calculating dissolved selenium loads from the delta to san ... · pdf fileselenium inputs to...
TRANSCRIPT
Calculating Dissolved Selenium Loads from the Delta to San Francisco Bay Using DSM2 May 1, 2014
Prepared by Tetra Tech
Introduction Selenium inputs to North San Francisco Bay from the Central Valley via the Delta are expected to be a significant source. Accurately estimating these loads is difficult due to the complex flows within Delta and tidal influences from the bay. Flow in the Delta is a mix of water from the Sacramento River, the San Joaquin River, eastern tributaries (Mokelume, Cosumnes, Calaveras Rivers, as well as other minor sources), agricultural return flows and tidal flows from the Bay (identified operationally as inputs from Martinez). The mix of water varies depending on location and time. The Sacramento River branch of the Delta receives a greater contribution from the Sacramento River. At the confluence of the Sacramento and San Joaquin Rivers at Antioch, the flow may be a more even mixture of water from different sources. A portion of the water in the Delta is also exported out through aqueduct for agricultural and urban uses in other parts of the state. Because of this diversion, a time-varying fraction of San Joaquin River and the Sacramento River waters reach the North Bay. Because the San Joaquin River has much higher selenium concentrations than the other major sources, it is important to estimate the contribution from the San Joaquin River to the North Bay. The Delta Simulation Model II (or DSM2) model provides a tool to estimate contributions from different sources to specific locations within the Delta, and was used to estimate the composition of flow at different locations.
The loads of selenium from the Sacramento and San Joaquin Rivers to the Delta were estimated using different methods in previous work (Tetra Tech, 2008). One of the approaches used selenium concentrations and flow at the Sacramento River at Freeport and San Joaquin River at Vernalis to estimate selenium loads from these two rivers separately and by accounting for loss of loads through the Delta aqueducts (Tetra Tech, 2008). The other approach estimated selenium loads by estimating loads at Sacramento River at Rio Vista and loads at San Joaquin River at the confluence by applying a Delta removal constant for concentrations in the San Joaquin River (Meseck, 2002). In this study, selenium loads from upstream were estimated as selenium loads from the Sacramento River at Rio Vista (representing Sacramento River branch of the Delta) and the San Joaquin River at the confluence (Antioch, representing San Joaquin River branch of the Delta), using the estimated selenium concentrations and flow from DSM2 at Rio Vista and at the confluence. The sum of loads from these two locations represents an estimate of total selenium inputs to the bay. The locations of these stations in the eastern portion of North San Francisco Bay are shown in Figure 1.
Figure 1 Locations of key stations in the eastern portion of San Francisco Bay (Source: GoogleMaps)
DSM2 Simulations DSM2 is a one-dimensional mathematical model for simulating the Sacramento – San Joaquin Delta hydrodynamics, water quality, and particle tracking. The DSM2 model calculates stages, flows, velocities, transport of individual particles, and mass transport processes for conservative and non-conservative constituents, including salts, water temperature, dissolved oxygen (DO), and dissolved organic carbon. The DSM2 model can be applied using a “fingerprinting” mode to predict the sources of water at a given location in the Delta.
In this study, the model was used to estimate dissolved selenium loads to the Delta from the Sacramento and San Joaquin River watersheds. The DSM2 model was run using the fingerprinting mode to estimate water composition at three output locations:
• Sacramento River at Rio Vista; • San Joaquin River at Antioch; and • Mallard Island.
These locations characterize the Central Valley loads as well as the concentrations in the eastern portion of San Francisco Bay (at Mallard Island). The DSM2 simulation considers contributions from five inputs including:
• Sacramento River at Freeport, • San Joaquin River at Vernalis, • Martinez, representing tidal inputs from the Bay to the Delta • East side tributaries, and • Agricultural return flows.
The DSM2 model was used to simulate volumetric contributions from these sources to the three output locations. The model was run for the period 1992- 2013.
Simulated Composition of Water The simulated composition of water at Sacramento River Rio Vista, San Joaquin River at Antioch and Mallard Island is shown in Figure 1 to Figure 3. For Rio Vista, the dominant source of flow is the Sacramento River. Contributions from other sources are generally low (Figure 1).
