Housatonic River Project Supplemental Modeling Studies Supplemental Modeling Studies Earl Hayter National Exposure Research Lab Athens, GA Earl Hayter National Exposure Research Lab Athens, GA scs1
Housatonic River Project
SupplementalModeling StudiesSupplementalModeling Studies
Earl HayterNational Exposure Research Lab
Athens, GA
Earl Hayter National Exposure Research Lab
Athens, GA
scs1
Slide 1
scs1 Please add name and affiliation susan, 4/6/2005
Housatonic River Project pg 2
OutlineOutline
One-Dimensional (1D) modeling
– EFDC1D – GSTARS-1D– HEC-6
Tests of the applicability of EFDC for the Housatonic River
– Benchmark tests of EFDC hydrodynamics and sediment transport
– Evaluation of alternative grid configurations
One-Dimensional (1D) modeling
– EFDC1D – GSTARS-1D – HEC-6
Tests of the applicability of EFDC for the Housatonic River
– Benchmark tests of EFDC hydrodynamics and sediment transport
– Evaluation of alternative grid configurations
Housatonic River Project pg 3
One-Dimensional Modeling - EFDC1DOne-Dimensional Modeling - EFDC1D EFDC1D (Hamrick 2001)– 1D hydrodynamic and sediment transport model – Capable of simulating flow and both cohesive and noncohesive
sediment transport – Applicable to streams, low-order rivers, and non-stratified tidal rivers
Features/capabilities of EFDC1D– Box- or reach-based spatial data structure (compatible with HSPF),
for representing 1D channel networks– Fully dynamic 1D equation solver for 1D momentum and continuity
equations, and generic 1D transport solver for salinity, temperature, multiple sediment and contaminant classes. Sources and sinks are represented.
EFDC1D (Hamrick 2001)EFDC1D (Hamrick 2001) – 1D hydrodynamic and sediment transport model – Capable of simulating flow and both cohesive and noncohesive
sediment transport – Applicable to streams, low-order rivers, and non-stratified tidal rivers
Features/capabilities of EFDC1DFeatures/capabilities of EFDC1D – Box- or reach-based spatial data structure (compatible with HSPF),
for representing 1D channel networks – Fully dynamic 1D equation solver for 1D momentum and continuity
equations, and generic 1D transport solver for salinity, temperature, multiple sediment and contaminant classes. Sources and sinks are represented.
Housatonic River Project pg 4
One-Dimensional Modeling - EFDC1DOne-Dimensional Modeling - EFDC1D
Features/capabilities of EFDC1D (continued)
– Time-varying upstream and lateral inflows and withdrawals, including corresponding sediment loads and rainfall
– Time varying downstream boundary conditions for stage – Uses water surface elevation-dependent descriptions of channel
cross-section area, surface width, and wetted perimeter– Representation of hydraulic structures such as dams and culverts – Includes settling, deposition and resuspension of multiple size
classes of cohesive and noncohesive sediments– Bed represented by multiple layers of mixed sediment classes – Sediment bed dynamically coupled to the cross-sectional area,
accounting for area changes due to deposition and resuspension
Features/capabilities of EFDC1D (continued)Features/capabilities of EFDC1D (continued)
– Time-varying upstream and lateral inflows and withdrawals, including corresponding sediment loads and rainfall
– Time varying downstream boundary conditions for stage – Uses water surface elevation-dependent descriptions of channel
cross-section area, surface width, and wetted perimeter – Representation of hydraulic structures such as dams and culverts – Includes settling, deposition and resuspension of multiple size
classes of cohesive and noncohesive sediments – Bed represented by multiple layers of mixed sediment classes – Sediment bed dynamically coupled to the cross-sectional area,
accounting for area changes due to deposition and resuspension
Housatonic River Project pg 5
One-Dimensional Modeling - GSTARSOne-Dimensional Modeling - GSTARS GSTARS-1D (Generalized Sediment Transport model for Alluvial River Simulation One Dimensional) (Yang et al. 2004)
– Hydraulic and sediment transport model
– Simulates steady or unsteady flows (with lateral inflows) in a singlechannel or channel network, and simulates cohesive and noncohesive sediment transport
GSTAR-1D’s capabilities– Simulates subcritical and supercritical flows and sediment transport
under unsteady conditions
– Simulates cohesive sediment settling, deposition, erosion, and consolidation processes
– Includes several noncohesive sediment transport equations, applicable to a wide range of hydraulic and sediment conditions
