Sustainable Hydraulics in the Era of Global Change – Erpicum et al. (Eds.)© 2016 Taylor & Francis Group, London, ISBN 978-1-138-02977-4

Sediment transport in the Schelde-estuary: A comparison betweenmeasurements, transport formula and numerical models

Y. PlanckeFlanders Hydraulics Research, Antwerp, BelgiumAntwerp University, Antwerp, Belgium

G. VosAntea Group, Antwerp, BelgiumFlanders Hydraulics Research, Antwerp, Belgium

ABSTRACT: The Schelde-estuary serves different estuarine functions and therefore faces managers withmultiple challenges: increasing tidal propagation vs. safety against flooding; sedimentation in the navigationchannel vs. port accessibility; changing dynamics vs. ecology. Within the Flemish-Dutch Long Term Vision forthe Schelde-estuary, a 4 year (2014–2017) research programme was defined, in which 8 topics will be dealt with(e.g. tidal penetration, risk for regime shift, sediment strategies, valuing ecology). Two fundamental tools arecrucial in answering the different questions towards the future management of the estuary: expertise/systemunderstanding and numerical models. Where the numerical models reproduce the hydrodynamics reasonablywell, sediment transport and the resulting morphological changes is still a big challenge. Therefore an extensivemonitoring campaign was performed in 2014, during which both hydrodynamic and sediment transport measure-ments were performed in the Schelde-estuary.The data allows to validate the existing numerical models, allowinga better assessment of the possibilities and limitations of the present numerical models. This paper describes thevalidation of a 2D numerical model, that is used to optimize the relocation strategy of non-cohesive sedimentsin the Beneden-Zeeschelde. The comparison of sediment transport of field data and numerical model resultsshow a rather promising agreement (i.e. differences of factor 2 to 3) for the Beneden-Zeeschelde. However,important differences (both in patterns, intensities and between different formula) were found when comparingtopo-bathymetric changes predicted by the morphological numerical model and observed bathymetric changes.

1 INTRODUCTION AND BACKGROUND

In order to supply managers with adequate answers,research tools (both expertise/system understandingand numerical models (Peters et al., 2006)) are crucialin answering the different questions. Where the numer-ical models reproduce the hydrodynamics reasonablywell, sediment transport and the resulting morpholog-ical changes is still a big challenge. These limitationstranslate into uncertainties of the results, and trig-ger the precautionary principle within environmentalimpact and appropriate assessment (within the scopeof the European Bird and Habitat Directive).

In order to collect appropriate datasets for validationof numerical models and to improve our system under-standing, good sediment transport measurements arevital. Nevertheless, measuring sediment transportremains one of the most challenging aspects in riverengineering (Plancke et al., 2012a; Thant et al., 2016).

Within the scope of a common Belgian-Dutchresearch programme (“Agenda for the future”), whichdeals with system management challenges for thenear future (e.g. tidal penetration, risk for regimeshift, sediment strategies, valuing ecology), additional

measurements were performed in order to vali-date state-of-the-art numerical models. This paperdescribes the measurements results, a comparison withclassic sediment transport formula and the validationof the numerical models.

2 THE SCHELDE-ESTUARY

The Schelde-estuary is a macro-tidal estuary with alength of 180 km in Flanders and the southern partof the Netherlands (Figure 1). The Vlakte van deRaan (“mondingsgebied”) connects the estuary withthe North Sea and should be seen as an integral partof the estuary. This part is a shallow water area withseveral channels. The Vlakte van de Raan (–20 KM to0 KM) connects to the Westerschelde (KM 0 to KM60), which has a multiple channel system, with ebband flood channels and intertidal sandbars in between.More up-estuary, near the Dutch-Belgian border, themorphological system changes into a single channelsystem, the Zeeschelde (KM 60 to KM 160).

