Overview of TM6:
Simulation of Selenium Fate and
Transport in North San Francisco
Bay
Limin Chen, Sujoy Roy, and Tom GriebTetra Tech, Inc., Lafayette, CA
Presentation to TMDL Advisory Committee
April 28, 2010
Overview� Goal: Develop tool to calculate selenium in water and
biota in response to different loads of selenium entering North San Francisco Bay
� Technical Review Process
� Modeling Approach
� Selenium Loads
� Example Calibration Results
� Predicted Loads and Concentrations
� Role of Boundary Conditions
� Model Scenarios
Technical Review Committee
(2007-2010)� Dr. Nicholas S. Fisher, State University of New York,
Stony Brook
� Dr. Regina G. Linville, California State Office of Environmental Health Hazard Assessment
� Dr. Samuel N. Luoma, Emeritus, U.S. Geological Survey
� Dr. John J. Oram, San Francisco Estuary Institute
The role of the Technical Review Committee was to provide expert
reviews of the modeling process as well as credible technical
advice on specific issues arising during the review.
Final TM-6 report includes their comments and our responses.
Model Structure
Total Particulate Selenium as a
Mix of Organic and Inorganic
Species (µg/g)
Uptake by bivalves
Uptake by predator species
ECoSModel
Results
DYMBAM
ModelTTF
SelenateDSe(VI)
SeleniteDSe(IV)
Organic
Selenide
DSe(-II)
Selenate +Selenite
PSe(IV+VI)
Elemental Se
PSe(0)
Organic Selenide
PSe(-II)
Bed Sediments
Dissolved species
Particulate species
Point Sources (Refineries, POTWs, Other Dischargers)
Contribute primarily to
suspended particulates, and
to dissolved phase in a limited way
Contribute to dissolved
phase
Estuary Water Column
River and
Tributary Loads
SelenateDSe(VI)
SeleniteDSe(IV)
Organic
Selenide
DSe(-II)
Selenate +Selenite
PSe(IV+VI)
Elemental Se
PSe(0)
Organic Selenide
PSe(-II)
Bed Sediments
Dissolved species
Particulate species
Point Sources (Refineries, POTWs, Other Dischargers)
Contribute primarily to
suspended particulates, and
to dissolved phase in a limited way
Contribute to dissolved
phase
Estuary Water Column
River and
Tributary Loads
Total Particulate Selenium as a
Mix of Organic and Inorganic
Species (µg/g)
Uptake by bivalves
Uptake by predator species
ECoSModel
Results
DYMBAM
ModelTTF
SelenateDSe(VI)
SeleniteDSe(IV)
Organic
Selenide
DSe(-II)
Selenate +Selenite
PSe(IV+VI)
Elemental Se
PSe(0)
Organic Selenide
PSe(-II)
Bed Sediments
Dissolved species
Particulate species
Point Sources (Refineries, POTWs, Other Dischargers)
Contribute primarily to
suspended particulates, and
to dissolved phase in a limited way
Contribute to dissolved
phase
Estuary Water Column
River and
Tributary Loads
SelenateDSe(VI)
SeleniteDSe(IV)
Organic
Selenide
DSe(-II)
Selenate +Selenite
PSe(IV+VI)
Elemental Se
PSe(0)
Organic Selenide
PSe(-II)
Bed Sediments
Dissolved species
Particulate species
Point Sources (Refineries, POTWs, Other Dischargers)
Contribute primarily to
suspended particulates, and
to dissolved phase in a limited way
Contribute to dissolved
phase
Estuary Water Column
River and
Tributary Loads
ECoS = Fate and transport modeling framework for selenium speciesDYMBAM = Dynamic Bioaccumulation Model for estimating bivalve concentrationsTTF = Trophic Transfer Factor, ratio between food and predator tissue concentration
Red dots:
approximate
locations of model
segments
Yellow pins:
sampling stations in
Cutter and Cutter
(2004) survey
Model domain starts
at Sacramento River
at Rio Vista and
extends to Golden
Gate
Study Domain
Model Components and Steps
in Calibration1. Salinity: relatively conservative (advection and dispersion)
2. Total Suspended Material: three components of PSP, BEPS and phytoplankton, result of advection, dispersion
3. Phytoplankton (Chl a): result of advection, dispersion, growth, respiration, and grazing
4. Dissolved selenium: selenite (SeIV), organic selenide (SeII), selenate (SeVI)
5. Particulate selenium: particulate elemental , particulate organic selenide , particulate adsorbed selenite + selenate
Transformation modeled as first order reactions; transformations include: uptake by phytoplankton, adsorption/desorption, oxidation, mineralization
Modeling Steps
� During model calibration, adjustable parameters were varied to obtain a best fit to the data; for evaluation, the model was run with the fitted parameters and compared with new data sets
� Model calibrated to data from 1999, and tested against datasets from 2001, 2005, 1998 and 1986
� Model applied in a predictive mode using historical hydrology and different load scenarios
� Tetra Tech worked with model developers (Shannon Meseck, John Harris) over the course of this work
Model Schematic
Point Sources, Tributaries, and South Bay Input
Sacramento River
at Rio Vista
San Joaquin River near
Delta
Seawater Exchange
North San Francisco BayGolden Gate
1-D model with 33 well-mixed cells representing the bay.