GSTARS-1D (Generalized Sediment Transport model for Alluvial River Simulation - One Dimensional) (Yang et al. 2004)
– Hydraulic and sediment transport model
– Simulates steady or unsteady flows (with lateral inflows) in a singlechannel or channel network, and simulates cohesive and noncohesive sediment transport
GSTAR-1D’s capabilities – Simulates subcritical and supercritical flows and sediment transport
under unsteady conditions
– Simulates cohesive sediment settling, deposition, erosion, and consolidation processes
– Includes several noncohesive sediment transport equations, applicable to a wide range of hydraulic and sediment conditions
Housatonic River Project pg 6
One-Dimensional Modeling - GSTARSOne-Dimensional Modeling - GSTARS
GSTAR-1D’s capabilities (continued)
– Simulates exchange of water and sediment between main channel and floodplain
– Simulates fractional sediment transport, bed sorting, and bedarmoring
– Computes channel width changes using theory of stream powerminimization
– Can simulate both point and nonpoint sources of flow and sediments
– Can simulate internal boundary conditions (e.g., time-stage tables,rating curves, weirs, bridges, and radial gates)
GSTAR-1D’s capabilities (continued)
– Simulates exchange of water and sediment between main channel and floodplain
– Simulates fractional sediment transport, bed sorting, and bedarmoring
– Computes channel width changes using theory of stream powerminimization
– Can simulate both point and nonpoint sources of flow and sediments
– Can simulate internal boundary conditions (e.g., time-stage tables, rating curves, weirs, bridges, and radial gates)
Housatonic River Project pg 7
One-Dimensional Modeling - SetupOne-Dimensional Modeling - Setup
Setup of 1D models– Model domain was represented with 203 surveyed cross
sections– Upstream flow boundary condition was the:
• measured 5-day hydrograph (starting at 2200 hours on 19 September 1999)
• sediment-discharge rating curve (used to specify sediment load time series)
– Downstream boundary condition used stage-discharge rating curve at Woods Pond Outlet
– Spatially-varying initial bed properties were specified using the measured grain size distributions and the cohesive sedimentproperties estimated from the Sedflume study
Setup of 1D models – Model domain was represented with 203 surveyed cross-
sections – Upstream flow boundary condition was the:
• measured 5-day hydrograph (starting at 2200 hours on 19September 1999)
• sediment-discharge rating curve (used to specify sedimentload time series)
– Downstream boundary condition used stage-discharge ratingcurve at Woods Pond Outlet
– Spatially-varying initial bed properties were specified using themeasured grain size distributions and the cohesive sedimentproperties estimated from the Sedflume study
Housatonic River Project pg 8
One-Dimensional Modeling - SetupOne-Dimensional Modeling - Setup
0.0
10.0
20.0
30.0
40.0
50.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Time (days)
Dis
char
ge (c
ms)
Housatonic River Project pg 9
One-Dimensional Modeling - ResultsOne-Dimensional Modeling - Results
282.0
284.0
286.0
288.0
290.0
292.0
294.0
0 2000 4000 6000 8000 10000 12000 14000 16000 18000
Distance (m)
Elev
atio
n (m
)
Bed surface GSTARS EFDC1D Woods Pond weir
Woods Pond weir sill elevation
Housatonic River Project pg 10
One-Dimensional ModelingOne-Dimensional Modeling
Conclusions– 1D models not suitable for simulating a major (e.g., 100
year flood) out-of-bank event
– Not possible to simulate meandering river with floodplain inundated with water (see next slide)
– During out-of-bank event, water flows across the floodplain (more or less orthogonally to the river channel) between meanders
– 1D model cannot represent this condition (simulation of wetting and drying) in a meandering river
ConclusionsConclusions – 1D models not suitable for simulating a major (e.g., 100
year flood) out-of-bank event
– Not possible to simulate meandering river with floodplain inundated with water (see next slide)
– During out-of-bank event, water flows across the floodplain (more or less orthogonally to the river channel) between meanders
– 1D model cannot represent this condition (simulation of wetting and drying) in a meandering river
Housatonic River Project pg 11
ExampleExample
Housatonic River Project pg 12
Tests of the Applicability of EFDC to theHousatonic River
Tests of the Applicability of EFDC to theHousatonic River
I. Benchmark tests of EFDC hydrodynamics and sedimenttransport
– Performed tests to confirm that EFDC accurately represents physicalprocesses controlling flow and sediment transport in the HousatonicRiver
