General Electric Housatonic River PCB BioaccumulationPCB Bioaccumulation Site, Rest of River Model Calibration:Model Calibration:

Document Food Chain Model (FCM)Food Chain Model (FCM)

Overview Meeting April 2005

Pittsfield, MA Gary Lawrence (EVS Environment Consultants)

Jonathan Clough (Warren Pinnacle Consulting)

Presentation TopicsPresentation Topics

1. FCM Overview2. Performance Summary3. Conceptual Model4. Structure and Application5. Model Calibration6. Sensitivity and Uncertainty7. Conclusions

1. FCM Overview 2. Performance Summary 3. Conceptual Model 4. Structure and Application 5. Model Calibration 6. Sensitivity and Uncertainty 7. Conclusions

Topic 1:FCM Overview

Topic 1:FCM Overview

Sunfish (bluegill)

Model TypeModel Type

• Mass balance model based on QEAFDCHN v.1.0 • Bioenergetic model (food sources, food quality, and

species activity are sensitive inputs)• Time-dependent • “Decoupled”

water/sediment• Fish age-classes

specified

• Mass balance model based on QEAFDCHN v.1.0 • Bioenergetic model (food sources, food quality, and

species activity are sensitive inputs) • Time-dependent • “Decoupled”

water/sediment • Fish age-classes

specified

Years

Purpose of Bioaccumulation ModelPurpose of Bioaccumulation Model

1. Identify how aquaticorganisms obtain PCBburdens (water vs.sediment)

1. Identify how aquaticorganisms obtain PCBburdens (water vs.sediment)

0

10

20

30

40

50

0 10 20 30 40 50

Tiss

ue P

CB

(mg/

kg w

w)

Time

2. Estimate the time needed for PCBs in biota tissue to be reduced to acceptablelevels.

2. Estimate the time needed for PCBs in biota tissue to be reduced to acceptablelevels.

Contaminants SimulatedContaminants Simulated

• Total PCBs (similar toAroclor 1260)

• 9 Individual congeners ¾ “Dioxin-like” PCBs:

(PCB-77, PCB-123,PCB-126)¾ High-concentration

PCB congeners:(PCB-101, PCB-118,PCB-138, PCB-156, PCB-177, PCB-183)

• Total PCBs (similar toAroclor 1260)

• 9 Individual congeners ¾ “Dioxin-like” PCBs:

(PCB-77, PCB-123,PCB-126) ¾ High-concentration

PCB congeners:(PCB-101, PCB-118,PCB-138, PCB-156, PCB-177, PCB-183)

1. “Field data” – measured PCB inputs2. “Linked model” with EFDC PCB inputs • Approach identifies uncertainties specific

to biological model components • Same biological parameter set was used

for both calibrations• We emphasize linked model results in this

presentation

1. “Field data” – measured PCB inputs 2. “Linked model” with EFDC PCB inputs • Approach identifies uncertainties specific

to biological model components • Same biological parameter set was used

for both calibrations • We emphasize linked model results in this

presentation

Two Applications of FCMTwo Applications of FCM

Topic 2:Model Performance

Topic 2:Model Performance

Largemouth bass

QAPP TargetsQAPP Targets • Calibration targets developed a priori (MFD;

Weston, 2004; Appendix I)• Not absolute criteria for acceptance/rejection

• Calibration targets developed a priori (MFD; Weston, 2004; Appendix I)

• Not absolute criteria for acceptance/rejection

tPC

B (m

g/kg

wet

)

PCBs in Invertebrates (Simulated VersusMeasured) – Linked FCM Model

PCBs in Invertebrates (Simulated VersusMeasured) – Linked FCM Model

Field Data (D-Net) (n=15)

FCM Epifauna

FCM Infauna FCM Water-Column Invertebrates

0

10

20

30

40

50

60 PS

A

Reac

h 5A

Reac

h 5B

Reac

h 5C

Reac

h 5D

Reac

h 6

Reac

h 5A

Reac

h 5B

Reac

h 5C

Reac

h 5D

Reac

h 6

Reac

h 5A

Reac

h 5B

Reac

h 5C

Reac

h 5D

Reac

h 6

Invertebrate Type/River Reach

tPC

B (m

g/kg

ww

)

tPC

B (m

g/kg

wet

)

(Predicted) (Predicted) (Predicted) (Predicted)

(Predicted) (Observed)

(Observed) (Observed)

Comparison of Measured Biota Tissue tPCBConcentrations to Linked Model SimulationsComparison of Measured Biota Tissue tPCBConcentrations to Linked Model Simulations

