BAY DELTA MODELING FORUM
REVIEW OF 1995 WORKSHOPS
FEBRUARY 16, 1996
CONTENTS
PAGE
1 Preface 1
2 Introduction 2
3. Delta Carriage Water Workshop 5
4. Bay Delta Standards Agreement Workshop 10
5. Economics Models Workshop 11
6. Bay Delta Monitoring Workshop 13
7. Toxics and Water Quality Workshop 22
8. Statewide Operations Workshp 28
9. Drinking Water Quality Workshop 35
10. Delta Modeling for End Users 40
11. Biological Models Workshop 45
PREFACE
This document is a compilation of abstracts received from
presenters at each of the nine Bay-Delta Modeling Forum
workshops that were held during 1995. E-mail addresses and
telephone numbers of the presenters are provided following the
individual abstracts should the reader wish to obtain more
information on the topic or the published abstract. The
workshops were well attended with numbers ranging from 35 to
120. Membership in the Forum increased throughout the year -
most memberships resulting from attendance at one or more of the
workshops. A retrospective of the 1995 workshops will occur at
the Annual meeting of the Bay-Delta Forum at Asilomar, Monterey,
February 27 - 28, 1996.
Nigel W.T. Quinn, Editor
Lawrence Berkeley National Laboratory /
US Bureau of Reclamation
Chair, Bay-Delta Forum Technical Committee
GOALS OF THE BAY-DELTA MODELING FORUM
The Forum is a statewide, non-profit, non-partisan,
"consensus"organization whose mission is to increase the
usefulness of models for analyzing water-related problems in the
San Francisco Bay, Sacramento-San Joaquin Delta, and Central
Valley system and other related California areas. The Forum
carries out this mission by:
o Providing a consensus-building atmosphere on water-related
issues;
o Maintaining a modeling clearinghouse that provides an open
forum for the exchange, improvement, and pooling of models,
modeling information, and professional resources (under
development);
o Assisting in mediating technical disputes involving
physical, chemical, biological, and economic modeling;
o Conducting impartial peer reviews of models in order to
document strengths and weaknesses, suggest improvements, and
identify appropriate applications;
o Seeking input from California water stakeholders and
decision makers about their modeling needs; and
o Providing educational opportunities through technical
conferences and workshops.
The Forum has over 20 institutional and 110 individual members
representing a wide range of agricultural, urban, environmental,
consulting, and academic modeling perspectives.
For more information, contact the Forum at:
Phone: (510) 231-9539
FAX: (510) 231-9414
E-mail: modelingforum@sfei.org
INTRODUCTION
The following briefing paper was originally written by Wim
Kimmerer to introduce the topic of the biological modeling
workshop and to provoke thought and discussion on some of the
key issues. I have included this paper in this compilation of
abstracts, after some minor editing, since much of what is
written is applicable to each and all of the models presented in
the nine workshops dealing with Bay-Delta issues during 1995.
Modeling in the Bay-Delta
Wim Kimmerer (SFSU)
What do we mean by modeling? Any concept or
simplification of a real thing is a model. For the purposes of
the workshops, however, we will confine the discussion to models
with some (however limited) predictive capability. Such models
can be arrayed along an axis of increasing empirical content in
one direction, and increasing knowledge of mechanisms in the
other. One end of the axis is anchored by purely empirical
models (e.g. regression models). The other is held down by
purely theoretical models. Models of greatest interest in the
Bay-Delta scientific community are probably closer to the middle
of this axis. They would be based on understanding of parts of
the prototype system, have some connection to data from the
prototype, and would be usable in predicting the response of the
prototype to conditions not previously observed. Models that
meet this description may not yet exist for the bay-delta
ecosystem. We will also confine the discussion to models of
predominantly open-water systems, as opposed to marshes and
riparian zones. In addition, we will emphasize the use of
models for management, while recognizing that models developed
for research may ultimately have uses in management.
Why model? Models are constructed for a variety of purposes
which can roughly be divided into research and management.
Research models are constructed either to determine the
consequences of a series of assumptions, or to translate data
from one context to another. In many cases model predictions
are compared with observations and the degree of "fit" of the
model is evaluated. The most useful research models are those
that do not do a good job of predicting, forcing researchers to
reevaluate and revise assumptions. Unfortunately, in many cases
an acceptable fit results in acceptance of the model by its
author, with the result that alternative, equally plausible
models are not tested.
Modeling for management has a different aim, which is to
predict the consequences of management actions given the best
estimate of how the system is believed to respond. Relatively
few alternative methods are available to predict the effects of
management activity or engineering changes to a natural system.
Most often managers rely on expert opinion for guidance. In
doing so, they fail to note that in fact they are relying on the
experts' conceptual (or other) models, but with the assumptions
not made explicit.
Constructing simulation models for management purposes has two
advantages over the more traditional use of expert judgement: 1)
assumptions are made explicit; and 2) the consequences of
uncertainty in functions, parameters, and data can be explored.
However, it is crucial to scale the complexity of the model to
the level of understanding of the system being modeled. For
example, it would make no sense to construct a detailed
population model to predict the response of delta smelt to their
environment until better knowledge of the controls on their
abundance became available (although a research model might be
useful in exploring those controls).
Model development can be quite expensive, especially where it
must be combined with data collection to provide input to the
model. However, for managing the bay-delta estuary and
tributaries, modeling is essential for:
o Exploring the relative costs and benefits of different actions
o Investigating the consequences of alternative
descriptions of the system
o Determining the key weaknesses in understanding of the system.
Finally, modeling of biological systems is required by several
statutes and regulations. Although these requirements may be
somewhat naive and overly optimistic, they nevertheless require
the Bay-Delta scientific community to examine carefully the
opportunities for use of models in management.
A mismatch between expectations and capabilities One
of the reasons to hold these workshops is a persistent mismatch
between what managers and engineers expect of modeling, and what
biologists think they can deliver. The high expectations of the
engineering community arise from extensive experience with
simulation of the physical environment (e.g. hydrology,
hydrodynamics, temperature), which can give generally realistic
predictions of physical conditions. Although models of
hydrodynamics may seem quite complex, in fact they embody only
one equation plus conservation of mass; their complexity arises
mainly from the necessity to parameterize turbulent mixing at
scales smaller than the length scale of the model cells. Models
of biological systems, on the other hand, must describe systems
for which the equations are at best poorly known.
To understand this, consider a model of a population of
annually-reproducing fish. This model would describe the
reproductive rate of the adult fish, and then the survival of
the young fish as they grow to maturity. The population grows
or decays at an annual rate equal to the product of the
reproductive rate and the survival proportions for each life
stage. At some point in the life cycle, there must be negative
feedback or "density dependence", by which the survival is
inversely related to population size. Without this feedback,
the population will grow or decay without limit. Although there
is plenty of evidence that such feedback exists, only rarely is
it possible to determine the mechanism or even at what stage of
the life cycle this feedback occurs. A model attempting to
predict how the population will change as a result of changes
in, say, egg survival will be completely unsuccessful if
density-dependent mortality occurs in the larval stage. As
another complication, consider that most models of populations
are trophic-dynamic, i.e. they describe changes in populations
based mainly on changes in food supply (and sometimes
predators). However, it is not clear that food supply is the
principal influence on specific populations. In the Bay-Delta
estuary, many populations vary positively with freshwater
outflow, but this variation is probably not a result of
covariation in food supply. A trophic-dynamic model would be
unsuccessful at describing how these populations vary with their
environment.
If biological models cannot be built from first principles,
how can we proceed? It seems to me that a logical next step is
to use the available data to construct models that incorporate
known mechanisms along with some empirical information. This
approach has been used for models of salmon smolt survival
through the delta and striped bass production, although in both
cases the models have been criticized on statistical grounds.
Examples of Models Modeling efforts to date in the bay-delta
system have been constrained by the questions being asked
(whether for management or research purposes) and the local
emphasis on effects of freshwater flow. These constraints have
led in the past to an emphasis on single-species models of
striped bass and salmon, and less emphasis than elsewhere on
lower trophic levels. Some biological models currently in
existence, listed by decreasing degree of empiricism, and
increasing knowledge of mechanisms (names in parentheses are
those who presented models at the biological modeling workshop)
Fish-X2 models (Kimmerer)
Salmon smolt survival (Williamson)
Survival models for threatened species (Botsford)
"Particle"-tracking models (Quinn, Cowan)
Salmon population models (Williamson)
Striped bass individual-based population model (Cowan)
South bay phytoplankton (Lucas)
Some biological models not now used in the Bay-Delta:
Rule-based simulation models
Coupled physical-biological models (e.g. nutrient-
phytoplankton-zooplankton models; these are
used extensively to describe open-ocean systems)
Material flow models (e.g. network models, trophic-
dynamic models)
Multi-species fishery models
Caveats for modeling
Modelers and their employers need to akeep the
following in mind in developing, testing, and applying
models:
1. The form of the model depends on the questions. The first
step in any modeling exercise should always be to decide
what questions the model will attempt to answer. This step
constrains the form, scale, and content of the model. A model
built without specific questions in mind is unlikely to perform
well at answering questions determined post hoc.
2. The map is not the territory. Modelers, with intimate
knowledge of the content of their models, do not usually
confuse model output with that of the prototype. Model users,
on the other hand, can readily be persuaded that the model is an
exact replica of the prototype with all of its complexity. It
is incumbent on modelers to disabuse model users of the notion
that the model is more than just a tool to be used in
conjunction with other tools.
3. Model validation requires an illogical statistical model
There is a lot of pseudo-rigor in the practice of
calibrating a model against one set of data and "validating"
it against another. Often this is done with some specific
criteria, based on the measurements at hand, as to how well the
model should fit the data, and sometimes with a statistical test
of the model's fit to the data. However, the model prediction
is the null hypothesis against which the data are tested.
Statistical tests are designed to distinguish between the data
and the null hypothesis, and the more data available, the more
precise that distinction can be. Therefore, it is in the
modeler's interest to have as few data as possible, with the
widest possible confidence intervals, to insure that the model
fits well. Collecting more data will practically guarantee
deviation of model from data.
4. Model validation is a flawed concept anyway If we
construct a model and "validate" it by comparing it with a
set of data, what have we done? In fact, all validation does
for us is give us a sense that model predictions are generally
in the right ballpark. It does not permit us to infer that the
model is a correct description of the system, or that model
predictions for other sets of inputs would also be correct.
There may be (and usually is) an infinite set of possible
alternative models, many of which could fit the data better or
be more true to the response of the prototype. Since the data
set available for validation is usually small and the number of
parameters and functions large, some reasonable fit could be
expected with a wide variety of alternative descriptions.
The future of modeling:
What will be the most fruitful paths for development of models
over the next 5-10 years?
Given the difficulties in developing realistic models of
biological systems alluded to above, it is clear than any
modeling effort will need to be realistic and clear about the
uncertainties. This includes not only uncertainties in
parameters and data (e.g. population indices), but also
uncertainties in structure and function. Thus, models for
management purposes should explicitly consider alternative
formulations and display prominently the differences among
alternatives.
Modeling will have to be done in close conjunction with
research and monitoring programs. This has not always been the
case in the past. Models have been developed by one group of
people, and data collected by the other, with insufficient
communication between groups. This results in a lack of
"ownership" for the models, with the consequence that they are
developed but not used. An alternative institutional framework
may be needed to insure that models are integrated fully with
other methods of investigation.
Consideration should be given to "meta-models", or models
incorporating sub-models with different scales and levels of
detail. For example, suppose a goal of a modeling exercise was
to assess the changes in all estuarine-dependent species
resulting from a specific change in the flow regime. A model
could be constructed to answer this question specifically, but
it would have limited applicability and flexibility. An
alternative approach would be to have individual models of
different subsystems (e.g. populations, races, regions), having
different levels of complexity, and that could be queried by an
overarching model. As understanding of the subsystems
developed, the individual models could be revised without
requiring a revision of the meta-model.
Finally, these workshops are a first step in bringing together
scientists and engineers involved in modeling both physical and
biological systems in the Bay-Delta. The current regulatory
framework, and the interest in solving environmental problems of
the Bay-Delta, suggest that modeling efforts will need to be
intensified in the future. This process would benefit by
frequent communication among interested parties. To the extent
that these workshops are successful, it may be useful to
continue holding conferences or workshops on related topics from
time to time.
E-mail : kimmerer@mercury.sfsu.edu
Phone : (415) 435-7118
----------------------------------------------------------------
Delta Carriage Water Workshop
Holiday-Inn Marine World
Vallejo
May 4, 1995
Organizers : Greg Gartrell (Contra Costa Water District)
Richard Denton (Contra Costa Water District)
Spreck Rosekrans (EDF) Welcome and introduction
On May 4, 1995, the Bay-Delta Modeling Forum sponsored an all
day workshop on "Flows, salinity and carriage water." Factors
addressed at the workshop included the accuracy of current
models in determining the flows required to meet salinity
standards, the effect of net flow on salinity in the Delta, ways
of improving current models, appropriate carriage water charges
for water transfer, and, tests for resolving flow/salinity
issues. Spreck Rosekrans (EDF) served as moderator. When the
State Water Project and Central Valley Project are not pumping,
a given amount of Delta outflow will be required to maintain a
given salinity at the intake to the Contra Costa Canal at Rock
Slough. Carriage water can be thought of as the additional
outflow that is assumed to be needed to maintain that same
salinity if the projects are pumping. The higher the exports,
the greater the amount of additional outflow that is assumed to
be required to maintain the same salinity. Carriage water is
assumed to be needed to counteract tidally-averaged reverse
flows in the lower San Joaquin River and through Threemile
Slough. This flow quantity is also referred to as QWEST where
QWEST is negative when the flows in these channels are reversed
(i.e. eastward and southward, respectively).
E-mail : spreck@edf.org
Phone : (510) 658-8008
Greg Gartrell (CCWD) Net flow models and tidal dispersion
Greg Gartrell reviewed the definition of carriage water and
the assumption that reverse flows on the lower San Joaquin leads
to increased salinity intrusion at Contra Costa Water District's
diversion point. The carriage water model assumes that
additional outflow is required to counteract this increase in
salinity. These assumptions have been used to justify
additional water transfer charges. The carriage water model
only allows salt transport by net reverse flows whereas tidal
flows and tidal dispersion actually dominate the hydrodynamic
and salinity transport process. CCWD attempted to verify the
carriage water model using historical Delta outflows and found
that the model greatly overestimates historical Rock Slough
chloride concentrations. If historical Delta outflows are input
to the carriage water model, the predicted Rock Slough chlorides
can be as high as 1700 mg/l, whereas historical chlorides have
seldom exceeded 250 mg/l. The error in the carriage water model
predictions of Rock Slough chlorides was found to have a strong
positive correlation with QWEST. The smallest error of estimate
occurred when the Rock Slough chlorides were assumed to just
depend on outflow only with no dependence on export level
(independent of QWEST). Greg showed how Rock Slough chlorides
could be closely correlated with Jersey Point salinity with a
14-day lag. Adding QWEST as a parameter did not improve the
correlation. Emmaton salinity on the other hand did appear to
react to changes in QWEST.
The carriage water model also assumes a fixed 20-80 net flow
split between Three Mile Slough and the flow in the San Joaquin
River at Jersey Point. Greg Gartrell presented net flow data
from the Fischer Delta Model, a numerical hydrodynamics and salt
transport model, that suggests that the flow split is not
constant but widely varying. The simulated Fischer Model flows
through Three Mile Slough were found to be well described by the
equation:
3MILE = 0.0845*(SAC - 4.94*SJR) + 805
where SAC and SJR are the net flows (in cfs) in the Sacramento
River and San Joaquin Rivers, respectively, immediately upstream
of Three Mile Slough. Salinity predictions at Rock Slough using
the Fischer Delta Model did appear to have a weak correlation
with QWEST. When the errors in salinity prediction were
accounted for,this QWEST dependence was significantly reduced.
In conclusion, Greg suggested that the assumed carriage water
effect is of the same order of magnitude as the scatter in the
field data and as such a 30% surcharge on water transfers could
not be justified.
E-mail : wrccwd@ccnet.com
Phone : (510) 688-8187
Rick Oltman (USGS) Flow measurements in the western Delta
Rick Oltman described the USGS's program for measuring flows
in the Delta using ultrasonic velocity meters (UVM). There is
currently a lack of measured flow data and the USGS measurements
will help to understand the tidal hydrodynamics of the Delta,
allow calibration and verification of flow models, and assist
real-time decision making by Project operators. UVMs have been
deployed or are proposed on Old and Middle River at the southern
end of Bacon Island, in the San Joaquin River at Stockton, at
Walnut Grove above and below the cross-channel, and four western
Delta sites for measuring Delta outflow (Rio Vista, Three Mile
Slough, Jersey Point and on Dutch Slough). Both 15-minute and
low-pass-filtered flow data were presented. Delta outflows were
found to be highly correlated with the filling and draining of
the Delta over the
spring-neap cycle, as evidenced by fluctuations in low-pass-
filtered tidal stage records. The UVM signals have been
calibrated by taking vertical profiles over the channel cross-
section using ship-based acoustic doppler equipment. The
calibration data remains in good agreement with the UVM data
over the full range of the tidal cycles. The preliminary USGS
work suggests that a fourth-order rating curve gives a
significantly better agreement between the instantaneous UVM
flows and the cross-sectional doppler profile measurements at
Jersey Point than a simple linear fit. At other sites a linear
fit is sufficient.
