68 hydrolink number 3/2018

The sediment load of rivers is affected by humanalterations, such as increased soil erosion dueto removal of native vegetation, roadconstruction, and other land disturbances,especially in steep upland areas (Figure 1).Sediment loads can also be increased by urban-ization and the resulting increased runoff. Assediment loads are transported downstreamthrough the water network they can be inter-rupted by natural and artificial means, includingmining of sand and gravel and trapping behinddams. Despite widespread increases in landdisturbance and consequent increased sedi-ment yields from upland areas, the sedimentloads of most major rivers have decreased inrecent decades as a result of extensive sedi-ment trapping by dams. This has led to accel-erated coastal erosion and loss of delta lands.

This article focuses on the effects of sedimenttrapping by dams and planning/managementopportunities to minimize these impacts and torestore downstream sediment supply tomaintain or restore geomorphic and ecologicalconditions. It is complementary to other articlesin this issue (e.g. Annandale et al., Efthymiou etal.), which explore structural and managementapproaches to reduce sediment trapping bydams, from a perspective of improving thesustainability of reservoir storage capacity forfuture generations.

Sediment trapping by damsDams typically store water by design, and storesediment as an unintended consequence,although some dams have been built as debrisbasins or sediment-control (sabo) dams. Dam-induced changes in flow regime are typicallyaccompanied by reductions in the river’ssediment load as reservoirs trap sediment,creating conditions of sediment starvationdirectly below the dam. Reservoirs trap 100% ofthe river’s bedload (coarse sediment movingalong the channel bed by rolling, sliding, andbouncing, consisting of gravel and sand), and apercentage of the suspended load (sand, silt,and mud held aloft in the water column), whichdepends on the ratio of the reservoir storagecapacity over the mean annual inflow of water.Storing water and sediment results in changes inflow and sediment load downstream of dams

(e.g. incision, narrowing, bed clogging andarmoring).

Dams that trap sediment but still release flowsthat are high enough to transport sedimentcreate sediment-starved, or ‘hungry water’downstream[2], so-called because these flowsstill have energy to transport sediment, but theirsediment loads have been trapped in thereservoir. This excess energy is expendeddownstream on bed and bank erosion, leadingto channel incision (downcutting) and conse-quent undermining of infrastructure (e.g.bridges, weirs) and loss of habitats throughchannel simplification.

However, hungry water does not occurdownstream of all dams. It depends on thebalance of flow and sediment supply.Reservoirs with large storage (relative to flow inthe river), built to redistribute water betweenseasons or even years, commonly reduce highflows, reducing the dynamism of the riverchannel downstream. Gravel beds, formerlymobilized every year or two, may go for yearswithout being moved, allowing fine sediment toaccumulate within the substrate (so-calledclogging process) and riparian vegetation toestablish in the active channel. Encroachingwoody riparian vegetation can lead to afeedback, where root establishment increasesthe resistance of the channel banks to erosion,so that dam-modified high flows are ever lesslikely to result in natural channel morpho-dynamics.

Large reservoirs may be capable of controllinga wide range of floods, and consequently canreduce the magnitude and frequency of floodsexperienced by the downstream channel. Thereduced flow may not transport sedimentdelivered to the river below the dam by tribu-taries, promoting channel aggradation andpotentially increasing flooding risk. Thus,depending on the balance between transportenergy available and sediment supply, someriver reaches below dams are in sedimentdeficit, some in sediment surplus[3].

The ecological consequences can be profound.The complexity of alluvial channel formsdepends upon the availability of coarse material(sand and gravel) that composes bars andriffles. In reaches starved of sediment byupstream dams, gravel is transporteddownstream without being replaced, resultingin loss of bars, riffles and beds, and with them,loss of channel complexity, resulting in asimplified ‘bowling-alley’ channel form lackingin habitats needed for fish and invertebrates.

