Long term evolution of a dam reservoir subjected to regular flushing events

Transcription

1 Adv. Geosci., 39, 89 94, 2014 doi: /adgeo Author(s) CC Attribution 3.0 License. Advances in Geosciences Open Access Long term evolution of a reservoir subjected to regular flushing events L. Guertault 1, B. Camenen 1, C. Peteuil 2, and A. Paquier 1 1 Irstea, UR HHLY, centre de Lyon-Villeurbanne, 5 rue de la Doua, CS 70077, Villeurbanne cedex, France 2 Compagnie Nationale du Rhône, 2 rue André Bonin, Lyon Cedex 04, France Correspondence to: L. Guertault (lucie.guertault@irstea.fr) Received: 11 October 2013 Revised: 20 January 2014 Accepted: 27 January 2014 Published: 1 April 2014 Abstract. An analysis of long term morphological evolution of Génissiat reservoir (France) is provided. First, a methodology for bathymetric data processing and reservoir sediment volume budget calculation is described. An estimation of global uncertainties in volume calculation is proposed. The reservoir bathymetric budget for several flushing events and interflush periods is presented, showing global decrease of deposited sediment volume with time. The spatial dynamics of reservoir subreaches is highlighted and typical patterns in flush and interflush periods are identified. 1 Introduction Reservoirs formed by s on natural Rivers can alter balance between sediment inflow and outflow (Morris and Fan, 1998). Because of ir low flow velocities, y trap a significant quantity of incoming sediment introducing downstream filling. Sediment issues do not only affect reservoirs through water storage loss but can also influence downstream patterns creating bed degradation, accelerating bank failure and scour formation. Among several technical solutions for reservoir desiltation, a common one used by operators is flushing. Reservoir flushing consists of lowering water level so that inflowing water is routed through reservoir with a hydraulic regime similar to natural Riverine condition (Brandt, 2000). Higher velocities and bed shear stresses increase sediment transport capacity and allow to erode part of material deposited in reservoir (Di Silvio, 2001). The Rhône River is one of major European Rivers. It flows from Swiss Alps through Switzerland and France to Mediterranean sea. The reach from Swiss border to Lyon is known as French Upper Rhône River (Fig.1). Output water from Lake Geneva is clear water with a controlled discharge. Two kilometers downstream is confluence with Arve River, which is main tributary of Upper Rhône River, providing almost all sediment load, which is estimated between 1 and 3 million tons per year. Two s are built on Swiss Rhône River: Verbois and Chancy Pougny s. They trap a large quantity of sediment from Arve River. Downstream 70 m high Génissiat and four additional run-of-river developments operated by Compagnie Nationale du Rhône (CNR) exist. Génissiat has been in operation since The reservoir is 23 km long with an initial volume of 56 million m 3. Nowadays it is filled with approximately 14 million m 3 of sediment. In upper part of reservoir, one can find a large valley ( Etournel site) with a secondary channel and several islands. Then River enters in a canyon where flow is confined and under hydraulic influence. Close to, water height is about 60 m. Granulometric surveys (Bouchard and Dumond, 2000; Lerch and Thizy, 2013) show a typical downstream fining in reservoir channel: gravel and coarse sand in upper part of reservoir to a predominance of silt in last 10 km of reservoir. A transversal grain sorting is also highlighted, as banks materials are finer than channel ones. These conditions lead us to consider that both bedload for coarse sediments and suspended load for fine sediments are involved in reservoir dynamics. 2 Flushing operations on Upper Rhône River Since early 20th century, regular flushes have been conducted on Upper Rhˆ0ne River to prevent flood hazards Published by Copernicus Publications on behalf of European Geosciences Union.