For the San Joaquin River at Antioch, the major source of flow is also the Sacramento River. However, contributions from the San Joaquin River, Martinez and east side tributaries are considerably larger than for Rio Vista (Figure 1). Contributions from Martinez are generally ~30%. Contributions from the San Joaquin River and east side tributaries are more significant during the wet years.
For Mallard Island, the dominant source of water is the Sacramento River at Freeport (Figure 3). Martinez contributions are also a significant source of water to Mallard Island. The contribution from the San Joaquin River to Mallard Island is generally low with some exceptions during the wet years. During the wet years, contribution from the San Joaquin River to Mallard Island can reach as high as 40% for limited periods. Contribution from the east side tributaries is generally low.
Figure 2 DSM2 simulated volumetric contributions from source waters to Rio Vista
0102030405060708090
100
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Volu
met
ric C
ontr
ibut
ion
(%)
Sacramento River at Rio Vista
Sac
Ag
East
MTZ
SJR
Figure 3 DSM2 simulated volumetric contributions from source waters to Vernalis
Figure 4 DSM2 simulated volumetric contributions from source waters to Mallard Island
Simulated Selenium Concentrations Simulated volumetric contributions from source waters in conjunction with selenium concentrations in the source water were used to calculate concentrations at given locations. Dissolved selenium concentrations from USGS for the Sacramento River at Freeport and the San Joaquin River at Vernalis (available for the time period of 2007 -2014) were used to calculate concentrations at Sacramento River at Rio Vista, San Joaquin River at Antioch and Mallard Island. Average dissolved selenium concentrations are 0.095 μg/l for Sacramento River at Freeport and 0.57 μg/l for San Joaquin River at Vernalis. Dissolved selenium concentrations for the Sacramento River at Freeport and the San Joaquin River at Vernalis are shown in Figure 4. The relationship between concentrations and daily flows are shown in Figure 5. There is minimal dependence on flow rate at Freeport, and a strong inverse relationship with flow at Vernalis. For the simulations performed here observed data, linearly interpolated between sampling dates, were used. Dissolved selenium concentrations for Martinez, east tributaries and agricultural
0102030405060708090
100
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Volu
met
ric C
ontr
ibut
ion
(%)
San Joaquin River at Antioch
Ag
East
Mtz
Sac
SJR
0102030405060708090
100
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Volu
met
ric c
ontr
ibut
ion
(%)
Mallard Island
Ag
East
MTZ
Sac
SJR
return flows used in the calculation were assumed to be 0.09 μg/L, 0.1 μg/L and 0.11 μg/L and were assumed constant. Concentrations at Yolo Bypass are assigned the same value as the Sacramento River at Freeport.
For the Sacramento River at Rio Vista and the San Joaquin River at Antioch, only the contribution from the Sacramento River, San Joaquin River, east side tributaries and agriculture return flow is used in the calculation. Therefore the calculated concentrations reflect contribution from the rivers only and not from the bay.
For the Sacramento River at Rio Vista, calculated dissolved selenium concentrations are shown in Figure 8. Estimated dissolved selenium concentrations are generally ~0.1 μg/l with occasional peaks greater than 0.1 μg/l.
For the San Joaquin River at Antioch, calculated dissolved selenium concentrations are shown in Figure 8. Estimated dissolved selenium concentrations are generally ~0.1 μg/l with some peak values greater than 0.2 μg/l. The calculated dissolved selenium concentrations at Antioch were compared to concentrations at Vernalis and with a reduction factor of 60% from the Vernalis concentrations as used in Meseck (2002). Calculated dissolved selenium concentrations were lower than values at Vernalis combined with the 0.6 reduction factor. This demonstrates that following the diversion in the South Delta and mixing with the Sacramento River, selenium concentrations at Antioch were significantly lower than Vernalis.
For Mallard Island, the calculated dissolved selenium concentrations are shown in Figure 8. Estimated dissolved selenium concentrations are generally ~0.1 μg/l with occasional peaks greater than 0.2 μg/l. The peaks are strongly associated with higher volumetric contributions from the San Joaquin River.
