Chester River Shallow Water Project –
SCHISM model results
Harry Wang, Joseph Zheng, Fei Ye,
Zhengui Wang, and Xiaonan Li
Virginia Institute of Marine Science,
College of William and Mary
Gloucester Point, VA 23062
[email protected];ph:804-684-7215
Chesapeake Bay Program sub-committee meeting presentation
4/23.2015
Acknowledgement:
(1) Grant supported by EPA Chesapeake Bay Program.
(2) SCHISM model development support by MD Environment service and MD Port authority
SCHISM: (Semi-implicit Cross-scale Hydroscience Integrated System
Model)
 A derivative product of SELFE, distributed with Apache v2 license
 Galerkin finite-element and finite-volume approach: generic unstructured grids (mixed
triangles and quadrangles)
 Semi-implicit time stepping: no mode splitting  large time step and no splitting errors
 Eulerian-Lagrangian method (ELM) for advection  more efficiency & robustness
 Hybrid SZ coordinates or LSC2 in the vertical
 Configurable
Cartesian or spherical coordinates
2D or 3D
Hydrostatic or non-hydrostatic
Mass conservative transport (implicit TVD2)
 Fully parallelized with domain decomposition (MPI) with good scalability
 Well-benchmarked inundation scheme for wetting and drying (NTHMP)
 Operationally tested and proven (NOAA, DWR, CWB…)
 Has evolved into a comprehensive modeling system
 Open source and driven by user community needs; our goal is to develop a verifiable
and comprehensive modeling system
Community model: www.schism.wiki
Upper Chesapeake Bay and Chester model grid
Upper Chesapeake Bay
model grid
Mixed quadrilateral and
triangular grid 000
Chester River
Chester model grid and Corsica River
Chester River
Corsica River
0 -40
The upper Chesapeake Bay model – Vertical grid
-50
0
5
10
15
20
25
-2
Master grid number
30
Cross channel transect
-4
0
-2
-6
z (m)
-8
-4
-6
-8
-10
-10
-12
4.8
5
5.2
5.4
5.6
Along transect distance (m)
-12
5.8
6
4
x 10
*Triangular grid for the bottom layer
Master grid
4.8 A0 reference grid for the hybrid
5 z-and-sigma grid
5.2
5.4
5.6
Along transect distance (m)
z (m)
-10
-20
-30
-40
-50
0
5
10
15
Master grid number
20
25
30
*Reference: Zhang et al (2015): A new vertical coordinate system for a 3D unstructured-grid model. Ocean Modeling, Vol. 85, p16-31
Cross channel transect
Vertical discretization in the Chester River
Along channel transect
Depth (m)
0
-5
-10
-15
0
1
2
3
4
5
Distance from Chester River mouth (m)
6
x 10
4
Cross channel transect near ET4.2
Depth (m)
0
-5
-10
-15
0
500
1000
1500
2000
2500
Distance from north to south (m)
3000
3500
4000
Chester River model setup
• Initial condition:
 Interpolated from observation in 2003
• Boundary condition:
 CH3D: elevation, salinity, temperature
 Schism Upper Bay model: velocity
• Fresh water inputs:
 Chester River fall-line load (provided by Bay Program)
 Susquehanna River fall-line load for the upper bay model
 Point and non-point sources provided by the HSPF watershed model
results
• Wind:
 Thomas Point station, spatially homogeneous throughout the
domain
• Heat, radiation and precipitation:
 North American Regional Reanalysis (NARR)
• Time step: 120 sec (hydrodynamic model); 360 sec (wind
wave model)
Chester River monitoring station
Model results
•
•
•
•
•
Elevation
Velocity
Salinity
Temperature
TSS
(*Model results from 2003-2006 has been submitted to management team of the
project, as of 03/2015)
Model results – tidal elevation
Amplitude (m)
Harmonic Constituents for NOAA station 8572955, Love Point Pier MD
0.18
Harmonic
Components
Amplitude
(m)
0.16
M2
0.170
0.14
K1
0.065
0.12
O2
0.060
0.1
N2
0.038
0.08
S2
0.020
0.06
P1
0.020
0.04
Q1
0.010
0.02
K2
0.010
0
Q1
O1
P1
N2
K1
Constituents
M2
S2
K2
Model results – M2 tidal range and phase spatial distribution
Phase
Phase
(degree)
(degree)
Phase
(degree)
Amplitude
Amplitude
(m)(m)
Amplitude
(m)
• Chester River
0.35
0.35
0.35
0.3
0.3
0.3
0.25
0.25
0.25
0.2
0.2
0.2
300
300
300
200
200
200
100
100
100
0
0
0
Tidal constituent (M2)
Tidal constituent (M2)
Tidal constituent (M2)
NOAA
NOAA
RUN06uc
NOAA
RUN06uc
Model
RUN06uc
Model
Model
mouth
mouth
mouth
LvPtPier Queenstown
LvPtPier Queenstown
LvPtPier Queenstown
CliffsPt
CliffsPt
CliffsPt
Chestertown Crumpton
Chestertown Crumpton
Chestertown Crumpton
mouth
mouth
mouth
LvPtPier Queenstown
LvPtPier Queenstown
LvPtPier Queenstown
CliffsPt
CliffsPt
CliffsPt
Chestertown Crumpton
Chestertown Crumpton
Chestertown Crumpton
Model results – surface water level (evidence of influence of the
wind)
• Chester River (ET4.2, XIH0077) and Corsica River (XHH4916)
Surface elevation
0.8
0.6
Elevation (m)
0.4
0.2
0
-0.2
ET4.2
XIH0077
XHH4916
-0.4
-0.6
100
101
102
103
104
105
106
Days since 2003-01-01
107
108
109
110
Model results – Velocity scatter diagram
Model results – Velocity magnitude
No velocity data for
comparison;
Attempt made to
contact
Douglas R. Levin of
Washington College
Model results – Velocity animation – Chester River
Days after 2003-01-01:
Model results – Velocity animation near Corsica River
Days after 2003-01-01:
Days after 2003-01-01:
Model results – Velocity distribution on an un-strucutured grid
Model results – salinity (with the CH3D b.c.)