2 90 L. Guertault et al.: Long term evolution of a reservoir subjected to regular flushing events FRANCE Génissiat Seyssel Rhône river Valserine river Verbois Arve river Chancy Pougny FRANCE Usses river bridge Lake Geneva Valserine river SWITZERLAND Pougny bridge Carnot bridge Génissiat Bognes Pyrimont bridge Usses river Etournel Défilé de l'ecluse Léaz 20 km 5 km river subreach Seyssel bed elevation (m) Etournel Défilé de l'ecluse Carnot Bridge march 1955 december 1975 april 1997 november 2012 Léaz distance from Génissiat Dam (km) Fig. 2. Evolution of Longitudinal profile of Génissiat reservoir. Fig. 1. Location of study site. in lowest parts of Geneva city due to bed aggradation of Verbois reservoir. A large quantity of sediment is removed from Verbois reservoir during a flush and supporting operations are carried out through French Upper Rhône River, particularly to prevent large deposition in Génissiat reservoir. Since 1981, CNR conducts environmental friendly flushing to release suspended sediment concentrations tolerable by fluvial environment downstream of Génissiat (Peteuil et al., 2013). In Fig. 2, evolution of longitudinal profile of Génissiat reservoir is presented for last 60 yr. From 1955 to 1975, seven Swiss-French flushing operations were carried out. Swiss and French reservoirs were flushed and n risen toger. At Génissiat, water was mainly evacuated through by half depth gate and surface spillway. A stable profile is observed in upper 10 km of reservoir while large deposition (20 m in 20 yr) occurred furr downstream of Génissiat reservoir. The bathymetric evolution was related to a reservoir management that did not promote sediment release downstream of Génissiat. As a consequence, annual mean deposition rate was about m 3 ; From 1975 to 1997, six flushing operations were carried out. The reservoir water level was lowered by about 10 m and n risen toger with Swiss reservoirs. Water was also briefly evacuated through bottom gate, promoting sediment release downstream of. The mean deposition rate has decreased and main morphological changes occurred in last 5 km; From 1997 to 2012, four flushing operations were carried out. The longitudinal profile was in a quasi equilibrium. Since 1997, flushing operations have been carried out following a specific procedure: first lowering Génissiat reservoir water level by about 20 m during first week of flush, and n rising it by 10 m during second week dedicated to Swiss flushes. The bottom gate is opened during almost all time to empty reservoir. Indeed, this gate opening allows to release downstream a higher concentration due to vertical gradient in concentration profile observed in reservoir and reduces reservoir filling. The objective of paper is to highlight spatial dynamics of Génissiat reservoir for flushing events and interflush periods to better understand hydro-sedimentary processes involved in its morphological evolution since A methodology to calculate those budgets is proposed. Uncertainty sources are listed to estimate global uncertainty. 3 Methodology for volume calculation and uncertainties estimation Bathymetric surveys using a mono beam echo sounder coupled to a DGPS are carried out by CNR before and after flushing events. Since 1984, data are sufficiently accurate and homogeneous to allow volume calculation using bathymetric difference. About 110 transversal profiles with an average number of points per section from 15 to 45 and with distances ranging from 20 to 500 m are measured. 3.1 Bathymetric data post processing A first step consists in completing all profiles with points based on topographic data or or profiles measured at same section assuming banks have not moved. The next step is to define same boundaries for each measured profile at a given cross section to use same projection axis to convert XYZ data into abscissa-elevation coordinate. Then distance between two consecutive sections L i can be estimated using GIS Tools. This distance is defined as length of curvilinear arc joining middle of two consecutive sections (Fig. 3). Adv. Geosci., 39, 89 94, 2014

