli et al., 2004), were extrapolat-ed to the set of pools and riffles in the entire reach. Antonelli et al.(2004) showed that the channel bed was relatively stable over the riffle(−150m3/y/km). In the pool, erosionwas estimated at−15,000m3/y/kmbefore 1965, and−6000m3/y/km from1965 to 1999. Extrapolationof these results indicates negative sediment balances of −20–26 Mm3

corresponding roughly to the BD period before 1960–1965 andbetween −8 and −14.4 Mm3 for the AD period. The riffles appear tobe relatively stable, andwe consider that this extrapolation yields an ac-ceptable sediment balance.

5.2.6. Mas Thibert subreach (5a)In the channel, the strong erosion during the BD period

(−5.1 Mm3, −14,700 m3/y/km) decreased after 1965 (−0.42 Mm3,−5800 m3/y/km), in good correlation with the sediment balance

estimates of the entire Grand Rhône reach. The balance in thefluvialmar-gins was +0.77 Mm3 (+1575 m3/y/km) in 1907–1965, but almost evenat+0.01 (+175m3/y/km) in 1965–1999. CoreMT,which has a length of3 m, recorded only atmospheric global fallout before 1960 (Fig. 4).

5.2.7. Delta lobeDeposition on the delta lobe off the mouth of the Grand Rhône

amounted to about +3.4 M m3/y between 1872 and 1975 andabout +1.2 Mm3/y between 1975 and 2006. These values are ofthe same order of magnitude as those proposed by Sabatier et al.(2006, 2009) for the period 1895–1975. Our results clearly confirmthe drastic reduction in sediment input from the Rhône to the seaby a factor of 2.8 before and after dam construction.

5.3. Synthesis on channel-width changes and sediment balance

In all the studied reaches, the highest rates of channel narrowing oc-curred at the end of the nineteenth century and beginning of the twen-tieth century. This narrowing was concomitant with managementworks designed to improve navigation. Beginningwithin a period rang-ing from 1870 to 1910, the rate of narrowing was very rapid and thendecreased between 1910 and 1950, becoming negligible after. After

Fig. 7. Erosion/accumulation and global sediment balances within each reach for the BD and AD periods. (A) Accumulation (in Mm3); (B) Erosion (in Mm3); (C) Relative accumulation(in Mm3/y/km); (D) Relative erosion (in Mm3/y/km).

36 M. Provansal et al. / Geomorphology 219 (2014) 27–41

the construction of the CNR dams in the 1970s, channel narrowingslowed down in reaches 4 and 5 and was insignificant or nil in reaches1, 2, and 3.

The fine-grained sediment balance within the channel marginsthroughout the studied reaches (Fig. 7A) is approximately +64.3 Mm3

(+0.1 to +0.7 Mm3/y), corresponding to +50.5.7 Mm3 (+0.5to +0.8 Mm3/y) for the BD period and +13.7 Mm3 (+0.4 Mm3/y), forthe AD period. This balance is almost certainly underestimated becauseof the incomplete data on distal bank sedimentation in reaches 1 to3 and the absence of detailed data in reach 5, except subreach 5a.Nevertheless these results propose orders of magnitude never publishedbefore for the lower Rhone. By extrapolation, we estimate a balance ofabout +198 Mm3 (+1.6 Mm3/y) of fine sediments stored over thesame period in the entire 300-km Lyon to the Mediterranean Sea Rhônereach (upstream of Lyon, the glacial geomorphic heritagemodifies alluvi-al storage; Bravard et al., 2008). This corresponds to the upper value pro-posed by CNR (+20 to +200 Mm3; Bravard and Clemens, 2008).

Regarding the channel itself, the sediment balance concerns mainlygravels and pebbles (D50 = 70 mm in reaches 1 to 3 and 15 mm inArles) down to the Terrin riffle in reach 5 (river PK 292–296) (Fig. 1),and coarse to medium sands downstream of this riffle to the mouth(Arnaud-Fassetta et al., 2003). The total balance over the study period(Fig. 7B) is between −51.4 and −68.4 Mm3, depending on the valuesused for the Grand Rhône reach, i.e., −33.39 to −37.39 Mm3 for theBD period, and−18.4 to−24.9Mm3 for the AD period. In consequence,

gravel mining over the last 20 years in the lower Rhône (2.2 Mm3) hashad little impact on this balance (IRS, 2000).

At the reach scale, the sediment balance is highly variable. The BDperiod in reaches 1 and 2 was characterised by a net positive balance(respectively, +9.63 and +28.1 Mm3), whereas reach 3 underwentnet erosion (−4.8 Mm3). The sediment balance in reaches 1 to 3 inthe AD period has been close to zero, whereas that of reach 4 has beenidentical before and after the construction of Vallabrègues Dam(+3.4 Mm3), the AD balance maintained by stronger erosion after1950 (−9.4 Mm3). Finally, the sediment balance of the Mas Thibertsubreach has been negative for the BD and the AD periods (−3.93 Mm3

and−2.17 Mm3, respectively).Overall, the sediment balance (accumulation–erosion) indi-

cates that the lower Rhône was characterised by net storage(+13.08 to +17.1 Mm3) during the BD period and net loss(−4.64 to −11.2 Mm3) during the AD period.

6. Interpretation and discussion

The foregoing results show significant changes in river channelmor-phology and sediment balance that can be broadly defined in termsof the pre-hydropower dam (BD) and post-hydropower dam (AD)periods. This enables an evaluation of the role of these structures vis-a-vis other control factors in affecting the sediment balance of thelower Rhône down to the mouth and shoreface.

Fig. 8. Three-dimensional evolution of the channel during the BD and AD periods. (A) Reach 2 (Avignon); (B) Reach 4 (Beaucaire–Arles). 1. CNR diversion canal in reach 2.

37M. Provansal et al. / Geomorphology 219 (2014) 27–41

The discussion will successively examine the following points:(i) pre-dam geomorphic change/sediment balance variations anddrivers of fluvial change; (ii) the impacts of dams and gravel miningin the course of the last 50 to 60 years; (iii) differences in geomorphicchange/sediment balance variations amongst the reaches; (iv) thedelta mouth sediment balance, and (v) a comparison of the RhôneRiver with other human-modified river systems.