Selenium Transformations
Selenate
Se(VI)
Organic
Selenide
Se(-II)
Selenite
Se(IV)
Dissolved SpeciesSelenate+ Selenite
Se(VI)+ Se(IV)
Organic
SelenideSe(-II)
Elemental Se
Se(0)
Selenate+ Selenite
Se(VI)+ Se(IV)
Organic
SelenideSe(-II)
Elemental Se
Se(0)
Organic Selenide
Se(-II)
PSP
Phyto-
plankton
BEPS
Mineralization, k1
Uptake
, k6
Mine
raliz
ation
, k 1
Minera
lization
, k1Ads/Des, a’, b
Ads/D
es, a
’, b
Uptake, k4
Uptake, k5
Oxidation, k2
Oxidation, k3
Advective/
Dispersive
Exchange
with Upper Cell
Advective/
Dispersive
Exchange with Lower
Cell
Bed Exchange
Represented by first-order rate constants.
Uptake by Bivalves
Time
Time
Time
Time
Se(0), particulate
Se(IV) + Se(VI),particulate
Se(-II),particulate
AE = 0.2AE = 0.45
AE = 0.54 to 0.8
C. amurensisconcentration
CmsskeCfIRAECwkudt
dCmss ×−××+×=
Cmss is selenium concentration in
tissue (µg/g), ku is the dissolved metal
uptake rate constant (L/g/d), Cw is the
dissolved metal concentration (µg/L),
AE is the assimilation efficiency (%), IR
is the ingestion rate (g/g/d), Cf is the
metal concentration in food (e.g.
phytoplankton, suspended particulate
matter, sediment) (µg/g), and ke is the
efflux rate (d-1).
Boundary Conditions are
Important
0 100Distance
C
Sources in the Bay
Seawater
boundary
Riverine
boundary
C, on the y-axis, represents a constituent being modeled. The model framework shown on the preceding slides involves the solution of a set of differential equations. These explain the shape of the curve. However, the boundary conditions also have an important effect on determining the actual magnitudes of C.
Year
1985 1990 1995 2000 2005
Se
leniu
m lo
ad
s (
kg/y
r)
0
2000
4000
6000
8000
10000
Riverine (kg/yr)
Refineries (kg/yr)
Tributaries (kg/yr)
Annual Selenium Loads
Dissolved Loads for Water Year
1999SJR @ Vernalis
2666
7%
Delta 365
South Bay
1607 176
SJR @ confluence POTWs
32% 3%
Sac. River @ Rio Vista Bay Exchange with Ocean Water
1502 5034
30%
Tributaries Refineries
820 559
16% 11%
Particulate Loads for Water
Year 1999SJR @ Vernalis
652
Delta
78 0
SJR @ confluence POTWs
11%
Sac. River @ Rio Vista Bay Exchange with Ocean Water
465 ~ 754 804 (32 BEPS)89%
Tributaries Refineries Bed Exchange
0 0 0.1
−−=
∑∑
∑∑
Xcal
Xobs
Xobs
XcalGOF 1*100(%)
Example Calibration 1: Salinity (1999)
Salinity
Observed (psu)
0 5 10 15 20 25 30 35P
redic
ted
(p
su)
0
5
10
15
20
25
30
35
y = 0.9272 x + 1.0404
Jan 21, 1999
Sa
lin
ity (
psu
)
0
5
10
15
20
25
30
35
April 14, 1999
Sali
nit
y (
ps
u)
0
5
10
15
20
25
30
35
May 7, 1999
Sa
lin
ity (
ps
u)
0
5
10
15
20
25
30
35
June 7, 1999
Distance (km)
0 20 40 60 80 100
Sa
lin
ity (
ps
u)
0
5
10
15
20
25
30
35
August 17, 1999
Sep 14, 1999
Oct 19, 1999
Nov 10, 1999
Distance (km)
0 20 40 60 80 100
r = 1.00GOF = 97.0%
r = 0.98GOF = 99.6%
r = 1.00GOF = 89.2%
r = 0.97GOF = 85.2%
r = 0.99GOF = 98.5%
r = 1.00GOF = 94.9%
r = 1.00GOF = 94.8%
r = 0.99GOF = 97.5%
Example Calibration 2: Chlorophyll a (1999)
Jan 21, 1999
Ch
l a (
µµ µµg
/L)
0
2
4
6
8
10
12
April 14, 1999
Ch
l a (
µµ µµg
/L)
0
2
4
6
8
10
12
May 7, 1999
Ch
l a (
µµ µµg
/L)
0
2
4
6
8
10
12
June 7, 1999
Salinity
0 5 10 15 20 25 30 35
Ch
l a (
µµ µµg
/L)
0
2
4
6
8
10
12
August 17, 1999
0
2
4
6
8