E-mail :
Phone :
Russ Brown Analysis of salinity measurements in the Delta
Jones and Stokes)
Russ Brown described his use of daily electrical conductivity
data to identify relationships with Delta flows. Russ assumed
that the effective Delta outflow (EO) that determines the mean
salinity for a given day could be given by:
EOt = EOt-1 + (Q - EOt-1) * (1 - exp(-EOt-1/175000)
where Q is the Delta outflow for the given day, and t and t-1
denote the present and previous day's values of EO. A similar
equation was used for monthly data. A negative exponential
equation was used to estimate the variation of measured Antioch
and Jersey Point EC with outflow. Russ suggested that the
western Delta salinity gradient is dominated by the effective Delta
outflow and that "carriage water" effects should be identified
as deviations from the basic negative exponential relationship
between EC at a station and effective Delta outflow.
Russ has made extensive use of field measurements of specific
conductance (EC), both in calibrating his conceptual model and
in learning how salinities and salinity gradients respond to
outflow. For example, EC data from western Delta stations can
be plotted against the EC at Benicia, representing the salinity
at the western end of Suisun Bay. Russ also provided extensive
plots of the ratio of mean daily EC at western Delta stations,
e.g. Jersey/Antioch and Jersey/Emmaton. Because the salinities
at these stations respond at different rates and different times
to increasing and decreasing outflow, these salinity ratios
showed a great deal of scatter.
E-mail : none
Phone : (916) 737-3000
Richard Denton (CCWD) Flow/salinity relationships, G-
model and reverse flows
Richard Denton presented a salinity-outflow model for the
western Delta and Suisun Bay (often referred to as the "G-
Model") and suggested that this model is an improvement on the
carriage water model currently used in DWR's Central Valley
Operations Model DWRSIM. Simulations of the chloride
concentrations at Rock Slough from the Fischer Delta Model and
the G-Model based on monthly DWRSIM outflows were compared and
found to be in good agreement when seawater intrusion was the
main source of salinity. Richard used the G-Model to show that
the carriage water model underestimated the Delta outflow needed
to meet the Rock Slough standard in dry and critical years but
overestimated Delta outflow requirements in wet and normal
years. In some wet years, the carriage water model in DWRSIM
required carriage water releases even though Rock Slough
chlorides were already well below the 250 or 150 mg/l
objectives.
Richard also briefly described another type of salinity-outflow
conceptual model that models salinity transport by tidal
exchange between different well-mixed pools (such as San Pablo,
and Suisun Bay). After each flood and ebb cycle of the model,
there is a net exchange of mass (whether salt or some other
contaminant) between each pair of linked pools. Richard showed
how this could be extended to include tidal exchange between
different "pools" within the Delta (e.g. a Jersey Point and
Emmaton pools connected by Threemile Slough). Because it
includes tidal effects, this "Flushing Model" would be an
improvement over the carriage water model which considers only
mean flows and neglects tidal exchange processes.
Richard also showed how hourly Delta outflow from a Fischer
Delta Model run with real tides could be independently simulated
using the mean daily outflow and an adjustment based on the net
filling and draining of the Delta (as represented by the low-
pass filtered variation in tidal height at Antioch). This
method also worked well when applied to hourly QWEST flows from
the Fischer Delta Model. These results were similar to those
reported earlier in the day by Rick Oltmann from his field UVM
measurements. By adjusting daily estimates of Net Delta Outflow
using low-pass filtered field measurements of tidal height at
Antioch, Richard was able to reproduce the daily variations in
salinity data measured in Suisun Bay. If only the daily
estimates of NDO are input to the G-Model the output salinities
do not show daily fluctuations but more closely resemble 14-day
averaged salinities.
Richard concluded by looking at how much of the remaining
variance in salinity at Jersey Point and Emmaton (after taking
into account variation with Delta outflow) could be attributed
to reverse flow or QWEST. Jersey Point specific conductance (or
EC) did not show any additional trend with QWEST. Richard also
showed how Rock Slough chlorides are closely correlated with
Jersey Point EC with a 14-day time lag. This correlation was
also independent of QWEST. Emmaton EC, on the other hand, did
show some dependence on QWEST. As the flow on the Sacramento
near Emmaton increased relative to the San Joaquin flow at
Jersey Point, the Emmaton EC decreased, as might be expected.
However, this did not result in a corresponding increase in
Jersey Point EC, perhaps because of the high degree of tidal
exchange through Threemile Slough.
E-mail : wrccwd@ccnet.com
Phone : (510) 688-8187
Francis Chung (DWR) Carriage water issues and indirect evidence
Francis Chung noted that the implications of carriage water
not existing would be that a lumped-parameter model (based only
on net Delta outflow) would be sufficient to describe salinity
variations in the Bay-Delta estuary. It would also follow that
the Delta cross-channel would have no effect on improving export
water quality. Francis invited the audience to join him in a
search for carriage water within time-series records of
historical flow and export operations and Delta salinities.
Francis defined carriage water as the amount of water needed in
excess of an increased export so as to maintain the same
salinity in the Delta, in other words, the marginal export cost.
Francis noted that the large variations in both salinity and
flow made it difficult to distill out the effect of carriage
water. There were several examples where the Delta outflow
remained relatively constant but the chlorides at Rock Slough
increased (e.g. Summer of 1994 when the chlorides increased from
50 mg/l to 230 mg/l). These increases roughly coincided with
increases in export pumping. However, an audience member
suggested that another explanation might be that the earlier
higher outflows had decreased the chloride concentrations and
the Rock Slough chlorides were merely approaching a new steady
state condition (e.g. the approximately constant outflow of
about 4,000 cfs corresponds to a steady state chloride
concentration of about 220 mg/l at Rock Slough).
Francis also used DWR's salinity-transport model DWRDSM to
test for carriage water effects. Increasing the export rate
while maintaining a constant Delta outflow caused an immediate
decrease in the predicted Rock Slough chlorides using DWRDSM but
the chlorides then slowly increased to reach a higher steady-
state value. The DWR's Artificial Neural Network model showed
very similar results. These simulated data results are
consistent with the existence for carriage water. However, more
detailed calibration and verification of these and other models
need to be carried out to confirm that they are accurately
simulating field conditions. No error analysis for the DWRDSM
or the ANN work was presented.
E-mail : chung@dop.water.ca.gov
Phone : (916) 653-5601
Ralph Finch (DWR) Neural networks and probabilistic flows
Ralph Finch presented results from his Artificial Neural
Network model. This "black box" model is non-linear and allows
multiple input parameters (such as Sacramento, San Joaquin, and
east side stream inflows to the Delta, Delta exports and
estimated consumptive use, gate positions and barriers, tidal
height, wind speed and barometric pressure). Because the model
is a black box, it does not provide physical insight into
salinity-outflow processes in the Delta. It is best suited for
interpolating existing ranges of data not for modeling a future
reconfigured Delta system. A wide array of transfer or fitting
functions are utilized within the model and allow any function
to be approximated with a finite number of discontinuities.
Ralph showed an example of the ability of the ANN model to fit
a compound sine wave function with added noise. He has applied
the model to predict Collinsville EC and Rock Slough chlorides,
the latter being more difficult because of the presence of
agricultural drainage in some months. Comparisons between
predicted and actual data still shows a high degree of scatter
but the model is being further refined. A member of the
audience referred to a recent Scientific American article that
advises giving the ANN model an initial hint. In the case of a
Bay-Delta model this might mean supplying net Delta outflow as
one of the input parameters. Delta outflow is known to explain
a significant portion of the variation of salinity in the
western Delta. Other independent parameters could then be
formed from the ratios of other flows and Delta outflow. A
disadvantage of inputing the separate contributions to Delta
inflow as input parameters is that this approach is that none of
these parameters can be dropped as part of a sensitivity
analysis without affecting their contribution to Delta outflow.
E-mail : rfinch@dop.water.ca.gov
Phone : (916) 653-8268
Spreck Rosekrans (EDF) Moderator : Roundtable discussion on
flows, salinity and carriage water
The roundtable discussion dealt with issues of whether the
existence or non-existence of carriage water had been
established by the speakers or if indeed it could be
established. There was some agreement that if carriage water
did exist its effect was smaller than previously assumed. The
meeting participants drew attention to the need for more
rigorous calibration and verification of these salinity-outflow
models, in particular, the inverse form in which a model is used
to predict the outflow needed to meet a given salinity
objective. There was agreement that this should be done under
the auspices of the Bay-Delta Modeling Forum, but the work load
of many of the presenters may not allow much time to be spent on
this effort.
There also appeared to be some confusion over the difference
between minimum required Delta outflow (to meet salinity
standards even without exports) and carriage water. CCWD's G-
Model predicts that more outflow is needed to meet Rock Slough
salinity objectives in dry and critical years than is predicted
by DWR's carriage water model. This means that the carriage
water model is underestimating the outflow needs in drier years
(with or without exports), not that an additional increment of
outflow is needed to counteract salinity intrusion caused by
reverse flows set up by Delta exports.
Consider, for example, the case where the Rock Slough standard
were only just being met (Delta in balance) and Delta exports
were suddenly increased by 2,000 cfs. To maintain the same
Delta outflow, the projects would need to release an additional
2,000 cfs. The G-Model assumes salinity only depends on outflow
so assumes the 2,000 cfs is sufficient to continue to just meet
the Rock Slough standard. Proponents of carriage water on the
other hand would maintain that the increase in upstream releases
should be closer to 2,600 cfs to provide enough additional flow
to carry water to the export pumps without drawing in additional
salt from the ocean (through increased reverse flow effects).
In dry and critical years, DWRSIM appears to be releasing less
water than needed for the Delta outflow alone (e.g. less than
the 2,000 cfs required in the above example)
E-mail : spreck@edf.org
Phone : (510) 658-8008
-------------------------------------------------------------------
Bay Delta Standards Agreement Workshop
Modeling and the Standards
Holiday Inn North East
Sacramento
May 23, 1995
Organizers : Spreck Rosekrans (EDF)
Richard Denton (CCWD)
Spreck Rosekrans Welcome and introduction
(Environmental Defense Fund)
No abstract received.
E-mail : spreck@edf.org
Phone : (510) 658-8008
Sushil Aurora (DWR) DWRSIM studies : December 15 impacts
No abstract received.
E-mail : sushil@dop.water.ca.gov
Phone : (916) 653-7921
Walter Bourez (WRMI) WRMISIM studies : December 15 impacts
No abstract received.
E-mail :
Phone : (916) 920-1811
Richard Denton (CCWD) Daily operations : Historical
reference and impacts
No abstract received.
E-mail : wrccwd@ccnet.com
Phone : (510) 688-8187
Dwight Russell (DWR) Effects of the standards on
Suisun marsh
Kamyar Guivetchi (DWR)
No abstract received.
E-mail :
Phone :
--------------------------------------------------------------------
Economics Models Workshop
115 Wellman Hall
University of California, Davis
June 21, 1996
Organizer : Jay Lund (UCD)
Overview of Economic Modeling Problems related to Bay-
Delta management
Farhad Farnam, DWR Water Use Forecasting
No abstract received.
E-mail :
Phone :
Steve Hatchett, CH2M-Hill Economic & Financial impacts
of reduced water availability
No abstract received.
E-mail :
Phone : (916) 920-0300
Chris Dumas, EPA Valuing Environmental Uses
Thomas Wegge, Jones & Stokes)
Larry Dale, Larry Dale Assoc.
No abstract received.
E-mail : ludvig@well.sf.ca.us
Phone : (510) 236-9630
Ray Hoagland, DWR DWR Economic Risk Model"
Brief description of traditional economic analysis framework
used by DWR for water supply project evaluation. Problems of
incorporating long-term demand management as well as shortage
contingency measures in
traditional framework. Evolution of risk analysis approach for
urban water supply benefit analysis within DWR. Outline of DWR
Economic Risk Model logic. List of types of subroutines used in
model code. Examples of format of model input and output.
E-mail :
Phone :
Roger Mann, CH2M-Hill Valuing Water Supply Reliability
No abstract received.
E-mail :
Phone :
Jay R. Lund, UCD Integrated Resources Planning for
Urban Water Supplies:
Mimi Jenkins, UCD Economic Reliability Analysis and
Institutional Uncertainties
Orit Wilchfort, UCD
The last decade has highlighted the many sources of
uncertainty in urban water supplies, and the many measures that
utilities can take to respond to actual and prospective
shortages. Water supply shortages are now seen as arising from
many causes, including drought, flood, earthquakes, growth in
water demands, and mandated operations for environmental
purposes. The occurrence of each cause is uncertain. True
system reliability estimates should include the most important
sources of uncertainty.
The last few droughts also have demonstrated the wide range of
measures that utilities can apply to mitigate or avoid water
shortages. These include traditional water supply augmentation
measures, long-term and short-term water conservation measures,
and an array of water transfer measures. Integrated reliability
analysis of urban water supplies should include the roles of
these measures as well.
The integrated analysis of complex water resource systems
requires the use of computer models. To demonstrate how such
analysis can be done, we have developed computer models of a
simplified East Bay Municipal Utility District system (Lund, et
al., 1995). A water supply "yield" simulation model
includes representation of the water supply system, streamflows,
reservoirs, and aqueducts along with system operating rules.
The results of running this model using a 72-year historical
streamflow record are a time-series of water availabilities or
water shortages. The supply system model is a fairly
conventional system "yield" model commonly used for planning and
operations studies, with the addition of pumping and treatment
costs. The resulting time series of shortages is then
transformed into a probability distribution of water shortages.
The system's response to shortages is modeled using an
optimization procedure called two-stage linear programming
(Lund, 1995). This approach allows for an integrated planning
of short-term and long-term water
conservation and water transfer measures to accommodate
shortages with a minimum average annual cost. The probability
distribution of water shortage from the supply yield model is an
input to this shortage response model. The results of this
model include shortage response plans as well as average annual
costs of responding to these shortages.
Together, these models of supply yield and shortage management
constitute an integrated model of system operations and
economics from a reliability perspective (Lund, et al., 1995).
The approach can be used for long-term planning of supply,
demand management, and water transfer measures
E-mail : jrlund@ucdavis.edu
Phone : (916) 752-5671
Richard E. Howitt, UCD The Effect of Water Markets on
the Cost of Water Reallocation
An undisputed outcome of the Bay Delta process will be
the reallocation of some quantity of surface water from
agriculture to Delta quality control. Central to the selection
of this quantity is the cost to
farmers and others of this reallocation. The results show that
the extent to which farmers and districts are able to trade
water, substantially effects the cost of the reallocations.
Comparing the impact of a 1 million acre foot
reallocation from the Delta Mendota and San Joaquin regions, a
reallocation based on fixed proportions of existing contracts
resulted on an average reduction of
$320.13 million in total revenue. Allowing a local market to
allocate the one million acre feet reduction within the San
Joaquin valley reduced the loss of total revenue by one third to
$195.4 million. The presentation
showed the impacts of a range of water reductions from 400
thousand acre feet to 1.3 million acre feet, with effects
measured in terms of labor loss and acres fallowed as well as
gross revenue.
E-mail : rhowitt@ucdavis.edu
Phone :
Bay Delta Monitoring Workshop
Monitoring and measurement needs for improved modeling
in the Bay-Delta watershed
Energy Commission Building
Sacramento
August 1, 1995
Organizers : Margaret Johnston (SFEI)
Lance Johnson (WWD)
Harlan Proctor (IEP) IEPŐs water quality program
Current long-term water quality monitoring is being conducted
under a program mandated by water right Decision 1485. This
Decision, issued by the State Water Resources Control Board in
1978, committed the Department of Water Resources and U.S.
Bureau of Reclamation to an extensive discrete sampling effort
supplemented by a net work of continuous, multi-parameter
recorders located throughout the Sacramento-San Joaquin Delta.
This program, however, was recently reviewed by an Interagency
Ecological Program committee and a proposed revision was
integrated with other IEP elements into a long-term IEP
Comprehensive Monitoring Plan. This Plan was adopted by the
SWRCB and included in their |Water Quality Control Plan (May,
1995). The IEP intends to petition the State Board to begin
implementing this new proposal in lieu of the 1978 mandate.
The basic concept of the IEP plan is a reduction of the
base-line discrete monitoring with more emphasis on the
continuous recorder network and expanded telemetry capabilities.
The resources gained by reducing the routine field sampling
could then be applied to an agenda of special projects that
would address specific problem areas.
E-mail :
Phone :
Mike Vassey (NOAA) Real-Time Environmental Monitoring
in the Upper San Francisco Bay
Estuary.
We report on two monitoring projects of significance to the
Bay Delta modeling community. The first is the National Ocean
Service (NOS) Partnership Project entitled ŇReal-Time
Environmental Monitoring in San Francisco and Suisun BaysÓ.
This was one of four national partnership projects funded by NOS
in 1994. The second is the NOS San Francisco Bay Demonstration
Project. This project is one of two national demonstration
projects initiated by NOS in 1995. The purpose of both projects
is to stimulate horizontal integratation of different divisions
and branches within NOS and to demonstrate the full range of
products and services available through NOS in cooperation with
local partners engaged in projects of national significance.
