11

A Turbidity Model For A Turbidity Model For Ashokan ReservoirAshokan Reservoir

Rakesh K. Gelda, Steven W. EfflerRakesh K. Gelda, Steven W. EfflerFeng Peng, Emmet M. OwensFeng Peng, Emmet M. Owens

Upstate Freshwater Institute, Syracuse, NYUpstate Freshwater Institute, Syracuse, NY

Donald C. PiersonDonald C. PiersonNew York City Department of Environmental ProtectionNew York City Department of Environmental Protection

2009 Watershed Science & Technical Conference2009 Watershed Science & Technical ConferenceSeptember 14September 14thth-15-15thth,,Thayer Hotel, West Point, New YorkThayer Hotel, West Point, New York

22

• network of 19 reservoirsnetwork of 19 reservoirs• three controlled lakesthree controlled lakes• Croton, Catskill, Delaware Croton, Catskill, Delaware systemssystems• watershed: 1930 miwatershed: 1930 mi22

• storage: 550 BGstorage: 550 BG• unfiltered supplyunfiltered supply• 1.2 BG/day1.2 BG/day

Ashokan ReservoirAshokan Reservoir• watershed: 257 miwatershed: 257 mi22

• storage: 130 BGstorage: 130 BG• Catskill Aq.: 600 MGDCatskill Aq.: 600 MGD• Turbidity < 8 NTU (90Turbidity < 8 NTU (90thth percentile; 1987-2008)percentile; 1987-2008)

*

33

Ashokan Reservoir

West Basin

East Basin

44West Basin

East Basin

Upper Gate Chamber

Bridge and Dividing Weir

55

Gates (4)

66

East BasinDiversion Wall

77

Upper Gate Chamber

88

Intake Structure

99

Turbidity ProblemTurbidity Problem stream channel and banks erosion – glacial and fluvial stream channel and banks erosion – glacial and fluvial

sediment; Esopus Creek 85% of the inflowsediment; Esopus Creek 85% of the inflow turbidity in waters leaving Ashokan Reservoir can be high turbidity in waters leaving Ashokan Reservoir can be high

following major runoff eventsfollowing major runoff events alum treatment before it enters Kensico – Nine alum events, alum treatment before it enters Kensico – Nine alum events,

524 days during 1987-2007524 days during 1987-2007 turbidity model to evaluate management alternativesturbidity model to evaluate management alternatives

1010

Features of Turbidity ModelFeatures of Turbidity Model

Two-dimensional (longitudinal-vertical), laterally Two-dimensional (longitudinal-vertical), laterally averaged transport framework (CE-QUAL-W2)averaged transport framework (CE-QUAL-W2)

State variables: Temperature (T) and turbidity (Tn)State variables: Temperature (T) and turbidity (Tn) Three size classes of TnThree size classes of Tn Source of Tn: external loadingSource of Tn: external loading Sinks: settling, export (via withdrawal, spill, waste Sinks: settling, export (via withdrawal, spill, waste

channel diversion)channel diversion) Two basins simulated separatelyTwo basins simulated separately

1111

Model Grid – West BasinModel Grid – West Basin

2

3

45

6

7

89

1011

1213

1415

1617 18

1920

2122

2324

2526

2728

27 segments (~330 m avg)47 layers (1 m)1 branch

Esopus Creek

dividing weirdividing weir

1212

Model Grid – East BasinModel Grid – East Basin

2 34

56

78

910

1112

1314

1516

1718

1920

2122

2324

2526

2728 29 30 3132 33 34 35 36 37

38

37 segments (~ 300 m avg)26 layers (1 m)1 branch

spill

dividing dividing weirweir

1313

Model Grid – Vertical LayersModel Grid – Vertical LayersDistance from weir (m)

0 2000 4000 6000 8000 10000 12000

Ele

vatio

n (m

)

130

140

150

160

170

180

div

idin

g w

eir

west basin east basin

1414

Turbidity (Tn)Turbidity (Tn) primary metric of quality for water suppliesprimary metric of quality for water supplies measure of light scattering by particles at 90° collection measure of light scattering by particles at 90° collection

angle, units of NTUangle, units of NTU

Tn Tn α α bb;; supported in peer-reviewed literature supported in peer-reviewed literature bb, Tn = , Tn = f f (particle concentration, size distribution, (particle concentration, size distribution,

composition, shape)composition, shape)

0°

90°

incident beam

scattered light

1Tn Tn αα

light scattering coefficient (b, m-1)

1

360

0bb

1515

Scattering (Scattering (bb) and Turbidity (Tn): ) and Turbidity (Tn): Behaves Like Intensive PropertiesBehaves Like Intensive Properties

mass balance calculations can be donemass balance calculations can be done well-established in optical literature (well-established in optical literature (Davies-Colley et al. 1993Davies-Colley et al. 1993))