The estuary is characterized by semi-diurnal tides,causing ebb and flood currents with importantsediment transports of both cohesive and non-cohesive

498

Figure 1. Overview of the Schelde-estuary and differentlocations where field measurements were performed.

sediments. The Schelde-estuary serves different estu-arine functions and therefore faces managers withmultiple challenges: increasing tidal propagation vs.safety against flooding; sedimentation in the naviga-tion channel vs. port accessibility; changing dynamicsvs. ecology.

To guarantee port accessibility, dredging and relo-cation works are performed on a daily basis. Inorder to perform these works, licenses are grantedby the responsible governments (Dutch governmentsin the Westerschelde, Flemish government in theZeeschelde). Dredged sediment are relocated withinthe estuary, in order to minimize possible effects.Within the Westerschelde mostly sand is dredged,while in the Beneden-Zeeschelde (near the Antwerpport area) both sand and silt/mud are dredged duringmaintenance works.

Although annual discharge and sediment transportmeasurements (going from continuous SPM mea-surements at several points to yearly sailed transectmeasurements) are performed within the scope ofthe MONEOS programme (Plancke et al., 2012b),additional measurements were necessary for modelvalidation.

3 METHODOLOGY

3.1 Field measurements

In 2014, several measurement campaigns were per-formed, during which both hydrodynamic and sedi-ment transport measurements were performed in theSchelde-estuary. At more than 10 locations (Figure 1)measurements were executed over a full tidal cycle(13h): 3 locations at the Vlakte van de Raan, 7 loca-tions at the Zeeschelde and 1 location at the Rupel, atributary of the Zeeschelde.

At each location one or two vessels were used.The first vessel was anchored during the measure-ment period, performing measurements at a fixedpoint in the estuary. Currents were measured usingADCP (vertical profile) and Aanderaa Current Meter(point), while sediment transport was measured using

both direct (Delft Bottle and pump samples) andindirect (OBS,ABS) techniques.The Delft Bottle tech-nique was used both near-bed (using frame) and inthe water column (suspended), at four different posi-tions (bed + 20 cm, bed + 40 cm, bed + 100 cm andbed + 200 cm). Measurements were executed contin-uously, with sampling times varying from 3 min-utes (at peak transport) to 15 minutes (near slackmoments). From these measurements total transportswere derived every 30 minutes. At those locationswhere a second vessel was available, additional tran-sects were sailed using ADCP (current and sedimenttransport fromADCP-backscatter). For the locations atthe Vlakte van de Raan, additional frames were placedat the bed, allowing long term (4 weeks) measurementsof hydrodynamics (currents and waves) and sedimenttransport.

The measurements are used to validate the availablenumerical models.A first project deals with large scalesediment management issues in the down-estuarinepart, Vlakte van de Raan (Van der Werf et al., 2015).A second project focusses on management strate-gies for the most up-estuarine part, Boven-Zeeschelde(Vanlede et al., 2015). A third project will detail thefuture sediment strategy in the Beneden-Zeescheldeand will be discussed in this paper. Both Delft3D andTELEMAC models are used in these different studies.

3.2 Numerical modelling

In order to receive a new license to relocate dredgedsediments in the Beneden-Zeeschelde, an appropriateassessment is necessary. Within this assessment thepossible effects of the relocation strategy must be eval-uated, with regard to the reference situation (presentstrategy). A preliminary study was performed by Flan-ders Hydraulics Research to optimize the relocationstrategy, both for sand and silt/mud, taking into accountpresent and future challenges (tidal penetration, possi-bility of a regime shift) and minimizing possible effectson primary production, birds and fishes, etc.

A 2D numerical model of the Beneden-Zeescheldewas set-up in Delft3D, based on existing state-of-the-art NeVla-model (Figure 2). Initially, it was opted tomake a distinction between the cohesive (silt/mud) andnon-cohesive (sand) sediment transport model, whichis state-of-the-art (Dam et al., 2013). Possible effectsare different: relocation of silt/mud will have a possi-ble effect on the suspended sediment concentration andtherefore will influence the primary production; relo-cation of sand will have an effect on the morphologyand therefore will influence the ecological valuablehabitats. However, as both sand and silt transport takeplace simultaneous in reality, the final model includedboth the cohesive and the non-cohesive sediment. Itwas found that mainly the cohesive sediment transportwas influenced by this approach, as the major part ofthe available transport capacity was consumed by thenon-cohesive sediment transport.