Similar to river channels, also coastal areas andespecially deltas depend on a supply ofsediment from the river system to maintain theirforms against the natural processes of subsi-dence and coastal erosion[4]. Where thesediment supply to coasts and deltas has beencut off by upstream dams (and/or other activ-ities such as in-channel mining), coastal landshave eroded back and subsided below sealevel at increasing rates, as documented for the

DAMS, SEDIMENT DISCONTINUITY,AND MANAGEMENT RESPONSESBY G. MATHIAS KONDOLF AND RAFAEL J. SCHMITT

Figure 1. Humanalterationsincreasing sedimentyields from theupland landscape,sediment trappingabove dams, andconsequences ofsediment starvationdownstream[1]

RESERVOIRSEDIMENTATION

69hydrolink number 3/2018

downstream channel[8]. Although this solutionreplaces the downstream sediment supply withthe same sediment transported by the river, it israrely done because of the costs and of logis-tical and legal impediments to dredging thedeltas and transporting the sediment aroundthe reservoir and dam. Where sediment (usuallygravel or gravel-sand mixtures) has beenmechanically added to the channeldownstream, the sediment has mostly beenderived from other sources, such as terracegravels, floodplain gravel pits, or in somecases, gravel mines on tributary streams.

Adding gravel to river channels below dams iscommonly termed gravel augmentation orgravel replenishment. It has been widely under-taken in North America and Europe, in the vastmajority of cases to restore spawning habitatfor fish, especially salmon or trout. In northernCalifornia between 1976 and 2013, over200,000 m3 was added to the Sacramento River(Figure 2), 30,000 m3 to the Trinity River, andover 45,000 m3 to Clear Creek. On the TrinityRiver, the first such projects were undertaken in1976 to create artificial riffles, with lines ofboulders across the stream to hold the gravel inplace. The river’s transport capacity was greatly

reduced by Trinity Dam, so the placed graveldid not immediately wash out, as occurred withsimilarly designed projects on the MercedRiver[10]. By the early 1990s, releases ofmorphogenic flows were coordinated withgravel augmentation[11]. Planners havemeasured the transport of gravel downstream ofTrinity Dam by morphogenic flows (and naturalfloods spilling over the dam) and sought tocompensate for these gravel losses from thereach with gravel additions. Thus, therestoration project evolved to have the explicitgoal of building of bars and complex channelhabitat through addition of coarse sediment andrelease of flows to transport and redeposit thesediment in natural channel forms; resultingecological benefits, such as processing partic-ulate organic matter, inducing hyporheicexchange, and creating thermal complexityhave been documented[12].

Similarly, on the Uda River below the MurouDam in Japan, sediment replenishment hasbeen undertaken to restore channel complexitysince 2006. In the first five years of therestoration program, natural flows spilling fromthe dam were sufficient to transport the addedsediment in the first year, but in the subsequentfour years, morphogenic flows were released toachieve desired sediment mobility[8].Increasingly, sediment is added to reachesbelow dams in Japan to support developmentof gravel bars and other complex channelfeatures[13].

As summarized by Ock et al.[13], suchrestoration efforts require systematic planningthat accounts for specific objectives and localrestrictions of the river basin, river and reservoircharacteristics, and coordinating “flushing flows(magnitude, frequency, and timing), determiningquantity (amount added) and quality (grain sizeand source materials) of coarse sediment, andselecting an effective implementation techniquefor adding and transporting sediment…”. Damsvary widely in their settings (e.g. flow, sedimentload, presence of tributaries downstream,channel slope), in their size relative to the riverflow, and in their design and operation (e.g. sizeand location of outlets, reservoir geometry). Toassess dam-induced disruptions to a pre-damsediment balance, a sediment budget[14] canprovide a framework within which to analyzeinformation on the sediment transport capacityof the river (with and without “morphogenicflows”) and the quantity and caliber of sedimentsupplied from tributaries and other downstreamsources, as a basis for specifying

Mississippi[5] and the Mekong[6]. For example,the Mekong delta was created by deposition ofabundant river sediments, as the coast built outmore than 250 km over the past 8,000 years,from the current location of Phnom Penh to itspresent configuration. After millenia of progra-dation, however, the delta has begun retreatingin the last two decades due to factors such asin-channel mining of sand and acceleratedsubsidence.