0

20

40

60

80

100

120

140

160

180 Reach 5A Reach 5B Reach 5C Reach 5D Woods Pond Reach 5A Reach 5B/5C Woods Pond

Invertebrates

Reac

h 5

D-N

et

Epi

faun

a

Infa

una

Wat

er-

Col

umn

Fish Species

Brown Bullhead

White Sucker

Cyprinid Minnow

Sunfish (Pumpkinseed)

Largemouth Bass

(Simulated) (Simulated) (Simulated)

(Simulated) (Measured)

(Measured)

(Simulated)

(Measured)

tPC

B (m

g/kg

ww

)

Error bars represent two standard errors

tPC

B (m

g/kg

wet

)

(Predicted)redicted)

(Predicted) (Predicted)

redicted) (Observed)

(Observed)bserved)

0

20

40

60

80

100

120

140 Reach 5A Reach 5B (P Reach 5C Reach 5D Woods Pond (P Reach 5A Reach 5B/5C Woods Pond (O

Invertebrates

Reac

h 5

Kic

k-Ne

t

Epi

faun

a

Infa

una

Wat

er-

Col

umn

Fish Species

Brown Bullhead

White Sucker

Cyprinid Minnow

Sunfish (Pumpkinseed)

Largemouth Bass

Comparison of Measured Biota Tissue tPCBConcentrations to Field Data Simulations

Comparison of Measured Biota Tissue tPCBConcentrations to Field Data Simulations

(Simulated) (Simulated) (Simulated)

(Simulated) (Measured)

(Measured)

(Simulated)

(Measured)

tPC

B (m

g/kg

ww

)

Error bars represent two standard errors

(Simulated)(Simulated)(Simulated)

(Simulated)(Measured)

(Measured)

(Simulated)

(Measured)

tPC

B (m

g/kg

lipi

d)

Comparison of Lipid Normalized tPCBConcentrations to Linked Model Simulations

Comparison of Lipid Normalized tPCB Concentrations to Linked Model Simulations

0

500

1000

1500

2000

2500

3000

3500

4000

4500

5000 Reach 5A (Predicted) Reach 5B (Predicted) Reach 5C (Predicted) Reach 5D (Predicted) Woods Pond (Predicted) Reach 5A (Observed) Reach 5B/5C (Observed) Woods Pond (Observed)

Invertebrates

Reac

h 5

Kic

k-Ne

t

Epifa

una

Infa

una

Wat

er-

Col

umn

Fish Species

Brown Bullhead

White Sucker

Cyprinid Minnow

Sunfish (Pumpkinseed)

Largemouth Bass

tPC

B (m

g/kg

lipi

d)

error bars reflect two standard errors

Types of Performance MeasuresTypes of Performance Measures

1.1. “Goodness-of-fit”“Goodness-of-fit” statisticsstatistics

2.2. Visual inspection ofVisual inspection of simulated versussimulated versus measured PCBs:measured PCBs: ¾¾ by fish ageby fish age ¾¾ by speciesby species ¾¾ by PSA reachby PSA reach ¾¾ by chemicalby chemical ¾¾ wet weight and lipid-wet weight and lipid-

normalized PCBsnormalized PCBs

3.3. Model bias checkModel bias check

Bias Percent error

R2

Mean average error Variation

Bias Measures Precision

Measures

Reach 5A (n =3 6) 5B (n = 15) 5BC (n = 80) 5C (n = 9)

6

(n = 107)

Mean Error (mg/kg wet) 32.40 17.93 6.22 15.83 -11.04

Mean Absolute Error (mg/kg wet) 47.93 21.92 22.60 29.65 27.22

Mean Percent Error 26% 25% -6% 11% -38%

Mean Absolute Percent Error 72% 42% 43% 45% 53%

“Goodness of Fit” Statistics“Goodness of Fit” Statistics

Central Tendency Sample Size n = 1 n >1 n >3 n >6

Number of Data Groupings 270 490 240 60

R2 (Untransformed) 0.59 0.73 0.85 0.93

R2 (log-log transformed) 0.88 0.94 0.95 0.96

tPC

B (m

g/kg

wet

)Largemouth Bass – tPCB Versus Age

(Linked Model)Largemouth Bass – tPCB Versus Age

(Linked Model)

Reach 5B, 5C, and 5D

0

100

200

300

400

0 2 4 6 8 10 12 14

Age (Measured)

