Embed Size (px)
Citation preview
Modeling Selenium in North San Francisco BayLimin Chen1, Shannon Meseck2, Sujoy B. Roy1, Thomas M. Grieb1, and Barbara Baginska3
1 Tetra Tech, Inc., 3746 Mt. Diablo Blvd., Suite 300, Lafayette, CA2National Marine Fisheries Service, 212 Rogers Avenue, Milford, Connecticut 06460
3San Francisco Bay Regional Water Quality Control Board, 1515 Clay Street, Oakland, CA 94612
AbstractThis poster describes the application of a numerical model of selenium fate and transport in North San Francisco Bay (NSFB), in support of the development of a selenium TMDL in this water body (see Baginska et al., this conference). The model builds on a pre-viously published application, and considers known point and non-point sources of selenium entering the bay, transport and mixing in the bay, and transforma-tion and biological uptake. The model considers the behavior and uptake of different dissolved and par-ticulate species. Dissolved species considered include selenate, selenite, and organic selenide. Particulate species considered include inorganic selenium (sel-enate plus selenite), organic selenide, and elemental selenium. Data on all these species is available for a set of sampling dates in the mid-1980s and late-1990s. The flows and selenium loads from the Sac-ramento and San Joaquin Rivers are dominant in the bay, although in the dry season, some of the point sources can become more important. Dissolved se-lenium concentrations (all species combined) in the NSFB are generally low (~0.2 ug/l). However, seleni-um present in particulate forms in the water column of the estuary bioaccumulates in filter feeders, such as bivalves, and then into predator organisms that feed on these bivalves. Selenium-associated impairment in NSFB is largely a consequence of elevated con-centrations in these predator organisms, specifically the white sturgeon and diving ducks. The modeling framework allows an examination of the relationship between selenium loads, in-bay concentrations, and biota concentrations to support TMDL development.
1. IntroductionThe San Francisco Bay Regional Board is developing a selenium Total Maximum Daily Load (TMDL) for North San Francisco Bay (NSFB), due to high concen-trations in some organisms. In support of the NSFB selenium TMDL, an estuary model (developed using the ECoS 3 framework) was used to simulate the se-lenium concentrations in water column and bioaccu-mulation of selenium in the NSFB. The model built upon the previous work of Meseck and Cutter (2006).
The location of NSFB and the starting point of mod-eling domain (the “head” of the estuary) are shown in Figure 1. The end of the modeling domain is at Golden Gate Bridge.
Modeling Domain
Figure 1. Boundaries of NSFB Selenium TMDL
2. MethodsThe model was applied in one-dimensional form to simulate several constituents including salinity, total suspended material (TSM), phytoplankton, dissolved and particulate selenium and selenium concentrations in bivalves and higher trophic organisms (Figure 2).
The biogeochemistry of selenium, including transfor-mations among different species of dissolved and par-ticulate selenium and bioaccumulation of selenium through foodweb, was simulated by the model. Species simulated by the model include selenite, selenate, and organic selenide. The particulate species simulated by the model include particulate organic selenium, par-ticulate elemental selenium, and particulate adsorbed selenite and selenate (Figure 3; Table 1). The uptake of dissolved selenium by phytoplankton includes uptake of three species (selenite, organic selenide and sele-nate). Bioaccumulation of particulate selenium to the bivalves was simulated using a dynamic bioaccumu-lation model (DYMBAM, Presser and Luoma, 2006), applied in a steady state mode. Bioaccumulation into bivalves considers the different efficiency of absorp-tion for different selenium species (Figure 4). Bioac-cumulation to higher trophic levels of fish and diving ducks was simulated using previously derived linear regression equations by Presser and Luoma (2006), and using estimates of trophic transfer factors (TTF) summarized from the literature (Presser and Luoma, personal communication, 2009).
The model was calibrated using data from 1999 and validated against data for the rest of the simulation period (2000-2008)
Model Components
Flow, Tides Sunlight, grazing
External Drivers
Salinity Phytoplankton
Riverine Loads
TSMGeneral
Processes in NSFB
Sediment erosion
Selenium Loads
Dissolved Se Se (IV),
Se(VI), (-II)
Particulate Se Se(0), Se(II), Se(IV)+Se(VI)
Zooplankton Bivalves
White Sturgeon
Diving Ducks
Selenium Specific Processes in NSFB
Figure 2. Model Components of ECoS3 Application in NSFB
Selenium Transformations SimulatedTransformations among dissolved selenium species are usually slow processes.
Particulate selenium can be associated with Perma-nently Suspended Particulates (PSP), phytoplankton, and Bed Exchangeable Particles (BEPS).
The uptake of dissolved selenium by phytoplankton depends species (greater in selenite and organic sel-enide than selenate).