6.1. Pre-dam geomorphic change/sediment balance variations and driversof fluvial change

The most significant feature of the BD period was the channel meta-morphosis that occurred in the lower Rhône between the end of thenineteenth century and the early twentieth century. This process wascharacterised by a shift in channel pattern from braiding to single-thread, accompanied by reduction in the width of the active channel,vertical and lateral accretion of the channel margins, and bed erosion.Within the five studied reaches, much of the geomorphic change oc-curred prior to 1950. Channel narrowing was two-thirds complete bythe 1920–1940s. This evolution occurred several decades prior to theconstruction of hydropower dams throughout the Rhône main stemand tributaries and prior to gravel mining in the Rhône and its tribu-taries. The rate of channel narrowing and the three-dimensionalchanges diminished and virtually ceased after 1947–1950 as the fluvialsystem approached a new equilibrium state (except in reach 4). Thisearly trend agrees with the results of previous studies (Arnaud-Fassetta,2003; Antonelli et al., 2004). Geomorphic evolution mainly resultedfrom two control factors that acted simultaneously: (i) an overall reduc-tion in sediment input from the catchment, and (ii) the joint influence

of engineering infrastructure designed to improve river navigation andthe spontaneous afforestation of the alluvial margins.

The reduction in suspended sediment flux since the end of the nine-teenth century highlighted by measurements in Arles and furtherconfirmed by the reduction in sedimentation at the mouth of theRhône (cf. 6.4 infra) may be attributed to several factors. By extrapola-tion to the 300 km reach of the Rhône valley from Lyon to the Mediter-ranean Sea, the storage of fine sediments in the alluvial margins of thelower Rhône during the BD period amounted to about 1.2 Mm3/y. Thisis a negligible quantity, being b10–12% of the total solid load in Arlesover the period (18.8, then 9.1 Mm3/y in the middle of the twentiethcentury) and, taken alone, does not suffice to explain the progressiveload diminution. We therefore propose the hypothesis that reforesta-tion and torrent control, which have very efficiently reduced potentialerosion and favoured sediment retention in the catchment since theend of the nineteenth century, have played a decisive role in the declinein suspended load. Thismechanism, characteristic of EuropeanMediter-ranean rivers (Liquette et al., 2009; Gracia-Ruiz, 2010), is commonlyneglected because of the difficulty of compiling relevant data on theeffects of reforestation, unlike the much better documented effects ofdam retention.

As far as the Rhône channel itself is concerned, the engineering in-stallations between the 1880s and 1920s, well before dams were con-structed, were aimed at producing a single-thread channel that wouldself-dredge its bed. This artificial deepening of the channel furtherstrongly enhanced the natural erosion occurring in response to a sedi-ment deficit. Despite the decrease in specific stream power (Bravard,2010), the amplitude and speed of channel degradation can be ex-plained by the high critical shear stresses coupled with a bedload finingtrend (decrease in grain size from Caderousse to the delta; Arnaud-

38 M. Provansal et al. / Geomorphology 219 (2014) 27–41

Fassetta et al., 2003). The reduction in large floods since the end of thenineteenth century (Pichard, 1995) enabled vegetation to colonise thefluvial margins. As vegetation grew dense, it increasingly favoured thetrapping of fines, therefore favouring growth of the size and elevationof banks while also stabilising bars, as demonstrated by GIS analysis ofreach 4 (Fig. 8B). Afforestation of the alluvial margins occurred quiteearly, as in reach 4 and subreach 5a, between 1906 and 1950 in the for-mer, and as early as in the latter (Fig. 6). The strong chronological coin-cidence of these changes with engineering works and the speed of theensuing morphological changes seem to indicate that these workshave been the main driver of these changes.

6.2. The impacts of dams and gravel mining in the course of the last50 years

As far as the AD period is concerned, the overall storage in Alpinereservoir dams (fine and coarse sediment) at the catchment scale is es-timated at about 200 Mm3, i.e., 4 Mm3/y (EDF unpublished data). Thisstorage concerns the entire grain size spectrum. The proportion ofcoarse sediments is likely to be high in the middle and upper reachesof the Rhône and their feeder catchments. No studies have beenconducted on this aspect of the Rhône basin, but over the last 10–20 years much of the temporarily stored fine sediments are evacuatedby artificial flash flows (Eyrolle et al., 2012). This is an important aspectas far as the Rhône sediment balance is concerned, even if pre-dam landuse changes and reforestation in the hitherto strongly eroded Mediter-ranean mountains had already significantly reduced the sediment thatcould potentially be stored by dams. The removal (mainly of coarse sed-iments) caused bymining activities between 1970 and 1990 has furtherexacerbated these effects. The estimated total extracted volume is about13.5Mm3 (IRS, 2000), i.e., 0.83Mm3/y from the Rhônemain stem (fromLyon to the sea) and 70.54 Mm3, i.e., 3.5 Mm3/y from tributaries, theMediterranean tributaries (mainly the Durance) having been the mostaffected (63 Mm3, i.e., 2.06Mm3/y). The combined total amount of sed-iment trapped by reservoir dams and extracted by mining representsonly a small fraction of the overall reduction in suspended load(−12 Mm3/y) and in bedload (about 10–15 Mm3/y).

Variations in river liquid discharge, especially the frequency of floodsexceeding 8000 m3/s, play a decisive role in solid transport (Pont et al.,2002). Floods of this magnitude (Fig. 2) were important in the course ofthe first half of the twentieth century, but the AD period coincidedwithone of several decades of weak floods (1960–1990). Since 1993, theRhône liquid discharge has risen once again to high levels comparableto those of the end of the nineteenth century. The meanmaximum dis-charge was 6217 m3/s before the transitional upstream dam-buildingperiod, and 6220 m3/s after. The secular decreasing trend in suspendedload cannot therefore be attributed to a change in liquid discharge.