Restoring flow for geomorphicprocesses below damsTo mitigate dam-induced impacts on sedimenttransport and channel processes, controlledhigh-flow releases designed to mimic the actionof natural floods are increasingly required inhydroelectric licenses for dams and as part ofprograms to restore river function. These delib-erate, high-flow releases constitute onecomponent of environmental flow requirementsfor maintenance of aquatic and riparian habitat.They reflect an evolution of environmentalrequirements from simple minimum flows toinclude periodic high flows to mimic floodeffects on channels or on ecological processes.While terminology varies (e.g. “flushing flows”,“channel maintenance flows”, “morphogenicflows”), the need for periodic high flows toaccomplish geomorphic goals has been widelyrecognized[7].

However, even if a post-dam flow regime was tomimic precisely the pre-dam flow regime, theriver system would still be severely altered bythe loss of its sediment load. Thus, for thedefinition of most beneficial morphogenic flows,it is critically important to take into account thesediment load available to the reachdownstream of the dam, such as sedimentsupplied from downstream tributaries.Increasingly, partial restoration of sediment loadis prescribed along with morphogenic flows.Coordinating morphogenic flows with sedimentaugmentations (i.e. supply, replenishment) isbecoming more common[8].

Managing sediment supply below existing damsTo partially restore sediment loads in aregulated stream, coarse material is mostcommonly added to downstream channels bymechanical means, and, to less extent, troughinduced riverbank erosion and failure[9]. Thesecoarser fractions preferentially deposit in deltasat the upstream end of reservoirs. In somecases, sediment has been mechanicallyremoved from reservoir deltas and placed in the

IAHR

G. Mathias Kondolf is a fluvialgeomorphologist and fellow at theCollegium, Institute for AdvancedStudies, University of Lyon,France, and Professor ofEnvironmental Planning at theUniversity of California Berkeley,where he teaches courses in

hydrology, river restoration, and environmental science.He researches human-river interactions, includingmanaging flood-prone lands, urban rivers, sediment inrivers and reservoirs, and river restoration(https://riverlab.berkeley.edu) and advises governmentsand non-governmental organizations on sustainablemanagement of rivers.

Rafael Schmitt is currently a PostDoc at the Natural Capital Projectand the Woods Institute for theEnvironment at StanfordUniversity, and is affiliated with UCBerkeley’s Riverlab. His researchfocusses on numerical modellingof catchment processes with a

special focus on sediment connectivity. Schmittdeveloped the CASCADE model for sedimentconnectivity, which he has since applied to strategicdam portfolio planning for sediment management inlarge river basins. His research has been awarded theChorafas Prize and the International HydropowerAssociation’s Young Researcher Award.

RESERVOIRSEDIMENTATION

70 hydrolink number 3/2018

“morphogenic flows” and, if needed, supplyingsediment to downstream reaches. Programs ofcoupled gravel additions and “morphogenicflows” are expensive and consequently notwidespread, but prescribing a “morphogenicflow” alone without accounting for sedimentsupply will usually not achieve ecological goalsenvisioned for the flows.

Designing dams to pass sedimentMechanically adding sediment downstream ofdams is expensive. It is more efficient to employgravity to deliver sediment to the channeldownstream of dams by passing sedimentthrough or around dams, for which a range oftechniques can be used[15,16,17].

