Post on 20-Jan-2016
description
1
River Sediment Calibration Rules
Gary Shenk
6/7/05
Modeling Subcommittee
2
Schedule
• Sediment calibration completed
• Initial nutrient calibration by August 1.
• Refinements will follow over the next 18 months.
3
Calibration Method
• Flow– Automatic calibration method developed
• Edge-of-Field Sediment– Automatic calibration method developed
• River Sediment– Two out of three ain’t bad– Developed calibration rules and applied them
by hand
4
critz
w
eMass
1
1
River Cohesive Sediment Simulation
SuspendedSediment
Bed Storage(unlimited)
Outflow
Sco
ur
Dep
osit
ion
Inflow
1*
crit
AreaM
5
critz
w
eMass
1
1
River Sediment Simulation
Bed Storage(unlimited)
Dep
osit
ion
6
River Sediment Simulation
Bed Storage(unlimited) S
cour
1*
crit
AreaM
7
River Sand Simulation
Jvelocityksand }{*][
Adjustments are then made to the bed depth
8
Calibration Rules
• All rivers segments upstream of a calibration point have identical parameters
• For nested rivers, this applies only to rivers downstream of any upstream gages.
Remington
Robinson
Rappahannock
1
9
Uncalibrated
10
Stau 97 percentileCtau 93SW .001 inches/secCW .0001 CM = 1 lb/ft^2/daySM = 1KS = .003 empiricalES = 4
High flow in the ballpark, but low flow is a problem. Lowest concentrations are mostly VSS, which is not yet simulated and are at LOD in the observed
11
Stau 97Ctau 93SW .001CW .0001CM = 1SM = 1KS = 3ES = 3
To deal with the lack of VSS in the simulation, use SAND to make the low concentrations work out. This can be easily reversed when VSS are simulated
12
Stau 99Ctau 96SW .001CW .0001CM = 1SM = 1KS = 3ES = 3
Reduce the frequency of scour to deal with over-simulation
13
Stau 99Ctau 96SW .001CW .0001CM = .5SM = .5KS = 3ES = 3
Reduce the effect of scour to deal with over-simulation
14
Stau 99.5Ctau 98SW .001CW .0001CM = .5SM = .5KS = 2.8ES = 3
Tighten up the simulation
15
Issues with simulation
• Low values dominate the CFD, but they are not meaningful from:– A load standpoint– An accuracy of observation standpoint– Simulation standpoint (no VSS)
• Flow not perfectly calibrated– If peak is missed by a day, then the
concentration simulation should not match the observed.
16
‘Windowed’ comparison
• If simulated or observed value is below 10 mg/l set it to 10 mg/l.
• Check simulation for 24 hours before and after observation and set simulated value to point closest to observation.
Of the highest observed and simulated peaks at all calibration stations, almost as many peaks occurred one day apart, but few occurred on two days apart
One day apart – 83% of the same day figure
Two days apart – 19% of the same day figure
17
I
Reported Error
18
Stau 97Ctau 93SW .001CW .0001CM = 1SM = 1KS = .003ES = 4
Starting point
19
Stau 97Ctau 93SW .001CW .0001CM = 1SM = 1KS = 3ES = 3
Sand Calibration
20
Stau 99Ctau 96SW .001CW .0001CM = 1SM = 1KS = 3ES = 3
Reduce Frequency of Scour
21
Stau 99Ctau 96SW .001CW .0001CM = .5SM = .5KS = 3ES = 3
Reduce effect of scour
22
Stau 99.5Ctau 98SW .001CW .0001CM = .5SM = .5KS = 2.8ES = 3
Tighten up calibration
23
Other Considerations: Load 1:1 plot, error vs Shear stress, Load Frequency
24
Choptank RiverLog Sediment Load (log kg/mo)
y = 0.7828x + 1.0642
R2 = 0.7312
3
3.5
4
4.5
5
5.5
6
6.5
7
7.5
3 3.5 4 4.5 5 5.5 6 6.5 7
Phase 5 Simulation log kg/mo
ES
TIM
AT
OR
log
kg
/mo
Comparison with Estimator Model
25
Legend
P5 Calibration Stations
# of Observations
0 - 100
101 - 300
301 - 500
501 - 1000
1001 - 5000
50
80
12
8
14
164
# of Stations
26
Plan
• Following calibration rules, team has calibrated all stations this month.– Jing Wu – Susquehanna, Coastal Plain,
Southern Rivers– Ross Mandel – Potomac, Patuxent, Choptank– Doug Moyer – James, Appomattox– Jason Pope – Rappahannock, York
27
Hope for automation
• Separate out sections of the CFD– Low part: sand - VSS– Highest part: silt– Mid-high part: clay
• PEST– Few parameters– Short run time ~ 1-2 minutes