EPA 903-R-04-004
July 2004
United States
Environmental
Protection Agency
Region III U.S. Army Corps of Engineers
Chesapeake Bay Engineer Research & Development Center
Program Office Environmental Laboratory
f
\
US Army Corps
of Engineers.
Engineer Research and
Development Center
The 2002
Chesapeake Bay
Eutrophication Model
Carl F. Cerco
Mark R. Noel
July 2004
-------
EPA903-R-04-004
July 2004
The 2002
Chesapeake Bay
Eutrophication Model
CARL F. CERCO and MARK R. NOEL
U. S. Army Corps of Engineers
Waterways Experiment Station
3909 Halls Ferry Road
Vicksburg, MS 39180
Prepared for: Chesapeake Bay Program Office
U.S. Environmental Protection Agency
410 Severn Avenue
Annapolis, MD 21401
This work was accomplished by ERDC through a cooperative agreement with
the U.S. Environmental Protection Agency Chesapeake Bay Program Office,
which was administered by the Baltimore District U. S. Army Corps of Engineers.
-------
Contents
Executive Summary vi
Introduction vi
Coupling with the Hydrodynamic Model ' vii
Boundary Conditions viii
Hydrology and Loads viii
Kinetics xi
Format of Model-Data Comparisons xiv
Zooplankton xv
Effects of Predation and Respiration on Primary Production xvi
Process-Based Primary Production Model xvii
Suspended Solids and Light Attenuation xviii
Tributary Dissolved Oxygen xix
Modeling Processes at the Sediment-Water xix
Dissolved Phosphate xxi
Statistical Summary of Calibration xxii
1 Introduction 1
The 2002 Chesapeake Bay Environmental Model Package 3
Expert Panels 4
This Report 5
Bibliography 5
2 Coupling with the Hydrodynamic Model
Introduction 7
The Hydrodynamic Model 7
Linkage to the Water Quality Model 10
Vertical Diffusion 11
References 20
-------
Chapter 4 • Hydrology and Loads
3 Boundary Conditions 22
Introduction 22
River Inflows 22
Lateral Inflows 25
Ocean Boundary Conditions 26
A Recommendation 37
References 37
4 Hydrology and Loads 38
Hydrology 38
Loads 42
Nonpoint-Source Loads 42
Point-Source Loads 49
Atmospheric Loads 60
Bank Loads 62
Wetlands Loads 71
Summary of All Loads 77
References 80
5 Linking in the Loads 81
Introduction 81
Nonpoint-Source Loads 81
Point-Source Loads 89
Other Loads 89
References 90
6 Water Quality Model Formulation 91
Introduction 91
Conservation of Mass Equation 91
State Variables 92
Algae 95
Organic Carbon 105
Phosphorus 108
Nitrogen 112
Silica 116
Chemical Oxygen Demand 117
Dissolved Oxygen 118
Temperature 120
Inorganic (Fixed) Solids 121
Salinity 121
Parameter Values 121
References 126
7 Introduction to the Calibration 128
The Monitoring Data Base 128
Comparison with the Model 131
References 138
-------
Chapter 4 • Hydrology and Loads
8 The Zooplankton Model 139
Introduction 139
Model Conceptualization 139
Zooplankton Kinetics 142
Interfacing with the Eutrophication Model 145
Parameter Evaluation 148
Observations 157
Model Results 157
Recommendations for Improvement 169
References 170
9 Analysis of Predation and Respiration on Primary Production .. 172
The N-P-Z Model 172
Basic Parameter Set 173
Phytoplankton with Respiration Only 174
Phytoplankton with Zooplankton 176
Phytoplankton with Quadratic Predation 178
10 Process-Based Primary Production Modeling in
Chesapeake Bay 180
Introduction 180
Chesapeake Bay 181
The Chesapeake Bay Environmental Model Package (CBEMP) 183
Data Bases 183
Model Formulation 186
Primary Production Equations 192
Parameter Evaluation 193
Results 199
Discussion 209
In Conclusion 215
References 216
11 Suspended Solids and Light Attenuation 220
Data Bases 220
The Light Attenuation Model 225
Solids Settling Velocities 228
Model Results 231
References 246
12 Tributary Dissolved Oxygen 247
Introduction 247
Dissolved Organic Carbon Mineralization Rate 250
Wetland Dissolved Oxygen Uptake 252
Nonpoint-Source Carbon Loads 252
Spatially-Varying Dissolved Organic Carbon Mineralization Rate 255
Reaeration Rate 258
Sensitivity to G3 Carbon 258
Sensitivity to Algal Predation Rate 260
- IV-
-------
Chapter 4 • Hydrology and Loads
Sensitivity to Vertical Diffusion Coefficient 262
Discussion 265
References 274
13 Modeling Processes at the Sediment-Water Interface 275
Introduction 275
Coupling With the Sediment Diagenesis Model 279
Parameter Specification 281
Sediment Model Results 283
Benthos Model Results 301
Results of the Submerged Aquatic Vegetation (SAV) Model 304
Results from the Benthic Algal Model 307
References 319
14 Dissolved Phosphate 321
Introduction 321
Dissolved Organic Phosphorus Mineralization 322
Sulfide Oxidizing Bacteria 323
Precipitation 324
Summary 326
References 330
15 Statistical Summary of Calibration 331
Introduction 331
Methods 331
Statistics of Present Calibration 333
Statistics of Model Improvements 335
Comparison with Other Applications 338
Graphical Performance Summaries 340
References 349
-v-
-------
Executive Summary
Introduction
Deterioration of water quality in Chesapeake Bay and associated losses of living
resources have been recognized as a problem for more than twenty years. An order-
of-magnitude increase in anoxic volume and a catastrophic decline in submerged
aquatic vegetation (SAV) were among the primary problems cited. Two decades
later, elimination of anoxia and restoration of SAV remain prime management
goals. Models have been employed as tools to guide management since the forma-
tion of the first water quality targets. Over time, as management focus has been
refined, models have been improved to provide appropriate, up-to-date guidance.
The Chesapeake 2000 Agreement called for a ten-fold increase in biomass of
oysters and other filter feeding organisms. At the same time, regulatory forces were
shaping the direction of management efforts. Regulatory agencies in Maryland
listed the state's portion of Chesapeake Bay as "impaired." The US Environmental
Protection Agency added bay waters within Virginia to the impaired list. Settlement
of a lawsuit required development of a Total Maximum Daily Load (TMDL) for
Virginia waters by 2011. To avoid imposition of an arbitrary TMDL, the Chesa-
peake 2000 Agreement specified removal of water quality impairments by 2010.
Impairments in the bay were defined as low dissolved oxygen, excessive chloro-
phyll concentration and diminished water clarity. A model recalibration was
undertaken, with emphasis on improved accuracy in the computation of the three
key indicators.
The Chesapeake Bay Environmental Model Package
Three models are at the heart of the Chesapeake Bay Environmental Model
Package (CBEMP). Distributed flows and loads from the watershed are computed
with a highly-modified version of the HSPF model. Nutrient and solids loads are
computed on a daily basis for 94 sub-watersheds of the 166,000 km2 Chesapeake
- VI -
-------
Executive Summary
Bay watershed. The CH3D-WES hydrodynamic model computes three-dimensional
intra-tidal transport on a grid of 13,000 cells. Computed loads and transport are
input to the CE-QUAL-ICM eutrophication model which computes algal biomass,
nutrient cycling, and dissolved oxygen, as well as numerous additional constituents
and processes. The eutrophication model incorporates a predictive sediment diagen-
esis component. Ten years, 1985-1994, are simulated continuously using time steps
of 5 minutes (hydrodynamic model) and 15 minutes (eutrophication model).
This Report
This report comprises the primary documentation of the eutrophication com-
ponent of the 2002 CBEMP. We concentrate here on portions of the model that
have undergone major revisions and on portions that have not been previously
documented. Aspects of the model that proved particularly troublesome are docu-
mented for reference against future improvements. We have minimized repetition
of previously-reported information and model results. Complete model results are
available on a CD-ROM that accompanies this report.
Coupling with the Hydrodynamic Model
CH3D-WES
Hydrodynamic model formulation is based on principles expressed by the
equations of motion, conservation of volume, and conservation of mass. Quantities
computed by the model include three-dimensional velocities, surface elevation,
vertical viscosity and diffusivity, temperature, salinity, and density.
The basic equations of CH3D-WES are solved via the finite-difference method
on a grid of discrete cells. The computational grid extends from the heads of tide of
major bay tributaries out onto the continental shelf. The grid contains 2961 cells,
roughly 4 km2 in area, in the surface plane. Number of cells in the vertical ranges
from one to nineteen. Surface cells are 2.14 m thick at mean tide. Thickness of all
sub-surface cells is fixed at 1.53 m. Total number of cells in the grid is 12,920.
Calibration and Verification
The hydrodynamic model was calibrated and verified against a large body of
observed tidal elevations, currents, and densities. The calibration process was
reviewed by an Expert Panel consisting of three university faculty members. Final
approval was obtained from the panel before the hydrodynamic model was used to
drive the water quality model.
Linkage to the Water Quality Model
Hydrodynamics for employment in the water quality model were produced for
ten years, 1985-1994. Each year was a single, continuous production run.
Computed flows, surface elevations, and vertical diffusivities were output at two-
hour intervals for use in the water quality model. The algorithms and codes for
linking the hydrodynamic and water quality models were developed over a decade
ago and have been tested and proved in extensive applications since then. In every
model application, the linkage is verified by comparing transport of a conservative
-Vll -
-------
Executive Summary
substance computed in each model. For estuarine applications, the transport tests
commonly take the form of comparing salinity computed by two models. In the
present application, salinity comparisons verify that the hydrodynamic and water
quality models are correctly coupled and that transport processes in the two models
are virtually identical.
Boundary Conditions
Boundary conditions must be specified at all open edges of the model grid.
These include river inflows, lateral flows, and the ocean interface.
Ocean Boundary Conditions
In the first version of the model, the open edge of the model grid was at the
entrance to Chesapeake Bay. In the Tributary Refinements phase of model develop-
ment, the grid was extended beyond the bay mouth, out onto the continental shelf.
The primary objective was to ensure that boundary conditions specified at the edge
of the grid were beyond the influence of conditions within the bay. The grid exten-
sion traded one set of problems for another. Model boundaries were moved from a
location with abundant observations to multiple locations at which little informa-
tion was available. In the present phase of the model study, specification of
boundary conditions at the edge of the grid proved especially problematic. The
strategy was developed in which conditions observed at the bay mouth were
extended to the edges of the grid. Kinetics were disabled outside the bay mouth to
prevent substance transformations.
Our Recommendation
Extension of the grid onto the continental shelf had two objectives. The first was
to move nutrient boundary conditions to a location beyond the influence of loads
within the bay. The second was to allow for coupling with a proposed continental
shelf model. The first objective was met, albeit with trade-offs. The proposed shelf
model has been postponed indefinitely. The extension of the grid produced enor-
mous difficulties for both the hydrodynamic and water quality modeling teams and
did not increase the accuracy of either model. Consequently, we recommend the
boundary be restored to the mouth of the bay in future model efforts.
Hydrology and Loads
Hydrology
Major sources of freshwater to the Chesapeake Bay system are the Susquehanna
River, to the north, and the Potomac and James Rivers to the west. Of these, the
Susquehanna provides by far the largest flow fraction, followed by the Potomac
and James. All tributaries exhibit similar seasonal flow patterns. Highest flows
occur in winter (December-February) and spring (March-May). Lowest seasonal
flows occur in summer (June-August) and fall (September-November) although
tropical storms in these seasons can generate enormous flood events.
-Vlll -
-------
Executive Summary
Loads
Loads to the system include distributed or nonpoint-source loads, point- source
loads, atmospheric loads, bank loads, and wetlands loads. Nonpoint-source loads
enter the system at tributary fall lines and as runoff below the fall lines. Point-
source loads are from industries and municipal wastewater treatment plants.
Atmospheric loads are from the atmosphere directly to the water surface. Atmos-
pheric loads to the watershed are incorporated in the distributed loads. Bank loads
originate with shoreline erosion. Wetland loads are materials created in and
exported from wetlands.
Nonpoint-Source Loads
Nonpoint-source loads are from Phase IV of the EPA Chesapeake Bay Water-
shed Model. Loads are provided on a daily basis, routed to surface cells on the
model grid. Routing is based on local watershed characteristics and on drainage
area contributing to the cell.
Largest nitrogen loads, by far, come from the Susquehanna River. Lesser loads
enter at the Potomac and James fall lines. Magnitude of nitrogen loading corre-
sponds to relative flows in these tributaries. The greatest phosphorus loads enter at
the Susquehanna, Potomac, and James fall lines. Phosphorus loads are not propor-
tional to flows in these tributaries. In 1990, phosphorus load in the Potomac was
70% of the load in the Susquehanna although Potomac flow was 25% of the
Susquehanna. Phosphorus load from the James amounted to half the load from the
Susquehanna although flow was less than 25% of the Susquehanna. Multiple
factors may account for the disparity between flow and load. No doubt watershed
characteristics and above-fall-line point-source loadings contribute. A speculation is
that a portion of the Susquehanna particulate phosphorus load is retained in the
Conowingo reservoir, just upstream of the fall line.
In 1990, the largest solids load was from the Potomac, followed by the Susque-
hanna and James. The relative importance of the major tributaries varies from year
to year, however, depending on occurrence of major storm events. For suspended
solids, the Conowingo reservoir acts as a settling basin to remove solids from the
Susquehanna before entering the bay. Although solids can be scoured from the
Conowingo at high flows, the predominant effect of the reservoir is to diminish
Susquehanna solids loads relative to the other major tributaries.
Point-Source Loads
Point-source loads were provided by the EPA Chesapeake Bay Program Office
in December 2000. These were based on reports provided by local regulatory
agencies. Loads from individual sources were summed into loads to model surface
cells and were provided on a monthly basis. Despite the provision of monthly
values, loads from Virginia were most often specified on an annual basis while
loads from Maryland varied monthly.
Point-source loads are concentrated in urban areas. Major nitrogen loads
originate in Northern Virginia/District of Columbia, Richmond Virginia, Baltimore
Maryland, and Hampton Roads Virginia. Of these, only loads to the Patapsco show
a monotonic decreasing trend. Loads to the upper Potomac and upper James
- IX-
-------
Executive Summary
suggest a decrease after 1990. Point-source nitrogen loads to the Back River and
the lower James show no trend.
Point-source phosphorus loads are concentrated in the same urban areas as
nitrogen although the relationship by size differs. In 1990, largest loads were from
Hampton Roads and Richmond Virginia. Northern Virginia /District of Columbia
contributed the least phosphorus load of the major urban centers. The same area
showed the least reduction of phosphorus loads over the simulation period, possibly
a consequence of load reductions prior to 1985. In the remaining urban areas,
point-source phosphorus loads were halved from 1985 to 1994.
Atmospheric Loads
Daily atmospheric loads for each surface cell were computed by the EPA
-Chesapeake Bay Program Office and provided in January 2000. Wet deposition of
ammonium and nitrate was derived from National Atmospheric Deposition
Program observations. Dry deposition of nitrate was derived from wet deposition
using ratios calculated by the Regional Acid Deposition Model. Deposition of
organic and inorganic phosphorus was specified on a uniform, constant, areal basis
derived from published values.
Bank Loads
Bank loads are the solids, carbon, and nutrient loads contributed to the water
column through shoreline erosion. Although erosion is episodic, bank loads can be
estimated only as long-term averages. The volume of eroded material is commonly
quantified from comparison of topographic maps or aerial photos separated by time
scales of years. Consequently, the erosion estimates are averaged over periods of
years.
Erosion rates for various portions of the system were summarized by the Balti-
more District, US Army Corps of Engineers. The summary indicates widespread
coverage but no estimates for the tidal fresh portions of the western tributaries, for
the Patuxent River, and for major eastern shore embayments. Extensive measures of
composition of eroded material are available for the major Virginia tributaries but
are sparse or absent elsewhere. In view of the missing coverage and high variance
in the observations, we decided to consider bank loading as a spatially and tempo-
rally uniform process. We found an erosion rate of 5.7 kg nr1 d"1 (0.7 ton ft"1 yr1)
produced reasonable solids computations in the model.
The spatial distribution of bank loads depends directly on the length of the
modeled shoreline. Segments receiving the greatest loads include the lower estu-
arine Potomac, Tangier Sound, and the tidal fresh portions of the major western
tributaries. One characteristic of bank nutrient loads is they are phosphorus-
enriched relative to organic matter. The modeled affinity of phosphorus for silt and
clay produces a nitrogen-to-phosphorus ratio in modeled loads of 0.5:1 while the
characteristic ratio in phytoplankton is 7:1.
Wetlands Loads
Wetlands loads are the sources (or sinks) of oxygen and oxygen-demanding
material associated with wetlands that fringe the shore of the bay and tributaries.
-x-
-------
Executive Summary
These loads are invoked primarily as an aid in calibration of tributary dissolved
oxygen. Wetlands areas adjacent to model surface cells were derived via GIS
analysis and provided by the EPA Chesapeake Bay Program in November 2001.
Loads to each cell were computed as the product of adjacent wetlands area and
areal carbon export or oxygen consumption.
A uniform carbon export of 0.3 g C m~2 d"1 was employed. A uniform oxygen
demand of 2 g O2 m~2 d"1 was employed. Segments receiving the largest carbon
loads and subject to the greatest oxygen consumption include the mid-portion of
the bay, Tangier Sound, several Eastern Shore tributaries, the middle and lower
James River, the tidal fresh York River, and the York River mouth.
Loading Summary
Loads from all sources were compared by for 1990, a year central to the simula-
tion period. Runoff in this year was moderate in the Susquehanna and James and
low in the Potomac.
Nonpoint sources dominated the nitrogen loads except in a few segments adja-
cent to major urban areas. In these regions, point sources contributed a significant
fraction of nitrogen loads. Atmospheric nitrogen loads were significant only in the
large, open segments of the mainstem bay and in the lower Potomac. Bank loads
were negligibly small throughout.
Nonpoint sources usually comprised the largest fraction of phosphorus loads but
were not so predominant as for nitrogen. As with nitrogen, point-source loads were
significant in urban areas and atmospheric loads contributed to substantially to
large open-water segments. A major contrast with nitrogen was in bank loading
which equaled or exceed nonpoint-source phosphorus loads in segments distant
from major inflows. These included the eastern embayments and the river-estuarine
transition segments of western tributaries.
Nonpoint sources dominate the solids loads in segments adjoining the inflows of
the Susquehanna, Potomac, and James. Otherwise, bank loads are the dominant
source. For 1990, bank loads contribute more solids, system-wide, than nonpoint
sources. Of course, the relative contribution of bank loads depends on annual
hydrology. In high-flow years, nonpoint sources may dominate. Still, the summary
indicates that control of bank erosion should be included in any solids management
plan.
Kinetics
The foundation of CE-QUAL-ICM is the solution to the three-dimensional mass-
conservation equation for a control volume. Control volumes correspond to cells on
the hydrodynamic model grid. Solution is via a finite-difference scheme using the
QUICKEST algorithm in the horizontal plane and a Crank-Nicolson scheme in the
vertical direction. Discrete time steps, determined by computational stability
requirements, are 15 minutes.
At present, the CE-QUAL-ICM model incorporates 24 state variables in
the water column including physical variables, multiple algal groups, two
zooplankton groups, and multiple forms of carbon, nitrogen, phosphorus and
silica (Table ES-1).
-xi -
-------
Executive Summary
Table ES-1
Water Quality Model State Variables
Temperature
Fixed Solids
Spring Diatoms
Microzooplankton
Dissolved Organic Carbon
Refractory Particulate Organic Carbon
Nitrate+Nitrite
Labile Particulate Organic Nitrogen
Total Phosphate
Labile Particulate Organic Phosphorus
Chemical Oxygen Demand
Dissolved Silica
Salinity
Freshwater Cyanobacteria
Other (Green) Algae
Mesozooplankton
Labile Particulate Organic Carbon
Ammonium+Urea
Dissolved Organic Nitrogen
Refractory Particulate Organic Nitrogen
Dissolved Organic Phosphorus
Refractory Particulate Organic Phosphorus
Dissolved Oxygen
Particulate Biogenic Silica
Algae
Algae are grouped into three model classes: cyanobacteria, spring diatoms, and
other green algae. The cyanobacteria distinguished in the model are the bloom-
forming species found in the tidal, freshwater Potomac River. Spring diatoms are
large phytoplankton that produce an annual bloom in the saline portions of the bay
and tributaries. The other green algae represent the mixture that characterizes saline
waters during summer and autumn and fresh waters year round.
Zooplankton
Two zooplankton groups are considered: microzooplankton and mesozoo-
plankton.
Organic Carbon
Three organic carbon state variables are considered: dissolved, labile particulate,
and refractory particulate. Labile and refractory distinctions are based upon the
time scale of decomposition. Labile organic carbon decomposes rapidly in the
water column or the sediments. Refractory organic carbon decomposes slowly,
primarily in the sediments, and may contribute to sediment oxygen demand years
after deposition.
-Xll -
-------
Executive Summary
Nitrogen
Nitrogen is first divided into available and unavailable fractions. Available refers
to employment in algal nutrition. Two available forms are considered: reduced and
oxidized nitrogen. Reduced nitrogen includes ammonium and urea. Nitrate and
nitrite comprise the oxidized nitrogen pool. Unavailable nitrogen state variables are
dissolved organic nitrogen, labile particulate organic nitrogen, and refractory partic-
ulate organic nitrogen.
Phosphorus
As with nitrogen, phosphorus is first divided into available and unavailable frac-
tions. Only a single available form, dissolved phosphate, is considered. Three forms
of unavailable phosphorus are considered: dissolved organic phosphorus, labile
particulate organic phosphorus, and refractory particulate organic phosphorus.
Silica
Silica is divided into two state variables: dissolved silica and particulate biogenic
silica. Dissolved silica is available to diatoms while particulate biogenic silica
cannot be utilized.
Chemical Oxygen Demand
Chemical oxygen demand is the concentration of reduced substances that are
oxidized by abiotic processes. The primary component of chemical oxygen demand
is sulfide released from sediments. Oxidation of sulfide to sulfate may remove
substantial quantities of dissolved oxygen from the water column.
Dissolved Oxygen
Dissolved oxygen is required for the existence of higher life forms. Oxygen
availability determines the distribution of organisms and the flows of energy and
nutrients in an ecosystem. Dissolved oxygen is a central component of the water-
quality model.
Salinity
Salinity is a conservative tracer that provides verification of the transport com-
ponent of the model and facilitates examination of conservation of mass. Salinity
also influences the dissolved oxygen saturation concentration and may be used in
the determination of kinetics constants that differ in saline and fresh water.
Temperature
Temperature is a primary determinant of the rate of biochemical reactions.
Reaction rates increase as a function of temperature although extreme temperatures
may result in the mortality of organisms and a decrease in kinetics rates.
-XIII -
-------
Executive Summary
Fixed Solids
Fixed solids are the mineral fraction of total suspended solids. Solids are consid-
ered primarily for their role in light attenuation.
Format of Model-Data Comparisons
The Monitoring Data Base
The water quality model was applied to a ten-year time period, 1985-1994. The
monitoring data base maintained by the Chesapeake Bay Program was the principal
source of data for model calibration. Observations were collected at 49 stations in
the bay and 89 stations in major embayments and tributaries. Sampling was
conducted once or twice per month with more frequent sampling from March
through October. Samples were collected during daylight hours with no attempt to
coincide with tide stage or flow. At each station, salinity, temperature, and
dissolved oxygen were measured in situ at one- or two-meter intervals. Samples
were collected one meter below the surface and one meter above the bottom for
laboratory analyses. At stations showing significant salinity stratification, additional
samples were collected above and below the pycnocline. Analyses relevant to this
study are listed in Table ES-2. The listed parameters are either analyzed directly or
derived from direct analyses.
Comparison with the Model
Time series comparisons of computations and observations were produced at 42
locations. These were selected to provide coverage throughout the system. At least
one station was selected within each of the Chesapeake Bay Program Segments
represented on the grid. Within the model code, daily-average concentrations were
derived from computations at discrete time steps (15 minutes). These were
compared to individual observations, at surface and bottom.
The spatial distributions of observed and computed properties were compared in
a series of plots along the axes of the bay and major tributaries. Three years were
selected for comparisons: 1985, 1990, and 1993. The year 1985 was a low-flow
year although the western tributaries were impacted by an enormous flood event in
November. Flows in 1990 were characterized as average while major spring runoff
occurred in 1993.
Model results and observations were averaged into four seasons:
Winter—December through February
Spring—March through May
Summer—June through August
Fall-September through November
The mean and range of the observations, at surface and bottom, were compared
to the average and range of daily-average model results.
The vertical distributions of observed and computed properties were compared at
selected stations in the bay and major tributaries. As with the longitudinal compar-
isons, selection and aggregation were required to produce a manageable volume of
information. Comparisons were completed for three years and were subjected to
-XIV-
-------
Executive Summary
Table ES-2
Constituents in Model-Data Comparisons
Constituent
Chlorophyll >a=
Dissolved Inorganic Nitrogen
Dissolved Inorganic Phosphorus
Dissolved Organic Nitrogen
Dissolved Organic Phosphorus
Dissolved Oxygen
Light Attenuation
Ammonium
Nitrate+Nitrite
Particulate Organic Carbon
Particulate Organic Nitrogen
Particulate Phosphorus
Salinity
Total Nitrogen
Total Organic Carbon
Total Phosphorus
Total Suspended Solids
Temperature
Dissolved Silica
Time -Series
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Longitudinal
X
X
X
X
X
X
X
X
X
X
X
X
X
Vertical
X
X
X
seasonal averaging as previously described. Parameters were limited to the three for
which detailed vertical observations were available: temperature, salinity, and
dissolved oxygen.
Zooplankton
This chapter details the formulation of zooplankton kinetics in the Chesapeake
Bay Environmental Model Package. Two zooplankton groups, microzooplankton
and mesozooplankton, were incorporated into the model during the tributary refine-
ments phase.
The present model represents zooplankton biomass within 50% to 100% of
observed values. Discrepancies between observations and model certainly indicate
-xv-
-------
Executive Summary
shortcomings in the model. A large portion of the discrepancies, however, must be
attributed to observational methodology and to the variance inherent in the popula-
tions. A high degree of accuracy is unlikely to be obtained but more realism and,
potentially, more accuracy can be added to the model. Suggestions for improve-
ment range from parameter re-evaluation through complete model reformulation.
A basic improvement in the model is to distinguish, through parameter values,
freshwater and estuarine zooplankton populations. Differentiation between indi-
vidual tributaries may also be appropriate. Differentiation of freshwater and
saltwater populations is readily justified and this approach will likely prove
successful.
Another potential improvement is to add a second mesozooplankton group. One
group would represent the winter-spring population; the second group would repre-
sent the summer population. No doubt, these two populations exist and can be
differentiated. The second group can be readily included and adds realism to the
model. The potential quantitative improvement in model computations cannot be
foreseen.
The final improvement is the most difficult. Add age structure to the mesozoo-
plankton model. Adults in the present model instantaneously reproduce adults. In
the most realistic model, adults would produce eggs. Eggs would hatch into larvae,
mature into juveniles and, later, into adults. A model of this sort offers the highest
probability of success in representing the time series of observed mesozooplankton.
A multi-stage population model requires tremendous resources in programming,
calibration, and execution. The additional effort is likely not worthwhile in the
present multi-purpose model. The improvement is highly recommended if the
present model is employed in an application that focuses largely on zooplankton.
Effects of Predation and Respiration
on Primary Production
A basic Nutrient-Phytoplankton-Zooplankton (NPZ) model is used to examine
effects of predation and respiration formulations on computed primary production.
Phytoplankton with Respiration Only
The simplest system contains phytoplankton and a nutrient with no predators or
predation term. Gross production increases as a function of respiration over much
of the range, then declines precipitously as respiration attains roughly 75% of the
maximum growth rate. Net production also exhibits a curvilinear relationship to
respiration with a peak at roughly 50% of the maximum growth rate. The potential
increase in production along with respiration is counter-intuitive. Since biomass
declines as respiration increases, a simultaneous decline in production is expected.
The key is to realize that production is the product of biomass and the nutrient-
limited growth rate. As respiration increases over much of its range, nutrients are
released from algal biomass, the nutrient limitation to growth is relaxed, and the
product of growth and biomass increases.
-xvi -
-------
Executive Summary
Phytoplankton with Zooplankton
A system which includes zooplankton but no higher-order predation terms is
examined next. Algal biomass decreases rapidly and approaches zero asymtotically
as predation rate is increased from the minimum feasible value. Surprisingly,
zooplankton biomass also diminishes as predation rate is increased. The steady
state solution indicates maximum biomass occurs at low predation rates in which
zooplankton slowly feed on a large algal standing stock. Primary production
responds in direct proportion to algal biomass. Maximum production occurs a
minimum predation rate. The steady-state solution for the basic parameter set indi-
cates little or no nutrient limitation. Consequently production, the product of
nutrient-limited growth and biomass, responds primarily to changes in biomass.
Phytoplankton with Quadratic Predation
Predation by higher trophic levels (other than zooplankton) on phytoplankton is
represented by a quadratic term in the algal mass-balance equation. Algal concen-
tration declines as predation rate increases. Despite the decrease in algal biomass,
primary production increases as predation is raised from minimum levels. The
increase occurs because the nutrient limit to production is relaxed as predation
liberates nutrients from algal biomass. Consequently, the product of nutrient-
limited growth rate and algal biomass increases despite the decrease in biomass.
The increasing trend in production continues until nutrients are no longer limiting.
At that point, production declines in proportion to biomass.
Process-Based Primary Production Model
Primary production calculations in the original version of the CBEMP were
consistent with characteristics of similar models. Computed production matched or
exceeded observed production in the turbidity maximum region, where nutrients
are abundant but light is limited. In the middle and lower portions of the bay,
where light attenuation is diminished but nutrients are sparse relative to the
turbidity maximum, computed production fell short of observed.
Management efforts in the Bay now require investigation of the effects of filter-
feeders in reducing eutrophication, so that the amount of production available to
these organisms must be represented. At the same time, water quality standards are
becoming more stringent such that accurate computations of chlorophyll and
nutrient concentrations cannot be ignored. Consequently, the CBEMP must now
represent both properties of the system and production rates.
A primary production model is described and compared to three observational
data bases: light-saturated carbon fixation, net phytoplankton primary production,
and gross phytoplankton primary production. The model successfully reproduces
the observations while maintaining realistic calculations of algal carbon, chloro-
phyll, limiting nutrient, and light attenuation. Computed primary production in
light-limited regions is proportional to the algal growth rate. Successful computa-
tion of primary production in nutrient-depleted waters depends on the formulation
and magnitude of the model predation term. Our quadratic formulation mimics a
predator population that is closely coupled to algal biomass and rapidly recycles
nutrients from algal biomass to available form.
-xvil -
-------
Executive Summary
Suspended Solids and Light Attenuation
Light Attenuation Model
We applied a model that related light attenuation to attenuation from water, from
inorganic solids, and from organic solids. Model parameters were determined for
each Chesapeake Bay Program Segment represented in the model domain. The
result was a model in which background attenuation was highest near the fresh-
water sources and lowest near the ocean interface. Solids near the fall lines were
characterized as having higher attenuation than solids in regions distant from
sources in upland watersheds. Our light attenuation model is similar to the model
developed as part of the second technical synthesis on submerged aquatic vegeta-
tion and water quality. Managers and other users of the two models, ours and "Tech
Syn II," should be confident that guidance obtained from the two models will be
consistent.
Suspended Solids
The principal suspended solids variable in the CBEMP was inorganic (fixed)
suspended solids. Organic solids were derived from particulate carbon variables
(phytoplankton, zooplankton, detritus) and added to inorganic solids for compar-
ison to observed total suspended solids.
Aside from loads, the distribution of inorganic solids in the water column was
determined by two settling velocities. One represented settling through the water
column and the other represented net settling to the bed sediments. Net settling was
always less than or equal to settling through the water column. The reduced magni-
tude of net settling represented the effect of resuspension. The employment of net
settling was a primary distinction between our own suspended solids model and a
true sediment transport model. Our model included no resuspension mechanism.
Once a particle was deposited on the bottom, it remained there.
Recommendations
Our present efforts probably represent the state of the art in use of a water-
quality model to guide management of water clarity. We recognize significant
improvements in monitoring and modeling are necessary to bring modeling of
water clarity up to the levels achieved in modeling nutrient cycling. The most
crucial need is for a rigorous, mechanistic three-dimensional sediment transport
model. The sediment transport model will accommodate the resuspension process,
which is missing in the current representation. Resuspension does not constitute a
new source of sediments to the system. Rather, it is a process that returns to the
water column sediments that originated in external loads and as internal production.
Our current net settling algorithm provides a mechanism to represent the spatial
distribution of suspended solids and, especially, features like the classic turbidity
maximum. The net settling algorithm is less useful, however, in the prediction of
solids responses to load reductions. As currently formulated, our net settling
algorithm ensures that load reductions produce reductions in computed sediment
concentrations. Resuspension may, however, counter or eliminate benefits gained
from load reductions. Consequently, management scenarios run with the present
model represent the maximum benefit to be gained from solids load reductions.
-xvill -
-------
Executive Summary
Tributary Dissolved Oxygen
Analysis and modeling of tributary dissolved oxygen has largely focused on
persistent or intermittent anoxia that occurs in bottom waters near the mouths of
four major western tributaries: the York, Rappahannock, Potomac, and Patuxent. A
tendency present in the model since the earliest application is the over-prediction of
surface dissolved oxygen in the tributaries, especially in the tidal freshwater
portions. This performance characteristic was overlooked when attention was
focused on bottom waters. During the present study, sponsors noted the discrepan-
cies between computed and observed surface dissolved oxygen concentrations and
asked for improved model performance. An extensive number of calibration and
sensitivity runs were performed while attempting to improve the model.
Two phenomenon vex the computation of dissolved oxygen in the western tribu-
taries. The first is the computation of excess dissolved oxygen in the tidal fresh
portions of the James and Potomac Rivers. The excess dissolved oxygen is the
result of an excess of computed production over consumption. We believe the
problem lies on the consumption side. Riverine organic carbon loads to the Virginia
tributaries are virtually unknown as are point-source carbon loads to all tributaries.
In addition, the James and Potomac receive loads from combined-sewer overflows
and urban runoff. Improved dissolved oxygen computations require improved infor-
mation on loading. Ideally, measures of respiration or of biochemical oxygen
demand in the water column should also be conducted.
The second vexing process is the occurrence of depressed surface dissolved
oxygen in the lower estuaries, notably the York, the Rappahannock, and the
Potomac. Our best explanation of the phenomenon is transfer of oxygen demand
and/or oxygen-depleted water from the bottom to the surface. The phenomena
cannot be represented by simple adjustments vertical mixing, however. The
phenomenon requires additional study and may be beyond modeling without
process-based field observations.
The York, Rappahannock, and Patuxent Rivers adjoin extensive tidal wetlands
which appear to influence water quality. Nutrients, as well as dissolved oxygen,
are, no doubt, exchanged between wetlands and channel. To represent the wetlands
physically, addition of wetting-and-drying to the hydrodynamic model is required.
A wetland biogeochemical module should be added to the water quality model.
And, as with so many processes, field investigations may also be necessary.
Modeling Processes at the Sediment-Water Interface
The Sediment Diagenesis Model
The need for a predictive benthic sediment model was revealed in a steady-state
model study that preceded Corps' modeling activity. For management purposes, a
model was required with two fundamental capabilities:
• Predict effects of management actions on sediment-water exchange processes,
and
• Predict time scale for alterations in sediment-water exchange processes.
A sediment model to meet these requirements was created for the initial three-
dimensional coupled hydrodynamic-eutrophication model. With some
modifications, the initial sediment model is employed in the present CBEMP.
-xix-
-------
Executive Summary
Cumulative distributions were created for the population of Chesapeake Bay
sediment-water flux observations and for corresponding computations from the
present model. Computed sediment oxygen demand exceeded observed throughout
the distribution. Median computed demand exceeded observed by more than
0.5 g m~2 d"1. Observed sediment ammonium release exceeded computed
throughout the distribution. Median observed release exceeded modeled by 10 mg
m~2 d"1. A number of explanations can be offered for these results. The excess of
sediment oxygen demand may be attributed to computed bottom dissolved oxygen.
Computed bottom water dissolved oxygen does not match the lowest observations
at all locations. As a result, computed sediment oxygen demand exceeds observed
because the model allows consumption of oxygen where no oxygen is present in
the bay. The excess of observed over computed ammonium release may indicate
more nitrogen should be deposited on the bottom. Or the occasional excess of
computed dissolved oxygen may be allowing more nitrification to take place in the
modeled sediments than in the observations.
The preponderance of observed and computed sediment-water nitrate fluxes are
essentially zero. Half the observed and computed sediment-water phosphate fluxes
are less than or effectively zero. In the upper half of the distribution, the observa-
tions show a gradual transition to sediment phosphorus release while the model
shows a much steeper gradient. Maximum phosphorus releases agree in both model
and observations. Observed sediment silica release greatly exceeds modeled. At
the median, modeled release is essentially zero while median observed release is
175 mg m~2 d"1. Maximum observed release exceeds 600 mg m~2 d"1 while the
maximum modeled release is 100 mg m-2 d"1.
The Benthos Model
For the "Virginia Tributary Refinements" phase of the model activities, a deci-
sion was made to initiate direct interactive simulation of three living resource
groups: zooplankton, benthos, and SAV. Benthos were included in the model
because they are an important food source for crabs, finfish, and other economi-
cally and ecologically significant biota. In addition, benthos can exert a substantial
influence on water quality through their filtering of overlying water. Benthos within
the model were divided into two groups: deposit feeders and filter feeders. The
deposit-feeding group represents benthos which live within bottom sediments and
feed on deposited material. The filter-feeding group represents benthos which live
at the sediment surface and feed by filtering overlying water.
Examination of present model results indicates the computation of filter feeders
closely resembles the original application. Computed deposit feeders have, perhaps,
increased since the original application. The increase in deposit feeders is inter-
preted to have negligible impact on model computations. The resemblance of
results between the present model and the Tributary Refinements model indicates
the activity of the benthos, as originally calibrated, is maintained in the present
model.
Benthic Algae
Benthic algae are considered to occupy a thin layer between the water column
and benthic sediments. Biomass within the layer is determined by the balance
-XX-
-------
Executive Summary
between production, respiration, and losses to predation. Nutrients from diagenetic
sediment production and from the overlying water are both available to the benthic
algae.
Computed benthic algal biomass ranges up to 3 g C m~2, in agreement with
measures conducted in a variety of systems. The primary determinate of algal
density is light. Algal biomass shows an inverse relationship to optical depth (total
depth x light attenuation) at the sediment-water interface. No algae are computed
above optical depth 5. The highest densities of computed benthic algae are found in
shallow water near the mouths of the lower western tributaries, along the lower
eastern shore, and in eastern embayments. Lesser densities occur in tidal fresh
waters and in other shoal areas.
Submerged Aquatic Vegetation (SAV)
An SAV submodel, which interacted with the main model of the water column
and with the sediment diagenesis submodel, was created for the "Virginia Tributary
Refinements" phase of the model activities. Three state variables were modeled:
shoots (above-ground biomass), roots (below-ground biomass), and epiphytes
(attached growth). Three dominant SAV communities, Vallisneria americana,
Ruppia maritima, or Zostera marina, were modeled.
When the phytoplankton production relationships and parameters in the present
eutrophication model were revised, corresponding changes were made to the
epiphyte component of the SAV model. Examination of the SAV component of the
model revealed that these changes, and perhaps others, had a substantial, detri-
mental, effect on computed SAV. Computed epiphytes overwhelmed the vegetation.
As a consequence, a re-calibration of the SAV model was completed. We endeav-
ored to bring epiphytes and SAV back into calibration while minimizing revisions
to the extensive model parameter suite. Changes were centered on the epiphyte loss
terms and on the SAV production-irradiance relationships. The computations in the
present model are consistent with, but not identical to, the original application. The
original application obeyed light attenuation criteria listed in the first SAV Tech-
nical Synthesis. During the re-calibration we verified that the present model
remains consistent with these criteria.
Dissolved Phosphate
An excess of computed dissolved phosphate, especially during summer, has been
a characteristic of the model since the original phase. While tuning the model to
effect an overall reduction in computed dissolved phosphate presents no problem,
reducing phosphate in summer while maintaining sufficient phosphate to support
the spring phytoplankton bloom is precarious.
We conducted an extensive number of sensitivity runs and process investigations
in order to calibrate dissolved phosphate in the present model. The final model cali-
bration incorporates dissolved organic phosphorus mineralization, uptake by sulfide
oxidizing bacteria, and precipitation. Introduction of the two uptake mechanisms as
well as alterations in multiple parameter values provided a reasonable representa-
tion of summer-average phosphate in the surface of the bay, especially during
years of dry to moderate hydrology. Considerable excess of computed phosphate
remained present in a wet year.
-xxi -
-------
Executive Summary
Phosphorus is a limiting nutrient in the mainstem bay in spring while nitrogen is
the primary limiting nutrient in summer. Consequently, our rough approach to
modeling dissolved phosphate in summer is acceptable. Still, a new phosphorus
model seems appropriate, especially for management of freshwater segments where
phosphorus is the more important nutrient.
The first step in a new phosphate model is to explicitly recognize phosphate
uptake by heterotrophic bacteria. Bacteria do not necessarily have to be incorpo-
rated into the model as a state variable. One reviewer suggested relating phosphate
uptake to organic carbon respiration, which is a bacterial process.
The second step in an improved phosphorus model is to explicitly differentiate
particulate inorganic phosphorus from particulate organic phosphorus through the
addition of a particulate inorganic phosphorus state variable.
A third step is to explore the utilization of dissolved organic phosphorus by
bacteria and phytoplankton.
A concluding modification is to implement realistic sediment transport
processes. No doubt, a distinct particulate inorganic phosphorus form exists and is
transported along with the solids with which it is associated. Our ability to simulate
solids transport with the present model is limited, however. Consequently, correct
representation of total phosphorus is impossible when solids distributions cannot be
reproduced.
Statistical Summary of Calibration
The calibration of the model involved the comparison of hundreds of thousands
of observations with model results in various formats. Comparisons involved
conventional water quality data, process-oriented data, and living-resources obser-
vations. The graphical comparisons produced thousands of plots which cannot be
assimilated in their entirety. Evaluation of model performance requires statistical
and/or graphical summaries of results.
Examination of statistical summaries (Table ES-3) requires a good deal of judge-
ment and interpretation. Generalizations and distinctions are not always possible. One
clear pattern is that the model overestimates, on average, surface chlorophyll. The
overestimation ranges from less than 1 mg m~3 to more than 2 mg m~3. The model
consistently underestimates salinity although the mean error is always less than 1 ppt.