3 L. Guertault et al.: Long term evolution of a reservoir subjected to regular flushing events 91 L i ΔS i+1,rd D i section i+1 ΔS i+1,fd ΔS i,rd L i-1 section i ΔS i,fd ΔS i+1,rg ΔS i-1,rd ΔS i,rg section i-1 ΔS i-1,fd ΔS i-1,rg Fig. 3. Representation of calculation variables S i,bk = S i,left + S i,right. 3.2 Volume calculation The volume calculation is based on bathymetric data differences between two dates t 1 and t 2. The main variables used are defined in Fig. 3. The overall volume budget V in reservoir is sum of all intermediate volumes V i associated to geometrical variations of cross section i between t 1 and t 2. M M V = V i = D i S i (1) i=1 i=1 where S i is area evolution of section i between t 1 and t 2 defined as difference between profile areas calculated at t 1 and t 2 and approximated using a trapezoidal description. D i is application length defined as sum of two half distances between consecutive sections and M is number of cross sections. 3.3 Uncertainty estimation There are two contributions to uncertainty: imprecision u P and bias u B. Imprecision is related to measuring instrument sensitivity and calculation methods. Bias is related to discrete description of bathymetric data and to fact that a given cross section cannot be representative of bed evolution around its location. The two uncertainty sources (bias and imprecision) are assumed independent. The total uncertainty u(v ) is written: u(v ) = M u 2 P (V i ) + u 2 B (V i ) (2) i=1 in application length u P (D i ). [up ] (S i ) 2 [ ] up (D i ) 2 u P (V i ) = V i + (3) S i D i We estimate that precision of GIS measuring tool is ±5 m, so that standard individual uncertainty is u P (D i ) = 5/ 3= 2.9 m, according to GUM (Joint Committee For Guides in Metrology, 2008). Imprecision u P ( S i ) in area is caused by imprecisions in measurement and calculation methodology. The different steps of methodology, more particularly cross sectional axis projection, allow to reduce imprecision and are neglected compared to measurement imprecision. Considering instruments used and measured parameters, relative imprecision of XY coordinates is neglected compared to relative imprecision on height. We estimate that average precision of sounder is ±5 cm. The individual standard imprecision uncertainty of height for a point is u P (z) = 0.05/ 3= m. [ ] N(t) 1 u P ( S i,t ) = u P (z) (x k+1,t x k,t ) 2 (4) Bias t=[t 1,t 2 ] k=1 Bias of volume related to section i u B (V i ) is due to lack of information between cross sections. When geometrical evolution is very different between consecutive sections, hyposis of a constant evolution of S i can be questioned. Moreover, banks are subjected to local phenomena such as bank failure which cannot be always captured. Consequently we suggest following formula for this uncertainty: u B (V i,l) = α i,l [ ( Si,l S i 1,l )L i 1 + ( S i+1,l S i,l )L i ] (5) where l = [bk,mc] for bank and main channel and α (i,l) is a coefficient function of position in space. As slope is a significant parameter driving morphological evolutions: α i,mc = ( p i,t1 + p i,t2 )/2, where p i,t1 and p i,t2 are respectively local slopes based on thalweg elevation at section i at times t 1 and t 2. α mc varies from to For banks, equilibrium slope of reservoir sediments is used, calculated using stability angle φ for silt and clay, α i,bk = tanφ = As volumes related to banks and channel are assumed independent, bias uncertainty of whole section is: u B (V i ) = u 2 B (V i,l ) (6) l=[mc,bk] Imprecision Imprecision u P (V i ) in volume related to section i is caused by imprecision on area u P ( S i ) and imprecision 4 Bathymetric balance of Génissiat reservoir The methodology is applied to estimate volume budgets of reservoir since Standard uncertainties are Adv. Geosci., 39, 89 94, 2014

4 92 L. Guertault et al.: Long term evolution of a reservoir subjected to regular flushing events Table 1. Sediment volume budget from 1984 to 2012 (Positive values correspond to deposited volumes and negative ones correspond to eroded volumes. The volume budget includes dredged volumes. Q F is annual maximum daily River discharge. Q F,Rhône = 800 m 3 s 1, Q F,Arve = 300 m 3 s 1 ) Period Volumes Duration with Q > Q F (10 3 m 3 ) (days) budget dredgings Rhône (Pougny) Arve (Geneva) 1984 Flush 855 ± ± Flush 673 ± ± Flush 431 ± ± Flush 795 ± ± Flush 374 ± ± Flush 672 ± ± Flush 357 ± ± Flush 1472 ± 60 calculated with methodology and multiplied by a coverage factor k = 2 to quantify m with a confident interval at 95 % (Joint Committee For Guides in Metrology, 2008). 4.1 Sediment volume budget from 1984 to 2012 The volume budget of reservoir is presented in Table 1. Uncertainties vary between 50 and m 3. On average, bias represents 70 % of uncertainty on total budget. Neverless, it yields most of variability in uncertainty. Indeed, bias and so uncertainty are higher when number of cross sections is low and when morphological evolution is not uniform along reservoir. From 1984 to 2012, global balance for flushing operations is positive. Those events last approximately two weeks but contribute to half volume budget of reservoir. Neverless, a high variability in Génissiat reservoir flushing events budget can be observed, with a negative budget in 2003 and a significant positive budget in This can be attributed to variability in both sediment input and reservoir management conditions; In general, volume