Bottom
Observation:
Surface
Bottom
Model:
Surface
* The depth of each station is the
maximum sampling depth
Model results – salinity (with the upper bay b.c.)
Bottom
Observation:
Surface
Bottom
Model:
Surface
*The depth of each station is the
maximum sampling depth
Model results – salinity stratification
Stratification (of salinity) equals bottom minus surface salinity at ET4.2)
12
With the CH3D b.c.
10
8
6
4
2
0
0
100
200
12
300
400
500
600
700
500
600
700
With the upper bay b.c.
10
8
6
4
2
0
0
100
200
Observation:
300
400
Model:
Model results – temperature (with the CH3D b.c.)
Bottom
Observation:
Surface
Bottom
Model:
Surface
*The depth of each station is the
maximum sampling depth
Model results – temperature stratification
Temperature (degree C)
Stratification (of temp) equals bottom minus surface temperature at ET4.2)
4
2
0
-2
-4
-6
0
100
200
300
400
Days from 2003-01-01
Observation:
Model:
500
600
700
Computational performance
Chesapeake Bay
model
Upper Bay model
Chester River model
Nodes (horizontal)
13426
13484
8325
Elements(horizontal)
21409
22268
15104
Average vertical
layers
25
23
17
Run time
(with 48 CPUs on
Hurricane Cluster)
Run time
(with 128 CPUs on
Hurricane Cluster)
614 times faster than
675 times faster than
real time
real time
(2 years of simulation
1288 times faster than
real time
finished in 1.2 day)
1064 times faster
than real time
869 times faster than
real time
(2 years of simulation
finished in 0.7 day)
*when the wind wave model was coupled, the run time reduced by factor of 5
1641 times faster than
real time
Additional effort: modeling TSS using a non-cohesive formulation
Erosion and Deposition formulation
Erosion
: 𝐸𝑞 = 𝐸0,𝑞
1 − 𝑝 𝑓𝑞
𝜏𝑠𝑓
𝜏𝑐𝑟,𝑞
− 1 , if 𝜏𝑠𝑓 > 𝜏𝑐𝑟,𝑞
𝐸0,𝑞 : bed erodibility constant
𝑝 : sediment porosity
𝑓𝑞 : volumetric fraction of sediment of class 𝑞
𝜏𝑐𝑟,𝑞 : critical shear stress (calculated internally)
𝜏𝑠𝑓 : bed shear stress
Deposition
: 𝐷𝑞 = 𝑤𝑠,𝑞 𝐶1
𝑊𝑠,𝑞 =
𝑣𝑎
𝑑50,𝑞
3
10.362 + 1.049𝐷∗,𝑞
1/2
− 10.36
𝑊𝑠,𝑞 : settling velocity
𝐶1 : sediment concentration
𝑣𝑎 : kinematic viscosity of water
𝑑50,𝑞 : median grain sediment diameter of class 𝑞
𝐷∗,𝑞 : dimensionless sediment diameter of class 𝑞
Some constants used in the model
𝐸0,𝑞 : 1.6e-4, 1.6e-4, 1.6e-4 (kg/m2/s)
𝑑50,𝑞 : 0.01, 0.02, 0.055 (mm)
𝑓𝑞 : 1/3 , 1/3 , 1/3
𝑝
: 0.4
Reference: Pinto et al. (2012): Development and validation of a 3D morphodynamic modeling system for non-cohesive sediments
Ocean Modeling, Vol. 57-58, p1-14.
Model results – TSS
Model results – TSS in the lower Chester River
The TSS in ET4.2 is under-predicted similar to the salinity !!!