Figure 5 Dissolved selenium concentrations for Sacramento River at Freeport and San Joaquin River at Vernalis used in calculation
0
0.05
0.1
0.15
0.2
0.25
2006
2007
2008
2009
2010
2011
2012
2013
2014
Diss
olve
d Se
(μg/
l)
Sacramento River at Freeport
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
2006
2007
2008
2009
2010
2011
2012
2013
2014
Diss
olve
d Se
(μg/
l)
San Joaquin River at Vernalis
Figure 6 Relationship between dissolved selenium concentrations and daily flow rates on the sampling dates, for Freeport and Vernalis.
Figure 7 Calculated dissolved selenium concentrations for Sacramento River at Rio Vista
y = 0.0261ln(x) - 0.1578R² = 0.1819
0
0.05
0.1
0.15
0.2
0.25
0 10000 20000 30000 40000 50000 60000 70000
Diss
olve
d Se
at F
reep
ort (
µg/l
)
Flow at Freeport(cfs)
y = -0.2ln(x) + 2.1519R² = 0.234
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
0 5000 10000 15000 20000 25000 30000
Diss
olve
d Se
at V
erna
lis (µ
g/l)
Flow at Vernalis ( cfs)
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
0.00
0.05
0.10
0.15
0.20
0.25
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
San
Joaq
uin
Rive
r Con
trib
utio
n (%
)
Diss
olve
d Se
at R
io V
ista
(μg/
l)
Dissolved Se at Rio Vista
San Joaquin River contribution(%)
Figure 8 Calculated dissolved selenium concentrations for San Joaquin River at Antioch, compared to concentrations at Vernalis, and concentration multiplied by a constant factor of 0.4 (representing a 60% reduction in concentration).
Figure 9 Calculated dissolved selenium concentrations at Mallard Island
0
10
20
30
40
50
60
70
80
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
San
Joaq
uin
Rive
r con
trib
utio
n (%
)
Diss
olve
d Se
at A
ntio
ch (μ
g/l)
Dissolved Se at Antioch
SJR contribution
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Diss
olve
d Se
(µg/
l)
Antioch Vernalis Vernalis *0.4
0
5
10
15
20
25
30
35
40
45
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
San
Joaq
uin
Rive
r Con
trib
utio
n (%
)
Diss
olve
d Se
at M
alla
rd (µ
g/l)
Dissolved Se at Mallard
SJR contribution
Calculated Selenium Loads to the Delta Model-simulated flow at Rio Vista is very close to the flow at Freeport (Figure 9). Calculated selenium loads at the Sacramento River at Rio Vista are shown on a daily basis (Figure 10) and for the dry and wet seasons of each water year (Figure 11). The wet season was defined as Oct 1st to Apr. 30th and the dry season was defined as May 1st to Sep. 30th, similar to previous approach in Tetra Tech (2008). Estimated dissolved selenium loads averaged 1,881 kg/yr for the entire period of 1993-2012.
Model-simulated flow at Antioch shows large variation across years (Figure 12). Flow at Antioch significantly increases during the wet years. Calculated selenium loads at Antioch are shown on a daily basis (Figure 13) and for the dry and wet seasons (Figure 14). Estimated dissolved selenium loads from Antioch ranged from 7 kg/yr to 3,907 kg/yr. Selenium inputs at this location reflect the variable mixture of inputs from the Sacramento River, the San Joaquin River, the east side tributaries and agricultural return flow.
Estimated total dissolved selenium loads from the Delta to the Bay are shown in Figure 15. The estimated loads compared to previous estimates for the same time period by Tetra Tech (2008) are slightly higher, particularly during the wet years. This is possibly due to the DSM2 model computing a higher contribution from the San Joaquin River during the wet years, plus the higher selenium concentrations from the San Joaquin River, resulting in even higher loads during wet years. A summary of dissolved selenium loads by year and season is listed in Table 1.
Dissolved selenium loads at Mallard Island could also be estimated using the estimated concentrations using the DSM2 finger printing results and DSM2 simulated flow at Mallard Island (tidally average flow and concentrations at daily basis). Estimated dissolved selenium loads at Mallard Island compared well to the estimated load at Rio Vista plus Antioch (Figure 16).