10
12
Sep 14, 1999
0
2
4
6
8
10
12
Oct 19, 1999
0
2
4
6
8
10
12
Nov 10, 1999
Salinity
0 5 10 15 20 25 30 35
0
2
4
6
8
10
12
r = 0.31GOF = 68.5%
r = 0.09GOF = 95.2%
r = 0.40GOF = 92.4%
r = 0.47GOF = 86.0%
r = 0.45GOF =66.8%
r = -0.04GOF = 48.2%
r = 0.07GOF = 31.8%
r = 0.83GOF = 82.2%
Chlorophyll a
Observed (µg/L)
0 2 4 6 8 10 12 14 16
Pre
dic
ted
(µ
g/L
)
0
2
4
6
8
10
12
14
16
R2 = 0.358
Salinity
0 5 10 15 20 25 30 35
Sele
nite
(µg
/L)
0.00
0.02
0.04
0.06
0.08
0.10
Salinity
0 5 10 15 20 25 30 35
Sele
nate
(µ
g/L
)
0.00
0.02
0.04
0.06
0.08
0.10
Salinity
0 5 10 15 20 25 30 35
Org
. S
ele
nid
e (
µg/L
)
0.00
0.02
0.04
0.06
0.08
0.10
r = 0.164GOF= 95.0%
r = 0.192GOF = 78.3%
r = 0.353GOF = 95.8%
Example Calibration 3: Dissolved Selenium (1999)
Salinity
0 5 10 15 20 25 30 35
Part
. S
eIV
+S
eV
I ( µ
g/L
)
0.000
0.005
0.010
0.015
0.020
0.025
Salinity
0 5 10 15 20 25 30 35
Part
. S
e0
(µ
g/L
)
0.000
0.005
0.010
0.015
0.020
0.025
Salinity
0 5 10 15 20 25 30 35
Part
. S
e-I
I ( µ
g/L
)
0.000
0.005
0.010
0.015
0.020
0.025
r = 0.801GOF = 92.1%
r = 0.676GOF = 83.5%
r = -0.021GOF = 73.5%
Example Calibration 4: Particulate Selenium (1999)
TSM Long-Term Evaluation at USGS
Stations
STN 3
1998 2000 2002 2004 2006 2008 2010
TS
M (
mg
/l)
0
50
100
150
200
250
Observed
Simulated
STN 6
Year
1998 2000 2002 2004 2006 2008 2010
TS
M (
mg
/l)
0
50
100
150
200
250
Observed
Simulated
STN 14
1998 2000 2002 2004 2006 2008 2010
TS
M (
mg
/l)
0
50
100
150
200
250
Observed
Simulated
STN 18
Year
1998 2000 2002 2004 2006 2008 2010
TS
M (
mg
/l)
0
50
100
150
200
250
Observed
Simulated
Evaluation of Chlorophyll a STN 3
1998 2000 2002 2004 2006 2008 2010
Ch
l a (
µµ µµg
/l)
0
2
4
6
8
10
12
14
16
18
20
Observed
Simulated
STN 6
Year
1998 2000 2002 2004 2006 2008 2010
Ch
l a (
µµ µµg
/l)
0
2
4
6
8
10
12
14
16
18
20
Observed
Simulated
STN 14
1998 2000 2002 2004 2006 2008 2010
Ch
l a (
µµ µµg
/l)
0
10
20
30
40
50
Observed
Simulated
STN 18
Year
1998 2000 2002 2004 2006 2008 2010
Ch
l a (
µµ µµg
/l)
0
2
4
6
8
10
12
14
16
18
20
Observed
Simulated
Suisun Bay
Suisun Bay
San Pablo Bay
Central Bay
Predicted Particulate Selenium Concentrations (1999)
November 11, 1999
0 5 10 15 20 25 30 35
Pa
rt. S
eIV
+ S
eV
I ( µ
g/g
)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Observed
Predicted
0 5 10 15 20 25 30 35
Part
. S
e0
(µ
g/g
)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 5 10 15 20 25 30 35
Part
. S
eII (
µg
/g)
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Salinity
0 5 10 15 20 25 30 35
To
tal P
art
. S
e (
µg
/g)
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Bivalve (C. amurensis) Concentrations
Year
1998 1999 2000 2001 2002 2003 2004 2005 2006
Cm
ss (
µg
/g)
0
5
10
15
20
25Observed
IR = 0.45, AE = 0.2,0.45, 0.8
IR = 0.65, AE = 0.2, 0.45, 0.8
IR = 0.65, AE = 0.2, 0.45, 0.54
IR = 0.85, AE = 0.2, 0.45, 0.80
White Sturgeon Concentrations
Year
80 85 90 95 00 05 10
Muscle
se
len
ium
concen
tration (
µg/g
)
0
10
20
30
40
50
Suisun Bay
San Pablo Bay
Estuary Mean
TTF = 1.7
Effect of Changing Boundary
Conditions
Salinity
0 5 10 15 20 25 30 35
Pa
rtic
ula
te
Se
len
ium
(µµ µµ
g/g
)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Observed
Lower Boundary