The Real-Time Environmental Monitoring Project was stimulated
by findings of the San Francisco Estuary Project recommending
the development of a real time monitoring system to track the X2
isohaline in the Suisun Bay region to manage flows for
ecological sustainability. After meeting with local agencies
and experts engaged in this issue, Dr. Kurt Hess and colleagues
of the Coastal and Estuarine Branch of OES designed a real time
monitoring system using one Acoustic Doppler Current Profiler
(ADCJP) and two pairs of Conductivity-Temperature (CT) sensors
that was intended to complement research currently underway by
US Geological Survey (USGS). The goal was to provide real time
data in the upper shallows of Grizzly Bay and Honker Bay to
correlate with data being collected in the main stem and
secondary channels of the Suisun Bay region by USGS. Data is
telemeter to a transmitting station on Benicia Bridge and then
piped to a communications facility at the California Maritime
Academy. It is available in real time by modem or a recorded
telephone message. This system went on line in January, 1995
and is still operational. Data has not yet been analysed in
conjunction with USGS.
John Klein and Karen Dennis of the Physical Environments
Characterization Branch of ORCA provided an analysis of
historical data to illustrate the historical context of salinity
variability in the Suisun region. DWR and USBR data from 1955
to l992 was broken down into dry, wet, and normal year-types and
analysed at hourly, daily, days to weeks, and monthly intervals
between high inflow conditions (February-April) and low inflow
conditions (July-September). They found that the most important
change in salinity structure is between normal and dry year
types during the high inflow season. The location of X2 is at
Port Chicago during normal years and 20 km to the east ear
Collinsville during dry years. Salinity variability at all time
scales is also more pronounced at Port Chicago during normal
years and these values correlate with data from grizzly Bay.
They also point out that there is very little shift of X2
between normal years and wet years, suggesting that increasing
flows has diminishing value for shifting X2 after a certain
threshold. They conclude that the dramatic shift in X2 between
dry and normal years could affect large areas of shallow
breeding habitat for threatened aquatic species and support the
need to obtain more real time data in these shallow areas to
correlate with values obtained from the main stem of Suisun Bay.
For further information, contact Michael Vasey of San Francisco
state University, the partner.
The San Francisco Bay Demonstration Project is designed to
build on to the real time monitoring system established in the
upper San Francisco Bay estuary but it shifts the emphasis from
environmental monitoring to a real item navigational monitoring
system that provides data on currents, water level, and
salinity. Three additional ADCPs are being installed at key
locations near Point Richmond, Oakland Harbor, and at the Golden
Gate. These will provide real time data to a communications
center that will be accessible by phone and modem. There is the
possibility that these stations will also provide environmental
data, such as salinity, temperature, and turbidity. Several
other NOAA efforts, including photogrammetric aerial photography
of the entire San Francisco Bay estuary shoreline, will also be
provided to the Bay Delta community. Installation operations
are currently underway. For further information, contact
Captain Richards, the Site Manager.
E-mail : mvassey@sfsu.edu
Phone : (415) 338-1957
Tom Gandesbery (SFBRWQCB) SFBRWQCB Monitoring
Current and future monitoring of San Francisco Bay waters
stems from the responsibility of the Regional Water Quality
Control Board to protect beneficial uses through its water
quality protection programs. Additionally, the law states that
in cases where objectives are not met, steps should betaken to
improve water quality. The ramifications of a recent court
decision invalidating the stateŐs numerical water quality
objectives for toxic pollutants in surface waters (also referred
to as state-wide plans) will be discussed.
Regional Monitoring of sediment water and tissue (bivalve
bioaccumulation) first began under the state Bay Protection and
Toxic Cleanup Program (BPTCP) in 1993 and later was assumed by
the San Francisco Estuary Institute via the San Francisco Bay
Regional Monitoring Program (RMP). The staffŐs efforts in the
area of regional monitoring will e discussed with the emphasis
on implementing bay monitoring stations which fill a dual role
of providing compliance information for a point source discharge
(e.g., sewage treatment plants) and data useful in a regional
context. Such an approach to monitoring may become more common
in the future. Several large dischargers are also required to
give partial support to local effects monitoring which have a
direct bearing on the effluent in question and the location and
characteristics of the discharge (e.g., studies of metals fate
near a sewage treatment outfall).
Areas of interest for surface water monitoring, and potential
numerical modeling, include: wasteload allocation, watershed
pollutants inputs and suspended sediment transport. Water
diversions in local watersheds and gathering comparable data on
effluent, ambient and background concentrations of elements will
also be important.
E-mail :
Phone :
Bruce Thompson (SFEI) San Francisco Estuary Regional
Monitoring Program for Trace Substances
The Regional Monitoring Program (RMP) began in 1993 through an
agreement between the SF Bay Regional Board and major
dischargers into the Estuary. RMP objectives include monitoring
the status and trends of contaminants in the Estuary, evaluating
compliance with water quality standards, and providing data for
other users.
Sampling and analyses are conducted through contracts. Samples
of Estuary water, sediment, and transplanted bivalve tissues are
collected 2 or 3 times a year from up to 24 stations located
between Coyote Creek in the South Bay, and the confluence of the
Sacramento and San Joaquin Rivers. Stations are generally
located in the main channels, but some stations are located at
the mouths of other major tributaries (Petaluma and Napa Rivers)
or in the shallows (Grizzly Bay).
Parameters measure include conventional water quality
parameters (salinity, suspended sediment, DOC, nutrients, etc.),
and sediment quality parameters (grain-size, TOC, sulfides), and
a full suite of contaminants in both water (dissolved and
particulate) and sediments (trace metals, chlorinated and
petroleum hydrocarbons, pesticides). Additionally, two aquatic
bioassays (mysids an larval bivalves), and two sediment
bioassays (amphipods and larval bivalves) area conducted twice
each year.
The RMP also includes several pilot studies (phytoplankton,
sediment transport, benthic invertebrates, tidal wetlands), and
several special studies (optimal sampling design, indicator
development, sediment toxicity). These additional studies are
often conducted in collaboration with researchers from other
agencies (USGS, DWR) and universities (UCD, UCB).
A descriptive Annual Report is produced by SFEI. However,
detailed analysis of relationships among variables or between
biological and contaminant measurements is not presented, as
only two years of RMP data have been collected so far. RMP data
could be used for model validation and testing. Data are
currently maintained at SFEI in an Oracle data base, and is
available upon request. Access to the data base will soon be
possible on the World Wide Web.
E-mail : brucet@sfei.org
Phone : (510) 231-9539
Larry Smith (USGS) USGS programs in the Bay and Delta
No abstract received.
E-mail :
Phone : (916) 979-2615
Bill Lowdermilk (DFG) San Joaquin River monitoring
No abstract received.
E-mail :
Phone :
Christopher Foe CRWQCB) Central Valley Regional Water
Quality Control Board Monitoring
Protection of water quality in the he Sacramento-San Joaquin
Delta Estuary is a high priority of the Central Valley Regional
Water Quality Control Board. Board staff hypothesize that
contaminant concentrations in the Estuary are the result of
either chemicals being transported into the Estuary on
mainstream Rivers or of releases from within the Estuary.
Ongoing monitoring programs are described for assessing aquatic
toxicity in the Sacramento and San Joaquin basins and in he
delta, for measuring metal concentrations entering the Estuary
the Sacramento watershed, and for measuring selenium, boron and
salt concentrations from the San Joaquin basin.
Aquatic toxicity Between 1988-90 the quality of all the major
types of water moving through both the Sacramento and San
Joaquin basins was screened with the EPA three species
freshwater bioassay test. Toxicity was identified in north
Valley reservoir releases, in storm runoff from both Cities and
agricultural areas and from releases of agricultural tailwater.
Follow-up studies by the U.S. Geological Survey, Department of
Pesticide Regulation and Regional Board are described which
determined the chemicals causing toxicity, their sources and
fate within the watershed. More recently, as part of the Bay
Protection Toxic cleanup program, the Regional Board in
cooperation with the University of California have begun to
collect and screen water samples for toxicity from the
freshwater portion of the Estuary.
Metals Metal water quality objectives are expected to be
adopted by the State Water resources Control Board as part of
the new Inland Surface Water Plan. Most earlier metal data was
not collected with clean techniques and is of questionable
value. Therefore, the Sacramento Regional County Sanitation
District the Sacramento County Water Agency and the city of
Sacramento have jointly established the Sacramento coordinated
Water Quality Monitoring Program (CMP). The primary objective
of the CMP is to collect high quality total recoverable and
dissolved metal data for the Sacramento and American Rivers.
Samples have been taken monthly since December 1993. In
addition, past rivering metal data has demonstrated that major
loads to the Estuary are associated with high flows of sediment
laden storm runoff water. The winter of 1995 was very wet. The
Regional Board has attempted to augment the CMP data set by
collecting daily during peak flows and twice weekly thereafter
water samples for metal analysis from both the Sacramento River
below the City of Sacramento and from the Yolo Bypass.
Selenium, boron and salt A monitoring program was established
in May 1985 to evaluate the effect of subsurface agricultural
drainage on water quality in both the Grasslands and San Joaquin
River. Grab samples are collected monthly from both the
Grasslands and San Joaquin River and are analyzed for a number
of constituents including salt, boron and selenium and
concentrations to assess compliance with water quality
objectives. In addition, a demonstration project is underway at
the Regional Board to assess the potential of using real time
monitoring to determine exceedances of the salt water quality
objective and the most cost effective method of increasing the
assimilative capacity of the San Joaquin River.
E-mail :
Phone : (916) 255-3113
Pat Coulston IEP) Workshop on Monitoring and
Measurement Needs for Improved
Modeling in The Bay-Delta Watershed
The Interagency Ecological Program for the Sacramento-San
Joaquin Estuary (IEP) is a 25-year old interagency research and
monitoring collaboration. The IEP, which now has nine member
agencies, has broad mission to evaluate the ecological resources
of the Estuary and ascertain he natural and human-caused
factors controlling the abundance and distribution of selected
fish and wildlife resources. The IEP presently is about a $12.0
million program with over 50 individual research and monitoring
elements. The subjects covered by these elements include the
distribution, abundance and survival of juvenile salmon in the
Estuary; studies of delta smelt life and splittail life history
characteristics and needs; indices of abundance for the various
life stages of key species such as striped bass sturgeon, and
bay shrimp; and the effectiveness of existing fish protective
facilities. The program is presently undergoing a comprehensive
evaluation and possible revision of activities, which will be
completed by the fall of 1995.
E-mail :
Phone : (209) 948-7800
Richard R. Johnston Anadromous Salmonid Monitoring in
The Upper Sacramento River
Monitoring of adult and juvenile chinook salmon Oncorhynchus
tshawytscha and steelhead trout O. mykiss in the upper
Sacramento River and its tributaries are described from Feather
River at river mile 80 to Keswick Dam at river mile 302.
Monitoring for other anadromous fish is briefly discussed. The
most prevalent technique for monitoring adult chinook among
twelve Sacramento River tributaries and the mainstem is the
snorkel survey, while juvenile monitoring most commonly employs
rotary screw traps of fyke nets.
E-mail :
Phone :
Joshua Collins (SFEI) A Working Model of Tidal Marsh
Development
A tidal marsh is a landscape that tends toward an average form
in dynamic equilibrium with regional changes in sediment and
water supplies, as affected by climate and land use. It is a
highly organized landscape with well-defined features that are
predictably distributed through space and over time. It is
almost equally maintained by abiotic and biotic processes that
vary in relative geomorphic influence along the tidal energy
gradient, useful proxies for which include tidal elevation or
distance from tidal source. The importance of biotic controls
increases with tidal elevation, distance upstream within
drainage networks, and distance away from channel banks. For
example, vertical accretion of the marsh plain away from
channels is dominated by peat production, whereas the form of
channels in cross-section and plan-view is controlled by tidal
deposition and scour of inorganic sediments.
Within the Unit Landscape, and under conditions of a rising
sea, the tidal marsh form is maintained through a compensatory
relationship between natural losses and gains in channel length.
Near the headward end of tidal channels, abiotic and biotic
controls fluctuate in dominance. Weak tides above Mean High
Water permit channel capture by vascular vegetation, resulting
in channel retrogression. Individual retrogression events are
incompetent to affect the cross-sectional area of the channel at
its tidal source. The tidal prism displaced from retrogressing
channels moves headward along the hydraulic gradient generated
by channel friction to other channels that consequently
elongate, such that the overall tidal prism and amount of
channels large and small are conserved.
The Unit Landscape Model predicts adjustments within a tidal
marsh drainage system of any process that alters the tidal
source. Local geomorphic controls are well understood.
Restricting the tidal source or moving it away from the interior
reaches of the system promotes retrogression, expansion of areas
that lack channels, and expansion of natural ponds on drainage
divides. Increasing the tidal source or moving it closer to the
interior reaches promotes headward erosion of channels and loss
of ponds.
Recent studies at SFEI have revealed that the plan form of
tidal marshlands varies predictably with aqueous salinity. For
example, tidal marshlands subject to strong springtime pulses of
freshwater have shorter channel networks and fewer but larger
ponds. These findings contribute to a template for tidal marsh
restoration that is sensitive to the influences of local and
regional salinity gradients.
However, large-scale geomorphic controls on tidal marsh form
and function are not well understood. For example, the response
of tidal marshland to major changes in the rate of sea level
rise cannot be predicted because it depends upon sediment
supply, and neither the source nor rate of sediment supply to
tidal marshland is understood at this time. Basic questions
remain unanswered. What is the suspended sediment load to tidal
marshlands and how does this vary with distance away from the
Golden Gate and upstream within subordinate estuaries, such as
the Petaluma River or Suisun Slough? How has sediment load
changed since the time of European contact in the region? What
proportions of the suspended sediment load come from local,
regional, or extra-regional sources? Answers to these questions
would tell us much about how the tidal marshlands are physically
connected to the rest of the estuary.
E-mail : collinsj@sfei.org
Phone : (510) 231-9539 Monitoring Workshop
The workshop participants broke into discussion groups after the
afternoon session. Each discussion group nominated a recorder
who was asked to summarize the deliberations of each group
during a plenary session at the end of the workshop. The
various groups were :
o Water Quality and Toxics
o Biological Modeling Needs
o Additional Monitoring And Measurement Needs For Bay-delta
Hydrodynamics
WATER QUALITY AND TOXICS GROUP
The group began the discussion by listing the types of models
that are being used for water quality and toxics modeling. They
include:
Hydrodynamic models that predict concentrations of dissolved
substance.
Geochemical fate and transport models.
Food chain transport of contaminants.
Biological effects of contaminants. These models require
input from hydrodynamic and fate and transport models.
Human health risk assessment models from exposure to
contaminants.
Then the group addressed three questions:
1. What improvements in current monitoring programs
would facilitate modeling?
The group felt that the current practice of monitoring monthly
or quarterly was conducted at to gross a time scale for
modeling which usually uses a daily, or less time step.
o Monitoring measurements collected monthly or quarterly could
provide information for calibration and verification of model
p redictions.
o Monitoring data from continuous recorders provides very good
information for models.
o Monitoring data such as that collected on pesticide pulses in
The Sacramento River by Foe and Kuivila also provides good
information for modelers.
o More empirical hydrodynamic data from monitoring was
identified as a priority.
o Information on the spatial distributions of contamination, for
example in Delta or river backwaters, or information on
point sources such as the North Bay Aqueduct would also be
useful.
o Monitoring the dispersion of dyes introduced to water would be
useful.
The development of conceptual models of Estuary hydrodynamics
would also be useful.
2. What do we need to do to get there?
Monitoring program managers need a better understanding of the
needs of modelers. Lists of parameters, time periods, etc.
could be provided to the Estuary monitoring programs so that
they can measure the appropriate state variables. For
biological models, knowledge of which contaminants and
concentrations are important in the Estuary, which species are
at risk, and which life stages are most impacted would help
modelers. Knowledge of population responses to contaminants
should also be conveyed to modelers.
3. How can modeling improve monitoring?
Knowledge of contaminant residence times or chemical half-life
would greatly facilitate interpretation of monitoring data.
Model output on contaminant fate, or illumination of geochemical
mechanisms would facilitate monitoring design and the
interpretation of monitoring data. Understanding the fate and
distribution of dissolved contaminants or nutrients can come
from hydrodynamics models; understanding the fate and
distribution of particulate-associated contaminants will come
from sediment transport models.
BIOLOGICAL MODELING NEEDS
At the August 1, 1995 session, the biologists were asked to
answer the following three questions:
1. What additional monitoring is needed?
2. How can existing programs be expanded/better coordinated to
get the needed data?
3. What data do modelers need to improve existing models?
1. What additional monitoring is needed?
The group had two thoughts with regard to this question. First,
that data needs vary, depending on the problem or answer you are
searching for, and therefore the data needs could be very
specific. The second thought was that monitoring needed to be
better designed so that the variance in individual parameters
and between parameters could be evaluated. This variance
questions needs to be settled before much in the way of useful
predictive models could be accomplished. In addition, the group
felt that the monitoring data collected should be useful in
establishing trend information (i.e. long time series) with the
most important result being able to detect changes between/among
organisms.