In addition to the aforementioned factors, a lag effect also resultsfrom bedload time travel from the upstream catchment to the lowerRhône reaches. Considering a transfer speed of 0.1 to 1 km/y in theupper Rhône or tributaries (Petit et al., 1996; Rollet, 2007), the impacton the lower Rhône River has probably been weak. This is not the casefarther upstream in the middle and upper Rhône or in the tributaries(i.e., 50 to 300 km upstream of the lower Rhône) where the sedimentdeficit caused by reservoir trapping favoured a variety of processesthat included bed remobilisation, bed incision and/or armouring, andreactivation of bank mobility (Rollet, 2007; Liébault et al., 2008;Chapuis, 2012). On the Rhône main stem, bedload transfer has alsobeen irregular, disrupted by local discontinuities (e.g., gorges) and theoverall decrease in specific stream power, as clearly demonstrated byBravard (2010). The geomorphic impacts expected from Alpine reser-voir dams and gravel mining have, thus, not really affected the lowerRhône owing to the inertia that characterises the downstream transla-tion of these impacts.

The story is a bit different for the run-of-the-river dams constructedby CNR, and which do not constitute obstacles during floods. About

0.5 Mm3/y would have been stored along the banks in the bypassedreaches in the Rhône valley, from Lyon to the sea, over the last 50–60 years, also a negligible quantity (5 to 8%) relative to the total fluxmeasured in Arles (9.1 to 6.1 Mm3/y) over the same period. However,bedload cannot pass over these dams, nor attain the bypassed channel(IRS, 2000; Bravard and Clemens, 2008). The channel morphology ofthe bypassed reaches 1, 2, 3, associated with the Caderousse, Avignon(Fig. 8A) and Vallabrègues Dams, is practically fixed due to the reducedmanaged discharge. The overall AD stability of these three reaches hasbeen further reinforced by the reduction in sediment input and inlarge floods over the period 1960–1990.

In contrast, reach 4 downstream of Vallabrègues Dam (Fig. 8B) hasbeen characterised by significant bed incision (−9.4 Mm3) and strongrecent accumulation in the margins (+12.8 Mm3). This trend disap-pears 10 km downstream, in the Grand Rhône reach (5) (Antonelliet al., 2004). Significant accumulation of sand and silt on the top of thebank in reach 4 may be due to flushing of trapped sediments throughthe dams during floods. Isotope dating (Fig. 4) has confirmed thatthese deposits are associated with large floods (50-to-100 yr RI floodevents) that occurred between 1993 and 2003 (Provansal et al., 2010).In reach 5 (Grand Rhône), bedload mobility is fastest downstream ofthe Terrin riffle (Fig. 1) from where sand has been readily transporteddownstream to generate the positive sediment balance at the mouth(5.1 Mm3, Fig. 7B).

6.3. Differences in geomorphic change/sediment balance variationsamongst the reaches

In addition to the impacts of changes at the catchment scale, of riverengineering, and of dams and mining, differences in sediment budgetsamongst the five reaches (Fig. 7C and D) can also be explained interms of lithological and structural controls that impacted the local flu-vial dynamics. Erosion of the alluvial bed has been lower in reaches 1and 2, where outcrops of limestone in the gorge limit such erosion,than in reaches 3 to 5 farther downstream where the channel cutsthrough soft Holocene deposits (Fig. 1). The GIS analysis shows the ef-fects of local discontinuities in reach 2 (Fig. 8A, 1902–1976 before thedam building), where bedload eroded upstream is blocked along thehard limestone outcrops of Avignon. Moreover, the large alluvial fanformed at the confluence with the Durance has created a local baselevel that reduces the bed slope of the Rhône River thereby weakeningits erosive capacity in reaches 1 and 2. Sediment accumulation hasalso been uneven. Accumulation in reaches 1 and 2 has been related re-spectively to channel narrowing at the downstream gorge and to theobstacle created by the alluvial fan at the confluence with the Durance.This retention of sediment explains the reduction in accumulationfurther downstream in reach 3.

6.4. The delta-lobe sediment balance

Sediment accumulation at the mouth occurred to the tuneof +3.4 Mm3/y prior to 1970, and +1.2 Mm3/y since then, i.e.,a more than 50% drop. The sandy fraction, which represents theupper 8 m of sedimentation, diminishes in the same proportions, from1.7 to 0.5 Mm3/y. These variations clearly highlight the strong link be-tween changes in the lower Rhône sediment budget and river-mouthsedimentation, with important implications in terms of secular deltaicgeomorphic change. The volume of coarse sediments eroded inthe lower Rhône during the pre-dam period was between 51.7 and64.1Mm3, depending on estimates for reach 5 (Fig. 7B). Budget calcula-tions (IRS, 2000) and dredging conducted by the CNR upstream of Arleshave shown that the rate of bedload displacement is b 0.2 Mm3/y, at aspeed of 0.2 km/y. This value is consistent with estimates proposed byPetit et al. (1996) and Bravard and Petit (2000). In the delta (reach 5),the transport of coarse sediments is limited due to the sharp reductionin slope. Pebbles and gravels were never transported beyond the Terrin

39M. Provansal et al. / Geomorphology 219 (2014) 27–41

riffle at PK 292–296 (Antonelli, 2002; Provansal et al., 2003) either inthe course of the last 130 years or during geological times (Vella et al.,2005). Because bedload is dominantly composed of gravels and pebblesupstream of the delta, we hypothesise that the coarse sediment erodedin reaches 1 to 4, more than 50–70 km upstream of the mouth, has notbeen transported down to the rivermouth, where only sands andmudsare present. We may therefore pursue this reasoning by consideringthat bedload extractions, all of which have been carried out upstreamof the Terrin riffle in the Rhône main stem and its tributaries, have notdirectly contributed to the reduction in sediment supply to the coast.

Consequently, bedload input to the mouth of the Rhône is largelyprovided by channel erosion in the last 36 kmdownstream of the Terrinriffle. Dense bathymetric surveys close to the mouth highlight the mo-bility of hydraulic sand dunes during flood discharges exceeding 2900m3/s (Vassas et al., 2007), providing evidence for sandy bedload trans-port down to the mouth. From these observations, we infer that thebed material eroded and transferred to the river mouth downstreamof the Terrin riffle is about 0.52 and 0.08 Mm3/y, prior to and afterdam building, respectively. These values are very low compared to thesediment accumulation at the mouth, thus suggesting that suspensionload constitutes the main agent of mouth accretion.