budget is positive during interflush periods. The negative budget for interflush period is not entirely representative because bathymetric data was not surveyed for upper four kilometers of reservoir before 1997 flush (and so for 1997 flush). The deposited volume depends on water and solid inflows natural variability during in- (10 3 m 3 ) volume of sediment within reservoir (spatially cumulated) Etournel Défilé Léaz de l'ecluse Chancy Pougny weir Carnot Bridge Bellegarde sur Valserine distance from Génissiat (km) 1984 flush flush flush flush flush flush flush flush Fig. 4. Spatially cumulated sediment volume budget of Génissiat reservoir since 1984 (circles represent sharp changes in sediment dynamics during flushes). terflush periods. Since 1991, dredging operations have been carried out in reservoir, particularly close to and part of bed material deposited re is thus removed. Moreover, from 1970 to 1995, mechanical extractions of gravel where operated in Etournel site. The total volume extracted is estimated to m Spatial dynamics of reservoir and effects of flushes from 1984 to 2012 This analysis allows a detailed description of spatial dynamics of deposition and erosion in reservoir for both interflush periods and flushing events. Figure 4 shows spatially cumulated volume budget of reservoir calculated from Pougny as a function of streamwise location for different flushes and interflush periods since The Génissiat is located at abscissa 0 and plotted value at this location corresponds to total volume deposited or eroded in reservoir. Homogeneous subreaches of River can be defined: In Etournel natural site reservoir dynamics are opposite during flush and interflush periods, showing respectively deposition and erosion. This behaviour explains stability of longitudinal profile in this reach; Downstream of Carnot Bridge, reach named Défilé de l Écluse is in an equilibrium state, may be because of narrow but homogeneous section; In Léaz and Gorges, deposition occurs during interflush periods and variable evolutions occur during flushes. For 2003 flush, se subreaches show a negative volume budget, while a significant deposition occured during 2012 flush; Adv. Geosci., 39, 89 94, 2014

5 During interflush period, re is also significant deposition since velocities are very low. 5.1 Bedload and suspended load transport The methodology allows to split total budget into bank and L. Guertault channel budgets al.: Long (Fig.5). term evolution of a reservoir 290 subjected At present, to regular sediment flushing input events at upper part of 93reser- voir is poorly known. It depends on upstream Swiss reser- (a) voir Continuous release for sedimentation fine sediments ofand silton and clay contribution causes of of floodplain reach downstream and banks Chancy-Pougny deposits while main for coarser channel shape An erosion-deposition is maintained by system repeated is flushing observed opera- in up flush 295 stream tions. gravel-bed During flushes, reach indicating banks area alternately dynamics dominated exposed by flush bedload and submerged, transport. This accelerating has been bank confirmed failureby rate. a hydrophone Typi- sed- Etournel Défilé Léaz 1984 flush de l'ecluse iments positioned cally, flushing at Pougny operations (Geay lead et al., to2013) bank erosion that has and shown interflush occurs periods in to this bank upper aggradation. reach even for relatively low 1997 flush flows. Since bed material extractions have been forbidden, 2000 flush Discussion Etournel site tends to aggrade with creation of larger that 1993 flush bedload 2003 flush islands and some clogging in secondary channels flush In Bedload lower and suspended reaches, sediment load transport transportprocesses are highly dependant of reservoir water level. For high wa Atter present, levels, corresponding sediment input to at interflush upper periods, part offlow reservoir progressively is poorly known. decrease It depends downstream on upstream from Léaz Swiss gorges reser- velocities (b) 305 to to voir Etournel Défilé Léaz. release A downstream for fine sediments fining is and observed on with contribution progressive of deposition of sediments in suspension depending on ir 1987 flush iments. An erosion-deposition system is observed in up flush de l'ecluse reach downstream Chancy-Pougny for coarser sed settling velocity. On contrary to clay and silt, sand is isgen- erally transported both in suspended load (graded) and bed bedload transport. This has been confirmed by a hydrophone stream gravel-bed reach indicating a dynamics dominated by 1990 flush 1993 flush load leading to more complex exchanges with bed positioned at Pougny (Geay, 2013) that has shown that 1997 of flush bedload During occurs flushing