SCOR Working Group 122 Mechanism of sediment retention in estuaries

Transcription

1 SCOR Working Group Mechanism of sediment retention in estuaries 3 5 September, 7 Sediment transport, erosion and deposition in Estuaries on different time and space scales C. L. Amos, C. Thompson, M. Villatoro, V. Venturini, R. Helsby G. Umgiesser, L. Zaggia, F. Zonta, A. Mazzoldi Centre for Coastal Processes, Engineering and Management University of Southampton, Southampton Oceanography Centre Southampton, UK ISMAR-CNR S. Polo 364,Venezia Italy

2 Space scale Global Regional Bottom-up modelling (SEDTRANS) Local days Hybrid modelling (PONTOS) years Top-down Modelling (SEDFLUX) Decades - centuries Time scale

3 management Modelling in sediment dynamics Regional geophysical surveys multibeam (swath) bathymetry Geophysical surveys Numerical simulation SEDFLUX PONTOS creation of hypotheses formulation of scientific questions Sampling and verification (Paleoceanography and climate change ) Numerical simulation SEDTRANS DELFT -3D Mike Sediment Dynamics In situ instrumentation Laboratory simulation Fundamental principles Numerical codes

4 Bottom-up modelling sediment transport in estuaries Types of models box models -D models (time, distance) -D models (x,y, depth-averaged; time versus x-axis) 3-D models (x,y, time; x,y,z) 4-D models (x,y,z, time) Box models - monitor continuity of mass with simple input/output controls simple and effective in sensitivity analyses, can be time-stepped easily no hydrodynamics, spatially-averaging, poor open boundaries -D models - spatial gradients in sedimentation, concentration easily set up, minimum hydrodynamics, offer practical solutions poor open boundary conditions -D models - provides links between sedimentation and hydrodynamics valuable insights into fundamental processes, engineering solutions possible labour intensive (99% rule), difficult to calibrate, difficult to programme

5 Short term estuarine retention Best method direct measurements + numerical/physical simulations + experience Next best method direct measurements + numerical simulations + experience Next best method direct measurements + experience Next best method numerical simulation + experience Etc, etc Long term predictions will become within reach when supercomputers are more generally available fundamental problems which remain to be solved are the schematization of the boundary conditions Calibration and verification of mathematical models require a detailed set of synoptic data. Much effort should be put in field surveys to obtain these data. (van Rijn, 999)

6 Fundamental equations Sediment continuity equation (3-D): (Closed system) S t + U S + V S + W S + WS s = x y z Mass continuity equation (3-D) (Exner): (Open to the bed) b s z ( e) = WS s t WS= D E where D is net deposition, E is net erosion, S is suspended sediment concentration, e is sediment porosity, b is sediment unit weight, h is bed elevation, W s is mean settling rate, and U,V,W are the three components of mean flow.

7 Instrumentation development to calibrate models - suspension benthic landers (RALPH) benthic flumes (Sea Carousel) remote sensing tools (backscatter) self-contained packages (Zedhead) acoustical optical electro-magnetic/peizoelectric x-ray time-series, usually from fixed location () Velocity field (3-D) () S t S t + U S + V x + W S + F s Mass settling rate (3) S + ( W W y + Fn = s S ) z Sediment concentration gradients (3-D) (6) Benthic flux (4) External sources/sinks (5) The sediment continuity equation = sediment budget

8 Morphological-dynamics and sediment dynamics suspension and bedload () only possible in closed system!!! (mesocosm) () bottom sampling (3) acoustic velocity, resistivity, CT scanning (4) acoustic altimeter, -axis sector scanning sonar, repetitive swath bathymetry (5i) Deposition (D) from cores (Pb or Cs 37 ) or sediment traps or altimeter measures (5ii) Erosion (E) (closed systems) or acoustic profiling or long-term swath bathymetry (6) Sediment traps, geophysical surveys of bedform migration Benthic flux () Sediment density () Porosity (3) Time-series of bed level change (4) W S + F F s = g s = D E = ( e ) h t Divergence in bedload transport rate (6) Deposition(D) Erosion(E) (5) The Exner equation

9 AGE The time that has elapsed since the water parcels enters an estuary ; Bolin and Rodhe (973). TRANSIT TIME The period between the times a particle entering and leaving an estuary ; Bolin and Rodhe (973). EXPOSURE TIME The total amount of time a particle spends in the domain ; Monsen et al. (). RESIDENCE TIME The time it takes for any water particles of the sample to leave the lagoon through its outlet to the sea ; Dronkers and Zimmerman (98); Prandle (984). The time required for the total mass of a conservative tracer originally within the whole or a segment of the estuary to be reduced to a factor /e ; Sanford et al. (99); Luketina, (998).