For all systems except the Potomac, computed mean summer, bottom dissolved
oxygen is within 1 g m~3 of the observed average. In the Potomac, computed mean
summer bottom dissolved oxygen is almost 2 g m~3 higher than observed. Careful
examination of model results indicates the region of greatest computed excess is in
the tidal fresh portion of the river, where observed bottom dissolved oxygen exceeds
5 g m~3. Excessive computed dissolved oxygen, surface and bottom, is a characteris-
tics of the present model in most tidal freshwater regions. Except in the James, the
model underestimates mean total phosphorus concentration. Underestimation of total
phosphorus has been a characteristic of the model since the earliest application. We
originally attributed the shortfall to omission of bankloads. In this version we include
bankloads of phosphorus but they are difficult to estimate accurately. The model also
omits resuspension of particulate phosphorus and has difficulty reproducing the
concentration of particulate phosphorus in the turbidity maximums. In the James, we
attribute the excess computed phosphorus to uncertainty in the large point-source
and distributed loads to this tributary.
- xxii -
-------
Executive Summary
Table ES-2 Statistical Summary of Calibration 1985-1994
Surface Chlorophyll, ug/L
Mean Error
Absolute Mean Error
Relative Error
Summer, Bottom Dissolved
Oxygen, mg/L
Mean Error
Absolute Mean Error
Relative Error
Light Attenuation, 1/m
Mean Error
Absolute Mean Error
Relative Error
Salinity, ppt
Mean Error
Absolute Mean Error
Relative Error
Total Nitrogen, mg/L
Mean Error
Absolute Mean Error
Relative Error
Total Phosphorus, mg/L
Mean Error
Absolute Mean Error
Relative Error
Mainstem Bay
-0.53
5.01
58.4
Mainstem Bay
0.32
1.47
35.7
Mainstem Bay
0.02
0.36
35.3
Mainstem Bay
0.71
1.97
11.8
Mainstem Bay
0.04
0.17
24.3
Mainstem Bay
0.005
0.014
37.6
James
-2.05
9.29
75.7
James
-0.09
2.43
36.6
James
-0.21
0.97
43.7
James
0.11
2.01
31.2
James
-0.16
0.42
44.6
James
-0.021
0.069
63.8
York
-1.68
4.71
60.1
York
0.37
1.18
22.8
York
0.09
0.84
41.9
York
0.95
1.84
14.5
York
0.01
0.23
33.1
York
0.012
0.036
49.2
Rappahannock
-2.55
8.22
81.4
Rappahannock
0.62
1.93
35.4
Rappahannock
-0.17
0.89
42.3
Rappahannock
0.01
1.49
18.3
Rappahannock
0.14
0.28
33.8
Rappahannock
0.001
0.036
52.6
Potomac
-1.85
7.45
80.2
Potomac
-1.31
2.13
40.5
Potomac
-0.02
1.03
45.2
Potomac
0.45
0.97
22.5
Potomac
0.32
0.61
31.9
Potomac
0.032
0.053
58.9
Patuxent
-1.53
8.15
65.4
Patuxent
-0.92
1.74
39.3
Patuxent
-0.20
0.84
38.4
Patuxent
0.16
1.69
17.5
Patuxent
-0.13
0.43
41.5
Patuxent
0.041
0.047
47.6
Examination of relative errors (Table ES-3) indicates that chlorophyll has the
greatest error, salinity the least. Relative error in chlorophyll prediction is 60% to
80% while relative error in salinity prediction is 10% to 20%. The chlorophyll error
reflects the difficulty in computing this dynamic biological component which can
attain unlimited magnitude. In contrast, salinity is purely physical and is bounded
at the upper end by oceanic concentration. The remaining components are in the
mid-range, 30% to 50%, with total phosphorus, perhaps exhibiting slightly higher
relative error. The higher error in phosphorus reflects the aforementioned difficul-
ties in evaluating loads, in simulating resuspension, and in representing particulate
phosphorus transport.
The mainstem bay is clearly superior in computations of salinity, total nitrogen,
and total phosphorus. The James River stands out as demonstrating the highest
-XXIII -
-------
Executive Summary
relative error in these components. We partially attribute the greater accuracy in the
mainstem to the relatively dense computational grid in this region. An additional,
and probably more significant influence, is that the mainstem is dominated by
internal processes while the tributaries are strongly influenced by point-source and
distributed loads. The point-source loads are incompletely described, especially in
the early years of the simulation and in the Virginia tributaries. Below-fall-line
distributed loads cannot be measured; they can only be computed by the watershed
model. We believe the uncertain loads, discharged into the constrained volumes of
the tributaries, are the major reason for higher relative error in the tributaries. The
James River stands out in this regard.
- xxiv -
-------
Introduction
Deterioration of water quality in Chesapeake Bay (Figure 1-1) and associated
losses of living resources have been recognized as problems for more than twenty
years (Flemer et al. 1983). An order-of-magnitude increase in anoxic volume and a
catastrophic decline in submerged aquatic vegetation (SAV) were among the
primary problems cited. Two decades later, elimination of anoxia and restoration of
SAV remain prime management goals. Models have been employed as tools to
guide management since the formation of the first water quality targets. Overtime,
as management focus has been refined, models have been improved to provide
appropriate, up-to-date guidance.
Three models are at the heart of the Chesapeake Bay Environmental Model
Package (CBEMP). Distributed flows and loads from the watershed are computed
with a highly-modified version of the HSPF model (Bicknell et al. 1996). These
flows are input to the CH3D-WES hydrodynamic model (Johnson et al. 1993)
which computes three-dimensional intra-tidal transport. Computed loads and trans-
port are input to the CE-QUAL-ICM eutrophication model (Cerco and Cole 1993)
which computes algal biomass, nutrient cycling, and dissolved oxygen, as well as
numerous additional constituents and processes. The eutrophication model incorpo-
rates a predictive sediment diagenesis component (DiToro and Fitzpatrick 1993).
The first coupling of these models simulated the period 1984-1986. Emphasis in
the model application was on examination of bottom-water anoxia. Results indi-
cated a decision to reduce controllable nutrient input by 40% (Baliles et al. 1987)
would reduce anoxic volume by 20%.
Circa 1992, management emphasis shifted from dissolved oxygen, a living-
resource indicator, to living resources themselves. In response, the computational
grid was refined to emphasize resource-rich areas (Wang and Johnson 2000) and
living resources including benthos (Meyers et al. 2000), zooplankton (Cerco and
Meyers 2000), and submerged aquatic vegetation (Cerco and Moore 2001) were
added to the model. The simulation period was extended from 1985 to 1994.
-------
Chapter 1 • Introduction
20 0 20 40 Kilometers
N
FIGURE 1-1. Chesapeake Bay
-2-
-------
Chapter 1 • Introduction
During this modeling phase, fixed solids were identified as major components of
light attenuation. Reductions in attenuation achieved solely through nutrient
controls on phytoplankton could not restore submerged aquatic vegetation system-
wide (Cerco and Moore 2001; Cerco et al. 2002).
In keeping with the emphasis on living resources, the Chesapeake 2000 Agree-
ment (Gilmore et al. 2000) called for a ten-fold increase in biomass of oysters and
other filter feeding organisms. In response, the computational grid was further
refined and plans were made to incorporate new living resources into the model. At
the same time, regulatory forces were shaping the direction of management efforts.
Regulatory agencies in Maryland listed the state's portion of Chesapeake Bay as
"impaired." The US Environmental Protection Agency added bay waters within
Virginia to the impaired list. Settlement of a lawsuit required development of a
Total Maximum Daily Load (TMDL) for Virginia waters by 2011. To avoid imposi-
tion of an arbitrary TMDL, the Chesapeake 2000 Agreement specified removal of
water quality impairments by 2010. Impairments in the bay were defined as low
dissolved oxygen, excessive chlorophyll concentration and diminished water clarity.
Management emphasis shifted from living resources back to living-resource indica-
tors: dissolved oxygen, chlorophyll, and clarity. A model recalibration was
undertaken, with emphasis on improved accuracy in the computation of the three
key indicators.
The 2002 Chesapeake Bay
Environmental Model Package
The framework of the original CBEMP remains intact although the components
have been substantially modified and improved over fifteen years. The watershed
model is now in Phase 4.3 (Linker et al. 2000). Documentation may be found may
be found on the Chesapeake Bay Program web site
http://www.chesapeakebay.net/modsc.htm. Nutrient and solids loads are computed
on a daily basis for 94 sub-watersheds of the 166,000 km2 Chesapeake Bay water-
shed and are routed to individual model cells based on local watershed
characteristics and on drainage area contributing to the cell. The hydrodynamic and
eutrophication models operate on a grid of 13,000 cells. The grid contains 2,900
surface cells (4 km2) and employs non-orthogonal curvilinear coordinates in the
horizontal plane. Z coordinates are used in the vertical direction which is up to 19
layers deep. Depth of the surface cells is 2.1 m at mean tide and varies as a func-
tion of tide, wind, and other forcing functions. Depth of sub-surface cells is fixed at
1.5 m. A band of littoral cells, 2.1 m deep at mean tide, adjoins the shoreline
throughout most of the system. Ten years, 1985-1994, are simulated continuously
using time steps of 5 minutes (hydrodynamic model) and 15 minutes (eutrophica-
tion model).
-3-
-------
Chapter 1 • Introduction
Expert Panels
Expert review has been part of the model activity since its commencement. For
the present study, several "Expert Panels" were assembled to review various aspects
of the model application. These teams were:
Hydrodynamics Expert Panel
Dr. Richard Garvine
College of Marine Studies
University of Delaware, Newark DE
Dr. Albert Y.Kuo
Virginia Institute of Marine Science
College of William and Mary, Gloucester Point, VA
Dr. Lawrence P. Sanford
University of Maryland Center for Environmental Science
Horn Point Laboratory, Cambridge MD
Primary Production Expert Panel
Dr. Lawrence W. Harding
University of Maryland Center for Environmental Science
Horn Point Laboratory, Cambridge MD
Dr. Raleigh Hood
University of Maryland Center for Environmental Science
Horn Point Laboratory, Cambridge MD
Dr. W. Michael Kemp
University of Maryland Center for Environmental Science
Horn Point Laboratory, Cambridge MD
Total Maximum Daily Load Expert Panel
Dr. Kevin Farley
Department of Environmental Engineering
Manhattan College, Riverdale NY
Dr. Wu-Seng Lung
Department of Environmental Engineering
University of Virginia, Charlotte sville VA
We gratefully acknowledge the advice and assistance provided by our experts.
Successful completion of this study would not have been possible without them.
-4-
-------
Chapter 1 • Introduction
This Report
This report comprises the primary documentation of the eutrophication com-
ponent of the 2002 CBEMP. The Chesapeake Bay model study has been
extensively documented since its earliest stages. We concentrate here on portions of
the model that have undergone major revisions and on portions that have not been
previously documented. Aspects of the model that proved particularly troublesome
are documented for reference against future improvements. We have minimized
repetition of previously-reported information and model results. The reader is
referred to the Bibliography, below, and to the EPA Chesapeake Bay Program web
site, http://www.chesapeakebay.net/modsc.htm, for additional information.
Bibliography
Baliles, G., Schaefer, W., Casey, R., Thomas, L., Barry, M, and Cole, K. (1987). "Chesa-
peake Bay Agreement'' United States Environmental Protection Agency Chesapeake Bay
Program, Annapolis MD.
Bicknell, B., Imhoff, I, Kittle, I, Donigian, A., Johanson, R., and Barnwell, T. (1996).
"Hydrologic simulation program - FORTRAN user's manual for release 11," United States
Environmental Protection Agency Environmental Research Laboratory, Athens GA.
Cerco, C., and Cole, T. (1993). "Three-dimensional eutrophication model of Chesapeake
Bay," Journal of 'Environmental Engineering, 119(6), 1006-10025.
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report EL-94-4, US Army Engineer Waterways Experiment Station, Vicks-
burg, MS.
Cerco, C. (1995a). "Simulation of long-term trends in Chesapeake Bay eutrophication,"
Journal of Environmental Engineering, 121(4), 298-310.
Cerco, C. (1995b). "Response of Chesapeake Bay to nutrient load reductions," Journal of
Environmental Engineering, 121(8), 549-557.
Cerco, C., and Cole, T. (1995). "User's guide to the CE-QUAL-ICM three-dimensional
eutrophication model," Technical Report EL-95-15, US Army Engineer Waterways Experi-
ment Station, Vicksburg, MS.
Cerco, C. (2000). "Phytoplankton kinetics in the Chesapeake Bay model," Water Quality
and Ecosystem Modeling, 1, 5-49.
Cerco, C., and Meyers, M. (2000). "Tributary refinements to the Chesapeake Bay Model,"
Journal of Environmental Engineering, 126(2), 164-174.
Cerco, C., and Moore, K. (2001). "System-wide submerged aquatic vegetation model for
Chesapeake Bay," Estuaries, 24(4), 522-534.
Cerco, C., Linker, L., Sweney, I, Shenk, G., and Butt, A. (2002). "Nutrient and solids
controls in Virginia's Chesapeake Bay tributaries," Journal of Water Resources Planning
and Management, 128(3), 179-189.
Cerco, C., Johnson, B., and Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model," ERDC TR-02-4, US Army Engineer Research and Development Center, Vicksburg,
MS.
Cerco, C., and Noel, M. (2003). "Managing for water clarity in Chesapeake Bay," Journal
of Environmental Engineering, in press.
-5-
-------
Chapter 1 • Introduction
DiToro, D., and Fitzpatrick, J. (1993). "Chesapeake Bay sediment flux model," Contract
Report EL-93-2, US Army Engineer Waterways Experiment Station, Vicksburg, MS.
DiToro, D. (2001). Sediment Flux Modeling, John Wiley and Sons, New York.
Flemer, D., Mackiernan, G., Nehlsen, W, and Tippie, V (1983). "Chesapeake Bay: A
profile of environmental change," U.S. Environmental Protection Agency, Region III,
Philadelphia, PA.
Gillmore, J., Glendening, P., Ridge, T., Williams, A., Browner, C., and Boiling, B. (2000).
"Chesapeake 2000 Agreement." United States Environmental Protection Agency Chesa-
peake Bay Program, Annapolis MD.
HydroQual Inc., (2000). "Development of a suspension feeding and deposit feeding benthos
model for Chesapeake Bay," Project No. USCE0410, HydroQual Inc., Mahwah NJ.
Johnson, B., Heath, R., Hsieh, B., Kim, K., and Butler, L. (1991). "Development and verifi-
cation of a three-dimensional numerical hydrodynamic, salinity, and temperature model of
Chesapeake Bay," HL-91-7, US Army Engineer Waterways Experiment Station, Vicksburg,
MS.
Johnson, B., Kim, K., Heath, R., Hsieh, B., and Butler, L. (1993). "Validation of a three-
dimensional hydrodynamic model of Chesapeake Bay," Journal of Hydraulic Engineering,
199(1), 2-20.
Linker, L., Shenk, G., Dennis, R., and Sweeney, J. (2000). "Cross-media models of the
Chesapeake Bay watershed and airshed," Water Quality and Ecosystem Modeling, 1(1-4),
91-122.
Meyers, M., DiToro, D., and Lowe, S. (2000). "Coupling suspension feeders to the Chesa-
peake Bay eutrophication model," Water Quality and Ecosystem Modeling, 1(1-4), 123-140.
Wang, H., and Johnson, B. (2000). "Validation and application of the second-generation
three-dimensional hydrodynamic model of Chesapeake Bay," Water Quality and Ecosystem
Modeling, 1(1-4), 51-90.
-6-
-------
Coupling with the
Hydrodynamic Model
Introduction
Modeling the physics, chemistry, and biology of the Bay required a package of
models. Transport processes were modeled by a three-dimensional hydrodynamic
model that operated independently of the water quality model. Transport informa-
tion from the hydrodynamic model was processed and stored on-line for
subsequent use by the water quality model.
The Hydrodynamic Model
CH3D-WES
The CH3D-WES (Computational Hydrodynamics in Three Dimensions—
Waterways Experiment Station) hydrodynamic model was a substantially revised
version of the CH3D model originally developed by Sheng (1986). Model formula-
tion was based on principles expressed by the equations of motion, conservation of
volume, and conservation of mass. Quantities computed by the model included
three-dimensional velocities, surface elevation, vertical viscosity and diffusivity,
temperature, salinity, and density. Details of the model formulation and initial
application to Chesapeake Bay were presented by Johnson et al. (1991).
Computational Grid
The basic equations of CH3D-WES were solved via the finite-difference
method. The finite-difference solution algorithm replaced continuous derivatives in
the governing differential equations with ratios of discrete quantities. Solutions to
the hydrodynamics were obtained using five-minute intervals for the discrete time
steps. The spatial continuum of the Bay was divided into a grid of discrete cells. To
achieve close conformance of the grid to Bay geometry, cells were represented in
curvilinear rather than rectangular coordinates. A z-plane grid was employed in
-7-
-------
Chapter 2 • Coupling with Hydrodynamic Model
which the number of vertical layers varied depending on local depth. Velocities
were computed on the boundaries between cells. Temperature, salinity, and density
were computed at the center of each cell.
The computational grid extended from the heads of tide of major bay tribu-
taries out onto the continental shelf (Figure 2-1). The shelf portion of the grid (30 x
150 km) was sized to capture large-scale circulation features outside the bay
mouth. The grid contained 2961 cells, roughly 4 km2 in area, in the surface plane
(Figures 2-1, 2-2). Number of cells in the vertical ranged from one to nineteen
(Figure 2-3). The maximum depth was in the shelf portion of the grid. Within the
bay, maximum number of layers was seventeen. Surface cells were 2.14 m thick at
mean tide. Variations in surface level caused by tides, wind, and other forcings
were represented by computed variations in thickness of the surface layer. Thick-
ness of all sub-surface cells was fixed at 1.53 m. Total number of cells in the grid
was 12,920.
Figure 2-1. The computational grid
-8-
-------
Chapter 2 • Coupling with Hydrodynamic Model
Figure 2-2. The computational grid in transformed coordinates.
Chesapeake Bay — Computational Grid
12320 Cells C2961 Surface CHIsJ
IE? layers deep
Surface layer 2.2S Meters deep
Sub—surface layer 1.525 Meters deep
Bad - 16.17.13,19 layflrs
Magentn - 11,12.13.1+.1S layer
Yellon - 6,7,8,9,10 la/ers
Cyan - 3,4,5 layers
Blue — 2 layers
GraBn - 1 layar
Figure 2-3. Number of cells in water column
-9-
-------
Chapter 2 • Coupling with Hydrodynamic Model
Calibration and Verification
The hydrodynamic model was calibrated and verified against a large body of
observed tidal elevations, currents, and densities. Details of the calibration proce-
dure were presented by Johnson and Nail (2001).
Linkage to the Water Quality Model
Hydrodynamics for employment in the water quality model were produced for
ten years, 1985-1994. Each year was a single, continuous production run. Initial
temperature and salinity for the first year were derived from observations. The
initial hydrodynamic field was obtained from a five-day spin-up period. Thereafter,
initial conditions for each year were taken from conditions computed at the end of
the previous year. Consequently, the hydrodynamic simulation was effectively a
continuous ten-year run, initialized only once.
Computed flows, surface elevations, and vertical diffusivities were output at
two-hour intervals for use in the water quality model. The two-hour hydrodynamics
were determined as arithmetic means of hydrodynamics computed on a five-minute
basis. The use of intra-tidal hydrodynamics contrasted with the earliest model
application (Cerco and Cole 1994) in which Lagrangian-average hydrodynamics
were stored at 12.4-hour intervals (Dortch et al. 1992).
The algorithms and codes for linking the hydrodynamic and water quality
models were developed over a decade ago (Dortch 1990) and have been tested and
proved in extensive applications since then. In every model application, the linkage
is verified by comparing transport of a conservative substance computed in each
model. For estuarine applications, the transport tests commonly take the form of
comparing salinity computed by two models. In the present application, salinity
comparisons verify that the hydrodynamic and water quality models are correctly
coupled (Figures 2-4, 2-5) and that transport processes in the two models are
virtually identical.
30
25
20
1-
Q.
o. 15
10
5
0
MAINSTEM BAY
Salinity
f~ "-Salinity Mainbay Surface
Summer 1985
: \\
_ -,J4K
Jl'X^-^-il
^ ^fi
1 ICM
0 100 200 300
Kilometers
Figure 2-4. Surface salinity along Chesapeake Bay axis computed by
hydrodynamic and water quality models.
-10-
-------
Chapter 2 • Coupling with Hydrodynamic Model
MAINSTEM BAY
30
25
20
0.
Q. 15
10
5
0
Salinity
: SalirtitvMairtbav Bottom
Summer 1985
- J K* T
7 K M ] i III ^
- " l4-*-|l..J j
\ ^ T "j5 ;
. *
-I
0 100 200 300
Kilometers
Figure 2-5. Bottom salinity along Chesapeake Bay axis computed by
hydrodynamic and water quality models.
Vertical Diffusion
One of the first steps in the present application was the comparison of model
results on the new grid with results on the previous grid, used for the Virginia Trib-
utary Refinements (Cerco et al. 2002). Model code, parameters, and loads were
identical. Only the grids differed. On the new grid, dissolved oxygen computed at
the bottom of the bay differed substantially from computations on the previous grid
(Figure 2-6). Origin of the discrepancy was eventually traced to the algorithm used
to compute vertical diffusion in the hydrodynamic model. The algorithm was
undergoing revision as part of the application to a new grid and a general re-verifi-
cation of the hydrodynamic model.
Once the importance of the vertical diffusion algorithm was apparent, sensi-
tivity analyses were conducted within both the hydrodynamic and water quality
models. The models were run for two years, 1985 and 1986, and assessments were
largely based on seasonal-average salinity and dissolved oxygen computations.
Formulation
The hydrodynamic model computed vertical viscosity (or friction) from first
principles using a two-equation turbulence closure scheme (Johnson et al. 1991).
Vertical diffusivity was determined through the Prandtl number. The Prandtl
number is the ratio of turbulent momentum transport to turbulent mass transport. In
unstratified water, the Prandtl number is commonly taken as unity. In stratified
water, turbulent mass transport is retarded more than momentum transport. A
-11 -
-------
Chapter 2 • Coupling with Hydrodynamic Model
MAINSTEM BAY, 12920 VS 10196 Cells
8
7
6
_j 5
"Si
£ 4
3
2
1
0
: DO Ma in bay Bottom /\
'- 1 985
- : i |
\ ' \ I '•
-\ i I I • f |
L * ' T '
~ ' * M
: - T | |
" -
i *
* . *
* " i
i '
-• . f
* , f ('-
1 * • H I 4
:, ,,,,.,,,,-,,. 14 \j 1, , ,
0 100 200 300
Kilometers
Figure 2-6. Bottom dissolved oxygen computed along bay axis on two grids.
variety of formulations have been proposed to describe the effect of stratification
on Prandtl number. Formulations tested in the hydrodynamic model included the
Bloss formulation (Bloss et al. 1988):
Az
%-TTIT^ <2-»
in which:
Dz = vertical diffusivity (cm2 s"1)
Az = vertical viscosity (cm2 s"1)
Rj = Richardson number
and the Munk-Anderson relationship (Munk and Anderson 1948):
Dz = Az
Ri)
1/2
x3/2
(2-2)
^ • Ri
The Richardson number is defined:
_ g Sp / SZ
Ri = --
p (SU / SZ)2
(2-3)
in which:
g = gravitational acceleration (cm2 s"1)
p = density of water (g cm"3)
U = horizontal velocity (cm s"1)
Z = vertical distance (cm)
-12-
-------
Chapter 2 • Coupling with Hydrodynamic Model
The Richardson number represents the ratio of turbulence suppression, via
vertical density gradient, to turbulence creation, via velocity shear. For any
Richardson number, the Bloss relationship provides lower vertical diffusivity than
the Munk-Anderson relationship (Figure 2-7).
Q
1 1.5 2
Richardson*
Figure 2-7. The effect of Bloss and Munk-Anderson formulations on vertical
diffusivity.
Hydrodynamic Model Results
Within the mainstem of Chesapeake Bay, visual assessment indicated the Bloss
relationship provided more favorable salinity computations (Figure 2-8). Salinity
computed using the Munk-Anderson relationship averaged 1.5 ppt lower than using
the Bloss relationship (Table 2-1). The mean difference between computed and
observed salinity was lower for the Bloss relationship although the root-mean-
square salinity error computed using the Bloss and Munk- Anderson relationships
was equivalent.
Within the tributaries, the difference in computed salinity based on the alternate
formulations was difficult to judge. Results suggested the Munk-Anderson relation-
ship provided superior results in the James River (Figure 2-9). Elsewhere, no
relationship was clearly superior to the other (Figures 2-10 to 2-13).
Water Quality Model Results
Computed summer bottom dissolved oxygen was universally 1 gm m~3 lower
using the Bloss relationship versus Munk-Anderson. In the bay (Figure 2-14) and
Potomac (Figure 2-18), the Bloss relationship clearly provided superior computa-
tions of bottom dissolved oxygen. In the James, the higher dissolved oxygen
computations computed by the Munk-Anderson relationship were closer to
observed (Figure 2-15). Elsewhere, neither relationship could be judged clearly
superior (Figures 2-16, 2-17, 2-19).
-13-
-------
Chapter 2 • Coupling with Hydrodynamic Model
Salinity
Julian day 24.3
;n Salinity Mainbay Surface
-70 0 70 MO 210 280 350
Kilo meters
Salinity
Julian doy 243
SoEinity Mainboy Bottom
-70 0 70 MO 210 2BO 350
Kilometers
Bloss
Solimty
Julian day 24.5
Salinity Moinbay Surface
-70 0 70 140 Z!0 280 350
Kilometers
Salinity
Julian doy 243
Sarinity Moinboy Bottom
-'0 0 70 MO 210 ?BO 350
Kilometers
Munk-Andersoii
Figure 2-8. Computed salinity along the axis of Chesapeake Bay, summer 1985,
for the Bloss (left) and Munk-Anderson (right) relationships.
Table 2-1. Statistical Comparison of Computations on Different
Grids and with Different Formulations for Prandtl Number
10,000 cell, VA
Trib
Refinements
12,000 cell with
Bloss scheme
12,000 cell with
Munk-Anderson
scheme
Salt, ppt
Mean Error
-0.99
-0.27
1.1
RMS Error
2.78
2.78
2.82
DO, mg/L
Mean Error
0.005
-0.057
-0.45
RMS Error
1.88
1.97
1.9
Temperature, C
Mean Error
0.74
0.69
0.67
RMS Error
1.93
1.77
1.65
-14-
-------
Chapter 2 • Coupling with Hydrodynamic Model
Salinity
Juhon day 245
Sciiniiy James Surface
SalmUy
Julian day 243
Salinity James Surface
-25 0 25 50 ?S 500 125 ISO 175
Kl'ometers
Salinity
Julian day 243
Salinity James Bottom
- )f> 0 2b JG 75 :sOO 125 150 i75
Kilometers
Sphnity
Julian day 243
Salinity James Bottom
- 2 5 0 25 50 75 100 12 5 150
Kilometers
Bloss
-25 0 25 50 75 100 125 150 175
Kilometers
Munk-Anderson
Figure 2-9. Computed salinity along the James River axis, summer 1985, for the
Bloss (left) and Munk-Anderson (right) relationships.
Salinity
Julian day 243
Saiinity York Surface
Salinity
Jufion day 243
Salinity York Surface
i/S-
o-
0 25 SO 75 100 125
K^orneters
Saiitlity
Julian day 243
Salinity York Bottom
-25 0 K 50 75 100 125
Kilometers
Bloss
Munk-Anderson
Figure 2-10. Computed salinity along the York River axis, summer 1985, for the
Bloss (left) and Munk-Anderson (right) relationships.
-15-
-------
Chapter 2 • Coupling with Hydrodynamic Model
Salinity
Julian day 243
Salinity Rap Surface
Salinity
Julian day 243
£i Salinity Rap Surface
-40 0 40 80 120 160
Kifometers
BO 120 160
K?lo meters
Salinity
Julian day 243
Salinity Rap Bottom
10 SO 120 WO
Ki'orneters
Bloss
0 <3Q 80 120 160
Kilometers
Munk-Anderson
Figure 2-11. Computed salinity along the Rappahannock River axis, summer 1985,
for the Bloss (left) and Munk-Anderson (right) relationships.
Salinity
Julian doy 243
£_, Salinity Potomac Surface
TV,
-60 0 60
Kilometers
Salinity
Julian day 243
Salinity Potomac Bottom
a.
o_ 0
-60 0 60 120
Kilometers
Salinity
Julian day 243
Solinity Potomac Bottom
-60 0 60 120 ISO
Kilometers
Bloss
0 60 120
Kilometers
Munk-Anderson
Figure 2-12. Computed salinity along the Potomac River axis, summer 1985, for
the Bloss (left) and Munk-Anderson (right) relationships.
-16-
-------
Chapter 2 • Coupling with Hydrodynamic Model
Salinity
Julian day 243
Salinity Potuxerst Surface
Salinity
Julian doy 243
aNnity Patuxent Surface
-10 0 10 20 30 40 50 60
Kifometers
Salinity
Julian -day 243
tn Salinity Potuxent Bottom
-10 0 10 2O 3Q 40 50 50
Kilometers
Salinity
Julian day 243
Salinity Paluxent Bottom
-10 0 10 20 30 40 50 60
Kiforneters
Bloss
-10 o n ;s 30 40 ;o 60
Kilometers
Munk-Anderson
Figure 2-13. Computed salinity along the Patuxent River axis, summer 1985, for the
Bloss (left) and Munk-Anderson (right) relationships.
"6) 4
E
MAINSTEM BAY
DO Mainbay Bottom
Summer 1 985
100 200
Kilometers
300
Figure 2-14. Computed bottom dissolved oxygen along the axis of
Chesapeake Bay, summer 1985, for the Bloss and Munk-Anderson rela-
tionships.
-17-
-------
Chapter 2 • Coupling with Hydrodynamic Model
11
10
8
7
_J
Bb fi
£
4
3
'i
1
0
JAMES RIVER, Bloss
Dissolved Oxygen
DO James Bottom
Julian Day 233
K ;/.-•..' ;.../
o so
Kitom
tr
(Cruise)
*
n
1C)
Q
8
"S) 6
£
1
0
ON NEW GRID ;
U
JAMES RIVER, SENS47 (Cruise) ON NEW GRID
Dissolved Oxygen
DO James Bottom
r juliar; Day 233
r ........••• . »
-. ... /. ••». -•
Kilometers
Figure 2-15. Computed bottom dissolved oxygen along the James River
axis, summer 1985, for the Bloss (left) and Munk-Anderson (right)
relationships.
YORK RIVER, Bloss (Cruise) ON NEW GRID
Dissolved Oxygen
DO fork Bottom
YORK RIVER, SENS47 (Cruise) ON NEW GRID
Dissolved Oxygen
00 York Bottom
Julian Day ?33 .
Figure 2-16. Computed bottom dissolved oxygen along the York River
axis, summer 1985, for the Bloss (left) and Munk-Anderson (right)
relationships.
-18-
-------
Chapter 2 • Coupling with Hydrodynamic Model
9
8
_j b
4
1
0
RAPPAHANNOCK RIVER, Bloss (Cruise) ON NEW G£
Dissolved Oxygen
DO Rap Bottom
• Julian 0 ay 233
*
n
: ' • ' D
# V""
1 *
i , .i
^ Kilon
n
u
&
a
7
£ 4
1
0
RAPPAHANNOCK RIVER, SENS47 (Cruise) ON NEW
Dissolved Oxygen
DO Rap Bottom
Julian Uay 2 J3
; * *
-
o M) 100 i.yj
Kilometers
Figure 2-17. Computed bottom dissolved oxygen along the Rappahannock River
axis, summer 1985, for the Bloss (left) and Munk-Anderson (right) relationships.
POTOMAC RIVER, Bloss (Cruise) ON NEW GRID
Dissolved Oxygen
DO Potomac Bottom
Julian Day 233
0.
£
POTOMAC RIVER, SENS47 (Cruise) ON NEW GRID
Dissolved Oxygen
DO Potomac Bottom
Figure 2-18. Computed bottom dissolved oxygen along the Potomac River axis,
summer 1985, for the Bloss (left) and Munk-Anderson (right) relationships.
-19-
-------
Chapter 2 • Coupling with Hydrodynamic Model
PATUXENT RIVER, Btoss (Cruise) ON NEW GRID
Dissolved Oxygen
DO Patuxent Bottom
Julian Day 233
PATUXENT RIVER, SENS47 (Cruise) ON NEW GRID
Dissolved Oxygen
DO Patuxent Bottom
Figure 2-19. Computed bottom dissolved oxygen along the Patuxent
River axis, summer 1985, for the Bloss (left) and Munk-Anderson (right)
relationships.
The Decision
The model team proposed using a mixed scheme in which the Bloss and Munk-
Anderson relationships were applied in the systems for which they provided the
best results. The first author's opinion was (and remains) that the use of the mixed
scheme was entirely appropriate. The existence of numerous relationships,
including Bloss, Munk-Anderson and others, indicates that no one relationship has
universal applicability. The Hydrodynamics Expert Panel insisted that the mixed
scheme should not be used. The Panel's opinion was that one relationship must be
applied system-wide. Their recommendation was accepted and the Bloss scheme
was selected for use. This scheme provided the best results in two systems, the
mainstem bay and the Potomac, for which computation of bottom-water anoxia was
crucial. Since dissolved oxygen was not a management issue in the James, the
consequences of employing the Bloss scheme there were considered acceptable.
References
Bloss, S., Lehfeldt, R., and Patterson, J. (1988). "Modeling turbulent transport in stratified
estuary," Journal of 'Hydraulic Engineering, 114(9), 1115-1133
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report El-94-4, US Army Engineer Waterways Experiment Station, Vicks-
burg MS.
Dortch, M. (1990). "Three-dimensional Lagrangian residual transport computed from an
intratidal hydrodynamic model," Technical Report EL-90-11, US Army Engineer Water-
ways Experiment Station, Vicksburg MS.
-20-
-------
Chapter 2 • Coupling with Hydrodynamic Model
Dortch, M., Chapman, R., and Abt, S. (1992). "Application of three-dimensional
Lagrangian residual transport," Journal of Hydraulic Engineering, 118(6), 831-848
Johnson, B., Kim, K., Heath, R., and Butler, L. (1991). "Development and verification of a
three-dimensional numerical hydrodynamic, salinity and temperature model of Chesapeake
Bay," Technical Report HL-91-7, US Army Engineer Waterways Experiment Station, Vicks-
burg, MS.
Johnson, B., and Nail, G. (2001). "A 10-year (1985-1994) simulation with a refined three-
dimensional numerical hydrodynamic, salinity, and temperature model of Chesapeake Bay
and its tributaries," June 2001 Draft Report, Coastal and Hydraulics Laboratory, US Army
Engineer Research and Development Center, Vicksburg MS.
Munk, W., and Anderson, E. (1948). "Notes on the theory of the thermocline," Journal of
Marine Research, 1, 276-295.
Sheng, P. (1986). "A three-dimensional mathematical model of coastal, estuarine and lake
currents using boundary-fitted grid," Report 585, ARAP Group of Titan Systems, Princeton
NJ.
-21 -
-------
Boundary Conditions
Introduction
Boundary conditions must be specified at all open edges of the model grid.
These include river inflows, lateral flows, and the ocean interface. Numerical treat-
ment of all boundary conditions is identical. When flow across a boundary is out of
the system, the concentration at the boundary is assigned the value computed at the
model cell immediately inside the boundary. When flow across a boundary is into
the system, concentration at the boundary is assigned a specified boundary value.
Diffusion across the boundary is considered to be zero. The method of assigning
boundary concentrations varies with the nature of the boundary.
River Inflows
Loads of carbon, nitrogen, phosphorus, and solids were specified at the river
inflows. Consequently, boundary concentrations of these substances were set to
zero so that no material entered the system other than the specified loads. Boundary
concentrations were required for substances not usually considered in the form of
loads. These included temperature, dissolved oxygen, phytoplankton, zooplankton,
and silica. Boundary concentrations were specified as monthly values (Tables 3-1-
3-7). Determination of the values depended on available observations.
Dissolved oxygen and temperature were based on mean monthly values in the
fall-line monitoring program. Zooplankton was not monitored at the river inflows.
Boundary conditions were based on nearby in-stream observations or adapted from
other rivers when no observations were available. Mean dissolved silica concentra-
tions were based on fall-line observations. Particulate silica concentrations were
derived from a 1994 observation program.
Limited chlorophyll observations were obtained from the fall-line data base.
These were converted to algal carbon, the model state variable, using a carbon-to-
chlorophyll ratio of 30. Phytoplankton in river inflows were assigned to the model
-22-
-------
Chapter 3 • Boundary Conditions
Table 3-1. Boundary Conditions at James and Appomattox Rivers
Month
1
2
3
4
5
6
7
8
9
10
11
12
Temperature,
oC
4.8
5.5
8.8
13.8
19.2
23.5
25.6
25
21.7
16.7
11.3
7
Chlorophyll,
ug/L
3
3
3
3
3
3
3
3
3
3
3
3
Micro Zoo, mg
C/L
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Meso Zoo, mg
C/L
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
Dissolved
Oxygen, mg/L
13.4
13
11.8
10.4
9.2
8.2
7.5
7.2
7.7
9
10.8
12.5
Dissolved
Silica, mg Si/L
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
3.4
Particulate
Silica, mg Si/L
0.16
0.26
0.19
0.15
0.18
0.1
0.07
0.15
0.04
0.05
0.08
0.11
Table 3-2. Boundary Conditions at Pamunkey River
Month
1
2
3
4
5
6
7
8
9
10
11
12
Temperature,
oC
5
6
9.2
14
19
22.9
24.6
23.7
20.4
15.7
10.7
6.8
Chlorophyll,
ug/L
3
3
3
3
3
3
3
3
3
3
3
3
Micro Zoo, mg
C/L
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Meso Zoo, mg
C/L
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
Dissolved
Oxygen, mg/L
13.4
13
11.3
9.4
7.8
6.9
6.6
6.6
7
8.1
10
12
Dissolved
Silica, mg Si/L
4.9
4.9
4.9
4.9
4.9
4.9
4.9
4.9
4.9
4.9
4.9
4.9
Particulate
Silica, mg Si/L
0.19
0.26
0.16
0.14
0.14
0.15
0.14
0.36
0.13
0.15
0.18
0.19
Table 3-3. Boundary Conditions at Mattaponi River
Month
1
2
3
4
5
6
7
8
9
10
11
12
Temperature,
oC
4.2
5.2
8.6
13.6
18.7
22.7
24.4
23.4
20
15
9.9
5.9
Chlorophyll,
ug/L
3
3
3
3
3
3
3
3
3
3
3
3
Micro Zoo, mg
C/L
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Meso Zoo, mg
C/L
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
Dissolved
Oxygen, mg/L
12.9
12.4
10.8
9.2
8
7.3
7
6.9
7.2
8.3
10.4
11.9
Dissolved
Silica, mg Si/L
3.64
3.64
3.64
3.64
3.64
3.64
3.64
3.64
3.64
3.64
3.64
3.64
Particulate
Silica, mg Si/L
0.17
0.29
0.14
0.1
0.07
0.12
0.04
0.26
0.12
0.1
0.12
0.18
-23-
-------
Chapter 3 • Boundary Conditions
Table 3-4. Boundary Conditions at Rappahannock River
Month
1
2
3
4
5
6
7
8
9
10
11
12
Temperature,
oC
3.3
4.1
7.8
13.3
19.2
23.8
26.1
25.3
21.6
16.1
10.2
5.5
Chlorophyll,
ug/L
3
3
3
3
3
3
3
3
3
3
3
3
Micro Zoo, mg
C/L
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Meso Zoo, mg
C/L
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
0.005
Dissolved
Oxygen, mg/L
13.3
13
11.9
10.5
9
7.8
7.3
7.6
8.6
10
11.5
12.7
Dissolved
Silica, mg Si/L
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
4.56
Particulate
Silica, mg Si/L
0.17
0.23
0.27
0.27
0.23
0.17
0.12
0.09
0.08
0.08
0.09
0.12
Table 3-5. Boundary Conditions at Potomac River
Month
1
2
3
4
5
6
7
8
9
10
11
12
Temperature,
oC
2.1
3.2
7.3
13.5
20
25.2
27.6
26.5
22.4
16.2
9.7
4.5
Chlorophyll,
ug/L
3
3
3
3
3
3
3
3
3
3
3
3
Micro Zoo, mg
C/L
0.01
0.01
0.01
0.01
0.02
0.02
0.04
0.04
0.02
0.01
0.01
0.01
Meso Zoo, mg
C/L
0.005
0.005
0.005
0.005
0.01
0.01
0.01
0.01
0.005
0.005
0.005
0.005
Dissolved
Oxygen, mg/L
14.7
14.4
13.1
11.4
9.7
8.5
8.1
8.3
9.1
10.6
12.3
13.8
Dissolved
Silica, mg Si/L
2
2
2
2
2
2
2
2
2
2
2
2
Particulate
Silica, mg Si/L
0.43
0.64
0.6
0.6
0.54
0.47
0.33
0.27
0.23
0.26
0.33
0.41
Table 3-6. Boundary Conditions at Patuxent River
Month
1
2
3
4
5
6
7
8
9
10
11
12
Temperature,
oC
3.8
4.6
7.8
12.6
17.7
21.7
23.6
22.8
19.6
14.8
9.7
5.7
Chlorophyll,
ug/L
3
3
3
3
3
3
3
3
3
3
3
3
Micro Zoo, mg
C/L
0.02
0.02
0.02
0.02
0.05
0.05
0.05
0.05
0.05
0.04
0.04
0.02
Meso Zoo, mg
C/L
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
0.01
Dissolved
Oxygen, mg/L
11.9
12.1
11.1
9.2
7.2
5.7
5
5
5.6
6.7
8.4
10.2
Dissolved
Silica, mg Si/L
3.7
3.7
3.7
3.7
3.7
3.7
3.7
3.7
3.7
3.7
3.7
3.7
Particulate
Silica, mg Si/L
0.65
0.65
0.65
0.65
0.65
0.65
0.65
0.65
0.65
0.65
0.65
0.65
-24-
-------
Chapter 3 • Boundary Conditions
Table 3-7. Boundary Conditions at Susquehanna River
Month
1
2
3
4
5
6
7
8
9
10
11
12
Temperature,
oC
1.9
2.3
5.8
11.7
18.3
23.8
26.9
26.6
23
17.2
10.6
5
Chlorophyll,
ug/L
5
5
5
5
5
5
5
5
5
5
5
5
Micro Zoo, mg
C/L
0
0
0.001
0.004
0.015
0.02
0.016
0.011
0.01
0.01
0.006
0.002
Meso Zoo, mg
C/L
0
0
0
0.001
0.002
0.002
0.001
0.001
0.001
0.001
0.001
0.001
Dissolved
Oxygen, mg/L
14.1
14.1
13.3
11.7
9.4
6.8
4.8
4.1
5.3
7.7
10.6
12.8
Dissolved
Silica, mg Si/L
2.2
2.14
1.98
1.8
1.25
1.08
0.94
0.93
1.22
1.28
1.77
2.39
Particulate
Silica, mg Si/L
0.53
0.53
0.54
0.55
0.42
0.4
0.37
0.35
0.44
0.39
0.49
0.65
Group 3 (summer, green algae) except at the Potomac River. There, during summer,
half the boundary concentration was assigned to model Group 1 (blue-green algae).