Shoreface sedimentation off themouth of theGrandRhône (Sabatieret al., 2006; Fanget et al., 2012) provides evidence for the progressivedecline in suspension load measured in Arles since the beginning ofthe twentieth century. This decrease (mean value of about 12 Mm3/y,i.e., 60% of the initial volume) is proportionally comparable to the dim-inution in the volume of deposits off the mouth (about 50%), thusunderscoring the strong relationship between the Rhône suspensionload and river-mouth sedimentation.

6.5. Comparison of the Rhône River with other human-modified riversystems

The lower Rhône is located downstream of a large catchment, par-tially influenced by Mediterranean climate, in a historical context of re-forestation and climatic changes characteristic of the mountainousEuropean countries since the end of the nineteenth century. In someMediterranean fluvial systems, lateral geomorphic mobility continuedat least during the course of the first half of the twentieth century andmuch of the geomorphic modification occurred after World War II, ashas been reported for the Ebro (Vericat and Batalla, 2005; Ollero,2010), the Tagus (Uribelarrea et al., 2003), and the Toscan rivers inItaly (Rinaldi, 2003), but also for non-Mediterranean rivers such as theSacramento in California (Michalkova et al., 2010; Micheli and Larsen,2010). This post-World War II period has been characterised by theimpacts of hydropower dams, especially given the fact that water indam reservoirs is also used for irrigation in the relatively dryMediterra-nean climatic context compared to more humid climates. Furtherimpacts have been generated by intense riverbed mining, as in the Po(Simeoni and Corbau, 2009) and in smaller coastal rivers of northeast-ern Spain (Rovira et al., 2005). By reducing sediment supply to thecoast, reservoir storage, estimated at more than 95% of the river loadfor the Nile (Stanley, 1996) and the Ebro (Guillen and Palanques,1992), has indirectly resulted in delta retreat and coastal erosion. Weshould stress the fact, however, that most of these studies have identi-fied neither the effects of local river management nor the timing ofthe propagation of the sediment deficit (Batalla and Kondolf, 2005;Ollero, 2010).

On other large European rivers such as the Rhine and the Danube,important observed changes occurred as early as in the lower Rhône(i.e., end of the nineteenth-beginning of the twentieth centuries,Hohensinner et al., 2004). On the lower Danube and delta, the relativelysmall reduction in sediment discharge (25–30%) following damming(the Iron–Gate Dams, 1884 and 1970) has been attributed to input ofdownstream tributaries (Salit et al., in press) and to increased erosionof the river bottom, banks, and bars in the lower course of the river

amplified by meander cutoffs in distributaries in the delta in the1990s (Jugaru Tiron et al., 2009).

Data of the type collated here for the Rhône are muchmore exhaus-tive than formany of the large rivers in Asia that are now iconic in termsof the effects of dams on their sediment flux. Wang et al. (2011) deter-mined sediment budgets for the five largest rivers in east and SoutheastAsia (Yellow, Yangtze, Pearl, Red, and Mekong), known to be importantcontributors of terrigenous sediment to thewestern Pacific Ocean. Timeseries of data on annual water discharges and sediment fluxes of theserivers covering the period 1950–2008 indicate that anthropogeniccauses involving dams and land use have been responsible for long-term (decadal scale) decreases in sediment flux to the ocean (Wanget al., 2011).

The nature and intensity of geomorphic and sedimentary impactsengendered by drivers of environmental change in human-modifiedriver systems are thus variable. They depend on catchment size and ge-ology, the history of land occupation and use, as well as on the climaticcontext, the chronology and degree of river management, and finally, atthe reach-scale, on process and sedimentary connectivity betweenneighbouring reaches, as has been shown here for the lower Rhône.This downstream part of the river combines impacts common notonly to large European rivers that have been heavily engineered in thenineteenth century, notably the Danube and the Rhine, but also tomany of the Mediterranean rivers on the more humid westernEuropean margins where early rural exodus and stabilisation oftorrents andmountain slopes occurred over the same period. Dischargeregulation and sediment trapping by dams, as well as aggregate extrac-tions since the end of the World War II, are factors in the geomorphictransformation of the lower Rhône but not determining ones. This orig-inal combination of pre- and post-dam impacts has also generated atime and space lag in geomorphic change between the Mediterraneantributaries, which are important sediment purveyors, and the Rhônemain stem.

7. Conclusion

Changes in the sediment balance of the lower Rhône River andits consequences on channel morphology have been discernible sincethe end of the nineteenth century and were most prominent in the be-ginning of the twentieth century. This early evolution was caused bytwo simultaneous control factors with combined effects. Engineeringworks on the river, aimed at improving navigation, led rapidly toanthropogenic fluvial metamorphosis, characterised by significantchannel narrowing, bed incision, and storage of fine sediment in thealluvial margins. This transformation was further reinforced by sponta-neous riparian afforestation. The impact of river engineeringworks wasexacerbated by an overall reduction in sediment inputs from the drain-age basin associated with land use changes that likely prevailed in theSouthern Alps from the end of the nineteenth century. These includedmassive rural exodus, a decline inmountain agriculture, and engineeredtorrent control followed by systematic reforestation. An important conse-quence of the changes that occurred on the lower Rhônewas a significantreduction in sediment delivery to the river mouth, with consequences onthe stability of the Rhône delta and adjacent shorelines.

Alpine reservoir dams andmining activities mainly in the tributariesover the last 50–60 years have had weaker impacts. The sources of sed-iment, torrential in origin, had already been exhausted before damswere constructed within the Alpine basins. Downstream transmissionof the geomorphic impacts of the reservoir damswas slowwithin an al-ready stabilised catchment. A reduction in the frequency of large floodsbetween 1960 and 1990 has further downplayed these impacts. Onthe main stem river, the cascade of CNR hydropower dams assuredstabilisation of the perturbed system by engendering segmentation ofbedload transfer. Only downstream of the last hydropower dam wasequilibrium not achieved in the second half of the twentieth century,such that the transfer of sandy bedload has persisted, assuring

40 M. Provansal et al. / Geomorphology 219 (2014) 27–41

maintenance of sand supply, albeit in diminished quantities, to theRhône delta. The lower Rhône River had already undergone significantlocal engineering transformations prior to the downstream propagationof the effects of these upstream dam installations.