SelenateSe(VI)
Organic SelenideSe(-II)
SeleniteSe(IV)
Dissolved SpeciesSelenate+ Selenite
Se(VI)+ Se(IV)
Organic Selenide
Se(-II)
Elemental SeSe(0)
Selenate+ SeleniteSe(VI)+ Se(IV)
Organic Selenide
Se(-II)
Elemental SeSe(0)
Organic SelenideSe(-II)
PSP
Phyto-plankton
BEPS
Mineralization, k1
Uptake, k6
Mineralization, k 1
Mineralization, k1Ads/Des, a’, b
Ads/Des, a
’, b
Uptake, k4
Uptake, k5
Oxidation, k2
Oxidation, k3
Advective/Dispersive Exchange with Upper
Cell
Advective/Dispersive Exchange with Lower
Cell
Bed Exchange
Figure 3. Transformations of dissolved and partic-ulate selenium
Table 1. Literature values for first order rate constants
Process Description Value Unit Reference
k1 P Se(-II) → D Se(-II)
Mineralization of particulate organic selenide
1.3× 10-5-5×10-2 d-1 Regeneration experiments (Cutter, 1992)
k2 D Se(-II) → D Se (IV)
Oxidation of dissolved organic selenide
1.0×10-3 - 81.0 d-1
Surface and deep Pacific water (Suzuki et al. 1979, cited in Cutter, 1992)
k3 D Se(IV) → D Se (VI)
Oxidation of dissolved selenite
2.4×10-6 d-1 Deep Pacific, Cutter and Bruland (1984)
k4 D Se(IV) → P Se (-II)
Uptake of dissolved selenite by phytoplankton
2.02-2.41 (15.8-18.8)
0.07-0.21*
(225.8-777.8)*
μmol Se (g chl)-1hr-
1 (l/g chl a/hr)
pmol Se (ug chl)-
1hr-1 (l/g chl a/hr)
Riedel et al. (1996) Baines et al. (2004)
k5 D Se(VI) → P Se (-II)
Uptake of dissolved selenate by phytoplankton
0.43-0.58 μmol (g chl)-1hr-1 Riedel et al. (1996)
k6 D Se(-II) → P Se (-II)
Uptake of dissolved organic selenide by phytoplankton
0.5 k4 μmol (g chl)-1hr-1 Baines et al. (2001)
a’ D Se(IV) → PSe (IV)
Mineral adsorption of selenite
0.1-0.8 l/g/d Zhang and Sparks (1990)
PSe(IV) → D Se(IV)
Desorption of adsorbed selenite
Kd/a’ d-1
Constant
b
Selenium Bioaccumulation through the 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
Figure 4. Bioaccumulation of Particulate Selenium in Bivalves
Model Boundary Conditions and External LoadsThe riverine inputs of flow from Sacramento River at Rio Vista are DAYFLOW records from the Interagen-cy Ecological Program (IEP; http://www.iep.ca.gov/dayflow/index.html). The San Joaquin River is mod-eled as a tributary, with flow derived as the difference between Net Delta Outflow Index (NDOI) and flow from the Sacramento River at Rio Vista (Figure 5).
Riverine inputs of TSM and chlorophyll a were mod-eled as flow multiplied by concentrations (Figure 6). TSM concentrations were modeled as a function of flow. Riverine chlorophyll a concentrations were ob-served data from USGS and Bay Delta and Tributary Project (BDAT).
Dissolved selenium inputs for selenate, selenite, and organic selenide were specified from the rivers as flow and selenium concentrations by species from two riv-ers (Figure 5). Different species of selenium concen-trations were derived using fitted functions based on observed data by Cutter and Cutter (2004). Delta re-moval constants were used in converting observed selenium concentrations at San Joaquin River at Ver-nalis to concentrations at the confluence. Particulate concentrations were based on data published by Dob-lin et al. (2006).
Selenium loads to the NSFB include point sources from refineries, municipal and industrial dischargers and tributaries. Point and non-point sources of sele-nium were added to the model domain at their dis-charge locations. Over the past two decades, there have been major declines in refinery loads due to improved wastewater treatment installed in 1998 and there is some evidence that San Joaquin River concentrations were lower in the late 1990s and beyond than in the 1980s.
Riverine inputs of selenium are the largest source to the NSFB (Figure 7).