Special consideration was required due to the model treatment of internal algal
nutrients. Within the water quality model, these were quantified implicitly, by ratio
to algal carbon. Within the watershed model, these were explicitly quantified as
part of the organic nitrogen and phosphorus loads. To keep the algal nutrients from
being added to the model twice, once explicitly as organic load and once implicitly
by ratio to algal carbon, the implicit algal nutrients were subtracted from the
explicit loads. For nitrogen, the quantity removed from each daily organic nitrogen
load was:
AlgalN = Anc • B • Q (3-1)
in which
Algal N = nitrogen load associated with algal carbon (g d"1)
Anc = algal nitrogen-to-carbon ratio (g N g"1 C)
B = algal carbon (g m~3)
Q = river inflow (m3 d"1)
An analogous quantity was removed from the organic phosphorus load.
Lateral Inflows
Lateral inflows, determined by the watershed model, entered at model cells
adjoining the shoreline. Dissolved oxygen in the lateral flows was at saturation
concentration. Temperature was equilibrium temperature. Dissolved and particulate
silica were specified at 3.67 and 0.19 g Si m"3, respectively. These were based on
system-wide characteristic values for river inflows. Concentrations of all other
substances in lateral flows were zero.
-25-
-------
Chapter 3 • Boundary Conditions
Ocean Boundary Conditions
Background
In the first version of the model (Cerco and Cole 1994), the open edge of the
model grid was at the entrance to Chesapeake Bay. Boundary concentrations were
initially based on an array of Chesapeake Bay Program monitoring stations coinci-
dent with the edge of the model grid. Inspection of the short record then available
suggested nitrogen and phosphorus concentrations at the mouth were declining in
response to load reductions within the bay. A requirement was imposed that the
model boundary conditions should respond to conditions within the bay. Conse-
quently, "mass balance" boundary conditions were developed for nitrogen and
phosphorus. The mass-balance boundary conditions determined the inflowing
concentrations of nitrogen and phosphorus as a weighted sum of outflowing
concentrations and concentrations in a hypothetical oceanic reservoir.
The mass-balance boundary conditions were successful but unnecessary. Sensi-
tivity runs (Cerco and Cole 1994) indicated the inflowing nutrient concentrations at
the mouth declined by only 5% (phosphorus) to 10% (nitrogen) following load
reductions of 90%. Moreover, the apparent reduction in nitrogen concentration at
the mouth was attributed to changes in methodology. (The reduction in phosphorus
remains unexplained to this day.)
In the Tributary Refinements phase of model development (Cerco et al. 2002),
the grid was extended beyond the bay mouth, out onto the continental shelf. The
primary objective was to ensure that boundary conditions specified at the edge of
the grid were beyond the influence of conditions within the bay. The grid extension
traded one set of problems for another. Model boundaries were moved from a loca-
tion with abundant observations to multiple locations at which little information
was available. Surveys conducted from May 1995 to August 1996 provided infor-
mation on key water quality constituents but detailed boundary conditions for the
simulation period, 1985-1994, could not be reconstructed. Boundary conditions for
major nitrogen and phosphorus constituents were specified as constant concentra-
tions throughout the simulation period. As a consequence, mean nitrogen and
phosphorus concentrations at the mouth of the bay were well represented but the
dynamics, especially of nitrogen, were not.
In the present phase of the model study, specification of boundary conditions at
the edge of the grid proved especially problematic. During the re-evaluation of the
hydrodynamic model and refinement of water quality model kinetics, unpredictable
behavior occurred in the portion of the grid outside the bay mouth. Both models
were in a transient state of development yet both models were expected to compute
dynamics in a region for which no data existed. Circulation and water quality
computed at the bay mouth were unsuited for calibration of the models within the
bay. For the water quality model, the strategy was developed in which conditions
observed at the bay mouth were extended to the edges of the grid. Kinetics were
disabled outside the bay mouth to prevent substance transformations. This strategy
carried over into the final calibration of the present model.
-26-
-------
Chapter 3 • Boundary Conditions
Assignment
The open boundary of the grid was divided into three faces (Figure 3-1). Obser-
vations from the northernmost monitoring station (CB7.4N) were imposed at the
northern face. Observations from the central monitoring station (CB7.4) were
imposed at the eastern face. Observations from the southernmost monitoring station
(CBS. IE) were imposed at the southern face.
Observations at the mouth were conducted once or twice per month. Salinity,
temperature, and dissolved oxygen were measured in-situ, at one-meter intervals,
from surface to bottom. Samples for laboratory analysis were collected one meter
below the surface and one meter off the bottom. At the center station (CB7.4), addi-
tional samples were collected above and below the pycnocline. For boundary
condition specification, samples were averaged by month and into three layers. The
surface mixed layer comprised the upper four model cells (depth < 6.7 m). The
pycnocline comprised the next four cells ( 6.7 m < depth < 12.8 m). The bottom
mixed layer comprised all cells below the upper eight (depth > 12.8 m). Missing
surface and bottom observations were filled with long-term monthly mean values.
Missing pycnocline observations were filled by linear interpolation between surface
and bottom.
CB7.4N
CB7.4
Northern Face
Eastern
Face
CB8.1E
1
Southern Face
Figure 3-1. The Ocean Boundary Region Including Monitoring Stations, Transect,
and Boundary Faces.
-27-
-------
Chapter 3 • Boundary Conditions
Although the principles for assigning boundary conditions were straightforward,
refinements to the observations were required and occasional exceptions to the
process occurred. Salinity boundary conditions were transferred exactly from the
hydrodynamic model (Johnson and Nail 2001) to ensure agreement in salinity
computations between the hydrodynamic and water quality models. Long-term
mean monthly values of dissolved organic and particulate phosphorus were
employed in 1985 and 1986 to overcome apparent analytical errors in the early
observations.
The model required specification of algal boundary conditions as carbonaceous
biomass. We assumed this biomass comprised 90% of the observed particulate
organic carbon. From November through May, phytoplankton at the ocean
boundary were assigned to the spring diatom group. In the remaining months,
oceanic phytoplankton were assigned to the summer, mixed group.
Observations of organic particulate matter consisted of viable phytoplankton and
other organisms as well as detritus. The model state variables were detrital matter.
Observations of particulate organic carbon, particulate organic nitrogen, and partic-
ulate phosphorus were corrected to remove algal carbon and nutrients. The
remaining detrital matter was considered to be 10% labile and 90% refractory.
Regular observations were available for ammonium while the corresponding
model state variable represented ammonium plus urea. Observations near the
mouth of the bay (Lomas et al. 2002) indicated urea was 0.01 g m~3. This concen-
tration was subtracted from observed dissolved organic nitrogen and added to
observed ammonium.
Limited observations were available on which to base boundary conditions for
zooplankton and particulate biogenic silica. Microzooplankton were assigned the
constant value 0.01 g C m~3 while mesozooplankton were varied from 0.002
(winter) to 0.028 (summer) g C m~3. Particulate biogenic silica was assigned the
constant value 0.068 g Si m~3 and corrected to remove the algal component, leaving
only detrital silica.
Results
Net Circulation. Computed net circulation was examined for three years, 1985,
1990, and 1993, at a transect drawn across the mouth of the bay (Figure 3-1). The
year 1985 is considered a low-flow year although the western tributaries were
affected by a major autumn storm event. Flows in 1990 were moderate. The year
1993 was characterized by high spring runoff followed by a dry summer.
Eulerian- (or arithmetic-) average volumetric flow was computed at each flow
face in the grid transect as:
Qbar = -«jQdt (3-2)
in which
Qbar = annual average volumetric flow (m3 s"1)
Q = instantaneous volumetric flow (m3 s"1)
t = time (s)
T = averaging interval (one year)
-28-
-------
Chapter 3 • Boundary Conditions
Results (Figures 3-2 to 3-4) conform to expectations. Net flows in the deepest
portion of the transect, driven by a longitudinal density gradient, are into the bay.
Flows out of the bay are confined to surface waters. Rotational effects tilt the level
of no net motion so that flows at the northern side of the transect tend to be into the
bay at all depths. A small irregularity occasionally occurs at one depth on the
northern side (Figures 3-3, 3-4). We cannot determine if this truly reflects a
i
r
f
Virtual
LOW IN
LOW OU~
-228
-249
-64
44
108
186
51
Mean FLOW (m"3/sec) at Mouth - 1985
~
-682
-613
•314
-62
162
298
361
403
396
398
-518
-556
-259
-113
-11
98
-398
-580
-255
-115
352
15
89
97
98
98
/
F
Annual Mean FLOW (m**3/sec) at Mouth - 1990
LOW OUT
-335
-307
-68
54
131
233
72
-899
-739
-339
-11
267
417
476
501
478
463
-783
-709
-292
-76
50
168
-691
-782
-357
-180
231
-61
36
55
63
80
Figure 3-2. Net Volumetric Flows (m3 s~1) at
the Mouth of the Bay for 1985. This view is into
the Bay from the Shelf. Positive flows are into
the Bay.
Figure 3-3. Net Volumetric Flows (m3 s~1) at
the Mouth of the Bay for 1990. This view is into
the Bay from the Shelf. Positive flows are into
the Bay.
Annual Mean FLOW (m**3/sec) at Mouth -1993
FLOW OUT
1 1 T~
294
44
121
231
64
-373
-52
231
385
449
482
466
465
-308
-95
34
152
-339
-166
51
68
74
Figure 3-4. Net Volumetric Flows (m3 s~1) at
the Mouth of the Bay for 1993. This view is into
the Bay from the Shelf. Positive flows are into
the Bay.
-29-
-------
Chapter 3 • Boundary Conditions
complex circulation pattern or if this irregularity is an artifact of the grid and/or
boundary conditions. Annual net flows in and out of the bay are roughly 3,000 and
6,000 m3 s"1, respectively (Table 3-8). The computed flows are in reasonable agree-
ment with net flows, 6,000 to 8,000 m3 s"1, measured during two months in 1971
(Boicourt 1973).
Table 3-8. Volume and Mass Transport at Bay Mouth
1985
In
Out
Net
1990
In
Out
Net
1993
In
Out
Net
Flow,
m3 sec"1
3262
-5025
-1762
3784
-6637
-2854
3707
-6564
-2857
DIN,
ton d 1
15
-54
-38
17
-40
-23
15
-58
-43
Total N,
ton d 1
89
-202
-113
87
-258
-171
91
-277
-186
DIP,
ton d 1
5.2
-4.3
0.9
5.5
-4.7
0.8
4.1
-4.7
-0.5
Total P,
ton d"1
10.2
-8.3
1.9
10.3
-10.3
-0.0
9.7
-9.5
0.2
Salinity Structure. Annual-mean computed salinity at the mouth (Figures 3-5 to
3-7) is influenced by the effects of density, rotation, and boundary conditions. The
rotational effects produce a tilting of the computed pycnocline. In the upper four or
five model layers (depth < 7 m), the water on the south side is fresher than water
on the north side at the same depth. This phenomenon is apparent in the observa-
tions as well. The observations suggest the depth of the computed freshwater plume
should be greater than the present representation but a more detailed analysis is
required. The apparent disparity in plume depths may be an artifact of comparing
averages of discrete observations with continuous model results.
In deeper waters (depth > 7 m), salinity computed on the north side is less than on
the south side at the same depth. This phenomenon is also suggested in the observa-
tions. The lower-salinity water originates in a plume that hugs the Delmarva peninsula
and enters the bay at the northern side of the mouth. Salinity boundary conditions on
the northern face, adjacent to the coast, were related to flow in the Delaware River
(Cerco et al. 2002) and were lower than elsewhere on the grid perimeter.
Computed Concentrations at Bay Mouth. Selected comparisons of computed
and observed concentrations for Station CBS. IE (Figure 3-1) are presented here.
This station is in a channel (20 m) on the southern side of the bay mouth. At this
location, net flow in the surface water (depth < 7 m) is usually out of the bay while
net flow in deeper water is into the bay (Figures 3-2 to 3-4). Complete time series
comparisons for all substances and stations across the mouth are included in an
appendix to this report.
-30-
-------
Chapter 3 • Boundary Conditions
Salinity (PPT) - Annual Average 1985
Figure 3-5. Annual-Average Computed
and Observed Salinity at the Mouth of
the Bay for 1985. This View is into the
Bay from the Shelf. Observations
(Circles) are Superimposed on Model
Values. Bottom observations were
collected in deep channels not entirely
represented by the model grid.
Salinity (PPT) -Annual Average 1990
Figure 3-6. Annual-Average Computed
and Observed Salinity at the Mouth of
the Bay for 1990. This View is into the
Bay from the Shelf. Observations
(Circles) are Superimposed on Model
Values. Bottom observations were
collected in deep channels not entirely
represented by the model grid.
Salinity (PPT) - Annual Average 1933
Figure 3-7. Annual-Average Computed
and Observed Salinity at the Mouth of
the Bay for 1993. This View is into the
Bay from the Shelf. Observations
(Circles) are Superimposed on Model
Values. Bottom observations were
collected in deep channels not entirely
represented by the model grid.
-31 -
-------
Chapter 3 • Boundary Conditions
The time series (Figures 3-8 to 3-29) compare instantaneous observations to
daily-average model computations for the surface and bottom samples. While the
time series allow the formation of quick judgements for some substances, forma-
tion of even qualitative opinions for other substances is difficult. Consequently,
statistical summaries (Table 3-9) are provided for key parameters. The summaries
are based on one-to-one comparisons of all observations at the three sample
stations across the mouth. For each sampling, multiple observations in a single
model cell, if any, were averaged and then compared to computations in the same
cell, averaged over the sample day.
The summaries show remarkable agreement between observed and modeled
salinity at the mouth. The model provides near perfect representation of mean
dissolved inorganic phosphorus and is within ten percent of mean observed total
phosphorus. Modeled mean total nitrogen at the mouth is high by roughly 20%
with the excess split between dissolved inorganic and organic forms. Computed
mean silica and chlorophyll at the mouth exceed observed by 50% or more but
exact specification of these boundary conditions is not considered critical.
Final Calibration -SENS 136
Salinity CBB.1 E Surface
Final Calibration - SENS 136
Salinity CBB.1E Bottom
0 I 2 3
S 6 7 8 9 10
Years
Figure 3-8. Computed and Observed Surface
Salinity at Station CB8.1E.
Figure 3-9. Computed and Observed Bottom
Salinity at Station CB8.1E.
Final Calibration - SENS 136
Salinity CBB.1E Surface
Final Calibration-SENS 136
Salinity CB8.1E Bottom
Years
Figure 3-10. Computed and Observed Surface
Total Suspended Solids at Station CB8.1E.
Figure 3-11. Computed and Observed Bottom
Total Suspended Solids at Station CB8.1E.
-32-
-------
Chapter 3 • Boundary Conditions
Final Calibration-SENS 136
Chlorophyll CBB.1E Surface
Figure 3-12. Computed and Observed Surface
Chlorophyll at Station CB8.1E.
Final Calibration - SENS 136
Chlorophyll CB8.1E Bottom
Figure 3-13. Computed and Observed Bottom
Chlorophyll at Station CB8.1E.
Final Calibration-SENS 136
Ammonium CB8.1 E Surface
Figure 3-14. Computed and Observed Surface
Ammonium at Station CB8.1E.
Final Calibration-SENS 136
Ammonium CBB.1E Bottom
am
0.08
OCK
_j Offli
E0.05
OtJi
0.03
0.02
0.01
2 3 •<
5 6
Years
; a 9 m
Figure 3-15. Computed and Observed Bottom
Ammonium at Station CB8.1E.
Final Calibration-SENS 136
Nitrate CBB.1E Surface
5 6
Years
Figure 3-16. Computed and Observed Surface
Nitrate at Station CB8.1E.
Final Calibration - SENS 136
Nitrate CB8.1E Bottom
Figure 3-17. Computed and Observed Bottom
Nitrate at Station CB8.1E.
-33-
-------
Chapter 3 • Boundary Conditions
Final Calibration-SENS 136
Total Nitrogen CB8.1 E Surface
08
ta
_1 O6
=•
E 05
0.4
0.3
OS
It I
U
12345
Year
7 8 9 10
Figure 3-18. Computed and Observed Surface
Total Nitrogen at Station CB8.1E.
Final Calibration - SENS 136
Total Nitrogen CB8.1 E Bottom
i
Figure 3-19. Computed and Observed Bottom
Total Nitrogen at Station CB8.1E.
Final Calibration-SENS 136
Dissolved Inorganic Phosphorus CBB.1E Surface
5 6
Years
Figure 3-20. Computed and Observed Surface
Dissolved Inorganic Phosphorus at Station
0.04
OO35
a 0:1
« 07!,
j?0.02
0015
a 01
0005
Final Calibration - SENS 136
Dissolved Inorganic Phosphorus CBB.1E Bottom
Figure 3-21. Computed and Observed Bottom
Dissolved Inorganic Phosphorus at Station
Final Calibration-SENS 136
Total Phosphorus CB8.1 E Surface
8 9 10
Figure 3-22. Computed and Observed Surface
Total Phosphorus at Station CB8.1E.
Final Calibration-SENS 136
Total Phosphorus CBB.1E Bottom
0.12
0.11
it !
0.03
O.OB
jO.07
!>
= OIK
0.04
0.03
LI u;-'
0.01
0
5
Years
Figure 3-23. Computed and Observed Bottom
Total Phosphorus at Station CB8.1E.
-34-
-------
Chapter 3 • Boundary Conditions
Final Calibration - SENS 136
Total Organic Carbon CB8.1 E Surface
Figure 3-24. Computed and Observed Surface
Total Organic Carbon at Station CB8.1E.
Final Calibration - SENS 136
Total Organic Carbon CB8.1 E Bottom
34 56
Years
Figure 3-25. Computed and Observed Bottom
Total Organic Carbon at Station CB8.1E.
Final Calibration -SENS 136
Dissolved Oxygen CBB.1 E Surface
Figure 3-26. Computed and Observed Surface
Dissolved Oxygen at Station CB8.1E.
Final Calibration - SENS 1 36
Dissolved Oxygen CBB.1 E Bottom
1 23
5
Years
89 10
Figure 3-27. Computed and Observed Bottom
Dissolved Oxygen at Station CB8.1E.
Final Calibration - SENS 136
Dissolved Silica CBB.1 E Surface
Figure 3-28. Computed and Observed Surface
Dissolved Silica at Station CB8.1E.
Final Calibration - SENS 136
Dissolved Silica CB8.1 E Bottom
Figure 3-29. Computed and Observed Bottom
Dissolved Silica at Station CB8.1E.
-35-
-------
Chapter 3 • Boundary Conditions
Table 3-9. Statistical Summary of Computed and Observed Concentrations at Bay Mouth
Constituent
Dissolved
Oxygen, mg/L
Chlorophyll, ug/L
Dissolved
Inorganic
Nitrogen, mg/L
Dissolved
Inorganic
Phosphorus, mg/L
Dissolved Silica,
mg/L
Salinity, ppt
Total Nitrogen,
mg/L
Total Phosphorus,
mg/L
Number
Obser-
vations
3909
951
983
979
994
3917
957
972
Observed
Mean
8.11
5.08
0.033
0.012
0.154
28.3
0.307
0.034
Model
Mean
8.20
7.56
0.071
0.012
0.270
28.3
0.363
0.031
Observed
Standard
Deviation
1.57
4.33
0.036
0.007
0.145
3.2
0.102
0.016
Model
Standard
Deviation
1.58
3.04
0.034
0.005
0.180
3.2
0.072
0.007
Observed
Maximum
12.80
26.90
0.249
0.042
0.962
34.4
0.717
0.158
Model
Maximum
13.50
20.74
0.189
0.033
0.928
34.1
0.655
0.066
Observed
Minimum
3.71
0.00
0.008
0.005
0.023
15.0
0.100
0.011
Model
Minimum
4.40
2.16
0.007
0.002
-0.016
15.4
0.176
0.016
Observed
Median
7.86
3.96
0.016
0.010
0.106
28.9
0.294
0.031
Model
Median
7.81
6.93
0.071
0.011
0.289
29.1
0.356
0.030
Net Transport. Computed net transport of key nutrients at the bay mouth was
examined for the same three years as net circulation. Eulerian- (or arithmetic-)
average transport of key nutrients was computed for each flow face in the grid
transect as:
F = —
C dt
(3-3)
in which:
F = net flux (kg s"1)
C = concentration (kg m~3)
Results (Table 3-9) indicate the bay exports 90 tons nitrogen per day, largely in
dissolved organic form. This quantity agrees closely with the export of 86 tons
N d"1 computed in the original model (Cerco and Cole 1994) and with an export of
126 tons N d'1 estimated by Boynton et al. (1995).
Computed net transport of phosphorus at the bay mouth is into the bay at a mean
rate less than 1 ton phosphorus per day (Table 3-9). No form, organic or inorganic
predominates. The present import is lower than both the previous model computa-
tion, 7.4 tons P d"1 (Cerco and Cole 1994), and the estimate based on a
system-wide nutrient budget, 11 tons P d"1 (Boynton et al. 1995). Explanation for
the decreased import may lie in underestimation of phosphorus in water entering
the bay through the mouth. While total phosphorus is well represented in an
average sense, peak concentrations at the bottom are missed (Figure 3-23).
Improved representation of these peaks should result in greater phosphorus trans-
port into the bay.
-36-
-------
Chapter 3 • Boundary Conditions
A Recommendation
Extension of the grid onto the continental shelf had two objectives. The first was
to move nutrient boundary conditions to a location beyond the influence of loads
within the bay. The second was to allow for coupling with a proposed continental
shelf model. The first objective was met, albeit with trade-offs. The proposed shelf
model has been postponed indefinitely.
In retrospect, extension of the boundaries to avoid influence of bay loads was
unnecessary. From 70% to 85% of the mix of water at the bay mouth originates in
the sea rather than in rivers, as evidenced by typical salinities of 25 to 30 ppt.
Riverine and point-source loads are attenuated during transport such that a 90%
load reduction results in only 5% to 10% concentration changes at the bay mouth
(Cerco and Cole 1994). The extension of the grid produced enormous difficulties
for both the hydrodynamic and water quality modeling teams and certainly did not
increase the accuracy of either model. Consequently, we recommend the boundary
be restored to the mouth of the bay in future model efforts.
Grid refinements also produce enormous difficulties. The present boundary
condition "works." The grid should not be refined solely to alter the location of the
boundaries. If future modeling requires grid changes, however, consideration
should be given to restoration of the boundary to the bay mouth.
References
Boicourt, W. (1973). "Circulation of water on the continental shelf from Chesapeake Bay to
Cape Hatteras," Ph.D. diss., Johns Hopkins University, Baltimore, MD.
Boynton, W., Garber, J., Summers, R., and Kemp, W. (1995). "Inputs, transformations, and
transport of nitrogen and phosphorus in Chesapeake Bay and selected tributaries,"
Estuaries, 18(1B), 285-314.
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report El-94-4, US Army Engineer Waterways Experiment Station, Vicks-
burg MS.
Cerco, C., Johnson, B., and Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model," ERDC TR-02-4, US Army Engineer Research and Development Center, Vicksburg
MS.
Johnson, B., and Nail, G. (2001). "A 10-year (1985-1994) simulation with a refined three-
dimensional numerical hydrodynamic, salinity, and temperature model of Chesapeake Bay
and its tributaries," June 2001 Draft Report, Coastal and Hydraulics Laboratory, US Army
Engineer Research and Development Center, Vicksburg MS.
Lomas, M., Trice, T, Gilbert, R, Bronk, D., and McCarthy, J. (2002). "Temporal and spatial
dynamics of urea concentrations in Chesapeake Bay: biological versus physical forcing,"
Estuaries, submitted.
-37-
-------
Hydrology and Loads
Hydrology
Major sources of freshwater to the Chesapeake Bay system are the Susquehanna
River, to the north, and the Potomac and James Rivers to the west (Figure 4-1). Of
these, the Susquehanna provides by far the largest flow fraction (64% of total
gauged flow) followed by the Potomac (19%) and James (12%). The remaining
western-shore tributaries, the
York (3%), the Rappahannock
Figure 4-1. Major tributaries of the Chesapeake
Bay.
-38-
andPatuxent (<1%)
contribute only small fractions
of the total freshwater to the
bay.
All tributaries exhibit similar
seasonal flow patterns.
Highest flows occur in winter
(December-February) and
spring (March-May). Lowest
seasonal flows occur in summer
(June-August) and fall
(September-November)
although tropical storms in
these seasons can generate
enormous flood events.
During the simulation period,
the Susquehanna exhibited peak
spring flows in 1993 and 1994
(Figure 4-2). The summer of
1994 was also a high-flow
season although, by contrast,
-------
Chapter 4 • Hydrology and Loads
onnn
1 Rnn
I ouu
^ cnn
I DUU
T3
C A Af\f\
O 1 4UU
u
dl
3; -i onn
"35
41" 1 nnn
(U
P onn
*- OUU
.0
•Q onn
3 DUU
U
/inn
onn
D Winter
•
D
n
Spring
Summe
Fall
-
-
;r
-,
-
n
L
]
ii
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 All
Figure 4-2. Susquehanna River Seasonal Median Flows 1985-1994.
summer of 1993 had below-average flows. The year 1985 was a low-flow year,
when all seasons are considered. Other years with extremely dry seasons included
1988 and 1991.
The Potomac (Figure 4-3) and James (Figure 4-4) followed the pattern of high
spring flows in 1993 and 1994. The James also exhibited a large spring pulse in
1987 that was not evident in the two other major tributaries. An enormous flood
event in November of 1985 influenced the James seasonal median but was
"averaged out" of the Potomac flow for that season. Dry seasons in the James
included the summers of 1986 and 1988. In the Potomac, the driest seasons were
1986, 1988, and 1991.
O
O
0)
JO
"5
0)
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 All
Figure 4-3. Potomac River Seasonal Median Flows 1985-1994.
-39-
-------
Chapter 4 • Hydrology and Loads
450
400
350
300
250
200
150
100
50
0
D Winter
• Spring
D Summer
DFall
nl
11
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 All
Figure 4-4. James River Seasonal Median Flows 1985-1994.
The remaining western-shore tributaries (Figures 4-5 to 4-7) exhibited individual
variation but followed the general pattern of high flows in spring 1993 and 1994.
Dry summer years varied but 1986 was consistently among the lowest flow in all
three tributaries.
120
100
8 80
-------
Chapter 4 • Hydrology and Loads
70
60
o
o
0)
0)
E
30
10
0
fh-
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 All
Figure 4-6. York River Seasonal Median Flows 1985-1994.
Median Patuxent River Flows
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 All
Figure 4-7. Patuxent River Seasonal Median Flows 1985-1994.
-41 -
-------
Chapter 4 • Hydrology and Loads
RET2
WT1
VTO
WT3
ET1
ET2
TF5
Loads
Loads to the system include distributed or nonpoint-source loads, point-source
loads, atmospheric loads, bank loads, and wetlands loads. Nonpoint-source loads
enter the system at tributary fall lines and as runoff below the fall lines. Point-
source loads are from industries and municipal wastewater treatment plants.
Atmospheric loads are from the atmosphere directly to the water surface. Atmos-
pheric loads to the watershed are incorporated in the distributed loads. Bank loads
originate with shoreline erosion. Wetland loads are materials created in and
exported from wetlands.
Nonpoint-Source Loads
Nonpoint-source loads are from Phase IV of the EPA Chesapeake Bay Water-
shed Model (WSM). The WSM is a modified version of the HSPF (Hydrologic
Simulation Program FORTRAN)
model (Bicknell et al. 1996). Docu-
mentation of the latest version of the
WSM may be found on the Chesapeake
Bay Program web site http://www.
chesapeakebay.net/modsc.htm. Loads
were provided in December 2000.
Some refinements to solids loads were
performed in February 2002 after
which loads were finalized.
Loads were provided on a daily
basis, routed to surface cells on the
model grid. Routing was based on local
watershed characteristics and on
drainage area contributing to the cell.
For reporting, loads are summed here
by Chesapeake Bay Program Segments
(GBPS, Figure 4-8). System-wide
summaries are for 1990, a year central
to the simulation period and with
typical flows. Individual segments are
selected for time-series presentation.
Nonpoint-Source Nitrogen Loads
Largest nitrogen loads, by far, come
into segment CB1 which includes the
input from the Susquehanna River
(Figure 4-9). Lesser loads enter at the
Potomac (TF2) and James (TF5) fall
lines. Magnitude of loads corresponds
to relative flows in these tributaries. In
the Susquehanna and Potomac, 70% of
the nitrogen load is in the form of
EE3
CBS
LE5
Figure 4-8. Chesapeake Bay Program Segments.
-42-
-------
Chapter 4 • Hydrology and Loads
Figure 4-9. Nonpoint-Source Total Nitrogen by GBPS, 1990.
nitrate while in the James 43% is nitrate. Segment ET6, on the lower Eastern
Shore, receives the greatest load not associated with a fall line, of which 80% is
nitrate.
Time series of major loads (Figures 4-10 to 4-13) demonstrate little correspon-
dence to flows. Correspondence may be obscured by the comparison of annual
mean loads with seasonal median flows. Above all, inspection indicates no trend in
nitrogen loading.
250000
onnnnn
^ 150000
1 nnnnn
morion
1
.1!
' 1*1
>>|.,-:
'^ i
98
5 1
TT-I
:'t
""!
* VJ
:=i
98
6 1
™
*°
V
;;
98
7 1
TTI
V'1
98
8 1
it 1
f '/
??'
98
9 1
;;'
,it!-
k
/f
"'n<
99
0 1
•7
•''.
•^•w
99
1 1
— !i
;:
/"
;>*
99
2 1
"
-,_
^-:-
|!
99
3 1
*-
^;
.,«
99
4
Figure 4-10. Annual Nonpoint-Source Nitrogen Loads in CB1, the Susquehanna
Fall Line
-43-
-------
Chapter 4 • Hydrology and Loads
80000 -
finnnn
| 4oooo
°>40
20000
1
m
IS
n
§1
i
98
.. .
5 1
y
ii'S*
JjJSJ
98
6 1
ra
•i-t
®$
•1
3
1
98
7 1
i"?
If;
*
8
jj
98
8 1
n
is
li
St
1
t$
.'of
98
9 1
—
t>*™
'>•:;.-
1
99
0 1
1
'?
ifc-p
99
1 1
us
""(•
1
8
;*
.^
99
••
2 1
;:£:
io
1
•||
:'.i:
If
99
3 1
1
s-'.f
!l
£
f:
i
99
4
Figure 4-11. Annual Nonpoint-Source Nitrogen Loads in TF2, the Potomac Fall
Line.
40000
25000
«
O)
isnnn
I UUUU
mnnn
0
•
1
p-
i*
:>|
4',
i i1
if"
A'*
'^
98
5 1
«i— «
,fl f
i£
98
6 1
•m
?»'
i:
Hi
98
7 1
fJtf"?
•^;
!*'
!_..'
98
8 1
IV
f,
•;.:j
H
H'
)"
tli
"n
?!
i.
98
r
9 1
f~
,.'•;
:1
,,<3
'•;i
_
99
0 1
' t
1-
M
J;'
t
99
1
1 1
PP-J
^
_lJ
99
2 1
J*T1
;J ,' ;
V
i ;•
Ihc^1'
llf
99
3 1
'-'-
r
*•
4:'
L*'4
99
4
Figure 4-12. Annual Nonpoint-Source Nitrogen Loads in TF5, the James Fall Line.
16000 -
14000
12000
10000 -
>,
f 8000
x.
cnnn
Af\t\n
2000
0
1
p5f
"J.
;* ..
•j.t
i|
N
98
5 1
p,
el
98
6 1
I
O.,
"•%
aL
il
1
98
7 1
^7
|l
s
98
8 1
1
®
'to
5i
II
S
38
n:,;,^
r
9 1
!i
1
I
1
S
99
0 1
t,
i,
i,(i
99
1 1
sy
11
IS
f«=J
|»
f**V
1
SI
i
?!
s
99
2 1
w
If
,;|f
^
.;$IH
||
99
3 1
1
|j;
B'V
«j|
Kj
r>v
i
99
4
Figure 4-13. Annual Nonpoint-Source Nitrogen Loads in ET6, the Lower Eastern
Shore.
-44-
-------
Chapter 4 • Hydrology and Loads
Nonpoint-Source Phosphorus Loads
As with nitrogen, the greatest phosphorus loads enter at the Susquehanna,
Potomac, and James fall lines (Figure 4-14). Less than 20% of the load in the
Susquehanna and Potomac is as dissolved phosphate. In the James, roughly 30% is
dissolved phosphate. Loads are not proportional to flows in these tributaries. In
1990, phosphorus load in the Potomac was 70% of the load in the Susquehanna
although Potomac flow was 25% of the Susquehanna. Phosphorus load from the
James amounted to half the load from the Susquehanna although flow was less than
25% of the Susquehanna. Reasons for the disparity between flow and load are not
apparent. No doubt watershed characteristics and above-fall-line point-source load-
ings play a role. A speculation is that a portion of the Susquehanna particulate
phosphorus load is retained in the Conowingo reservoir, just upstream of the fall
line. Away from the three major tributaries, the largest nonpoint-source phosphorus
load is in segment LE5, the lower James River.
Figure 4-14. Nonpoint-Source Total Phosphorus by GBPS, 1990.
Inspection of time series in the major load segments indicates no trend in the
Susquehanna (Figure 4-15), the Potomac (Figure 4-16), and the lower James
(Figure 4-18). At the James fall line (Figure 4-17, loads from 1985 to 1989 oscil-
late between highest and lowest in the ten-year record. From 1990 to 1994, the
loads settle into relatively constant, moderate values.
Results indicated no consistent difference between estimates (Figures 4-19 to
4-21) and none could be validated as true.
Nonpoint-Source Total Organic Carbon Loads
Organic carbon is not a state variable in the WSM. A variety of approaches to
obtaining nonpoint-source carbon load were investigated. These included use of a
constant concentration, determination of concentration via MVUE (minimum vari-
ance unbiased estimator) regression of concentration versus flow, and determination
of load as a ratio to WSM organic nitrogen load. Results indicated no consistent
-45-
-------
Chapter 4 • Hydrology and Loads
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-15. Annual Nonpoint-Source Phosphorus Loads in CB1, the
Susquehanna Fall Line.
12000 -
1 nnnn
;>•
ra
O)
^
onnn
0
1
—
^
•
98
5 1
„—»
98
T
6 1
n
-"'-''
98
T
7 1
1
98
8 1
=,
Ei
98
9 1
pmy
£,
99
1
0 1991 1
S
B
99
2 1
=
••
Si»
••
_
E
99
3 1
Sjjjs
99
4
Figure 4-16. Annual Nonpoint-Source Phosphorus Loads in TF2, the Potomac Fall
Line.
12000 !
10000
8000
6000
4000
2000
0
H
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-17. Annual Nonpoint-Source Phosphorus Loads in TF5, the James Fall
Line.
-46-
-------
Chapter 4 • Hydrology and Loads
1800
1400 -
S- 1000 '
T3
* 800 - pi
pnn
400 •
onn
0 •
!<•
i1
].<
1
p^l
;*i
biJ
~r
;«;
\
p.
a .
^
41
V:
>"i
^
f"
O
1
,,>:
S';
' iv
'('
1
i
»"•
^
$
r^
• i
,' /
'f;
,?'
"
kid
1 -
|
r11
^p
/v^
tf
'„
fV
its
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-18. Annual Nonpoint-Source Phosphorus Loads in LE5, the Lower
James River.
'OOOOO"DMVUE
600000 i"WSM
, " Constant
^nnnnn i
>- 400000 -
C3
^
^)
v <>rinnnn
200000 '
100000 +
0 • -
i
i
i
i
I
35
8
6
-
8
7
-
8
r
8
8
9
9
p
1
0 i
31
32
_
93
—
9
4
-
Figure 4-19. Susquehanna Total Organic Carbon Loads by Regression (MVUE),
by Ratio to Watershed Model (WSM) and by Constant Concentration (Constant).
600000 -,
500000
85 86 87 88 89 90 91 92 93 94
Figure 4-20. Potomac Total Organic Carbon Loads by Regression (MVUE), by
Ratio to Watershed Model (WSM) and by Constant Concentration (Constant).
-47-
-------
Chapter 4 • Hydrology and Loads
140000 ,
120000 -.-
>, 80000 ;
ro
3?
& 60000 -
40000 ;
20000 -
•-
85
Jr
86
[^
(i
(I
h
1
87 88
Q MVUE
• WSM
q Constant
n
—
, r
\
. *
i -4
j. L,
-,
89 90 91 92
1
—
:~
-
'
93 94
Figure 4-21. James Total Organic Carbon Loads by Regression (MVUE), by Ratio
to Watershed Model (WSM) and by Constant Concentration (Constant).
difference between the estimates (Figures 4-19 to 4-2land none could be validated
as true. Ratio to the WSM was selected since this method allowed for computation
of carbon load reductions in response to watershed management. Carbon-to-
nitrogen ratios were obtained from fall-line observations posted by the USGS
Chesapeake Bay River Input Monitoring Program http://va.water.usgs.gov/chesbay/
RIMP/). Observations were available for Maryland tributaries only. From these, a
carbon-to-nitrogen ratio of 12 was specified system-wide except for the Potomac
and Choptank. Ratios of 8 and 17.5 were specified for these fall lines, respectively.
The distribution of nonpoint-source carbon loads across the system follows the
patterns of nitrogen and phosphorus; largest loads are at the Susquehanna,
Potomac, and James fall lines (Figure 4-22).
600000
% 3UI
.•
Figure 4-22. Nonpoint-Source Total Organic Carbon by GBPS, 1990.
-48-
-------
Chapter 4 • Hydrology and Loads
Nonpoint-Source Total Suspended Solids
In 1990, the largest solids load was from the Potomac, followed by the Susque-
hanna and James (Figure 4-23). These loads show no relationship to relative flows
in the three tributaries. For suspended solids, the Conowingo reservoir acts as a
settling basin to remove solids from the Susquehanna before entering the bay
(Donoghue et al. 1989; Schubel 1968). Although solids can be scoured from the
Conowingo at high flows, the predominant effect of the reservoir is to diminish
Susquehanna solids loads relative to the other major tributaries. Away from the fall
lines, largest solids loads are received by segment LE5, the lower James River.
3500
2500
500
n
Figure 4-23. Nonpoint-Source Total Suspended Solids by GBPS, 1990.
On an annualized basis (Figures 4-24 to 4-27), solids loads show little correla-
tion to flows although these correlations must exist. The averaging methods remove
obvious dependencies. In the Potomac and James, largest annual loads occur in
1985. The loads are dominated by a November 1985 flood event that is obscured
by the median statistic. Perhaps the major conclusion drawn from the time series is
that solids loads in the major sources bear little correlation with each other.
Point-Source Loads
Point-source loads were provided by the EPA Chesapeake Bay Program Office
in December 2000. These were based on reports provided by local regulatory agen-
cies. Loads from individual sources were summed into loads to model surface cells
and were provided on a monthly basis. Despite the provision of monthly values,
loads from Virginia were most often specified on an annual basis while loads from
Maryland varied monthly. In the Potomac River, loads on the Virginia side were
-49-
-------
Chapter 4 • Hydrology and Loads
6000000
5000000
3000000
Figure 4-24. Annual Nonpoint-Source Suspended Solids Loads in CB1, the
Susquehanna Fall Line.
12000000
n
Figure 4-25. Annual Nonpoint-Source Suspended Solids Loads in TF2, the
Potomac Fall Line.
-50-
-------
Chapter 4 • Hydrology and Loads
7000000 -,
&
T3
5
i — 1
198E
|-|
1986
1987
I 1
1988
1
98E
| 1
1990 1991 1992 1993 1994
Figure 4-26. Annual Nonpoint-Source Suspended Solids Loads in TF5,
the James Fall Line.
400000 -
33
1-1
n
—
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-27. Annual Nonpoint-Source Suspended Solids Loads in LE5,
the Lower James River.
-51 -
-------
Chapter 4 • Hydrology and Loads
annual loads while loads on the Maryland side were monthly loads. Examination of
the Maryland loads indicates considerable variance (Figure 4-28) which is lacking
in the specification of Virginia loads.
For reporting, loads are summed here by Chesapeake Bay Program Segments
(CBPS, Figure 4-8). System-wide summaries are for 1990, a year central to the
simulation period. Individual segments are selected for time-series presentation.
900
700
600
ro
T3
200
100
/ /
Figure 4-28. Monthly (1994) Point-Source Phosphorus Loads to Model Cell 2846
on the Maryland Shore of the Potomac River.
Point-Source Nitrogen Loads
Point-source loads are concentrated in urban areas (Figure 4-29). Major nitrogen
loads originate in Northern Virginia/District of Columbia (TF2), Richmond Virginia
(TF5), Baltimore Maryland (WT5, WT4), and Hampton Roads Virginia (LE5). Of
these, only loads to the Patapsco (WT5, Figure 4-32) show a monotonic decreasing
trend. Loads to the upper Potomac (TF2, Figure 4-30), and upper James (TF5,
Figure 4-31) suggest a decrease after 1990. Point-source nitrogen loads to the Back
River (WT4, Figure 4-33) and the lower James (LE5, Figure 4-34) show no trend.
-52-
-------
Chapter 4 • Hydrology and Loads
a • • P
•=3- LT. r- • ^ouuu
TO
^
^>
v; ovnnn
-1- z/uuu
ocnnn
ZDUUU
ocnnn
ZDUUU
o/innn
f
f i
*.**
I,
-V' '
/,,
v
i;»
«• .' ;
j
>
,
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-30. Annual Point-Source Nitrogen Loads in TF2, Northern Virginia/District
of Columbia
-53-
-------
Chapter 4 • Hydrology and Loads
30000 -,
90000
ro
~o A cnnn
-~. 1 DUUU
O)
1 nnnn
IUUUU
1
' f
\
;;|
!:
98
5 1
t1
i'p;'
•'•?,
3
98
6 1
I
(,j'iV
7 %
JV
'I*1
98
1
7 1
liiS'i
I1!''
I
/
n k 5
'j|i;
98
8 1
L ^i
9 k
(l
|'
ii
'$
98
9 1
n
it]
iy
I "f
>
99
0 1
i V
'i1'
' '"< ]
"•'1
99
1 1
Fl
S;
If
>;',>
99
2 1
Tl
1
$i
. . !
99
1
3 1
„
K
!
J
9
I '
'I-
<'
9
!
4
Figure 4-31. Annual Point-Source Nitrogen Loads in TF5, Richmond Virginia Vicinity.
20000 -,
iftnnn
-j cnnn
1 DUUU
1 yinnn
1 4UUU
1 onnn
I ZUUU
^*
(0
T3 -i nnnn
onnn
OUUU
onnn
DUUU
yinnn
^tUUU
onnn
1
ii'i,1
>','
ii
l,>|(
'I.1!''
98
5 1
n ' '
:|
;j
:i
. ,/
ii .,
:?i
ti''(1.
^V
98
6 1
Si'^J
!','i'i
^\ *
;.;,;
;^
>yt L
j't f
. / f'LJ
tt^f
98
7 1
__
'rf; i
j^
i't * f
i':i
,iii'
.;:,;(
>''l
,ri-l
f ii
,H^'
98
8 1
ijn
t |
^
9
*s
>;i
8
9 1
ili
;j|
il[
i'i
\ u'
j
$•
99
0 1
TT— I
jj
J
:i^
!'i;i':
'•I'
u' tl'
99
i^i !r pi
-;>,.„ ,, ,': ,,",,
;';"' Iv'i; '.V'1/
1 1992 1993 1994
Figure 4-32. Annual Point-Source Nitrogen Loads in WT5, Patapsco River,
Baltimore Maryland.