Finally, the detailed quantitative morphological and sediment-balance analysis in space and time conducted in this study has enableda clear hierarchy of control factors on the lower Rhône River within thesource-to-sink fluvial continuum. Recognition of this point is funda-mental in terms of sustainable management of fluvial systems. It per-mits identification of different courses of action that can be taken inorder to improvemanagement strategies relating to economic activities,flood risks, and ecology (Downs and Thorne, 2000; Dufour et al., 2008;Glenn et al., 2008).

Acknowledgements

The authors acknowledge the input of several Masters students onthe geomorphic evolution of the lower Rhône River. This work expandson previously published studies and includes results from several un-publishedMasters and Ph.D. studies conducted at Aix-Marseille Univer-sity. Thanks to P. Pentsch and N. André-Poyaud for their help in drawingthe figures. Our thanks to Michal Tal and to four anonymous reviewersfor their constructive reviews. Editor-in-Chief Richard Marston is alsothanked for his salient constructive remarks on the revisedmanuscript.This work was supported by the French National Research Agency(ANR) through the ANR807-EXTREMA project (contract # ANR-06-VULN-005, 2007–2011, Vulnérabilité, milieux et climat) and by theresearch cluster of the Observatoire des Sediments du Rhone (OSR).

References

Anthony, E.J., 2014. Chapter 13 — Deltas. In: Masselink, M., Gehrels, R. (Eds.), Coastal En-vironments and Global Change. John Wiley and Sons, pp. 299–337.

Anthony, E.J., Julian, M., 1999. Source-to-sink sediment transfers, environmental engi-neering and hazard mitigation in the steep Var River catchment, French Riviera,southeastern France. Geomorphology 31, 337–354.

Antonelli, C., 2002. Flux sédimentaires et morphogénèse récente dans le chenal du Rhôneaval. Ph.D. Thesis , Aix-Marseille Univ., Aix-en-Provence, France.

Antonelli, C., Provansal, M., Vella, C., 2004. Recent morphological changes of a channel in adeltaïc environment. The case of the Rhône River, France. Geomorphology 57,385–402.

Arnaud-Fassetta, G., 2003. River channel changes in the Rhône Delta (France) since theend of the Little Ice Age; geomorphological adjustment to hydroclimatic changeand natural resource management. Catena 51 (2), 141–172.

Arnaud-Fassetta, G., Quisserne, D., Antonelli, C., 2003. Downstream grain-size distributionof superficial bed material and its hydro-geomorphological significance in a large andregulated river: the Rhône River in its delta area (France). Géomorphologie: Relief,Processus, Environnement, 1 pp. 33–50.

Batalla, R.J., Kondolf, G.M., 2005. Hydrological effects of dams and water diversions onrivers of Mediterranean-climate regions: examples from California. In: Garcia, C.,Batalla, R.J. (Eds.), Catchment Dynamics and River Processes: Mediterranean andOther Climate Regions. Elsevier, Amsterdam, pp. 197–211.

Bravard, J.P., 1986. Etude d'un hydrosystème fluvial, le Haut-Rhône français. In: CNRS (Ed.),Lyon, France.

Bravard, J.P., 2002. The adjustments of fluvial systems to the decrease in water andsediment fluxes following mountain reafforestation. La Houille Blanche-RevueInternationale de l'eau, 3 pp. 68–71.

Bravard, J.P., 2010. Discontinuities in braided patterns: the River Rhône from Geneva tothe Camargue delta before river training. Geomorphology 117, 219–233.

Bravard, J.P., Clemens, A., 2008. Le Rhône en 100 questions. In: GRAIE (Ed.), Lyon, France(295 pp.).

Bravard, J.P., Peiry, J.L., 1993. La disparition du tressage fluvial dans les Alpes françaisessous l'effet de l'aménagement des cours d'eau (19–20e siècle). Z. Geomorphol. 88,67–79.

Bravard, J.P., Petit, F., 2000. Les cours d'eau. Dynamique du système fluvial. , Armand-Colin, Paris, France (222 pp.).

Bravard, J.P., Landon, N., Peiry, J.L., Piégay, H., 1999. Principles of engineering geomorphol-ogy for managing channel erosion and bedload transport, examples from Frenchrivers. Geomorphology 31, 291–311.

Bravard, J.-P., Provansal, M., Arnaud-Fassetta, G., Chabert, S., Gaydou, P., Dufour, S.,Richard, F., Valleteau, S., Melun, G., Passy, P., 2008. Un atlas du paléoenvironnementde la plaine alluviale du Rhône, de la frontière suisse à la mer, EDYTEM. Cahiers dePaléoenvironnement, 6 pp. 101–116.

Brierley, G., Fryirs, K., 2009. Don't fight the site: three geomorphic considerations incatchment-scale river rehabilitation planning. Environ. Manag. 43, 1201–1218.

Brunier, G., Anthony, E.J., Goichot, M., Provansal, M., Dussouillez, P., 2014. Recentmorphological changes in the Mekong and Bassac river channels, Mekong Delta:the strong impact of riverbed mining, and implications for delta destabilisation.Geomorphology (in press).

Chapuis, M., 2012. Mobilité des sédiments fluviaux grossiers dans les systèmes fortementanthropisés: éléments pour la gestion de la basse vallée de la Durance. Ph.D. Thesis ,Aix-Marseille Univ., Aix-en-Provence, France.

Chen, Z., Wang, Z., Finlayson, B., Chen, J., Yin, D., 2010. Implications of flow control by theThree Gorges Dam on sediment and channel dynamics of the middle Yangtze(Changjiang) River, China. Geology 11, 1043–1046.

Comiti, F., Da Canal, M., Surian, N., Mao, L., Picco, L., Lenzi, M.A., 2011. Channel adjust-ments and vegetation cover dynamics in a large gravel bed river over the last200 years. Geomorphology 125, 147–159.

Detriché, S., Rodriguez, S., Macaire, J.J., Bonté, Ph., Bréhéret, J.G., Bakyono, J.P., Jugé, Ph.,2010. Caesium 137 in sandy sediments of the River Loire (France): assessment ofan alluvial island evolving over the last 50 years. Geomorphology 115, 11–22.

Downs, P.W., Thorne, C.R., 2000. Rehabilitation of a lowland river: reconciling flooddefence with habitat diversity and geomorphological sustainability. J. Environ.Manag. 58, 249–268.