Q1, b1, Tn1

Q2, b2, Tn2

Q, b, Tn

example

Q

bQbQb 2211

Q

TnQTnQTn 2211

Q = Q1 + Q2

1616

Turbidity: As the Model State VariableTurbidity: As the Model State Variable

Tn is the regulated parameterTn is the regulated parameter

disadvantages of TSS (a gravimetric measurement) as an disadvantages of TSS (a gravimetric measurement) as an alternative (would have to rely on Tn = alternative (would have to rely on Tn = kk · TSS) · TSS)

differences in particle size and composition dependencies of Tn differences in particle size and composition dependencies of Tn and TSSand TSS

Tn, Tn, bb (scattering) and (scattering) and cc (beam attenuation) measurements more (beam attenuation) measurements more preciseprecise

limitations in temporal and spatial resolution; e.g., robotic and limitations in temporal and spatial resolution; e.g., robotic and rapid profiling capabilities for Tn and rapid profiling capabilities for Tn and cc

pore size for TSS measurements too large (1.7 µm)pore size for TSS measurements too large (1.7 µm) variation in relationship between Tn and TSS in time and space variation in relationship between Tn and TSS in time and space

(i.e., (i.e., kk is not really a constant) is not really a constant)

Tn, [and Tn, [and cc] supported in peer-reviewed literature, without ] supported in peer-reviewed literature, without published critical commentspublished critical comments

1717

Model InputsModel Inputs Model testing period: 2003-2007Model testing period: 2003-2007

supported by UFI’s intensive (Robohut on Esopus Creek, in-supported by UFI’s intensive (Robohut on Esopus Creek, in-reservoir robots) and DEP’s routine monitoring datareservoir robots) and DEP’s routine monitoring data

constrained by the availability of operations dataconstrained by the availability of operations data Additional (secondary) validation period: 1995-2002Additional (secondary) validation period: 1995-2002

Operations data Operations data Hydrologic inputs/outputsHydrologic inputs/outputs Loading of turbidityLoading of turbidity Creek temperature Creek temperature Meteorological dataMeteorological data

1818

In-Reservoir Robots: Example, 2007In-Reservoir Robots: Example, 2007

site 2

site 1.4

site 3.1

site 4.2

April – November (June in 2007)depth-profiles every 6 hoursdepth interval 1 m

1919

In-Reservoir Rapid ProfilingIn-Reservoir Rapid ProfilingExample, 11/30/2006Example, 11/30/2006

00

09

01

020304

0507

08

10

11

12

13

1415

1617

18

1920

2122

06

23

2425

262728

29

3031323334

after major runoff eventsdepth interval 0.25 m

2020

Example of Driving Conditions and Example of Driving Conditions and Reservoir Response: June 2006Reservoir Response: June 2006

(a)

Q (

m3 ·s

-1)

10

100

1000

(b)

Tn

(NT

U)

10

100

1000

(c)

6/24/06 6/28/06 7/2/06 7/6/06 7/10/06 7/14/06

Te

mp

era

ture

(°C

)

12

18

Esopus Creek

1 m, site 1.4 (f)

6/24/06 6/28/06 7/2/06 7/6/06 7/10/06 7/14/06

Tn,

w (

NT

U)

0

50

100

(d) Tn, site 3.12040

40

6080100

160140120100 80

8060

Dep

th (

m)

10

20

(e) Tn, site 4.2

20

2020

40606040

40

4040

Dep

th (

m)

10

20

2121

Turbidity-Causing ParticlesTurbidity-Causing Particles

(b)

Particle Size (d, µm)

0.1 1 10

Cum

ulat

ive

b m(6

60)

(%)

0

20

40

60

80

100

(a)

F(d

)(p

artic

les·

L-1·µ

m-1

)

106

107

108

109

1010

1011

Esopus Creek

Ashokan - West basin

Esopus Creek

Ashokan - West basin

1.7 µm

Four Features:Four Features:1.1. number concentrationnumber concentration2.2. size distributionsize distribution3.3. compositioncomposition4.4. shapeshape

Individual Particle Analysis Individual Particle Analysis (IPA) Technology(IPA) Technology

• 75-80% clay75-80% clay• Tn associated with 1-10 µTn associated with 1-10 µ• sub-µ particles unimportantsub-µ particles unimportant• TSS filter pore size 1.7 µm; TSS filter pore size 1.7 µm;

misses some turbidity misses some turbidity causing particlescausing particles

April 2005

bbmm(660) – minerogenic particle scattering coefficient, m(660) – minerogenic particle scattering coefficient, m-1-1