2. How can existing programs be expanded/better
coordinated to get the needed data?
Although we did not discuss the how question, the group felt
that there was significant opportunity to expand and better
coordinate data collection efforts. Although we did not
specifically discuss an ecological community monitoring program
specifically, there was discussion about better integrating the
data collection efforts.
3. What data do modelers need to improve existing
models?
The group felt that two major improvements were necessary.
First, that some type of real time monitoring effort was needed
to possibly support real time (predictive?) operations decision
making. This decision making model would have to integrate a
number of physical and biological parameters at a scale and time
step that has not been attempted in this system. Secondly, the
group felt that there must be a major effort to develop models
that mimic the behavior of organisms. For example, the particle
tracking model could be revised to reflect the behavior of
organisms when they are eggs and larvae and then be able to
shift and mimic the behavior of free swimming fry and in the
case of salmon smolts. This modification would eliminate a
number of invalid assumptions and provide decision makers with a
more realistic and predictive tool from which to work.
ADDITIONAL MONITORING AND MEASUREMENT NEEDS
FOR BAY-DELTA HYDRODYNAMICS
The Hydrodynamics breakout group focused on the question: What
are the additional monitoring and measurement needs for
hydrodynamic modeling in the Bay and Delta? Because the
participants were primarily interested in Delta modeling, more
emphasis was placed on Delta monitoring and measurement needs
(including Suisun Bay) than upstream tributaries or San
Francisco Bay needs.
The following measurement and monitoring needs were
identified. The order reflects the approximate priority that
the group assigned to these needs.
1. Measurement of key Delta flows
Accurate field measurements of key Delta flows are needed for
validating hydrodynamic models and modeling salinity and
biological system responses to flow. Of particular interest are
the Delta outflow at Chipps Island and other key regulatory
locations such as Rio Vista, Vernalis and the QWEST stations
(Jersey Point, Threemile Slough and Dutch Slough). The Delta
outflow is currently estimated from Delta inflows and exports
and estimates of Delta depletions without accounting for tidal
variations in channel storage. The ultrasonic velocity meter
program carried out by the USGS is helping to fill this need but
additional meters and additional funding is needed to ensure
this program continues.
2. Flow splits
The flow splits at key Delta locations may play an important
role in salinity and fish transport in the Delta. Measurements
are needed in the vicinity of the Delta Cross-channel and
Georgiana Slough, at Three Mile Slough, near the entrance to
Clifton Court and in New York Slough (near Pittsburg), head of
Old River, Turner Cut, Connection Slough and Dutch Slough.
3. Baldwin Ship Channel Impacts
The effect of proposed increases in the depth and width of the
shipping changes in San Pablo and Suisun Bay on salinity,
entrapment zone location, and flow distribution is presently
unknown. Numerical models can be used to simulate the effect of
proposed changes in San Francisco Bay bathymetry but detailed
field data will need to be collected both before and after
dredging to verify the modeling results.
4. Common Tide gage datum
Many of the existing tide gages are not tied to a common
vertical datum. Global Positioning System (GPS) survey data
should be used to determine the datum for each gage, e.g. NGDV
level. Accurate leveling of tide gages is important as small
errors in tidal boundary elevations can drive large volumes of
flow.
5. In-Delta consumptive use
Modeling of Delta flows is hampered by uncertainties in
estimates of the volume of water diverted to Delta islands,
returned from Delta islands, seeping onto Delta islands through
levees, evaporated from Delta channels and open-water areas or
lost by evapotranspiration by riparian vegetation. DWR has
recently completed a reanalysis of Delta island drainage and
consumptive use (Estimation of Delta Island Diversions and
Return Flow, DWR Division of Planning, February 1995). More
work is needed.
6. Delta precipitation data
The distribution of precipitation across the Delta is unknown
because of the limited number of published precipitation
records. This information is needed for modeling direct
rainfall onto Delta channels and open-water areas and for
calculations of Delta consumptive use. Additional
meteorological stations are needed throughout the Delta. Some
additional rainfall gages may already exist; these need to be
identified and located.
7. Channel geometry
Detailed measurements of channel cross-sections and general
channel bathymetry is needed to identify possible channel
constrictions and to allow the use of irregular cross-sections
and bottom slopes in hydrodynamic and salinity transport
modeling. DWR has begun this task as part of an IEP
Hydrodynamics project but the existing data are sparse. Some of
the data are out of date and of uncertain accuracy. More
hydrographic surveys are needed throughout the Bay and Delta.
It may be possible to use helicopter air-borne laser systems for
this task.
8. Rim flows
While detailed measurements of the main inflows to the Delta
from the Sacramento and San Joaquin Rivers are available, the
inflows from the smaller rivers and streams such as the east
side streams, and Yolo Bypass are known in sufficient detail or
accuracy.
9. Three-Dimensional data for boundary conditions for 3-D
(or 2-D) numerical models
Application of three-dimensional models to Suisun Bay and other
parts of the Bay/Delta system will enhance our knowledge of
salinity stratifications, gravitational circulation
(longitudinal and transverse) and the effect of bottom
bathymetry on mixing and circulation. However, detailed time
series of the cross-sectional variations in flow and salinity
are needed for use as boundary conditions for this type of
modeling effort.
WETLANDS DISCUSSION GROUP
The goal of this breakout session was to identify:
(1) what is the connection between wetlands and aquatic
ecosystems; and
(2) what should be monitored, potentially modeled and why.
For the purpose of this session, we opted to focus on tidally
influenced wetlands (as opposed to diked wetlands, for example)
adjacent to the San Francisco Bay estuary. These wetlands are
influenced by a tidally driven hydroperiod that correlates with
adjacent aquatic environments that are currently the focus of
Bay Delta modeling. Tidal wetlands are a source for aquatic
nutrients and a sink for suspended sediments and potentially
toxic runoff from adjacent uplands. A dendritic system of tidal
channels connects these wetlands to peripheral tidal mudflats
and deeper open water.
A large array of terrestrial, semi-aquatic, and aquatic
organisms inhabit or utilize tidal wetlands and adjacent
habitats. Although these wetlands are clearly important to
aquatic food webs, linkage between tidal wetlands and adjacent
aquatic ecosystems is not well documented and there are numerous
areas of this interface where careful long-term research is
needed.
It was generally agreed that there are several tidal wetland
elements that should be monitored to create a framework for
modeling the connection between tidal wetlands and aquatic
ecosystems. These can conveniently be divided between physical
and biological elements. Physical elements include parameters
that relate hydrodynamics with wetland geomorphology, e.g.
bathymetry of channel systems, velocity of channel flows,
hydroperiod relationship of marsh plain and marsh divides to
channels and ponds, field capacity, depth of water table,
sediment transport, patterns of erosion and deposition,
microtopography of tidal wetland landscapes, evaporation (mass
balance), dispersion coefficients, toxic loading, water quality,
and salinity. Biological elements include a careful selection
of organisms that utilize an array of tidal wetland habitats.
These include shorebirds, resident song birds, water fowl,
patterns of wetland vegetation, selected vascular plant species,
resident mammals, terrestrial invetebrates, benthic
invertebrates, and both juvenile and adult fish species. A
number of these organisms can be of value for assessing
bioaccumulation of toxics, evaluating habitat degradation,
comparing experimental wetland creation projects with reference
sites, and so on. More also needs to be known about energy
transfer in these systems and the contribution of wetlands to
aquatic food webs. Monitoring of physical and biological
elements should be done by on-site sampling and larger scale
photo interpretation of land forms and biotic patterns.
The consensus of the group was that strategic monitoring of
selected parameters drawn from this array of physical and
biological elements could be of value in developing
landscape-scaled models of the wetland and aquatic estuarine
interface. Such models could address issues such as the
functional value of tidal and wetland ecosystems to aquatic
ecosystem health, the optimal size and complexity of wetlands to
provide these functional values, wetland restoration needs and
potential projects, and management of endangered species
metapopulations. Ideally, these models will be developed on
geographic information systems that relate geographic features
to scientific data bases. These models might have predictive
value in understanding the potential impacts of different annual
and seasonal estuarine regimes (e.g. wet years versus dry years)
on the associated wetland biota, as well as many other analytic
applications. Consequently, management for the conservation of
target species, habitats, and biotic communities could be
enhanced by proactive planning and focused interventions based
upon insights provided by these models
------------------------------------------------------------------
Toxics and Water Quality Workshop
Lawrence Berkeley National Laboratory
Berkeley
August 25, 1995
Organizers : Nigel Quinn (LBNL
Margaret Johnston (SFEI)
Greg Gartrell (CCWD)
Nigel Quinn (LBNL/USBR) Welcome and introduction
The workshop was organized to systematically deal with issues
starting in the upper watersheds and then moving to the Delta
and San Francisco Bay. It was noted at the start of the
workshop that there is pitifully poor interaction at present
between water quality models that describe the river basins and
models that address similar issues in the Bay-Estuary. A
successful outcome of the workshop was considered to be a
clearer appreciation by attendees of how we might work towards
more comprehensive water quality and toxics models of the whole
upper watershed and Bay-Delta system.
Questions to be addressed at the workshop were the following :
(1) How do the upper watersheds affect water quality and
toxics loading in the Bay-Delta.
(2) What models are being used to better understand water
quality and toxics in the Bay-Delta.
(3) What models are being used to make decisions in the Bay-
Delta.
(4) What are the limitations of existing models and how can
they be improved.
E-mail : nquinn@mp405a.mp.usbr.gov
Phone : (916) 979-2325 ; (510) 486-7056
Ray Krone (UCD) Overview of hydraulic and sediment
transport modeling in the San
Francisco Bay system
The fishery, particularly the migratory fishery, has declined to
tragically low levels since 1943. The decline has been
attributed to overfishing, loss of spawning beds, and diversion
to water project pumps. There are large fish kills every year in
the Bay system, indicating that water quality is also a cause of
the decline. Migratory fish must pass through the Bay system to
reproduce, making their survival during this passge vital to
sustaining the fishery. Water quality is a function of waste
discharges, average residence time of the water, and transport
of sediment into and through the system. This seminar presents
descriptions of the roles of flows and sediments in maintanance
of water quality for the fishery.
Diversion of water from the 60,000 square mile drainage
through the Central Valley has reduced flows to the bays to a
fraction of their 1943 levels, and has changed the temporal
distribution of the flows from high winter and spring runoff and
very low summer flows to much lower winter and spring flows and
summer flows just sufficient to keep acceptable salinities in
the Delta. Water circulation in the estuaru between San Pablo
Bay and the Delta, the region where ocean and river water mix,
is very sensitive to fresh water outflow. Heavy industries line
this waterway and the removal of their wastes depends on water
circulation. This waterway was also the principal migration
route for fish.
Central Valley drainage is also the principal source of
suspended sediments that enter the Bay system. It supplied 76 %
of the suspended sediment in 1960, and by 1990 the supply had
dimished to 66 %. The total annual supply was reduced from 4.4
million to 2.6 million tons. Suspended sediment is important to
water quality because it adsorbs dissolved toxic substances and
carries them to the ocean or to deposits in the bays. It
provides a large assimilative capacity for wastes. Suspended
sediment also inhibits the penetration of sunlight, limiting
summer growths of algae.
Recognition of the importance of fresh water outflows to
management of water quality and to sustaining the fishery is
beginning to be apparent. Effective management will require the
development of predictive models that can be used to determine
optimum water resource operation and water qulaity for the
fishery.
E-mail : rkrone@ucd.edu
Phone : (916) 733-2555
Charlie Kratzer (USGS) Transport and aquatic toxicity of
Chris Foe, (CRWQCB) diazanon in California's Central Valley
Each January and February approximately a million pounds of
dormant orchard spray insecticides are applied to about half a
million acres of orchards in California's Central Valley to
control wood boring insects. Diazinon is the most commonly used
dormant spray insecticide. During 1991-94, diazinon was detected
in the Sacramento and San Joaquin Rivers following rainfall in
January and February. Three mechanisms are believed to be
responsible for the movement of diazinon from orchards to
waterways: surface runoff from sprayed orchards, drift during
application to nearby waterways and land with subsequent surface
runoff, and volatilization and scavenging of atmospheric
diazinon by rain. The first mechanism is believed to be
principally responsible for the pulses of diazinon observed in
Central Valley waterways after storms.
In a February 1993 study, the U.S. Geological Survey (USGS)
found high diazinon concentrations in creeks of the western San
Joaquin Valley, the Merced River, and the San Joaquin River near
Vernalis. Two distinct peaks in diazinon concentration occurred
at Vernalis. On the basis of estimated traveltimes, the first
peak was considered to be from westside sources. Although the
rising limb of the Merced River hydrograph was not sampled, a
high diazinon concentration on the falling limb (2.5 ug/l) and
the traveltimes make the Merced River the likely source of the
second Vernalis peak. However, the Tuolumne and Stanislaus
Rivers were sampled only once each during the storm hydrograph
and a high diazinon concentration on the falling limb could also
account for much of the Vernalis load.
In 1994, the USGS sampled the Merced, Tuolumne, and Stanislaus
Rivers and the San Joaquin River near Vernalis throughout the
hydrographs of a late January and an early February storm. In
both storms, the Tuolumne River had the highest concentrations
of diazinon and contributed the largest load to the San Joaquin
River. On the basis of frequent sampling at Vernalis and
estimated traveltimes, westside creeks were significant sources
early in the storms while the Merced River was later. The
Stanislaus River was a very small source (<5% of the Vernalis
load) in both storms. In all storm sampling, the USGS found that
diazinon concentrations are highly variable and require frequent
sampling to adequately describe the concentration curve and
estimate loads.
A pulse of diazinon was followed from the Sacramento River
at Sacramento to Martinez during the February 1993 storm. The
peak concentration at Sacramento was 0.39 ug/l and at Martinez
was 0.12 ug/l about a week later. During the same storm, a pulse
of diazinon was followed in the San Joaquin River from Vernalis
to Stockton. The peak concentration at Vernalis was 1.1 ug/l and
at Stockton was 0.81 ug/l about 2 days later. In both cases, as
the diazinon pulse moved seaward, the maximum concentration
decreased and the pulse dispersed over time, in part because of
tidal diffusion.
Results of 7-day bioassays indicate that Sacramento River
water at Rio Vista was acutely toxic to Ceriodaphnia dubia
(water flea) for 3 consecutive days and San Joaquin River water
at Vernalis for 12 consecutive days during the February 1993
storm. Bioassay mortality corresponded with the highest diazinon
concentrations at both sites, and diazinon appears to explain
much of the observed mortality. Daily diazinon concentrations in
the Sacramento River during this period were 0.19 to 0.28 ug/l;
the San Joaquin River concentrations were 0.15 to 1.1 ug/l.
E-mail : E-mail :
Phone : (916) 979-2615 Phone : (916) 255-3113
Curtis Oldenburg (LBNL) SELECT : A platform for
environmental remediation analysis
The SELECT project at Lawrence Berkeley National Laboratory is
developing software that integrates state-of-the-art models for
site characterization, exposure, health risk assessment,
remediation cost, and subsurface transport. SELECT brings the
best science to the decision maker, allowing the selection of
environmental remediation strategies that maximize health-risk
reduction and minimize cost. The framework being developed
will allow flexibility in the choice of models for each of the
components of the analysis. This flexibility will allow SELECT
to be applicable to a wide variety of environmental
contamination problems. SELECT will run on PCs and serve as a
platform from which the different components of a remediation
analysis can either be performed independently or in an
integrated manner. The design of SELECT emphasizes rigorous and
defensible analyses within each component. As such, SELECT is
designed to be used by specialists in the various component
fields. The framework ensures that the data, interpretations,
and results of analyses for each component will be available in
a common format for use by the other components of SELECT. The
complete analysis can then be reviewed by site managers to help
in decision making.
In the SELECT prototype, we have analyzed a case of TCE
contamination at McClellan AFB. The analysis includes a 3-D
transport simulation with LBNL's T2VOC multiphase transport
simulator. Concentration predictions calculated by T2VOC for
various remediation scenarios are passed to the CalTOX
multimedia exposure model to calculate exposure to nearby
residents. These predicted exposures over time give rise to a
potential carcinogenic risk due to exposure to TCE at the
residence. The cost of the various remediation scenarios is
compared with the predicted risk to show cost and risk as
decision making criteria. Current work includes analysing
uncertainty in the transport simulation and propagating this
uncertainty through the SELECT methodology.
E-mail : CMOldenburg@lbl.gov
Phone : (510) 486-7419
G. Fred Lee Water quality modeling issues in the Delta
(Fred Lee & Associates)
The Sacramento - San Joaquin River delta is experiencing serious
water quality degradation problems due to the input of chemical
constituents upstream and within the Delta. At this time, the
load response relationship for these various constituents -
problems are poorly understood. A review is presented of some
of the delta-related water quality problems that need attention.
These include the diazinon-caused toxicity to zooplankton and
the significance of zooplankton dealth for a several week period
each spring on the fish and aquatic life-related designated
beneficial uses of Delta waters.