Reach 5BC FCM Reach 5B FCM Reach 5C FCM Reach 5D

tPC

B (m

g/kg

ww

)

PC

B (m

g/kg

wet

)Sunfish – tPCB Versus Age

(Linked Model)Sunfish – tPCB Versus Age

(Linked Model)

Reach 5B, 5C, and 5D

0

20

40

60

80

100

0 1 2 3 4 5 6 7

Age (Estimated)

Reach 5BC FCM Reach 5B FCM Reach 5C FCM Reach 5D

tPC

B (m

g/kg

ww

)

PCB

(mg/

kg w

et)

White Sucker – tPCB Versus Age(Linked Model)

White Sucker – tPCB Versus Age(Linked Model)

Reach 5B, 5C, and 5D

0

50

100

150

200

250

0 1 2 3 4 5 6 7 8 9 Age (Measured or Estimated)

Reach 5B FCM Reach 5B FCM Reach 5C FCM Reach 5D

tPC

B (m

g/kg

ww

)

mg/

kg w

et w

eigh

t

PCB Congeners in Invertebrates: FCMSimulations Versus Field MeasurementsPCB Congeners in Invertebrates: FCM

Simulations Versus Field Measurements

0.0001

0.001

0.01

0.1

1

10 P

CB

77

PC

B10

1

PC

B11

8

PC

B12

6

PC

B13

8

PC

B12

3

PC

B15

6

PC

B17

7

PC

B18

3

PCB Congener

Mean of Reach 5 D-Net Samples (n=15) Average of Modeled Reach 5 Epifauna

coplanar coplanar coplanar

PC

B (m

g/kg

ww

)

PCB

183

(mg/

kg w

et)

(Predicted) (Predicted) (Predicted) (Predicted)

Predicted) (Observed)

(Observed)Observed)

PCB-183: FCM Simulations Versus Field Measurements (Linked Model)

PCB-183: FCM Simulations Versus Field Measurements (Linked Model)

0

0.5

1

1.5

2

2.5

3

3.5

4 Reach 5A Reach 5B Reach 5C Reach 5D Woods Pond ( Reach 5A Reach 5B/5C Woods Pond (

Invertebrates

Rea

ch 5

D-N

et

Epifa

una

Infa

una

Wat

er-

Col

umn

Fish Species

Brown Bullhead

White Sucker

Cyprinid Minnow

Sunfish (Pumpkinseed)

Largemouth Bass

(Simulated)

(Simulated)

(Simulated) (Simulated)

(Simulated)

(Measured) (Measured)

(Measured)

PCB

-183

(mg/

kg-w

w)

PCB

77 (m

g/kg

wet

)

(Predicted) (Predicted) (Predicted) (Predicted)

Predicted) (Observed)

(Observed)Observed)

PCB-77: FCM Simulations Versus Field Measurements (Linked Model)

PCB-77: FCM Simulations Versus Field Measurements (Linked Model)

0.000

0.002

0.004

0.006

0.008

0.010

0.012

0.014 Reach 5A Reach 5B Reach 5C Reach 5D Woods Pond ( Reach 5A Reach 5B/5C Woods Pond (

Invertebrates

Rea

ch 5

D-N

et

Epifa

una

Infa

una

Wat

er-

Col

umn

Fish Species

Brown Bullhead

White Sucker

Cyprinid Minnow

Sunfish (Pumpkinseed)

Largemouth Bass

(Simulated)

(Simulated) (Simulated) (Simulated)

(Simulated)

(Measured) (Measured)

(Measured)

PCB

-77

(mg/

kg-w

w)

Check for Model BiasCheck for Model Bias

1000 Overprediction by 5-fold

•• Looked forLooked for systematic over-systematic over- oror under-predictionsunder-predictions

No bias observed•• No bias observed as function of:as function of: ¾¾ PCB typePCB type

¾¾ TrophicTrophic levellevel ¾¾ Age or lipidAge or lipid S

imul

ated

con

cent

ratio

n (m

g/kg

ww

)Pr

edic

ted

Fish

Tis

sue

Con

cent

ratio

ns (m

g/kg

wet

)

Overprediction by 2-fold Perfect fit line

100 Underprediction by 2-fold Underprediction by 5-fold Unique Age, Species, Reach Mean (1 < n < 6) tPCB

10 Mean (n > 6)

1

Principal PCB 0.1 congeners

0.01

0.001 Coplanar PCB congeners

0.0001 0.0001 0.001 0.01 0.1 1 10 100 1000

Measured concentration (mg/kg ww)Observed Concentrations (mg/kg wet)