in this events, upper reach progressive even for relatively lowering low of to in 2000 flush flows. reservoir Sincewater bed material level allows extractions to erode have been bed forbidden, material in reaches Etournel where site tends an accelerated to aggradeflow with is isestablished. creation of larger The reser flush islands voir draw and some down clogging creates in a flushing secondary channel. channels. The banks stabil flush ity In decreases lower and reaches, collapse. sediment Then, transport when processes reservoir are is isrisen highly during dependant Swiss ofs reservoir flushes, water sediment level. For transport high wa-ininthterlower levels, subreaches corresponding is isrelated to interflush to to periods, reservoir flowhydrodynamics. velocities distance distance upstream upstream Génissiat Génissiat (km) Fig. 5. Spatially cumulated sediment volume budget (km) in Génissiat progressively As a consequence, decrease downstream second week from Léaz of of gorges flush toisis gener- reservoir since 1984 separating main channel (a) and banks (b). Fig Spatially cumulated sediment volume budget in Génissiat. ally Acharacterized downstream fining by a large is observed deposit with of offine progressive sediments de-coming 320 from of Swiss sediments flushes, inas as suspension shown in infig depending Fig4 downstream on irfrom settling circles. velocity. The Onlocation contrary of of toextend clay and of of reservoir since 1984 separating main channel (a) and banks (b) 320position silt, this this sand deposit is generally on transported reservoir both water in suspended level during load depends In reach, since it corresponds to immediate mainupstream channel reach contribution of, is about deposition 80 % isof load flushing leading event. to more complex exchanges with bed. global budget trend in ans soreach prevails dynamics to overall during reservoir both flush- dy- During No significant flushing events, change ofprogressive grain lowering (graded) second and phase bed- of of The grainsize sizedistribution of of of namics. ing events and interflush periods. During flushes, intense deposition is generally observed. This deposition In In upper subreaches of reservoir, contribu- reaches through where an last accelerated reservoir river water bedlevel along-stream allows to erode reservoir bed material has hasbeen in beenobserved last years. flow Since is established. Since re re isobviously The reser- some could result from second phase of flushing operation, when rise of reservoir water level leads de- some tion tion of of banks in in volume budget is negligible, exvoir transport draw down of of coarse creates sediments, a flushing this channel. thismay maybe The beexplained banks sta- explainedby byde- to deposition of sediments coming from Swiss reservoirs. During interflush period, re is also significant deposition since velocities are very low. volume of sediment for main channel ( 10 3 m 3 ) volume of of sediment for for banks ( 10 3 m 3 ) The methodology allows to split total budget into bank and channel budgets (Fig. 5). The main channel contribution is about 80 % of global budget and prevails to overall reservoir dynamics. In upper subreaches of reservoir, contribution of banks in volume budget is negligible, except in case of a local significant bank failure (interflush , 20 km upstream of ). In gorges (13.5 km upstream of ), contribution of banks increases. bility decreases and leads to ir collapse. Then, when reservoir is risen during Swiss s flushes, sediment transport in lower subreaches is related to reservoir hydrodynamics. As a consequence, second week of flush is generally characterized by a large deposit of fine sediments coming from Swiss flushes, as shown in Fig. 4 downstream from circles. The location of extend of this deposit depends on reservoir water level during second phase of flushing event. No significant change of grain size distribution of river bed along-stream reservoir has been observed through last 15 yr. Since re is obviously some transport of coarse sediments, this may be explained by deposits of fine sediments over coarser layers. Some two meter deep cores close to confirmed occurrence of alternate layers of clay, silts and sands (Bouchard and Dumond, 2000). Adv. Geosci., 39, 89 94, 2014

