Morphological Modeling of the Alameda Creek Flood Control Channel

1
Morphological Modeling of theAlameda Creek Flood
Control Channel
Rohin Saleh, Alameda County Flood Control
District Søren Tjerry, Ph.D., DHI Portland,
Oregon David E. Rupp, Ph.D., DHI Portland, Oregon
(presenter)
11th December 2008 Alameda County
2
Background

• San Francisco Public Utilities Commission (SFPUC)
  has decommissioned the Sunol and Niles dams on
  Alameda Creek.
• Temporary increase in sediment discharged to
  Alameda Creek.
• The dams and their removal are under the
  jurisdiction of SFPUC, while the Alameda County
  Flood Control and Water Conservation District
  (District) is responsible for Alameda Creek and
  the Alameda Creek Flood Control Channel (ACFCC).

3
Background

• Additional sediment deposition in channel system
  due to increased sediment supply to the creek.
• Increased inundation in the advent of a flood
  could result from additional sediment influx,
  although the volume and timing of sediment
  discharged to the creek is unknown.
• This poses a burden for the District responsible
  for the creek and the flooding that occurs along
  the creek.

4
Objective and Approach

• Objective
• Quantify the largest increase in the 100-yr
  floodplain that can be experienced as the
  sediment pulse from the dam removal migrates
  through the Alameda Creek Flood Control Channel
  (ACFCC).
• Approach
• Develop MIKE 21C graded sediment morphological
  model.
• Calibrate model by matching observed
  morphological development.
• Apply the model with and without dam removal to
  determine the morphological developments in the
  two cases.
• Develop MIKE FLOOD floodplain model that can
  simulate the 100-yr floodplain around the ACFCC
  as function of the updated ACFCC bathymetry
  (i.e., feed updated bathymetry to floodplain
  model).

5
Objective and Approach

• Objective
• Quantify the largest increase in the 100-yr
  floodplain that can be experienced as the
  sediment pulse from the dam removal migrates
  through the Alameda Creek Flood Control Channel
  (ACFCC).
• Approach
• Develop MIKE 21C graded sediment morphological
  model.
• Calibrate model by matching observed
  morphological development.
• Apply the model with and without dam removal to
  determine the morphological developments in the
  two cases.
• Develop MIKE FLOOD floodplain model that can
  simulate the 100-yr floodplain around the ACFCC
  as function of the updated ACFCC bathymetry
  (i.e., feed updated bathymetry to floodplain
  model).

6
Outline

• What is MIKE 21C?
• Examples of MIKE 21C applications
• A MIKE 21C model of the ACFCC
• Calibration of the MIKE 21C morphological model

7
MIKE 21C model features

• Depth-integrated hydrodynamic model that can
  simulate quasi-steady and dynamic flow fields
  with time-varying boundary conditions.
• Curvilinear grids (follow streamline curvature).
• Graded sediment (up to 16 grain sizes from fine
  sand to coarse gravel).
• Cohesive sediment (clay, silts).

8
MIKE 21C model features cont.

• Suspended load model with several transport
  formulas available accounts for helical flow and
  adaptation in space (AD equation).
• Bed-load model with several transport formulas
  available accounts for helical flow and bed
  slope.
• Dynamic update of the bed level true
  morphological model.
• Dynamic update of bed composition by grain size.
• Parallel code! Can use as many processors as you
  have available. Allows computations on fine
  grids, over long time, with many sediment
  fractions.

9
Hydrodynamics Snake River ADCPPine Bar, below
Hells Canyon Dam, 24,300 cfs
ADCP
MIKE 21C
10
Curvilinear hydrodynamics
Fish resting poolsExample from San Lorenzo
Creek, complex flow fields
Fish resting pool
11
Jamuna River, Bangladesh
12
Jamuna River, Bangladesh, Q45,000 m3/s, d500.16
mm, eroding sandy banks, 10 km width (decreasing
towards the Ganges confluence), 100 km reach
modeled (200 km North-South in Bangladesh).
0 years
30 years
3 years
6 years
12 years
24 years
18 years
Simulation of braiding
13
Example of morphological model calibration

• Observed morphological development over 9 years
  based on 10 bathymetry surveys.
• Un-calibrated means we assume that standard
  sediment transport formulas are valid.
• Calibrated model has substantially (up to 60
  times) higher sediment transport than what
  accepted formulas yield.
• Calibration is critical. The calibrated model
  matches the observations incredibly well this is
  what an accurate morphological model can do
  when calibrated!