10 (input to estuary) /year 3 Mm (erosion from est Western Scheldt estuary (volumetric changes of sand) Years cm/century SL rise 6 cm/century SL rise Prior to 986 export of sand from estuary to ebb tidal delta After 986 import of sand into estuary

11 Western Scheldt, Belgium DELFT-3D Flood and ebb dominance due to interplay of Semi-diurnal Amplitude ratio: M /M 4, M /M 6 Phase shift: ( 4 ), 3( 6 ) Major change in inner estuary Major changes on central estuary Major changes in inner estuary Changes throughout estuary

12 Sedimentation/erosion import/export

13 The open boundary problem The open boundary offers the greatest difficulty to measure and to predict accurately note the change from importing to exporting with time import and export linked to S and more importantly to S(x) an estuary can change trapping efficiency depending on S gradient caused by wave resuspension dredging tidal variations construction, etc

14 Sensitivity analysis of sediment flux rates at open boundary S - sediment concentration dominates flux increases as S increases W s - flux increases with increased settling rate U d - deposition threshold flux increases as threshold increases U e - erosion threshold flux increases as threshold increases NOTE: S dominates the signal so should be the focus of measurement, input and calibration in model

15 4 4 3 M+M4 no phase shift M+M4 45 degree advance M+M4 45 degree retard 3 M+M6 no phase shift M+M6 45 degree advance M+M6 45 degree retard - - Tidal amplitude Time (hours) Tidal amplitude Time (hours) 3 M+M4 (no phase advance) M+M4 (45 phase advance) M+M4 (45 phase retard) 4 3 M+M6 (no phase advance) M+M6 (45 phase advance) M+M6 (45 phase retard) - - Relative t - Advance phase = increase flood transport Retard phase = increase ebb transport Time (hours) Relat - Advance phase = increase flood transport Retard phase = increase flood transport Time (hours)

16 tidal constituent M M+S M+S M+S M+S+M4 M+S+M4+M6 M+S+M4+M6 M+S+M4+M6 M+S+M4+M6 M+S+M4+M6 M+S+M4+M6 M+S+M4+M6 phase (S) -45(S) (S) -45(S) 45(M4) -45(M4) 45(M6) -45(M6) relative transport E

17 tidal constituent M amplitude(m). phase 35 period(hours).4 S.7 3 Mean tidal flow (m/s) (flood) Time (hours) sand transport rate (97-973) sand transport rate (99-9) N K K O P M S N K (ebb) K O P Relative t : potential net export = 9. x 4 m 3 year 9: potential net export = 7.9 x 4 m 3 year Time (hours)

18 Mean tidal flow (m/s) (flood) Time (hours) sand transport rate (97-973) sand transport rate (99-9). -.5 (ebb) -. Relative t : potential net export = 9. x 4 m 3 year 9: potential net export = 7.9 x 4 m 3 year Time (hours)

19 R l ti h i ht. Surface sampling Towed sand trap Valeport OBS Sand trap (Lido Inlet, September 6) Epi-benthic sand trap Benthic sand trap.... Mean concentration (mg/l) profile profile profile 3 profile 4 profile 5 profile 6 profile 7 profile 8 profile 9 profile profile profile. Lido Sud (fixed ADCP site) Shaded region is the shadow zone of ADCP PROFILE # R-squared /m b log C = ( z) ( z) m h z = H a = mlog( C) + b b m Suspension number Our results /m = -.93 C = concentration (sand) Z = relative height h = sample height H = water depth a = reference conc. height m W U Ws = ( U s * * = k =.4; = ).93