Another important issue is the excessive fertilization of
Delta waters by nitrogen and phosphorus compounds from the Delta
watershed that cause excessive fertility in water supply
reservoirs that store delta waters. This excessive fertility
significantly degrades the raw water quality for municipalities
that use Delta waters as a water supply. There is need to
develop load response models that can be used to reliably assess
the potential impact of controlling nitrogen and/or phosphorus
inputs to the Delta on domestic water supply quality.
The relative significance of in-Delta vs. specific Delta
watershed sources of dissolved organic carbon as a precursor for
trihalomethanes should be determined. Of particular importance
is the relative significance of various types of DOC sources.
It may be possible through an understanding of the loads from
various types of sources of DOC to develop load response models
to relate the impact of controlling DOC from a particular source
on THM formation for water utilities that use Delta water as a
raw water source.
A discussion will also be presented on the importance of
focusing so-called water quality models on true water quality
issues. At this time, so-called water quality models are
typically chemical constituent models that do not reliably
assess water quality impacts. True water quality models must
incorporate the biological responses as the output from the
model.
E-mail : gfredlee@aol.com
Phone : (916) 753-9630
Lawrence Smith (USGS) Transport modeling in the south
Ed Gross (Stanford University) San Francisco Bay
Lower bounds during summer for mean residence times of
substances introduced at sewage discharge locations in South San
Francisco Bay are 65 to 95 days in the area south of the San
Mateo Bridge, of which 40 to 65 days are spent in the area south
of Dumbarton Bridge. Residence times are longest for the most
southern release sites, and release location accounts for most
of the reported range of values. The mean component of summer
wind and the spring-neap tidal cycle have secondary, but
demonstrable effects on residence times. Summer winds appear to
increase residence times relative to no wind, primarily in the
area south of Dumbarton Bridge. Residence times increase by a
few days during the part of the spring-neap tidal cycle that the
mean volume of South Bay is increasing, and subsequently
decrease by a few days as the mean volume of the Bay decreases.
These residence time estimates are the first estimates made
that include the effects of tides and wind in transporting these
substances through South Bay. They were made using a model
TRIM-2D, that actually calculates tides, tidal currents and the
resultant transport of dissolved substances. The residence time
estimates are lower bounds for metals residence times in South
Bay becuase they do not include the effects of biochemical or
geochemical filters that lengthen residence times of introduced
metals.
E-mail :
Phone : (916) 979-2615
Carl Chen (Systech) Modeling of copper in San
Francisco Bay
An existing two dimensional estuary model was modified to
incorporate processes important to the transport and fate of
copper in San Francisco Bay. These processes include advection,
dispersion, partitioning with suspended particles, settling and
resuspension of absorbed copper. A systematic calibration of
these processes was made. The model results matched
hydrodynamics (I.e. Tidal stages, time lag of slack waters and
currents), total dissolved solids, total suspended solids, total
copper, dissolved copper, and sediment copper. The model
predicted that a reduction of copper load in storm water runoff
which occurred in the winter months would lower copper
concentrations in the summer months. With current copper
loadings, the models simulated the more recent copper
concentration observed in the Bay, which was lower than the
those measured in historic past. This water quality improvement
was consistent with the source reduction programs implemented by
municipal treatment plants since the early 1980Ős. The treated
effluents may contribute inflows important to the recovery of
South Bay from previous contaminations.
E-mail :
Phone : (510) 335-1780
Susan Paulsen (CalTech) Use of tracers in transport
modeling
Understanding mixing or the distribution of salinity and other
water quality parameters within the Delta requires detailed
knowledge of water circulation patterns. The objective of this
work is to develop a method to resolve flow distribution and
water quality questions in surface waters using inductively
coupled plasma-mass spectrometry (ICP-MS), a technique capable
of measuring the concentrations of seventy-five elements in
aqueous samples rapidly and to very low detection limits; this
technique is used to fingerprint specific water sources and to
provide estimates of the fractions of various fingerprinted
waters in water samples containing a mixture of source waters.
In order to model or trace flows accurately in the
environment, several
conditions must be met:
o source smust have distinct chemical or elemental signatures
o signatures must not be altered to any
great extent by chemical and/or biological reactions or by
physical changes during mixing (I.e. Mixing should be
conservative or very nearly conservative);
o source concentrations must not vary significantly on timescales
shorter than the mixing timescales of a system.i
The use of ICP-MS enables rapid and accurate analysis so that
elements which meet these conditions can be easily identified.
Within the Delta, samples have been collected to begin to
determine which elements most nearly meet these conditions. It
has been determined that concentrations of many elements do vary
significantly between sources of water in the Delta, including
between the waters of the Sacramento River, the San Joaquin
River, and San Pablo Bay. Studies of mixing at the confluences
of freshwater rivers (confluences of the Sacramento/American and
the Sacramento/Feather Rivers)
have helped to determine which elements behave conservatively in
freshwater environments, and laboratory mixing studies of the
various end members have further helped to narrow the list of
tracer elements. Data collection is ongoing to determine
variations in elemental concentrations over both long and short
timescales. Preliminary work indicates that several elements
may meet these conditions, including Li, B, Na, Mg, K, Ca, Sc,
Br, Rb, Sr, Ba, Pr, and U. Additional study, including
characterization of the effects of particles and salinity
changes on element behavior during mixing, should allow accurate
determination of actual flow patterns and levels of mixing
within the Delta and may allow the determination of actual net
Delta outflow.
This technique has been successfully applied to several
simpler systems. The mixing ratio of two influent sources in
the effluent of a water treatment plant (Weymouth Treatment
Plant, MWD) has been accurately predicted, as has mixing in the
Napa River estuary. Additionally, elements have been added to
the inflow to a major southern California reservoir to determine
mixing rates and turnover times.
E-mail :
Phone : (818) 395-4348
Peter Zawislanski (LBNL) Selenium cycling in inter-tidal
zones of the Carquinez Strait
Selenium inputs to the North Bay consist of riverine-Se and
refinery-Se. The input of Se from the rivers, mostly the San
Joaquin, into the Delta, consists of primarily dissolved
selenate, some dissolved selenite and organo-Se, and minor
particulate-Se, at a total annual load of roughly 1000 kg.
Refineries contribute about 2500 kg Se per year, mostly as
selenite. Most of the Se entering the Bay from the Delta exits
via the Golden Gate. Based on Se concentrations on suspended
sediments and depositional rates of those sediments, about 300
kg of adsorbed Se is immobilized in Bay muds and marshes.
However, the details of Se immobilization are not well
understood. This especially applies to dissolved Se which is
taken up by aquatic organsisms or which may be reduced in
sediments in intertidal zones. Furthermore, the fate of Se
subsequent to immobilization in sediments is not well
understood. The rate at which Se is converted to elemental
forms is particularly important, because elemental Se is
unavailable to biological organisms. Our current understanding
of selenium cycling in intertidal sediments is as follows. Se
enters the system in both dissolved and adsorbed forms.
Adsorbed selenium drops out of suspension with solids; the
relative contribution of SPM-Se is proportional to sedimentation
rate at any given point, I.e., greatest in the subtidal zone,
intermediate in mudflats, and lowest in marshes. However, the
chance for resuspension is also lowest in the marsh, meaning
that SPM-SE which is deposited there is largely immobile.
Dissolved Se, on the other hand, requires contact with sediments
and higher organic matter in order to be immobilized.This
suggests that much if not most of the dissolved selenium which
enters the marsh becomes sequestered on the sediments. Once Se
enters the intertidal sediment system, its reduction to
elemental forms begins. We have observed elemental Se
comprising about 25% of the total Se in the top 2 cm of
sediment, and over 50% in the 10-20 cm deep layer of sediment.
The rest of the Se is primarily associated with organic matter,
with very little (<5%) being dissolved. This suggests that
reduction and adsorption of Se in the shallow sediments is
rapid. Preliminary redox measurements near the mudflat/martsh
boundary show that the shallow sediments, even at 2.5 cm, are at
a potential favoring elemental Se (~ -50mV). It is likely that
in the higher (more inland) areas of the marsh, soils are more
oxidized near the surface, due to a longer period of aeration in
between tides. Understanding the rates of Se immobilization
will greatly enhance the ability to model the movement of Se
through the Bay and its potential bioavailability.
E-mail : pzawislanski@lbl.gov
Phone : (510) 486-4157
Rainer Hoenicke (SFEI) Potential uses of trace
contaminant data in the development and
verification of pollutant transport models
A simple conceptual model is presented that can form the basis
for questions related to contaminant flux. Existing data may
already be sufficient to be used in biomagnification modeling,
forecasting recoevry time frames, and in turn influence
refinement of monitoring and research programs. Theoretical
models can also be tested with existing field data that can
yield insights into structural or parametization errors. Data
examples from the regional Monitoring program for Trace
Substances and a recent study on Contaminant levels in Fish
Tissue from San Francisco Bay are used to illustrate potential
uses in modeling recovery of specific ecosystem health
indicators.
E-mail : hoenicke@sfei.org
Phone : (510) 231-9539
------------------------------------------------------------------------
Statewide Operations Workshop
Secretary of State Building Auditorium
Sacramento
September 21-22, 1995
Organizers : George Barnes (DWR)
Harold Meyer (WRMI)
George Barnes (DWR) Introduction
No abstract received
E-mail : gbarnes@dop.water.ca.gov
Phone : (916) 653-5924
Dan Sheer (IEP) Overview of Operations Modeling
No abstract received
E-mail :
Phone :
Jeff Lefkoff (HCI) Optimization approach to
operations modeling
In contrast to purely physical models, such as hydrodynamic
and ground water models, operations models compute water
managernent decisions- This creates a fundamental challenge for
the models' logical structure. Decision making is a complex
process based on human judgement, a process that can be
represented only approximately using mathematical logic.
Calibration and verification is often not possible, since the
models are applied to presumed conditions that have no
associated historical record.
Operations model have been constructed using two alternative
mathematical approaches. The simulation approach utilizes
decision rules, checks, and adjustments to compute operational
decisions. The optimization approach uses formal algorithms to
maximize some operational objective. Neither approach can claim
to prescribe procedures that operators would actually follow
given the presumed conditions-both approaches can, however,
provide useful representations of the outcome of operational
decisions.
In this study, an optimization approach was taken to develop
an operations model of the State Water Project. The model uses
a non-linear optimization algorithm to compute the value of all
operational decisions during each time step, including reservoir
releases, Delta exports, and SWP deliveries. Decisions are made
that maximize long-term SWP yields while maintaining mass
balances and meeting regulatory standards and SWP contract
terms. The model user specifies upper and lower bounds on the
decision variables, which reflect facility capacities,
regulatory limits, or demands. Non-linear components of the
model include salinity standards in the Delta and the cutback
provisions of SWP delivery contracts.
To demonstrate model capabilities, a pair of model runs was
implemented to measure the increase in yield associated with
extension of the Folsom-South Canal to the California Aqueduct.
The effects of the new facility is measured as the difference in
project yield between the "with-facility" and 'without-facility"
model run. The only difference in implementation of the two
runs is the facility itself -- no adjustments to rule curve
parameters or other input data are needed. Preliminary results
indicate substantial increases in yield attributable to the new
facility.
E-mail : hcidvs@netcom.com
Phone : (916) 756-0925
Lenore Thomas (USBR) San Joaquin River Area Simulation
Huxley Madeheim (USBR) Model (SANJASM)
Ramona Swafford (USBR)
SANJASM is a computer model specific to the San Joaquin River
basin simulating the operations of the major reservoirs from the
headwaters of the San Joaquin River to the Cosumnes River.
Reservoir operating rules and demands for water for multiple
purposes can be varied to determine impacts throughout the river
system. It also accounts for imports to the San Joaquin River
due to deliveiies from the Delta Mendota Canal (CVP) and the
State Water Project (SWP). The model operates on a monthly time
increment for a simulation period which can be varied.
Presently the model simulates the period from l922 through 1992.
The model is capable of reproducing the operations of
Millerton Lake and Mendota Pool on the San Joaquin River;
Hensley Lake on the Fresno River, Buchanan Lake on the
Chowchilla River; Lake McClure (Now Exchequer) on the Merced
River; Lake Eleanor, Lake Lloyd, Hetchy Hetchy Reservoir and Don
Pedro Reservoir on the Tuolumne River; New Melones, Turloch and
Goodwin Reservoirs on the Stanislaus Rivcr; New Hogan Lakc on
the Calaveras River; and Camanche Reservoir on the Mokelumne
River.
Flood control operating rules are based on the Corps of
Fngineers flood control criteria when applicable.
Accretions and depletions not resulting from irrigation and M&I
uses are accounted for in a monthly time series data file.
Irrigation and M&I demands at reservoirs can be adjusted based
on deficiencies correlated to water year types or specified each
month in a time series file containing values developed outside
of the program. Return flows due to irrigation and M&I can be
based on year type and applied to applicable points within the
river system.
Ground water pumping demand can also be expressed as a
function of water year type. Pumping can be varied based on
deficiencies in reservoir deliveries or can be supplied
regardless of other water sources. Water supplies
provided by ground water pumping may be accounted for and
applied to applicable points within the river system. Releases
for fish are handled in a number of ways. In some cases, fish
releases are determined using program code specific to a certain
project; fish releases may also be tied to water year types or
specified in a monthly time series file. Water quality releases
from New Melones are computed based on the flow from Westside
sources and on the quantity and quality of water from other
sources above the confluence of the Stanislaus and San Joaquin
Rivers.
E-mail : E-mail : E-mail
Phone : (916) 979-2282 Phone : (916) 979-2281 Phone : (916) 979-2274
Derek Hilts (USBR) Overview of USBR's Planning Model of
the California State Water project and
Federal Central Valley Project, PROSIM
This presentation describes very briefly six aspects of
Reclamation's PROject SIMulation model (PROSIM). Following these
descriptions two additional items are discussed:
1. Recent comparisons with California Department of Water
Resources' planning model, DWRSIM.
2. Future enhancements to the PROSIM model.
The first aspect of PROSIM described is the model's general
characteristics - PROSIM is a rule and demand driven model of
the CVP and SWP. It works with monthly water volumes. It is a
conjunctive use tool to the extent groundwater pumping can be
set or determined internally (although groundwater heads are not
tracked). It is a tool, not a decision maker. It should be
used in a comparative, regional sense only. Like all computer
programs, it is subject to the Garbage In --> Garbage Out
principle.
The second aspect of PROSIM described is the geographic
coverage of the model. This description is divided into Water
Supply and Water Use. Water supply from all areas tributary to
the San Francisco Bay-Delta (Delta) is reflected in PROSIM.
Water usage in the Sacramento Basin, including the Delta, and
areas south of the Delta supplied by the CVP and SWP is
reflected in PROSIM.
The third aspect of PROSIM described is the physical (real
world) facilities constituting the system the model attempts to
represent. First a map depicting the real world facilities is
discussed. Then a schematic depicting PROSIM's representation
of those facilities is discussed.
The fourth aspect of PROSIM described is the input used by
PROSIM. A layout of the numerous input files feeding into
PROSIM is presented, followed by a brief discussion of a couple
of the files.
The fifth aspect of PROSIM described is the output from
PROSIM. A layout of the relationships between four
post-processors and the direct PROSIM output is presented,
followed by a brief example of output each post-processor
creates.
The sixth aspect of PROSIM described is the general
computational process the model uses in simulating the system
previously described. A top down approach is used, i.e., most
general description is followed by a more detailed description
of a part of the general description.
Two comparisons with DWRSIM are described. The first focuses
on the difference in overall water gain or loss in the
Sacramento Basin between the two models and the insignificance
of this difference. The second focuses on the difference in
estimated impacts of recent proposals for Delta standards using
the two models. Again, the insignificance of this difference is
discussed, particularly in comparison with recent historical
data.
Lastly, some short term and some long tem plans and/or desires
related to improving PROSIM are described. These include code
modifications to enhance PROSIM's representation of the real
world, spatially, as well as temporally. Also included is
coupling PROSIM with other models.
E-mail :
Phone : (916) 979-2279
Russ Brown (Jones & Stokes) Adjusted daily historical
operations simulation
No abstract received
E-mail :
Phone : (916) 737-3000
Lloyd Peterson (USBR) Monthly forecast model.
U.S.B.R. Mid-Pacific Regional Office's Central Valley Operations
Office's (CVO) monthly forecast model is applied at least
monthly to develop targets for reservoir operations (Trinity,
Shasta, Folsom, New Melones, and San Luis) and pumping
operations in the San Joaquin-Sacramento Delta. The format of
the model is a Lotus for Windows spreadsheet with a user
interface that allows for highly interactive calibration. Very
few rules for operation and no water quality relationships are
explicitly coded. The rules and constraints are applied by the
CVO staff as they observe resultant flows, flow indices, and
reservoir levels as shown by model graphics and tables. This
model is dependent on the skills of the water operations staff.