Topic 3:Conceptual Model

Topic 3:Conceptual Model

White sucker

Reach Pomeroy Avenue 5B

RoRoaarriinngg BrBrooookkEast/West Branch LOCATOR MAP

Confluence FCM Spatial DomainFCM Spatial DomainSpatial DomainHolmes Road

MAof Food Chain Yokun Brook CTModel (FCM)Reach Sackett

5A Brook

Reach 5C

N

W E

Pittsfield S

WWTP 0.2 0 0.2 0.4 Miles

Woods Pond Headwaters

Reach 5B

LEGEND New Lenox Reach

Road 6 Backwater RegionsWoods Pond FCM Domain Main Channel

Footbridge 10-Year Floodplain Woods Pond #

Dam

Habitat TypesHabitat Types

5A5A 5B5B

6 (Woods Pond)6 (Woods Pond)5C/5D5C/5D

UpstreamUpstream

DownstreamDownstream

Conceptual Model – Upstream PSAConceptual Model – Upstream PSA

Conceptual Model – Downstream PSAConceptual Model – Downstream PSA

Topic 4:Model Structure and Application

Topic 4:Model Structure and Application

Brown bullhead

Bioaccumulation ProcessesBioaccumulation Processes

• PCB Uptake: a. Dietary uptakeb. Respiration

• PCB Elimination c. Depuration (elimination via gill and feces)d. Growth dilution

• Fish modeled as “time-dependent”; invertebrates modeled as “steady-state”:

• PCB Uptake: a. Dietary uptake b. Respiration

• PCB Elimination c. Depuration (elimination via gill and feces) d. Growth dilution

• Fish modeled as “time-dependent”; invertebrates modeled as “steady-state”:

a d

b c

iiij

n

j ijcui

i vGKvCcKdt dv )(

1 +−+= ∑

=

α

b a c & d

dc ba

loss uptakev i +

+ ==

Model LinkagesModel Linkages

Exposure Inputs from EFDC:1. PCB in bed sediment

organic carbon2. PCB in water column

organic matter (POM)3. Dissolved PCB in sediment

pore water4. Dissolved PCB in water

column

Exposure Inputs from EFDC: 1. PCB in bed sediment

organic carbon 2. PCB in water column

organic matter (POM) 3. Dissolved PCB in sediment

pore water 4. Dissolved PCB in water

column

PCB Exposure Inputs (EFDC)

Temperature (HSPF)

Bioaccumulation Model (FCM)

2

4

1 3

• Linked model – Individual cells were aggregated by subreach

• Reach 5A – Correction to sediment tPCB made to account for low OC samples and bioavailability differences

• Linked model – Individual cells were aggregated by subreach

• Reach 5A – Correction to sediment tPCB made to account for low OC samples and bioavailability differences

Exposure Data ProcessingExposure Data Processing

PCB Exposure Inputs(Congeners)

PCB Exposure Inputs(Congeners)

-2 -1 0 1 2 3 4 5 Sum of Congeners (Log ug/g OC)

-3

-2

-1

0

1

PC

B 7

7 (L

og u

g/g

OC

)

Log[PCB-77] = -2.8584 + 0.6445 • Log[tPCB] n = 45 R2 = 0.680 p = 3.3E-12

-2 -1 0 1 2 3 4 5 Sum of Congeners (Log ug/g OC)

-3

-2

-1

0

1

2

3

PC

B 1

01 (L

og u

g/g

OC

)

Log[PCB-101] = -1.3722 + 0.9212 • Log[tPCB] n = 109 R2 = 0.984 p = 1.0E-15

Sediment PCB-101

Sediment PCB-77

Model ParameterizationModel Parameterization

• Energy balance parameters (e.g., sediment energydensity, lipids, protein, etc.) (C.4);

• Fish respiration rates (C.8); • Fish growth rates (C.6); • Invertebrate “rate parameters” (growth rates,

respiration rates, and PCB elimination rates) (C.5,C.7, C.10);

• Chemical-specific parameters (e.g., PCBassimilation efficiency, partitioning constants) (C.9);

• Invertebrate/fish feeding preferences (C.11-C.13).

• Energy balance parameters (e.g., sediment energydensity, lipids, protein, etc.) (C.4);

• Fish respiration rates (C.8); • Fish growth rates (C.6); • Invertebrate “rate parameters” (growth rates,

respiration rates, and PCB elimination rates) (C.5,C.7, C.10);

• Chemical-specific parameters (e.g., PCBassimilation efficiency, partitioning constants) (C.9);

• Invertebrate/fish feeding preferences (C.11-C.13).

