DSD-INT 2014 - Delft3D Users Meeting - Keynote Lecture 2014 - Dynamic Deltas, Rudy Slingerland,...

1

Understanding Delta Dynamics using Delft3D

Rudy Slingerland

Penn State University

& Alberto Canestrelli, Doug Edmonds, Alex Burpee, Jim Cederberg, Jim Best, Dan Parsons, James Syvitski,

Bert Jagers, Aukje Spruyt, Mart Borsboom, Fei Xing, William Nardin, Sergio Fagherazzi

FESD DELTARES

2

Low concentration two-phase mixtures of water and sediment

particles subject to a free surface flow bounded by a granular

medium

Interface between flowing mixture and granular medium moves as a

result of a continuous exchange of sediment particles

Feedbacks between flow and interface produce the system

dynamics

The general problem of morphodynamics--- determine the motion of

the interface for given boundary and initial conditions

Deltas as a Morphodynamic System

--Giovanni Seminara

3

10 km

Lena Delta

Siberia


4


Gilgel Abay River Delta in Tana Lake

Ethiopia

5


Wax Lake Delta

Louisiana

6


Last Chance Delta

of the Ferron Sandstone,

Ferron, Utah

(90 Ma)

12 m

7

General Question

What morphodynamic processes and feedback

loops cause river-dominated deltas to self-

organize into these diverse natural geomorphic

forms?

NB: Discussion restricted to river-dominated deltas!

8

Our Conjecture

These variables determine:

a) number of distributaries

b) rugosity of shoreline;

c) frequency and shape of

delta lobes;

d) lengths and sinuosities of

channels; and

e) topographic roughness of

floodplains

These variables control a delta’s

internal stratigraphy

[McKeown et al. 2004]

Delta Form

(Q,Q , , ,A)s sf P Veg

s

s

QQP

Veg

A

where: = water discharge

sediment dischargeproportion of cohesive sediment

= a measure of sediment binding by vegetation accommodation space

9

Collect field observations to identify important

variables and possible feedbacks

Conduct morphodynamic numerical experiments to

determine causes and effects

The value of

Delft3D…..

Plan of Approach

10

How can we simulate flow and sediment transport at

minutes to hours to obtain deposits that accumulate over

hundreds to thousands of years?

Is a morphological scaling factor the only solution?

“Attempting to extract the dynamics at higher levels from

comprehensive modeling of everything going on at lower

levels is . . . like analyzing the creation of La Boheme as

a neurochemistry problem” (Paola, 2000).

Problem of “geologic” versus “engineering” time- and length-scales

11

Louisiana Deltas

5 km

Delta Attributes of Interest

• number of active channels

• distance between bifurcations

• bifurcation angle

• discharge ratio

• bar shapes

• planform

Important Features and Variables

bifurcation

a

b

c

levee

crevasse

turbulent plane jet L

W

a

Qb/Qc

12

5 km

1 4

3 2

A

A’

A A’

Fossil foreset

delta lobe sequence

channel facies foreset facies topset facies

foreset

concavity

topset

roughness

condensed

horizon rugosity

of shoreline

Important Features and Variables

Delta Attributes of Interest

• relative proportions of facies

• number of lobe sequences

• foreset shapes

• topset roughness

13

Start with observations from one delta of one sediment caliber

Study area: Mossy Delta of the Saskatchewan River No waves

No tides

Known history

A

14 1947

Growth of the Mossy Delta

1 km

15 1953

16 1968

17 2003

18

RMB

Distributary network is built by two processes: Bifurcations around mouth bars

Avulsions through crevasses

Network is inherited, but subsequently modified

This network (and 10 others) are organized such that: Bifurcation angle ≈ 65o

Distance to next BIF: L / W ≈ 14

Discharge ratios:

but RARELY 1

Observations

W = channel

width

Ql Qs

a

1.5 / 4b cQ Q

b c

19

Experimental design for building a mouth bar

seven computational layers are used in the vertical

constant basin depth (h = 5 m) with small amplitude (ah=0.05h)

random disturbances

no waves, tides, Coriolis acc., or salinity; D50 = 0.2 mm

constant water temperature

HLES model is used for horizontal turbulence

180 numerical experiments

Open boundaries on N, E, and S

Morphodynamic causes of this network structure

freshwater discharge

equlibrium sediment flux

1/2

1

1

1 1500 m

1 30 m

25 75 m s

0.5 4 m s

1 800

f

o

mor

B

h

c

U

f

outlet depths:

friction factors:

outlet velocities:

morph scaling:

outlet widths:

20

Morphodynamic causes…

Turbulent jet expands

and (sometimes) “flaps”

Suspended sediment is

advected to sides and

grows levees

Levees confine jet and

locus of jet expansion

shifts forward

Bed- & suspended-

loads build mid-channel

bar

When bar reaches

about 7/10s depth, back

pressure deviates

flowlines and bifurcation

forms

Depth-averaged velocity (m/s)

0

3

1

-1

0

-3

Bed elevation (m) 0

-3

Edmonds and Slingerland, 2007; Canestrelli, Nardin, Edmonds, Fagherazzi, and Slingerland (2014)

21

Morphodynamic causes… Not all jets are unstable (flap); it depends upon:

2

fc BS

hJet stability number, and distributary Reynolds No.:

Re o

B

U B

Canestrelli, Alberto, et al. "Importance of frictional effects and jet instability on the morphodynamics of river mouth bars and levees.