– Two of the more important hydrodynamic processes in the river are: • the impact of meandering on in-channel flow • flow onto the floodplain
Three evaluations were performed, comparing model simulations toexperimental data for in-channel and out-of-bank flow conditions:• Case 1 – Out-of-bank flow: straight channel and overbank flow onto
vegetated floodplain (Thornton et al., 2000)
• Case 2 – Out-of-bank flow: meandering channels and overbank flow ontofloodplain (Shiono and Muto, 1998)
• Case 3 – In-channel flow: 180-degree horseshoe bend (Yen and Lee, 1995)
I. Benchmark tests of EFDC hydrodynamics and sedimenttransport
– Performed tests to confirm that EFDC accurately represents physicalprocesses controlling flow and sediment transport in the HousatonicRiver
– Two of the more important hydrodynamic processes in the river are: • the impact of meandering on in-channel flow • flow onto the floodplain
Three evaluations were performed, comparing model simulations toexperimental data for in-channel and out-of-bank flow conditions: • Case 1 – Out-of-bank flow: straight channel and overbank flow onto
vegetated floodplain (Thornton et al., 2000)
• Case 2 – Out-of-bank flow: meandering channels and overbank flow ontofloodplain (Shiono and Muto, 1998)
• Case 3 – In-channel flow: 180-degree horseshoe bend (Yen and Lee, 1995)
Housatonic River Project pg 13
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Case 2 – Out-of-bank flow: meandering channels and overbank flow onto floodplain (Shiono and Muto, 1998)
– Shiono and Muto (1998) designed flume experiments to describe thedistributions of shear stresses and turbulence in meander channels for in-channel and out-of-bank flows
– Constructed hydraulic flume (10.8-m long, 1.2-m wide, and 0.33-mdeep) as a rectangular cross-section representation of a river valleywith a bottom slope of 0.001
– Constructed a series of meander waves with varying sinuositycharacteristics in the floodplain
– Measured flow velocities at sections w/in ½ meander wavelength of the fourth and fifth meanders using a laser-Doppler anemometer
– Using this experimental setup, they conducted detailed measurementsof secondary flow and shear stresses in the meander channel for a range of out-of-bank flows.
Case 2 – Out-of-bank flow: meandering channels and overbank flow onto floodplain (Shiono and Muto, 1998)
– Shiono and Muto (1998) designed flume experiments to describe thedistributions of shear stresses and turbulence in meander channels for in-channel and out-of-bank flows
– Constructed hydraulic flume (10.8-m long, 1.2-m wide, and 0.33-mdeep) as a rectangular cross-section representation of a river valleywith a bottom slope of 0.001
– Constructed a series of meander waves with varying sinuositycharacteristics in the floodplain
– Measured flow velocities at sections w/in ½ meander wavelength of the fourth and fifth meanders using a laser-Doppler anemometer
– Using this experimental setup, they conducted detailed measurementsof secondary flow and shear stresses in the meander channel for a range of out-of-bank flows.
Housatonic River Project pg 14
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Case 2 – Out-of-bank flow: meandering channels and overbank flow onto floodplain (Shiono and Muto, 1998)
EFDC Setup
• Curvilinear-orthogonal grid used to represent the hydraulic flume • The main channel width of 0.15-m was defined using five lateral cells
with a width of 0.03-m each. Vertical resolution of the floodplain and channel was defined using four layers.
• Bottom roughness was assigned a uniform value of 0.1 mm• The bottom slope of the floodplain and river channel was set at 0.001
m/m, the same slope used in the flume• The depth of flow in the flume was controlled by a tailgate. In EFDC, a
simulated weir was used to control depth as a function of the discharge rate.
Case 2 – Out-of-bank flow: meandering channels and overbank flow onto floodplain (Shiono and Muto, 1998)
EFDC Setup
• Curvilinear-orthogonal grid used to represent the hydraulic flume • The main channel width of 0.15-m was defined using five lateral cells
with a width of 0.03-m each. Vertical resolution of the floodplain and channel was defined using four layers.
• Bottom roughness was assigned a uniform value of 0.1 mm • The bottom slope of the floodplain and river channel was set at 0.001
m/m, the same slope used in the flume • The depth of flow in the flume was controlled by a tailgate. In EFDC, a
simulated weir was used to control depth as a function of the discharge rate.