-54-
-------
Chapter 4 • Hydrology and Loads
8000 -,
7nnn
Rnnn
^nnn
>,
(0
T3 /innn
^nnn
ouuu
onnn
zuuu
mnn
1
"v"""' *
X
^ •* i
98
5 1
"" ^ *i
/
,«'-\
98
6 1
1 ;"""
^
,;'
- «4
98
7 1
"T
V
', ^
•rj
98
8 1
i{ i
•?,
\'
','
98
9 1
v-r
|^
'If1*
99
0 1
__.
!-''
99
1 1
™"V"T"
99
2 1
3
V^
5^'
.,";„'
-,'_
99
3 1
'..
99
4
Figure 4-33. Annual Point-Source Nitrogen Loads in WT4, Back River, Baltimore
Maryland.
9000 -,
snnn
ynnn
cnnn
DUUU
>i cnnn
ra ouuu
^
•59 A nnn
onnn
onnn
1 nnn
I UUU
1
__
t] \
>,"•
^ \
i1
98
5 1
&'
1 ',
"5
" J
j*
') ^
:?
98
6 1
:
v
r.
^%
\
98
7 1
• '.
/
; '
•f
i
98
8 1
7™™1™
-?
;:.
-•
98
9 1
_
^ r
' \
,
"«,
99
0 1
,v
1
""
99
1 1
T"
•'
% '
99
2 1
_r_
;•
^ „ i
99
3 1
""•"""'•.
IV-
1
99
4
Figure 4-34. Annual Point-Source Nitrogen Loads in LE5, Hampton Roads Virginia.
-55-
-------
Chapter 4 • Hydrology and Loads
Point-Source Phosphorus Loads
Point-source phosphorus loads are concentrated in the same urban areas as
nitrogen although the relationship by size differs (Figure 4-35). In 1990, largest
loads were from Hampton Roads and Richmond Virginia. Northern Virginia/
District of Columbia contributed the least phosphorus load of the major urban
centers. The same area showed the least reduction of phosphorus loads over the
simulation period (Figure 4-36), possibly a consequence of load reductions prior to
1985. In the remaining urban areas, point-source phosphorus loads were halved
from 1985 to 1994 (Figures 4-37 to 4-40)
III
LLJUJLULULLI
H-I— i— i— i— LL u_ u_ u_ u_ LU i
Figure 4-35. Point-Source Phosphorus Loads by GBPS, 1990.
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-36. Annual Point-Source Phosphorus Loads in TF2, Northern Virginia/
District of Columbia.
-56-
-------
Chapter 4 • Hydrology and Loads
2500 n
onnn
1 *^nn
ro
5
1 nnn
m"in
n
::::
• — i
p 1
i i i i i i i i i
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-37. Annual Point-Source Phosphorus Loads in TF5, Richmond
Virginia Vicinity.
900 -
Ann
700
enn
DUU
>•» cnn -
5
51 400
^nn
onn
100
1
ET';
j™ ",
^':
£T'
}m»^
}m»^
98
5 1
n
„;* !ii
jjj""!1,
"""*
98
6 1
m
115""'
if
98
7 1
py^
»— -""HI
••-!
:|
;,,,^l
J-S
98
8 1
•Kij
$»•»
Si-
'»«£ "
'"<"€'
98
9 1
.
V-
,«„,
y
99
0 1
99
1 1
r|
Sf
I
i
"•C||
_!|
99
2 1
=k
J
1
a|
99
3 1
|?|-
13::
^ ,„
;:'
99
4
Figure 4-38. Annual Point-Source Phosphorus Loads in WT5, Patapsco River,
Baltimore Maryland.
-57-
-------
Chapter 4 • Hydrology and Loads
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-39. Annual Point-Source Phosphorus Loads in WT4, Back River,
Baltimore Maryland.
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-40. Annual Point-Source Phosphorus Loads in LE5,
Hampton Roads Virginia.
Point-Source Total Organic Carbon
Organic carbon is not commonly monitored or reported in point-source loads. As
part of the Tributary Refinements phase of this study, effluent from selected point
sources in Virginia was analyzed for parameters included in the model. Total
organic carbon in effluent showed enormous variation (Figure 4-41). When
carbon concentration was normalized by total nitrogen concentration, the variance
diminished but was still substantial (Figure 4-42). Based on these analyses, a
-58-
-------
Chapter 4 • Hydrology and Loads
140 -
1 9n
1 nn
-" an
o
01 en
Af\ ...
on
/
/
—
>
-------
Chapter 4 • Hydrology and Loads
70000 -
fr
-a
03
^
rwi ™ ™ ™ 1 cm m m m
' ' ' ' ! ' ! ' ' ' ' ' ' ' '
OjQjLQCDLDCDLD£DLL3LULL3l — *~~ 1 — 1— 1— 1— 1— 1 — \— \ — LLJSJJLLJSJJL
U O U O O O U O LLJ LLJ LLJ LLJ K LLJ LU LLJ LLJ LLJ LLJ LLJ LLJ _j _i _3 _i _
„ 1
i i i i i i i
J 1- (- 1- (- 1- U_ LL
jLULULULULUt— h-
LY (Y. (Y. Ct LY
m El
i i i
LL LL U
1— 1— h
• 1 ..
= ^^^^S
Figure 4-43. Point-Source Organic Carbon Loads by GBPS, 1990.
Atmospheric Loads
Atmospheric loads are deposited directly on the water surface. These originate in
precipitation (wetfall) and as particle deposition (dryfall). Daily atmospheric loads
for each surface cell were computed by the EPA Chesapeake Bay Program Office
and provided in January 2000.
Wet deposition of ammonium and nitrate was derived from National Atmos-
pheric Deposition Program observations. Data from 15 stations in the watershed,
collected from 1984-1992, was employed to relate concentration to precipitation,
month, and location. For each rainfall event in the calibration period, atmospheric
loads were computed as the product of calculated concentration and precipitation
volume.
Dry deposition of nitrate was derived from wet deposition using ratios calculated
by the Regional Acid Deposition Model (Chang et al. 1987). The atmospheric load
of organic nitrogen to water surfaces was determined as the product of concentra-
tion and rainfall volume. Deposition of organic and inorganic phosphorus was
specified on a uniform, constant, areal basis derived from published values.
For reporting, loads are summed here by Chesapeake Bay Program Segments
(CBPS, Figure 4-8). System-wide summaries are for 1990, a year central to the
simulation period.
Although areal atmospheric nitrogen loads vary on a spatial basis, total loading
to each CBPS is primarily a function of surface area; the largest segments receive
the greatest loads (Figure 4-44). These include segments in the central bay, the
-60-
-------
Chapter 4 • Hydrology and Loads
eastern embayments, and the lower portions of major western tributaries.
Variability, on an annual basis, is roughly 40% of the mean load and exhibits no
obvious trend (Figure 4-45). The spatial distribution of atmospheric phosphorus
loads (Figure 4-46) is identical to the distribution of nitrogen loads and is
temporally constant.
4000 -
>
Rf
^ ZUUU
^
1
-
n r
_ _ 0 n II 0 D n D - fl
n
. D . n -. D n D _ 0 • fl , . . _ _ 1 . . .
^ r^ ^ ^ ir> ^ f^ f^ ^ <^ ^ ^ c> c^ ^ ^ ^o '^ t^ o^ a) ^ c^ <^ ^ LO ^ c^ ^ ^ ^o ^ r^ ^ ^ 10 ^ ^ r^ ^-i ^ LO (JD \^- o^
cacoincncBffitomsuLLisui— -^i— i— I— i— i— I— i— i— LLJLJJLJJLLJLLJI— I— i— I— i— u_ u_ u_ u_ u_ LJJ tr tr tr tr t~ tr tr tr
UUUUUQOUUJLULlJ^^LlJLlJLULlJLlJLULlJLU^^^^^UJLlJLlJLUm
Figure 4-44. Atmospheric Nitrogen Loads by GBPS, 1990.
5000
4500
4000
3500
3000
2500
2000
1500
1000
500
0
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994
Figure 4-45. Annual Atmospheric Nitrogen Loads in CBS, Central Chesapeake Bay.
-61 -
-------
Chapter 4 • Hydrology and Loads
Susquehanna River
1.1 million cy/yr
Potomac River ______
1.5 million cy/yr [Virgin
James River
1.0 million cy,
Figure 4-47. Annual Shoreline Erosion in Chesapeake Bay System (US Army Engineer
District, Baltimore 1990).
-62-
-------
Chapter 4 • Hydrology and Loads
Susquehanna River
1.1 million cy/yr
Potomac River
1.5 million cy/yr
James River
1.0 million cy
Figure 4-47. Annual Shoreline Erosion in Chesapeake Bay System (US Army Engineer
District, Baltimore 1990).
(Table 4-1), we decided to consider bank loading as a spatially and temporally
uniform process. Loads to each surface cell were computed:
BankLoad = L • E • Fsc • BankAdj • C (4-1)
in which:
Bank Load = load to surface cell (kg d"1)
L = length of cell adjacent to shoreline (m)
E = erosion rate (kg solids m"1 shoreline d"1)
Fsc = fraction silt and clay in total volume eroded (0 < Fsc < 1)
BankAdj = calibration factor used to adjust bank loads (1)
C = nutrient or carbon concentration associated with solids (kg kg"1)
Only cells adjacent to the shore received loads. In portions of tributaries which
were one cell wide, the shoreline length was twice the cell length. Otherwise, the
shoreline length was the cell length. We specified erosion rate based on a mean
value and used parameter BankAdj to tune the solids concentration in the model.
Alternately, we could have specified erosion rate based on observations, but not
-63-
-------
Chapter 4 • Hydrology and Loads
NAUTICAL MILES
0 5 10 15 20 25
0 10 20 30 40 SO
KILOMETERS
Figure 4-48. Sample Sites for Bank Load Composition (from Ibison et al. 1992).
-64-
-------
Chapter 4 • Hydrology and Loads
Table 4-1
Composition of Bank Solids Loads
Mean
Median
Standard Deviation
Maximum
Minimum
N
Model
Load,
(kgm'd1)
11.4
8.6
8.8
34.0
0.8
43
5.7
Gravel
(%)
20.3
16.3
16.0
71.9
0.7
255
Sand,
(%)
17.0
16.3
14.1
60.3
0.1
255
Silt,
(%)
60.9
63.1
26.0
98.7
1.6
255
68
Clay,
(%)
1.8
0.1
5.6
41.1
0.0
255
Total
Phosphorus,
(mg/g)
0.20
0.07
0.44
5.31
0.00
255
0.68
Inorganic
Phosphorus,
(mg/g)
0.028
0.003
0.154
2.243
0.000
255
Total
Nitrogen,
(mg/g)
0.37
0.20
0.60
6.44
0.00
255
0.33
Particulate
Carbon,
(mg/g)
4.35
1.22
8.43
70.90
0.03
255
4.1
necessarily the mean, in which case parameter BankAdj would be unnecessary.
Concentration is equal to unity when loading of solids is considered.
We found an erosion rate of 5.7 kg nr1 d"1 (0.7 ton ft"1 yr1) produced reasonable
solids computations in the model. Our rate is two-thirds the median or one-half the
mean of rates reported by Ibison et al. 1992 (Table 4-1). The reported rates are
primarily for eroding banks. Our lesser rate, employed system-wide, may reflect
the lower shorelines along the eastern shore and elsewhere, relative to selected
eroding banks. The fraction of silt and clay and the carbon and nitrogen composi-
tion of the solids, as employed in the model, followed closely the observed means.
Model phosphorus composition was considerably higher than the observed mean.
The higher concentration was specified to represent an affinity of phosphorus for
silt and clay particles rather than sand and gravel.
Summary of Model Loads
The spatial distribution of bank loads (Figures 4-49^1-52) depends directly on
the length of the model shoreline in each CBPS. Segments receiving the greatest
loads include the lower estuarine Potomac (LE2), Tangier Sound (EE3), and the
tidal fresh portions of the major western tributaries (TF2-TF5).
One characteristic of bank nutrient loads is they are phosphorus-enriched relative
to organic matter. The ratio of observed median nitrogen to phosphorus in bank
loads is roughly 3:1 (Table 4-1) while the characteristic ratio in phytoplankton is
7:1 (Redfield et al. 1966). The modeled affinity of phosphorus for silt and clay
produces a nitrogen-to-phosphorus ratio in modeled loads of 0.5:1 (Figures 4-49,
4-50).
Sensitivity to Bank Loads
Sensitivity to bank nutrient loads was examined during the model calibration
process. In SENS65, nitrogen and phosphorus content of solids were as employed
in the final calibration (Table 4-1). In SENS66, these were multiplied three-fold.
-65-
-------
Chapter 4 • Hydrology and Loads
350 -,
inn
ouu
re
V -fen
1 nn
o c
n
n
n
S3 SSE
II
1
-,
R
s§Ea[3p
,n
n
°p ppppf:
1
In!
n
pi
1 n
in
ppcasisiapp
r
p
oouoou ij, o: a: or:
i
pp
UJ LU
t£ o:
n
1 JtLn .fill
ESESEIiiiillii
Figure 4-49. Total Nitrogen Bank Loads by GBPS
(£ EC K DC EC
Figure 4-50. Total Phosphorus Bank Loads by GBPS
-66-
-------
Chapter 4 • Hydrology and Loads
Figure 4-51. Total Organic Carbon Bank Loads by GBPS.
1000000
onnnnn
orinnnn
ouuuuu
700000
cnnnon
~c cnnnnn
•~ ouuuuu
n
jinnnnn
200000
100000
•
nn
B
n
n
II
__ ' iii_i ' '^_
UOOOOOOOUJUJUJUJ
1
n
1
=SEEESE
In
Lu 111
B
_
r
III
n
li
a: K &: ct oc
1
nnn
mnimmil
Figure 4-52. Suspended Solids Bank Loads by GBPS
Due to subsequent changes in parameters and inputs, these runs do not represent
the model in its calibrated state but do provide useful insights.
The effect of a three-fold increase in bank nitrogen loads on the bay is nearly
invisible (Figure 4-53) while a three-fold increase in bank phosphorus loads results
in substantial increases in bay total phosphorus (Figure 4-54). As with the bay, trib-
utary sensitivity is largely to phosphorus loads rather than nitrogen. Sensitivity is
not uniform across tributaries. Response of the Potomac (Figure 4-55) and James
(Figure 4-56) is limited. The Patuxent (Figure 4-57) and York (Figure 4-58) exhibit
moderate responses in their upper reaches while the Rappahannock (Figure 4-59)
shows high sensitivity throughout.
-67-
-------
Chapter 4 • Hydrology and Loads
175
1.5
1.JS
ors
0.5
025
0
MAINSTEM BAY, SENSES (NEW HYD
Total Nitrogen
Total N Mainbay Surface
: Julian Day 3153
ii::ri ic-j
Kilometers
SENS65 = BASE
SENS66 = LOAD X 3
R
^
0)
(Croi&e) 0
j
A
300
I '
075
0.5
0.25
a
^T
i
:M BAY, SENS66 (Cruise) ON NEW GRID
DflWI
ainbay Surface
y 3153
1
1
>--tr
^-rr^ •- • •
TIT
100 200 300
Kilometers
Figure 4-53. Sensitivity of Chesapeake Bay to Nitrogen Bank Loads
(Summer 1993).
0.07
006
OOb
0.03
00!
0.01
0
MAINSTEM BAY, SENS65 (NEW HYDRO) (Cruise) 0
Total Phosphorus
Total P Mainbay Surface
Julian Day 3153
100
Kilometers
SENS65 = BASE
SENS66 = LOAD X 3
300
lioM
om
003
001
0
i — -.
STEM BAY, SENS6
Phosphorus
P Mainbay Surface
iDay 3151
j
|1|
6 (Cruise
•:
)0
N NEW GRID
^
1 . , . . 1 , . , , 1 III
0 100 200 300
Kikmwters
Figure 4-54. Sensitivity of Chesapeake Bay to Phosphorus Bank Loads
(Summer 1993).
-68-
-------
Chapter 4 • Hydrology and Loads
JAMES RIVER, SENS65 (NEW HYDRO) (Cruise) ON N
Total Phosphorus
Total P James Surface
ER, SENS66 (Cruise) ON NEW GRID
ihorus
es Surface
3153
SENS65 = BASE
SENS66 = LOAD X 3
-
50 100
Kilometers
Figure 4-55. Sensitivity of James River Bank Loads (Summer 1993).
YORK RIVER, SENS65 (NEW HYDRO) (Cruise) ON N
Total Phosphorus
Total P York Surface
Julian Day 3153
SENS65 = BASE
SENS66 = LOAD X 3
X,
\
1
103
0.0&
0
: ri
[R, SENS6
ihorus
( Surface
3153
\J
,,-••""
I
6(<
Jr
uiseJONNEWGRID
\h\
0 50 103
Kilometers
Figure 4-56. Sensitivity of York River to Phosphorus Bank Loads (Summer 1993).
-69-
-------
Chapter 4 • Hydrology and Loads
RAPPAHANNOCK RIVER, SENS65 (NEW HYDRO) (C
Total Phosphorus
O.i
0.175
0.15
0.125
1 0.1
0.075
0.05
0.025
Total P Rap Surface
- Julian Day 3153
-
-
-
)i t
JJi -1
L ____^^J*" —
i
0 03 100
Kilometers
SENS65 = BASE
SENS66 = LOAD X 3
!
f
0.125
1
0.1
0,075
0.05
O.OS5
H
150
INOCK RIVER, SENS66 (Cruise) ON NEW
ihorus
Surface
3153
UrHr,
C 50 100 150
Kilometers
Figure 4-57. Sensitivity of Rappahannock River to Phosphorus Bank Loads
(Summer 1993).
I
POTOMAC RIVER, SENS65 (NEW HYDRO) (Cruise) (j)
Total Phosphorus
Total P Potomac Surface
Julian Day 31 S3
100
Kilometers
RIVER, SENS66 (Cruise) ON NEW GRID
ihorus
>mac Surface
3153
SENS65 = BASE
SENS66 = LOAD X 3
I
50 100
Kilometers
Figure 4-58. Sensitivity of Potomac River to Phosphorus Bank Loads
(Summer 1993).
-70-
-------
Chapter 4 • Hydrology and Loads
0.2
0.18
0.16
0.14
0.12
fo,
CI.C6
0.06
0.04
OKI
PATUXENT RIVER, SENS65 (NEW HYDRO) (Cruise)
Total Phosphorus
Total P Patuxent Surface
Julian Day 31 S3
RIVER, SENS66 (Cruise) ON NEW GRID
ihorua
ixent Surface
3153
SENS65 = BASE
SENS66 = LOAD X 3
Figure 4-59. Sensitivity of Patuxent River to Phosphorus Bank Loads
(Summer 1993).
Wetlands Loads
Wetlands loads are the sources (or sinks) of oxygen and oxygen-demanding
material associated with wetlands that fringe the shore of the bay and tributaries.
These loads are invoked primarily as an aid in calibration of tributary dissolved
oxygen.
Quantifying the influence of wetlands on estuarine water quality has been an
interest of scientists and engineers for decades. New methodologies have provided
improved estimates but precise specification of wetlands effects remains impos-
sible. The best that can be done is to bracket or "range" possible effects and then
select values for use in the model from the range of potential values.
Neubauer et al. (2000) measured annual net macrophyte production of 1.4-2 g
C m~2 d"1 in a tidal freshwater Pamunkey River marsh. Although episodic flooding
or scouring of marshes may produce large loads, over long periods no more carbon
can be exported from a marsh than is produced. Consequently, the Neubauer et al.
study provides an upper range for marsh carbon export. Actual carbon export
should be much less than annual net production due to burial; carbon import may
be necessary to balance combined effects of burial and respiration (Neubauer et al.
2001). In the model, a uniform carbon export of 0.3 g C m~2 d"1 was employed.
Neubauer at al. (2000) calculated belowground marsh respiration of 516-723 g
C m~2 yr1—equivalent to 3.8-5.3 g oxygen equivalents m~2 d"1. Sediment oxygen
consumption should be less than total respiration due to gas evolution, burial of
sulfide, and other processes. A summary of sediment oxygen demand in the system
-71 -
-------
Chapter 4 • Hydrology and Loads
water column (DiToro 2001) indicates consumption of oxygen in sediments rarely
exceeds 3 g O2 m~2 d"1 while measures in salt marshes (Cai et al. 1999) indicate
sediment oxygen flux of 1-1.3 g O2 m~2 d"1. In the model, a uniform oxygen
demand of 2 g O2 m~2 d"1 was employed.
Summary of Wetland Loads
Wetlands areas adjacent to
model surface cells were derived
via GIS analysis and provided
by the EPA Chesapeake Bay
Program in November 2001.
Loads to each cell were
computed as the product of adja-
cent wetlands area and areal
carbon export or oxygen
consumption.
Wetlands are distributed
throughout the bay system.
Basins with the greatest
wetlands area (Figure 4-60)
include the lower Eastern Shore,
the lower James River, the York
River, the upper Rappahannock
River, and the upper Patuxent
River. Segments receiving the
largest carbon loads (Figure
4-61) and subject to the greatest
oxygen consumption include the
mid-portion of the bay (CBS),
Emergent Wetland Areas
Associated with the
Water Quality Model
Wwland AKN> m Acre
n •mill
•a:: •:=::
aoc s -
MOD
10000 J'-fin
WqcdIUl slip
Figure 4-60. Emergent Wetlands in
Chesapeake Bay System
1000000
100
i[i[^------
CJUUUQQCJtJUJUJLlJUJl-LlJLlJLULlJUJLlJLlJLlJ — 1_ l_l_l_lUJLULlJLlJLU
........... Ill (£ DC ££ [£ K
Figure 4-61. Wetland Carbon Loads by GBPS
-72-
-------
Chapter 4 • Hydrology and Loads
Tangier Sound (EE3), several Eastern Shore tributaries (ET6, ET7, ET8), the
middle and lower James River (RETS, LE5), the tidal fresh York River (TF4), and
the York River mouth (WE4).
Sensitivity to Wetland Loads
Our ability to judge the impact of wetland loads on total organic carbon was
clouded by large variance in the observations and by discontinuities apparently
due to procedural changes. Our judgement was that moderate carbon loads,
1 g C m~2 d"1, produced model results that were high relative to observations
(Figure 4-62). The sensitivity of computed dissolved oxygen, the parameter of
prime interest, to wetland carbon loads was negligible (Figure 4-63). Consequently,
we assigned a small wetland carbon load (0.3 g C m~2 d"1) which had little impact
on computations. Our focus shifted to direct oxygen consumption in wetlands.
The response of the water column to wetland oxygen uptake depends on the
ratio of wetland area to water surface area. CBPS with large surface areas show
little response to wetlands along their perimeters (Figures 4-64, 4-65). Narrow
reaches adjoining extensive wetlands show the greatest sensitivity. These include
tidal fresh portions of the York (Figure 4-66), Rappahannock (Figure 4-67), and
Patuxent (Figure 4-68) Rivers.
SENS1040NNEWGRI0
1 •'M i iq inn jrt n
* TQC TF4 2 Surtea
SENS 104 -NO WETLAND LOADS
SENS 109 - WETLAND CARBON LOADING
SENS1090NNEWGR10
1 ttl Cr'Vi! C sthi'ii
* TQC TF4 „' Surface
,*
-------
Chapter 4 • Hydrology and Loads
!::•
SENS1040NNEWGRID
Dissolved Oxygen
Dissolved Oxygen TF4.2 Surface
IF4.Z Lltetil
•*•;
_
s. •>'• • Vs v\ -
7»•? > V • t >:
SENS 104 - NO WETLAND LOADS
SENS 109-WETLAND CARBON LOADING
I I
SENS1090NNEWGRID
Dissolved Oxygen
Dissolved Oxygen TF4.2 Surface
&F4.3 Jfcelll \ ji * (1 fl
PM
jji-ifCJ
/ -• .
•* * *
•; .; , •% > ^ * v
• :
> "
Figure 4-63. Sensitivity of Dissolved Oxygen in Tidal Fresh York River to Wetland
Carbon Loading (1 g nrr2 d~1).
SENS1Q40NNEWGRID
Dissolved Oxygen
Dissolved Oxygen EE3.1 Surface
EE3.1 Level 1
SENS 104 - NO WETLAND LOADS
SENS110 - WETLAND OXYGEN UPTAKE
SENS1100NNEWGRID
Dissolved Oxygen
J Oxygen EE3.1 Surface
EE3.1 Level 1 .
> 1 fi a it'
Years
Figure 4-64. Sensitivity of Tangier Sound to Wetland Dissolved Oxygen Uptake
(2 g m-2 d-1).
-74-
-------
Chapter 4 • Hydrology and Loads
SENS1040NNEWGRID
Dissolved Oxygen
Dissolved Oxygen WE4.2 Surface
WE4.J Level 1
SENS 104 - NO WETLAND LOADS
SENS 110 - WETLAND OXYGEN UPTAKE
I
SENS1100NNEWGRID
Dissolved Oxygen
Dissolved Oxygen WE4.2 Surface
WE4.I Level 1
Figure 4-65. Sensitivity of York River Mouth to Wetland Dissolved Oxygen Uptake
(2 g m-2 d-1).
SENS1040NNEWGRID
Dissolved Oxygen
Ived Oxygen TIF4.2 Surface
1F4.? Llteiil
SENS1100NNEWGRID
Dissolved Oxygen
Dissolved Oxygen TF4.2 Surface
"F4.3J Livel A
SENS 104 - NO WETLAND LOADS
SENS 110 - WETLAND OXYGEN UPTAKE
Figure 4-66. Sensitivity of Tidal Fresh York River to Wetland Dissolved Oxygen
Uptake (2 g nr2 d'1).
-75-
-------
Chapter 4 • Hydrology and Loads
SENS 104 ON NEW GRID
Dissolved Oxygen
TF3.3 Surface
SENS11DONNEWGRID
Dissolved Oxygen
Dissolved Oxygeh TF3.3 Surface
TF3.3 Le^el
SENS 104 -NO WETLAND LOADS
SENS110 -WETLAND OXYGEN UPTAKE
Figure 4-67. Sensitivity of Tidal Fresh Rappahannock River to Wetland Dissolved
Oxygen Uptake (2 g nr2 d~1).
SENS1G40NNEWGRID
Dissolved Oxygen
Dissolved Oxygen TF1.7 Surface
TF1.7, Level 1 *i
SENS 104 - NO WETLAND LOADS
SENS110 - WETLAND OXYGEN UPTAKE
SENS1100NNEWGRID
Dissolved Oxygen
Dissolved Oxygen TN ,7 Surface
TF1.n Lawsl
4 6
Yeare
Figure 4-68. Sensitivity of Tidal Fresh Patuxent River to Wetland Dissolved
Oxygen Uptake (2 g nr2 d~1).
-76-
-------
Chapter 4 • Hydrology and Loads
Summary of All Loads
Loads from all sources were compared by CBPS for 1990, a year central to the
simulation period. Runoff in this year was moderate in the Susquehanna and James
(Figures 4-1, 4-3) and low in the Potomac (Figure 4-2).
Nonpoint sources dominated the nitrogen loads (Figure 4-69, Table 4-2) except
in a few segments adjacent to major urban areas including Richmond (TF5),
Hampton Roads (LE5), Washington (TF2), and Baltimore (WT4, WT5). In these
regions, point sources contributed a significant fraction of nitrogen loads. Atmos-
pheric nitrogen loads were significant only in the large, open segments of the
mainstem bay (CB3-CB7, EE3) and in the lower Potomac (LE2). Bank loads were
negligibly small throughout.
100000
onnnn
pnnnn
ynnnn
cnnnn
03
-o
"*- f^nnnn
&
dnnnn
•snnnn
onnnn
•innnn
n
CB1 load is
" 195,980 of which
99.6% is Nonpoint
Source
1
^uUOnUlliLDn^nUUn^ .JnJ _Hn_nP
^c*i^**Lnico*d-Ln^t-T-tori--03
1— 1— h— h- h- 1— HI— LLl UJ LLl LLJ LJJ h- 1— 1— h— h- Lj_Lj_U_Lj_U_LLj|— (— h- t~|— t~t~t~
miijmmmLiJLLiLU_i_i_i_i_iLULiJLiJLLiLiJi— i— i— i— i— ^ -' ----- ~~ '- >=;<-<
[rcrttcrLr ^>>>>>>>>
Figure 4-69. Summary of Nitrogen Loads for the Year 1990.
Table 4-2
System-wide Load Summary, 1990
Nitrogen
Phosphorus
Carbon
Solids
Point Source,
(kg/day)
80321
4245
160647
Nonpoint
Source,
(kg/day)
359506
24600
1014435
11298430
Atmosphere,
(kg/day)
26184
1927
Bank Load,
(kg/day)
4218
8695
52424
12786991
Wetlands,
(kg/day)
1355742
-77-
-------
Chapter 4 • Hydrology and Loads
Nonpoint sources usually comprised the largest fraction of phosphorus loads
(Figure 4-70, Table 4-2) but were not so predominant as for nitrogen. As with
nitrogen, point-source loads were significant in urban areas and atmospheric loads
contributed substantially to large open-water segments. A major contrast with
nitrogen was in bank loading which equaled or exceed nonpoint-source phosphorus
loads in segments distant from major inflows. These included the eastern embay-
ments (EE1-EE3) and the river-estuarine transition segments of western tributaries
(RET1-RET5).
8000
7nnn
finnn
m
"~- *innn
O)
^nnn
•i nnn
n Bank Load
LJ Atmosphere
• Nonpoint Source
n Point Source
_i-iTl
Hfl
' > '.' ' <
BflJlRaLiy
rn m rn m en fjj in in K- • i—
jH^J^
Qy .-
ll
ft in iri in ci SS "i "l "I '"'i l{i tr, S tn if, in 1 1 1 £ t S S 1 1 1 1 1 1 1
L£ tt tt Ct LC
Figure 4-70. Summary of Phosphorus Loads for the Year 1 990.
Nonpoint sources dominate the carbon loads (Figure 4-71, Table 4-2) in
segments adjoining the inflows of the Susquehanna (CB1), Potomac (TF2), and
James (TF5). Point sources contribute a significant fraction in segments adjacent to
major urban areas including Richmond (TF5), Hampton Roads (LE5), Washington
(TF2), and Baltimore (WT4, WT5). Otherwise, wetlands are the major source of
carbon loads to the system. An assessment of loads provides an incomplete picture
of the carbon budget, however. A complete assessment requires inclusion of carbon
contributed by primary production as well as loads.
Nonpoint sources dominate the solids loads (Figure 4-72) in segments adjoining
the inflows of the Susquehanna (CB1), Potomac (TF2), and James (TF5). Other-
wise, bank loads are the dominant source. For 1990, bank loads contribute more
solids, system-wide, than nonpoint sources (Table 4-2). Of course, the relative
contribution of bank loads depends on annual hydrology. In high-flow years,
nonpoint sources may dominate. Still, the summary indicates that control of bank
erosion should be included in any solids management plan.
-78-
-------
Chapter 4 • Hydrology and Loads
700000 -1
finnnnn
Rnnnnn
*? Annnnn
"Q
o
c& 'snnnnn
9nnnnn
1 nnnnn
n
u
I
DLjUUoDiiaU
- . -i "i -it l! . if i !" - ''I! — IN f
jUJCUUJUJUJUJ£l^yJUJL
n Wetlands
LJ Bank Load
• Nonpoint Source
H Point Source
n I |
j 111
,._.DJy_QUiD nJoaLoDnDQlllluL ni __
I ~- ,' 5 (-1 ft, -t-- u ! -li f- - K ' ' "1 — <'J f 1 -ff U I — <"J ^1 T|- 1 — I --t U i T* — |-| fy( ^. L, , ;, , [-- . i| i
J! — i 1 I i 1 i t UJLUUJUJUJt 1 > ( 1 LL. U, LL. LL U_ UJ ! 1 t t 1 i t
j LU !. UJ 1JJ UJ UJ UJ L1J UJ UJ -j _J _J ~i _j UJ UJ UJ 111 Lu 1 t~ t™ h- ^ ,» .* ,.^ „> ^ -* «.=*'*
Figure 4-71. Summary of Carbon Loads for the Year 1990.
4000 -,
3500
Figure 4-72. Summary of Solids Loads for the Year 1990.
-79-
-------
Chapter 4 • Hydrology and Loads
References
Bicknell, B., Imhoff, J., Kittle, J., Donigian, A., Johanson, R., and Barnwell, T. (1996).
"Hydrologic simulation program—FORTRAN user's manual for release 11," United States
Environmental Protection Agency Environmental Research Laboratory, Athens GA.
Cai, W., Pomeroy, L., Moran, M., and Wang, Y. (1999). "Oxygen and carbon dioxide mass
balance for the estuarine-intertidal marsh complex of five rivers in the southeastern U.S.,"
Limnology and Oceanography, 44(3), 639-649.
Chang, J., Brost, R., Isaksen, L, Madronich, S., Middleton, P., Stockwell, W., and Walcek,
C. (1987). "A three-dimensional Eulerian acid deposition model— physical concepts and
formulation," Journal of Geophysical Research, 92, 14681-14700.
DiToro, D. (2001). Sediment Flux Modeling. Wiley-Interscience, New York, NY.
Donoghue, J., Bricker, O., and Olsen, C. (1989). "Particle-bourne radionuclides as tracers
for sediment in the Susquehanna River and Chesapeake Bay," Estuarine, Coastal and Shelf
Science, 29, 341-360.
Ibison, N., Frye, C., Frye, J., Hill, C., and Burger, N. (1992). "Eroding bank nutrient verifi-
cation study for the lower Chesapeake Bay," Department of Conservation and Recreation,
Division of Soil and Water Conservation, Gloucester Point VA.
Neubauer, S., Anderson, L, Constantine, J., and Kuehl, S. (2001). "Sediment deposition and
accretion in a mid-Atlantic (U.S.A.) tidal freshwater marsh," Estuarine, Coastal, and Shelf
Science, 56, 000-000.
Neubauer, S.,Miller, W., and Anderson, I. (2000). "Carbon cycling in a tidal freshwater
marsh ecosystem: a carbon gas flux study," Marine Ecology Progress Series, 199, 13-30.
Redfield, A., Ketchum, B., and Richards, F. 1966. "The influence of organisms on the
composition of sea-water," in The Sea Vol. II, Interscience Publishers, New York, NY, pp
26-48.
Schubel, J. (1968). "Turbidity maximum of the northern Chesapeake Bay," Science, 161,
1013-1015.
US Army Corps of Engineers Baltimore District. (1990). "Chesapeake Bay shoreline
erosion study feasibility report," Baltimore MD.
-80-
-------
Linking in the Loads
Introduction
Major loads were mapped to model cells by the sponsor before they were
provided to the model team. The linkage process between loads and model
consisted of converting quantities in the loads into appropriate model state vari-
ables. The present chapter describes development of conversion guidelines and lists
parameter values employed in the conversions.
Nonpoint-Source Loads
Linking of watershed model loads focused on the nitrogen and phosphorus
state variables (Table 5-1). The watershed model loads distinguished particulate
inorganic phosphorus from organic phosphorus. Discussion with the watershed
model team indicated particulate inorganic phosphorus was utilized as a way of
modeling scour from the bottom during high-flow intervals. No effort was made to
truly distinguish particulate inorganic from organic phosphorus forms. The Tribu-
tary Refinements phase of water quality model application (Cerco et al. 2002)
distinguished particulate organic and particulate inorganic phosphorus forms. This
distinction was not carried forward into the present model. Consequently, the
watershed model particulate inorganic and organic phosphorus forms were
combined before mapping into water quality model variables. Since particulate
phosphorus in the water quality model consists of both organic and inorganic
forms, we have dropped the specification "organic" when referring to water quality
model particulate phosphorus.
The watershed model produced loads of total suspended solids. The corres-
ponding water quality model variable was fixed (inorganic) solids. Linkage with
the watershed model required removal of the volatile (organic) solids from the total
solids load. Volatile solids were still input to the water quality model but in the
form of particulate organic carbon.
-81 -
-------
Chapter 5 • Linking the Loads
Table 5-1
Variables in Loads and Water-Quality Model
Load Variable
Ammonium
Nitrate
Organic Nitrogen
Dissolved Phosphate
Organic Phosphorus
plus
Particulate Inorganic
Phosphorus
Total Organic Carbon
Total Suspended Solids
Maps to
— >
>
->
>
->
>
->
Model Variable
Ammonium
Nitrate
Dissolved Organic Nitrogen,
Labile Particulate Organic Nitrogen,
Refractory Particulate Organic Nitrogen
Phosphate
Dissolved Organic Phosphorus,
Labile Particulate Phosphorus,
Refractory Particulate Phosphorus
Dissolved Organic Carbon,
Labile Particulate Organic Carbon,
Refractory Particulate Organic Carbon
Fixed Solids
Nonpoint-source total organic carbon loads were derived by ratio to watershed
model organic nitrogen loads. The total carbon loads also required mapping into
water quality model state variables.
Guidelines for splitting total organic forms contained in the loads into dissolved
and particulate forms for the water quality model were based on observations
collected in the Chesapeake Bay River Input Monitoring Program. Data were
obtained from the program web site: http://va.water.usgs.gov/chesbay/RIMP/.
Inspection of the data base indicated collection of relevant observations at Virginia
fall lines commenced in 1996. Relevant observations from earlier years were avail-
able for Maryland fall lines but detection levels and reporting increments were
poor. Circa 1996, improvements were made to analytical methods and reporting.
Consequently, analysis of Maryland observations was restricted to data collected
from 1996 onwards. In all cases, reports of "less than" were converted to the detec-
tion level e.g. <.01 became 0.01. At the time our analysis was conducted,
observations were available through 1999.
Nitrogen and Phosphorus
For the Virginia fall lines, suspended (particulate) nitrogen and phosphorus
forms were reported directly. Dissolved organic nitrogen was determined as
dissolved nitrogen less ammonium less nitrate+nitrite. Dissolved organic phos-
phorus was determined as dissolved phosphorus less dissolved inorganic
phosphorus.
-82-
-------
Chapter 5 • Linking the Loads
For the Maryland fall lines, dissolved organic nitrogen was determined as total
Kjeldahl nitrogen (filtered) less ammonium. Particulate organic nitrogen was deter-
mined as total Kjeldahl nitrogen (whole) less total Kjeldahl nitrogen (filtered).
Dissolved organic phosphorus was determined as dissolved phosphorus less
dissolved inorganic phosphorus. Particulate phosphorus was determined as total
phosphorus less dissolved phosphorus.
Inspection indicated dissolved organic nitrogen and phosphorus concentrations
were independent of flow (Figures 5-1-5-6). Concentrations were similar at all
fall lines except the Susquehanna which exhibited lower dissolved organic
2E
2
E
1 -
0.5 -
• suspended N
• DON
»
• *
• " * " * *
* • . * .*
1000 10000 100000 1000000
Flow (CFS)
Figure 5-1. Observed particulate and dissolved organic nitrogen at the James
River.
0.7
0.6 ^
0.5
dO.4 -
Q_
g
E 0.3 -
0.2
0.1 ^
suspended P
OOP
0
1000
£•»&•
•-••*
10000 100000
Flow (CFS)
1000000
Figure 5-2. Observed particulate and dissolved organic phosphorus at the James
River fall line.
-83-
-------
Chapter 5 • Linking the Loads
3.5
3 -
2.5 -
2
l.5-
1
0.5 -
0
1000
PON
DON
10000 100000
Flow (CFS)
1000000
Figure 5-3. Observed particulate and dissolved organic nitrogen at the Potomac
River fall line.
1.4 -i
1.2 -
1 -
=J 0.8
0.
D)
E 0,6
0.4
0.2
0
10
• PP (mg/L)
• OOP (mg/L)
•
«
* • * • *
*
•" * *• .
f. ]*"•*• • j
00 10000 100000 100<
)000
Flow (CFS)
Figure 5-4. Observed particulate and dissolved organic phosphorus at the
Potomac River fall line.
0.3
0.25 -F
0.2
,0.15 -
0.1 -
0.05
PP(mg/L)
D°P (mg
4 4
• •
/*
1000
10000 100000
Flow (CFS)
1000000
Figure 5-5. Observed particulate and dissolved organic nitrogen at the
Susquehanna River fall line.
-84-
-------
Chapter 5 • Linking the Loads
10000 100000
Flow (CFS)
1000000
Figure 5-6. Observed particulate and dissolved organic phosphorus at the
Susquehanna River fall line.
concentrations than elsewhere (Table 5-2). Particulate forms increased as a function
of flow. Frequently a break was apparent in particulate nutrient concentrations that
suggested initiation of bottom scour of particulates. These insights led to
specification of constant dissolved organic concentrations in distributed sources.
An algorithm (Figure 5-7) was developed to convert loads into dissolved and
particulate fractions.
Particulate Inorganic Phosphorus. An analysis was conducted to distinguish
organic and inorganic particulate phosphorus fractions. While the water quality
model does not presently incorporate these distinctions, results are reported for
future reference.
The EPA Chesapeake Bay Program Office conducted a special monitoring
program in 1994 to measure particulate inorganic phosphorus (PIP) at the fall lines
and in stream. We paired fall-line PIP observations with particulate phosphorus
observations, collected on the same day, in the USGS data base. A total of 54 pairs
were produced, half of which were in the Potomac River. The median ratio of PIP
to particulate phosphorus was 0.50. That is, half of the particulate phosphorus load
at the fall lines was in inorganic form. No effect of flow on composition of particu-
late phosphorus was evident but this finding was limited to the range of flows
sampled. Sampling was concentrated in summer and at low to moderate flows.
Total Organic Carbon
Total organic carbon observations were available only at the Maryland fall
lines. These showed a tendency for organic carbon concentration to increase with
flow (Figures 5-8, 5-9). No data were available to partition total organic carbon into
particulate and dissolved fractions. Both load and fractions were obtained by ratio
to particulate and dissolved organic nitrogen using ratios presented in the chapter
entitled "Hydrology and Loads."
-85-
-------
Chapter 5 • Linking the Loads
Table 5-2
Dissolved Organic Nutrients and Volatile Solids at Major Fall Lines
Organic N,
mg/L
Median
Mean
Std. Dev.
Range
Number of
Observations
Model
Organic P,
mg/L
Median
Mean
Std. Dev
Range
Number of
Observations
Model
Volatile
Solids,
fraction
Median
Mean
Range
Number of
Observations
Model
James
0.20
0.22
0.14
0.02 - 1 .25
127
0.25
0.009
0.010
0.010
0 - 0.06
127
0.0085
0.19
0.40
0.35
0.09 - 1
124
0.12
York
0.27
0.29
0.11
0 - 0.84
279
0.25
0.010
0.011
0.012
0-0.13
279
0.0085
0.38
0.47
0.31
0.08 - 1
268
0.12
Rappa-
hannock
0.21
0.26
0.29
0 - 2.25
87
0.25
0.008
0.012
0.020
0-0.17
87
0.0085
0.17
0.40
0.36
0.09 - 1
83
0.12
Potomac
0.26
0.28
0.11
0 - 0.92
241
0.25
0.010
0.015
0.016
0-0.17
241
0.0085
0.12
Patuxent
0.29
0.30
0.11
0.01 -0.62
79
0.25
0
0.003
0.018
0-0.12
79
0.0085
0.12
Susque-
hanna
0.16
0.18
0.11
0-0.93
83
0.16
0.005
0.008
0.017
0-0.16
83
0.005
0.12
-86-
-------
Chapter 5 • Linking the Loads
Estimate dissolved load as
dissolved concentration X
flow
Is dissolved load < total
load from watershed
model?