Inve

rtebr

ate

Res

pira

tion

Res

pira

tion

Rat

e (k

J/g-

wet

day

)

10

1

0.1

0.01

Other OdonateCaddisfly

Amphipod Zooplankton

Oligochaete

Polychaete Mayfly Shelled

Epifauna Mosquito

Chironomid

Organism Type

Individual Result Median of Group

Topic 5:Model Calibration

Topic 5:Model Calibration

Fallfish

General Calibration StrategyGeneral Calibration Strategy simulate fine-tune

compare

Kept most parameters uncalibrated•• Kept most parameters uncalibrated ((e.g., usede.g., used initial estimate from literature review)initial estimate from literature review) Some small-scale refinements to initial parameter•• Some small-scale refinements to initial parameter estimates made to improve modelestimates made to improve model Parameters always constrained to ranges of•• Parameters always constrained to ranges of scientifically plausible valuesscientifically plausible values

Calibration #1 – Fallfish Respiration Rate

Calibration #1 – Fallfish Respiration Rate

• No fallfish respiration data • Initially assumed to be similar to small cyprinids• Revised: average of white sucker and dace

• No fallfish respiration data • Initially assumed to be similar to small cyprinids • Revised: average of white sucker and dace

Fallfish [cyprinid benthivore]

Shiner

White Sucker [large benthivore]

Dace

[Small cyprinids]

Calibration #2 – Chemical Assimilation Efficiency

Calibration #2 – Chemical Assimilation Efficiency

Invertebrates Fish

Dietary Assimilation Efficiency Normalized to Food Assimilation Efficiency (Parkerton, 1993)

Calibration #3 – LargemouthBass Dietary Matrix

Calibration #3 – LargemouthBass Dietary Matrix

Prey Age

Class 0+

Age Class

1+

Age Class

2+

Age Class

3+

Age Class

4+

Age Class

5+

Age Class

6+

Age Class

7+

Age Class

8+

Age Class

9+

Water column inverts 30 20 15 10 0 0 0 0 0 0

Epifauna inverts 25 20 20 20 15 15 15 15 15 15

Age 0+ cyprinids 25 10 5 5 5 5 5 5 5 5

Age 1+ cyprinids 0 15 10 10 10 10 10 10 10 10

Age 2+ cyprinids 0 10 20 20 20 15 15 15 15 15

Age 3+ cyprinids 0 0 0 10 10 15 15 15 15 15

Age 0+ sunfish 10 15 15 10 5 5 5 5 5 5

Age 1+ sunfish 0 0 0 5 15 15 15 15 15 15

Age 0+ gamefish 10 10 15 5 5 5 5 5 5 5

Age 1+ gamefish 0 0 0 5 15 15 15 15 15 15

Rea

ch 5

A L

arge

mou

th B

ass

• Site-specific growth data available (annual increments)

• Warmwater fish growth is limited to warm months (>10ºC)

• Model input file revised to constrain growth to Julian dates with T > 10ºC

• No change to model code required • Improved fit for YOY fish

• Site-specific growth data available (annual increments)

• Warmwater fish growth is limited to warm months (>10ºC)

• Model input file revised to constrain growth to Julian dates with T > 10ºC

• No change to model code required • Improved fit for YOY fish

Calibration #4 – Seasonal Fish Growth

Calibration #4 – Seasonal Fish Growth

• Depression of BSAFs for some coplanarPCBs:– due to differences in rates of metabolism

in fish and/or its underlying food web– potential differences in chemical

bioavailability (e.g., small amount of sootcarbon) (Burkhard et al., 2004)

• Application of empirical scaling factors forPCB-77 and PCB-123 – for fish, noassumption about mechanism for BSAFdepression

• Depression of BSAFs for some coplanarPCBs: – due to differences in rates of metabolism

in fish and/or its underlying food web – potential differences in chemical

bioavailability (e.g., small amount of sootcarbon) (Burkhard et al., 2004)

• Application of empirical scaling factors forPCB-77 and PCB-123 – for fish, noassumption about mechanism for BSAFdepression

Calibration #5: CoplanarPCB Adjustment

Calibration #5: CoplanarPCB Adjustment

Topic 6:Sensitivity and Uncertainty

Topic 6:Sensitivity and Uncertainty

Epifaunal invertebrates

• Iterative procedure:1. Process screening (MFD)2. Test reach model3. Preliminary calibration4. “Final” calibration5. Validation

• Both qualitative and quantitative approaches.