E-mail :
Phone : (916) 979-2196
George Barnes (DWR) Overview of planned DWRSIM enhancements
Sushil Aurora (DWR)
During the last few years it has been felt that DWRSIM model
capabilities have fallen short in meeting growing modeling
needs. The objectives for the model enhancements goal were set
forth as follows:
- Ability to operate non-CVP/SWP water systems
- Ability to incorporate evolving water management options
- More detailed and accurate CVP, SWP and Delta operations
- Improve model input/output access through data management
system
- Improve model availability and documentation of model
- Develop a more efficient, flexible, state-of-the-art
simulation model
In order to both prioritize the enhancements to DWRSIM and
better serve the users, a need assessment survey was carried out
by the project consultants, WRMI and David Ford. The following
list sums up the results of this survey:
ENHANCEMENT RANK
- Refine Carriage Water 1
- Sacramento Basin Conjunctive Use 2
- Expand San Joaquin River System 3
- CVP/SWP - one system operation 4
- Water Transfers from Sacramento Basin 5
- Front-end and Back-end Utilities 6
- Real-time SWP/CVP Delivery Deficiency 7
- Isolated Delta Transfer 8
- Develop Shorter Time Step Logic 9
- Stochastic Hydrologic Input 10
- Incorporate Yuba/Bullards Bar System 11
- Connect CVGSM with DWRSIM 12
- Rice Field Flooding Operations 13
- Replace HEC-3 with New Engine 14
- Expand CVP DMC and Joint Reach 15
- Feather River Service Area Deficiencies 16
- Delta Islands Storage Facility 17
- Streamline Delta Operations 18
In addition, the following enhancements were added to the
list by individual users to meet their specific needs.
- West Side Conveyance/storage facilities
- Folsom South Canal & EBMUD Div.
- Automate single/multiple-year at a time capability
- Optimize for power supply for CVP and SWP
- Smaller delivery location
- Operate San Luis as single res. with 2 users
- mprove ability to reoperate Sac. & SJV delivery
- Documentation
Some of the enhancements have already been completed and are
being incorporated into the model version to be released to the
public within a few months.
E-mail : sushil@dop.water.ca.gov
Phone : (916) 653-7921
Bill Smith (DWR) Input/output system.
No abstract received
E-mail : bsmith@dop.water.ca.gov
Phone : (916) 653-6079
Susan Lee (DWR) Merging STANSIM with the DWRSIM.
This presentation describes the current STANSIM model, the
purpose of the merge, and the features of this new STANSIM
model. The current STANSIM model is an independent model that
is used to simulate the Stanislaus River and the lower San
Joaquin River systems. This model was not dynamically linked to
DWRSIM model because contribution from the San Joaquin River
system was never required to meet any Delta standards. The
December 15 Water Quality Control Plan required that the San
Joaquin River contribute to meeting the X2 standard. In order
to meet this new X2 requirement, the STANSIM model was merged
into the DWRSIM model. The merged STANSIM retained the
capabilities of determining the actual flows to meet the
Stanislaus fish flow standards, the Vernalis pulse flow
standards, the Vernalis water quality standards. In additon,
two new features resulted from the merged STANSIM: the
determination of the X2 flow requirements and the SJR flow share
in meeting both pulse flow and X2 flow requirements.
E-mail : sue@water.ca.gov
Phone : (916) 653-6868
Robert Leaf (DWR) Annual Delivery Decisions in the
Simulation of the California State
Water Project and Federal Central Valley
Project using DWRSIM
This paper presents a new annual contractor delivery
determination procedure for the California Department of Water
Resources' State Water Project (SWP) and Central Valley Project
(CVP) system planning model: DWRSIM. The procedure attempts to
mimic the delivery decision process the department uses in
operating the SWP. The procedure uses the calendar year as the
delivery timeframe in which estimates of the annual delivery are
based on water system storage and runoff forecasts. Delivery
decisions are made on 1-January of each year and updated each
month until 1- May, at which time the delivery level is fixed
for the remaining seven months. Increasing "firmness" of
decisions is achieved by varying forecast exceedence levels and
using a "no-reduction" rule. Decisions are based explicitly on
runoff forecasts and storage levels, but implicitly incorporate
unknown factors that influence the decision process. These
factors may include: water quality and flow standards for, and
restrictions on exports from, the Sacramento-San Joaquin Delta;
water system facilities, level of hydrology; and the anticipated
error range of forecasts. Two relationships are used in making delivery
decisions: a user-defined Delivery versus Carryover Risk Curve,
and a Water Supply Index versus Demand Index relationship.
Using forecast data, incorporating Delta constraints and
standardizing the modeled decision processare distinct benefits
of this procedure.
E-mail : rleaf@dop.water.ca.gov
Phone : (916) 653-6868
David Ford (WRMI) New DWRSIM engine.
No abstract received
E-mail :
Phone :
Susan Lee (DWR) Delta transfer and in-Delta storage
facilities modeling in the DWRSIM.
This presentation describes the reason for modeling a Delta
Transfer facility and a In-Delta Storage facility in the DWRSIM
model and the current status of this new enhancement. As a
future planning tool, the Department of Water Resources would
like to see how these facilities would impact the Sacramento-San
Joaquin Delta. Currently, the only development that has been
done is designing the model network representation of the
conceptual diagram of the proposed facilities. Because the
operation rules for these facilities have not been determined,
actually implementation into the DWRSIM model has not started.
E-mail : sue@dop.water.ca.gov
Phone : (916) 653-6868
Devinder Sandhu (DWR) Expanded California Aqueduct and
Delta Mendota canal enhancements
Purpose of presentation was to inform DWRSIM model users and
potential users that the South of the Delta schematic that is
modelled has been revised to better represent the State Water
Project(SWP) and Central Valley Project(CVP). The Delta-Mendota
Canal (DMC) has been expanded to represent the different reaches
and the varying channel capacities, i.e. the channel capacity of
DMC varies from 4600 CFS in Upper DMC to 4200 CFS just above
O'Neil Forebay. A new feature is that CVP demands are broken
into user categories of Agriculture, Exchange, Refuge and Losses
; this feature provides the model users with more control in
adjusting CVP deliveries by being able to adjust a combination
of Agriculture, Municipal Industrial, Exchange, Refuge and
Losses.
The joint-reach of the California Aquedust is modelled to
allow modelling of potential water transfers. The channel
capacities of both CVP and SWP shares of the joint-reach are now
represented. The South Bay Aqueduct has been expanded and allows
user to better see the various users.
E-mail : sandhu@dop.water.ca.gov
Phone : (916) 653-9795
Ralph Finch (DWR) Carriage Water in DWRSIM
DWRSIM needs to relate Delta salinity standards as flow
values. The function that does this should be able to capture
import Delta dynamics, model future conditions, and be fast.
The current MDO routine is inccurate on individual days of
certain months, is not well documented, and is dffficult to
adapt to planned future conditions. Tberefore a replacement is
needed. A traditioual numerical Delta model such as DSM2 would
be far too slow to connect to DWRSIM. Therefore, a new
technology called Artificial Neural Networks (ANNS) is being
used to replace the MDO routine. ANNs are non-linear, blackbox
models that can capture Delta dynamics and are fairly fast.
For non-structural, operational changes (e.g. pumping), the
ANN can be calibrated on either historical or DSM2 output data.
For structural changes (e.g. barriers), the ANN must be
calibrated on DSM2 output with the planned changes implemented
in the DSM2 run.
To connect the ANN to DWRSIM, different networks will be
calibrated to produce salinities at the location specified in
tht standard (Dl485, Monterey Agreement, etc.), given current
and previous months' Sacramento and rim flows. A numerical
"wrapper" around the ANN inverts the problem and retums tht
current month's Sacramento flow to DWRSIM, given other variables
(past flows, and a set EC value from the standard).
Ideally, one would minimize the volume of Sacramento flows over
some time ptriod, while meeting all salinity requirements.
Currently no optimization is performed. Rather, the ANN systern
begins to meet the EC requimments looking forward 2-3 months in
advance, allowing time for flows to change gradually, and
avoiding abrupt, large changes in flows.
Carriage Water is the amount of water needed above an
increased export so as to not increase salinity in the Delta,
that is, it would require an increase in Net Delta Outflow to
hold salinity constant given an export increase. An alternate
term for Carriage Water could be the "Marginal Export Cost".
The ANN does not explicitly calculate Carriage Water. Instead,
when a final Sacrarnento flow is computed, one more call must be
made to the ANN with the Exports set to zero. Any change to MDO
is by definition Carriage Water.
E-mail : rfinch@dop.water.ca.gov
Phone : (916) 653-8268
Tariq Kadir (DWR) Linking a groundwater hydrology
model with DWRSIM.
Hydrology Development (HD) is the process of estimating Central
Valley future water supplies for areas tributary to the Delta
for use in DWRSIM planning studies. A basic assumption is that
the historical measured precipitation would occur in the same
pattern and quantity in simulated studies. The development
process modifies historical measured stream flows to reflect
future conditions. Three models are used as tools in the
development process: The Consumptive Use (CU) model, the
Depletion Analysis (DA) model, and the COMP model. The HD
process itself is structured and follows step by step
instructions for utilizing the CU, DA, and COMP models, area by
area, culminating in a set of IN's and YD's for use in DWRSIM.
It is a lengthy and complicated procedure and previously has
involved the manual manipulation of hundreds of data tables. A
new model, CUDACOMP, has been developed to simplify the process.
CUDACOMP is a FORTRAN program that incorporates the CU, DA, and
COMP models, and includes all the instructions for carrying out
a full hydrology development. The program will also modify an
existing DWRSIM main.dat file, updating it for any changes
resulting from the new HD process. Advantages of using CUDACOMP
include: streamlining the HD process, reducing human error, and
a quick turnaround in developing DWRSIM input data subject to
changes in land use (15 minutes on a 486 66 MHz PC, versus
approximately one week using the manual approach).
E-Mail : kadir@water.ca.gov
Phone : (916) 653-3513
Walter Bourez (WRMI) San Joaquin River system in
DWRSIM.
E-mail :
Phone :
Sushil Arora (DWR) Other DWRSIM new enhancements in
Bill Smith (DWR) progress.
E-mail : bsmith@dop.water.ca.gov
Phone : (916) 653-6079
Bill Smith (DWR) DWRSIM User Interface
The presentation covers three changs in the DWRSIM user
interface. The changes include a new, simpler, self documenting
format for user input job or simulation control type data, and
the use of the Corp of Engineer's Data Storage System (DSS)
software to handle time series based DWRSIM input-ouput. The
bulk of the presentation was spent describing the initial
release of a menu driven DWRSIM Output Analysis System that uses
the DSS software. The system will allow users to create tables
and plots of DWRSIM output data, both on screen and in hard
copy, with no knowledge of how the data is stored. Near future
additions will include additional types of tables and plots and
the ability to automatically generate tables and to make plots
of comparisions or differences between two or more DWRSIM
studies.
DWRSIM Availability
The presentation covers a number of new features that we are in
the process of adding to DWRSIM and r-he time schedule we are
working towards. The new features include:
A new simulation control input format
Merger of STANSIM model functionality
Enhanced California Aqueduct and Delta Mendota Canal schematic
New SWP/CVP delivery logic with forecasting
New hypertext - based documentation system
The availablity of the DWRSIM model and support programs was
also discussed.
E-mail : bsmith@dop.water.ca.gov
Phone : (916) 653-6079
----------------------------------------------------------------------
Drinking Water Quality Workshop
Energy Commission Hearing Room A
Sacramento
October 3, 1995
Organizers : Paul Hutton (DWR)
Francis Chung (DWR)
Paul Hutton (DWR) Welcome and introduction
Questions to be addressed in the Workshop are :
(1) What are the most important water quality issues in the Bay-
Delta.
(2) What models are available to evaluate Bay-Delta drinking
water quality issues.
(3) How can models be used to evaluate Bay-delta drinking
water quality issues.
E-mail : hutton@water.ca.gov
Phone : (916) 653-5666
Paul Maitski (SCVWD) Bay-Delta drinking water
quality issues
Bay-Delta Drinking Water Quality Issues are affected by two
items, drinking water quality regulations and source water
contaminants. Drinking water is regulated by use of primary and
secondary Maximum Contaminant Levels (MCLs), and treatment
techniques. Primary MCLs address health concerns (both acute
and chronic) while secondary MCLs address aesthetic concerns
such as taste and odor. The law that governs drinking water
quality regulations is the Safe Drinking Water Act which was
amended in 1986 to address a wide range of concerns. However,
in recent years, the focus has narrowed to balancing the
disinfection process to maximize public health protection from
both pathogens and disinfection byproducts (DBPs). To address
pathogens, the Surface Water Treatment Rule was promulgated that
made use of the multi-barrier approach filtration and
disinfection regulations. A regulatory negotiation process was
used to address DBPs which resulted in three rules: the
Disinfectants-Disinfection Byproduct (DDBP) Rule, Enhanced
Surface Water Treatment Rule (ESWTR), and the Information
Collection Rule (ICR). The Stage 1 DDBP Rule set MCLs for
Trihalomethanes (THMs), Haloacetic Acids, and Bromate of 80
parts per billion (ppb), 60 ppb, and 10 ppb, respectively. It
also established a treatment techniques of Total Organic Carbon
(TOC) removal prior to disinfection. The ESWTR will further
regulate pathogens, possibly including Cryptosporidium. Five
alternatives are being reviewed including log removals for each
treatment plant based on source water monitoring. The ICR will
require monitoring of source water quality and treatment plant
processes for pathogens and DBPs. This information will be used
to finalize the ESWTR and a Stage 2 DDBP Rule.
The contaminants of concern in the Delta are taste and odor
constituents, turbidity, pathogens, and DBP precursors. With
the current regulatory focus, the most important contaminants
for Delta modelers are the DBP precursors, specifically bromide
and TOC. Although pathogens are also very important, data is
limited and difficult to obtain. Bromide is a component of THMs
and other chloro/organic DBPs. The mixed chloro/bromo species
of THMs pose a greater health risk than chloroform or bromoform.
Bromide also forms bromate in the ozonation process (the primary
alternative to chlorination) which is a more potent carcinogen
than THMs. Bromide enters the Delta through seawater intrusion.
TOC is naturally present in all surface waters and is converted
to THMs and other DBPs in the chlorination process. A major
contributor of TOC in the Delta is agricultural return water
from flood irrigation of Delta islands that have high amounts of
peat soils.
E-mail :
Phone : (408) 265-2600
Rick Woodward (DWR) Municipal Water Quality
Investigations Program
The mission of the Municipal Water Quality Investigations
(MWQI) Program is to determine and evaluate the sources of
contaminants that affect the drinking water quality of the
Delta. The objectives of the program is to alert water agencies
about current and potential contaminants in Delta water
supplies, to assist water supply agencies in planning,
protecting, and improving drinking water sources and water
supply facilities, and to document water quality under a variety
of hydrologic conditions for studying water transfer
alternatives, water quality standards, and predictive modeling
capabilities.
The MWQI Program began in 1983 as the Delta Health Aspects
Monitoring Program. Under this program, 20 sampling sites,
including 17 Delta channels and 3 agricultural drains, were
routinely monitored to investigate sources of contaminants in
Delta waters. From the activities of this program, it was found
that 1) peat soils contain high concentrations of organic
compounds, which may serve as precursors for Trihalomethane
(THM) and other disinfection by-products (DBPs) formation during
drinking water treatment processes; 2) bromides enter the Delta
during episodes of saltwater intrusion and increase brominated
THM production; 3) pesticides and industrial chemicals are
detected infrequently in Delta water and when detected, are at
very low concentrations which do not exceed drinking water
standards; 4) sodium is rarely a problem in Delta export water;
and 5) selenium is barely measurable in Delta export water.
To further investigate the sources of organic compounds
from agricultural drainage, the program implemented an intensive
investigation entailing routine sampling of 54 agricultural
drains in 1987. The findings of the Delta Island Drainage
Investigation included: 1) Seasonal farming activities affect
the amount of organic matter that is carried off by drain water.
Drainage volumes are highest in the late fall and early winter
when fields are flooded to leach out salt accumulations, and
during the summer when irrigation is increased. 2) The
increases in dissolved
organic carbon (DOC) and THM precursor concentrations in the
Delta channel waters are mostly from drainage discharges. 3)
High DOC and THM formation potential levels are associated with
the organic content of the drained soils.
The MWQI Program recognizes the benefit of using models as
a tool used in predicting changes in water quality during
decision making processes. Models supported by the MWQI Program
include the Delta Trihalomethane Formation Potential Model, the
Delta Island Consumptive Use Model, and the Delta Simulation
Model. MWQI resources available for modeling support include
continued data collection and monitoring at selected sampling
sites in the Delta, development and maintenance of a database
for Delta water quality, and MWQI funds allocated directly for
modeling by the Division of Planning.
E-mail : rwoodward@water.ca.gov
Phone : (916) 327-1636
Francis Chung (DWR) Simulating DBP precursor transport
in the Sacramento-San Joaquin
Delta
This paper focuses on DWR's most recent efforts to simulate
disinfection by-product (DBP) precursor transport within the
Delta. Work to date has been limited to DOC transport. While
precursor monitoring has been conducted in the Delta for 15
years, only recently has the Department been collecting data at
short time intervals (e.g. daily) at key locations. More
frequent sampling, along with recent enhancements to the
Department's Delta Island Consumptive Use (DICU) model, have
provided an opportunity to validate DWR's hydrodynamics and
water quality model, DWRDSM. In this study, daily tides,
hydrology and water quality are provided as boundary conditions
to the model. The entire water year 1993 was selected as the
simulation period.