20 .6.4 A - Bedload transport rate - Chioggia inlet G s =.6( U U crit ) 3 kg / m / s( Chioggia).. G s =.5( U U crit ) 3 kg / m / s( Lido) Gadd et al. (976) relationship for fine sand coefficient from.73 to 7.33 (wave-dominated shelf) Ws/U* is evaluated as.4*m = Excess current velocity (U-U crit ) 3 B - Sand concentrations - Chioggia inlet PROFILE U (m/s) STAGE OF TIDE /m 3.77 EBB -.49 R l ti h i ht.. Profile Profile 3 Profile 4 Profile 5 Profile 6 Profile 7 Profile 8 Profile 9 Profile Profile Profile FLOOD FLOOD FLOOD FLOOD EBB EBB Sand concentration (mg/l).6.68 EBB EBB

21 Lido Inlet - September, 6 sand trap survey Friction velocity (U*, m/s).3 surface sand epibenthic sand benthic sand Ws/U * given by van Rijn et al. (997) as.5 (turbulent rough) Ws/U* given by Bagnold (966) as.5 (turbulent rough) Using measured values from Lido yields a value of.38 (turbulent smooth) Using measured values from Chioggia yields a value of.9 (turbulent smooth) Settling Settling rate measured in NOC sedimentation column (.8 m) U * derived from TKE method on ADV (5 Hz) time-series (75 data points) Linear (inverse) relationship between friction velocity (U * ) and settling rate (W s ) Inverse relationship i.e. Faster flows have finer grains!!! W U W U s * s * =.38 (Lido) =.9 (Chioggia)

22 Summary of results (Ws/U*) Venice lagoon, 6 Measured from SSC profiles Measured from samples D* = Lido Inlet Chioggia Inlet.9.3.

23 * s /U W. D* Roe (7) natural beach sand (sieved) epi-benthic, Venice lagoon, Lido (6) van Rijn (993), (D*/4) Bagnold (966) (V'up =.56V'), (D*/8) Bagnold (966) (V'up =.6V') (D*/) benthic, Venice lagoon, Lido (6) Benthic, Chioggia (6) Epi-benthic traps, Chioggia (6) Surface traps, Chioggia (6) Blair (7) Jamaica beach carbonate sand (susp Blair (7) Jamaica beach quartz sand (suspens Ridley (4) natural sand (suspension) Ridley (4) fish feed pellets Ridley (4) bakelite granules D* is dimensionless grain diameter Ws = still water fall velocity U* = critical friction velocity D W U * s * W U s * = = crit, susp crit, susp s ( ) g D *, D * = const., D = D * = const. s < * > W gd s s 5 W gd.333, D s 5 *, D d 5 < * >

24 Ws/U* = /D* Lower than van Rijn (4/D*) Critical Shields parameter for suspension lower than Bagnold (966) Yields good separation between bedload and suspended load samples from Lido inlet. crit.5, susp =.43D*, D* <

25 Laboratory controls Lido inlet, Venice lagoon (all samples) Chioggia inlet, Venice lagoon (epi-benthic samples) Jamaica beach sand suspension threshold. * s /U W ( d. 5) * d 5 (mm) W s = U

26 Fundamental issues and questions regarding estuarine retention short term prediction Can we accurately define the transport pathways, depo-centres, and concentration profiles of () fines, and () sands in estuaries? Do we understand the links between sand and fines along a transport pathway? Can we predict morphodynamical evolution? Do we understand the link between morphodynamical changes and hydrodynamics (waves and tidal currents)? What is the true relevance of biological feedbacks to estuarine evolution What controls the trapping efficiency (F) of an estuary? Estuarine filtering efficiency (F) where Q loss is the net export from the inlets while Q total is the total mass balance and Q totsl = Q onshore + Q offshore + Q longshore F Qloss = [ ] Q total

27 Fundamental issues and questions regarding estuarine retention long term prediction Largely empirical based Can we accurately define the morphological evolution of an estuary? Does predicted morphological change fit with regime based theories? Does predicted evolution fit with theories of sediment dynamics? What are the long-term boundary conditions (sediment type, supply)? Hard to differentiate causality (processes) easier to programme and faster to execute Can be used for what if scenarios (such as SLR) Requires expert validation and interpretation