The numerical model has a grid resolution in thezone of interest of 25 m (L) by 15 m (W) (width of

499

Figure 2. Overview of model grid: existing NeVla grid(green) and detail (3 × 3 refined) grid for this study (red).

the estuary in this zone ∼500 m). It is driven by awater level time-series at the down-estuarine bound-ary, while a discharge time-series (generated from theNeVla-model) was defined at the up-estuarine bound-ary. A uniform Manning bed roughness was applied(0.025 m1/3/s), resulting in a good agreement of waterlevels in the study area.

A sensitivity analysis was performed on the sedi-ment transport model (both cohesive as non-cohesive)and the morphological model. As the additional mea-surements focused on non-cohesive transport, only thispart is discussed in the paper.

3.3 Sediment transport formula

Several formula are available to describe the sedimenttransport in rivers and estuaries. Within this study itwas chosen to use both Engelund-Hansen formula (1)and Van Rijn formula (2a and 2b)), which are bothsuited for non-cohesive sediment transport of finesand. Further information on these formula can befound in (Engelund & Hansen, 1967) and (Van Rijn,1993).

It can be seen that concept of these formula is differ-ent (total load qt vs. bed qb and suspended qs load) andthe different effect of both mean flow velocity u (power5 vs. power 3) and median grain size d50 (inverse vs.proportional) between these formula.

4 RESULTS

4.1 Field measurements

Figure 3 shows the hydrodynamic conditions for thedifferent locations in the Zeeschelde. The tidal char-acteristics change significantly along the estuary:the most down-estuarine locations of the Zeeschelde

Figure 3. Overview of hydrodynamic conditions: waterlevel (top) and flow velocities (bottom) in the Zeeschelde.

Figure 4. Overview of sediment transport in the Zeeschelde.

(Liefkenshoek, Oosterweel) have an almost sym-metrical tide; more up-estuary the tidal asymme-try increases, most pronounced at Schellebelle andSchoonaarde. The tidal range increases from the NorthSea up to Driegoten due to the funnel shape of theestuary. More up-estuary it decreases due to the damp-ing effect of the undeeper channels. The asymmetry inthe vertical tide is translated into the horizontal tide(flow velocities). Highest flow velocities are found atOosterweel (ebb and flood phase) and Driegoten (floodphase).

Figure 4 shows the sediment transport for the dif-ferent locations in the Zeeschelde. The sediment D50ranges from very fine sand (near-bed “DBF”) to veryfine silt or mud (suspended “SUSP”).

Sediment transport patterns are different for differ-ent locations, which is related to the position along the

500

estuary and its position on the transect. For Driegotenthe sediment transport shows a maximum 1 to 3 hoursafter low water. The sediment during this period ismuddy, with a lot of flocs. Where the flow velocitiesin this period are low, the peak is probably related to thetechnique of the Delft Bottle: while this technique issuited for sand, it was found that during period with lowflow velocities, currents through the bottle are insuf-ficient to transport mud through the bottle, leading toanomalies in the measured transport.

4.2 Numerical model

The numerical model was validated with the availablefield measurements. Where the simulated period wasnot identical to the moment of the field measurements(due to limited availability of boundary conditionsof fresh water discharge), a tidal cycle with simi-lar characteristics was searched during the simulationperiod. A good agreement was found, although smalldifferences are possible.

Figure 5 (top panel) compares the measured andmodelled flow velocities at Kruibeke. It can be seenthat during ebb (HW-360’ to HW + 30’) good agree-ment is found. During flood, peak velocities are overes-timated in the model.A possible explanation was foundin (1) a steeper rising of the tide in the model and (2)a difference in position along the cross-section due togrid size resolution (measurement was performed out-side the navigation channel and thus relatively close tothe bank (steeper bed slope)).