Model Boundary ConditionsFlow
Year98 99 00 01 02 03 04 05 06 07
Sac
ram
ento
Riv
er @
Rio
Vis
ta F
low
(m3/s
)
0
2000
4000
6000
8000
10000
12000
Year
98 99 00 01 02 03 04 05 06 07
San
Joaq
uin
Riv
er F
low
to N
SFB
(m3/s
)
0
200
400
600
800
1000
1200
1400
1600
1800
2000
TSM
Year
98 99 00 01 02 03 04 05 06
TSM
load
s (k
g/d)
0
1e+7
2e+7
3e+7
4e+7
5e+7Sacramento River at Rio Vista
Year
98 99 00 01 02 03 04 05 06 07
TSM
load
s (k
g/d)
0.0
2.0e+6
4.0e+6
6.0e+6
8.0e+6
1.0e+7
1.2e+7
San Joaquin River at Confluence
Chlorophyll a
SJR
1998 1999 2000 2001 2002 2003 2004 2005 2006
Tota
l Chl
a (k
g/d)
0
50
100
150
200
250
300
350
Sac
1998 1999 2000 2001 2002 2003 2004 2005 2006
Tota
l Chl
a (k
g/d)
0
500
1000
1500
2000
2500
Figure 6. Boundary conditions of flow, TSM and chlorophyll a
Selenium Loads
Figure 7. Annual selenium loads from riverine (Sacramento River + San Joaquin), refineries and local tributaries for prior to refinery clean-up (1986 and 1998) and post refinery clean-up (1999-2005) used in the model. Refinery loads for 1986 and 1998 are from Meseck (2002).
3. ResultsEvaluation of salinity, TSM, and chlorophyll a sim-ulation for the low flow year 2001 suggested good agreement of simulated salinity versus observed val-ues for different months across the year (Figure 8, 9, and 10). Overall values for goodness of fit (GOF) for these months are between 71.5-97.9% for salinity, 36.4 – 99.4% for TSM, and 53.7 – 95.7% for chloro-phyll a. The location of the estuarine turbidity maxi-mum (ETM) was simulated well for most months in 2001, particularly for June and July 2001. For about two months, chlorophyll a concentrations were under predicted near the Central Bay, similar to the pattern in the calibration. For the evaluation period, simulat-ed correlation coefficient (r) is 0.92-1.00 for salinity in 2001, 0.68 – 0.97 for TSM in 2001, and 0.02-0.79 for chlorophyll a in 2001 (Figure 11).
−−=∑∑
∑∑
XcalXobs
XobsXcal
GOF 1*100(%)
Simulated TSM and chlorophyll a concentrations were also evaluated against long-term data from the USGS monitoring stations. The model-simulated chloro-phyll a and TSM concentrations were evaluated against long-term data at four stations, station 3 (Suisun Bay), 6 (Suisun Bay), 14 (San Pablo Bay) and 18 (Central Bay), respectively. The model was able to capture the seasonal patterns in chlorophyll a concentrations and TSM (Figure 12 and Figure 13) relatively well. The model was able to capture the peaks and lows in both TSM and chlorophyll a concentrations.
Salinity Validation (2001)6 Feb 2001
0 20 40 60 80 100 120
Salin
ity
0
5
10
15
20
25
30
35
26 Feb 2001
0 20 40 60 80 100 120
Salin
ity
0
5
10
15
20
25
30
35
27 Mar 2001
Distance (km)
0 20 40 60 80 100 120
Salin
ity
0
5
10
15
20
25
30
35
24 Apr 2001
0 20 40 60 80 100 120
Salin
ity
0
5
10
15
20
25
30
35
19 Jun 2001
0 20 40 60 80 100 120
Salin
ity
0
5
10
15
20
25
30
35
17 Jul 2001
Distance (km)
0 20 40 60 80 100 120
Salin
ity
0
5
10
15
20
25
30
35
11 Sep 2001
Salin
ity
0
5
10
15
20
25
30
35
16 Oct 2001
0 20 40 60 80 100 120
Salin
ity
05
10
15
20
25
30
35
27 Nov 2001
Distance (km)
0 20 40 60 80 100 120
Salin
ity
0
5
10
15
20
25
30
35
r = 0.97GOF = 74.1%
r = 0.99GOF = 82.9%
r = 0.99GOF = 82.2%
r = 0.98GOF = 90.7%
r = 0.99GOF = 97.9%
r = 0.99GOF = 91.3%
r = 1.00GOF = 71.5 %
r = 0.99GOF = 95.6%
r = 0.97GOF = 77.2%
0 20 40 60 80 100 120
Figure 8. Evaluation of simulated monthly salinity profiles for a low flow year 2001 (Data source: USGS)
TSM Validation (2001)6 Feb 2001
0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
26 Feb 2001
0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
27 Mar 2001
Distance (km)0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
24 Apr 2001
0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
19 Jun 2001
0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
17 Jul 2001
Distance (km)0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