For smaller dams, the most sustainableapproach (where feasible) is to pass thesediment load around or through the dam.Water can be diverted to an off-channelreservoir only during lower flows, when water isrelatively sediment free, while allowingsediment-laden floodwaters to pass by in themain river. A sediment bypass can divert part ofthe incoming sediment-laden waters into atunnel around the reservoir, so they never enterthe reservoir at all, but rejoin the river below thedam. Sediment can also be sluiced bymaintaining sufficient velocities through thereservoir to let it pass through without allowing itto deposit. Alternately, the reservoir can bedrawn down to scour and re-suspend sedimentin the reservoir and transport it downstream.This involves complete emptying of thereservoir through low-level gates. Densitycurrent venting makes use of the higher densityof sediment-laden water. Opening dam bottomoutlets when denser turbidity currents passthrough the reservoir can maintain them intactand allow them to exit the reservoir via theoutlets, carrying most of their sediment withthem. Sluicing, flushing, and density currentventing pass sediments in suspension, whichtend to be the finer fractions of the sedimentload but can include significant sand. Sluicingand flushing work best on reservoirs that arenarrow, have steep channel gradients, andhave storage that is small relative to the riverflow. Otherwise, back water zones might form inwider reservoirs where the hydrodynamic forcesare insufficient to mobilize sediment. Flushinghas been effective on reservoirs that impoundless than 4% of the mean annual inflow[18]

(Figure 3). Large reservoirs with year-to-yearcarry-over storage are poor candidates for suchsediment pass-through approaches.

It is generally most efficient to take sedimentmanagement into account at the outset of thedesign and planning the operation of dams, sothat dams are equipped from the outset tosuccessfully sluice or slush sediment (e.g., withsufficiently large low-level outlets), and theoperations are planned to account for someperiods of reduced power generation (or otherfunctions) to allow sediment to be passed.Retrofits to allow sediment passing through

existing dams may be possible, but often raisesafety concerns. Bypasses can be safely builtaround existing dams without threatening theintegrity of the dam.

Minimizing sediment trapping throughstrategic dam planningStrategic dam planning at the river basin scaleis an often-overlooked opportunity to minimizesediment trapping in dams, with benefits for the

Figure 3. Plot of projects from diverse environments and with different sediment management strategies(flushing (squares), sluicing (triangles), excavation/dredging, check dams or no strategy (circles)).Reservoir life is indicated by the ratio between the reservoir storage capacity and the mean annual inflowsediment to the reservoir. Successful implementation have been in cases characterized by impoundmentratios (reservoir storage capacity divided by mean annual runoff to the reservoir) of 0.04 or less. Usingthe data, a simple linear regression relates the reservoir life to the impoundment ratio (linearity coefficienta = 2.45×103). (Figure developed by Tetsuya Sumi, adapted from Kondolf et al.[17], used by permissionof AGU/John Wiley & Sons)

Figure 2. Gravel replenishment below Keswick Dam. To balance the sediment starvation created bytrapping in Shasta and Keswick Dams, gravel is deposited from dump trucks down the bank of theSacramento River, creating a cone to be eroded by subsequent high flows. (a) Remote-sensing compositeimage of site, showing gravel pile emplaced (15 April 2015), and (b) subsequently eroded (24 May 2017)(Google Earth). (c) Gravel augmentation has been ongoing here for decades, as reflected in a much-reproduced photo from January 1989 by Kondolf.

RESERVOIRSEDIMENTATION

71hydrolink number 3/2018

sand load. The actual portfolio built to date isthe result of project-by-project construction ofdams, without a strategic trade-off analysis orplanning (Figure 4). As a result, the current damportfolio produces 51% of the basin’s hydro-electric capacity while trapping 91% of its sandload, mostly because of early construction ofdownstream dams in the Sre Pok and Se Sanbasins[19] (Figure 4), the tributaries contributingmost of the basins sand load[20], with highsediment trapping and very little potential forsustainable sediment management. The currentportfolio, resulting from project-by-projectdevelopment, has also similar generation coststhan the optimal alternatives[19]. In an effort topreserve remaining connectivity of sedimentsources in the basin, the Natural HeritageInstitute (as US-based NGO) and the NationalUniversity of Laos developed a plan (adoptedby the Laotian government) to site newhydropower dams in the Se Kong River basinonly upstream of existing dams. The planfollows a strategic analysis for planned and builtdams to minimize additional sediment trappingin the basin[21]. The example of the lowerMekong tributaries is a call for action for thestakeholders involved in planning and financingthe global boom in dam development.