Dufour, S., Piégay, H., 2009. From the myth of a lost paradise to targeted river restora-tion: forget natural references and focus on human benefits. River Res. Appl. 25,568–581.

Dufour, S., Provansal, M., Raccasi, G., 2008. Can flood risk management restore ecogeo-morphologic functioning of fluvial margins along the lower Rhone River(France SE)? 4th ECRR Conference on River Restoration, Italy, Venice S. Servolo Is-land, pp. 349–356.

Dugas, F., 1989. Etude du transit sédimentaire par charriage dans le Rhône entreBeaucaire et Arles. Technical report. Institut Français de Recherche Scientifiquepour le Développement en Coopération et Compagnie Nationale du Rhône, Montpel-lier, France.

Ericson, J., Vörösmarty, C.J., Dingman, S.L., Ward, L.G., Meybeck, M., 2006. Effective sea-level rise in deltas: causes of change and human dimension implications. Glob. Planet.Chang. 50, 63–82.

Eyrolle, F., Louvat, D., Metivier, J.M., Rolland, B., 2005. Origins and levels of artificial radio-nuclides within the Rhône River waters (France) for the last forty years: towards anevaluation of the radioecological sensitivity of river systems. Radioprotection 40 (4),435–446.

Eyrolle, F., Antonelli, C., Raimbault, P., Boullier, V., Arnaud, M., 2010. SORA: a high frequen-cy flux monitoring station at the lower Rhône River. 39th International Congress ofCIESM — the Mediterranean Science Commission. Venice, Italy.

Eyrolle, F., Radakovitch, O., Raimbault, P., Charmasson, S., Antonelli, C., Ferrand, E., Aubert,D., Raccasi, G., Jacquet, S., Gurriaran, R., 2012. Consequences of hydrological events onthe delivery of suspended sediment and associated radionuclides from the RhôneRiver to the Mediterranean Sea. J. Soils Sediments 12, 1479–1495.

Fanget, A.S., Bassetti, M.A., Arnaud, M., Chiffoleau, J.F., Cossa, D., Goineau, A., et al., 2012.Historical evolution and extreme climate events during the last 400 years onthe Rhone prodelta (NW Mediterranean). Mar. Geol.. http://dx.doi.org/10.1016/j.margeo.2012.02.007.

Ferrand, E., Eyrolles, F., Radakovitch, O., Provansal, M., Dufour, S., Vella, C., Raccasi, G.,Gurrarian, R., 2012. Historical levels of heavy metals and artificial radionuclides re-constructed from overbank sediment records in lower Rhône River (South-EastFrance). Geochim. Cosmochim. Acta 82, 163–182 (special Issue on EnvironmentalRecords of Anthropogenic Impacts).

Glenn, E., Hucklebridge, P.K., Hinojosa-Huerta, O., Nagler, P.L., Pitt, J., 2008. Reconcilingenvironmental and flood control goals on an arid-zone river: case study of the limi-trophe region of the lower Colorado River in the United States and Mexico. Environ.Manag. 41, 322–335.

Gracia-Ruiz, J.M., 2010. The effects of land uses on soil erosion in Spain: a review. Catena81, 1–11.

Gregory, K.J., Macklin, M.G., Walling, D.E. (Eds.), 2006. Past Hydrological Events Related toUnderstanding Global Change. Catena Special Issue (66). Elsevier, Amsterdam (187pp.).

Guillen, J., Palanques, A., 1992. Sediment dynamics and hydrodynamics in the lowercourse of a river highly regulated by dams: the Ebro River. Sedimentology 39,567–579.

Gurnell, A.M., Peiry, J.L., Petts, G.E., 2003. Using Historical Data in Fluvial Geomorphology.In: Kondolf, M., Piégay, H. (Eds.), Tools in Geomorphology. Wiley, pp. 77–103.

Hinderer, M., 2012. From gullies to mountains belts: a review of sediments budgets atvarious scales. Sediment. Geol. 280, 21–59.

HistRhône Databank, 2013. Base de données historiques des crues et des inondationsdans le bas Rhône. http://histrhone.cerege.fr.

Hoffmann, T., Thorndycraft, V.R., Brown, A.G., Coulthard, T.J., Damnati, B., Kale, V.S.,Middelkoop, H., Notebaert, B., Walling, D.E., 2010. Human impact on fluvial regimesand sediment flux during the Holocene: review and future research agenda. Glob.Planet. Chang. 72, 87–98.

Hohensinner, S., Habersack, H., Jungwirth, M., Zauner, G., 2004. Reconstruction of thecharacteristics of a natural alluvial river–floodplain system and hydromorphologicalchanges following human modifications: the Danube River (1812–1991). River Res.Appl. 20 (1), 25–41.

Hohensinner, S., Herrnegger, M., Blaschke, A.P., Habereder, C., Haidvogl, G., Hein, T.,Jungwirth, M., Weiss, M., 2008. Type-specific reference conditions of fluvial land-scapes: a search in the past by 3D-reconstruction. Catena 75, 200–215.

Hooke, J.M., 2006. Human impacts on fluvial systems in the Mediterranean region.Geomorphology 79, 311–335.

IRS, 2000. Etude globale pour une stratégie de réduction des risques dus aux crues duRhône. Technical report, 1. Etude du transport solide. SOGREAH, Grenoble, France.

41M. Provansal et al. / Geomorphology 219 (2014) 27–41

Jorda, M., 1985. La torrentialité holocène des Alpes françaises du sud, facteursanthropiques et paramètres naturels de son évolution. Cahiers ligures de Prehistoireet Protohistoire, 2 pp. 49–70.

Jugaru Tiron, L., Le Coz, J., Provansal, M., Panin, N., Raccasi, G., Dramais, G., Dussouillez,2009. Flow and sediment processes in a cutoff meander of the Danube Delta duringepisodic flooding. Geomorphology 106, 186–197.

Jungwirth, M., Muhar, S., Schmutz, S., 2002. Re-establishing and assessing ecologicalintegrity in riverine landscapes. Freshw. Biol. 47, 867–887.

Kettner, A.J., Syvitski, J.P.M., 2009. Fluvial responses to environmental perturbations inthe Northern Mediterranean since the Last Glacial Maximum. Quat. Sci. Rev. 28,2386–2397.