No
Yes
Dissolved load = total
load from watershed
model
Particulate load = 0
Dissolved load =
estimated load
Particulate load =
total load minus
dissolved load
Figure 5-7. Determination of dissolved organic and particulate loads from water-
shed model total organic loads.
14 -j
12 \
10 1
O
O
£ 6
* *••«t
0
1000
10000 100000
Flow (cfs)
1000000
Figure 5-8. Observed total organic carbon versus flow at Potomac fall line.
-87-
-------
Chapter 5 • Linking the Loads
35
30
25
O 20
O
|» 15
10
5
1000
10000 100000
Flow (cfs)
1000000
Figure 5-9. Observed total organic carbon versus flow at Susquehanna fall line.
Fixed Solids
Observations of total, fixed, and volatile solids were available at the Virginia fall
lines. These observations indicated the solids fractionation was dependent on flow
(Figure 5-10). At low flows, solids were almost entirely volatile (organic). The frac-
tion volatile solids decreased to a constant value as flow increased. Also, the river
with the lowest flow, the York, had the greatest median fraction volatile solids
(Table 5-2). Since the greatest solids loads enter during storm events in major tribu-
taries, we assigned our fraction volatile solids to reflect high flows. The watershed
model total solids loads were reduced by 12% to provide fixed solids loads for the
water quality model.
Labile and Refractory Participates
Particulate matter in nonpoint-source loads must be split into labile and refrac-
tory portions to correspond to water quality model variables. The definition of
labile organic matter derives from the sediment diagenesis model. In the sediment
model, labile organic matter has a decay coefficient of 0.035 d"1 @ 20 °C. At that
rate, almost 90% of labile organic matter decays away in sixty days. Little guidance
exists for partitioning particulate loads into labile and refractory fractions. A set of
preliminary experiments conducted during the first portion of this study (Cerco and
Cole 1994) indicated the labile fraction of particles at the fall line varied from 0.2
(carbon and phosphorus) to 0.6 (nitrogen). Experiments with the early model indi-
cated these fractions adversely affected model calibration. Sediment diagenesis of
labile particles deposited near the fall lines produced unrealistically large nutrient
fluxes into the water column. As a result, the labile fraction of particles in nonpoint
sources was set to zero. This fractionation was carried over into the present model.
-88-
-------
Chapter 5 • Linking the Loads
1000
10000 100000
Flow (CFS)
1000000
Figure 5-10. Observed fraction volatile solids at James River fall line.
Point-Source Loads
Linking point-source loads to the water quality model involves the same tasks as
the nonpoint-source loads. Loads of total organic nutrients and carbon must be
split into dissolved, labile particulate, and refractory particulate forms.
As part of the Tributary Refinements portion of the model study, seven point-
source discharges in Virginia were sampled on two occasions in 1994. Effluent was
analyzed specifically to provide quantities necessary to map reported effluent
quantities into model state variables. The particulate fraction assigned to model
loads closely reflected the median value determined from the 1994 observations
(Table 5-3).
Little guidance was available to split particulates into labile and refractory frac-
tions. A set of preliminary experiments conducted during the first portion of this
study (Cerco and Cole 1994) indicated the labile fraction of particles from point
sources was low for phosphorus (0.02) and moderate for nitrogen (0.5) and carbon
(0.7). The values for carbon and nitrogen appeared unreasonable. Effluent from
point sources is treated and is not expected to be highly reactive. Consequently,
labile fractions assigned in the model were 0.15 for carbon, 0.15 for nitrogen, and
0.07 for phosphorus.
Other Loads
Bank loads were assumed to consist entirely of refractory particulate matter.
Wetlands carbon loads were split equally into dissolved organic, labile particulate,
and refractory particulate organic carbon. The organic fraction of atmospheric loads
was split equally into dissolved, labile particulate, and refractory particulate forms.
-89-
-------
Chapter 5 • Linking the Loads
Table 5-3
Participate Fraction of Organic Matter at Point Sources
Facility
Richmond
Petersburg
Hopewell
Allied
Chesterfield
Allied Hopewell
Chesapeake
Corp
Fredericksburg
Median
Model
NPDES
63177
25437
66630
5312
5219
3115
25127
PON/TON
0.53
0.68
0.05
0.35
0.75
0.64
0.22
0.53
0.5
PP/TOP
0.59
0.37
0.45
0.79
0.76
0.60
0.78
0.60
0.6
POC/TOC
0.23
0.14
0.12
0.37
0.36
0.20
0.17
0.20
0.2
References
Cerco, C, Johnson, B., Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model," ERDC TR-02-4, US Army Engineer Research and Development Center, Vicksburg
MS.
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report EL-94-4, US Army Engineer Waterways Experiment Station,
Vicksburg MS.
-90-
-------
Water Quality Model
Formulation
Introduction
CE-QUAL-ICM was designed to be a flexible, widely-applicable eutrophication
model. Initial application was to Chesapeake Bay (Cerco and Cole 1994). Sub-
sequent additional applications included the Delaware Inland Bays (Cerco et al.
1994), Newark Bay (Cerco and Bunch 1997), and the San Juan Estuary (Bunch et
al. 2000). Each model application employed a different combination of model
features and required addition of system-specific capabilities. This chapter
describes general features and site-specific developments of the model as presently
applied to the water column of Chesapeake Bay.
Conservation of Mass Equation
The foundation of CE-QUAL-ICM is the solution to the three-dimensional
mass-conservation equation for a control volume. Control volumes correspond to
cells on the model grid. CE-QUAL-ICM solves, for each volume and for each state
variable, the equation:
(6-D
in which:
Vj = volume of jth control volume (m3)
Cj = concentration in j"1 control volume (g m"3)
t, x = temporal and spatial coordinates
n = number of flow faces attached to jth control volume
Qk = volumetric flow across flow face k of jth control volume (m3 s"1)
Ck = concentration in flow across face k (g m"3)
Ak = area of flow face k (m2)
Dk = diffusion coefficient at flow face k (m2 s"1)
Sj = external loads and kinetic sources and sinks in jth control volume (g s"1)
-91 -
-------
Chapter 6 • Water Quality Model Formulation
Solution of Equation 6-1 on a digital computer requires discretization of the
continuous derivatives and specification of parameter values. The equation is
solved using the QUICKEST algorithm (Leonard 1979) in the horizontal plane and
a Crank-Nicolson scheme in the vertical direction. Discrete time steps, determined
by computational stability requirements, are 15 minutes.
State Variables
At present, the CE-QUAL-ICM model incorporates 24 state variables in the
water column including physical variables, multiple algal groups, two zooplankton
groups, and multiple forms of carbon, nitrogen, phosphorus and silica (Table 6-1).
Table 6-1
Water Quality Model State Variables
Temperature
Fixed Solids
Spring Diatoms
Microzooplankton
Dissolved Organic Carbon
Refractory Particulate Organic Carbon
Nitrate+Nitrite
Labile Particulate Organic Nitrogen
Total Phosphate
Labile Particulate Organic Phosphorus
Chemical Oxygen Demand
Dissolved Silica
Salinity
Freshwater Cyanobacteria
Other (Green) Algae
Mesozooplankton
Labile Particulate Organic Carbon
Ammonium+Urea
Dissolved Organic Nitrogen
Refractory Particulate Organic Nitrogen
Dissolved Organic Phosphorus
Refractory Particulate Organic Phosphorus
Dissolved Oxygen
Particulate Biogenic Silica
Algae
Algae are grouped into three model classes: cyanobacteria, spring diatoms, and
other green algae. The grouping is based upon the distinctive characteristics of each
class and upon the significant role the characteristics play in the ecosystem.
Cyanobacteria, commonly called blue-green algae, are characterized by their abun-
dance (as picoplankton) in saline water and by their bloom-forming characteristics
in fresh water. Cyanobacteria are unique in that some species fix atmospheric
nitrogen although nitrogen fixers are not predominant in the Chesapeake Bay
system. The Cyanobacteria distinguished in the model are the bloom-forming
species found in the tidal, freshwater Potomac River. They are characterized as
having negligible settling velocity and are subject to low predation pressure. The
picoplankton are combined with the other green algae group since insufficient data
-92-
-------
Chapter 6 • Water Quality Model Formulation
exists to consider them separately. Spring diatoms are large phytoplankton that
produce an annual bloom in the saline portions of the bay and tributaries. Settling
of spring diatom blooms to the sediments may be a significant source of carbon for
sediment oxygen demand. Diatoms are distinguished by their requirement of silica
as a nutrient to form cell walls. Algae that do not fall into the preceding two groups
are lumped into the heading of green algae. The other green algae represent the
mixture that characterizes saline waters during summer and autumn and fresh
waters year round. Non-bloom forming diatoms comprise a portion of this mixture.
Zooplankton
Two zooplankton groups are considered: microzooplankton and mesozoo-
plankton. The microzooplankton can be important predators on phytoplankton and
they are one of the prey groups for mesozooplankton. Mesozooplankton consume
phytoplankton and detritus as well as microzooplankton. The mesozooplankton are
an important prey resource for carnivorous fmfish such as Bay Anchovy.
Zooplankton were included in the model as a first step towards computing the
effect of eutrophication management on top-level predators.
Organic Carbon
Three organic carbon state variables are considered: dissolved, labile particulate,
and refractory particulate. Labile and refractory distinctions are based upon the
time scale of decomposition. Labile organic carbon decomposes on a time scale of
days to weeks while refractory organic carbon requires more time. Labile organic
carbon decomposes rapidly in the water column or the sediments. Refractory
organic carbon decomposes slowly, primarily in the sediments, and may contribute
to sediment oxygen demand years after deposition.
Nitrogen
Nitrogen is first divided into available and unavailable fractions. Available refers
to employment in algal nutrition. Two available forms are considered: reduced and
oxidized nitrogen. Reduced nitrogen includes ammonium and urea. Nitrate and
nitrite comprise the oxidized nitrogen pool. Both reduced and oxidized nitrogen are
utilized to fulfill algal nutrient requirements. The primary reason for distinguishing
the two is that ammonium is oxidized by nitrifying bacteria into nitrite and, subse-
quently, nitrate. This oxidation can be a significant sink of oxygen in the water
column and sediments.
Unavailable nitrogen state variables are dissolved organic nitrogen, labile partic-
ulate organic nitrogen, and refractory particulate organic nitrogen. The dissolved
organic nitrogen state variable excludes urea which is directly available as an algal
nutrient.
Phosphorus
As with nitrogen, phosphorus is first divided into available and unavailable
fractions. Only a single available form, dissolved phosphate, is considered. The
model framework allows for exchange of phosphate between dissolved and
-93-
-------
Chapter 6 • Water Quality Model Formulation
particulate (sorbed to solids) forms but this option is not implemented in the
present application. Three forms of unavailable phosphorus are considered:
dissolved organic phosphorus, labile particulate organic phosphorus, and refractory
particulate organic phosphorus.
Silica
Silica is divided into two state variables: dissolved silica and particulate biogenic
silica. Dissolved silica is available to diatoms while particulate biogenic silica
cannot be utilized. In the model, particulate biogenic silica is produced through
diatom mortality. Particulate biogenic silica undergoes dissolution to available
silica or else settles to the bottom sediments.
Chemical Oxygen Demand
Chemical oxygen demand is the concentration of reduced substances that are
oxidized by abiotic processes. The primary component of chemical oxygen demand
is sulfide released from sediments. Oxidation of sulfide to sulfate may remove
substantial quantities of dissolved oxygen from the water column.
Dissolved Oxygen
Dissolved oxygen is required for the existence of higher life forms. Oxygen
availability determines the distribution of organisms and the flows of energy and
nutrients in an ecosystem. Dissolved oxygen is a central component of the water-
quality model.
Salinity
Salinity is a conservative tracer that provides verification of the transport compo-
nent of the model and facilitates examination of conservation of mass. Salinity also
influences the dissolved oxygen saturation concentration and may be used in the
determination of kinetics constants that differ in saline and fresh water.
Temperature
Temperature is a primary determinant of the rate of biochemical reactions. Reac-
tion rates increase as a function of temperature although extreme temperatures may
result in the mortality of organisms and a decrease in kinetics rates.
Fixed Solids
Fixed solids are the mineral fraction of total suspended solids. Solids are consid-
ered primarily for their role in light attenuation.
The remainder of this chapter is devoted to detailing the kinetics sources and
sinks and to reporting parameter values. For notational simplicity, the transport
terms are dropped in the reporting of kinetics formulations.
-94-
-------
Chapter 6 • Water Quality Model Formulation
Algae
Equations governing the three algal groups are largely the same. Differences
among groups are expressed through the magnitudes of parameters in the equa-
tions. Generic equations are presented below except when group-specific
relationships are required.
Algal sources and sinks in the conservation equation include production, metab-
olism, predation, and settling. These are expressed:
— B= G-BM-Wa • — B-PR ((- ^
8 t ( SzJ (6-2)
in which:
B = algal biomass, expressed as carbon (g C m~3)
G = growth (d-1)
BM = basal metabolism (d"1)
Wa = algal settling velocity (m d"1)
PR = predation (g C nr3 d'1)
z = vertical coordinate
Production
Production by phytoplankton is determined by the intensity of light, by the avail-
ability of nutrients, and by the ambient temperature.
Light
The influence of light on phytoplankton production is represented by a chloro-
phyll-specific production equation (Jassby and Platt 1976):
(6-3)
in which:
PB = photosynthetic rate (g C g'1 Chi d'1)
PBm = maximum photosynthetic rate (g C g"1 Chi d"1)
I = irradiance (E m"2 d"1)
Parameter Ik is defined as the irradiance at which the initial slope of the produc-
tion vs. irradiance relationship (Figure 6-1) intersects the value of PBm
Ik=P^ (6-4)
a
in which:
a = initial slope of production vs. irradiance relationship (g C g"1 Chi (E m"2)"1)
-95-
-------
Chapter 6 • Water Quality Model Formulation
o
CD
350
300
250 -
200 -
0 150
•3
m
CL
100 -
50 -
PBm
0 25 50 75 100 125 150 175 200
I (E / m2 / d)
Figure 6-1. Production versus irradiance curve.
Chlorophyll-specific production rate is readily converted to carbon specific growth
rate, for use in Equation 6-2, through division by the carbon-to- chlorophyll ratio:
G =
CChl
(6-5)
in which:
CChl = carbon-to-chlorophyll ratio (g C g"1 chlorophyll a)
Specification of the carbon-to-chlorophyll ratio is detailed in a subsequent
chapter entitled "Primary Production."
Nutrients
Carbon, nitrogen, and phosphorus are the primary nutrients required for algal
growth. Diatoms require silica, as well. Inorganic carbon is usually available in
excess and is not considered in the model. The effects of the remaining nutrients on
growth are described by the formulation commonly referred to as "Monod kinetics"
(Figure 6-2; Monod 1949):
f(N) =
D
KHd + D
(6-6)
in which:
f(N) = nutrient limitation on algal production (0 < f(N) < 1)
D = concentration of dissolved nutrient (g m"3)
KHd = half-saturation constant for nutrient uptake (g m"3)
-96-
-------
Chapter 6 • Water Quality Model Formulation
f(N) = 0.5 When N = KH
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
N/KH
Figure 6-2. Monod formulation for nutrient-limited growth.
Temperature
Algal production increases as a function of temperature until an optimum
temperature or temperature range is reached. Above the optimum, production
declines until a temperature lethal to the organisms is attained. Numerous func-
tional representations of temperature effects are available. Inspection of growth
versus temperature data indicates a function similar to a Gaussian probability curve
(Figure 6-3) provides a good fit to observations:
= e"KTgl'(T"Topt)2whenT < Topt
= e"KTg2 ' (Topt"T)2 when T > Topt
in which:
T = temperature (°C)
Topt = optimal temperature for algal growth (°C)
KTgl = effect of temperature below Topt on growth (°C~2)
KTg2 = effect of temperature above Topt on growth (°C~2)
(6-7)
1.2 T KTg1 = 0.004
KTa2 = 0.006
Tm = 20
0.8
1
10 15 20
DEGREES C
25
30
Figure 6-3. Relation of algal production to temperature.
-97-
-------
Chapter 6 • Water Quality Model Formulation
Constructing the Photosynthesis vs. Irradiance Curve
A production versus irradiance relationship is constructed for each model cell at
each time step. First, the maximum photosynthetic rate under ambient temperature
and nutrient concentrations is determined:
PBm(N,T) = pBm • f(T)
D
KHd + D
(6-8)
in which:
PBm(N,T) = maximum photosynthetic rate under ambient temperature and nutrient
concentrations (g C g"1 Chi d"1)
The single most limiting nutrient is employed in determining the nutrient limitation.
Next, parameter Ik is derived from Equation 6-4. Finally, the production vs. irra-
diance relationship is constructed using PBm(N,T) and Ik. The resulting production
versus irradiance curve exhibits three regions (Figure 6-4). For I » Ik, the value of
the term I / (I2 + Ik2)1/2 approaches unity and temperature and nutrients are the
primary factors that influence production. For I « Ik, production is determined
solely by irradiance I. In the region where the initial slope of the production versus
irradiance curve intercepts the line indicating production at optimal illumination,
I = Ik, production is determined by the combined effects of temperature, nutrients,
and light.
300
-T 250
T3
s
o
-: 200
oi
O
O)
150
V)
'in
o>
o>
"o
100
£ so
B
--- PBm(NT) = 300
- PBm(N,T) = 225
PBm(N.T) = 150
20
40
60
80
100
Irradiance (E m~2 d"1)
Figure 6-4. Effects of light and nutrients on production versus irradiance curve,
determined fora = 8 g C g-1 Chi (E nrr2)-1.
-98-
-------
Chapter 6 • Water Quality Model Formulation
Irradiance
Irradiance at the water surface is evaluated at each model time step. Instanta-
neous irradiance is computed by fitting a sin function to daily total irradiance:
T n rr • (n * DSSR"\ (, „
lo = • IT • sin (6-9)
2 • FD I, FD } ^ }
in which:
lo = irradiance at water surface (E m~2 d"1)
IT = daily total irradiance (E m~2)
FD = fractional daylength (0 < FD < 1)
DSSR = time since sunrise (d)
lo is evaluated only during the interval:
(6-10)
in which:
DSM = time since midnight (d)
Outside the specified interval, lo is set to zero.
Irradiance declines exponentially with depth below the surface. The diffuse
attenuation coefficient, Kd, is computed as a function of color and concentrations
of organic and mineral solids.
Respiration
Two forms of respiration are considered in the model: photo-respiration and
basal metabolism. Photo-respiration represents the energy expended by carbon
fixation and is a fixed fraction of production. In the event of no production (e.g. at
night), photo-respiration is zero. Basal metabolism is a continuous energy expen-
diture to maintain basic life processes. In the model, metabolism is considered to
be an exponentially increasing function of temperature (Figure 6-5). Total
respiration is represented:
KTb'(T"Tr)
R = Presp • G + BM • e
in which:
Presp = photo-respiration (0 < Presp < 1)
BM = metabolic rate at reference temperature Tr (d"1)
KTb = effect of temperature on metabolism (°C"1)
Tr = reference temperature for metabolism (°C)
-99-
-------
Chapter 6 • Water Quality Model Formulation
KT = 0.069 / degree C 2.5 T
-10 -7.5 -5 -2.5 0 2.5 5 7.5 10
Figure 6-5. Exponential temperature relationship employed for
metabolism and other processes.
Predation
The predation term includes the activity of zooplankton, filter-feeding benthos,
and other pelagic filter feeders including planktivorous fish. Formulation and
results of the zooplankton computation appear elsewhere in this report. Details of
the benthos computations may be found in HydroQual (2000) and at
http://www.chesapeakebay.net/model.htm. Predation by other planktivores is
modeled by assuming predators clear a specific volume of water per unit biomass:
PR=F«B«M
(6-12)
F = filtration rate (m3 g"1 predator C d"1)
M = planktivore biomass (g C m"3)
Detailed specification of the spatial and temporal distribution of the predator
population is impossible. One approach is to assume predator biomass is propor-
tional to algal biomass, M = y B, in which case Equation 6-12 can be rewritten:
PR = y • F • B2
(6-13)
Since neither y nor F are known precisely, the logical approach is to combine
their product into a single unknown determined during the model calibration proce-
dure. Effect of temperature on predation is represented with the same formulation
as the effect of temperature on respiration. The final representation of predation,
including zooplankton, is:
PR = -
B
KHsz + B
• RMsz • SZ
(6-14)
B
KHlz + B
RMsz = microzooplankton maximum ration (g algal C g"1 zoo C d"1)
SZ = microzooplankton biomass (g C m"3)
KHsz = half saturation concentration for carbon uptake by microzooplankton
(g C m-3)
-100-
-------
Chapter 6 • Water Quality Model Formulation
RMlz = mesozooplankton maximum ration (g algal C g"1 zoo C d"1)
LZ = mesozooplankton biomass (g C m"3)
KHlz = half saturation concentration for carbon uptake by mesozooplankton
(g C m-3)
Phtl = rate of predation by other planktivores (m3 g"1 C d"1)
Predation by filter-feeding benthos is represented as a loss term only in model cells
that intersect the bottom.
Accounting for Algal Phosphorus
The amount of phosphorus incorporated in algal biomass is quantified through
a stoichiometric ratio. Thus, total phosphorus in the model is expressed:
TotP = PO4 + Ape • B + OOP + LPOP + RPOP (6-15)
TotP = total phosphorus (g P m"3)
PO4 = dissolved phosphate (g P m"3)
Ape = algal phosphorus-to-carbon ratio (g P g"1 C)
DOP = dissolved organic phosphorus (g P m"3)
LPP = labile particulate organic phosphorus (g P m"3)
RPP = refractory particulate organic phosphorus (g P m"3)
Algae take up dissolved phosphate during production and release dissolved
phosphate and organic phosphorus through respiration. The fate of phosphorus
released by respiration is determined by empirical distribution coefficients. The fate
of algal phosphorus recycled by predation is determined by a second set of distribu-
tion parameters.
Accounting for Algal Nitrogen
Model nitrogen state variables include ammonium+urea, nitrate+nitrite,
dissolved organic nitrogen, labile particulate organic nitrogen, and refractory
particulate organic nitrogen. The amount of nitrogen incorporated in algal biomass
is quantified through a stoichiometric ratio. Thus, total nitrogen in the model is
expressed:
TotN = NH4 - Urea + NO23
(6-16)
+ Anc • B + DON + LPON + RPON
TotN = total nitrogen (g N m"3)
NH4-Urea = ammonium+urea (g N m"3)
NO23 = nitrate+nitrite (g N nr3)
Anc = algal nitrogen-to-carbon ratio (g N g"1 C)
DON = dissolved organic nitrogen (g N m"3)
LPON = labile particulate organic nitrogen (g N m"3)
RPON = refractory particulate organic nitrogen (g N m"3)
As with phosphorus, the fate of algal nitrogen released by metabolism and
predation is represented by distribution coefficients.
-101 -
-------
Chapter 6 • Water Quality Model Formulation
Algal Nitrogen Preference
Algae take up ammonium+urea and nitrate+nitrite during production and release
ammonium+urea and organic nitrogen through respiration. Nitrate+nitrite is inter-
nally reduced to ammonium before synthesis into biomass occurs (Parsons et al.
1984). Trace concentrations of ammonium+urea inhibit nitrate reduction so that, in
the presence of multiple nitrogenous nutrients, ammonium+urea is utilized first.
The "preference" of algae for ammonium+urea is expressed by an empirical func-
tion (Thomann and Fitzpatrick 1982):
= NH4-Urea
+ NH4-Urea
NO23
(KHn + NH4 - Urea) • (KHn + N023)
KHn
(6-17)
(NH4-Urea
n + N023)
in which
PN = algal preference for ammonium uptake (0 < Pn < 1)
KHn = half saturation concentration for algal nitrogen uptake (g N m~3)
The function has two limiting values (Figure 6-6). When nitrate+nitrite is absent,
the preference for ammonium+urea is unity. When ammonium+urea is absent, the
preference is zero. In the presence of ammonium+urea and nitrate+nitrite, the pref-
erence depends on the abundance of both forms relative to the half-saturation
constant for nitrogen uptake. When both ammonium+urea and nitrate+nitrite are
abundant, the preference for ammonium+urea approaches unity. When ammo-
nium+urea is scarce but nitrate+nitrite is abundant, the preference decreases in
magnitude and a significant fraction of algal nitrogen requirement comes from
nitrate+nitrite.
1.1 -1
1 -
0.9 -
| 0.8-
fe 0.7 -
M—
£ n a
ol 0.6 -
| 0.5 -
° 0.4 -
| 0.3-
0.2 -
0.1 -
C
NH4/KH = 0.5
NH4/KH = 1.0
>W NH4/KH = 2.0
i'^>~~— NH4/KH-5.0
""-- NH4 / KH - 10
\ v_
\
"•---.-._. -•
KH = 0.01 gm N / m3
)1 23456789 10
NO3 / KH
Figure 6-6. Algal ammonium preference.
-102-
-------
Chapter 6 • Water Quality Model Formulation
Effect of Algae on Dissolved Oxygen
Algae produce oxygen during photosynthesis and consume oxygen through
respiration. The quantity produced depends on the form of nitrogen utilized for
growth. More oxygen is produced, per unit of carbon fixed, when nitrate is the
algal nitrogen source than when ammonium is the source. Equations describing
algal uptake of carbon and nitrogen and production of dissolved oxygen (Morel
1983) are:
106 C02 + 16 NHl + H2POi + 106 H20-->
(6-18)
protoplasm + 106 02 + 15 H+
106 CO2 + 16 NO3 + H2POi + 122 H2O + 17 H+-->
(6-19)
protoplasm + 138 Q2
When ammonium is the nitrogen source, one mole oxygen is produced per mole
carbon dioxide fixed. When nitrate is the nitrogen source, 1.3 moles oxygen are
produced per mole carbon dioxide fixed.
The equation that describes the effect of algae on dissolved oxygen in the model is:
— DO = [(1.3-0.3 • PN) • P-(l-FCD) • BMJ • AOCR • B (6-20)
St
in which:
FCD = fraction of algal metabolism recycled as dissolved organic carbon
(0 < FCD < 1)
AOCR = dissolved oxygen-to-carbon ratio in respiration (2.67 g O2 g"1 C)
The magnitude of AOCR is derived from a simple representation of the respira-
tion process:
The quantity (1.3 - 0.3 • PN) is the photosynthesis ratio and expresses the molar
quantity of oxygen produced per mole carbon fixed. The photosynthesis ratio
approaches unity as the algal preference for ammonium approaches unity.
Accounting for Algal Silica
The amount of silica incorporated in algal biomass is quantified through a stoi-
chiometric ratio. Thus, total silica in the model is expressed:
TotSi = Dsil + Asc . B + PBS (6-22)
-103-
-------
Chapter 6 • Water Quality Model Formulation
TotSi = total silica (g Si nr3)
Dsil = dissolved silica (g Si m~3)
Asc = algal silica-to-carbon ratio (g Si g"1 C)
PBS = particulate biogenic silica (g Si nr3)
As with the other nutrients, the fate of algal silica released by metabolism and
predation is represented by distribution coefficients.
Salinity Toxicity
The Cyanobacteria represented in the model are freshwater organisms that cease
production when salinity exceeds 1 to 2 ppt (Sellner et al. 1988). The effect of
salinity on Cyanobacteria was represented by a mortality term in the form of a
rectangular hyperbola:
STOX1 = STF1
(6-23)
KHstl + S
in which
STOX1 = mortality induced by salinity on Cyanobacteria (d"1)
STF1 = maximum salinity mortality on Cyanobacteria (d"1)
S = salinity (ppt)
KHstl = salinity at which mortality is half maximum value (ppt)
The spring diatom bloom is limited to saline water. The limiting mechanism is
not defined but appears to be related to salinity. The upstream limit of the spring
bloom was defined in the model by introducing a mortality term at low salinity:
STOX2 = STF2 • - (6-24)
KHst2 + S v '
in which
STOX2 = mortality induced by freshwater on spring diatoms (d"1)
STF2 = maximum freshwater mortality on spring diatoms (d"1)
S = salinity (ppt)
KHst2 = salinity at which mortality is half maximum value (ppt)
The salinity-related mortality (Figure 6-7) is added to the basal metabolism.
Representation of Blue-Green Algal Mats
Microcystis commonly form algal mats. Exact representation of algal mats is
beyond the present capacity of the model. Mat formation was approximated by
restricting Cyanobacteria growth to the surface model layer.
-104-
-------
Chapter 6 • Water Quality Model Formulation
Cyanobactena
- Spring Diatoms
10
15 20
Salinity (ppt)
25
30
35
Figure 6-7. Salinity toxicity relationship.
Organic Carbon
Organic carbon undergoes innumerable transformations in the water column.
The model carbon cycle (Figure 6-8) consists of the following elements:
• Phytoplankton production and excretion
• Zooplankton production and excretion
• Predation on phytoplankton
• Dissolution of particulate carbon
• Heterotrophic respiration
• Denitrification
• Settling
Algal production is the primary carbon source although carbon also enters the
system through external loading. Predation on algae by zooplankton and other
organisms releases particulate and dissolved organic carbon to the water column. A
fraction of the particulate organic carbon undergoes first-order dissolution to
dissolved organic carbon. Dissolved organic carbon produced by excretion, by
predation, and by dissolution is respired at a first-order rate to inorganic carbon.
Particulate organic carbon which does not undergo dissolution settles to the bottom
sediments.
Zooplankton kinetics are detailed in a separate chapter. Kinetics of the organic
carbon state variables are described below.
-105-
-------
Chapter 6 • Water Quality Model Formulation
Dissolved
Inorganic Carbon
photosynthesis
Figure 6-8. Model carbon cycle.
Dissolution and Respiration
Organic carbon dissolution and respiration are represented as first-order
processes in which the reaction rate is proportional to concentration of the
reactant. An exponential function (Figure 6-5) relates dissolution and respiration
to temperature.
In the model, a Monod-like function diminishes respiration as dissolved oxygen
approaches zero. As oxygen is depleted from natural systems, oxidation of organic
matter is effected by the reduction of alternate oxidants. The sequence in which
alternate oxidants are employed is determined by the thermodynamics of oxidation-
reduction reactions. The first substance reduced in the absence of oxygen is nitrate.
A representation of the denitrification reaction can be obtained by balancing stan-
dard half-cell redox reactions (Stumm and Morgan 1981):
4 NO3
CH2O-->2
H2O + 5 CO2
(6-25)
Equation 6-25 describes the stoichiometry of the denitrification reaction. The
kinetics of the reaction, represented in the model, are first-order. The dissolved
organic carbon respiration rate, Kdoc, is modified so that significant decay via
denitrification occurs only when nitrate is freely available and dissolved oxygen is
depleted (Figure 6-9). A parameter is included so that the anoxic respiration rate is
slower than oxic respiration:
-106-
-------
Chapter 6 • Water Quality Model Formulation
1 -,
_ 0.8 -
O
-S n 7
X.
S- 0.4 -
1 0.3 -
Q
0.2 -
0.1 -
C
_____ " """ DO / KHodoc = 0.0
^.. •-'""' DO / KHodoc = 1.0
,-' DO / KHodoc = 10.
/
/ _- - -
- '' ,.^'"
) 1 23456789 10
NO3 / KHndn
Figure 6-9. Effect of dissolved oxygen and nitrate on denitrification.
Denit = -
KHodoc
N03
KHodoc + DO KHndn+ N03
AANOX • Kdoc
(6-26)
in which:
Denit = denitrification rate of dissolved organic carbon (d"1)
Kdoc = first-order dissolved organic carbon respiration rate (d"1)
AANOX = ratio of denitrification to oxic carbon respiration rate
(0 < AANOX < 1)
KHodoc = half-saturation concentration of dissolved oxygen required for oxic
respiration (g O2 m~3)
KHndn = half-saturation concentration of nitrate required for denitrification
(g N m-3)
Dissolved Organic Carbon
The complete representation of dissolved organic carbon sources and sinks in the
model ecosystem is:
— DOC = FCD • R • B + FCDP • PR + Klpoc • LPOC
§ t
+ Krpoc • RPOC-
DO
KHodoc + DO
Kdoc • DOC - DENIT • DOC
(6-27)
-107-
-------
Chapter 6 • Water Quality Model Formulation
in which:
DOC = dissolved organic carbon (g m~3)
LPOC = labile particulate organic carbon (g m~3)
RPOC = refractory participate organic carbon (g m~3)
FCD = fraction of algal respiration released as DOC (0 < FCD < 1)
FCDP = fraction of predation on algae released as DOC (0 < FCDP < 1)
Klpoc = dissolution rate of LPOC (d'1)
Krpoc = dissolution rate of RPOC (d'1)
Kdoc = respiration rate of DOC (d"1)
Labile Participate Organic Carbon
The complete representation of labile particulate organic carbon sources and
sinks in the model ecosystem is:
5
LPOC = FCL • R • B + FCLP • PR - Klpoc • LPOC
8 t (6-28)
-Wl • — LPOC
8z
in which:
FCL = fraction of algal respiration released as LPOC (0 < FCL < 1)
FCLP = fraction of predation on algae released as LPOC (0 < FCLP < 1)
Wl = settling velocity of labile particles (m d"1)
Refractory Particulate Organic Carbon
The complete representation of refractory particulate organic carbon sources
and sinks in the model ecosystem is:
— RPOC = FCR • R • B + FCRP • PR-Krpoc • RPOC
5 l (6-29)
-Wr • — RPOC
Sz
in which:
FCR = fraction of algal respiration released as RPOC (0 < FCR < 1)
FCRP = fraction of predation on algae released as RPOC (0 < FCRP < 1)
Wr = settling velocity of refractory particles (m d"1)
Phosphorus
The model phosphorus cycle (Figure 6-10) includes the following processes:
• Algal uptake and excretion
• Zooplankton excretion
• Predation
• Hydrolysis of particulate organic phosphorus
• Mineralization of dissolved organic phosphorus
• Settling and resuspension
-108-
-------
Chapter 6 • Water Quality Model Formulation
Figure 6-10. Model phosphorus cycle.
External loads provide the ultimate source of phosphorus to the system.
Dissolved phosphate is incorporated by algae during growth and released as phos-
phate and organic phosphorus through respiration and predation. Dissolved organic
phosphorus is mineralized to phosphate. A portion of the particulate organic phos-
phorus hydrolyzes to dissolved organic phosphorus. The balance settles to the
sediments. Within the sediments, particulate phosphorus is mineralized and recy-
cled to the water column as dissolved phosphate.
Hydrolysis and Mineralization
Within the model, hydrolysis is defined as the process by which particulate
organic substances are converted to dissolved organic form. Mineralization is
defined as the process by which dissolved organic substances are converted to
dissolved inorganic form. Conversion of particulate organic phosphorus to phos-
phate proceeds through the sequence of hydrolysis and mineralization. Direct
mineralization of particulate organic phosphorus does not occur.
Mineralization of organic phosphorus is mediated by the release of nucleoti-
dase and phosphatase enzymes by bacteria (Ammerman and Azam 1985; Chrost
and Overbeck 1987) and algae (Matavulj and Flint 1987; Chrost and Overbeck
1987; Boni et al. 1989). Since the algae themselves release the enzyme and since
bacterial abundance is related to algal biomass, the rate of organic phosphorus
mineralization is related, in the model, to algal biomass. A most remarkable prop-
erty of the enzyme process is that alkaline phosphatase activity is inversely
proportional to ambient phosphate concentration (Chrost and Overbeck 1987; Boni
-109-
-------
Chapter 6 • Water Quality Model Formulation
et al. 1989). Put in different terms, when phosphate is scarce, algae stimulate
production of an enzyme that mineralizes organic phosphorus to phosphate. This
phenomenon is simulated by relating mineralization to the algal phosphorus
nutrient limitation. Mineralization is highest when algae are strongly phosphorus
limited and is least when no limitation occurs.
The expression for mineralization rate is:
Kdop = Kdp +
KHp
• Kdpalg • B
(6-30)
in which:
Kdop = mineralization rate of dissolved organic phosphorus (d"1)
Kdp = minimum mineralization rate (d"1)
KHp = half-saturation concentration for algal phosphorus uptake (g P m~3)
PO4 = dissolved phosphate (g P m~3)
Kdpalg = constant that relates mineralization to algal biomass (m3 g"1 C d"1)
Potential effects of algal biomass and nutrient limitation on the mineralization
rate are shown in Figure 6-11. When nutrient concentration greatly exceeds the
half-saturation concentration for algal uptake, the rate roughly equals the minimum.
Algal biomass has little influence. As nutrient becomes scarce relative to the half-
saturation concentration, the rate increases. The magnitude of the increase depends
on algal biomass. Factor of two to three increases are feasible.
Exponential functions (Figure 6-4) relate mineralization and hydrolysis rates to
temperature.
Precipitation and Bacterial Uptake
Functions to represent precipitation and uptake by sulfide-oxidizing bacteria
were added during the model calibration procedure. These functions are described
in a subsequent chapter.
3.5
3
0.5
Kmin = 0.1 /day
Kalg = 0.2 m"3 / gm C / day
0 1
7 8
N/KH
—I 1 1 1 1 1
10 11 12 13 14 15
Figure 6-11. Effect of algal biomass and nutrient concentration on
phosphorus mineralization.
-110-
-------
Chapter 6 • Water Quality Model Formulation
Dissolved Phosphate
The mass-balance equation for dissolved phosphate is:
— PO4 = Kdop • DOP-APC • G • B , ,
5 t W-jlj
+ APC • [FPI • BM «B + FPIP • PR]
in which:
FPI = fraction of algal metabolism released as dissolved phosphate (0 < FPI < 1)
FPIP = fraction of predation released as dissolved phosphate (0 < FPIP < 1)
Dissolved Organic Phosphorus
The mass balance equation for dissolved organic phosphorus is:
— DOP = APC • (BM • B • FPD + PR • FPDP) + Klpop • LPOP ff- »o,
8 t (o-3zj
+ Krpop • RPOP - Kdop • OOP
in which:
DOP = dissolved organic phosphorus (g P m~3)
LPOP = labile particulate organic phosphorus (g P m~3)
RPOP = refractory particulate organic phosphorus (g P m~3)
FPD = fraction of algal metabolism released as DOP (0 < FPD < 1)
FPDP = fraction of predation on algae released as DOP (0 < FPDP < 1)
Klpop = hydrolysis rate of LPOP (d'1)
Krpop = hydrolysis rate of RPOP (d'1)
Kdop = mineralization rate of DOP (d"1)
Labile Particulate Organic Phosphorus
The mass balance equation for labile particulate organic phosphorus is:
— LPOP = APC • (BM • B • FPL + PR • FPLP)-Klpop • LPOP ,, ...
5 t > v v (6-33)
e
-Wl • — LPOP
5z
in which:
FPL = fraction of algal metabolism released as LPOP (0 < FPL < 1)
FPLP = fraction of predation on algae released as LPOP (0 < FPLP < 1)
-111 -
-------
Chapter 6 • Water Quality Model Formulation
Refractory Particulate Organic Phosphorus
The mass balance equation for refractory particulate organic phosphorus is:
e
— RPOP = APC • (BM • B • FPR+PR • FPRP)-Krpop • RPOP
-Wr • — RPOP
5z
in which:
FPR = fraction of algal metabolism released as RPOP (0 < FPR < 1)
FPRP = fraction of predation on algae released as RPOP (0 < FPRP < 1)
Nitrogen
The model nitrogen cycle (Figure 6-12) includes the following processes:
• Algal production and metabolism
• Predation
• Hydrolysis of particulate organic nitrogen
• Mineralization of dissolved organic nitrogen
• Settling
• Nitrification
• Denitrification
Am
mineralization
i
moniu
hUrea
1 I
* Z<
Dissolved
Organic
Nitrogen
mil
nitrification
uptake
Two
soplankton
Groups
Nitrate+
Nitrite «
respiration
predation
Three Algal
Groups
1 1
Labile Refracto
Particulate Parti cuh
Organic Organi
Nitrogen Nitroge
hydrolysis
leralization
r 1
Sediments
nitrification
ry
ite
c
n
«— 1
Figure 6-12. Model nitrogen cycle.
-112-
-------
Chapter 6 • Water Quality Model Formulation
External loads provide the ultimate source of nitrogen to the system. Available
nitrogen is incorporated by algae during growth and released as ammonium+urea
and organic nitrogen through respiration and predation. A portion of the particulate
organic nitrogen hydrolyzes to dissolved organic nitrogen. The balance settles to
the sediments. Dissolved organic nitrogen is mineralized to ammonium+urea. In an
oxygenated water column, a fraction of the ammonium+urea is subsequently
oxidized to nitrate+nitrite through the nitrification process. In anoxic water,
nitrate+nitrite is lost to nitrogen gas through denitrification. Particulate nitrogen
that settles to the sediments is mineralized and recycled to the water column,
primarily as ammonium+urea. Nitrate+nitrite moves in both directions across the
sediment-water interface, depending on relative concentrations in the water column
and sediment interstices.
Nitrification
Nitrification is a process mediated by specialized groups of autotrophic bacteria
that obtain energy through the oxidation of ammonium to nitrite and oxidation of
nitrite to nitrate. A simplified expression for complete nitrification (Tchobanoglous
and Schroeder 1987) is:
NH^ + 2 O2-->NO3 + H2O + 2 H+ C6'35)
The simplified stoichiometry indicates that two moles of oxygen are required to
nitrify one mole of ammonium into nitrate. The simplified equation is not strictly
true, however. Cell synthesis by nitrifying bacteria is accomplished by the fixation
of carbon dioxide so that less than two moles of oxygen are consumed per mole
ammonium utilized (Wezernak and Gannon 1968).
The kinetics of complete nitrification are modeled as a function of available
ammonium, dissolved oxygen, and temperature:
ATT^ NH4-Urea f
NT = - • - • f(T) • NTm 1
KHont + DO KHnnt + NH4-Urea
in which:
NT = nitrification rate (g N m~3 d"1)
KHont = half-saturation constant of dissolved oxygen required for nitrification
(g 02 m-3)
KHnnt = half-saturation constant of NH4 required for nitrification (g N m"3)
NTm = maximum nitrification rate at optimal temperature (g N m"3 d"1)
The kinetics formulation (Figure 6-13) incorporates the products of two Monod-
like functions. The first function diminishes nitrification at low dissolved oxygen
concentration. The second function expresses the influence of ammonium concen-
tration on nitrification. When ammonium concentration is low, relative to KHnnt,
nitrification is proportional to ammonium concentration. For NH4 = KHnnt, the
reaction is approximately first-order. (The first-order decay constant NTm/KHnnt.)
When ammonium concentration is large, relative to KHnnt, nitrification approaches
a maximum rate. This formulation is based on a concept proposed by Tuffey et al.
(1974). Nitrifying bacteria adhere to benthic or suspended sediments. When ammo-
nium is scarce, vacant surfaces suitable for nitrifying bacteria exist. As ammonium
-113-
-------
Chapter 6 • Water Quality Model Formulation
,1
z
^
2
1 -I
0.8 -
0.7 -
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1 -
0_
C
DO/KHont = 0.1
DO/KHont= 1.0
DO/KHont = 5.0
DO/KHont= 10.