Figure 5 (bottom panel) compares the measuredand modelled sediment transport rates at Kruibeke.It should be mentioned that sediment transport mea-surements were executed at several (4) points over thevertical profile, and an assumption (profile derivedfrom relative profiles of all measurements divided in30’-blocks of the tidal cycle) had to be made to inte-grate this towards a transport over the full vertical.It can been seen that both the Engelund-Hansen andthe Van Rijn formula give similar patterns over time,with higher values from the Van Rijn formula at peaktransport. When model results are compared with themeasurements, some similarities can be seen, althoughdifference are present: At the start of the ebb phase(HW-300’) a peak is found in the measurements, whichdoes not occur in the model. However, it is not certainthat this peak is related to natural transport (an impor-tant increase in flow velocity coincides with this peak),or other factors (e.g. re-suspension due to ships) playa role. During the next phase of the ebb, a similar pat-tern is found, with modelled transport rates being twiceas high as measured values. During the first phase ofthe flood period (until HW + 200’), modelled trans-port rates are lower than measured values, and at peaktransport differences are even larger. This is proba-bly related to the difference in flow velocity at thismoment, as sediment transport rates relate to the 5thpower (Engelund-Hansen) of the velocity. Althoughdifferences exist, it was concluded that the model gave

Figure 5. Comparison between measurements and numer-ical model of flow velocity (top) and sediment transport(bottom) at Kruibeke.

Figure 6. Difference (left) and consistency (right) map ofbathymetry of Beneden-Zeeschelde over period 2009–2014[red = erosion, green = sedimentation].

a reasonable agreement with regard to non-cohesivesediment transport.

In the next phase, morphological updates were acti-vated in the model. To evaluate the model, severaltopo-bathymetric maps (interval of one year) of theZeeschelde were used and both difference and consis-tency (to identify yearly variation in sedimentation-erosion patterns) maps (Figure 6) were derived.

From these maps, it can be seen that several zones inthe Beneden-Zeeschelde are eroding (red), while otherzones are characterized by sedimentation (green).

501

Figure 7. Difference map (bathymetry) from numericalmodel using Van Rijn (left) and Engelund-Hansen (right)formula [red = erosion, green = sedimentation].

Morphological changes were computed by the numer-ical models (Figure 7), using both Engelund-Hansenand Van Rijn formula. It can be seen that (1) patternsdon’t show a good agreement, (2) that the intensityof bathymetric changes is larger in the model than inreality and (3) different formula give very importantdifferences in erosion-sedimentation patterns. It canbe seen that the Van Rijn formula predict erosion inthe channels and accretion near the banks, while theEngelund-Hansen formula predicts the opposite.

Based on these results, it was decided that sce-narios within the project to optimize the relocationstrategy could not be done using the morphologicalmodel. Therefore different scenarios were investigatedby making changes in the initial bathymetry (addingsediment to reproduce the relocation) and analyzingchanges in flow patterns and sediment transport rates.

5 CONCLUSIONS

In 2014, a series of field measurement campaignswere organized to collect datasets for numerical modelcalibration and validation with regard to sedimenttransport and morphology. In the past, the models havebeen extensively calibrated and validated for hydrody-namics, but due to lack of available measurement data,sediment transport was never really validated.As mea-suring sediment transport remains very challenging,results have an uncertainty due to (a) measurementstechniques (e.g. sensitivity of indirect techniques tosediment properties, bio-fouling), (b) measurementexecution (e.g. errors made during direct samplingin sample collection), (c) field conditions (externalfactors e.g. ships) and (d) data post-processing (e.g.calculation of vertical profile based on discrete data).

Within the scope of several projects, these newdatasets were used to validate numerical models.Within this paper the results of a validation wasdescribed for a 2D numerical model in Delft3Dused to optimize the relocation strategy of dredgednon-cohesive sediments in the Zeeschelde.