Kiss, T., Fiala, K., Sipos, G., 2008. Alterations of channel parameters in response to riverregulation works since 1840 on the Lower Tisza River (Hungary). Geomorphology98, 96–110.

Kondolf, G.M., 1997. Hungry water: effects of dams and gravel mining on river channels.Environ. Manag. 533–551.

Liébault, F., Piégay, H., 2002. Causes of 20th century channel narrowing in mountain andpiedmont rivers of southeastern France. Earth Surf. Process. Landf. 227, 425–444.

Liébault, F., Piégay, H., Frey, P., Landon, N., 2008. Tributaries and the management ofmain-stem geomorphology. In: Rice, S.P., Roy, A.G., Rhoads, B.L. (Eds.), River Conflu-ences, Tributaries and the Fluvial Network, 12. John Wiley and Sons, pp. 243–270.

Liquette, C., Canals, M., Ludwig, W., Arno, P., 2009. Sediment discharge of the rivers ofCatalonia, NE Spain, and the influence of human impacts. J. Hydrol. 366, 76–88.

Luo, X.L., Zeng, E.Y., Ji, R.Y., Wang, C.P., 2007. Effects of in-channel sand excavation on thehydrology of the Pearl River Delta, China. J. Hydrol. 343, 230–239.

Maillet, G., Raccasi, G., Provansal, M., Sabatier, F., Antonelli, C., Vella, C., Fleury, J., 2007.Transferts sédimentaires dans le bas Rhône depuis le milieu du XIXème siècle, essaide quantification. Géog. Phys. Quatern. 61, 39–53.

Maillet, G., Poizot, E., Sabatier, F., Mear, Y., Vella, C., 2011. Pattern of sediment transport ina microtidal river mouth using geostatistical sediment-trend analysis. J. Sediment.Res. 81, 138–152. http://dx.doi.org/10.2110/jsr.2011.8.

Michalkova, M., Piégay, H., Kondolf, G.M., Greco, S., 2010. Longitudinal and temporalevolution of the Sacramento River between Red Bluff and Colusa, California (1942–1999). Earth Surf. Process. Landf. 36, 257–272.

Micheli, E.R., Larsen, E.W., 2010. River channel cutoff dynamics, Sacramento River,California, USA. River Res. Appl. 27, 328–344. http://dx.doi.org/10.1002/rra.1360.

Milliman, J.D., Farnsworth, K.L., 2011. River Discharge to the Coastal Ocean. A GlobalSynthesis, Cambridge University Press, UK (392 pp.).

Miramont, C., Jorda, M., Pichard, G., 1998. Evolution historique de la morphogenèse et dela dynamique fluviale d'une rivière méditerranéenne: l'exemple de la moyenneDurance (France du sud-est). Géog. Phys. Quatern. 52, 381–392.

Ollero, A., 2010. Channel changes and floodplain management in the meandering middleEbro River, Spain. Geomorphology 117, 247–260.

Pardé, M., 1925. Le régime du Rhône, étude hydrologique, Etude et travaux de l'Institut deGéographie Alpine de l'Université de Grenoble, 2 vol. Thesis , Grenoble Univ.,Grenoble, France.

Petit, F., Poinsard, D., Bravard, J.P., 1996. Channel incision, gravel mining and bedloadtransport in, the Rhône River upstream of Lyon, France (“canal de Miribel”). Catena26, 209–226.

Petts, G.E., Gurnell, A.M., 2005. Dams and geomorphology: research progress and newdirections. Geomorphology 71, 27–47.

Phillips, J.D., Slattery, M.C., Musselman, Z.A., 2004. Dam-to-delta sediment inputs andstorage in the lower Trinity River, Texas. Geomorphology 62, 17–34.

Phillips, J.D., Slattery, M.C., Musselman, Z.A., 2005. Channel adjustments of the lowerTrinity River, Texas downstream of Livingston Dam. Earth Surf. Process. Landf. 30,1419–1439.

Pichard, G., 1995. Les crues sur le bas Rhône de 1500 à nos jours. Pour une histoirehydroclimatique. Méditerranée 82, 105–116.

Piégay, H., Walling, D.E., Landon, N., He, Q., Liébault, F., Petiot, R., 2004. Contemporarychanges in sediment yield in an alpine montane basin due to afforestation (theupper Drôme in France). Catena 55, 183–212.

Piégay, H., Hupp, C.R., Citterio, A., Dufour, S., Moulin, B., Walling, D.E., 2008. Spatial andtemporal variability in sedimentation rates associated with cutoff channel infilldeposits: Ain River, France. Water Resour. Res. 44 (5), W05420.

Pont, D., Simonnet, J.P., Walter, A.V., 2002. Medium-term changes in suspended sedimentdelivery to the ocean: consequences of catchment heterogeneity and river manage-ment (Rhône River, France). Estuar. Coast. Shelf Sci. 54, 1–18.

Poulos, S.E., Collins, M.B., 2002. Fluviatile sediment fluxes to the Mediterranean Sea: aquantitative approach and the influence of dams. In: Jones, S.J., Frostick, L.E. (Eds.),Sediment Flux to Basins: Causes, Controls and Consequences. Special PublicationsGeological Society, London, 191, pp. 227–245.

Provansal, M., Vella, C., Arnaud-Fassetta, G., Sabatier, F., Maillet, G., 2003. Role of thefluvial sediment input in the mobility of the Rhône delta coast, (France).Géomorphologie: Relief, Processus, Environnement, 4 pp. 271–282.

Provansal, M., Villiet, J., Eyrolle, F., Raccasi, G., Gurriaran, R., Antonelli, C., 2010. High-resolution evaluation of recent bank accretion rate of the managed Rhone: a casestudy by multi-proxy approach. Geomorphology 117, 287–297.

Provansal, M., Ferrand, E., Eyrolle, F., Raccasi, G., Monaco, M., Gurriaran, R., 2012. Spatialvariability in sedimentation rates and artificial radionuclide storage in alluvialbanks of the lower Rhône River. Aquat. Sci. 74, 735–750.

Raccasi, G., 2008. Mutations géomorphologiques récentes du Rhône aval, Recherches envue de la restauration hydraulique et de la gestion des crues. Ph.D. Thesis , Aix-Marseille Univ., Aix-en-Provence, France.