_____——-
_—- — -:~~~~:~--""""
__--7.:''-:-"""~"
_-.<"-""" ._._-—--- — • — "~
,-'-'' -—~'"~~
t'ff'^ - '
) 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
NH4 / KHnnt
Figure 6-13. Effect of dissolved oxygen and ammonium concentra-
tion on nitrification rate.
concentration increases, bacterial biomass increases, vacant surfaces are occupied,
and the rate of nitrification increases. The bacterial population attains maximum
density when all surfaces suitable for bacteria are occupied. At this point, nitrifica-
tion proceeds at a maximum rate independent of additional increase in ammonium
concentration.
The optimal temperature for nitrification may be less than peak temperatures that
occur in coastal waters. To allow for a decrease in nitrification at superoptimal
temperature, the effect of temperature on nitrification is modeled in the Gaussian
form of Equation 6-7.
Effect of Denitrification on Nitrate
The effect of denitrification on dissolved organic carbon has been described.
Denitrification removes nitrate from the system in stoichiometric proportion to
carbon removal:
— NO3 = -ANDC • Denit • DOC
5t
(6-37)
in which:
ANDC = mass nitrate-nitrogen reduced per mass dissolved organic carbon
oxidized (0.933 g N g-1 C)
-114-
-------
Chapter 6 • Water Quality Model Formulation
Nitrogen Mass Balance Equations
The mass-balance equation for nitrogen state variables are written by summing
all previously-described sources and sinks:
Ammonium+Urea
A NH4-Urea = ANC • [(BM • FNI-PN • P) • B + PR • FNIPJ ,, ,R,
§ t (o-joj
+ Kdon • DON-NT
in which:
FNI = fraction of algal metabolism released as NH4 (0 < FNI < 1)
PN = algal ammonium preference (0 < PN < 1)
FNIP = fraction of predation released as NH4 (0 < FNIP < 1)
Nitrate+Nitrite
-^- N023=-ANC • (1-PN) • P • B + NT
5 t (6-39)
-ANDC • Denit • DOC
Dissolved Organic Nitrogen
— DON = ANC • (BM • B • FND + PR • FNDP) + Klpon • LPON ., ._
8 t (6-4U)
+ Krpon • RPON - Kdon • DON
in which:
DON = dissolved organic nitrogen (g N m~3)
LPON = labile particulate organic nitrogen (g N m~3)
RPON = refractory particulate organic nitrogen (g N m~3)
FND = fraction of algal metabolism released as DON (0 < FND < 1)
FNDP = fraction of predation on algae released as DON (0 < FNDP < 1)
Klpon = hydrolysis rate of LPON (d'1)
Krpon = hydrolysis rate of RPON (d'1)
Kdon = mineralization rate of DON (d"1)
Labile Particulate Organic Nitrogen
— LPON = ANC • (BM • B • FNL+PR • FNLP)-Klpon • LPON (
5 t
. Wl • — LPON
5z
in which:
FNL = fraction of algal metabolism released as LPON (0 < FNL < 1)
FNLP = fraction of predation on algae released as LPON (0 < FNLP < 1)
-115-
-------
Chapter 6 • Water Quality Model Formulation
Refractory Particulate Organic Nitrogen
— RPON = ANC • (BM • B • FPR + PR • FPRN)-Krpon • RPON
8 t
o
-Wr • — RPON
5z
in which:
FNR = fraction of algal metabolism released as RPON (0 < FNR < 1)
FNRP = fraction of predation on algae released as RPON (0 < FNRP < 1)
Silica
The model incorporates two siliceous state variables, dissolved silica and partic-
ulate biogenic silica. The silica cycle (Figure 6- 14) is a simple one in which
diatoms take up disolved silica and recycle dissolved and particulate biogenic silica
through the actions of metabolism and predation. Particulate silica dissolves in the
water column or settles to the bottom. A portion of the settled particulate biogenic
dissolves within the sediments and returns to the water column as dissolved silica.
Sources and sinks represented are:
• Diatom production and metabolism
• Predation
• Dissolution of particulate to dissolved silica
• Settling
Dissolved Silica
Diatoms
predation
Two Zooplankton
Groups
Particulate
Biogenic Silica
recycle
Sediments
Figure 6-14. Model silica cycle.
-116-
-------
Chapter 6 • Water Quality Model Formulation
Dissolved Silica
The kinetics equation for dissolved silica is:
— Dsil = (FSAP • PR - P) • ASC • B + Ksua • PBS „
in which:
Dsil = dissolved silica (g Si m~3)
PBS = participate biogenic silica concentration (g Si m~3)
FSAP = fraction of diatom silica made available by predation (0 < FSAP < 1)
ASC = algal silica-to-carbon ratio (g Si g"1 C)
Ksua = particulate silica dissolution rate (d"1)
Participate Biogenic Silica
The kinetics equation for particulate biogenic silica is:
c*
— PBS = (BM + (i - FSAP) • PR) • ASC • B
8t
-Wpbs — PBS-Ksua • PBS
Sz
(6-44)
in which:
Wpbs = biogenic silica settling rate (m d"1)
An exponential function (Figure 6-4) describes the effect of temperature on
silica dissolution.
Chemical Oxygen Demand
Chemical oxygen demand is the concentration of reduced substances that are
oxidized through abiotic reactions. The source of chemical oxygen demand in
saline water is sulfide released from sediments. A cycle occurs in which sulfate is
reduced to sulfide in the sediments and reoxidized to sulfate in the water column.
In freshwater, methane is released to the water column by the sediment model.
Both sulfide and methane are quantified in units of oxygen demand and are treated
with the same kinetics formulation:
A COD = — • Kcod • COD (6-45)
5 t KHocod + DO
in which:
COD = chemical oxygen demand concentration (g oxygen-equivalents m"3)
KHocod = half-saturation concentration of dissolved oxygen required for exertion
of chemical oxygen demand (g O2 m"3)
Kcod = oxidation rate of chemical oxygen demand (d"1)
An exponential function (Figure 6-5) describes the effect of temperature on
exertion of chemical oxygen demand.
-117-
-------
Chapter 6 • Water Quality Model Formulation
Dissolved Oxygen
Sources and sinks of dissolved oxygen in the water column (Figure 6-15)
include:
• Algal photosynthesis
• Atmospheric reaeration
• Algal respiration
• Heterotrophic respiration
• Nitrification
• Chemical oxygen demand
Atmospheric
Oxygen Ye aeration
heterotrophic
respiration
respiration
a
o
IS
o
-------
Chapter 6 • Water Quality Model Formulation
In freeflowing streams, the reaeration coefficient depends largely on turbulence
generated by bottom shear stress (O'Connor and Dobbins 1958). In lakes and
coastal waters, however, wind effects may dominate the reaeration process
(O'Connor 1983). For Chesapeake Bay, a relationship for wind-driven gas
exchange (Hartman and Hammond 1985) was employed:
Kr = Arear • Rv • Wms1
(6-47)
in which:
Arear = empirical constant (0.1)
Rv = ratio of kinematic viscosity of pure water at 20 °C to kinematic viscosity of
water at specified temperature and salinity
Wms = wind speed measured at 10 m above water surface (m s"1)
Hartman and Hammond (1985) indicate Arear takes the value 0.157. In the
present model, Arear is treated as a variable to allow for effects of wind sheltering,
for differences in height of local wind observations, and for other factors.
An empirical function (Figure 6-16) that fits tabulated values of Rw is:
Rv = 0.54+ 0.0233 • T-0.0020 • S
(6-48)
in which:
S = salinity (ppt)
T = temperature (°C)
1.4
1.3
1.2 -
1.1
1
* Table, S=0
• Table, S=35
Equation, S-0
Equation, S 35
15
Degrees C
20
25
Figure 6-16. Computed and tabulated values of Rv.
-119-
-------
Chapter 6 • Water Quality Model Formulation
Saturation dissolved oxygen concentration diminishes as temperature and
salinity increase. An empirical formula that describes these effects (Genet et al
1974) is:
DOs = 14.5532 - 0.38217 • T + 0.0054258 • T2
-CL • (1.665 x 10"4-5.866 x 10"6 • T + 9.796 X 10'8 • T2) ^
in which:
CL = chloride concentration (= salinity/1.80655)
Mass Balance Equation for Dissolved Oxygen
c
— DO = AOCR • [(1.3-0.3 • PN) • P-(l-FCD) • BM] • B
§ t
-AONT • NT — • AOCR • Kdoc • DOC ((- -m
KHodoc + DO (6-50)
D° • Kcod • COD + — • (DOs - DO)
KHocod + DO H
in which:
AOCR = oxygen-to-carbon mass ratio in production and respiration
(=2.67g02g-1C)
AONT = oxygen consumed per mass ammonium nitrified (= 4.33 g O2 g"1 N)
Temperature
Computation of temperature employs a conservation of internal energy equa-
tion that is analogous to the conservation of mass equation. For practical purposes,
the internal energy equation can be written as a conservation of temperature equa-
tion. The only source or sink of temperature considered is exchange with the
atmosphere. Atmospheric exchange is considered proportional to the temperature
difference between the water surface and a theoretical equilibrium temperature
(Edinger et al. 1974):
o „ KT
T = • (Te - T) ff- c n
5 t p • Cp • H I6'51)
in which:
T = water temperature (°C)
Te = equilibrium temperature (°C)
KT = Heat exchange coefficient (watt m~2 °C"1)
Cp = specific heat of water (4200 watt s kg"1 °C"1)
p = density of water (1000 kg m~3)
-120-
-------
Chapter 6 • Water Quality Model Formulation
Inorganic (Fixed) Solids
The only kinetics transformation of fixed solids is settling:
— ISS = -Wiss • — ISS (6-52)
in which:
ISS = fixed solids concentration (g m~3)
Wiss = solids settling velocity (m d"1)
Salinity
Salinity is modeled by the conservation of mass equation with no internal
sources or sinks
Parameter Values
Model parameter evaluation is a recursive process. Parameters are selected from
a range of feasible values, tested in the model, and adjusted until satisfactory agree-
ment between predicted and observed variables is obtained. Ideally, the range of
feasible values is determined by observation or experiment. For some parameters,
however, no observations are available. Then, the feasible range is determined by
parameter values employed in similar models or by the judgement of the modeler.
A review of parameter values was included in documentation of the first applica-
tion of this model (Cerco and Cole 1994). Parameters from the initial study were
refined, where necessary, for the Virginia Tributary Refinements (Cerco et al. 2002)
and refined again for the present model. A complete set of parameter values is
provided in Table 6-2. Subsequent chapters describe derivation of parameters that
were not in the original study.
-121 -
-------
Chapter 6 • Water Quality Model Formulation
Table 6-2
Parameters in Kinetics Equations
Symbol
AANOX
ANC
AOCR
AONT
ARC
Areaer
ASC
BM
FCD
FCDP
FCL
FCLP
FOR
FCRP
FNI
FNIP
FND
Definition
ratio of anoxic to oxic respiration
nitrogen-to-carbon ratio of algae
dissolved oxygen-to-carbon ratio in
respiration
mass dissolved oxygen consumed per
mass ammonium nitrified
algal phosphorus-to-carbon ratio
empirical constant in reaeration equation
algal silica-to-carbon ratio
basal metabolic rate of algae at reference
temperature Tr
fraction of dissolved organic carbon
produced by algal metabolism
fraction of dissolved organic carbon
produced by predation
fraction of labile particulate carbon
produced by algal metabolism
fraction of labile particulate carbon
produced by predation
fraction of refractory particulate carbon
produced by algal metabolism
fraction of refractory particulate carbon
produced by predation
fraction of inorganic nitrogen produced by
algal metabolism
fraction of inorganic nitrogen produced by
predation
fraction of dissolved organic nitrogen
produced by algal metabolism
Value
0.5
0.135 (spring),
0.1 75 (other)
2.67
4.33
0.01 75 (green),
0.01 25 (other)
0.078
0.0 (cyan),
0.4 (spring),
0.3 (green)
0.03 (cyan),
0.01 (spring),
0.02 (green)
0.0
0.15
0.0
0.65
0.0
0.2
0.55
0.4
0.2
Units
0 < AANOX < 1
g N g 1 C
9 O2 g"1 C
9 02 g 1 N
g P g 1 c
g si g 1 c
d1
0< FCD< 1
0
-------
Chapter 6 • Water Quality Model Formulation
Table 6-2 (continued)
Parameters in Kinetics Equations
Symbol
FNDP
FNL
FNLP
FNR
FNRP
FPD
FPDP
FPI
FPIP
FPL
FPLP
FPR
FPRP
FSAP
Kcod
Kdoc
Kdon
Kdp
Kdpalg
KHn
Definition
fraction of dissolved organic nitrogen
produced by predation
fraction of labile particulate nitrogen
produced by algal metabolism
fraction of labile particulate nitrogen
produced by predation
fraction of refractory particulate nitrogen
produced by algal metabolism
fraction of refractory particulate nitrogen
produced by predation
fraction of dissolved organic phosphorus
produced by algal metabolism
fraction of dissolved organic phosphorus
produced by predation
fraction of dissolved inorganic phosphorus
produced by algal metabolism
fraction of dissolved inorganic phosphorus
produced by predation
fraction of labile particulate phosphorus
produced by algal metabolism
fraction of labile particulate phosphorus
produced by predation
fraction of refractory particulate
phosphorus produced by algal metabolism
fraction of refractory particulate
phosphorus produced by predation
fraction of dissolved silica produced by
predation
oxidation rate of chemical oxygen demand
dissolved organic carbon respiration rate
dissolved organic nitrogen mineralization
rate
minimum mineralization rate of dissolved
organic phosphorus
constant that relates mineralization rate to
algal biomass
half-saturation concentration for nitrogen
uptake by algae
Value
0.2
0.2
0.25
0.05
0.15
0.25
0.4
0.5
0.75
0.0
0.07
0.0
0.03
0.5
20
0.011 to 0.075
0.025
0.15
0.4
0.02 (cyan),
0.025 (other)
Units
0< FNDP< 1
0< FNL< 1
0
-------
Chapter 6 • Water Quality Model Formulation
Table 6-2 (continued)
Parameters in Kinetics Equations
Symbol
KHndn
KHnnt
KHocod
KHodoc
KHont
KHp
KHs
KHst
Klpoc
Klpon
Klpop
Krpoc
Krpon
Krpop
Ksua
KTb
KTcod
KTg1
Definition
half-saturation concentration of nitrate
required for denitrification
half-saturation concentration of NH4
required for nitrification
half-saturation concentration of dissolved
oxygen required for exertion of COD
half-saturation concentration of dissolved
oxygen required for oxic respiration
half-saturation concentration of dissolved
oxygen required for nitrification
half-saturation concentration for
phosphorus uptake by algae
half-saturation concentration for silica
uptake by algae
salinity at which algal mortality is half
maximum value
labile particulate organic carbon
dissolution rate
labile particulate organic nitrogen
hydrolysis rate
labile particulate organic phosphorus
hydrolysis rate
refractory particulate organic carbon
dissolution rate
refractory particulate organic nitrogen
hydrolysis rate
refractory particulate organic phosphorus
hydrolysis rate
biogenic silica dissolution rate
effect of temperature on basal metabolism
of algae
effect of temperature on exertion of
chemical oxygen demand
effect of temperature below Tm on growth
of algae
Value
0.1
1.0
0.5
0.5
1.0
0.0025
0.0 (cyan),
0.03 (spring),
0.01 (green)
0.5 (cyan),
2.0 (spring)
0.02 to 0.1 5
0.12
0.24
0.005 to 0.01
0.005
0.01
0.1
0.032
0.041
0.005 (cyan),
0.0018 (spring),
0.0035 (green)
Units
gNm3
g N m 3
g O2 m 3
g O2 m 3
g O2 m 3
gPm3
g Si m 3
ppt
d1
d1
d1
d1
d1
d1
d1
0Q-1
d1
oc-2
continued
-124-
-------
Chapter 6 • Water Quality Model Formulation
Table 6-2 (continued)
Parameters in Kinetics Equations
Symbol
KTg2
KThdr
KTmnl
Kind
KTnt2
KTsua
NTm
Phtl
PmB
Presp
STF
Topt
Tmnt
Tr
Trhdr
Trmnl
Trsua
Wa
Definition
effect of temperature above Tm on growth
of algae
effect of temperature on hydrolysis rates
effect of temperature on mineralization
rates
effect of temperature below Tmnt on
nitrification
effect of temperature above Tmnt on
nitrification
effect of temperature on biogenic silica
dissolution
maximum nitrification rate at optimal
temperature
predation rate on algae
maximum photosynthetic rate
photo-respiration fraction
salinity toxicity factor
optimal temperature for growth of algae
optimal temperature for nitrification
reference temperature for metabolism
reference temperature for hydrolysis
reference temperature for mineralization
reference temperature for biogenic silica
dissolution
algal settling rate
Value
0.004 (cyan),
0.006 (spring),
0.0 (green)
0.032
0.032
0.003
0.003
0.092
0.1 to 0.5
0.0 (cyan),
0.1 to 0.2
(spring),
0.5 to 2 (green)
100 (cyan),
300 (spring),
350 (green)
0.25
0.3 (cyan),
0.1 (spring)
29 (cyan),
16 (spring),
25 (green)
30
20
20
20
20
0.0 (cyan),
0.1 (other)
Units
oQ-2
= c-1
0Q-1
oQ-2
oc-2
o,,-!
g N m 3 d 1
m3 g 1 C d 1
g C g 1 Chi d"1
0 < Presp < 1
d1
°C
°C
°c
°c
°c
°c
m d"1
continued
-125-
-------
Chapter 6 • Water Quality Model Formulation
Table 6-2 (continued)
Parameters in Kinetics Equations
Symbol
Wl
Wr
Wiss
Wpbs
a
Definition
settling velocity of labile particles
settling velocity of refractory particles
settling velocity of fixed solids
settling velocity of biogenic silica
initial slope of production vs. irradiance
relationship
Value
0.1
0.1
1 to 4
0.1
3.15 (cyan),
8.0 (other)
Units
md1
md1
m d 1
md1
g C g 1 Chi
(E m 2) 1
References
Ammerman, J., and Azam, F. (1985). "Bacterial 5'-nucleodase in aquatic ecosystems: a
novel mechanism of phosphorus regeneration," Science, 227, 1338-1340.
Boni, L., Carpene, E., Wynne, D., and Reti, M. (1989). "Alkaline phosphatase activity in
Protogonyaulax Tamarensis," Journal of plankton research, 11, 879-885.
Bunch, B., Cerco, C., Dortch, M., Johnson, B., and Kim, K. (2000). "Hydrodynamic and
water quality model study of San Juan Bay and Estuary," ERDC TR-00-1, U.S. Army Engi-
neer Research and Development Center, Vicksburg MS.
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report EL-94-4, US Army Engineer Waterways Experiment Station, Vicks-
burg, MS.
Cerco, C., Bunch, B., Cialone, M., and Wang, H. (1994). "Hydrodynamic and eutrophica-
tion model study of Indian River and Rehoboth Bay, Delaware," Technical Report EL-94-5,
US Army Engineer Waterways Experiment Station, Vicksburg, MS.
Cerco, C., and Bunch, B. (1997). "Passaic River tunnel diversion model study, Report 5,
water quality modeling," Technical Report HL-96-2, US Army Engineer Waterways Experi-
ment Station, Vicksburg, MS.
Cerco, C., Johnson, B., and Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model," ERDC TR-02-4, US Army Engineer Research and Development Center, Vicksburg,
MS.
Chrost, R., and Overbeck, J. (1987). "Kinetics of alkaline phosphatase activity and phos-
phorus availability for phytoplankton and bacterioplankton in Lake Plubsee (north German
eutrophic lake)," Microbial Ecology, 13, 229-248.
Edinger, J., Brady, D., and Geyer, J. (1974). "Heat exchange and transport in the environ-
ment," Report 14, Department of Geography and Environmental Engineering, Johns
Hopkins University, Baltimore, MD.
Genet, L., Smith, D., and Sonnen, M. (1974). "Computer program documentation for the
Dynamic Estuary Model," US Environmental Protection Agency, Systems Development
Branch, Washington, DC.
Hartman, B., and Hammond, D. (1985). "Gas exchange in San Francisco Bay," Hydrobi-
ologia 129, 59-68.
-126-
-------
Chapter 6 • Water Quality Model Formulation
HydroQual (2000). "Development of a suspension feeding and deposit feeding benthos
model for Chesapeake Bay," Project USCE0410, prepared for US Army Engineer Research
and Development Center, Vicksburg MS.
Jassby, A., and Platt, T. (1976). "Mathematical formulation of the relationship between
photosynthesis and light for phytoplankton," Limnology and Oceanography 21, 540-547.
Leonard, B. (1979). "A stable and accurate convection modelling procedure based on quad-
ratic upstream interpolation," Computer Methods in Applied Mechanics and Engineering,
19, 59-98.
Matavulj, M., and Flint, K. (1987). "A model for acid and alkaline phosphatase activity in a
small pond," Microbial Ecology, 13, 141-158.
Monod, J. (1949). "The growth of bacterial cultures," Annual Review of Microbiology 3,
371-394.
Morel, F. (1983). Principles of Aquatic Chemistry, John Wiley and Sons, New York, NY,
150.
O'Connor, D., and Dobbins, W. (1958). "Mechanisms of reaeration in natural streams,"
Transactions of the American Society of Civil Engineers, 123, 641-666.
O'Connor, D. (1983). "Wind effects on gas-liquid transfer coefficients," Journal of the Envi-
ronmental Engineering Division, 190, 731-752.
Parsons, T, Takahashi, M., and Hargrave, B. (1984). Biological oceanographicprocesses.
3rd ed., Pergamon Press, Oxford.
Sellner, K., Lacoutre, R., and Parrish, C. (1988). "Effects of increasing salinity on a
Cyanobacteria bloom in the Potomac River Estuary," Journal of Plankton Research, 10, 49-
61.
Stumm, W, and Morgan, J. (1981). Aquatic chemistry. 2nd ed., Wiley- Interscience, New
York.
Thomann, R., and Fitzpatrick, J. (1982). "Calibration and verification of a mathematical
model of the eutrophication of the Potomac Estuary," HydroQual Inc., Mahwah, NJ.
Tchobanoglous, G., and Schroeder, E. (1987). Water quality, Addison Wesley, Reading,
MA.
Tuffey, T, Hunter, J., and Matulewich, V. (1974). "Zones of nitrification", Water Resources
Bulletin, 10, 555-564.
Wezernak, C., and Gannon, J. (1968). "Evaluation of nitrification in streams," Journal of the
Sanitary Engineering Division, 94(SA5), 883-895.
-127-
-------
Introduction to
Calibration
The Monitoring Data Base
The water quality model was applied to a ten-year time period, 1985-1994. The
monitoring data base maintained by the Chesapeake Bay Program (Chesapeake
Bay Program 1989, 1993) was the principal source of data for model calibration. A
summary of the monitoring program, as it operated from 1985 to 1994, is provided
below. Details of the data base and information on the current program may be
found at the Chesapeake Bay Program web site
http://www.chesapeakebay.net/data/index.htm.
Observations were collected at 49 stations in the bay and 89 stations in major
embayments and tributaries (Figure 7-1). Sampling was conducted once or twice
per month with more frequent sampling from March through October. Samples
were collected during daylight hours with no attempt to coincide with tide stage or
flow. Sampling periods were designated as "cruises." Under ideal circumstances,
cruises were completed in two to three days. Inclement weather or equipment fail-
ures occasionally produced cruises as long as two weeks.
At each station, salinity, temperature, and dissolved oxygen were measured in
situ at one- or two-meter intervals. Samples were collected one meter below the
surface and one meter above the bottom for laboratory analyses. At stations
showing significant salinity stratification, additional samples were collected above
and below the pycnocline. Analyses relevant to this study are listed in Table 7-1.
The listed parameters are either analyzed directly or derived from direct analyses.
The major portion of the data base was obtained from the sponsor, in SAS trans-
port format, in June 1996. Observations from 1991 through 1995 were updated in a
second transmission from the sponsor in June 1998. We discovered that observa-
tions in the Potomac River and Maryland eastern shore were not complete. These
were provided by Maryland Department of Environment in March 1999. Additional
observations were obtained from multiple sources. Descriptions are provided as
these observations are cited.
-128-
-------
Chapter 7 • Introduction to Calibration
Pennsylvania^.
Siiscjuehanna H .
Maryland
Baltimore WT»-K.
Pntapsco R.
__Wasi inot on
puij H cm.ip cwj S
pt-ank R .
Virgxnaa
Rirfiraond York H,
Iftz
TTS.2
•Virginia
Figure 7-1. Chesapeake Bay Program monitoring stations.
-129-
-------
Chapter 7 • Introduction to Calibration
Table 7-1
Direct or Derived Laboratory Analyses
Grouping
Phosphorus
Nitrogen
Carbon
Other
Parameter
Total Phosphorus
Total Dissolved Phosphorus
Particulate Phosphorus
Orthophosphate
Dissolved Inorganic Phosphorus
Dissolve Organic Phosphorus
Total Nitrogen
Total Dissolved Nitrogen
Particulate Organic Nitrogen
Total Kjeldahl Nitrogen, Whole/Filtered
Nitrate + Nitrite
Ammonium
Dissolved Inorganic Nitrogen
Dissolved Organic Nitrogen
Total Organic Nitrogen
Total Organic Carbon
Dissolved Organic Carbon
Particulate Organic Carbon
Silica, filtered
Total Suspended Solids
Chlorophyll >a=
Code
TP
TOP
PHOSP
P04F
DIP
OOP
TN
TON
PON
TKNW, TKNF
N023
NH4F
DIN
DON
TON
TOC
DOC
POC
SI
TSS
CHLA
-130-
-------
Chapter 7 • Introduction to Calibration
Comparison with the Model
Time Series
Time series comparisons of computations and observations were produced at
42 locations (Figure 7-2). These were selected to provide coverage throughout the
system. At least one station was selected within each of the Chesapeake Bay
Program Segments (CBPS) represented on the grid (Figure 7-3). This segmentation
scheme was based on salinity and circulation patterns. Within the model code,
daily-average concentrations were derived from computations at discrete time steps
(15 minutes). These were compared to individual observations, at surface and
bottom, for parameters listed in Table 7-2.
We wished to provide comparisons for the model cell corresponding to location
and depth at which samples were collected. For the surface sample, locating the
cell was straightforward. For the bottom sample, finding the corresponding model
cell was more difficult. The depth of the model grid usually represented the average
depth beneath the area covered by the surface cell. Consequently, grid depth was
less than the maximum prototype depth within the surface cell area. The moni-
toring program usually sampled at the maximum local depth, often within a
channel or deep hole. Depth of the bottom sample sometimes exceeded the
maximum depth of the model grid. In the event the sample depth exceeded the grid
depth, the vertical samples were interpolated into one-meter depth increments and
the model was matched to the corresponding interpolated data. For the few cases in
which grid depth exceeded the sample depth, the bottom sample was compared to
the bottom cell. No extrapolation was employed.
Time series comparisons of selected stations and parameters are provided in the text
of this report. Complete results are provided on CD-ROM that accompanies printed
copies of the report and are archived at the Chesapeake Bay Program Office.
Figure 7-2. Locations for time series comparisons.
-131 -
-------
Chapter 7 • Introduction to Calibration
Figure 7-3. Chesapeake Bay Program Segments.
-132-
-------
Chapter 7 • Introduction to Calibration
Table 7-2
Constituents in Model-Data Comparisons
Constituent
Chlorophyll >a=
Dissolved Inorganic Nitrogen
Dissolved Inorganic Phosphorus
Dissolved Organic Nitrogen
Dissolved Organic Phosphorus
Dissolved Oxygen
Light Attenuation
Ammonium
Nitrate + Nitrite
Particulate Organic Carbon
Particulate Organic Nitrogen
Particulate Phosphorus
Salinity
Total Nitrogen
Total Organic Carbon
Total Phosphorus
Total Suspended Solids
Temperature
Dissolved Silica
Time-Series
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Longitudinal
X
X
X
X
X
X
X
X
X
X
X
X
X
Vertical
X
X
X
Longitudinal Comparisons
The spatial distributions of observed and computed properties were compared in
a series of plots along the axes of the bay and major tributaries (Figures 7-4, 7-5).
The calibration period encompassed nearly 200 cruises. Reducing this number of
surveys into a manageable volume of comparisons required selection and aggrega-
tion. Three years were selected for comparisons: 1985, 1990, and 1993. The year
1985 was a low-flow year although the western tributaries were impacted by an
enormous flood event in November. Flows in 1990 were characterized as average
while major spring runoff occurred in 1993.
Model results and observations were averaged into four seasons:
• Winter—December through February
• Spring—March through May
• Summer—June through August
• Fall—September through November
-133-
-------
Chapter 7 • Introduction to Calibration
Figure 7-4. Longitudinal transects in the bay and major tributaries.
Figure 7-5. Longitudinal transects in the bay and major tributaries on transformed
grid.
-134-
-------
Chapter 7 • Introduction to Calibration
Conventional arithmetic means were calculated for the observations. Model
results were subjected to a process denoted as "cruise averaging." Within each
season, model results were considered only during intervals coinciding with sample
cruises. Cruise averaging diminished discrepancies between model and observa-
tions attributed to consideration of model results for periods when no data was
collected. Daily averages of model results were computed within the model code.
Cruise averaging was completed in a postprocessor. The postprocessor also
extracted the maximum and minimum computed daily averages.
The mean and range of the observations, at surface and bottom, were compared
to the cruise average and range of daily-average model results. A parameter list is
presented in Table 7-2. The longitudinal axes largely followed the maximum depths
represented on the model grid. Only stations located exactly on the transect
(Figures 7-6 to 7-11) were considered for comparison with the model. In the event
the depth of the bottom sample and grid depth did not coincide, observations were
treated as described for the time series.
Longitudinal comparisons of selected stations and parameters are provided in the
text of this report. Complete results are provided on CD-ROM that accompanies
printed copies of the report and are archived at the Chesapeake Bay Program Office.
CB8 CB7 CB6 CE
Ji
<_>
O
I i
E -E £ 2
O .G O b
—3>- Cd Q_
y ^ ^ £ ^ ^ ^
•n r-L K \n iri in
StD DO CO LO CD
•LJ 0 LJ 0 0
w
0 25 50 75 100 125
5 CB4 CEJ3 CB2 CB1
0)
-£ •£ % 8
| I S |
^ o >, 5
n -^ D J3
CL u m OL
r-1^,- Jj aH^^^, _ ^
in -*iri -J- •+ •+ ^ rO i*i eg i^J-H
CD tuCQ Cu m 11) tD CdCQCD DJCQ
UUtfC? U O O U (J CJ tJO
111 . _ II
II
1 " " ttntnt
tt
w
150 175 ZOO Z25 250 275 300 325 350
Kilometers
Figure 7-6. Elevation view of Chesapeake Bay transect showing kilometers from
mouth, sample stations, grid bathymetry, and Chesapeake Bay Program Segments.
-135-
-------
Chapter 7 • Introduction to Calibration
CB8
LE5
RET5
TF5
bj UJ UJ
. .
e e
e e e
-30 -20 -10 0 10 20 JO 40 50 60 70 SO BO 100 110 120 130 WO 150 160
Kilometers
Figure 7-7. Elevation view of James River transect showing kilometers from
mouth, sample stations, grid bathymetry, and Chesapeake Bay Program Segments.
c
20
37
C
_
BE
10
c
WE
I
EN
Jj
s
10
n
a
^ •
LE4 RET4 TR-
w T pf ^ >N
g g g £ £
1 1
__ U
20 30 40 50 60 70 80 90 10D 110 11D 130
Kilometers
Figure 7-8. Elevation view of York River transect showing kilometers from mouth,
sample stations, grid bathymetry, and Chesapeake Bay Program Segments.
-136-
-------
Chapter 7 • Introduction to Calibration
CB7 CB6
LE3
RET3
TF3
tn in
-30 -20 -10 0 10 20 30 40 50 60 70 SO BO 100 110 120 130 1+O 150 160
Kflo meters
Figure 7-9. Elevation view of Rappahannock River transect showing kilometers
from mouth, sample stations, grid bathymetry, and Chesapeake Bay Program.
CB5
LE2
RET2
TF2
zn
20 40 50 80 100 120 140 160 1BO 200
Kilometers
Figure 7-10. Elevation view of Potomac River transect showing kilometers from
mouth, sample stations, grid bathymetry, and Chesapeake Bay Program Segments.
-137-
-------
Chapter 7 • Introduction to Calibration
CB5
-10 -5
LE1 RET1 TF1
s ,_
*"i ** m tM — -^ r*- if> "*t ^
CO UJ bJLd u]^ EZ JZ C El
0 5 10 15 20 25 30 35 +D 45 50 55 80 65 70 75 HO
Kilo meters
Figure 7-11. Elevation view of Patuxent River transect showing kilometers from
mouth, sample stations, grid bathymetry, and Chesapeake Bay Program Segments.
Vertical Comparisons
The vertical distributions of observed and computed properties were compared at
selected stations in the bay and major tributaries. As with the longitudinal compar-
isons, selection and aggregation were required to produce a manageable volume of
information. Comparisons were completed for three years and were subjected to
seasonal averaging as previously described. Parameters were limited to the three for
which detailed vertical observations were available: temperature, salinity, and
dissolved oxygen.
Examination of the seasonally-averaged vertical data frequently showed a break
or jump near the bottom. The break was an artifact of the sampling program. Due
to varying topography and sampling methods, the bottom was not always found at
the same depth. Consequently, averages computed near the bottom sometimes
contained fewer samples than averages closer to the surface.
Vertical comparisons of selected stations and parameters are provided in the text
of this report. Complete results are provided on CD-ROM that accompanies printed
copies of the report and are archived at the Chesapeake Bay Program Office.
References
Chesapeake Bay Program. (1989). "Chesapeake Bay Basin Monitoring Program Atlas,"
CBP/TRS 34/89, US Environmental Protection Agency, Chesapeake Bay Liaison Office,
Annapolis MD.
Chesapeake Bay Program. (1993). "Guide to Using Chesapeake Bay Program Water
Quality Monitoring Data," CBP/TRS 78/92, Annapolis MD.
-138-
-------
The Zooplankton
Model
Introduction
Many of the earliest eutrophication models (e.g. DiToro et al. 1971; DiToro and
Matystik 1980) included one or more zooplankton groups as state variables. Later
efforts omitted zooplankton (e.g. Thomann and Fitzpatrick 1982). At present,
eutrophication models sans zooplankton are widely accepted (e.g. Cerco and Cole
1993). The reasons for dropping zooplankton are unclear. Lack of observations and
difficulty in calibration are two possibilities.
The present chapter details the formulation of zooplankton kinetics in the
Chesapeake Bay Environmental Model Package. Zooplankton were incorporated
into the model during the tributary refinements phase (Cerco, Johnson, and Wang
2002; Cerco and Meyers 2000). The primary reason was to advance the model into
the realm of living resources. Although zooplankton have no commercial value,
they are a prime food source for commercially-valuable fmfish. Addition of
zooplankton to the model framework was a first step towards modeling the effects
of eutrophication management on top-level predators. A secondary goal of
modeling zooplankton was to improve model accuracy in the computation of algal
biomass and other parameters.
Model Conceptualization
A conceptual food web for the bay exhibits seven components (Figure 8-1). At the
base of the food web are the phytoplankton. The phytoplankton are preyed upon by
zooplankton and by herbivorous finfish (in Chesapeake Bay, menhaden). Dissolved
organic carbon (DOC) secreted by plankton and produced from detritus is consumed
by heterotrophic bacteria which, along with phytoplankton, form a food source for
the microzooplankton. Microzooplankton are one of the prey groups for the meso-
zooplankton which also consume phytoplankton and detritus. The mesozooplankton
are a primary food source for carnivorous finfish (in Chesapeake Bay, anchovy).
-139-
-------
Chapter 8 • The Zooplankton Model
Algae
Dissolved
Organic C. art)on
13 a clena
1'aiticulate
Organic C a it on
Fish
M e s o -
zooplankton
Figure 8-1. The conceptual carbon cycle.
The art in modeling is to determine which of the food web components are to be
modeled and to describe the carbon and nutrient transfers between the groups. As
the primary producers, the phytoplankton must be modeled. The mesozooplankton
are the primary food source for the bay anchovies. Modeling the prey biomass and
mass flows into the fmfish are among the primary goals of the introduction of
living resources into the model suite, so mesozooplankton must be modeled. The
microzooplankton have no commercial value nor are they a primary food source for
the fmfish. They are an important prey for the mesozooplankton, however. More
importantly, they can be significant predators on the phytoplankton. Omission of
the microzooplankton requires a hybrid predation term on the phytoplankton in
which predation by mesozooplankton is computed but predation by microzoo-
plankton is specified. In view of the problems associated with a formulation of this
sort it is easier to include microzooplankton in the model than it is to omit them.
Heterotrophic bacteria are crucial components of the ecosystem. Their role in
water-column respiration and in nutrient recycling is conventionally represented as
first-order organic matter decay processes. These representations are adequate and,
in view of the complication of modeling both bacterial biomass and activity, can be
left unchanged. Any problem with omitting bacteria lies not with inadequate repre-
sentation of bacterial processes but with omission of a microzooplankton food
source. For the present, the omission is circumvented by allowing microzoo-
plankton to graze directly on DOC. The predation on mesozooplankton by fmfish is
handled as a predation term which closes the mesozooplankton equation.
-140-
-------
Chapter 8 • The Zooplankton Model
Model Carbon Cycle
Once the conceptual model has been formulated, the new state variables can be
added to the previously-established (Cerco and Cole 1994) model carbon cycle.
The resulting cycle (Figure 8-2) includes: three algal groups (spring diatoms, green
algae, and freshwater bloom-forming cyanobacteria); two zooplankton groups
(mesozooplankton and microzooplankton); two detritus groups (labile and refrac-
tory particulate organic carbon); and dissolved organic carbon. In the absence of
fmfish biomass, mass is conserved by routing consumed mesozooplankton back to
the detrital and dissolved organic carbon pools.
Figure 8-2. The model carbon cycle.
Additional Cycles
The eutrophication model also simulates cycling of nitrogen, phosphorus, silica,
and dissolved oxygen. The flows of nitrogen and phosphorus through the system
largely resemble the flow of carbon. Diagrams for all these cycles can be derived
from inspection of the revised carbon cycle (Figure 8-2) and from diagrams
provided by elsewhere in this report.
-141 -
-------
Chapter 8 • The Zooplankton Model
Zooplankton Kinetics
The Algal Production Equation
Effects of zooplankton on phytoplankton are computed in the predation term of
the algal production equation:
5 Bx [ 5
= Gx - BMx - WSx •
• Bx-PRx
(8-1)
5 t | 5 z
in which:
Bx = biomass of algal group x (g C m~3)
Gx = growth rate of algal group x (d"1)
BMx = basal metabolic rate of algal group x (d"1)
WSx = settling rate of algal group x (m d"1)
PRx = predation on algal group x (g C m"3 d"1)
z, t = vertical (m) and temporal (d) coordinates
Zooplankton Production Equation
Each zooplankton group is represented by an identical production equation. The
two groups are distinguished largely by parameter evaluation.
7
=[Gz-BMz-Mz] • Z-PRz
(8-2)
8 t
in which:
Z = zooplankton biomass (g C m"3)
Gz = growth rate of zooplankton group z (d"1)
BMz = basal metabolic rate of zooplankton group z (d"1)
Mz = mortality (d"1)
PRz = predation on zooplankton group z (g C m"3 d"1)
Growth Rate
Grazing is represented by a maximum ration formulation equivalent to the
familiar Monod formulation used to represent algal nutrient uptake. Grazing is not
equivalent to growth, however. Not all prey grazed is assimilated. Grazing also
requires effort which results in respiration above the basic metabolic rate. The
representation of growth rate which incorporates grazing, assimilation, and
respiratory loss is:
PAz
Gz = - • RMAXz • Ez • (1-RFz) • f(T) (8-3)
KHCz + PAz V ' ^ ' y '
in which:
PAz = prey available to zooplankton group z (g C m"3)
KHCz = prey density at which grazing is halved (g C m"3)
RMAXz = maximum ration of zooplankton group z
(g prey C g"1 zooplankton C d"1)
Ez = assimilation efficiency of zooplankton group z (0 < E < 1)
RFz = fraction of assimilated prey lost to respiration (0 < RF < 1)
f(T) = effect of temperature on grazing
-142-
-------
Chapter 8 • The Zooplankton Model
Available Prey
The computation of available prey incorporates two principles:
1) A constant, between zero and unity, determines the utilization of a prey group
by a predator,
2) A threshold density exists below which prey is not utilized.
The available portion of an algal group, for example, is determined:
BAxz = Max (fix - CTz, 0) (8-4)
in which:
BAxz = The portion of algal group x available to zooplankton group z (g C m~3)
CTz = The threshold concentration below which prey will not be utilized by
zooplankton group z (g C m"3)
Prey Available to Microzooplankton. Microzooplankton are conceived to graze
on dissolved organic carbon (a surrogate for bacteria), three algal groups, and
organic detritus. The total prey available to microzooplankton is:
PAsz = UDsz • DOCAsz + Z UBxsz • BAxsz
(8-5)
+ ULsz • LPOCAsz + URsz • RPOCAsz
in which:
PA = prey available to microzooplankton (g C m"3)
UDsz = utilization of dissolved organic carbon by microzooplankton
UBxsz = utilization of algal group x by microzooplankton
ULsz = utilization of labile particulate organic carbon by microzooplankton
URsz = utilization of refractory particulate organic carbon by microzooplankton
DOCAsz = dissolved organic carbon available to microzooplankton (g C m"3)
BAxsz = algal group x available to microzooplankton (g C m"3)
LPOCAsz = labile particulate organic carbon available to microzooplankton
(g C m-3)
RPOCAsz = refractory particulate organic carbon available to microzooplankton
(g C m-3)
Prey Available to Mesozooplankton. Mesozooplankton are conceived to graze on
three algal groups, microzooplankton, and organic detritus. The total prey available
to mesozooplankton is:
PAlz = Z UBxlz • BAxlz + USZlz • SZA
(8-6)
+ ULlz • LPOCAlz + URlz • RPOCAlz
in which:
SZ = microzooplankton biomass (g C m"3)
Definitions and notation for remaining terms largely follow those for microzoo-
plankton.
-143-
-------
Chapter 8 • The Zooplankton Model
Utilization of Each Prey Group. The fraction of the total ration removed from
each prey group is determined by the fraction of each utilizable prey group relative
to total utilizable prey.
Basal Metabolism
Basal metabolism of both zooplankton groups is represented as an exponentially
increasing function of temperature:
BMz = BMREFz • e
KTBz • (T-Trz)
(8-7)
in which:
BMREFz = metabolic rate of zooplankton group z at temperature Trz (d"1)
T = temperature (°C)
KTBz = effect of temperature on metabolism of zooplankton group z (°C"1)
Mortality
Both zooplankton groups are subject to mortality at low dissolved oxygen concen-
trations. The mortality term is zero until dissolved oxygen falls below a threshold
(Figure 8-3). Thereafter, mortality increases as dissolved oxygen decreases:
DOREF 1
Mz = MZEROz • 1 -
DOCRITzJ
(8-8)
in which:
Mz = mortality of zooplankton group z (d"1)
MZEROz = mortality at zero dissolved oxygen concentration (d"1)
DOCRITz = threshold below which dissolved-oxygen-induced mortality occurs
(g DO m-3)
DOREF = dissolved oxygen concentration when DO < DOCRIT, otherwise zero
(g DO m-3)
o
o:
HI
N
1
0.9 -
0.8 -
0.7
0.6 -
0.5 -
0.4 -
0.3 -
0.2 -
0.1
-0.1
DOCRIT = 2.0
34567
Dissolved Oxygen (mg/l)
10
Figure 8-3. Dissolved oxygen mortality function.