The purpose of this study is three-fold: (1) to validate the
DWRDSM against observed DOC data, (2) explain periodic
fluxuations in the observed concentrations of precursors at
drinking water diversions, and (3) demonstrate the use of the
model as a decision support tool for managing water quality in
the Bay-Delta. The model shows excellent validation at key
locations throughout the Delta.
E-mail : chung@water.ca.gov
Phone : (916) 653-5601
Ken Tanjii (UCD) Agricultural return water
quality model
Opportunities exist for real-time water quality modeling of
agricultural return flows from Delta Islands to serve as inputs
to hydrodynamic models. Net channel depletions are heavily
dependent on ]low well hydrology is simulated within the
islands. Moreover, drain water quality also depends on the
nature of the hydrologic model. Suggestions are made to extend
current soil moisture-ET based DICU Model by splitting siphon
diversions and channel seepage, routing @page into soil moisture
and drain ditch, adding a shallow ground water compartment, and
multi-layered rootzone. These extensions will provide a more
sound description of Delta Island hydrology. The constituents of
concern in return flows include EC, DOC, nitrates and
pesticides. Instead of assigning historical monthly
discharges, a recommendation is made to compute these quality
parameters on a real-time monthly basis. And since the
concentrations and mass emission rates of these quality
parameters are affected by their distribution in the soil
profiles, a suggestion is made to consider multi-layered soil
conditions. Substantial progress has been made in validating a
hydrosalinity mode) of the multilayered root zone. Water flows
in sub- and surface irrigated croplands are simulated based on
monthly reaching fractions (LF) in rootzone quartiles. The LF
is defined as the ratio of depths of' irrigation drainage to
effective crop irrigation. LFs ire calculated by assuming a
40-30-20- 10% water extraction in the quartiles to meet crop ET.
Soil and water salinity in terms of EC are calculated by a
mixing cell -evapoconcentration model. The degree of
evapoconcentration of soil water is the inverse of LF. The
hydrosalinity model is being validated by measured field data
from the corn salt tolerance study conducted on Terminous Island
in 1979-91 by the USSL and UC. The accumulation of salts in the
upper soil layers from subiriigation as well as in the lower
soil layers from sprinkler irrigation with irrigation water ECs
ranging from 0.2 to 6 dS/m have been partly validated. Modeling
salt leaching practices through leaching irrigations and
precipitation are well underway. The goal of real-time monthly
simulation of return flows and salinity appear to he imminently
achievable. Although a substantial data base on the sources and
sinks of DOC within Delta Islands are not available, DOC
modeling in root zone quartiles are being explored.
E-mail : kktanjii@ucdavis.edu
Phone : (916) 752-6540
Russ Brown General method for THM
(Jones & Stokes) prediction
Trihalomethane (THM) concentrations in treated drinking water
from the Delta depend on the export water quality (i.e dissolved
organic carbon [DOC] and bromide [Br]) and on the water
treatment methods used (e.g. chlorination). Analysis of DWRŐs
MWQI data from the Delta and several water quality experiments
conducted to support preparation of the Draft EIR/EIS for the
SWRCB on the proposed Delta Wetlands project has led to the
development and application of a monthly Delta drainage water
quality model (DeltaDWQ) that simulates concentrations of Br and
DOC in Delta agricultural drainage and in Delta exports. The
DOC concentrations in Delta drainage and exports depend on river
inflow DOC concentrations and source loading of DOC from
agricultural lands.
Comparison of MWQI measurements of trihalomethane formation
potential (THMFP) and MWDŐs simulated distribution system assays
(SDS) for THM concentration revealed some generalized
relationships between DOC concentration, chlorine dose, bromide
concentration, and the resulting THM concentrations. The
distribution of THM molecules can also be approximated from
probability arguments once the bromine incorporation factor (n)
is estimated. The chlorine dose (Cl2) relative to the DOC
concentration governs the amount of carbon that will be
incorporated in THM molecules (C-THM yield). A maximum yield of
about 2% DOC is observed. A half-saturation relationship is
assumed to describe the C-THM yield:
C-THM (ug/l) = DOC (mg/l) * 20 * Cl2 (mg/l)/DOC/[ 5 + Cl2/ DOC]
The bromine incorporation (n) , which is the average number of
halogen sites of the THM molecules containing bromine, is
determined using a similar half-saturation relationship with the
bromine saturation value, which is the ratio of bromine
molecules compared to halogen sites of the THM molecules:
Br saturation = Br (ug/l) / [C-THM (ug/l) * 20]
The maximum bromine incorporation is 3, with a half-saturation
value of 2. The concentration of THM is then:
THM (ug/l) = C-THM (ug/l) * (10 + 3.75 * n)
This general method for THM prediction provides a sound basis
for environmental impact assessment and planning studies where
the general patterns and effects of new facilities, modified
operations, alternative Delta water quality objectives, or
other management decisions require evaluation and comparison of
expected Br, DOC, and treated THM concentrations.
E-mail : none
Phone : (916) 737-3000
Leslie Palencia (MWDSC) SuperCAMP - California
Aqueduct Modeling Package
The Metropolitan Water District of Southern California (MWDSC),
in conjunction with the Harvey Mudd College Engineering Clinic
Program, developed a computer model to predict water quality
changes in the California Aqueduct. The computer model is
entitled SuperCAMP, which stands for Super California Aqueduct
Modeling Package. SuperCAMP is a stand-alone Microsoft Windows
Application. Currently, the model can predict the concentration
of a selected chemical at a specific time or location along the
Aqueduct, given the concentration of the chemical in the inputs
to the Aqueduct, and the locations of inputs and outputs along
the Aqueduct. To date, the model has been verified with
historical data for two time periods (January-April 1992 &
June-September 1991) and two constituents (sulfate and
electrical conductivity). MWDSC plans to continue testing the
model to evaluate effects of groundwater pump-ins to Aqueduct
water quality.
E-mail : lpalencia@mwd.dst.ca.us
Phone : (909) 392-2915
Zaid Chowdhury EPA WTP and WATERCO$T models
(Malcolm Pirnie)
no abstract submitted
E-mail : (602) 241-1770
Phone : chowduryz@aol.com
Paul Hutton (DWR) Using artificial neural networks to
predict trihalomethane formation potential
This paper focuses on DWR's most recent efforts to model THM
formation potential with the aid of artificial neural networks
(ANNs). The model presented in this paper will be used as a
post-processor for DWR's Delta hydrodynamics and water quality
model, DSM2.
The Stuttgart Neural Network Simulator, a public domain
computer code, was used in conjunction with a 900-point data set
(developed by the Metropolitan Water District of Southern
California) to train an ANN for predicting values of total THM
and bromine incorporation factor. The ANN was given input
values for bromide concentration, the product of dissolved
organic carbon concentration and ultraviolet absorbance at 254
nm, available chlorine dose defined as chlorine dose minus 7.6
times ammonia concentration as nitrogen, reaction time,
temperature andpH. A feed-forward ANN, with five neurons in the
first hidden layer and three neurons in the second hidden layer,
was trained to minimize the sum-of-squares error between
observed and predicted values. A log sigmoid function was
specified as the activation or transfer function. Prior to
training, input calibration data were log transformed. Output
calibration data were not log transformed. However, values of
total THM were scaled between 0.2 and 0.8 and values of bromine
incorporation factor were scaled between 0.01 and 0.8.
Model predictions for total THM formation and individual
species concentrations give superior predictions to those
produced by traditional statistical models, both in calibration
and validation. The model gives predictions within 10 percent
error, 40 to 70 percent of the time, and generally give
predictions within 30 percent, error 96 to 99 percent of the
time. The trained feed-forward networks were tested extensively
for sensitivity and demonstrate reasonable sensitivities over
wide ranges of input variables. It is recommended that this
model be incorporated into the EPA Water Treatment Plant model,
if this latter model is used for studying Delta drinking water
quality management alternatives.
E-mail : hutton@water.ca.gov
Phone : (916) 653-5666
---------------------------------------------------------------
Delta Modeling for End Users
Army Corps of Engineers Bay-Delta Model
Sausalito.
November 30, 1995
Organizers : Francis Chung (DWR)
Steve Monismith (Stanford University)
George Nichol (USCOE)
The theme of the workshop was one-dimensional modeling of the
Delta from the perspective of potential users (though the status
of multi- dimensional modeling efforts was also described). The
morning talks discussed the motivation for performing modeling
and described issues important to modeling, such as data
availability and peripheral utilities and functions. The
afternoon talks described the 1-D Delta models currently in use
or planned for use, and an effort was made to formally compare
1-D Delta models. Finally, the workshop closed with a
discussion of what end users would like in the models, and where
the models are now.
E-mail : chung@dop.water.ca.gov
Phone : (916) 653-5601
Francis Chung (DWR) Why Model?
Why model? Simulation modeling can answer questions about a
physical system that cannot be answered by field experiments or
historical data. Hydrodynamic modeling can answer questions
about flooding and low stages, channel velocities for scour,
sedimentation, and neutrally buoyant particles, and flow as a
transport mechanism for consumptive use and water quality
constituents.
Water quality modeling can address issues such as ocean and
land salts for salinity standards, THM precursors for drinking
water quality, and non-conservative constituents such as
temperature, nutrients, BOD, phytoplankton, and toxics for a
wide range of environmental and drinking water quality topics.
Estuarine biology can be dealt with using a particle tracking
model with either neutrally buoyant or active particles.
Although data and behavioral mechanisms are lacking, a particle
can help with real-time monitoring and designing biological
monitoring programs. Hydrodynamic modeling results can be used
in water and land quality
models, and in particle models; water quality results can be
used in particle models.
It is helpful to consider a model as a system, consisting of
input data (historic and generated), the model engine, and a
user interface which assists users in preparing input and
analyzing output. A weakness in any area will impair users'
ability to perform realistic, accurate, and timely studies.
E-mail : chung@dop.water.ca.gov
Phone : (916) 653-5601
Stephen Monismith Multi-Dimensional Modeling
(Stanford Univ)
No abstract submitted
E-mail : monismith@cive.stanford.edu
Phone : (415) 723-4764
Karl Jacobs (DWR) Data availability
Data is a key component of modeling and should have three
attributes: 1) easy to access, 2) easy to use and 3) accurate
enough to obtain the necessary results. The potential accuracy
of the models used to predict events in the Bay/Delta are
improving, resulting in the need to better understand the field
data used to run, calibrate and verify these models. Therefore,
modelers need to review the field data relative to the accuracy
required to produce the needed modeling results. Two ways to
complete this review are: 1)Reviewing QA/QC guidelines for the
data's collection and 2) by interviewing the data collector.
Within the Bay/Delta community the Interagency Ecological
Program has started making data available on the Internet
through utilization of a browser (URL wwwiep.water.ca.gov).
These data are easy to access, intuitive to use and information
describing how the data is collected QA/QC and a contact list is
made available to the user in the form of metadata sets. Data
is continuing to be uploaded to this file server. Data types
include fisheries, benthic, phytoplankton, zooplankton,
hydrodynamic, water quality and meteorological. Some data on
the server is the time series data used frequently by modelers.
To further support the modeling effort, time series data and
other data used to operate the forthcoming DWRDSM2 will be made
available on the server using HCCDSS. Agencies contributing
data to the server include USGS, USBR, DWR, DFG and USFWS.
In addition to the field and modeling data, a digital library
specializing in Bay/Delta technical reports, and information on
the IEP is also being prepared.
E-mail :
Phone :
Ralph Finch (DWR) Hydrologic models; User interfaces
Auxiliary issues such as data and user interfaces are important
to the overall modeling system. Currently DSM uses rectangular
channel cross-sections. Irregular-shaped cross-sections can give
better simulation accuracy and allow better results in studies
involving channel dredging or reshaping. The DWR has assembled
channel bathymetry data from several agencies and transformed it
to a common file format, horizontal coordinate system (UTM), and
vertical datum (NGVD). A Windows-based viewer was developed to
examine the data and draw cross-sections. An IEP group will be
helping to check the data and correct errors, and add needed
features to the viewer.
A delta model may be run in different modes of simulation. The
Historical mode uses historical (observed) input for flows and
boundary stage and salinity. For example, one could use data
from water year 1989 for a dry year simulation, and 1995 for a
wet year. The Generic or Plan mode uses representative inputs
for flows. For example, DWRSIM flow output could be used as
input to a Delta model to study effects at a certain level of
hydrology. Finally, the Steady-State mode uses fixed flows as
inputs. As an example, one could average all wet year types
together to arrive at an average wet year flow.
For the Generic and Steady-State modes, a synthetic downstream
boundary condition must be used at Martinez (in the case of
DSM). Currently the DWR uses a 19-year hourly average tide for
stage. For salinity the SALDIF2 program is used to calculate
daily average salinity at Martinez, given a Net Delta Outflow as
input. The daily salinity is partitioned to hourly salinity
values using Kristof coefficients, developed some years ago at
the US Bureau of Reclamation. Problems with these routines are:
the SALDIF2 finite difference method uses only 8 segments
between the Golden Gate and Sacramento; the procedures are not
well documented; they cannot account for effects of wind or
barometric pressure, and the 19-year mean tide is too simple for
some studies.
A proposed solution is to use a unified synthetic downstream
boundary generator. Stages would be first generated using
astronomical tides, then modified for flow, wind, and barometric
pressure with an artificial neural network (ANN) trained on
historical data. Salinity would be generated on an average
daily basis using a neural network with NDO as input, followed
by another neural network with the synthetic hourly stage as
input to generate hourly salinities.
Good model user interfaces are important. A good user
interface will:
- make the model accessible to both the novice and experienced
user
- prevent the user from misusing the model
- provide prompts for what input data is required
- not impose irrelevant burdens such as aligning data in
certain columns
- provide helpful run-time error diagnostics
- provide for graphical examination of output
As a result, the process of preparing model studies and
understanding output will be faster and less prone to errors.
E-mail : rfinch@dop.water.ca.gov
Phone : (916) 653-8268
Paul Hutton (DWR) Delta Island Diversions And
Returns
An important auxiliary ingredient to a successful Delta modeling
system is the ability to represent island diversion and return
flows and quality. Delta diversions and returns are important
for at least three reasons. First, they have significant
effects on water supply. During a typical "critical" water
year, net channel depletions (diversions minus returns) account
for a tenth of Delta water use. During a peak irrigation month
such as July, net channel depletions can account for a third of
Delta water use. A second reason diversions and returns are
important is that they have significant effects on drinking
water quality. Return flows are typically high in salinity and
dissolved organic carbon. Dissolved organic carbon is a
surrogate measure of humic materials that are precursors to
disinfection by- product formation in drinking water. A third
reason diversions and returns are important is that they impact
aquatic habitat. Channel diversions indirectly degrade aquatic
habitat through reduction in Delta outflow. Channel diversions
directly degrade aquatic habitat by entraining larvae and
juvenile fish. The significance of entrainment is poorly
understood but appears to be species specific.
Little directly measured data on Delta island diversions and
returns are available for input to hydrodynamics, transport and
particle tracking models. Therefore, one must rely on indirect
methods to characterize diversions and returns. DWR's Delta
Island Consumptive Use (DICU) model is driven with input data
for precipitation, land use, and pan evaporation. Assumptions
include, but are not limited to, soil moisture budgets, crop
evapotranspiration, irrigation and leaching schedules,
irrigation efficiencies, seepage and return quality. The
current version of DICU provides reasonable predictions as
evidenced by comparison with 1954-55 drain volume data. A
recent verification of DWRDSM's ability to simulate dissolved
organic carbon transport during water year 1993, using daily
changing hydrology and a real tide, strongly suggests that DICU
is providing reasonable input to the Delta model.
The DICU model is currently undergoing significant revisions.
Noteworthy projects include:
(1) development of a salinity and dissolved organic carbon
predictor through contract with U.C. Davis that will be
dynamically linked with the DSM2-QUAL module,
(2) adoption of a temperature-based method of predicting crop
evapotranspiration, and
(3) development of annually varying land use data for 142
different computational subareas.
Model revisions are scheduled for completion by July 1996. The
model will then be debugged and its link with DSM2-QUAL will be
tested. The new model will be released to the public upon
completion of this phase.
Current Delta Models
Don Smith (RMA) RMA Model
(no abstract provided)
E-mail :
Phone :
Greg Gartrell, CCWD Fischer Model Version 10
The historical development of the model was reviewed. The most
recent changes included: simplification of the code, multiple
components, and addition of a Boundary salinity generator
(Version 8), code revision that eliminates the "leakage" by
correcting the algoithm for channel junctions using a method
first proposed in 1986 (Version 9) and update of the geometry
with added accounting (Version 10). The original calibration
with real tides was reviewed, as well as the most recent
calibration. Problems inherent with Delta models were discussed.
These include data accuracy, the sparcity of data (spatially)
and especially the lack of flow data. Problems with phase
shifts in measured data were identified and discussed. The
accuracy of the time of measurements can significantly affect
comparisons. Finally, it was shown that inaccuracies in channel
parameters, including friction factors, can significantly
affect flows while not changing elevations in a discernible
fashion. The serious implication of this when flow data are
lacking was discussed.