• Iterative procedure: 1. Process screening (MFD) 2. Test reach model 3. Preliminary calibration 4. “Final” calibration 5. Validation

• Both qualitative and quantitative approaches.

Sensitivity/UncertaintyApproach

Sensitivity/UncertaintyApproach

Sensitivity Analysis - MethodsSensitivity Analysis - Methods

• Automated sensitivity procedures – varied parameters by 10% and 33%

• “Local sensitivity analysis” method (deterministic)• Sensitivity = percent change in model output

relative to percent change in parameter• Separate sensitivities calculated for each trophic

level• High sensitivity parameters propagated to formal

uncertainty assessment

• Automated sensitivity procedures – varied parameters by 10% and 33%

• “Local sensitivity analysis” method (deterministic) • Sensitivity = percent change in model output

relative to percent change in parameter • Separate sensitivities calculated for each trophic

level • High sensitivity parameters propagated to formal

uncertainty assessment

PctChanged ticDeterminis

MinResMaxRes

ySensitivit⎟ ⎠ ⎞

⎜ ⎝ ⎛

⋅ −

= 2

Highest Bass Sensitivities (tPCB)Highest Bass Sensitivities (tPCB) Parameter Sensitivity

Bass Gamma for Respiration 131.7%

Bass Food Assimilation Efficiency 125.3%

Bass Chemical Assimilation Efficiency 113.8%

Bass Respiration Rate 79.6%

Particulate Energy Density 49.9%

tPCBs Adsorbed to POM 46.8%

Water Column Invertebrate Respiration 45.2%

W.C. Invertebrate Chemical Elimination Rate 44.8%

W.C. Invertebrate Food Assimilation Efficiency 44.0%

W.C. Invertebrate Chemical Assim.Efficiency 41.3%

Sunfish Assimilation Efficiency 36.1%

Epifauna Assimilation Efficiency 33.9%

Sunfish Food Assimilation Efficiency 33.8%

Sunfish Chemical Assimilation Efficiency 33.8%

Epifauna Chemical Assimilation Efficiency 31.8%

Bass Feeding Preference – Sunfish 31.3%

tPCB, Kow 30.8%

• Monte Carlo modeling of 10 groups of sensitive input parameters

• Deterministic calibrated output compared to non­calibrated output distribution (by species/reach)

• Issues:– Incertitude vs. variability (i.e., individuals versus central

tendencies)– Some professional judgment in distributions necessary– Correlated inputs (e.g., assimilation efficiency, dietary

matrices)

• Monte Carlo modeling of 10 groups of sensitive input parameters

• Deterministic calibrated output compared to non-calibrated output distribution (by species/reach)

• Issues: – Incertitude vs. variability (i.e., individuals versus central

tendencies) – Some professional judgment in distributions necessary – Correlated inputs (e.g., assimilation efficiency, dietary

matrices)

Uncertainty Analysis -Methods

Uncertainty Analysis ­Methods

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

Proportion per B

ar

0.00

0.01

0.02

0.03

0.04

0.05

Proportion per B

ar

tPCB in Largemouth Bass V, Reach 5C, mg/kg wet

5 6 7 8 Total PCBs, Log Kow

0

100

200

300

400

500

600

700

800

Cou

nt

0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 Bullhead: Mult. Factor, Respiration

0

100

200

300

400

500

Cou

nt

Sample Input: log KOW

0 50 100 150 200 0

100

200

300

400

500

600

Cou

nt

0.00

0.01

0.02

0.03

0.04

0.05

0.06

Proportion per B

ar

Monte-Carlo Uncertainty ModelMonte-Carlo Uncertainty Model

Sample Output Distribution: Bass Tissue Concentrations

Deterministic simulation

(calibrated)

Probabilistic simulation

(non-calibrated)

tPCB in Reach 5C Largemouth Bass Age 5+ (mg/kg ww)

Sample Input: Bullhead Respiration

Topic 7:Conclusions

Topic 7:Conclusions

Water column invertebrates

ConclusionsConclusions

• Reasonably predicted PCB burdens at multiple levels of the aquatic food chain (tPCBs, congeners)

• Calibration performance criteria specified in the MFD/QAPP were achieved

• Reasonably predicted PCB burdens at multiple levels of the aquatic food chain (tPCBs, congeners)

• Calibration performance criteria specified in the MFD/QAPP were achieved