-144-
-------
Chapter 8 • The Zooplankton Model
Predation on Zooplankton
Mesozooplankton graze on microzooplankton. In addition, grazing on mesozoo-
plankton by organisms not represented in the model (jellyfish, finfish) is
considered. Representation of predation by organisms that are not modeled is a
classic problem. Our approach results in a quadratic formulation that closes the
mesozooplankton system.
Assume the predators clear a specific volume of water per unit biomass. Then
predation on zooplankton is the product of the clearance rate, prey abundance, and
predator abundance:
PRlz = F • LZ • HTL (8-9)
in which:
F = clearance rate (m3 g"1 predator C d"1)
LZ = mesozooplankton biomass (g C m"3)
HTL = predator biomass (g C m"3)
In the absence of detailed data regarding the temporal and spatial predator distri-
bution, a reasonable assumption is that predator biomass is proportional to prey
biomass, HTL = a • LZ. In that case, Equation 8-9 can be re-written:
Since neither a nor F are known precisely, the logical approach is to combine
their product into a single unknown, PHTlz, determined during the model calibra-
tion procedure. In addition to closure, the quadratic predation term adds desirable
stability to the potential oscillatory system represented by algae and zooplankton
alone.
Interfacing with the Eutrophication Model
The basic principles of the zooplankton model have been outlined above. Since
this chapter is the first documentation of the zooplankton model, additional detail
of the interfacing of the zooplankton component with the remaining model state
variables is presented here.
Effect of Zooplankton on Carbon
The rate of total carbon uptake by zooplankton is the product of the maximum
ration and biomass, modified by any existing food limitation:
Rz = - — - • RMAXz • f(T) (8-11)
KHCz + PAz
in which:
Rz = ration of zooplankton group z (g prey C g"1 zooplankton C d"1)
The rate at which carbon is recycled to the environment is determined by the
fraction of the ration not assimilated, by the mortality, and by predation from
higher trophic levels. The recycling of all zooplankton consumed by predators
-145-
-------
Chapter 8 • The Zooplankton Model
enforces mass conservation on the system. Mass conservation implies that
zooplankton carbon permanently removed from the system due to harvest of preda-
tors is negligibly small. The carbon recycle rate is:
CRRATEz = [(l-Ez) • Rz+Mz] • Z + PRz (8-12)
in which:
CRRATEz = rate of carbon recycling by zooplankton group z (g C m~3 d"1)
Once the total carbon uptake and recycle are defined, the uptake and release of
each carbonaceous state variable can be obtained. The uptake is determined by the
ratio of utilization of a single component to total available carbon. The distribution
of released carbon is determined by a set of empirical coefficients. The effect of
microzooplankton on dissolved organic carbon, for example, is:
UDsz • DOCAsz
O t _r ASZ
in which:
FDOCsz = fraction of carbon recycled to the dissolved organic pool by microzoo-
plankton (0 < FDOC < 1)
Effect of Zooplankton on Algae
Algae are a fraction of the total carbon uptake. No carbon is recycled to the algal
pool, however. The effect of mesozooplankton on algal group 2, for example, is:
8 B2 UB21z • B2Alz „,
- = -- • Klz • LZ /o iA\
8 t PAlz O14)
Effect of Zooplankton on Dissolved Oxygen
Zooplankton consume dissolved oxygen through the respiratory cost of assimi-
lating food and through basal metabolism:
-[Ez • RFz • Rz + BMz] • AOCR • Z (8-15)
8 t
in which:
Aocr = ratio of oxygen consumed to carbon metabolized (2.67 g DO g"1 C)
Effect of Zooplankton on Nitrogen
The computation of zooplankton effects on nutrients must account for differing
composition of zooplankton and prey. An additional complication is that the model
considers organic nitrogen and phosphorus to exist independently from organic
carbon. In reality, these exist as organic matter composed of carbon, nitrogen, and
phosphorus, among other elements. We consider that particulate organic nitrogen
and phosphorus are consumed in proportion to detrital carbon.
-146-
-------
Chapter 8 • The Zooplankton Model
First, evaluate the nitrogen-to-carbon ratio in the total prey consumed. For meso-
zooplankton, this is:
ANCPlz =
[ZANCx • UBxlz • BxAlz + ANCsz • USZlz • SZAlz] / PAlz
(8-16)
fULlz • LPON • LPOCAlz URlz • RPON • RPOCAlzl
+ + / PAlz
|_ LPOC RPOC J
in which:
ANCPlz = nitrogen-to-carbon ratio in prey of mesozooplankton (g N g"1 C)
ANCx = nitrogen-to-carbon ratio in algal group x (g N g"1 C)
ANCsz = nitrogen-to-carbon ratio in microzooplankton group x (g N g"1 C)
Additional terms follow previous definitions and notation.
All nitrogen consumed is recycled except for the amount assimilated. Additional
nitrogen is recycled through respiration, mortality, and predation on zooplankton.
For mesozooplankton, for example, the nitrogen recycle rate is:
NRRATElz =
[ANCPlz-ANClz • Elz • (1-RFlz)] • Rlz • LZ
(8-17)
+ [BMlz + Mlz] • ANClz • LZ + PRlz • ANClz
in which:
NRRATElz = nitrogen recycled by mesozooplankton (g N m~3 d"1)
ANClz = nitrogen-to-carbon ratio of mesozooplankton (g N g"1 C)
Zooplankton do not take up dissolved organic nitrogen or ammonium. Recycle
of these constituents is determined by the total recycle rate and empirical distribu-
tion coefficients. Ammonium recycle by mesozooplankton, for example, is:
^yy1 = NRRATElz • FNH4lz (8_18)
in which:
FNH4lz = fraction of nitrogen recycled to the ammonium pool by mesozooplankton
(0 < FNH4lz < 1)
For particulate organic nitrogen, the recycle rate is the difference between
detritus consumed and recycled. The effect of mesozooplankton on labile particu-
late organic nitrogen, for example, is:
5 LPON ULlz • LPOCAlz LPON „,
= - — • • Rlz • LZ fo 1 P.X
5 t PAlz LPOC (8-19)
+ NRRATElz • FLPONlz
-147-
-------
Chapter 8 • The Zooplankton Model
Effect of Zooplankton on Phosphorus
The effect of zooplankton on phosphorus is analogous to the effects described
for nitrogen.
Effect of Zooplankton on Silica
Zooplankton consume silica solely through the uptake of phytoplankton. All
silica consumed is recycled. Computation of the effect of zooplankton on silica
requires summation of silica consumed and allocation to the two external silica
pools. The effect of microzooplankton on dissolved silica, for example, is:
5 SA SASCx • UBxsz • BxAsz C7 _.
= • Rsz • SZ • FRSAsz (8-20)
5 t PAsz v '
in which:
ASCx = silica to carbon ratio of algal group x (g Si g"1 C)
FRSAsz = fraction of silica recycled to dissolved pool by microzooplankton
(0 < FRSAsz < 1)
Parameter Evaluation
Parameters in the zooplankton model (Tables 8-1, 8-2) were adapted from
published values, when available, and adjusted to provide improved model results.
Published values were not available for a number of empirical parameters which
were evaluated largely through a recursive calibration procedure. The evaluation
procedure for significant parameters is detailed below.
Composition
The classic Redfield ratios for zooplankton (Redfield et al. 1963) indicate the
following ratios: nitrogen-to-carbon = 0.19 g N g"1 C; phosphorus to carbon = 0.025
g P g"1 C. A summary by Parsons et al. (1984) indicates median values of 0.22 g N
g"1 C and 0.017 g P g"1 C for nitrogen-to-carbon and phosphorus-to-carbon. Model
composition for both zooplankton groups closely reflects these values.
Microzooplankton Ration
The microzooplankton community in the mainstem bay consists largely of
rotifers and tintinnids as well as juvenile forms of mesozooplankton and other
organisms (Brownlee and Jacobs 1987). A survey of specific grazing rates
(Table 8-3) indicates maximum rates in excess of 2 d"1 are observed.
The effect of temperature on grazing (Figure 8-4) was described with a function
of the form:
f(t) = e-KTgl' (T-Topt)2 when T < Topt
(8-21)
= e'KTg2' (T-Topt)2 when T > Topt
in which:
T = temperature (°C)
Topt = optimal temperature for grazing (°C)
KTgl = effect of temperature below optimal on grazing (°C~2)
KTg2 = effect of temperature above optimal on grazing (°C~2)
-148-
-------
Chapter 8 • The Zooplankton Model
Table 8-1
Parameters in Microzooplankton Model
Parameter
ANCsz
AOCRsz
APCsz
BMREFsz
CTsz
DOCRITsz
Esz
FDOCsz
FDONsz
FDOPsz
FLPOCsz
FLPONsz
FLPOPsz
FNH4sz
FPO4sz
FRPOCsz
FRPONsz
FRPOPsz
FRSAsz
KHCsz
KTBGsz
KTGsz!
Description
nitrogen to carbon ratio
ratio of oxygen consumed to carbon
metabolized
phosphorus to carbon ratio
basal metabolism at reference temperature
carbon threshold for grazing
concentration below which dissolved oxygen
mortality occurs
assimilation efficiency
fraction of carbon recycled to dissolved organic
pool
fraction of nitrogen recycled to dissolved
organic pool
fraction of phosphorus recycled to dissolved
organic pool
fraction of carbon recycled to labile particulate
organic pool
fraction of nitrogen recycled to labile particulate
organic pool
fraction of phosphorus recycled to labile
particulate organic pool
fraction of nitrogen recycled to dissolved
inorganic pool
fraction of phosphorus recycled to dissolved
inorganic pool
fraction of carbon recycled to refractory
particulate organic pool
fraction of nitrogen recycled to refractory
particulate organic pool
fraction of phosphorus recycled to refractory
particulate organic pool
fraction of silica recycled to dissolved pool
prey density at which grazing is halved
effect of temperature on basal metabolism
effect of sub-optimal temperature on grazing
Value
0.2
2.67
0.02
0.254
0.01
2.0
0.3
0.25
0.20
0.40
0.50
0.25
0.07
0.40
0.50
0.25
0.15
0.03
0.55
0.05
0.069
0.0035
Units
g N g1 C
g DO g'1 C
g P g 1 c
d1
g Cm3
g DO m 3
0
-------
Chapter 8 • The Zooplankton Model
Table 8-1 (continued)
Parameters in Microzooplankton Model
Parameter
KTGsz2
MZEROsz
RMAXsz
RFsz
TMsz
TRsz
UB1sz
UB2sz
UBSsz
UDsz
ULsz
URsz
Description
effect of super-optimal temperature on grazing
mortality at zero dissolved oxygen
concentration
maximum ration
fraction of assimilated prey lost to respiration
optimal temperature for grazing
reference temperature for basal metabolism
utilization of algal group 1
utilization of algal group 2
utilization of algal group 1
utilization of dissolved organic carbon
utilization of labile particulate organic carbon
utilization of refractory particulate organic
carbon
Value
0.025
4.0
2.25
0.5
25
20
0.0
1.0
1.0
0.1
0.1
0.1
Units
oQ-2
d1
g prey C g'1
zooplankton C d 1
0
-------
Chapter 8 • The Zooplankton Model
Table 8-2
Parameters in Mesozooplankton Model
Parameter
ANCIz
AOCRIz
APCIz
BMREFIz
CTIz
DOCRITIz
Elz
FDOCIz
FDONIz
FDOPIz
FLPOCIz
FLPONIz
FLPOPIz
FNH4lz
FP04lz
FRPOCIz
FRPONIz
FRPOPIz
FRSAIz
KHCIz
KTBGIz
KTGIz!
Description
nitrogen to carbon ratio
ratio of oxygen consumed to carbon metabolized
phosphorus to carbon ratio
basal metabolism at reference temperature
carbon threshold for grazing
concentration below which dissolved oxygen
mortality occurs
assimilation efficiency
fraction of carbon recycled to dissolved organic
pool
fraction of nitrogen recycled to dissolved organic
pool
fraction of phosphorus recycled to dissolved
organic pool
fraction of carbon recycled to labile particulate
organic pool
fraction of nitrogen recycled to labile particulate
organic pool
fraction of phosphorus recycled to labile
particulate organic pool
fraction of nitrogen recycled to dissolved
inorganic pool
fraction of phosphorus recycled to dissolved
inorganic pool
fraction of carbon recycled to refractory
particulate organic pool
fraction of nitrogen recycled to refractory
particulate organic pool
fraction of phosphorus recycled to refractory
particulate organic pool
fraction of silica recycled to dissolved pool
prey density at which grazing is halved
effect of temperature on basal metabolism
effect of sub-optimal temperature on grazing
Value
0.2
2.67
0.02
0.186
0.05
2.0
0.30
0.25
0.20
0.40
0.50
0.25
0.07
0.40
0.50
0.25
0.15
0.03
0.55
0.175
0.069
0.008
Units
g N g 1 C
g DO g 1 C
g P g 1 c
d1
g Cm3
g DO m 3
0< E < 1
0< FDOC< 1
0< FDON < 1
0
-------
Chapter 8 • The Zooplankton Model
Table 8-2 (continued)
Parameters in Mesozooplankton Model
KTglz2
KTPRIz
MZEROIz
PHTIz
RMAXIz
RFIz
TMIz
TPRIz
TRIz
UB1sz
UB2sz
UBSsz
ULsz
URsz
USZIz
effect of super-optimal temperature on grazing
effect of temperature on predation by higher
trophic levels
mortality at zero dissolved oxygen
concentration
predation by higher trophic levels
maximum ration
fraction of assimilated prey lost to respiration
optimal temperature for grazing
reference temperature for predation by higher
trophic levels
reference temperature for basal metabolism
utilization of algal group 1
utilization of algal group 2
utilization of algal group 1
utilization of labile particulate organic carbon
utilization of refractory particulate organic carbon
utilization of microzooplankton
0.03
0.069
4.0
2.0
1.75
0.07
25
20
20
0.0
1.0
1.0
0.1
0.1
1.0
oc-2
OQ-1
d1
m3 g 1 C d 1
g prey C g 1
zooplankton C d"1
0< RF< 1
°C
°C
°c
0
-------
0
10
25
30
35
15 20
Degrees C
Figure 8-4. Effect of temperature on microzooplankton grazing.
Mesozooplankton Grazing
The mesozooplankton community of the mainstem bay consists largely of three
species of copepods: Acartia tonsa, Eurytemora affinis, and Acartia hudsonica
(Brownlee and Jacobs 1987, White and Roman 1992b). Of these, an abundance of
data exists for Acartia tonsa. Consequently, parameterization of the mesozoo-
plankton component was based largely on this species.
Observations collected by White and Roman (1992a) provide an excellent basis
for evaluation of several model parameters. The observations are based on the body
weight of an adult copepod female. Ingestion, growth, and respiration were
converted to carbon specific rates (Table 8-4) using a mass 2.6 ug C individual"1
derived from White and Roman (1992b).
Table 8-4
Ingestion, Growth, and Respiration for an Adult Female Acartia
Tonsa (after White and Roman 1992a)
Temp (°C)
16.8
23.5
26.9
27.6
17.1
18.7
19.3
20
27.2
29.4
26
24.2
15.8
Growth
(cgc/
female/
d)
1.16
0.67
2.07
0.68
0.64
1.58
1.89
1.91
1.06
1.21
2.01
2.3
0.71
Phyto
Ingest
(cgc/
female/d)
0.69
2.26
3.35
6.86
0.28
0.99
0.01
9.2
4.71
2.4
4.69
5.82
0.68
MicroZ
Ingest
(cgc/
female/d)
1.15
0.36
Respi ratio
n (fig C/
female/d)
0.7
0.98
0.9
0.93
0.72
0.8
0.81
0.82
1.12
1.2
0.98
0.89
0.64
Growth
(1/d)
0.45
0.26
0.80
0.26
0.25
0.61
0.73
0.73
0.41
0.47
0.77
0.88
0.27
Phyto
Ingest
(1/d)
0.27
0.87
1.29
2.64
0.11
0.38
0.00
3.54
1.81
0.92
1.80
2.24
0.26
Respiration
(1/d)
0.27
0.38
0.35
0.36
0.28
0.31
0.31
0.32
0.43
0.46
0.38
0.34
0.25
-153-
-------
Chapter 8 • The Zooplankton Model
The observations indicate maximum specific ingestion rates of 2 to 3 d"1. Early
experiments with the model indicated the computed phytoplankton population was
not sustainable with ingestion greatly in excess of 2 d"1. Consequently a value of
1.75 d"1 was selected. The observations are forAcartia tonsa which predominates
from May through October (White and Roman 1992b). To account for grazing by
the cold-water species, a piecewise temperature function was devised that provided
grazing in winter and spring in excess of observations for Acartia (Figure 8-5).
3.5 -
^ 2.5
£ 1.5
0.5 -
Figure 8-5. Effect of temperature on mesozooplankton grazing.
Basal Metabolism
Ikeda (1985) presented a relationship for routine metabolism:
Y = a
M
(8-22)
in which:
Y = oxygen consumption (uL O2 individual"1 hour1)
M = body mass (mg C individual"1)
T = temperature (°C)
a, b, c, d = empirical constants
Ikeda linearized his equation through logarithmic transformation and evaluated
parameters via linear regression.
-154-
-------
Chapter 8 • The Zooplankton Model
Ikeda's relationship is, no doubt, convenient for employment in terms of conven-
tionally-measured quantities. The equation is a dimensional nightmare, however.
Several pages of conversion factors eventually yield the more tractable relationship:
BM =
0.072 • e'
0.06 . (T-20)
M
0.165
(8-23)
in which:
BM = carbon specific metabolic rate (d"1)
M = body mass (mg C individual"1)
An interesting insight is that Ikeda's original relationship indicates respiration
per individual increases with body mass while the converted relationship indicates
specific respiration decreases as a function of body mass (Figure 8-6).
Using body masses of 0.0012 mg DW and 0.008 mg DW for a rotifer and adult
Acartia (White and Roman 1992b) and the conversion 0.4 mg C mg"1 DW yields
the basal metabolic rates employed in the model (Tables 8-1, 8-2). The multiplier
0.06 in Equation 8-23 was increased to 0.069 in the model to provide a rounded,
classic Q10 of 2.
0.6 n
0.5
Acartia Hudsonica nauplii (0.2]
— Acartia Hudsonica adult (4.0)
Acartia Tonsa nauplii (0.1 2)
Acartia Tonsa adult (3.2)
Rotifer (0 48)
Figure 8-6. Effect of body size (|jg C) and temperature on specific, basal
metabolism. Size from White and Roman (1992b).
-155-
-------
Chapter 8 • The Zooplankton Model
Respiration Fraction and Efficiency
Data presented by White and Roman (1992a) indicates respiration accounts for
40 to 50% of total microzooplankton and phytoplankton consumed by Acartia. This
includes both basal metabolism and the respiratory cost of feeding. Basal metabo-
lism by Ikeda's formula (0.185 d"1) is roughly half the specific respiration derived
from White and Roman's data (Table 8-4). These proportions indicate the respira-
tory fraction is the remaining half of specific respiration or 20 to 25% of prey
consumed. We found a lower fraction, 7% of prey consumed, provided reasonable
results when employed in the mesozooplankton computations. A higher value, 50%
of prey consumed, was employed for microzooplankton to reflect the effect of body
size on respiration.
The same observations indicate growth and respiration account for 87 to 95% of
total ingestion. Roughly 10% of total ingestion is unaccounted for, indicating effi-
ciency is roughly 90%. Modeled efficiencies are only one third of that value,
however. While the observed efficiency seems high (White and Roman assume
80%), the model values are unrealistically low. Reduced efficiencies were initially
assigned to the model based on need. Efficiencies of 80 to 90% produced wide
oscillations in computed populations and resulted in collapse of the phytoplankton
population. Lower efficiencies resulted in damped oscillations and more stable
communities.
Later insights indicated why lower efficiencies are appropriate. In the model,
adult zooplankton instantaneously reproduce adult zooplankton. In reality, repro-
duction is in the form of eggs. Eggs hatch into juvenile forms that eventually
mature into adults. The true growth process installs temporal lags in the system and
reduces efficiency as eggs and juveniles are lost to predation and other processes.
While the temporal lags are difficult to introduce into the present model, the losses
that occur to eggs and juveniles are simulated though employment of relatively low
efficiencies.
Additional Parameters
Selection of the dissolved oxygen threshold for mortality was guided by the
observation that concentrations below 1 gm DO m~3 result in reduced survival of
copepod adults and inhibited hatching of Acartia tonsa eggs (Roman et al. 1993).
Sellner et al. (1993) noted neither rotifers nor copepods grazed heavily on
Microcystis. Consequently, utilization of Group 1 algae, which represent cyanobac-
teria in the tidal fresh Potomac River, was set to zero. White and Roman's (1992a)
observations indicate consumption of microzooplankton by mesozooplankton is
comparable to consumption of phytoplankton. Consequently utilization of micro-
zooplankton was set to unity.
Selection of half-saturation and threshold concentrations for mesozooplankton
was guided by interpretation of numerous studies (Table 8-5). Smaller concentra-
tions were employed for microzooplankton to reflect their smaller body size.
-156-
-------
Chapter 8 • The Zooplankton Model
Table 8-5
Mesozooplankton Threshold and
Half Saturation Concentrations
mg C /L
0.01
0.7
0.05
0.175
0.2
Comment
for growth
egg production
clearance rates
decrease
A. nauplii growth
ingestion
Source
Kiorboe and
Nielson (1994)
Durbin et al.
(1983)
Paffenhofer
and Stearns
(1988)
Berggreen et
al. (1988)
Kiorboe et al.
(1985)
Observations
Zooplankton biomass was monitored at 27 stations throughout the mainstem and
tributaries (Figure 8-7). Mesozooplankton were sampled monthly, using a 202 |^m
net towed obliquely from bottom to surface, and quantified as dry weight. For
comparison with the model, mesozooplankton biomass was converted from dry
weight to carbon using the ratio 0.4 gm C gnr1 DW. Microzooplankton were
sampled monthly, in the Maryland portion of the bay only, using a 44 um net. Five
samples from above the pycnocline were combined into "surface mixed layer"
composites. Five samples from below the pycnocline were combined into "bottom
mixed layer" composites. The investigators computed total biomass, as carbon,
from species counts using standard biomass for individuals of each species. The
data base, extending from mid-1984 through 1994, was provided by the Chesa-
peake Bay Program Office (Maryland) and by investigators at Old Dominion
University (Virginia).
Model Results
Harmonic Analysis of Observations
The zooplankton observations exhibit an enormous amount of variability that
confounds visual interpretation and model-data comparisons. To provide insight,
basic harmonic analysis was applied to the observations. First, observations were
grouped into four units:
Unit 1—Mainstem Bay (Segments CB2-CB7, EE3)
Unit 2—Tidal Fresh (Segments CB1, TF1-TF5)
Unit 3—Transition Zone (Segments RET 1-RETS)
Unit 4— Lower Estuary (Segments LE1-LE5)
-157-
-------
Chapter 8 • The Zooplankton Model
CB1.T
.1
Zooplankton
monitoring
stations
XEA659
\
RET
E3.1
meso only
micro + meso
TF5.-5
BG .4
CB7.3
B7,4
Figure 8-7. Zooplankton sampling stations.
For each unit, a relationship was proposed that included annual, semi-annual,
and seasonal cycles:
log10(Z) =
sin( " * +9)
365
(8-24)
. 27t • d . 271 • d
+ A3 • sin( + cp) + A4 • sin( + w)
182.5 91.25
in which:
Z = zooplankton (g C m~3)
Aj = mean of log-transformed concentration
A2, A3, A4 = amplitude of annual, semi-annual, and seasonal components
0,(j>,ip = phase lags
d = Julian day
Linear regression was used to evaluate the amplitudes and phase lags.
-158-
-------
Chapter 8 • The Zooplankton Model
The proposed relationship accounted for 25% to 30% of the mesozooplankton
log-variance in the mainstem bay and in the lower estuaries. Only 10% to 15% of
the variance could be explained in the tidal fresh and transition regions. The
analysis indicated that seasonal patterns exist in the estuarine areas but seasonally
accounts for only a small portion of the total variance. In the remaining regions,
virtually no seasonal pattern exists in the aggregate data. A reconstruction of the
annual time series, obtained by substituting the regression coefficients into
Equation 8-24, indicates a bi-modal pattern. Maximum biomass occurs in spring
while a secondary maximum occurs in summer (Figure 8-8). These peaks indicate
the spring bloom ofEurytemora and A. hudsonica and the summer population of
A. tonsa.
The harmonic analysis of microzooplankton strongly contrasted with mesozoo-
plankton. The proposed relationship accounted for 25% of the log-variance in the
tidal fresh and transition regions but less than 10% of the log-variance in the estu-
arine areas. Within the tidal fresh and transition regions, the reconstructed time
series shows a single annual peak that corresponds to the annual temperature cycle
(Figure 8-9). In the lower estuaries and mainstem bay, the microzooplankton time
series has no structure and is essentially "white noise."
0.03 -
0.02 —
CO
£
o
0.01 -
Mesozooplankton
Observed
Modeled
-Temperature
-25
— 20
o
60
120
~i '—i
180 240
Day
~i—'—r
300 360
Figure 8-8. Annual time series of observed mesozooplankton, computed
mesozooplankton, and temperature. Observations from CB2-CB7 and EE3.
Computations from surface at station CB5.2.
-159-
-------
Chapter 8 • The Zooplankton Model
Microzooplankton
Observed
-30
0.04 -
300
360
Figure 8-9. Annual time series of observed microzooplankton, computed micro-
zooplankton, and temperature. Observations from TF1-TF2 and RET1—RET2.
Computations from station TF1.7.
Model Time Series
Model results were compared to observed time series at all stations for which
data were available. Comparisons were consistent with sampling procedure.
Mesozooplankton were sampled in oblique vertical casts that provided a vertically-
integrated sample. Modeled mesozooplankton was vertically averaged for
comparison with the observations. Microzooplankton observations were composite
samples representing "above pycnocline" and "below pycnocline" concentrations.
Model results from individual cells were combined into above- and below-
pycnocline values. The observed pycnocline depth varies spatially and temporally
and is not always distinct. For comparison with the observations, the model pycno-
cline was assumed to occur at a consistent depth of 6.7 m (upper four model
layers). In the event the total depth was less than 6.7 m, the entire model water
column was combined into the above pycnocline value. A sampling of results is
presented here. Complete comparisons are provided in the CD-ROM that accompa-
nies this report.
The time series of observed and modeled zooplankton are difficult to interpret,
primarily due to the large variance inherent in the observations. Some basic obser-
vations are possible, however. The model does not capture the abundance of
microzooplankton in the tidal fresh regions. Peak concentrations in excess of
0.1 gm m"3 commonly occur in tidal fresh waters while the model peaks are
-160-
-------
Chapter 8 • The Zooplankton Model
roughly less than half the observed (Figure 8-10). Within the lower estuaries,
microzooplankton are less abundant and both model and observations show similar
peak abundance, roughly 0.05 gm m~3 (Figure 8-11). Within the mainstem of the
bay, observed peak microzooplankton abundance is roughly twice as high above
pycnocline than below pycnocline. The model demonstrates lesser difference
above and below pycnocline and is closer to the observed below-pycnocline
values (Figures 8-12, 8-13).
Final Calibration - SENS 1 36
Mierozooplankton TF1,7 Above Pycnocline
0 18
0,10
0.1*5
0.12
-« 0 !
A
DOB
0.06
00-'.
•II
-
*
'. 9
- » * *
• •
* .
* *
N; *,"<•>•', . ;• "'*• *
r "i1 it • • \ ;,» * • . ^* * ;* -
1 I 2 .1 - 1 n 7 ft 9 TO
Years
Figure 8-10. Time series of computed and observed microzooplankton at
Station TF1.7 in the tidal fresh Patuxent River.
Final Calibration - SENS 136
Microzooplankton LE2.2 Abow Pycnocline
007
006
OOb
004
am
ao-f
001
A
L
-
" i
«
i
-
• *
.*
\
\ ^
*
»
»
#
«
..;,,J
i
3
1
r 1 *
i p
*
* *
1 1 1
i
»>
*
1*1 *
i
•
La*j i , ' i H
°s /^ sy \ ] ! '*, |
, ••/ 1 ', , 1 ". .'.J 1 v ,
p
*
•
••;
»
»
V
A | 1
i *, ,.'i* t:i
ft 1
Years
Figure 8-11. Time series of computed and observed microzooplankton at
Station LE2.2 in the lower Potomac estuary.
-161 -
-------
Chapter 8 • The Zooplankton Model
0,0?
006
005
ifto o-s
E
0,03
0.02
0,01
Final Calibration - SENS 136
Microzooplankton CB3.3C Above Pycnocline
* *
*
r « » •
: . *
*
'. *
:»" I
. * ? i •
•i*'* *».'» I * i *
: » * I* r *'» "i * I
**i A? < ' '^. -'1'! /u • *% . ^' ' fr i ='C *J ' '
> 1 23-156789 10
Years
Figure 8-12. Time series of computed and observed microzooplank-
ton, above pycnocline, at Station CB3.3C in the mainstem
Chesapeake Bay.
Oi35
OOJ
0025
_,0.02
E
0.0 i
0005
Final
Calibration -
SENS 136
Microzooplankton CB3,3C Below Pyenocline
1
•
-
-
L
'i
i '
'*
0
9
m
i
i, W
^
,
|
I
i | V
' .*
I
' '
P-J.
J
J
I
*
I
•
4
» s aJ*
1 ' "
1 1 ^
I 1
J '.*!
*
I 1 m~ %>
V B* ', i j
2 3
w
i
J
I
*•
4
*
» *
i
! ' i
r
j • i ' . i. i
1 r
.
»f j
I,**8 i '- * i
v-i 4 i_.; i IT !
1 * y i * f 1^ " * !! * '• ! *^ ; !
J.I.-.TJ, {M.'V/i."'*!],
5 8 7 & 9 10
Years
Figure 8-13. Time series of computed and observed microzooplankton,
below pycnocline, at Station CB3.3C in the mainstem Chesapeake Bay.
-162-
-------
Chapter 8 • The Zooplankton Model
Inspection indicates enormous differences in mesozooplankton between the tidal
fresh Patuxent and other estuaries (Figures 8-14, 8-15). If the observations are
correct, the Patuxent is a unique system that requires unique treatment. For the
majority of tidal fresh and transition waters, computed mesozooplankton peaks,
roughly 0.06 gm m~3, are nearly double observed peaks (Figures 8-15, 8-16).
Within the lower estuaries, observed and computed maximum values are compa-
rable at roughly 0.05 gm m~3 (Figure 8-17). Agreement between computed and
observed mesozooplankton continues for the mainstem bay (Figure 8-18).
Final Calibration - SENS 136
OS
03
"ft
0.2
O.I
Mesozooplankton TFU Vertical Average
•
»
*
* *
t 1 2 3 ** -4 5 6 ^7 8 9 10
Years
Figure 8-14. Time series of computed and observed mesozooplankton at
Station TF1.7 in the tidal fresh Patuxent River.
0.08
006
0.05
—i
"goo*
E
a 0:1
uu;>
OUI
Final Calibration - SENS 1 36
Mesozooplankton TF5.S Verticil Average
I
; I i
i
- 1 ,
i t
, \
r *
(
P'- :-:/h
I i
J I
I I
i
1 1, i
• ; . » ;•• ,
Years
Figure 8-15. Time series of computed and observed mesozooplankton at
Station TF5.5 in the tidal fresh James River.
-163-
-------
Chapter 8 • The Zooplankton Model
Final Calibration - SENS 136
Mesozooplankton RET3,1 Vertical Average
0.07
am*
a ab
iJiDQA
E
oa:i
00?
UO!
-
:
-
i
-
-
0
1 ,
[
i • k'
4 1
i
1
» m
I
i
I
I
|t*
T;i'i \mf
2 .3
f
i
!
!t
»
•
is .
'4
§%N
^ *
ft -J >'i
*
i
!
i
J I 1
i N
i>;
1 i
1
f,' v ;•'
4 5 6*7 8 9 10
Years
Figure 8-16. Time series of computed and observed mesozooplankton at
Station RET3.1 in the Rappahannock River transition from fresh to salt water.
Final Calibration - SENS 1 36
Mesozooplankton LE5,§ Vertical Average
0,06
0.05
004
1
*?O.QJ
E
0.02
0.01
~
-
-
_
'1
*
*
*
I
\
41 i
1
i
i
\
i. j|j-
Jj/-
f
m
*
9
i
i '
i/.;-
z
f
m
.•!' '
i
, •
.
y
>
s 1
.'l^l'jf''
3*5
Years
i
j 1 ! -
i f f
It r! u
i
( j
i
' '
r j1 ,
f / !jfj •''''. :
A.^v^'. ../... ;
6 7 8 9 10
Figure 8-17. Time series of computed and observed mesozooplankton at
Station RET3.1 in the Rappahannock River transition from fresh to saltwater.
-164-
-------
0,07
mv,
Od!i
0,04
0,02
ddi
Chapter 8 • The Zooplankton Model
Final Calibration - SENS 136
Mesozooplankton CB4.3C Vertical Average
i,
I I'
.1 J
• 1
2 3
Yaarm
678
10
Figure 8-18. Time series of computed and observed mesozooplankton at
Station CB4.3C in the mainstem Chesapeake Bay.
Effect of Dissolved Oxygen
The model includes a mortality factor to account for the impact of anoxia on
zooplankton (Equation 8-8). The vertical tows used to sample mesozooplankton do
not allow distinction of surface and bottom populations in regions where anoxia
occurs. The model computes substantial differences between surface and bottom
populations in the presence of anoxia. At a mid-bay station, computed surface and
bottom populations of mesozooplankton are nearly identical from January thorough
April (Figure 8-19). In April the populations diverge and mesozooplankton at the
bottom are nearly extinct from June through August. In September, the bottom
population recovers and is equivalent to the surface by November. The mesozoo-
plankton time series at bottom is consistent with the time series of computed
dissolved oxygen (Figure 8-20). The divergence between surface and bottom popu-
lations begins when computed bottom dissolved oxygen plunges towards zero.
Mesozooplankton at the bottom remain at low levels until the autumn turnover
period and gradually recover to surface levels throughout the autumn.
Harmonic Analysis of Model
The time series of computed zooplankton was subjected to the same harmonic
analysis and reconstruction as the observations. Variance in the computed time
series was much less than in the observations and the time series relationship
accounted for much more variance in the computed time series (R2 = 0.58 to 0.83)
than in the observed time series.
The computed time series of both microzooplankton and mesozooplankton
exhibit semi-annual cycles with two peaks per year (Figures 8-8, 8-9). For
-165-
-------
Chapter 8 • The Zooplankton Model
567
Month
10 11 12
Figure 8-19. Time series of computed mesozooplankton, surface and bottom, at
Station CBS.2 in the mainstem Chesapeake Bay for 1993.
0 1
10 11 12
Figure 8-20. Time series of computed dissolved oxygen, surface and bottom, at
Station CBS.2 in the mainstem Chesapeake Bay for 1993.
-166-
-------
Chapter 8 • The Zooplankton Model
microzooplankton, the semi-annual cycle conflicts with the observed annual cycle
in tidal fresh water. For mesozooplankton, the semi-annual cycle is consistent with
observations but computed peaks lag observed by 90 days.
Analysis of model results from the Virginia Tributary Refinements phase (Cerco
and Meyers 2000) indicated the mid-year dip in microzooplankton was due to
grazing by mesozooplankton. The dip in mesozooplankton was attributed to anoxia
in the mainstem bay. An alternate explanation for the dips in both populations may
lie with the temperature dependence of the grazing and respiration functions.
Grazing in both populations peaks at roughly 25°C (Figures 8-4, 8-5) while metab-
olism increases indefinitely as a function of temperature (Equation 8-7).
Consequently, at high mid-summer temperatures, an excess of metabolism over
grazing may cause a decline in computed zooplankton.
Cumulative Distributions
Little or no correspondence exists between individual observations and model
computations. In view of the large, random variance in the observations, lack of
one-to-one correspondence is expected. An alternate, more informative view can be
obtained by comparing the cumulative distributions of observed and computed
zooplankton. The computed distributions are based on model computations corre-
sponding to sample locations and days. The computed distributions do not
represent the population of computations.
The lower portions of the distributions of computed and observed microzoo-
plankton agree well within the tidal fresh and transition regions (Figure 8-21). The
median observed value exceeds the model by 50% while the upper range of the
observed distribution greatly exceeds the model. Excess of observed over computed
peaks was noted in the time series as well (Figure 8-10). Within the lower estuaries
Final Calibration - SENS 1 36
10'
jlQ2
O
os
ID1
M icroz ooplankton
Tidal Fresh and Transition
-
-. j.,,--
:
i
! i
j J , : J 1_ i- _L -1 J J. L > _l J J 1
25 50 75 100
Percent Less than
Figure 8-21. Cumulative distribution of computed and
observed microzooplankton in tidal fresh and transition
regions (CB1, TF1, TF2, RET2).
-167-
-------
Chapter 8 • The Zooplankton Model
and mainstem bay, the observed microzooplankton exceed computed by 50% to
100% throughout most of the distributions (Figures 8-22, 8-23).
At any percentile of the distributions, computed mesozooplankton exceed
observed by 50% to 100% in the tidal fresh (Figure 8-24), transition (Figure 8-25),
and lower estuary (Figure 8-26) regions. Agreement between computed and
observed mesozooplankton is excellent throughout the distribution in the mainstem
bay (Figure 8-27).
Final Calibration - SENS 136
ID'"
_l
t>
O!
ID1
10*
M icroz ooplankton
Lower Estuaries
;
...--"""
•
i
•uL_j_^_i^_u_a_L___J_-_a_a___l__i_i_j__L_l
25 50 76 100
Percent Less Hi an
Final Calibration - SENS 1 36
ID1
_l
U
OB
10*
ID'
Microzooplankton
Mainstem Bay
:
• .---•""'"
-
'(
'-
26 50 75 1*9
Percent Less ti in
Figure 8-22. Cumulative distribution of computed and
observed microzooplankton in lower estuaries (
LE1, LE2).
Final Calibration - SENS 136
Mesozooplan kton
Tidal Fresh Regions
Model
Figure 8-23. Cumulative distribution of computed and
observed microzooplankton in the mainstem bay
(CB2-CB5).
u
OS
ID1
25 50 76
Percent Less tian
100
Final Calibration - SENS 1 36
10'
CD
10'
Mesozooplan kton
Transition Regions
Model
-
-/'
25 50 75 100
Percent Less ti an
Figure 8-24. Cumulative distribution of computed and
observed mesozooplankton in tidal fresh regions
(CB1.TF1-TF5).
Figure 8-25. Cumulative distribution of computed and
observed mesozooplankton in transition regions
(RET1-RET5).
-168-
-------
Chapters • The Zooplankton Model
ID'
MO-
Q5
TO'
Final Calibration - SENS 136
Mesozooplankton
Lower Estuaries
Model
50
Percent Less than
1*0
Final Calibration - SENS 1 36
,o-
10'*
0
E
10'
10*
Me sozooplan kton
Mainstem Bay
r
,,.-=~~=
X"
i X''
J_ -' ' -1 L L J. . • Lj JL J^ .. i .._1__L J. ...1
25 5B 75 i»
Percent Less than
Figure 8-26. Cumulative distribution of computed and
observed mesozooplankton in lower estuaries (LE1-LE5).
Figure 8-27. Cumulative distribution of computed and
observed mesozooplankton in the mainstem bay
(CB2-CB7).
Recommendations for Improvement
The present model represents zooplankton biomass within 50% to 100% of
observed values, as determined by comparisons of cumulative distributions.
Discrepancies between observations and model certainly indicate shortcomings in
the model. A large portion of the discrepancies, however, must be attributed to
observational methodology and to the variance inherent in the populations. A high
degree of accuracy is unlikely to be obtained but more realism and, potentially,
more accuracy can be added to the model. Suggestions for improvement range
from parameter re-evaluation through complete model reformulation.
A basic improvement in the model is to distinguish, through parameter values,
freshwater and estuarine zooplankton populations. Differentiation between indi-
vidual tributaries may also be appropriate. Differentiation of freshwater and
saltwater populations is readily justified and this approach will likely prove
successful.
The effects of temperature on grazing and respiration should be reviewed and the
cause of the mid-year population dips re-established. If the dips are due to the
temperature functions, re-evaluation of the parameters that describe temperature
effects on grazing is called for.
Another potential improvement is to add a second mesozooplankton group.
One group would represent the winter-spring population; the second group would
represent the summer population. No doubt, these two populations exist and can be
differentiated. The second group can be readily included and adds realism to the
model. The potential quantitative improvement in model computations cannot be
foreseen.
-169-
-------
Chapters • The Zooplankton Model
The final improvement is the most difficult. Add age structure to the mesozoo-
plankton model. As previously noted, adults in the present model instantaneously
reproduce adults. In the most realistic model, adults would produce eggs. Eggs
would hatch into larvae, mature into juveniles and, later, into adults. A model of
this sort offers the highest probability of success in representing the time series of
observed mesozooplankton. A multi-stage population model requires tremendous
resources in programming, calibration, and execution. The additional effort is likely
not worthwhile in the present multi-purpose model. The improvement is highly
recommended if the present model is employed in an application that focuses
largely on zooplankton.
References
Berggreen, U., Hansen, B. and Kiorboe, T. (1988). "Food size spectra, ingestion and growth
of the copepod Acartia tons a during development: implications for determination of
copepod production," Marine Biology, 99, 341-352.
Brownlee, D., and Jacobs, F. (1987). "Mesozooplankton and microzooplankton in the
Chesapeake Bay," Contaminant Problems and Management of Living Chesapeake Bay
Resources, Majumdar, S., Hall, L., and Austin, PL, eds. The Pennsylvania Academy of
Science, Philadelphia.
Cerco, C., Johnson, B., and Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model," ERDC TR-02-4, U.S. Army Engineer Research and Development Center, Vicks-
burg, MS.
Cerco, C., and Meyers, M. (2000). "Tributary refinements to the Chesapeake Bay Model,"
Journal of Environmental Engineering, 126(2), 164-174.
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report EL-94-4, U.S. Army Engineer Waterways Experiment Station,
Vicksburg, MS.
Cerco, C., and Cole, T. (1993). "Three-dimensional eutrophication model of Chesapeake
Bay," Journal of Environmental Engineering, 119(6), 1006- 10025.
Dagg, M. (1995). "Ingestion of phytoplankton by the micro- and mesozooplankton commu-
nities in a productive subtropical estuary," Journal of Plankton Research, 17(4), 845-857.