28 Shallow Water Finite Element Model (SHYFEM) THE NUMERICAL MODEL U t V t fv + fu U + t t gh gh V t + RU x + RV y = + F + F x y = = LIDO INLET GULF OF VENICE LIDO INLET U V = = h h uz vz Barotropic transport MALAMOCCO INLET MALAMOCCO INLET c H b R = u + v Friction term CHIOGGIA INLET F p x H = p ^ H + f x dz + x h = ps + g z ()dz z B W ( ) x x Wind stress, non-linear advective terms... CHIOGGIA INLET MODEL DOMAIN F.E. grid of the Adriatic Sea and Venice Lagoon (87 nodes, 569 elements, spatial resolution varying from 3 Km to 5 m)

29 THE TRANSPORT TIME SCALES RESIDENCE TIME (Eulerian approach) THE TIME REQUIRED FOR EACH ELEMENT OF THE DOMAIN TO REPLACE THE MASS OF A CONSERVATIVE TRACER, ORIGINALLY RELEASED, WITH NEW WATER TRACER [C] IN THE LAGOON = % S = S (t = ) TIDE AND WIND ACTION DRIVES IT OUT THE BASIN S S S us vs + + = K H + t x y y x TIME DECAY OF THE TRACER CONCENTRATION DEFINITION OF THE REMNANT FUNCTION (Takeoka, 984 a,b) r ( x, y, t ) = S ( x, y, t ) S DEFINITION OF THE twater RESIDENCE TIME (x, y ) = r ( x, y, t )dt

30 A) Station V time (minutes) B) Station V Pa Initial profile of ce SSC measured SSC predicted by Sedtrans5 3 5 Pa SSC measured 4.5 SSC predicted by Sedtrans5 3 Initial. profile of ce time (minutes) A comparison of the cohesive sediment algorithm with field data collected with the Sea Carousel in Venice lagoon (Stations and 3, February 999). The initial profile of critical erosion threshold t ce, and the time-series of applied bed shear stress t, measured SSC, and SSC predicted by Sedtrans5 are shown.

31 . (A) Engelund-Hansen SIB93 SIB8 Venice. (B) Einstein-Brown SIB93 SIB8 Venice. (C) Bagnold SIB93 SIB8 Venice... E-3 E-3 E-3 E-4 E-4 E-4 E-5 E-5 E-5 E-5 E-4 E-3.. Q measured [kg/ms] E-5 E-4 E-3.. Q measured [kg/ms] E-5 E-4 E-3.. Q b measured [kg/ms].. E-3 E-4 E-5 (D) Yalin SIB93 SIB8 Venice E-5 E-4 E-3.. Q measured [kg/ms].. E-3 E-4 E-5 (E) Van Rijn SIB93 SIB8 Venice E-5 E-4 E-3.. Q measured [kg/ms] A comparison between the measured rates of sediment transport and the rates computed according to the five non-cohesive transport equations in Sedtrans5. Solid dots are data from the 993 deployment over medium sand (SIB93), triangles are data from the 98 deployment over fine sand (SIB8), and open circles are data from Venice (6). The solid line indicates perfect agreement; the dashed lines represent the factors.5 and.

32 THE RETURN FLOW FACTOR (b) FROM TIDAL PRISM METHOD: THE AVERAGE WATER RESIDENCE TIME IS DEFINED AS: THE TRANSPORT TIME SCALES I av TV av = ( b)p b IS THE RETURN FLOW FACTOR. IT EXPRESSES THE FATE OF THE WATER ONCE IT IS OUTSIDE THE EMBAYMENT DEFINITON OF THE RFF (b) SIMULATION I, THE MASS OF THE TRACER EXITED THE LAGOON CAN RETURN TO THE EMBAYMENT. Av IS COMPUTED II SIMULATION II, THE MASS OF THE TRACER EXITED THE LAGOON IS SET TO ZERO, (b=). IS COMPUTED b ( x y) TV = av P THE RFF IS DEFINED AS: ( x, y) ( x, y) ( x, y) =, av = b