Another approach for estimating selenium loads to the Bay is to subtract selenium loads exported through the aqueducts from the loads to the Delta (i.e., sum of Freeport, Yolo, Vernalis and east-side tributary loads). The results from this mass balance approach are compared to the estimated loads at Rio Vista plus Antioch and the estimated loads at Mallard Island (Figure 17 and Figure 18). Although the loads from the different approaches are comparable, the values from the mass balance approach are slightly higher; this is associated with the flow volumes associated with the individual inputs, and a result of losses of water volume in the Delta, likely through consumptive use on the Delta islands.
Discussion The DSM2 modeling framework can be adapted to estimate the contribution of different watershed sources to the Delta, using upstream boundary stations where the loads can be calculated directly through measurements.
Because of the extensive calibration that DSM2 has been subjected to over the past two decades, there is credibility associated with the mixing and tidal hydrodynamic processes represented by the model. The model is a more robust tool to estimate the contributions of the riverine loads to the bay. Although
the long term annual loads are in the same range as previously estimated (Tetra Tech, 2008), the model illustrates some important seasonal patterns. In particular, the contribution of the San Joaquin River can be elevated during the wet months of wet years, such that high concentrations at Mallard Island may be observed. In contrast the dry year loads from the San Joaquin River to the bay are very small compared to the Sacramento River loads. This finding is consistent with observation of high selenium concentrations in January 2005 and 2006 that were reported by Davis et al. (2012), and likely reflects the greater contribution of San Joaquin flows during this period. However, this is a limited data set and has not been continued since the 2005 and 2006 sampling.
Going forward, additional sampling at a regular frequency at Mallard Island is recommended (in addition to the USGS sampling being performed at Freeport and Vernalis), because this location is a good representation of the relative contribution of Sacramento and San Joaquin Rivers. The relative contributions vary in a complex manner, because higher concentrations occur during wet years, this modeling suggests that worst case conditions for selenium in the bay—at least in the context of Central Valley loads—may not occur during dry or drought years.
Figure 10 Model simulated flow at Sacramento River at Rio Vista
0
50,000
100,000
150,000
200,000
250,000
300,000
350,000
400,000
450,000
500,000
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Rio
Vist
a flo
w (c
fs)
Freeport
Rio Vista
Figure 11 Estimated dissolved selenium loads at Sacramento River at Rio Vista
Figure 12 Estimated dissolved selenium loads at Sacramento River at Rio Vista by water year and season. The water year type classification is based on DWR’s water year hydrologic classification for the Sacramento Valley (cdec.water.ca.gov/cgi-progs/iodir/WSIHIST). W: Wet, AN: Above normal, BN: Below normal, D: Dry, C: Critical year.
0
20
40
60
80
100
120
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Diss
olve
d Se
load
s (kg
/day
)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
1993
(AN
)19
94(C
)19
95(W
)19
96(W
)19
97(W
)19
98(W
)19
99(W
)20
00(A
N)
2001
(D)
2002
(D)
2003
(AN
)20
04(B
N)
2005
(AN
)20
06(W
)20
07(D
)20
08(C
)20
09(D
)20
10(B
N)
2011
(W)
2012
(BN
)
Diss
olve
d Se
load
s (kg
/sea
son) Wet Season
Dry Season
Figure 13 Model simulated flow at San Joaquin River at Antioch
Figure 14 Estimated dissolved selenium loads at San Joaquin River at Antioch
0
20000
40000
60000
80000
100000
120000
140000
160000
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
SJR
at A
ntio
ch F
low
(cfs
)
0
10
20
30
40
50
60
70
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
Diss
olve
d Se
load
s at A
ntio
ch (k
g/d)
Figure 15 Estimated dissolved selenium loads at Antioch by water year and season. The water year type classification is based on DWR’s water year hydrologic classification for the San Joaquin Valley (cdec.water.ca.gov/cgi-progs/iodir/WSIHIST). W: Wet, AN: Above normal, BN: Below normal, D: Dry, C: Critical year.