11 Sep 2001
0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
16 Oct 2001
0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
27 Nov 2001
Distance (km)0 20 40 60 80 100 120
TSM
(mg/
L)
0
20
40
60
80
100
r = 0.75GOF = 65.0%
r = 0.88GOF = 69.0%
r = 0.80GOF = 36.4%
r = 0.97GOF = 99.4%
r = 0.84GOF = 38.1%
r = 0.92GOF = 62.6%
r = 0.79GOF = 92.0%
r = 0.68GOF = 43.7%
Figure 9. Evaluation of simulated monthly TSM pro-files for a low flow year 2001 (Data source: USGS)
Chlorophyll a Evaluation (2001)6 Feb 2001
0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
26 Feb 2001
0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
27 Mar 2001
Distance (km)0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
24 Apr 2001
0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
19 Jun 2001
0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
17 Jul 2001
Distance (km)0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
11 Sep 2001
0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
16 Oct 2001
0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
27 Nov 2001
Distance (km)0 20 40 60 80 100 120
Chl
orop
hyll
a (µ
g/L)
02468
1012141618
r = 0.29GOF = 77.0%
r = 0.22GOF = 91.8%
r = 0.78GOF = 95.0%
r = 0.33GOF = 92.3%
r = 0.02GOF = 95.7%
r = 0.79GOF = 78.5%
r = -0.50GOF = 61.4%
r = 0.33GOF = 70.1%
r = 0.74GOF = 53.7%
Figure 10. Evaluation of simulated monthly chlo-rophyll a profiles for a low flow year 2001 (Data source: USGS)
TSM Long-term Evaluation STN 3
1998 2000 2002 2004 2006 2008 2010
TSM
(mg/
l)
0
50
100
150
200
250ObservedSimulated
STN 6
Year
1998 2000 2002 2004 2006 2008 2010
TSM
(mg/
l)
0
50
100
150
200
250ObservedSimulated
STN 14
1998 2000 2002 2004 2006 2008 2010
TSM
(mg/
l)
0
50
100
150
200
250ObservedSimulated
STN 18
Year
1998 2000 2002 2004 2006 2008 2010
TSM
(mg/
l)
0
50
100
150
200
250ObservedSimulated
Figure 11. Simulated time series of TSM concen-trations compared to observed data from USGS at stations 3 (Suisun Bay), 6 (Suisun Bay), 14 (San Pablo Bay) and 18 (Central Bay).
Chlorophyll a Long-term Validation
STN 3
1998 2000 2002 2004 2006 2008 2010
Chl
a (µ
g/l)
0
2
4
6
8
10
12
14
16
18
20ObservedSimulated
STN 6
Year1998 2000 2002 2004 2006 2008 2010
Chl
a (µ
g/l)
0
2
4
6
8
10
12
14
16
18
20ObservedSimulated
STN 14
1998 2000 2002 2004 2006 2008 2010
Chl
a (µ
g/l)
0
10
20
30
40
50ObservedSimulated
STN 18
Year1998 2000 2002 2004 2006 2008 2010
Chl
a (µ
g/l)
0
2
4
6
8
10
12
14
16
18
20ObservedSimulated
Figure 12. Simulated time series of phytoplankton concentrations compared to observed data from USGS at stations 3 (Suisun Bay), 6 (Suisun Bay), 14 (San Pablo Bay) and 18 (Central Bay).
Model HindcastThe model hindcast for 1998 (prior to refinery seleni-um load controls) suggested that for June (high flow) and October 1998 (low flow), the model-simulated sa-linity, TSM and chlorophyll a compared well to the ob-served values (Figure 13). The model hindcast was able to simulate the ETM during October 1998. Hindcast of dissolved selenium for 1998 suggested the model was able to simulate the relatively conservative mixing behavior of selenium during high flow and the mid-estuarine peaks during low flow (Figure 14). Simulat-ed selenium concentrations on particulates compared well with the observed values (Figure 15).
June 1998
Distance (km)0 20 40 60 80 100
Salin
ity
0
5
10
15
20
25
30
35October 1998
Distance (km)0 20 40 60 80 100 120
Salin
ity
05
10
15
20
25
30
35
June 1998
Salinity0 5 10 15 20 25 30 35
TSM
(mg/
l)
0
10
20
30
40
50
60
70
80October 1998
Salinity0 5 10 15 20 25 30 35
TSM
(mg/
l)
0
10
20
30
40
50
60
June 1998
Salinity0 5 10 15 20 25 30 35
Chl
orop
hyll
a (µ
g/l)
0
2
4
6
8
10October 1998
0 5 10 15 20 25 30 35
Chl
orop
hyll
a (µ
g/l)
0
2
4
6
8
10
Salinity
Figure 13. Model simulated profiles of salinity, TSM and chlorophyll a compared to observed val-ues for June and October 1998.