Housatonic River Project pg 15
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Housatonic River Project pg 16
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Housatonic River Project pg 17
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Tests of the Applicability of EFDC to theHousatonic River – Case 2
Case 2 – Out-of-bank flow: meandering channels and overbank flow onto floodplain (Shiono and Muto, 1998)
Conclusions– Magnitude and direction of simulated flow demonstrated good agreement
with the experiments both for flow within the river channel and onto the floodplain
– Reproduced the shift in alignment of the surface layer flow from the channel meander toward the floodplain as the depth of flow increased
– Simulated flow in the channel bottom layer was consistent with the magnitude and direction of the measured velocities
– The 4-layer EFDC model was able to represent the characteristic secondary circulation resulting from channel meanders
– These comparisons confirm the ability of EFDC to properly represent flow in a meandering channel and the transition from the channel to out-of-bank flow onto a floodplain
Case 2 – Out-of-bank flow: meandering channels and overbank flow onto floodplain (Shiono and Muto, 1998)
Conclusions – Magnitude and direction of simulated flow demonstrated good agreement
with the experiments both for flow within the river channel and onto the floodplain
– Reproduced the shift in alignment of the surface layer flow from the channel meander toward the floodplain as the depth of flow increased
– Simulated flow in the channel bottom layer was consistent with the magnitude and direction of the measured velocities
– The 4-layer EFDC model was able to represent the characteristic secondary circulation resulting from channel meanders
– These comparisons confirm the ability of EFDC to properly represent flow in a meandering channel and the transition from the channel to out-ofbank flow onto a floodplain
Housatonic River Project pg 18
Tests of the Applicability of EFDCto the Housatonic River - Grid
Tests of the Applicability of EFDCto the Housatonic River - Grid
II. Evaluation of alternative grid configurations
– Systematically tested different Cartesian and curvilinear-orthogonal grids
– Purpose - to evaluate how the grid structure affects:• simulation of in-channel river flow and transport of solids • computational speed
– Selected a 1 km long section of the river just upstream of New Lenox Road as a “Test Reach”
– Performed velocity measurements using an ADCP for comparison between simulated and measured velocities
II. Evaluation of alternative grid configurations
– Systematically tested different Cartesian and curvilinear-orthogonal grids
– Purpose - to evaluate how the grid structure affects: • simulation of in-channel river flow and transport of solids • computational speed
– Selected a 1 km long section of the river just upstream of New Lenox Road as a “Test Reach”
– Performed velocity measurements using an ADCP for comparison between simulated and measured velocities
Housatonic River Project pg 19
Tests of the Applicability of EFDCto the Housatonic River - Grid
Tests of the Applicability of EFDCto the Housatonic River - Grid
Housatonic River Project pg 20
Tests of the Applicability of EFDCto the Housatonic River - Grid
Tests of the Applicability of EFDCto the Housatonic River - Grid
Housatonic River Project pg 21
Tests of the Applicability of EFDCto the Housatonic River - Grid
Tests of the Applicability of EFDCto the Housatonic River - Grid
Housatonic River Project pg 22
Tests of the Applicability of EFDCto the Housatonic River - Grid
Tests of the Applicability of EFDCto the Housatonic River - Grid
Housatonic River Project pg 23
Tests of the Applicability of EFDCto the Housatonic River - Grid
Tests of the Applicability of EFDCto the Housatonic River - Grid
Housatonic River Project pg 24
Tests of the Applicability of EFDCto the Housatonic River - Grid
Tests of the Applicability of EFDCto the Housatonic River - Grid
II. Evaluation of alternative grid schemes CONCLUSIONS
– Decision of final grid configuration based on: • Representation of the physics of in-channel and out-of-bank flow
and transport of sediment • Computational efficiency – need to conduct long-term simulations
of baseline conditions and remedial alternatives– Final Grid
• A curvilinear-orthogonal grid of resolution similar to the tested 20m Cartesian grid
• Designed to conform to the topography of the floodplain and theshoreline of the river channel
• Eliminated unneeded grid cells that would have been presentusing a Cartesian grid
• Allowed for the design of an efficient and computationally time-effective modeling framework
II. Evaluation of alternative grid schemes CONCLUSIONS
– Decision of final grid configuration based on: • Representation of the physics of in-channel and out-of-bank flow
and transport of sediment • Computational efficiency – need to conduct long-term simulations
of baseline conditions and remedial alternatives – Final Grid
• A curvilinear-orthogonal grid of resolution similar to the tested 20m Cartesian grid
• Designed to conform to the topography of the floodplain and theshoreline of the river channel
• Eliminated unneeded grid cells that would have been presentusing a Cartesian grid
• Allowed for the design of an efficient and computationally time-effective modeling framework
Housatonic River Project pg 25
Model LinkageModel Linkage Hydrodynamic,
Sediment Transport, &
PCB Fate Model EFDC
Dissolved and sorbed PCBs in water column and sediment
PCB Bioaccumulation
Model FCM
(QEAFDCHN V1.0)
PCBs in Biota
Temperature
Flow (and
Solids*)
Watershed Model
HSPF
Housatonic River Project pg 26
• Models are approximations of reality; they can not preciselyrepresent natural systems
• There is no single, accepted statistic or test that determines whether or not a model is valid
• Both graphical comparisons and statistical tests are required inmodel calibration and validation
• Models cannot be expected to be more accurate than the errors (confidence intervals) in the input and observed data
• A ‘weight-of-evidence’ approach was used in calibration of PSA models
• Models are approximations of reality; they can not preciselyrepresent natural systems
• There is no single, accepted statistic or test that determines whether or not a model is valid
• Both graphical comparisons and statistical tests are required inmodel calibration and validation
• Models cannot be expected to be more accurate than the errors (confidence intervals) in the input and observed data
• A ‘weight-of-evidence’ approach was used in calibration of PSA models
CALIBRATION RATIONALE ‘Basic Truths’ in Modeling Natural Systems