14
Morphological Modeling of the Alameda Creek Flood
Control Channel
Model Development and Calibration
15
Alameda Creek

• Alameda Creek and Alameda Creek Flood Control
  Channel (ACFCC)

Dams removed, Autumn 2006
ACFCC
Niles gage
SF Bay
Active rubber dams
16
Longitudinal Profile of Thalweg, ACFCC
(Tidally influenced)
17
Curvilinear Grid of Flood Control Channel
Grid Resolution Grid Extent Longitudinal
300 ft 200 x 10 cells Transversal 30
90 ft
18
Bathymetry of Flood Control Channel
Elevation (m)
Elevation (m)
19
ACFCC MIKE 21C Model Properties

• Transport formula Engelund and Hanson
• 10 non-cohesive sediment grain sizes
• 0.125 mm to 64 mm
• Cohesive sediment lt 0.063 mm
• Time step 2 seconds
• Simulation period Oct. 2003 Sep. 2013
• (158 million time steps in 2 days real time)

20
Initial Bed Sediment Conditions
Bed Sediment Particle Size Distribution at Niles
Gage

• Bed Sediment Particle Size Distribution
• Sediment thickness

• Assumptions
• Distribution the same throughout ACFCC at time
  0.
• Sediment layer thickness 0.4 m

21
Model Calibration Challenges

• Largest discharge events dominate deposition and
  erosion.
• No suspended sediment and bedload measurements
  for largest events.
• Therefore, boundary conditions are uncertain!

22
Model Calibration Challenges
At least 14 events between 1999 and 2007 exceeded
3,000 cfs
23
Model Calibration Challenges
Initial attempt

• Generate rating curves for sediment inflow based
  on least-squares fit to available data.
• Apply rating curves to discharge time series at
  Niles Gage to calculate sediment inflow for
  calibration period (2003 to 2007).
• Failure! Less estimated sediment inflow than
  measured deposition.

24
Measured Cumulative Change in Sediment (2003 to
2007)
Zone II
Zone I
Zone III
25
Model Calibration
Assumptions

• Nearly all non-cohesive sediment input to the
  ACFCC between 2003 and 2007 was deposited in
  Zones I and II.
• Deposition in the tidally-influenced zone (Zone
  III) was mostly cohesive between 2003 and 2007.
• Both the SF Bay and Alameda Creek are cohesive
  sources to Zone III.

26
Model Calibration
Methodology

• Adjust non-cohesive sediment rating curve
  parameters so sediment influx matches total
  deposition in upper channel (Zones I and II).
• Adjust sediment transport factor (per grain size)
  to achieve match to cumulative longitudinal
  deposition pattern.
• Adjust cohesive sediment parameters and SF Bay
  cohesive sediment concentration to achieve match
  to cumulative longitudinal deposition pattern in
  tidally-influenced channel (Zone III).

27
Observed and Modeled Suspended Sediment Transport
vs. Discharge
Cohesive
Non-cohesive
28
Observed and Modeled Bedload Transport vs.
Discharge
29
Calibration Results Following Steps 2 and 3
Cumulative Deposition (2003 2007)
30
Model Calibration Results
Measured and modeled change in bed level in the
upper ACFCC between 2003 and 2007
Measured
Modeled
31
Model Calibration Results
Measured and modeled change in bed level in the
lower ACFCC between 2003 and 2007
Measured
Modeled
32
Simulated morphology 2003 - 2013
Rubber Dams
Upper channel
Lower channel
33
Simulated morphology 2003 - 2013
Difference in bed level due to presence of rubber
dams
Upper channel
Lower channel
34
Conclusions and Future Work

• Conclusions
• Successfully simulated general morphological
  development in the ACFCC between 2003 and 2007.
• Have a calibrated model that will permit us to
  evaluate scenarios.

• Next step
• Simulate additional sediment due to removal of
  Sunol and Niles Dams.

35
Thank You!