Salinity0 5 10 15 20 25 30 35
Sele
nite
(µg/
L)
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
Salinity0 5 10 15 20 25 30 35
Sele
nate
(µg/
L)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
Salinity0 5 10 15 20 25 30 35
Org
anic
Sel
enid
e (µ
g/L)
0.000.020.040.060.080.100.120.140.160.18
Salinity0 5 10 15 20 25 30 35
Sele
nite
(µg/
L)
0.005
0.010
0.015
0.020
0.025
0.030
0.035
0.040
Salinity0 5 10 15 20 25 30 35
Sele
nate
(µg/
L)
0.00
0.02
0.04
0.06
0.08
0.10
0.12
0.14
Salinity0 5 10 15 20 25 30 35
Org
anic
Sel
enid
e (µ
g/L)
0.000.020.040.060.080.100.120.140.160.18
June 1998
June 1998
June 1998
October 1998
October 1998
October 1998
r = 0.035GOF = 65.0%
r = 0.722GOF = 55.4%
r = 0.505GOF = 4.6%
r = 0.476GOF = 96.7%
r = 0.282GOF = 91.6%
r = 0.571GOF = 47.1%
Figure 14. Model simulated dissolved selenium by species as a function of salinity compared to ob-served values for June 1998 and October 1998.
October 1998
Salinity0 5 10 15 20 25 30 35
Part
. Se
(µg/
g)
0.0
0.5
1.0
1.5
2.0
2.5ObservedSimulated
June 1998
Salinity0 5 10 15 20 25 30 35
Part
. Se
(µg/
g)
0.0
0.5
1.0
1.5
2.0
2.5ObservedSimulated
Figure 15. Simulated particulate selenium for high flow (June 1998) and low flow (October 1998) in 1998.
Model simulated selenium concentrations on particulates and biotaPredicted selenium concentrations in Corbula amu-rensis near Carquinez Strait as a function of time were compared to data from Stewart et al. (2004) and are shown in Figure 16 for a range of ingestion rates. Dif-ferent ingestion rates of particulate selenium by Cor-bula amurensis and assimilation efficiencies for or-ganic selenium were used in the simulation. Predicted ranges in bivalve selenium concentrations are between 2 – 22 μg/g.
Simulated selenium concentrations in muscle tis-sues and liver of white sturgeon and greater scaup us-ing TTF and regression equations from Presser and Luoma (2006) were compared to observed values in the NSFB (Figure 17 and 18).
Simulated Selenium in Bivalves
Year
1998 1999 2000 2001 2002 2003 2004 2005 2006
Cm
ss (µ
g/g)
0
5
10
15
20
25ObservedIR = 0.45, AE = 0.2,0.45, 0.8IR = 0.65, AE = 0.2, 0.45, 0.8IR = 0.65, AE = 0.2, 0.45, 0.54IR = 0.85, AE = 0.2, 0.45, 0.80
Figure 16. Simulated selenium concentrations in bivalve Corbula amurensis near the Carquinez Strait compared to observed values from Stewart et al. (2004; station 8.1).
Simulated Selenium Concentrations in White Sturgeon
Year
80 85 90 95 00 05 10
Mus
cle
sele
nium
con
cent
ratio
n (µ
g/g)
0
10
20
30
40
50
Year
80 85 90 95 00 05 10
Live
r sel
eniu
m c
once
ntra
tion
(µg/
g)
0
10
20
30
40
50
60
70
80
Suisun BaySan Pablo BayEstuary Mean
Suisun BaySan Pablo BayEstuary Mean
Figure 17. Model predicted selenium concentra-tions in muscle tissue and liver of white sturgeon at Suisun Bay and San Pablo Bay compared to ob-served values (White et al., 1988, 1989, Urquhart et al., 1991, USGS and SFEI), using TTF = 1.7 and equation from Presser and Luoma (2006).
Simulated Selenium Concentrations in Greater Scaup
San Pablo Bay
1980 1990 2000 2010
Sel
eniu
m c
once
ntra
tions
in G
reat
er s
caup
(mus
cle,
µ g/
g)
0
5
10
15
20
25
30
35
40
Suisun Bay
1980 1990 2000 2010
Sel
eniu
m c
once
ntra
tions
in G
reat
er s
caup
(mus
cle,
µ g/
g)
0
5
10
15
20
25
30
35
40
Figure 18. Model predicted selenium concentra-tions muscle tissue of diving ducks (dry weight; Greater Scaup) compared to observed data in San Pablo Bay and Suisun Bay, respectively (White et al., 1988, 1989; Urquhart et al., 1991; SFEI), us-ing TTF = 1.8.
4. Evaluation of Loading ScenariosA total of 10 scenarios with different combinations of load reductions from different sources were evalu-ated (Table 2).The results suggested that changes in dissolved selenium concentrations due to load reduc-tion in dissolved selenium are more significant than particulate selenium concentrations (Figure 19). With reductions of in dissolved selenium to natural loads, particulate selenium and selenium in bivalves showed some responses (1.2 μg/g decreases in selenium in bi-valves).
