Sedflume Data Are Collected At My Site: So What the Hell Do I Do Now

1
Sedflume Data Are Collected At My Site So What
the Hell Do I Do Now?

• Second Iowa Workshop on Large Rivers
• C. Kirk Ziegler, Ph.D., P.E.
• Quantitative Environmental Analysis, LLC
• Montvale, NJ

October 8, 2004
2
Schematic of SEDFLUME
3
SEDFLUME Core Data
s
20
40
57
28
80
2
113
Applied ?
160
?crit f(z)
Approx current velocity (cm/s)
4
Potential Uses of SEDFLUME Data

• Estimate relative stability of bed at different
  locations
• This estimate must be made judiciously
• Input information for a sediment transport model
• Erosion rates
• Critical shear stress

5
Applying a SEDFLUME-Based Model

• SEDFLUME erosion rate data cannot be used in
  conventional sediment transport models
• For example ECOM-SED, EFDC, RMA-2
• An algorithm (SEDZLJ) has been developed that can
  use SEDFLUME data

6
SEDZLJ Algorithm
Esus
Dsus
Ebed
Dbed
Bed-Load Layer
Ta f(D50 , ?)
D50 , Fk
Active Layer
Parent Bed
7
SEDZLJ Algorithm Suspended Load

• Erosion flux for size-class k
• Esus,k ?k Fk E for ? gt ?ce
• where E gross erosion rate
• ? qsus / qtot f(u / Ws,k)
• Deposition flux for size-class k
• Dsus,k Pdep,k Ws,k Csus,k

8
SEDZLJ Algorithm Bed Load

• Net bed flux for size-class k
• Ebed,k Dbed,k 1 (Cbed / Ceq) Ebed,k
• 1 (Cbed / Ceq) (1 - ?) Fk E

9
SEDZLJ Algorithm Model Parameters

• Settling speed and effective particle size
• Ws,k f(Deff,k)
• Bed bulk properties
• ?s (dry or bulk density)
• D50
• Fk
• Bed erosion properties (SEDFLUME data)
• E
• ?ce

10
SEDZLJ Challenges Effective Particle Size

• How many size classes?
• Typically, three size classes are used
• How to specify Deff ?
• Can treat Deff as a calibration parameter
• How to determine composition of incoming sediment
  loads?
• Generally, sparse data are available for sediment
  load composition

11
SEDZLJ Challenges Bed Properties

• How to specify spatial distribution of ?s , D50
  and Fk?
• Generally, sparse data sets make it difficult to
  use conventional interpolation or extrapolation
  methods
• A method was developed for the Upper Hudson River
  sediment transport model that appears to be
  robust
• Used to spatially distribute D50 and Fk in
  noncohesive grid elements

12
Underlying Hypothesis

• Local noncohesive bed properties are primarily
  determined by local energy regime
• High energy ? coarser bed
• Low energy ? finer bed
• It is assumed that bottom shear stress (?) is
  representative of local energy
• The basic hypothesis in this analysis is
• As a first-approximation D50 f(?)

13
Step 1 Generate ? Distribution

• Run hydrodynamic model at nominal high flow
  (e.g., bank-full flow)
• Calculate ? in noncohesive bed elements
• Normalize ? with respect to ?max
• ?n ?/?max

14
Step 3 Determine D50 Function

• Assume a functional relationship between D50 and
  ?n
• Previous experience suggests that this
  relationship is of the form
• D50 A exp(B ?mn)
• The coefficients A, B and m are adjusted so that
  the predicted D50 distribution is in reasonable
  agreement with the observed distribution of D50

15
Step 4 Specify D50 Distribution for Model Input

• Use relationship to specify D50 for each grid
  element
• D50 (i,j) A exp B ?mn(i,j)

16
Step 5 Specify Fk Distributions for Model Input

• Use site data to develop correlations between Fk
  and D50
• Fk f(D50)
• Use these relationships to specify Fk for each
  grid element

17
SEDZLJ Challenges Erosion Properties

• Need to specify spatial distributions of erosion
  properties
• Vertical distribution
• Horizontal distribution
• Method 1 use interpolation techniques to specify
  E(x,y,z) and ?ce(x,y,z)
• Method 2 assume that
• E A ?n ? gt ?ce
• where A(x,y), n(x,y), ?ce(z)