E-mail : wrccwd@ccnet.com
Phone : (510) 688-8187
Parviz Nader (DWR) DSM-1 (An Overview)
The DSM (Delta Simulation Model) is a modified version of FDM
(Fischer Delta Model version 7E). FDM 7E was delivered to DWR in
1987. The package consisted of 2 separate models, one for
hydrodynamic simulations, and the other for water quality
studies. The package also came with a grid representation of
the Sacramento - San Joaquin Delta, which utilized 196 channels
and 152 junctions. The model had the capability of simulating
rectangular cross- sections, and prismatic channels (i.e. the
size of the cross- section did not change along the length of
the channel).
The FDM had been criticized for utilizing channel dimensions
which were somewhat different from those in the field. In 1989
it was decided to replace the grid representation of the Delta
with that of the DWR/RMA model, since it was closer to the real
Delta. The updated grid had 496 channels and 417 junctions. A
finer grid simply made it possible to better capture the changes
in the bathymetry.
The Delta is a very complex hydraulic system. It was soon
realized that in order to accurately represent the conditions in
the field, certain enhancements and modifications were
necessary. Starting from late 1988, a series of changes were
initiated by the staff in the Modeling support Branch at DWR.
These changes were designed to bring in more flexibility and
reduce the limitations of the existing code. The core of these
enhancements were developed from early 1989 to the middle of
1992.
A graphical package was developed called DGUI (Delta Graphical
User Interface) which made it possible to display both the DSM
computed and the observed data in a variety of ways. Various
types of observed data (e.g. stage, EC , salinity) has been
stored in the database. DGUI has proved to be an extremely
valuable tool, especially in calibration and verification of
both models.
Both the hydrodynamic and the water quality models were
calibrated and later verified on a Deltawide basis in 1990-91
using 1988 data. Parts of the historical data were used in
calibration, and the remaining parts were used in verification
of two models. Since then various attempts have been made to
verify the model using more recent data as they became
available.
E-mail : pnader@dop.water.ca.gov
Phone : (916) 653-5601
Francis Chung (DWR) DSM - 2
DSM2 is designed to be a replacement for DSM (Delta Simulation
Model). In 1992 a project was initiated to develop a new set of
models with capabilities surpassing those of DSM, with the goal
of having far less limitations, and a lot more flexibility.
After extensive review, two models were selected based on their
capabilities and superior numerical formulation. The most
importantfeatures offered by the new models are:
1- Irregular cross-sections and nonprismatic channels
2- elimination of the numerical leakage
3- More versatile flood modeling
4- Multiple conservative and non-conservative constituents
5- Baro-clinic term (density driven flow)
These two models have gone through major modifications in
order to incorporate any of required features not available in
their original versions.
In addition a new model was developed as a part of DSM2
called PTM (Particle Tracking Model), which can help determine
the fate and transport of the biomass. The model attempts to
include both the deterministic and the random motion of the
particles. The path taken by each particle is monitored and
recorded. Extensive effort has been placed to hook up all the
three models to a common input system, thus making it easier for
the users to learn to run them, and reduce the possibility of
having any errors.
Future plans include:
1- Development of the irregular channel cross-sections.
2- Calibration and verification of the models
3- Incorporation of THM formulation
4- Development of a Dynamic link with the Land Model
5- Introduction of Fish behavior in PTM
E-mail : chung@dop.water.ca.gov
Phone : (916) 653-5601
Ted Roeffs (USBR) The Peer Review Process
E-mail :
Phone : (916) 979-2278
--------------------------------------------------------------------
Biological Models Workshop
BIOLOGICAL MODELS: CURRENT STATUS AND POSSIBILITIES
Contra Costa Water District
December 12, 1995
Organizer : Wim Kimmerer, SFSU
OBJECTIVES
To describe the current status of modeling of biological issues
To clarify the capabilities and limitations of biological
models, particularly in contrast to models of estuarine physics
To determine where efforts might be most fruitful in the near
future
Wim Kimmerer, SFSU Introduction: goals for the day, and
what do we mean by modeling?
The first Bay-Delta Modeling Forum workshop on biological models
was held successfully in spite of travel problems imposed by the
season's first violent winter storm. Although attendance was
limited to about 45 people, interest was high and there was
considerable discussion of the workshop topic. The briefing
paper, prepared to stimulate discussion at the biological
modeling conference appears in the introduction on page 2 of
this document.
E-mail : kimmerer@mercury.sfsu.edu
Phone : (510) 525-9073
Randall L. Brown, DWR Why Managers Need Models?
For purposes of this discussion I define managers as those
individuals using information provided by their technical staff
to allocate or protect resources such as water and fish. In the
Department of Water Resources these individuals would typically
be division chiefs, deputy directors or the director. Managers
at these levels are normally not familiar with the details of
models and rely on their staff specialists for explanations
related to modeling assumptions as well as the models' strengths
and weaknesses.
I cite biological models to illustrate some of the points
being made. The points are equally valid for physical models.
The models themselves can be conceptual, empirical or
mechanistic; they all are useful to managers. Conceptual models
convey a qualitative understanding of how the system functions
and are generally an essential first step towards quantitative
models. Empirical models are often developed by exploring
several statistical techniques to link environmental
measurements (such as flow or concentration of a potential
toxicant) with the abundance of a particular organism.
Mechanistic models use biological and physical principles, often
in combination with field measurements, to quantitatively
describe variations in the physical and biological systems. In
all useful models, the key components are bright minds and good
environmental data bases.
Models allow managers and their staff to:
o Better understand the physical and biological systems.
Implicit in this, is an appreciation of any limitations in this
understanding. For example, a witness at a hearing before the
State Water Resources Control Board could describe a
conceptual model of the null zone in Suisun Bay. To the extent
that the witness provided a conceptual model that has wide
agreement in the technical community, the model can provide the
board, its staff, or other participants in the proceedings a
common understanding of an important physical feature of the
system - a feature that may be an essential component of their
ultimate decision. Unfortunately, there is often incomplete
agreement in the technical community and the next witness may
describe another view of how the null zone works. It is thus
important that the Modeling Forum and similar activities in the
Interagency Ecological Program provide a means through
which we achieve common understanding and consensus. To the
extent possible we need to limit the amount of arguing before
managers making critical decisions about the estuary.
o Play "what if" games to evaluate the relative benefits of
different management strategies. For example a chinook salmon
life cycle model could be employed to evaluate the relative
impacts of upstream and delta diversions and the ocean fishery
on the abundance of the four chinook races using Central Valley
streams. Synthesize large data sets into comparatively simple
models.
o Determine where important data sets are missing and
evaluate the need to design and conduct programs to collect
these data. Models allow managers to bring physical, chemical
and biological information into the political arena in a form
that can be used to accomplish a goal deemed desirable by a
diverse set of interest groups. For example the fish/X2 model
was a key component of the technical basis for the Delta accord
as was a conceptual model of mixing, circulation and particle
accumulation in Suisun Bay. The fish/X2 models and the
conceptual model of circulation and mixing are not definitive
and may not even be correct but managers were comfortable enough
with them to reach agreement on an interim solution for the
Bay/Delta. An important point is that the managers recognized
that they knew enough to make a decision. They also recognized
that there was more to learn and encouraged agencies staff to
obtain additional information.
Along with modeling results managers need an appreciation of
the strengths and weaknesses of the models and their results.
To this end modelers must:
o Provide a list of assumptions used in the model, including
some idea as to the validity of the assumptions.
o Provide an estimate of the confidence intervals around the
results.
o If appropriate include regression coefficient and
statistical significance; the significance level is useful in
assessing the model's statistical reliability and the
regression coefficient allows the manager an opportunity to
evaluate the chances that a particular action will achieve the
desired result.
o Indicate if it is appropriate to use the model in a
predictive sense. Many models can describe what happened with
reasonable accuracy but are less useful when they attempt to
predict the future - a future which is unlikely to have the
same environmental conditions occurring during model
development. For instance, environmental standards in
D1485 to protect striped bass were based on a flow/pumping
model. Unfortunately by 1978, the year in which Decision 1485
was issued, conditions in the estuary had changed to the extent
that expected benefits, in this case the summer abundance of
young striped bass, did not materialize under the new standards.
o Provide a balanced view of the models' strengths and
weaknesses. Do not be overly optimistic or pessimistic.
o Provide alternative models for observed phenomena. This is
especially important when the dependent
and independent variables both vary with some common factor
such as time. Dig deeper in the data set to convince yourself
and the manager that the results are not spurious.
o Indicate that models are most useful when the results are
considered in the context of the complete set of available
information, including best professional judgement (and even
common sense).
o The bottom line is that models are valuable tools for managers,
but more than model printouts are needed to
help make tough decisions.
o Validating and enhancing a physical habitat model of chinook
salmon pre-smolt production Sam Williamson, National
Biological Service
We have supplemented our existing stream habitat (PHABSIM) and
water temperature (SNTEMP) models with new capabilities. We
have also been predicting and validating stream habitat and fish
population conditions on a river segment- by-segment basis using
a weekly time step within one year for the purpose of better
annual flow scheduling and reservoir management planning. In
the conceptual model for the Trinity River, the perceived
importance of factors restricting the number of exiting pre-
smolts was: 1) physical habitat; 2) food abundance (stream
biological productivity); 3) biophysical requirements; 4)
reproduction requirements; 5) behavioral (crowding stress)
factors; and 6) cover availability. With concurrence of local
river biologists that helped build the conceptual model, the
computer model concentrates on physical habitat (with its
significant effects on movement, mortality, and fish food
production), water temperature (effects on mortality and
individual growth), and seasonal factors (effects on movement
and maturation); effects of food abundance, biophysical
requirements other than temperature, reproduction requirements,
behavioral factors, and cover availability were not represented
in the model. We used the model to design and improve young-of-
year population data collection efforts as well as probe
hypotheses that physical habitat significantly influences
movement, growth, and mortality of salmonid fishes, but have not
been able to judge relative severity of limiting factors. We
are continuing a model validation and enhancement process to
evaluate numerous proposed flow schedules based on surveyed
number of spawned-out females and simulated egg/alevin, fry, and
pre-smolt growth, movement and mortality as affected by measured
or anticipated flows and water temperatures.
E-mail :
Phone :
Loo Botsford, Population modeling of
(UC Davis) threatened species in the Central Valley
and Columbia watersheds
Both empirical and mechanistic models have limitations. Because
of problems such as intra-series correlation and multiple tests,
empirical models inevitably exhaust the degrees of freedom
available. On the other hand, there is never enough data to
completely parameterize mechanistic models. Because
anthropogenic impacts on the delta over the past 50 years were
not designed as a statistical experiment, we will not be able to
obtain statistically concrete results, nor draw certain
conclusions. Managers must realize this. With regard to
Columbia River salmon, two types of mechanistic models have
evolved, one obtains distributions of cohort survival from the
responses of individuals to hydrodynamics and hydrology, and the
other estimates population persistence from those
distributions. This two-pronged approaches seem appropriate in
the bay/delta. A promising recent development on the Columbia
is a shift in focus from model outcomes to identifying the
appropriate model structure. Maintaining an emphasis on the
empirical basis of model relationships seems like a good idea
for the Sacramento-San Joaquin system.
E-mail :
Phone :
Jim Cowan, Univ. of South Alabama Flow, water management
Kenny Rose, Oak Ridge National Lab. and recruitment of striped bass in the
Chris Enright, DWR Sacramento-San Joaquin River Delta, CA
Numbers of striped bass, Morone saxatilis, in the estuarine
Sacramento-SanJoaquin River Delta have declined since the mid-
1970's to low levels in recent years. Water management practices
and drought, toxins, and the introduction of exotic species all
have been implicated in the decline, although none are without
detractors. We have developed a bioenergetically-driven,
individual-based population model of striped bass that includes
site-specific values for a suite of model inputs configured to
represent feeding and growth environments in two spawning and
three larval and juvenile nursery locations within the Delta
estuary. Baseline environmental inputs and transport of eggs
and larvae within the estuary are modified via the history of
introduction of exotics, and by flow in simulations using a 2-D
hydrodynamic model of Delta flow, to realistically represent
water management practices, changes in zooplankton community
dynamics and climatological trends over the period of striped
bass decline. Results suggest that larvae and juveniles in the
estuary are food limited, and that probability of recruitment
success is highest when flow favors rapid transport of larvae
into richer nursery grounds in Suisun Bay. Infrequent wet years
since the late 1970's and changes in water management practices,
e.g., large diversions of water from the estuary, have
contributed to the striped bass decline.
E-mail :
Phone :
Jim Quinn, U.C. Davis Tracking striped bass
distributions through the delta using a model
An interdisciplinary modeling group in the Center for
Ecological Health Research at the University of California,
Davis, is developing a model for simulating population processes
of species in the Bay and Delta. Flows and localwater quality
are treated with finite elements models developed in part by
IanKing, Gerald Orlob and John DeGeorge, from the Department of
Civil Engineering. Depending upon location, the grid may be
one-, two- or three-dimensional. Currently, the model's
geographic coverage extends from the Golden Gate up the
Sacramento River system to many of the large dams. Calibration
continues using historical flow patterns from several years in
the 1980s and temperature and salinity records.
Superimposed on the hydrodynamic model, we have developed a
particle tracking model. Initially, we envision the particles
as representing individuals or cohorts of striped bass from egg
release through approximately 14 mm feeding larvae, although
the formulation is potentially applicable to many species
resident in the water column. Growth, development rate, and
mortality are predicted as a function of salinity, temperature,
and locations from empirically estimated parameters and
functional forms. The model can treat simple differences in
transport resulting from animal behavior, such as vertical
location in the water column, and can potentially treat complex
interactions among environmental influences, such as increased
predation resulting from increased development time caused
pollutant effects. The striped bass model is being calibrated
and parameterized using historical data from Fish and Game
monitoring efforts, which can be used to estimate such factors
as residence time in the delta under varying flow regimes.
The model differs from the individually based model presented
by Jim Cowan in several ways. It treats less detail in the
biology of individual fish, but is much more explicit
spatially. Demographic rates are estimated from field
measurements and correlations with environmental forcing
functions (flow, temperature, etc.) rather can derived from an
energetics calculation. At this point, it allows less direct
interaction with waterborne constituents (e.g., depletion of
planktonic food organisms.) Nevertheless, we believe that the
model and associated visualization tools will be useful to
managers in understanding the consequences of flow and water
quality on fish recruitment.
E-mail :
Phone :
Wim Kimmerer (SFSU) Empirical modeling
Empirical models are constructed statistically from data, using
techniques such as regression or analysis of variance. Although
such models are driven by data, they embody a substantial amount
of knowledge in terms of the independent variables used, form of
transforms or link functions, and treatment of outliers.
Empirical models attempt to fit a line or surface to data.
Outliers can be treated either explicitly by removing them on
the basis of information about the nature of these data, or
implicitly using robust statistical methods. The form of the
function used to fit the data is largely driven by underlying
knowledge as well, although there are functions that have little
or no information content (e.g. spline curves). Log transforms
or link functions are often used for biological data, because
biological processes tend to be exponential; log functions
provide straight-line fits and also homogenize the variance of
many biological relationships.
Advantages of empirical models are:
1) They do not require much knowledge of mechanisms; and
2) They can be developed from monitoring data.
Empirical models are limited in that:
1) History may not be a guide to the future;
2) They apply only within the range of data; and
3) They provide little insight into mechanisms.
Examples of empirical models were presented, focusing on the
"fish-X2" relationships underlying the December 1994 agreement.
E-mail : kimmerer@mercury.sfsu.edu
Phone : (510) 525-9073
Panel Discussion
A discussion was held by the entire group of speakers with
participation by the audience. The following general themes
emerged from this discussion and from the talks themselves.
1. Strongly predictive biological models are still a long way
off, and for purposes of predicting effects of some
management actions, simple (e.g. empirical) models may be as
useful as more complex mechanistic models. Decisions must be
made in the presence of uncertainty.
2. There is a real value in having alternative models
attempting to describe the same phenomenon using different
concepts and input data. To the extent that alternative models
produce the same result, that result will be more robust than a
similar result from a single model. When models produce
different results, it is incumbent on modelers and
stakeholders to find out why.
3. An essential ingredient of a successful smulation modeling
program is a commitment to the long term, and an integration
of modelng with data collection and research. In any new
program, modeling should ideally begin before data
collection, because of the long time to develop new models, and
because model development identifies data gaps. In an
integrated program, management can help in modeling by providing
contrast in the data.
4. In many modeling exercises, model development may be more
valuable than the finished product interms of advancing the
state of knowledge.
5. A wide variety of models is being used in the Bay-Delta-
river system. One class of models missing from this suite is
the broad class of trophic-dynamic models. However, pursuing
this class of models may not prove fruitful because the
ecosystem of San Francisco Bay, in contrast to other aquatic
ecosystems, does not seem to be strongly driven by the supply of
nutrients. More fruitful avenues may be species-specific models
that focus on limiting factors for each species.
6. Model development can take advantage of recent developments
including: 1) the use of the Internet either to make model
output available or to provide access to the models themselves;
2) data assimilation methods to force models to track newly
developing data; and 3) the use of increased computing power on
the average desktop to make the more complex models available
to anybody (e.g. particle tracking or individual-based models).
*********************