A scenario of increasing San Joaquin River flow to flow observed at Vernalis was also evaluated. Increas-ing flow to the Vernalis flow resulted in significant in-creases in dissolved and particulate selenium concen-trations (Figure 20 and 21).
Table 2. Load Change Scenarios Tested Using the Model
Scenario Description Loading factors as a fraction of base case loads, unless specified as a concentration in µg/l1
Riverine particulate selenium loads
Dissolved selenium loads
BEPS PSP Phyto Sac. SJR. Ref. Trib. POTWs 1 Base Case 1 1 1 1 1 1 1 1
2 Removal of all point source loads (refineries, POTWs), and local tributary loads
1 1 1 1 1 0 0 0
3 30% reduction in refinery and San Joaquin River loads, dissolved only
1 1 1 1 0.7 0.7 1 1
4 50% reduction in all point sources (refineries, POTWs), local tributaries and San Joaquin River loads, dissolved only
1 1 1 1 0.5 0.5 0.5 0.5
5 Increase dissolved selenium loads from San Joaquin River by a factor of 3, particulate loads remain the same as the base case
1 1 1 1 3 1 1 1
6 Decrease dissolved selenium loads from San Joaquin River by a factor of 50%, particulate loads remain the same as the base case
1 1 1 1 0.5 1 1 1
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
3 3 3 1 1 1 1 1
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
0.5 0.5 0.5 1 1 1 1 1
9 Increase San Joaquin River particulate loads by 3x, other loads stay the same
1 1 1 1 1 1 1 1
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
1 1 1 1 0.2 µg/l
0 Sac. R. levels
0
1 Base case loads are not constant through time in the simulations. When a load change is imposed, this means that the entire time series of load inputs is multiplied by the same factor.
Scenario of Increasing San Joaquin River Flow
Salinity
0 5 10 15 20 25 30 35
Dis
solv
ed
Sel
eniu
m (µ
g/L)
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
ObservedNormal SJR flowVernalis flow
Salinity
0 5 10 15 20 25 30 35
Par
t. S
e (µ
g/L)
0.005
0.010
0.015
0.020
0.025
0.030
Figure 20. Predicted dissolved and particulate se-lenium for different San Joaquin River discharge during a low flow period (November 11, 1999).
Salinity
0 5 10 15 20 25 30 35
Par
t. S
e (µ
g/g)
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Normal SJR FlowVernalis FlowObserved
Figure 21. Predicted particulate selenium concen-tration (μg/g) under estimated San Joaquin River flow at the confluence compared to the prediction for flow at the confluence set to the Vernalis flow rate.
AcknowledgementsWe thank Greg Cutter of Old Dominion University for providing selenium speciation data in the estuary, USGS for providing data on selenium in white sturgeon, and SFEI for providing data on selenium in Greater Scaup.
The work described in this poster was supported by the Western States Petroleum Association and performed in collaboration with the Regional Board.
The work presented in the poster is also reported in a Technical Memorandum, currently under review of by spe-cially constituted Technical Review Committee (Nicholas Fisher, Regina Linville, Samuel Luoma, John Oram). We thank the members of this committee for providing comments on various aspects of model development over the past 18 months.
5. Uncertainties and Data Collection NeedsSelenium speciation data: Selenium speciation data exist for 1999. After 1999, all the selenium data in the Bay are as total and dissolved selenium. Speciation data along the salinity gradient under different flow conditions are needed. Selenium contents on partic-ulates from the rivers (above the influence of the es-tuary) are lacking.
Selenium loads: Selenium loads for different species from the Delta and tributaries need to be better char-acterized. Using selenium speciation data for the Sac-ramento River at Freeport and San Joaquin River at Vernalis gave good predictions in dissolved selenium concentrations in the Bay. However due to the com-plexity of the Delta system and the potential trans-
formations occurring in the Delta, selenium loads from the Delta remain uncertain. Loads from local tributaries are more significant during high flow than low flow. Uncertainties remain in selenium concen-trations and speciation in the tributaries.
Role of phytoplankton and bacteria in selenium uptake: It is recommended that uptake of dissolved selenium by dominant species of phytoplankton and bacteria in the NSFB to be studied under the ambient sele-nium concentrations of the NSFB.
Bioaccumulation into the higher trophic levels (fish and birds): Uncertainties are associated with feeding patterns of the predators due to the migratory na-ture of certain species (such as surf scoter). Data with good correspondence of time and space in bivalves and predators are needed.
6. ConclusionsThe model is able to simulate key aspects of physical and biological constituents that affect selenium con-centrations. The model simulates salinity, TSM, and phytoplankton well for different hydrological condi-tions.