33 SCENARIO: TIDE AND SIROCCO WIND (7 m/s) LAGUNA DI VENEZIA 6 m**3/s -6 m**3/s 5 m**3/s -5 m**3/s 74 m**3/s -4 m**3/s MARE ADRIATICO 74 m**3/s

34 SCENARIO: TIDE AND SIROCCO WIND (7 m/s) TRANSPORT TIME SCALE WRT Total NBn 6 ± 8 NBc 38 ± 9 CB 3 ± 6 3 ± WTT 8 ± 9 37 ± 4.3 ± 6 ± 8±6 ± 6 6 ± 7 RFF SB ± 5 3 ± 6 ± 5

35 SCENARIO: TIDE AND BORA WIND ( m/s) LAGUNA DI VENEZIA -73 m**3/s 73 m**3/s -4 m**3/s 4 m**3/s -9 m**3/s -4 m**3/s MARE ADRIATICO -9 m**3/s

36 SCENARIO: TIDE AND BORA WIND ( m/s) TRANSPORT TIME SCALE Total NBn NBc CB SB WRT 4±4 3± ± 5±3 7±3 WTT 4± 9±4 4± 3± ± RFF

37 RESIDENCE TIME VS TRANSIT TIME TRAPPING FACTOR SIROCCO + TIDE Total NBn NBc CB SB WRT 6 ± 8 38 ± 9 3 ± 6 ± 6 ± WTT 8 ± 9 37 ± 4 ± 6 ± 5 3 ± 6 BORA +TIDE Total NBn NBc CB SB WRT 4±4 3± ± 5±3 7±3 WTT 4± 9±4 4± 3± ± THEY SEEM TO BE CORRELATED! BORA + TIDE

38 SCENARIO: TIDE AND SIROCCO WIND (7 m/s) SIROCCO + TIDE Total NBn NBc CB SB WRT 6 ± 8 38 ± 9 3 ± 6 ± 6 ± WTT 8 ± 9 37 ± 4 ± 6 ± 5 3 ± 6 V3 V CF =.9

39 SSC measured SSC predicted by Sedtrans5 3 PD % =.46 F = 6% V4 SC.8.4 PD % =.3 F = 93% V6 SC PD =.87 F = 69% % PD =.35 F = 54% % PD =. F = 78% % PD =.64 F = 69% % V SC V SC V3 SC V3 SC PD % =. F = 7% PD % =.7 F = 83% PD % =.5 F = % PD % =.74 F = 99% V5 SC V5 SC V6 SC V7 SC PD % =.98 F = 95% PD % =.94 F = 9% PD % =. F = 88% PD % =.97 F = 6% V8 SC V MF V3 MF V4 MF A comparison of the cohesive sediment algorithm with field data collected with the Sea Carousel (SC) and the field MiniFlume (MF) at different stations in Venice lagoon, February 999. For each experiment, the time-series of SSC measured and predicted by Sedtrans5, the standard deviation of the proportional difference (s PD ), and the time percentage when the difference is less than % (F % ) are shown..5 3 PD =.44 F = 8% % V4 SC.5..5 PD % =.85 F = 98% V7 SC PD % =.6 F = 99% V5 MF Time (minutes)

40 3 Amplitudes equal, M+S 4 3 M tide only M+S M+S+M4 M+S+M4+M6 Tidal amplitude - - M tide only M+ S tide Time (hours) Tidal amplitude Amplitudes equal, no phase differences Time (hours) M+M4 (no phase shift) M only (no residual) M+S (neap/spring modulation) with first overtide with second overtide Relative t Time (hours) Relative t Time (hours)

41 R l ti h i ht. Sand trap (Lido Inlet, September 6) Surface sampling Towed sand trap Valeport OBS Epi-benthic sand trap Benthic sand trap.... Mean concentration (mg/l) profile profile profile 3 profile 4 profile 5 profile 6 profile 7 profile 8 profile 9 profile profile profile. Lido Sud (fixed ADCP site) Shaded region is the shadow zone of ADCP PROFILE R-squared offset /m log C = ( z) ( z) m = mlog( C) + b b m Suspension number Our results /m = -.77 Suspension throughout water column at peak flows Stratification at Ht and LT m W U Ws = ( U s * = m.7 m =.56; k * ).77 =.4; =