Sensitivity exercises have indicated the importantinfluence of several numerical parameters. The com-parison of field data and numerical model resultsshow a rather promising agreement (i.e. differencesof factor 2 to 3) for the Beneden-Zeeschelde. Duringcertain moments of the tidal cycle sediment trans-port patterns are reproduced rather well, while atcertain moment differences become larger. Where sed-iment transport is calculated with difference formulas(e.g. Engelund-Hansen, Van Rijn) using flow veloci-ties as an important parameter, minor differences inflow velocity are translated into large differences insediment transport.

Finally a comparison with observed topo-bathymetric changes was made using the morpho-logical module. Importance differences were found,both in patterns, intensities and between the 2 for-mula. Therefore, it was decided that scenarios withinthe project to optimize the relocation strategy couldnot be done using the morphological model. Differentscenarios were investigated by making changes in theinitial bathymetry (adding sediment to reproduce therelocation) and analyzing changes in flow patterns andsediment transport rates.

ACKNOWLEDGMENT

The field measurements were executed with hydromet-ric vessels from DAB Vloot. The crew and colleaguesperforming the measurements deserve a special wordof appreciation, as measuring a full tidal cycle (13h)can be very challenging.

Additionally Sanne Ceuppens and Matthias Dar-cis (Antwerp University) are mentioned as they madelarge contribution during the sensitivity analysis.

REFERENCES

Dam, G.; Poortman, S.; Bliek, B.; Plancke, Y. (2013). Long-term modelling of dredging strategies on morpho- andhydrodynamic developments in the Western Scheldt.

Engelund, F., Hansen, E. (1967). A Monograph On Sed-iment Transport in Alluvial Streams. Teknisk Forlag,Copenhagen, Denmark.

Peters, J.J.; Cunge, J.A.; Plancke,Y.M.G. (2006). Role of prac-tinioners and hydroinformatics in the decision-makingprocess for morphological management of an estuary, in:Gourbesville, P. et al. (Ed.) (2006). Proceedings of the7th International Conference on Hydroinformatics 2006,Nice, 4-8 September 2006. pp. [1–8].

Plancke, Y.; Paridaens, K. (2012a). Comparison of mea-surement techniques for monitoring sediment transportunder field conditions, in the Scheldt estuary, in: (2012).Hydraulic Measurements and Experimental Methods2012 Conference (HMEM 2012), Snowbird, Utah,August12–15, 2012. pp. [1–8].

Plancke, Y.; Vanlierde, E.; Taverniers, E.; Mostaert, F.(2012b). Monitoring of physical parameters withinthe scope of the Dutch-Flemish integrated monitor-ing program, in: (2012). Hydraulic Measurements andExperimental Methods 2012 Conference (HMEM 2012),Snowbird, Utah, August 12–15, 2012. pp. [1–6].

502

Thant, S.; Plancke, Y.; Claeys, S. (2016). Sediment transportmeasurements in the Schelde-estuary: how do acous-tic backscatter, optical transmission and direct samplingcompare? (proposed for the 4th IAHR Europe Congress).

Vanlede, J.; Smolders, S.; Maximova, T.; Teles, M.J. (2015).The unstructured Scaldis model: a new 3D high resolu-tion model for hydrodynamics and sediment transport inthe tidal Scheldt, in: (2015). Scheldt Estuary: physics andintegrated management - Special Session on of the 36thIAHR WORLD CONGRESS, 28 June–3 July, 2015, TheHague, the Netherlands.

van der Werf, J.; van Oyen, T.; De Maerschalck, B.; Nnafie,A.; van Rooijen, A.; Taal, M.; Tonnon, P.K.; Verwaest, T.;deVet, L.;Vroom, J.;Van der Wegen, M. (2015). Modelingthe hydro-morphodynamics of the mouth of the Scheldtestuary - Special Session on of the 36th IAHR WORLDCONGRESS, 28 June – 3 July, 2015, The Hague, theNetherlands.

Van Rijn, L.C. (1993). Principles of sediment transport inrivers, estuaries and coastal seas. Aqua Publication, TheNetherlands, Amsterdam.

503