The model hindcast using hydrological and selenium loads data for 1986 and 1998 suggests the model was able to simulate physical parameters and selenium in the estuary relatively well. The model was able to simulate the mid-estuarine peaks in selenite for low flow of 1986 and 1998. This indicates the location and magnitude of the selenium input from point sources and the transport and transformation of selenium are represented well in the model. Simulated particulate selenium concentrations also compared well with the observed values.
When dissolved loads, including point sources and local tributary contributions, are reduced, there are corresponding decreases in the dissolved concentra-tions, but less significant changes in particulate spe-cies concentrations.
The overall sensitivity of the estuary to load chang-es from local tributaries and point sources is greater during dry months, especially during a dry year, i.e., for a given load change factor, greater change is ob-served during the dry periods.
The modeling approach is able to capture the key fea-tures of selenium behavior in the system at a level that is consistent with the data. This model as currently set up can be used to explore management options in the context of the TMDL. Analysis of new speciation data with the model will be very useful.
ReferencesBaines, S.B., N.S. Fisher, M.A. Doblin, and G.A. Cutter. 2001. Uptake of dissolved organic selenides by marine phytoplankton. Limnology and Oceanography 46: 1936-1944.
Baines, S.B., N.S. Fisher, M.A. Doblin, G.A. Cutter, L.S. Cutter, and B.E. Cole. 2004. Light dependence of selenium uptake by phytoplankton and implications for predicting selenium incor-poration into food webs. Limnology and Oceanography 49(2): 566-578.
Cutter, G.A. 1992. Kinetic controls on metalloid speciation in seawater. Mar. Chem. 40, 65-80.
Cutter, G.A., and L.S. Cutter. 2004. Selenium biogeochemistry in the San Francisco Bay estuary: changes in water column be-havior. Estuarine, Coastal and Shelf Science 61: 463-476.
Doblin, M.A., S.B. Baines, L.S. Cutter, and G.A. Cutter. 2006. Sources and biogeochemical cycling of particulate selenium in the San Francisco Bay estuary. Estuarine, Coastal and Shelf Sci-ence 67: 681-694.
Meseck, S.L. 2002. Modeling the biogeochemical cycle of sele-nium in the San Francisco Bay. A dissertation submitted to the faculty of Old Dominion University in partial fulfillment of the Degree of Doctor of Philosophy.
Meseck, S.L. and G.A.Cutter, 2006. Evaluating the biogeochem-ical cycle of selenium in San Francisco Bay through modeling. Limnology and Oceanography 51(5): 2018-2032.
Presser, T.S. and S. N. Luoma. 2006. Forecasting Selenium Dis-charges to the San Francisco Bay-Delta Estuary: Ecological Ef-fects of a Proposed San Luis Drain Extension. USGS Profession-al paper 1646.
Riedel, G.F., Sanders, J.G., and C.C. Gilmour. 1996. Uptake, transformation, and impact of selenium in freshwater phyto-plankton and bacterioplankton communities. Aquat. Microb. Ecol. 11, 43-51.
Stewart, R.S., S.N. Luoma, C.E. Schlekat, M.A. Doblin, and K.A. Hieb, 2004, Food web pathway determines how selenium affects ecosystems: A San Francisco Bay case study: Environmental Sci-ence and Technology, v. 38, no. 17, p. 4519-4526.
Urquhart, K.A.F., K. Regalado et al. 1991. Selenium verification study, 1988-1990 (report 91-2-WQWR), A report from the Cali-fornia Department of Fish and Game to the California State Wa-ter Resources Control Board; California Sate Water Resources Control Board Sacramento, California, 94 p. and 7 appendices.
White, J.R., Hofmann, P.S., Hammond, D., and S. Baumgartner. 1988. Selenium verification study, 1986-1987, A report from the California Department of Fish and Game to the California Sate Water Resources Control Board: California Sate Water Resourc-es Control Board Sacramento, California, 60 p. and 8 appendi-ces.
White, J.R., Hofmann, P.S., Urquhart, K.A.F., Hammond D., and S. Baumgartner. 1989. Selenium verification study, 1987-88, A report from the California Department of Fish and Game to the California State Water Resources Control Board: California State Water Resources Control Board, Sacramento, California, 81 p. and 11 appendices.
Zhang, Y., Sparks, D.L., 1990. Kinetics of selenate and selenite adsorption/desorption at the Geothite/Water interface. Envi-ron. Sci. Technol. 24, 1848-1856.