" Journal of Geophysical Research (2014)

22

Distance to river mouth bar, L (and therefore spacing of bifurcations) depends upon jet stability number, S

Bifurcation angle set by bar width when it stalls

, friction factor, channel

half-width, and water depth

f fB

S c c Bh

h


Canestrelli, Alberto, et al.

"Importance of frictional effects and

jet instability on the morphodynamics

of river mouth bars and levees.

" Journal of Geophysical Research

23


Wr=1.7

n =180

Why are the discharges down bifurcate

arms asymmetric?

Qr = 3.7

Qr = 1.4

Qr = Qlargert/Qsmaller

/larger smallerQ Q

24


Use Delft3D-FLOW to determine stable equilibrium

bifurcations

Find stable equilibrium combinations of Shields Number,

Qa, aspect ratio, B/Ha, & friction factor, Ca

Qa Ba Ca

25


Conclusion: Only asymmetric bifurcations are stable

equilibrium solutions to flow and sediment partitioning

Implication: If bifurcations are asymmetric, then a delta

cannot grow self-similarly. The result is the growth and

abandonment of delta lobes

Results….

1 5/ 2.a gC

Mossy Delta, SK

( )a a

B

HQ

Edmonds, D. A., and R. L.

Slingerland (2008),

Stability of delta

distributary networks and

their bifurcations, Water

Resources Research

26

For the class of deltas prograding into shallow water (depths

typically less than the depths of the distributary channels) with

minimal waves and tides:

The basic building block is the river mouth bar whose dimensions,

spacing, and distribution of water and sediment to the delta perimeter

are determined by the hydrodynamics of the turbulent plane jet, and the

morphodynamics at channel bifurcations

But we haven’t yet explained the variation in delta planforms

To summarize so far….

Goose River Delta, Lake Melville, Labrador Un-named Delta, Lake Tana, Ethiopia

27

Another Conjecture

Delta Form ( , , , , )s sf Q Q P Veg A

Earlier we hypothesized that…..

We know in modern deltas that the number of distributary

channels N, increases with maximum monthly discharge

Qmax, (Syvitski & Saito 2007):

Here we conjecture that Q and Qs are simply scaling

variables. It is actually sediment properties, degree and type

of vegetation, and the amount of space to store sediment

that determine a delta’s form.

0.75

maxN Q

28

critical shear stresses required for

sand/mud ratios re-erosion of cohesive sediments

90% sand-dominated

50% mixed

10% mud-dominated

0.25 N m-2 low-cohesion

1.75 N m-2 medium-cohesion

3.25 N m-2 high-cohesion

Approach

• Initial bathymetry

• 25 m2 cell size

• No waves, tides, Coriolis, salinity

• Constant water temperature

• Q = 1000 m3 s-1

• Six sediment sizes (300, 150, 80, 32, 13,

7.5 µm)

• Qs = 0.1 kg m-3

• Morph scale factor = 175

• Open boundaries on top and sides

Grow nine deltas of varying sediment cohesion using Delft3D:

Model design:

6

4

2

0 0 2 4 6 8

km water depth (m)

3

2

1

0

29

Expt 2: Delta Planform as a Function of Sediment Properties— Mud-dominated

Results

bed e

levation (

m)

1

0

-1

-2

30

Expt 2: Delta Planform as a Function of Sediment Properties--- Sand-dominated

Results

bed e

levation (

m)

1

0

-1

-2

31

Results

Be

d E

leva

tio

n

32

Some Speculation

…..“the marginal marine environment was characterized

by the more common presence of braid-delta systems

during the Precambrian than in Phanerozoic time.”

Explanation: vegetation increases the apparent cohesion

of a sediment. That is why deltas on the young Earth

were braid-deltas.

33

Experimental Design

Relative base level fall

(RBLF) applied at open

ocean boundaries

flat basin floor

150 m-wide feeder

channel

Q = 1000 m3 s-1

Qs = 0.17 m3 s-1

five grain sizes from 25

to 275 mm

How do the depth of the receiving basin and rate of sea level fall affect delta form?

34

Rate of Base Level Fall (mm yr-1)

0 5 10

Initia

l B

asin

Wa

ter

De

pth

(m

)

20

1

2 8

4

35

Example of Delft3D Stratigraphy

Initial basin depth = 8 m Rate of sea level fall = 10 mm/yr

A

A’

A A’ D50

36


and vegetation.

The basic building block of river-dominated deltas is the river mouth

bar whose dimensions, spacing, and distribution of water and

sediment to the delta perimeter are determined by the

hydrodynamics of the turbulent plane jet, and the morphodynamics

at channel bifurcations

Grain size is a major determinant of delta characteristics. Highly

cohesive sediment creates resistant levees that confine the flow,

thereby causing sediment deposition basinward of the levee termini

and progradation of channels far into the basin

Less cohesive sediment creates levees that are “leaky” and water is

fed to the entire delta topset through numerous crevasses.