Higher Boundary
Scenarios
Examined
Scenario Description1 Base case
2 Removal of all point source loads (refineries, POTWs), and local tributary loads
3 30% reduction in refinery and San Joaquin River loads, dissolved only
4 50% reduction in all point sources (refineries, POTWs), local tributaries and San Joaquin River loads, dissolved only
5 Increase dissolved selenium loads from San Joaquin River by a factor of 3, particulate loads remain the same as the base case
6 Decrease dissolved selenium loads from San Joaquin River by a factor of 50%, particulate loads remain the same as the base case
7 Increase particulate selenium loads associated with PSP, BEPS, and phytoplankton from Sacramento River by a factor of 3, dissolved loads remain the same as the base case
8 Decrease particulate selenium loads associated with PSP, BEPS, and phytoplankton from Sacramento River by a factor of 50%, dissolved loads remain the same as the base case
9 Increase San Joaquin River particulate loads by 3x, other loads stay the same
10 A natural load scenario, where the point sources are zero, the local tributary loads and speciation are at Sacramento River values, and the San Joaquin River is at 0.2 µg/l, at current speciation
Impact on
Dissolved Se
High Flow Month (April, 1999)
0 1 2 3 4 5 6 7 8 9 10
Dis
so
lve
d S
e (
µg
/l)
0.0
0.1
0.2
0.3
Low Flow Month (November, 1999)
0 1 2 3 4 5 6 7 8 9 10
Dis
so
lve
d S
e (
µg
/l)
0.0
0.1
0.2
Dry Year Dry Month (July, 2001)
0 1 2 3 4 5 6 7 8 9 10
Dis
so
lve
d.S
e (
µg
/l)
0.0
0.1
0.2
Impact on
Particulate Se
High Flow Month (April, 1999)
0 1 2 3 4 5 6 7 8 9 10
Part
. S
e (
µg/g
)
0.0
0.5
1.0
1.5
2.0
Low Flow Month (November, 1999)
0 1 2 3 4 5 6 7 8 9 10
Pa
rt.
Se (
µg/g
)
0.0
0.5
1.0
1.5
2.0
Dry Year Dry Month (July, 2001)
0 1 2 3 4 5 6 7 8 9 10
Part
. S
e (
µg/g
)
0.0
0.5
1.0
1.5
2.0
Summary of Model Results� The model is able to simulate key aspects of physical and biological
constituents that affect selenium concentrations.
� During calibration, the model was able to fit the patterns in concentrations of dissolved and particulate selenate and selenite well, although it performed less well for the organic fractions. The model was also able to represent the observed variation in biota concentrations.
� The model is a valuable tool to explore selenium transport, fate, and bioaccumulation in the bay, and can be applied in analyses in support of the TMDL, as demonstrated through a set of example scenarios.
� A modeling study provides an opportunity to synthesize information from the system, and in doing so, highlights unknowns that may have a bearing on model predictions.
� This report presents a set of data needs for further evaluation such as characterization of boundary conditions, selenium loads from major sources, recent water column concentrations and speciation, as well as biota concentrations.
Impact on
Bivalve Se
High Flow Month (April, 1999)
0 1 2 3 4 5 6 7 8 9 10
Cm
ss S
e (
µg
/g)
0
5
10
15
20
25
30
35
Low Flow Month (November, 1999)
0 1 2 3 4 5 6 7 8 9 10
Cm
ss S
e (
µg
/g)
0
5
10
15
20
25
30
35
Dry Year Dry Month (July, 2001)
0 1 2 3 4 5 6 7 8 9 10
Cm
ss S
e (
µg
/g)
0
5
10
15
20
25
30
35