High Flow Month (April, 1999)
0 1 2 3 4 5 6 7 8 9 10
Par
t. S
e (µ
g/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
Par
t. S
e (µ
g/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
Par
t. S
e (µ
g/g)
0.0
0.5
1.0
1.5
2.0
High Flow Month (April, 1999)
0 1 2 3 4 5 6 7 8 9 10
Cm
ss S
e (µ
g/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 (µ
g/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 (µ
g/g)
0
5
10
15
20
25
30
35
High Flow Month (April, 1999)
0 1 2 3 4 5 6 7 8
Dis
solv
ed S
e (µ
g/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
solv
ed S
e (µ
g/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
solv
ed.S
e (µ
g/l)
0.0
0.1
0.2
9 10
Figure 19. Model predicted selenium concentrations muscle tissue of diving ducks (dry weight; Greater Scaup) compared to observed data in San Pablo Bay and Suisun Bay, respectively (White et al., 1988, 1989; Urquhart et al., 1991; SFEI), using TTF = 1.8.
Daily flow reported in DAYFLOW
Flow
Suspended sediment -Function of flow rate at Rio Vista, USGS dataBed sediment15% of suspended sediment, no direct observations Phytoplankton-Based on chlorophyll a concentrations, BDAT and USGS dataSelenium on particulates-For suspended and bed sediment, use range from 0.3 to 0.8 ug/g, range from Doblin et al. (2006)-For phytoplankton, use Se:C ratio of 15.9x10-6, Baines et al. (2004)
Selenate, Se(VI)Selenite, Se(IV)Organic selenide, Se(-II)Data from Cutter and Cutter (2004) for WY 1999
Particulate Conc.Dissolved Conc.
Daily flow reported in DAYFLOW
Flow
Suspended sediment -Function of flow rate at Rio Vista, USGS dataBed sediment15% of suspended sediment, no direct observations Phytoplankton-Based on chlorophyll a concentrations, BDAT and USGS dataSelenium on particulates-For suspended and bed sediment, use range from 0.3 to 0.8 ug/g, range from Doblin et al. (2006)-For phytoplankton, use Se:C ratio of 15.9x10-6, Baines et al. (2004)
Selenate, Se(VI)Selenite, Se(IV)Organic selenide, Se(-II)Data from Cutter and Cutter (2004) for WY 1999
Particulate Conc.Dissolved Conc.
Calculated from Net Delta Outflow minus flow in Rio Vista, both reported in DAYFLOW
Flow
Suspended sediment-Function of flow rate at four Delta locations: Twitchell Island, Jersey Point, Antioch Ship Channel, and Old River, USGS dataBed sediment15% of suspended sediment, no direct observations Phytoplankton-Based on chlorophyll a concentrations at Twitchell Island, USGS dataSelenium on particulates-For phytoplankton, use Se:C ratio of 15.9x10-6, Baines et al. (2004)-For suspended and bed sediments, use Kd from Delta stations sampled by Doblin et al. (2006)
For WY 1999, Cutter and Cutter (2004) data from Vernalis, with a removal fraction (~50-70%, dependent on species)Selenate, Se(VI)Selenite, Se(IV)Organic selenide, Se(-II)
Particulate Conc.Dissolved Conc.
Calculated from Net Delta Outflow minus flow in Rio Vista, both reported in DAYFLOW
Flow
Suspended sediment-Function of flow rate at four Delta locations: Twitchell Island, Jersey Point, Antioch Ship Channel, and Old River, USGS dataBed sediment15% of suspended sediment, no direct observations Phytoplankton-Based on chlorophyll a concentrations at Twitchell Island, USGS dataSelenium on particulates-For phytoplankton, use Se:C ratio of 15.9x10-6, Baines et al. (2004)-For suspended and bed sediments, use Kd from Delta stations sampled by Doblin et al. (2006)
For WY 1999, Cutter and Cutter (2004) data from Vernalis, with a removal fraction (~50-70%, dependent on species)Selenate, Se(VI)Selenite, Se(IV)Organic selenide, Se(-II)
Particulate Conc.Dissolved Conc.
Suspended sediment, USGS dataDissolved selenium (speciated) from Cutter and Cutter (2004)
Particulate Conc.Dissolved Conc.
Suspended sediment, USGS dataDissolved selenium (speciated) from Cutter and Cutter (2004)
Particulate Conc.Dissolved Conc.
Suspended sediment, USGS dataSelenium on particulates-Estimate overall Kd from data in Doblin et al. (2006)
Total selenium, SWAMP data, mid-1980’s to presentSpeciation of dissolved selenium, 1999-2000, Cutter and Cutter (2004)
Particulate ConcDissolved Conc.
Suspended sediment, USGS dataSelenium on particulates-Estimate overall Kd from data in Doblin et al. (2006)
Total selenium, SWAMP data, mid-1980’s to presentSpeciation of dissolved selenium, 1999-2000, Cutter and Cutter (2004)
Particulate ConcDissolved Conc.
Dissolved and Particulate Se Load to Bay
Aqueduct Export
Figure 5. Flow, dissolved and particulate selenium concentrations and loads at the riverine boundary