37


and vegetation.

These different processes respectively create elongate birdsfoot

deltas with rugose shorelines and topographically rough floodplains

and fan deltas with smooth shorelines and flat floodplains

Delta form also depends upon the initial depth of the receiving

basin. Deeper basins promote fewer active distributaries and more

infrequent lobe switching

These shapes create relatively unique clinoform geometries and

facies

38

But….can we trust Delft3D Predictions? Why does Delft3D produce

asymmetric deltas given symmetric

BCs?

How can we believe the predictions

when the channels are poorly

resolved because of large cell sizes

and the channel pattern reflects the

rectangular grid?

39

But….can we trust Delft3D Predictions? Why are the hydraulic geometries of Delft3D’s self-formed

channels different from real channels (even when cell size is

considered)?

39

40

But….can we trust Delft3D Predictions? How robust is our answer if we ignore vegetation?

41

Towards a new Delft3D….

….built by Deltares and you…

…with help from the US National Science Foundation’s FESD

Delta Dynamics Collaboratory & CSDMS

FESD: Frontiers in Earth System Dynamics

The Delta Dynamics Collaboratory

12 PIs, 7 institutions, $5 million, 5 years

The overall objective is to develop tested, high-

resolution, quantitative models incorporating

morphodynamics, ecology, and stratigraphy to predict

river delta dynamics over engineering to geologic

time-scales, and to specifically address questions of

system dynamics, resiliency, and sustainability

42

A mass-conservative, staggered, immersed

boundary model for shallow flows on complex

geometries Implements an immersed boundary method that frees land/water

boundaries from constraints of a rectangular grid

Models lateral bank erosion

Contains a sub-grid vegetation-flow interaction module

Team: [Alberto Canestrelli, Aukje Spruyt, Bert

Jagers]—hydro- & morpho-dynamics, [Fei Xing,

James Syvitski, Doug Edmonds, William Nardin]—

vegetation, Rudy Slingerland (cheerleader)

A New Multidimensional Ecomorphodynamic Model: Delft3D+

43

Immersed boundary method for land/water

boundaries

hybrid cut-cell/ghost-cell method


cut-cells are used in the continuity

equation in order to conserve mass

Cut-cell =>Computational volume is tracked

ghost cells are used for the

momentum equations

Ghost-fluid => extra points are added

44

Test Case: 3D uniform flow in a channel not

oriented with grid


Cartesian solution for depth-

averaged flow velocity along

centerline

Vertical

distribution of

horizontal velocity

near bank

45

Can now treat lateral bank erosion…..

Use wall shear stress to predict particle-by-particle bank

erosion (e.g. Darby & Thorne formulation)


cohesive non-cohesive

dx dx

dry cells wet cells

E

( )n

b cE k

E

46

A sub-grid vegetation-flow interaction module

based upon the Baptist et al. (2005) equations

Vegetation modeled as rigid cylinders characterized by plant

height, density, stem diameter, and drag coefficient in the model.

Vertical flow velocity profile is divided into a constant zone of

flow velocity inside the vegetated part and a logarithmic velocity

profile above for submerged vegetation


47


Results


Adding vegetation

increases the local

fraction of sediment

deposited inside a marsh

but…..

the vegetative roughness

also forces more water

into the channels, leading

to more erosion in the

channels and more water

by-passing the marsh

surface RS(nc) = the ratio between vegetated and non-

vegetated sand deposition

intermediate vegetation

height maximizes

sedimentation on bars

Nardin & Edmonds, Nature Geoscience 7, 722–726 (2014)

48


Turf Erosion Module

Team: Fei Xing & James Syvitski

A new module calculates the critical shear stress needed to rip

up turf

turf parameters are 3D variables


( )wc root cohesion

Stress balance on a cube of turf…..

where: wc = wave-current shear

root = root strength

cohesion = sediment strength

wc

root cohesion

49

An individual-based community model for

predicting fish productivity on an evolving coastal

delta

Team: Paul Venturelli, Manuel Garcia-Quismondo

Objectives

develop an individual-based community model to predict fish

productivity on an evolving delta

determine the structural features of a delta that are highly

correlated to fish productivity

develop a mechanism for evaluating alternative restoration

scenarios in terms of fish productivity


50

An individual-based fish productivity cellular model


Layer 1: bathymetry from Delft3D+

Layers 2 and 3:

a) water depths, velocities, and

temperatures from Delft3D+

a) vegetation (prescribed)

Layer 4: individual-based

fish model - 400 x 400 m sub-grid

- 5 species

- Dt = 1 hour

- feed and grow

- swim about

- reproduce

- die

51

In Conclusion….

DSD-INT 2014 - Delft3D Users Meeting - Keynote Lecture 2014 - Dynamic Deltas, Rudy Slingerland,...

Engineering

Transcript of DSD-INT 2014 - Delft3D Users Meeting - Keynote Lecture 2014 - Dynamic Deltas, Rudy Slingerland,...