Figure 16 Estimated total dissolved selenium loads to the Bay by water years. The water year type classification is based on DWR’s water year hydrologic classification for the San Joaquin Valley (cdec.water.ca.gov/cgi-progs/iodir/WSIHIST). W: Wet, AN: Above normal, BN: Below normal, D: Dry, C: Critical year.
0
500
1000
1500
2000
2500
3000
3500
4000
4500
1993
(W)
1994
(C)
1995
(W)
1996
(W)
1997
(W)
1998
(W)
1999
(AN
)20
00(A
N)
2001
(D)
2002
(D)
2003
(BN
)20
04(D
)20
05(W
)20
06(W
)20
07(C
)20
08(C
)20
09(B
N)
2010
(AN
)20
11(W
)20
12(D
)
Diss
olve
d Se
load
s (kg
/sea
son)
Wet Season
Dry Season
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
1993
(W)
1994
(C)
1995
(W)
1996
(W)
1997
(W)
1998
(W)
1999
(AN
)20
00(A
N)
2001
(D)
2002
(D)
2003
(BN
)20
04(D
)20
05(W
)20
06(W
)20
07(C
)20
08(C
)20
09(B
N)
2010
(AN
)20
11(W
)20
12(D
)
Diss
olve
d Se
Loa
ds (k
g/yr
)
Wet season
Dry Season
Table 1. Summary of calculated dissolved selenium loads to the Bay.
Water Year (October 1 to September 30)
Type Dry season (kg/season)
Wet season (kg/season)
Total (kg/yr)
1993 W 639 1761 2400 1994 C 223 543 765 1995 W 2106 4417 6523 1996 W 800 2884 3684 1997 W 381 5747 6128 1998 W 2298 5691 7989 1999 AN 518 2650 3169 2000 AN 502 2138 2641 2001 D 287 659 946 2002 D 299 886 1185 2003 BN 622 1221 1843 2004 D 317 1614 1931 2005 W 981 1104 2086 2006 W 1784 4453 6236 2007 C 294 600 894 2008 C 230 896 1126 2009 BN 252 492 744 2010 AN 457 1212 1669 2011 W 770 2158 2928 2012 D 299 536 835
Figure 17 Comparison of estimated dissolved selenium loads as sum of Rio Vista and Antioch and estimated loads at Mallard Island.
Figure 18 Comparison of estimated dissolved selenium loads as sum of Rio Vista and Antioch, estimated loads at Mallard Island (using tidally average flow calculated by DSM2 and estimated concentrations), and estimated loads as mass balance of Freeport + Vernalis + Yolo + East – export loads.
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Diss
olve
d Se
load
s (kg
/yr)
Rio Vista + Antioch
Mallard
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Diss
olve
d Se
load
s (kg
/yr)
Rio Vista + Antioch
Mallard
Freeport +Yolo+ Vernalis + East - export
Figure 19 Top. Estimated dissolved Se loads to the Delta from individual sources. Middle. Loads exported through aqueducts. Bottom. Loads entering the Bay (Loads to the Delta – Exports).
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Diss
olve
d Se
load
s (kg
/yr)
Freeport+ Yolo
Vernalis
East
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Diss
olve
d Se
load
s (kg
/yr)
Export
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
Diss
olve
d Se
load
s (kg
/yr)
Freeport +Yolo+ Vernalis + East -export
References Meseck, S.L. 2002. Modeling the biogeochemical cycle of selenium in the San Francisco Bay. Ph.D. Dissertation, Old Dominion University.
David, N., Gluchowski, D.C, Leatherbarrow, J.E, Yee, D., and McKee, L.J, 2012. Evaluation of Loads of Mercury, Selenium, PCBs, PAHs, PBDEs, Dioxins, and Organochlorine Pesticides from the Sacramento-San Joaquin River Delta to San Francisco Bay. A Technical Report of the Sources Pathways and Loading Work Group of the Regional Monitoring Program for Water Quality: SFEI Contribution #681. San Francisco Estuary Institute, Oakland, CA. 49 pp. http://www.sfei.org/sites/default/files/Mallard_Island_Report_120216.pdf
Tetra Tech, 2008. Technical memorandum 2: North San Francisco Bay selenium data summary and source analysis. Prepared for: Regional Water Quality Control Board San Francisco Bay Region.