Räsänen, T.A., Koponen, J., Lauri, H., Kummu, M., 2012. Downstream hydrological impactsof hydropower development in the Upper Mekong Basin. Water Resour. Manag. 26,3495–3513.

Rinaldi, M., 2003. Recent channel adjustments in alluvial rivers of Tuscany, Central Italy.Earth Surf. Process. Landf. 28, 587–608.

Rodrigues, S., Breheret, J.G., Macaire, J.J., Moatar, F., Nistoran, D., Juge, Ph., 2006. Flow andsediment dynamics in the vegetated secondary channels of an anabranching river:the Loire River (France). Sediment. Geol. 186, 89–109.

Rollet, A.J., 2007. Etude et gestion de la dynamique sédimentaire d'un tronçon fluvial àl'aval d'un barrage: le cas de la basse vallée de l'Ain. Ph.D. Thesis , Lyon III Univ.,Lyon, France.

Rollet, A.J., Piégay, H., Bornette, G., Dufour, S., Persat, H., 2013. Assessment of conse-quences of sediment deficit on a gravel river-bed downstream of dams in restorationperspectives: application of a multicriteria, hierarchical, and spatially explicit diagno-sis. River Res. Appl. http://dx.doi.org/10.1002/rra.2689.

Rovira, A., Batalla, R.J., Sala, M., 2005. Response of a river sediment budget after historicalgravel mining (the lower Tordera, NE Spain). River Res. Appl. 21, 829–847.

Sabatier, F., Maillet, G., Provansal, M., Fleury, T.-J., Suanez, S., Vella, C., 2006. Sediment bud-get of the Rhône delta shoreface since the middle of the 19th century. Mar. Geol. 234,143–157.

Sabatier, F., Samat, O., Ullmann, A., Suanez, S., 2009. Connecting large-scale coastal behav-iour with coastal management of the Rhone delta. Geomorphology 107, 79–89.

Salit, F., Arnaud-Fassetta, G., Zaharia, L., Madelin, M., Beltrando, G., 2014. The influence ofriver training on channel changes during the 20th century in the Lower Siret River,Danube catchment (Romania). Géomorphologie: relief, processus, environnement(in press).

Sclafert, Th., 1959. Cultures en Haute Provence. Déboisements et pâturages au Moyen-Age, S.E.V.P.E.N., Paris (271 pp.).

Siché, I., Arnaud-Fassetta, G., 2014. Anthropogenic actions since the end of the Little IceAge: a critical factor controlling the last ‘fluvial metamorphosis’ of the Isonzo River(northern Italy, Adriatic). Méditerranée (in press).

Simeoni, U., Corbau, C., 2009. A review of the Delta Po evolution (Italy) related to climaticchanges and human impacts. Geomorphology 107, 64–71.

SMAVD, 2003. Le bassin versant de la Durance — Etude monographique des fleuves etgrandes rivières de France. Technical report. Syndicat Mixte d'Aménagement de laVallée de la Durance, Mallemort, France.

SOGREAH, CESAME, TELEOS, 2001. Moyenne et basse Durance-Schéma d'aménagementet de gestion. Technical report. Syndicat Mixte d'Aménagement de la Vallée de laDurance, Mallemort, France.

Stanley, D.J., 1996. Nile delta: extreme case of sediment entrapment on a delta plain andconsequent coastal land loss. Mar. Geol. 129, 189–195.

Steiger, J., Gurnell, A.M., Ergenzinger, P., Snelder, D., 2003. Sedimentation in the riparianzone of an incising river. Earth Surf. Process. Landf. 26, 91–110.

Surian, N., Rinaldi, M., 2003. Morphological response to river engineering and manage-ment in alluvial channels in Italy. Geomorphology 50, 307–326.

Syvitski, J.P.M., Vörösmaty, C.J., Kettner, A.J., Green, P., 2005. Impacts of humans on theflux of terrestrial sediments to the global coastal ocean. Science 308, 376–380.

Syvitski, J.P.M., Kettner, A.J., Overeem, I., Hutton, E.W.H., Hannon, M.T., Brakenridge, G.R.,Day, J., Vörösmarty, C., Saito, Y., Giosan, L., Nicholls, R.J., 2009. Sinking deltas due tohuman activities. Nat. Geosci. 2, 681–686.

Uribelarrea, D., Perez-Gonzalez, A., Benito, G., 2003. Channel changes in the Jarama andTagus rivers (central Spain) over the past 500 years. Quat. Sci. Rev. 22, 2209–2221.

Vassas, C., Sabatier, F., Vella, C., 2007. Résultats préliminaires sur le taux de transport àl'embouchure du Grand Rhône (SE de la France). La Houille Blanche 4, 35–40.

Vella, C., Fleury, T.J., Raccasi, G., Provansal, M., Sabatier, F., Bourcier, M., 2005. Evolution ofthe Rhône delta plain in the Holocene. Mar. Geol. 222, 235–265.

Vella, M.A., Ghilardi, M., Diouf, O., Parisot, J.C., Hermitte, D., Provansal, M., et al., 2013.Géoarchéologie du Rhône dans le secteur du Pont Saint-Bénézet (Avignon, Provence,France) au cours de la secondemoitié du deuxièmemillénaire apr. J.-C.: étude croiséede géographie historique et des paléo-environnements. Géomorphologie: Relief,Processus, Environnements, 3 pp. 287–310.

Vericat, D., Batalla, R.J., 2005. Sediment transport in a highly regulated fluvial systemduring two consecutive floods (lower Ebro River, NE Iberian Peninsula). Earth Surf.Process. Landf. 30, 385–402.

Walling, D.E., He, Q., 1997. Use of fallout 137Cs in investigations of overbank sedimentdeposition on river floodplains. Catena 29, 263–282.

Wang, H., Saito, Y., Zhang, Y., Bi, N., Sun, X., Yang, Z., 2011. Recent changes of sedimentflux to the western Pacific Ocean from major rivers in East and Southeast Asia.Earth Sci. Rev. 108, 80–100.

Warner, R., 2006. Natural and artificial linkages and discontinuities in a Mediterraneanlandscape: some case studies from the Durance Valley, France. Catena 66, 236–250.

Wohl, E., 2006. Human impacts to mountain streams. Geomorphology 79, 217–248.