6 94 L. Guertault et al.: Long term evolution of a reservoir subjected to regular flushing events 5.2 Dynamics of reservoir during interflush periods At Génissiat, only water intakes for provision of hydropower plants are operated during interflush periods. Those intakes are located at upper third of water level close to. Thus, as flow velocities are also quite low, sediment output is usually very low. Hydrological conditions that may be significant for sediment transport have been reported in Table 1 for Arve River at Geneva and Rhône River at Pougny. Indeed, a flood on Arve River may provide a significant sediment supply to Upper Rhône River and lead to deposition in Verbois, Chancy-Pougny and Génissiat reservoirs. A flood on Upper Rhône River may erode sediments in three reservoirs, but Génissiat DAM management mitigates sediment transport in Génissiat reservoir. If dredged volumes are taken into account in Génissiat reservoir sediment budget, we can notice that two interflush periods with highest deposited volume ( and ) are characterized by several floods on Arve River and no significant event in Rhône River. 5.3 Dynamics of reservoir during flushes A specific reservoir management procedure is operated during flushing events. Before 1997 flush, a small part of Génissiat reservoir bed material was eroded before released sediments from Swiss s flushes got into reservoir and were partly deposited. Those patterns can be observed on Fig. 4 for 1984 to 1993 flushes. The middle subreaches of reservoir encountered no significant evolution during this period. The downstream third of reservoir is characterized by large sediment deposits released from Swiss reservoirs that occurred at end of flush. Since 1997, first phase of flush allowed to erode reservoir bottom from Léaz gorges to. During second phase, efficiency of transfer of fine sediments downstream Génissiat is closely related to management. Different scenarios occurred between 1997 and 2012 (Fig. 4). During 1997 and 2003 flushes, reservoir was operated with planed water level during second phase and incoming sediment where mainly released downstream, with deposition only close to, as highlighted by circles on Fig. 4. During second phase of 2000 and 2012 flushes, reservoir water level was higher due to unfavourable hydrological conditions and led to a significant deposition of incoming material, beginning from Léaz gorges. 6 Conclusions and perspectives A methodology was implemented to calculate sediment volume budget in a reservoir based on bathymetric data described by cross sections. An estimation of uncertainties for those volumes was proposed. This methodology was used for 16 bathymetric campaigns to estimate sediment budgets of Génissiat reservoir during flushes and interflush periods from 1984 to The spatial budget has allowed to divide reservoir into four subreaches with homogeneous patterns. Furr analysis of budget combined to hydrosedimentary data has highlighted mechanisms of sediment transport in reservoir: bedload may prevail in upstream part while suspended load of sand, silt and mud prevail in downstream part. The morphological evolution of reservoir during flush and interflush periods has been related to reservoir management and sediment input. In order to better understand local aspects and driving processes of reservoir dynamics, an analysis based on reservoir hydro-sedimentary parameters is needed. 1-D hydro-sedimentary models will be applied to reproduce both fine and coarse sediments dynamics, during flushing events and for long term periods. Moreover, some significant efforts should be made to get a better estimate of incoming sediment fluxes in reservoir using a turbidimeter and a hydrophone installed at Pougny. Acknowledgements. Authors want to thank all technicians from Irstea and CNR for collecting data. References Bouchard, J. and Dumond, L.: Exploitation of results of coring campain in Génissiat reservoir, Barrage de Génissiat: Exploitation des résultats de la campagne de carottages, Tech. rep., EDF, 2000 (in French). Brandt, S.: A review of reservoir desiltation, Int. J. Sediment Res., 15, , Di Silvio, G.: Basic classification of reservoir according to relevant sedimentation processes, in: 29th IAHR World Congress, Bejing, China, , Geay, T.: Passive hydrophone monitoring of bedload transport in gravel bed rivers [Mesure hydrophone du transport solide par charriage dans les rivières], PhD sis, Joseph Fourier University, Grenoble, France, (in French), Joint Committee For Guides in Metrology: Evaluation of measurement data Guide to expression of uncertainty in measurement, Tech. rep., BIPM, Lerch, C. and Thizy, R.: 2012 flush sediments analysis, Analyses des sédiments: Chasse 2012, Tech. rep., CNR, 2013 (in French). Morris, G. and Fan, J.: Reservoir Sedimentation Handbook: Design and Management of Dams, Reservoirs and Watersheds for Sustainable Use., McGraw-Hill, Peteuil, C., Fruchart, F., Abadie, F., Reynaud, S., Camenen, B., and Guertault, L.: Sustainable management of sediment fluxes in reservoir by environmental friendly flushing: case study of Génissiat on Upper Rhône River (France), in: ISRS Kyoto, Japan, (CDRom), , Adv. Geosci., 39, 89 94, 2014