2222

““Turbidity” Size-Classes for ModelTurbidity” Size-Classes for Model

0.1 1 10

0

20

40

60

80

100

Cum

ulat

ive b m

(660

) (%

)

Particle size (m)

ClassClass size size (µm)(µm)

Size Size rangerange

velvel

(m/d)(m/d)

11 11 < 1.75< 1.75 0.0750.075

22 3.143.14 1.75-1.75-5.755.75

0.750.75

33 8.118.11 > 5.75> 5.75 5.05.0

ClassClass Q ≤ 40 Q ≤ 40 mm33/s/s

Q > 40 Q > 40 mm33/s/s

11 10%10% 10%10%

22 65%65% 45%45%

33 25%25% 45%45%

Fractions in Esopus CreekFractions in Esopus CreekStokes Law:

18/)( 2dgSV wp

coefficient specification constrained by coefficient specification constrained by reality of particle characteristics as reality of particle characteristics as obtained from IPAobtained from IPA

Esopus Creek

2323

Hydrothermal Model PerformanceHydrothermal Model Performance

Predicted Tw (°C)

0 5 10 15 20 25

Obs

erve

d T

w (

°C)

0

5

10

15

20

25(b)2003-2007r2 = 0.98RMSE = 1.20 °C

Predicted Tw (°C)

0 5 10 15 20 25

Obs

erve

d T

w (

°C)

0

5

10

15

20

25(c)1995-2002r2 = 0.95RMSE = 1.99 °C

* withdrawal temperature (Tw)* importance of withdrawal depth information

2003-2007 1995-2002

2424

Turbidity Model PerformanceTurbidity Model Performance* withdrawal turbidity (Tn,w)* importance of detailed monitoring of forcing conditions

2003-2007

Predicted Tn,w (NTU)10 100

Ob

serv

ed T

n,w (

NT

U)

10

100

r2 = 0.81 RMSE = 9.5 NTU

1995-2002

Predicted Tn,w (NTU)10 100

Ob

serv

ed T

n,w (

NT

U)

10

100

r2 = 0.48 RMSE = 13.2 NTU

2525

Turbidity Model PerformanceTurbidity Model Performance

site 3.1, 0-5 m avg

Tn

(N

TU

)

0

50

100

150

observedpredicted

site 3.1, 5-10 m avg

Tn

(N

TU

)

0

50

100

Q (

m3 ·s

-1)

0

400

800

withdrawal

10/1/05 11/1/05 12/1/05 1/1/06 2/1/06 3/1/06 4/1/06 5/1/06 6/1/06 7/1/06

Tn

,w (

NT

U)

0

50

100

alumtreatment

2626

Turbidity Model Turbidity Model PerformancePerformance

10

10

10

10

20

20

20

30

4050

6070

40

50

8090

100

Distance from dividing weir (km)

0 1 2 3 4 5 6 7 8

Ele

vatio

n (m

)

150

165

180

10

10

1020

20

20

10

30

4050

1020

30

60

2010

60

10

20

20

10

70

70

8080

20

80

Distance from dividing weir (km)0 1 2 3 4 5 6 7 8

Ele

vatio

n (m

)

150

165

180

observed Tn on 6/30/06 (east basin)

predicted Tn on 6/30/06 (east basin)

East Basin6/30/2006

2727

Alum treatment events

*

* Normalized RMSE (Gelda and Effler, 2007)

performance for well monitored years consistent with that reported for Schoharie Reservoir (Gelda and Effler, 2007)

Performance SummaryPerformance Summary

2828

SummarySummary

2-D model CE-QUAL-W2 as transport framework2-D model CE-QUAL-W2 as transport framework Turbidity as a state variableTurbidity as a state variable Characterization of turbidity-causing particlesCharacterization of turbidity-causing particles Three size classesThree size classes Model performed well in simulating in-reservoir and Model performed well in simulating in-reservoir and

withdrawal temperature and turbiditywithdrawal temperature and turbidity Model is suitable for evaluating management alternativesModel is suitable for evaluating management alternatives Future research: resuspension, particle-based modeling Future research: resuspension, particle-based modeling

including aggregationincluding aggregation

Gelda, R. K., S. W. Effler, F. Peng, E. M. Owens and D. C. Pierson, 2009. Turbidity Gelda, R. K., S. W. Effler, F. Peng, E. M. Owens and D. C. Pierson, 2009. Turbidity model for Ashokan Reservoir, New York: Case Study. J. Environ. Eng. model for Ashokan Reservoir, New York: Case Study. J. Environ. Eng. 135:135: 885-895. 885-895.

e-mail: RKGelda@UpstateFreshwater.Orge-mail: RKGelda@UpstateFreshwater.Org

2929

Ashokan ReservoirEast Basin Spillway4/2005