Compared to the current ad-hoc development ofindividual dams, strategic planning will involvemore careful, basin-scale assessments of damimpacts and benefits. It might also result insituations, where different objectives, such asfish-migration and sediment transport, or thenational interests of riparian countries to eachmaximize their generation, are in conflict.However, our increasing ability to model manydomains of river ecologic and morphodynamicprocesses on network scales allows us toevaluate many different planning alternativesand to take informed decisions regarding whichproject portfolio to develop.

Unfortunately, most dams have been (andcontinue to be) built on an individual, project-by-project basis, without analysis of cumulativeeffects of multiple dams on a river network,much less strategic planning to minimizeimpacts. In these cases, maintaining habitatdownstream of dams could involve a combi-nation of morphogenic flows, sediment augmen-tation, and adding large wood. Especially wherenew dams are build, decision makers should beaware that such measures can provide somemitigation but will also require continuous invest-ments to provide lasting improvements ofecologic conditions. Strategic planning mighthence require to forego developing someprojects with the largest short term economicreturn from a perspective of reducing costs ofmitigation measures over the decadal life-time ofsingle dams. For very large rivers, such as theMekong, cumulative dam sediment trapping andthe related impacts on the river system might,however, well exceed what can be possiblymitigated with such approaches mostly testedfor smaller rivers in temperate climates. Wheremitigation measures are feasible, a simplesediment budget and assessment ofgeomorphic processes and habitat conditionsshould be conducted before undertakingrestoration actions. The sediment budget shouldcompare downstream sediment supply withenergy available to transport it, to ascertain if thereach has a sediment deficit or surplus, and towhat degree. Likewise, assessing post-damchannel adjustments and their implications foraquatic habitat will inform potential options for

restoration. n

AcknowledgementsThis article is adapted in large measure fromparts of the chapter ‘Dams and ChannelMorphology’ by G.M. Kondolf, R. Loire, H.Piégay, and J.-R. Malavoi in the forthcoming

dam infrastructure and the downstream riversand coasts. Such planning should involverecognizing the spatial heterogeneity in naturalsediment transport, cumulative effects onsediment supply of multiple dams in a rivernetwork and consequent geomorphicimpacts[19]. New dams should be located insuch a way, that the final dam portfoliominimizes disruption of sediment transport. Inaddition, each individual dam should bedesigned to maximize its ability to passsediment around or through the reservoir[20].Overall, there is large, but so far mostly missed,potential to develop and manage dams moresustainably for both reservoirs and rivers.

Throughout the developing world there is anexplosion of dam building, motivated largely bya push for hydroelectricity, with an anticipateddoubling of global hydroelectric capacity withinthe next two decades. As demonstrated for themajor downstream tributary of the MekongRiver (the Sre Pok, Se San, and Se Kongsystem, drainage basins located in Laos,Cambodia, and Vietnam), strategic damplanning could have resulted in a dam portfolioproducing 68% of the basin’s hydroelectricpower potential while trapping only 21% of its

IAHR

Figure 4. Power generation and sediment trapping from dam building in the Sre Pok, Se San, and SeKong rivers (the ‘3S basin’), the largest downstream tributary to the Mekong River. (a) The current 3Sdam portfolio includes twenty-one (21) dams built or under construction (black squares), and twenty-one(21) more at various planning stages (white diamonds). (b) Increased power generation capacity andcumulative sediment trapping with construction of the current dam portfolio and alternative portfolioswith an optimal trade-off between sediment trapping and power production (grey circles). The arrowindicates a dam portfolio with higher power production but lower sediment trapping compared to thecurrent portfolio (see arrow). Optimal portfolios were identified based on analysis of 17,000 alternativedam portfolios (not shown). The optimal portfolio compares favorably to the currently planned devel-opment because of a different spatial configuration of dams in the network. (c) The current dam portfolioincludes dams downstream in the Sre Pok and Se San. (d) The alternative, optimal portfolio relies moreon dams in the headwaters and on lower sediment-yield portions of the basin. The optimal portfoliogreatly reduces environmental impacts and reservoir sedimentation, and also produces higher economicbenefits.

Continues in page 76

RESERVOIRSEDIMENTATION