DiToro, D., and Matystik, W (1980). "Mathematical models of water quality in large lakes
part 1: Lake Huron and Saginaw Bay," EPA-600/3-80-056, Environmental Research Labo-
ratory, Office of Research and Development, US Environmental protection Agency, Duluth
MN.
DiToro, D., O'Connor, S., and Thomann, R. (1971). "A dynamic model of the phyto-
plankton population in the Sacramento-San Joaquin Delta," Nonequilibrium systems in
water chemistry, American Chemical Society, Washington, DC, 131-180.
Durbin, E., Durbin, Ann G., Smayda, Thomas J., and Verity, Peter G. (1983). "Food limita-
tion of production by adult Acartia tonsa in Naragansett Bay, Rhode Island," Limnology
and Oceanography, 28(6), 1199-1213.
Ikeda, T. (1985). "Metabolic rates of epipelagic marine zooplankton as a function of body
mass and temperature," Marine Biology 85, 1-11.
Kiorboe, T, and Neilson, T. (1994). "Regulation of zooplankton biomass and production in
a temperate, coastal ecosystem. 1. Copepods," Limnology and Oceanography, 39(3), 493-
507.
-170-
-------
Chapters • The Zooplankton Model
Paffenhofer, G., and Stearns, D. (1988). "Why is Acartia tonsa (Copepoda: Calanoida)
restricted to nearshore environments?," Marine Ecology Progress Series, 42, 33-38.
Parsons, T., Takahashi, M., and Hargrave, B. (1984). Biological Oceanographic Processes,
3rd ed., Pergamon Press, Oxford.
Redfield, A., Ketchum, B., and Richards, F. (1966). "The influence of organisms on the
composition of sea-water." The Sea Volume II. Interscience Publishers, New York, 26-48.
Roman, M., Gauzens, A., Rhinehart, W., and White, J. (1993). "Effects of low oxygen
waters on Chesapeake Bay zooplankton," Limnology and Oceanography 38(8), 1603-1614.
Sellner, K., Brownlee, D., Bundy, M., Brownlee, S., and Braun, K. (1993). "Zooplankton
grazing in a Potomac River cyanobacteria bloom," Estuaries 16(4), 859-872. White, J., and
Roman M. (1992a). "Egg production by the calanoid copepod Acartia tonsa in the mesoha-
line Chesapeake Bay: the importance of food resources and temperature," Marine Ecology
Progress Series 86, 239-249.
Thomann, R., and Fitzpatrick, J. (1982). "Calibration and verification of a mathematical
model of the eutrophication of the Potomac Estuary," HydroQual Inc., Mahwah, NJ.
White, J., and Roman M. (1992b). "Seasonal study of grazing by metazoan zooplankton in
the mesohaline Chesapeake Bay," Marine Ecology Progress Series 86, 251-261.
-171 -
-------
Analysis of Predation
and Respiration on
Primary Production
The N-P-Z Model
Start by considering a system consisting of a single nutrient form, one phyto-
plankton group, and one zooplankton group. This system is commonly referred to
as an N-P-Z (nutrient, phytoplankton, zooplankton) model. The system is well-
mixed, has a volume V, and is flushed at flow rate Q. For analytical purposes,
examine the system at steady state. Employing the formulations for growth, respira-
tion, and predation employed in the Chesapeake Bay Environmental Model
Package yields the following equations:
J_.G-Rbl.B-P.["-^-l.Z
Khb + N J [Khz + Bj
(9-1)
-Phtb • B2-— • B = 0
V
B - P-Rz • Z-Phtz • Z2-— • Z = 0 (9~2)
- Anb •
Khz + B | V
N
G-Rb
• B + Anb • Rz • Z + Anb • Phtb • B2
(9-3)
Anb • Phtz • Z2 + — • [No-N]=0
-172-
-------
Chapter 9 • Analysis of Predation and Respiration of Primary Production
in which:
N = nutrient concentration (g N m~3)
No = influent nutrient concentration (g N m~3)
B = algal biomass (g C m~3)
Z = zooplankton biomass (g C m~3)
G = algal growth rate (d"1)
P = zooplankton specific feeding rate (d"1)
Rb = algal respiration rate (d"1)
Rz = zooplankton respiration rate (d"1)
Khb = half-saturation constant for algal nutrient uptake (g N m~3)
Khz = half-saturation constant for algal uptake by zooplankton (g C m~3)
Phtb = constant representing predation on algae by higher trophic levels
(m3 g-1 C d-1)
Phtz = constant representing predation on zooplankton by higher trophic levels (m3
g-1 c d-1)
Anb = nutrient fraction contained in algal and zooplankton biomass (g N g"1 C)
Q = volumetric flow through system (m3 d"1)
V = system volume (m3)
Gross primary production in the N-P-Z system is computed:
GPP =
- - - ,n ..
Khb+N (9-4)
in which:
GPP = gross primary production (g C m"3 d"1)
No exact analytical solution for the complete set of equations exists. Moreover
the number of terms and constants confound examination of numerical solutions.
We will proceed by examining successive sets of reduced equations. Analytical
solution of these is often simplified by substitution of a mass-conservation equation
for one of the three original equations. At steady state, the mass of nutrient flowing
into the system must equal the mass flowing out:
Q • No = Q • (N + Anb • B + Anb • Z) (9-5)
Basic Parameter Set
The purpose of this analysis is to examine the effects of respiration and preda-
tion terms on system composition and production. The analysis employs the basic
parameter set listed in Table 9-1. These parameters are characteristic of model
parameters for a summer algal population grazed on by mesozooplankton. Nitrogen
is the nutrient of interest. System residence time is ten days.
-173-
-------
Chapter 9 • Analysis of Predation and Respiration of Primary Production
Table 9-1
Basic Parameter Set
Parameter
No
G
Rb
Rz
Khb
Khz
Anb
Q/V
Value
1 g m 3
2d1
0.1 d1
0.5 d 1
0.05 g m 3
0.1 g m3
0.1 g g 1
0.1 d
Phytoplankton with Respiration Only
The simplest system contains phytoplankton and a nutrient with no predators or
predation term. For this system, Equation 9-1 becomes:
N
Khb + N
• G-Rb
;-Q.B = o
V
(9-6)
Employing the mass-conservation equation, solutions are readily obtained:
No-N
B =
Anb
(9-7)
Khb
N = -
V
(9-8)
Since respiration is the independent variable, it is worthwhile to examine net
production as well as gross:
NPP
N
Khb + N
• G - Rb • B
(9-9)
in which:
NPP = net primary production (g C m"3 d"1)
Sensitivity to respiration is examined over the range 0.01 d"1 < Rb < 1.8 d"1
(larger values result in negative nutrient concentrations indicating no feasible
steady state solution). As expected, algal biomass declines as respiration increases
(Figure 9- 1). Nutrient concentration increases as algal biomass decreases.
-174-
-------
Chapter 9 • Analysis of Predation and Respiration of Primary Production
10
00
0.1
0.01
0.5 1
Respiration (1/d)
1.5
Figure 9-1. Algal biomass and nutrient concentration for a system with algal
respiration only.
Gross production increases as a function of respiration over much of the range,
then declines precipitously as respiration attains roughly 75% of the maximum
growth rate (Figure 9-2). Net production also exhibits a curvilinear relationship to
respiration with a peak at roughly 50% of the maximum growth rate. The potential
increase in production along with respiration is counter-intuitive. Since biomass
declines as respiration increases, a simultaneous decline in production is expected.
The key is to realize that production is the product of biomass and the nutrient-
E
O
3
Q_
14 -i
12 -
10
;
J
fi
~ o
fi
4
2
0
-GrossP
NetP
0.5 1
Respiration (1/day)
1.5
Figure 9-2. Gross and net primary production for a system with algal respiration
only.
-175-
-------
Chapter 9 • Analysis of Predation and Respiration of Primary Production
limited growth rate. As respiration increases over much of its range, nutrients are
released from algal biomass, the nutrient limitation to growth is relaxed, and the
product of growth and biomass increases.
Phytoplankton with Zooplankton
The system which includes zooplankton but no higher-order predation terms is
examined next. The equation set includes Equation 9-2 (with Phtz = 0),
Equation 9-3 (with Phtb = 0), and Equation 9-5. Algal biomass can be obtained
directly from Equation 9-2:
Khz •
B = -
Vj (9-10)
v
Zooplankton biomass is readily expressed from the mass-conservation relation-
ship:
No - N - Anb • B (9-11)
Anb
Nutrient concentration is expressed by a quadratic relationship:
a • N2 + b • N + c = 0 (9-12)
in which:
a = Rz + —
b = Anb • B • [G-Rb] + Rz • [-No + Anb • B + Khb]
o (9'14)
+ — • [-No + Khbl
V L J
c = Khb • -Anb • Rb • B-Rz • No + Rz • Anb • B-— • No (9-15)
The solution for N is obtained via the classic quadratic formula. Since the rela-
tionship is quadratic, two values of N result for any parameter set. One value is
consistently negative for the basic parameter set (Table 9-1) and is not considered.
Negative values of B and/or Z can result, as well, from specification of predation
rate, P. Sensitivity results are shown for the range 0.7 d"1 < P < 2 d"1 that results in
feasible concentrations for all parameters.
-176-
-------
Chapter 9 • Analysis of Predation and Respiration of Primary Production
Algal biomass decreases rapidly and approaches zero asymtotically as predation
rate is increased from the minimum feasible value (Figure 9-3). Surprisingly,
zooplankton biomass also diminishes as predation rate is increased. The steady
state solution indicates maximum biomass occurs at low predation rates in which
zooplankton slowly feed on a large algal standing stock.
Primary production responds in direct proportion to algal biomass (Figure 9-4).
Maximum production occurs a minimum predation rate. The steady-state solution
for the basic parameter set indicates little or no nutrient limitation and the small
increase in nutrient concentration at higher predation rates has little effect on
1.8
1.6
1.4
TO 1 H
m0.6
0.4
0.2
0
0.6 0.8 1 1.2 1.4
Predation (1/d)
1.6
1.8
Figure 9-3. Algal biomass, zooplankton biomass, and nutrient concentration for a
system with algae, zooplankton, and a nutrient.
1.2 -,
1 H
T?
1= 0.8 }
O
ra
c 0.6 H
•—
-o 0.4 \
o
CL
0.2
0.6 0.8 1 1.2 1.4
Predation (1/d)
1.6
1.8
Figure 9-4. Gross primary production for a system with algae, zooplankton,
and a nutrient.
-177-
-------
Chapter 9 • Analysis of Predation and Respiration of Primary Production
growth. Consequently production, the product of nutrient- limited growth and
biomass, responds primarily to changes in biomass.
Phytoplankton with Quadratic Predation
Predation by higher trophic levels (other than zooplankton) on phytoplankton is
represented by a quadratic term in the algal mass-balance equation. The quadratic
formulation is derived by assuming the biomass of predators is linearly propor-
tional to the biomass of prey. The proportionality constant and the specific feeding
rate are lumped into the parameter Phtb.
The effect of the quadratic predation term is examined by solving a simplified
algal mass-balance equation:
• G-Rb • B-Phtb • B2-— • B = 0 (9.16)
Khb + N V
and the mass-conservation equation.
The solution for algal biomass is quadratic with the following coefficients:
a = Anb • Phtb (9-17)
b = Anb • [-G + R]-Phtb • [Khb + No]+Anb • —
c = No • G - [Khb + No]
Q
v
(9-19)
For the basic parameter set, one solution for B results in consistently negative
nutrient concentrations. The second solution indicates an algal concentration that
declines as predation rate increases (Figure 9-5). Mass conservation dictates
nutrient concentration must increase as biomass decreases.
Despite the decrease in algal biomass, primary production increases as predation
is raised from minimum levels (Figure 9-6). The increase occurs because the
nutrient limit to production is relaxed as predation liberates nutrients from algal
biomass. Consequently, the product of nutrient-limited growth rate and algal
biomass increases despite the decrease in biomass. The increasing trend in
production continues until nutrients are no longer limiting. That is, until N/(Khb +
N) approaches unity. At that point, production declines in proportion to biomass.
-178-
-------
Chapter 9 • Analysis of Predation and Respiration of Primary Production
10
£
O)
z
CD"
0.1
0.01
0.2
0.4
0.6
0.8
Predation (m /g C/d)
Figure 9-5. Algal biomass and nutrient concentration for a system with quadratic
predation and no zooplankton.
14
12 n
0.2 0.4 0.6
Predation (m3/g C/d)
0.8
Figure 9-6. Gross primary production for a system with quadratic predation
and no zooplankton.
-179-
-------
Process-Based
Primary Production
Modeling in
Chesapeake Bay
Introduction
Primary production models may be placed in two categories. We call the first
category "observation based." Observation-based models employ observations of a
photoadaptive variable, chlorophyll, and perhaps other quantities, to quantify depth-
integrated production at the time and location of the observations. This class of
models has been extensively reviewed by Behrenfield and Falkowski (1997).
We call the second category "process based." Process-based models quantify
production as a function of computed algal biomass and computed processes that
may include growth (as affected by temperature, nutrients, and light), respiration,
predation, and transport. Parameters in process-based models are commonly
assigned so that computations match one or more observed quantities. The assign-
ment process is commonly denoted "calibration."
We have noticed that process-based models frequently encounter difficulty in
simultaneously matching observed biomass and production. Most commonly,
process-based models provide reasonable representations of biomass but fall short
in computing production. Less commonly, the models adequately compute produc-
tion but misrepresent biomass and/or growth-limiting nutrient concentration.
Doney et al. (1996) modeled phytoplankton, zooplankton, and nutrients in a sub-
tropical region near Bermuda. They noted the model showed "skill in capturing the
major features of the annual chlorophyll field." Computed primary production was
almost always less than observed, however. Model performance was weakest
"during late summer, when the model cannot supply enough nutrients to support
the high production observed."
McGillicuddy et al. (1995) showed computed and observed vertical profiles of
primary production in the North Atlantic. In ten of thirteen cases, primary produc-
tion was under-computed at the surface, where light is abundant and nutrients are
scarce. In deeper waters, where light is attenuated and nutrients are more abundant,
model-data comparisons were much improved.
Fasham et al. (1990) applied a model that included phytoplankton, zooplankton,
bacteria and nutrients to ocean Station "S" near Bermuda. They calibrated their
-180-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
model to match observed primary production. During the summer, computed nitrate
concentrations were much less than observed. Their simulations "greatly overesti-
mated the phytoplankton biomass in the summer and autumn."
Moll (1998) simulated chlorophyll, phosphate, and primary production in the
North Sea. Observed primary production was well-matched by the model. At
several stations, the computed phosphate agreed with observed while at other
stations computed exceeded observed by several hundred percent. At all stations,
the computed concentration exceeded the specified half-saturation concentration by
two to three times. Consequently, the effect of nutrient limitation on computed
production was minimal.
Based on the cited studies, process-based models have the following characteris-
tics. Observed production can be matched when light rather than nutrients appears
to limit production (the North Atlantic at depths greater than 20 to 30 m, the North
Sea as modeled). In regions where nutrients appear to be more limiting than light
(North Atlantic surface waters, the sub-tropical Atlantic), observed production
cannot be computed without compromising computations of algal biomass and
nutrients (Ocean Station "S").
The objective of the present study is to create a process-based model of Chesa-
peake Bay that simultaneously matches observations of phytoplankton biomass,
limiting nutrient concentration, light attenuation, and primary production.
Chesapeake Bay
Chesapeake Bay is an extensive estuarine system located on the east coast of the
United States of America (Figure 10-1). The mainstem of the bay extends 300 km
from the Susquehanna River, at the head, to the Atlantic Ocean, at the mouth. Mean
depth of the mainstem bay is 8 m although a deep trench with depths to 50 m runs
up the center. The Susquehanna provides the majority of freshwater flow (64 %)
and nutrient loading (Malone et al. 1988, Boynton et al. 1995.) Virtually all
remaining runoff and loads originate in several major western tributaries. The bay
and major tributaries are classic examples of partially-mixed estuaries by
Pritchard's (1967) classification.
The physics, chemistry, and biology of the bay have been extensively studied for
decades. Significant studies relative to primary production include nutrient
(Boynton et al. 1995) and carbon budgets (Kemp et al. 1997), examinations of
nutrient limitations (Fisher et al. 1999, Malone et al. 1996, Fisher et al. 1992), and
measures of primary production (Harding et al. 2002, Smith and Kemp 1995,
Harding et al. 1986).
These studies describe a system in which maximum phytoplankton biomass and
maximum production are out of phase. Peak biomass occurs during the spring
bloom while peak production occurs concurrent with the summer temperature
maximum (Malone et al. 1988, Smith and Kemp 1995, Malone et al. 1996, Harding
et al. 2002). During the spring bloom, phosphorus and silica tend to be the limiting
nutrients while nitrogen is the primary limiting nutrient in summer (Fisher et al.
1992, Malone et al. 1996, Fisher et al. 1999). Peak nitrogen loads delivered during
spring runoff are coupled to summer production through a nutrient trapping
mechanism. Nitrogen present is spring runoff is taken up during the phytoplankton
bloom, deposited in bottom sediments, and recycled to the water column by
temperature-induced diagenesis (Malone et al. 1988).
-181 -
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
20 0 20 40 Mlometers
Figure 10-1. Chesapeake Bay showing longitudinal axis and segments CB2,
CB4, CB7.
-182-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
The Chesapeake Bay Environmental Model Package
(CBEMP)
The CBEMP is a system of interactive models including a three-dimensional
hydrodynamic model (Johnson et al. 1993, Wang and Johnson 2000), a eutrophica-
tion model (Cerco and Cole 1993, Cerco and Meyers 2000), and a sediment
diagenesis model (DiToro 2001). The CBEMP was developed primarily to assist
management of eutrophication within Chesapeake Bay and tributaries. Since
management involves attaining water quality standards, emphasis in the model was
placed on representing observed properties of the system including nutrients,
chlorophyll, dissolved oxygen, and light attenuation.
By our classification, the CBEMP is a process-based model. Primary production
calculations in initial versions of the CBEMP were consistent with characteristics
of other process-based models (Cerco 2000). That is, computed production
matched or exceeded observed in the turbidity maximum region of the bay where
nutrients are abundant but light is limited. In the middle and lower portions of the
bay, where light attenuation is diminished but nutrients are sparse relative to the
turbidity maximum, computed production fell short of observed.
Management efforts in the Bay now require investigation of the effects of filter-
feeders, especially oysters and menhaden, in reducing eutrophication (Gilmore et
al. 2000) so that the amount of production available to these organisms must be
represented. At the same time, water quality standards are becoming more stringent
such that accurate computations of chlorophyll and nutrient concentrations cannot
be ignored. Consequently, the CBEMP must now represent both properties of the
system and production rates. The present chapter details formulation and parame-
terization of a model which meets these requirements.
Data Bases
Carbon Fixation
Phytoplankton photosynthetic rates were monitored 12 to 18 times per year at six
stations in the upper portion of the mainstem bay (Figure 10-2). Methodology was a
14C technique outlined by Strickland and Parsons (1972). Replicate surface-layer
composite samples were incubated within six hours of collection. Incubation was
conducted for an hour or more at in-situ temperature and constant, saturating, light
intensity (> 250 uE m~2 s"1). Carbon fixation rates were reported as jig C L"1 h"1.
The data base for the period 1985-1994 was obtained from the US Environ-
mental Protection Agency Chesapeake Bay Program monitoring data base
http://www.chesapeakebay.net/data/index.htm. For consistency with other measures
of production and model units, reported rates were converted to g C nr3 d"1. This
process involved units conversion only. Carbon fixation reported here as g C nr3 d"1
represents a short-term, nearly-instantaneous rate and should not be confused with
daily carbon fixed.
-183-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
40 Kilometers
Figure 10-2. Location of sampling stations for carbon fixation,
algal biomass, and carbon-to-chlorophyll ratio.
Net Primary Production
Net production was measured via a 14C method conducted on deck at in- situ
temperatures (Harding et al. 2002). Neutral density screening was used on a series
of replicate bottles to simulate in-situ light intensity from the surface to 1% light
level. Incubations were conducted over a 24-hour period. More than 160 observa-
tions, conducted throughout the mainstem bay (Figure 10-3), were provided for the
period 1987-1994. Integral daily primary production was reported as mg C m~2.
Gross Primary Production
Gross production was measured by an oxygen evolution method (Smith and
Kemp 1995). Planktonic oxygen production was measured on deck, at in-situ
temperature and five irradiance levels, over a period of four to five hours. Produc-
tion of oxygen at each light level was normalized to chlorophyll. From these
measures, a photosynthesis versus irradiance (P vs. I) curve was created. The P vs.
I curve was employed to integrate production over the depth of the euphotic zone
-184-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Net Production
Gross Production
20
20 40 Kilometers
Figure 10-3. Location of gross and net primary production stations.
based on vertical attenuation of light and chlorophyll distribution. Daytime net
production was estimated as the product of depth-integrated production (g O2 m~2
h"1) and number of daylight hours when incident light exceeded 100 |iE m~2 s"1.
Respiration rates were measured as decreases in oxygen concentration in bottles
incubated for four to five hours in the dark at in-situ temperatures. Volumetric rates
were multiplied by depth to obtain vertically integrated respiration rates. The
summed absolute values of net daytime production and daytime respiration resulted
in estimates of daily gross production.
Roughly 60 measures, collected in the interval 1989-1992, were collected at
three stations (Figure 10-3). Daily gross production was reported as g O2 m~2. For
consistency with other measures and with model units, gross production was
converted to carbon equivalents using the relationship:
O2 mol O2
mol C
m
32 g O2 1-4 mol O2 mol C
12 g C_g_C
m
(10-1)
-185-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
The photosynthetic quotient was based on empirical comparisons of production
measured by 14C methods and by oxygen evolution (Harding et al. 2002).
Measures of light-saturated assimilation ratio, reported as ug O2 ^ig"1 Chi h"1,
were produced in the course of the production measures. For consistency with other
measures and with model units, the assimilation ratio was converted to carbon
equivalents using the relationship:
Q2 mol Q2 mol C 12 g C 24 h _ g C
Chi h 32 g Q2 1.4 mol O2 mol C d g Chi d
(10-2)
Algal Biomass
Phytoplankton species counts and chlorophyll analyses were conducted at
monthly intervals on composite samples collected at six locations in the upper bay
(Figure 10-2). Depending on the depth of the water column, the composites repre-
sented "above pycnocline," "below pycnocline," or "water column." Observations
were provided by the principal investigator (Richard Lacouture, Smithsonian Envi-
ronmental Research Center). Prior to the data transfers, algal counts were converted
to carbon using published values of biomass per individual. Division of carbona-
ceous biomass by chlorophyll concentration provided estimates of phytoplankton
carbon-to-chlorophyll ratio.
Chlorophyll, Nutrients, and Associated Observations
The Chesapeake Bay Program conducted 12 to 18 sample cruises per year at 50
stations throughout the mainstem bay. Chlorophyll, nutrient concentrations, and
other observations for calibrating the model were obtained from an on-line data
base, http://www.chesapeakebay.net/data/index.htm. Methods and metadata are
available at the same site.
Model Formulation
The eutrophication portion of the CBEMP considers three algal groups and two
zooplankton groups. These groups interact with simulated cycles of carbon,
nitrogen, phosphorus, and silica. Mass-balance equations for 24 state variables are
solved on a three-dimensional computational grid of over 12,000 cells. Details
provided here focus on recent developments required to simulate primary produc-
tion. Additional descriptions of model formulation and applications may be found
elsewhere (Cerco and Cole 1993, Cerco and Meyers 2000).
Conservation of Mass Equation
The foundation of CE-QUAL-ICM is the solution to the three- dimensional
mass-conservation equation for a control volume. Control volumes correspond to
cells on the model grid. CE-QUAL-ICM solves, for each volume and for each state
variable, the equation:
s (10-3)
i <5 xk
-186-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
in which:
Vj = volume of jth control volume (m3)
Cj = concentration in j"1 control volume (g m"3)
t, x = temporal and spatial coordinates
n = number of flow faces attached to jth control volume
Qk = volumetric flow across flow face k of jth control volume (m3 s"1)
Ck = concentration in flow across face k (g m"3)
Ak = area of flow face k (m2)
Dk = diffusion coefficient at flow face k (m2 s"1)
Sj = external loads and kinetic sources and sinks in jth control volume (g s"1)
Solution to the mass-conservation equation is via the finite-difference method.
For notational simplicity, the transport terms are dropped in the succeeding
reporting of kinetics formulations.
Phytoplankton Kinetics
The model simulates three algal groups. The blue-green algal group represents
the Microcystis that are found only in the tidal freshwater portion of the Potomac
River, one of the western tributaries. The spring algal group comprises the diatoms
that dominate saline waters from January to May. The summer algal group repre-
sents the assemblage of flagellates, diatoms and other phytoplankton that dominate
the system from May to December. Each algal group is represented by identical
formulations. Differences between groups are determined by parameter
specifications.
Algal sources and sinks in the conservation equation include production, respira-
tion, predation, and settling. These are expressed:
— B = fo-R-Wa—} B-PR (10-4)
5t ( § z) ^ '
in which:
B = algal biomass, expressed as carbon (g C m"3)
G = growth (d-1)
R = respiration (d"1)
Wa = algal settling velocity (m d"1)
PR = predation (g C nr3 d'1)
Production
Production by phytoplankton is determined by the intensity of light, by the avail-
ability of nutrients, and by the ambient temperature.
Light
The influence of light on phytoplankton production is represented by a
chlorophyll-specific production equation (Jassby and Platt 1976):
(10-5)
-187-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
in which:
PB = production (g C g-1 Chi cH)
PBm = maximum photosynthetic rate (g C g"1 Chi d"1)
I = irradiance (E m~2 d"1)
Parameter Ik is defined as the irradiance at which the initial slope of the produc-
tion vs. irradiance relationship (Figure 10-4) intersects the value of PBm:
pBm
Ik =
in which:
a = initial slope of production vs. irradiance relationship (g C g"1 Chi (E m"2)"1)
Chlorophyll-specific production rate is readily converted to carbon-specific
growth rate, for use in Equation 10-4, through division by the carbon-to-
chlorophyll ratio:
-T.B
G =
CChl
in which:
CChl = carbon-to-chlorophyll ratio (g C g"1 chlorophyll a)
(10-7)
300
PBm(N.T) = 300
PBm(N.T) = 225
PBm(N,T) = 150
20
40
60
80
100
Irradiance (E m"2 d"1)
Figure 10-4. Production versus irradiance relationship showing regions of light
limitation, nutrient limitation, and mixed limitations.
-188-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Nutrients
Carbon, nitrogen, and phosphorus are the primary nutrients required for algal
growth. Diatoms require silica, as well. Inorganic carbon is usually available in
excess and is not considered in the model. The effects of the remaining nutrients on
growth are described by the formulation commonly referred to as "Monod kinetics"
(Monod 1949):
«N)=
KHd + D (10-8)
in which:
f(N) = nutrient limitation on algal production (0 < f(N) < 1)
D = concentration of dissolved nutrient (g m~3)
KHd = half-saturation constant for nutrient uptake (g m~3)
Temperature
Algal production increases as a function of temperature until an optimum
temperature or temperature range is reached. Above the optimum, production
declines until a temperature lethal to the organisms is attained. Numerous func-
tional representations of temperature effects are available. Inspection of growth
versus temperature curves indicates a function similar to a Gaussian probability
curve provides a good fit to observations:
f(T) = eKTgl ' (T-Topt)2 when T < Topt
(10-9)
in which:
T = temperature (°C)
Topt = optimal temperature for algal growth (°C)
KTgl = effect of temperature below Topt on growth (°C~2)
KTg2 = effect of temperature above Topt on growth (°C~2)
Constructing the Photosynthesis vs. Irradiance Curve
A production versus irradiance relationship is constructed for each model cell at
each time step. First, maximum photosynthetic rate under ambient temperature and
nutrient concentrations is determined:
pBm(N,T) = pBm • f(T) • - - - (10-10)
KHd + D
in which:
PBm(N,T) = Maximum photosynthetic rate at ambient temperature and nutrient
concentrations (g C g"1 Chi d"1)
The single most limiting nutrient is employed in determining the nutrient
limitation.
-189-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Next, parameter Ik is derived from Equation 10-6. Finally, the production vs.
irradiance relationship is constructed using PBm(N,T) and Ik. The resulting produc-
tion versus irradiance curve exhibits three regions (Figure 10-4). For I » Ik, the
value of the term I / (I2 + Ik2)1/2 approaches unity and temperature and nutrients are
the primary factors that influence production. For I « Ik, production is determined
solely by and irradiance I. In the region where the initial slope of the production
versus irradiance curve intercepts the line indicating production at optimal illumi-
nation, I Ik, production is determined by the combined effects of temperature,
nutrients, and light.
Respiration
Two forms of respiration are considered in the model: photo-respiration and
basal metabolism. Photo-respiration represents the release or leakage of carbon
fixed during the photosynthetic process (Goldsworthy 1970) and is represented as a
fixed fraction of production. Basal metabolism is the continuous energy expendi-
ture to maintain basic life processes. In the model, metabolism is considered
to be an exponentially increasing function of temperature. Total respiration is
represented:
R = Presp • G + BMr • eKTb'(TTr) (10-11)
in which:
Presp = photo-respiration (0 < Presp < 1)
BMr = metabolic rate at reference temperature Tr (d"1)
KTb = effect of temperature on metabolism (°C"1)
Tr = reference temperature for metabolism (°C)
Predation
The predation term includes the activity of zooplankton, filter-feeding benthos,
and other pelagic filter feeders including planktivorous fish. Formulation and
results of the zooplankton and benthos computations may be found in Cerco and
Meyers (2000). Predation by other planktivores is modeled by assuming predators
clear a specific volume of water per unit biomass:
PR = F»B»M (10-12)
in which:
F = filtration rate (m3 g"1 predator C day"1)
M = planktivore biomass (g C m"3)
Detailed specification of the spatial and temporal distribution of the predator
population is impossible. One approach is to assume predator biomass is propor-
tional to algal biomass, M = yB, in which case Equation 10-12 can be rewritten:
PR = y • F • B2
(10-13)
-190-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Since neither y nor F are known precisely, the logical approach is to combine
their product into a single unknown determined during the model calibration proce-
dure. Effect of temperature on predation is represented with the same formulation
as the effect of temperature on respiration. The final representation of predation,
including zooplankton, is:
PR = - .RMsz«SZ
KHsz + B
(10-14)
B
KHlz + B
RMlz«LZ + Phtl»B2
in which:
RMsz = microzooplankton maximum ration (g algal C g"1 zoo C d"1)
SZ = microzooplankton biomass (g C m~3)
KHsz = half saturation concentration for carbon uptake by microzooplankton
(g C m-3)
RMlz = mesozooplankton maximum ration (g algal C g"1 zoo C d"1)
LZ = mesozooplankton biomass (g C m"3)
KHlz = half saturation concentration for carbon uptake by mesozooplankton
(g C m-3)
Phtl = rate of predation by other planktivores (m3 g"1 C d"1)
Predation by filter-feeding benthos is represented as a loss term only in model cells
that intersect the bottom.
The Model Nitrogen Cycle
Nitrogen is first divided into available and unavailable fractions. Available refers
to employment in algal nutrition. Two available forms are considered: reduced and
oxidized nitrogen. Reduced nitrogen includes ammonium and urea. Nitrate and
nitrite comprise the oxidized nitrogen pool. Both reduced and oxidized nitrogen are
utilized to fulfill algal nutrient requirements. The primary reason for distinguishing
the two is that ammonium is oxidized by nitrifying bacteria into nitrite and, subse-
quently, nitrate. This oxidation can be a significant sink of oxygen in the water
column and sediments.
In the presence of multiple nitrogenous nutrients, ammonium and urea are
utilized first (McCarthy et al. 1977). In the model, the preference for reduced
nitrogen is expressed by an empirical function (Thomann and Fitzpatrick 1982)
with two limiting values (Figure 10-5). When oxidized nitrogen is absent, the pref-
erence for reduced nitrogen is unity. When reduced nitrogen is absent, the
preference is zero. In the presence of reduced and oxidized nitrogen, the preference
depends on the abundance of both forms relative to the half-saturation constant for
nitrogen uptake. When both forms are abundant, the preference for reduced
nitrogen approaches unity. When reduced nitrogen is scarce but oxidized nitrogen is
abundant, the preference decreases in magnitude and a significant fraction of algal
nitrogen requirement comes from oxidized nitrogen.
-191 -
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
1.1 -I
1 -
0.9 -
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Gross Primary Production
The relationship for depth-integrated daily gross primary production is:
CChl
• f(N) • f(T) • B • dz • dt
(10-16)
in which:
GPP = daily gross primary production (g C m~2)
D = daylength (1 day)
Zp/o = depth at which irradiance is 1% of surface irradiance (m)
I(t) = time-varying irradiance (E m"2 d"1)
Algal biomass, nutrient limitation, temperature effects, irradiance, and carbon-to-
chlorophyll ratio are all based on model computations at the sample location and at
1.5 m depth increments.
Net Primary Production
The relationship for depth-integrated daily net primary production is:
i%
JJ
0 0
f f
pm
f(N) • f(T)
(l-Presp)-BMr(T)
(10-17)
• B • dz • dt
in which:
NPP = daily net primary production (g C m"2)
Algal biomass, nutrient limitation, temperature effects, irradiance, and carbon-to-
chlorophyll ratio are all based on model computations at the sample location and at
1.5 m depth increments.
Parameter Evaluation
Parameters are reported here for the spring and summer algal groups which
inhabit the mainstem of Chesapeake Bay. Model parameters are based on published
values reported in a variety of units. For comparison with the model, reported
parameters are converted to model units of meter, gram, and day.
Algal Production Parameters
Maximum photosynthetic rates and their temperature dependence were based on
observations collected by Harding et al. (1986) and by Smith and Kemp (1995).
The observed rates were subject to in-situ nutrient limitations. Since the maximum
photosynthetic rates employed by the model are for nutrient-unlimited situations,
parameter values (Table 10-1) were specified at the upper range of reported rates
(Figure 10-6) based on the assumption that lower observations represented nutrient-
limited conditions.
-193-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Table 10-1
Phytoplankton Parameters
Symbol
Wa
PBm
a
Khn
Khp
Kris
Topt
KTg1
KTg2
Presp
BMr
Tr
KTb
Phtl
Definition
settling velocity
maximum
photosynthetic
rate
initial slope of
production vs.
irradiance
relationship
half-saturation
concentration for
nitrogen uptake
half-saturation
concentration for
phosphorus
uptake
half-saturation
concentration for
silica uptake
optimal
temperature for
algal growth
effect of
temperature
below Topt on
growth
effect of
temperature
above Topt on
growth
photo-respiration
metabolic rate at
reference
temperature
reference
temperature for
metabolism
effect of
temperature on
metabolism
rate of predation
by other
planktivores
Units
m d 1
g C g 1 Chi d 1
g C g 1 Chi
(E nr2)-1
g N m 3
g P m 3
g s m 3
°C
oc-2
oc-2
d"
°c
oc-1
m3 g 1 C d 1
Spring Group
0.1
300
8
0.025
0.0025
0.03
16
0.0018
0.006
0.25
0.01
20
0.0322
0.1 to 0.2
Summer Group
0.1
350
8
0.025
0.0025
25
0.0035
0
0.25
0.02
20
0.0322
0.5 to 2
continued
-194-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Table 10-1 (continued)
Phytoplankton Parameters
a
b
c
minimum
carbon-to-
chlorophyll ratio
incremental
carbon-to-
chlorophyll ratio
effect of light
attenuation on
carbon-to-
chlorophyll ratio
g C g 1 Chi
g C g 1 Chi
m
30
150
1.18
30
90
1.19
700
600
500 -I
Harding etal. 1986
Smith & Kemp 1995
Model (Spring)
Model (Summer)
Figure 10-6. Observed and modeled photosynthetic rates.
Respiration
Laws and Chalup (1990) report values of 0.03 d"1 for basal metabolism and 0.28
for photo-respiration. Other investigators have found that from 15% to 35% of
carbon fixed is lost to metabolism (Groeger and Kimmel 1989; Kiddon et al. 1995).
Model values (Table 10-1) were based on these reports.
Predation
Parameters for the zooplankton model have been reported elsewhere in this
report. The parameter that determines predation by other planktivores (Table 10-1)
was determined empirically to fit model results to observed algal biomass and
production. Substantially lower rates were used for the spring algal group than for
the summer algal group. The differential predation rates were based on the life
cycle of Atlantic menhaden (Durbin and Durbin 1975; Rippetoe 1993; Luo et al.
-195-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
2001), a species represented by the predation term. Menhaden enter the bay in
spring and initially feed almost entirely on zooplankton. As they mature, gill rakers
develop to filter phytoplankton. Menhaden leave the bay in late fall and spawn in
coastal ocean waters. Consequently, predation pressure exerted by menhaden on
phytoplankton occurs primary in summer and early autumn, concurrent with domi-
nance by the modeled summer algal group.
Algal Composition
Carbon-to-Chlorophyll Ratio. Carbon-to-chlorophyll ratios obtained from the
enumerations and chlorophyll analyses varied over an enormous range. The
extreme values were well outside the ratios reported for algal cultures which range
from 20:1 to 333:1 (Cloern et al. 1995). More then 70% of the enumerations indi-
cated carbon-to-chlorophyll ratios less than 75:1 and the most common values were
between 25:1 and 50:1 (Figure 10-7).
Carbon-to-Chlorophyl! Ratio (gig)
Figure 10-7. Histogram of observed carbon-to-chlorophyll ratios.
We noted two characteristics of the data set. First, carbon-to-chlorophyll ratio
was related to disk visibility (Figure 10-8) and, therefore, inversely related to light
attenuation. For example, at Station CB2.2, in the turbidity maximum, median
carbon-to-chlorophyll ratio was 33 while median disk visibility was 0.7 m. At
Station CB5.2, roughly 140 km below the turbidity maximum, median carbon-to-
chlorophyll ratio was 89 while median disk visibility was 1.8 m. Summary of
reports of algae grown in culture indicates carbon-to-chlorophyll ratio is propor-
tional to irradiance (Cloern et al. 1995). This proportionality was apparent in our
data set as an increase in carbon-to-chlorophyll ratio as a function of depth of light
penetration into the water column.
-196-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
.c
O
*7
O
100
90 -
80
70 -
60 -
50
40
30 -
20 -
10 -
0 -I
•
•
*
•
*
«
c ;. « *
»
i i s
0 0,5 1 1.5 2
Disk Visibility (m)
Figure 10-8. Algal carbon-to-chlorophyll ratio versus disk visibility. Data points
represent median values at 14 stations in the bay and adjacent tributaries.
A second characteristic, apparent at stations that experience a spring algal bloom,
was a higher carbon-to-chlorophyll ratio during the bloom months (February-May)
versus other months (Figure 10-9). The difference was too large to assign to differ-
ences in disk visibility in spring versus other months. We attributed the distinction in
carbon-to-chlorophyll ratio between bloom months and other months as a taxonomic
property of bloom species versus other species (Chan 1980).
We desired to represent carbon-to-chlorophyll ratio with a function that was
consistent with available observations and with quantities computed in the model.
We proceeded as follows. The carbon-to-chlorophyll ratios were paired with disk
visibility measures collected on the same day. These were obtained from the Chesa-
peake Bay Program monitoring data base. This produced a set of 2462 paired
observations. Disk visibility was converted to light attenuation via the relationship:
1.4
(10-18)
Ke = -
DV
in which:
Ke = diffuse light attenuation (m"1)
DV = disk visibility (m)
Regression was used to evaluate parameters (Table 10-1) in a proposed relationship:
CChl = a + b • e
c'Ke
in which:
a = minimum carbon-to-chlorophyll ratio (g C g"1 Chi)
b = incremental carbon-to-chlorophyll ratio at zero light attenuation (g C g"1 Chi)
c = effect of light attenuation on carbon-to-chlorophyll ratio (m)
-197-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
8 9 10 11 i
Figure 10-9. Carbon-to-chlorophyll ratio at station CBS.2 in the mainstem bay.
Data points are monthly median, 10th and 90th percentiles.
Observations characteristic of the spring bloom (February-May at mainstem and
lower estuary stations) were treated separately from the rest resulting in distinct
relationships for the model spring and summer algal groups (Table 10-1).
Nutrients. Algal nitrogen-to-carbon and phosphorus-to-carbon ratios were
initially set to Redfield composition (Redfield et al. 1966). Adjustments were
made to improve model fit to observed chlorophyll, nutrients, and production
(Table 10-1). Nitrogen composition remains close to Redfield composition while
the phosphorus stoichiometry of the spring group reflects the limiting nature of
this nutrient during the spring bloom. The silica fraction of the spring group
(Table 10-1) was specified within the range of reported values (D'Elia et al. 1983;
Parsons et al. 1984). Silica fraction of the summer group (Table 10-1) was specified
at a low value since diatoms comprise only a fraction of the summer species.
Nutrient Uptake. Reported half-saturation concentrations for algal nutrient
uptake in Chesapeake Bay are 0.001 to 0.008 g N m~3 for ammonium (Wheeler et
al. 1982) and 0.003 to 0.053 g P nr3 for phosphate (Taft et al. 1975). Both studies
comprised a limited number of observations conducted over a short term. The half-
saturation concentrations reported by Wheeler et al. (1982) are much less than the
range of values commonly reported for neritic phytoplankton. Means of values
summarized by Eppley et al. (1969) are in the range 0.028 to 0.052 g N nr3,
depending on substrate composition and plankton division. Model half-saturation
constants, 0.025 g N nr3 and 0.0025 g P nr3, correspond with the lower end of
reported ranges (Eppley et al. 1969; Taft et al. 1975).
The model half-saturation concentration for silica uptake by spring diatoms, 0.03
g Si nr3, is well within reported values, 0.02 to 0.082 g Si nr3, for oceanic diatoms
(Davis et al. 1978; Parsons et al. 1984). A lower half saturation concentration was
-198-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
specified for the summer group to emphasize nitrogen limitation during this season
(Fisher et al. 1999).
Algal Settling Rates. Reported algal settling rates typically range from 0.1 to 5
m d'1 (Bienfang et al.1982; Riebesell 1989; Waite et al. 1992). In part, this varia-
tion is a function of physical factors related to algal size, shape, and density
(Hutchinson 1967). The variability also reflects regulation of algal buoyancy as a
function of nutritional status (Bienfang et al. 1982; Richardson and Cullen 1995)
and light (Waite et al. 1992). The algal settling rate employed in the model repre-
sents the net effect of all factors that result in downward transport of
phytoplankton. The settling rate employed, 0.1 m d"1, was derived through tuning
of observed chlorophyll and nutrient concentrations and is at the lower end of
reported rates.
Results
The CBEMP was applied to the ten-year period 1985-1994. The model was run
continuously throughout the simulated period. Boundary conditions and loads were
updated on a daily or monthly basis. Integration time step was fifteen minutes.
Output from the model was produced at one-day increments.
Spatial Distribution of Production
Carbon Fixation. Computed and observed primary production parameters were
compared along an axis (Figure 10-1) extending from the mouth of the bay (km 0)
to the Susquehanna River (km 325).
Observed carbon fixation shows a
minimum at km 275, in the turbidity
maximum region of the bay, and a
maximum immediately downstream of
the turbidity maximum (Figure 10-10).
Nutrients in the upper bay are abun-
dant, and the measures were
conducted under uniform, saturating
light intensity. Consequently, the fixa-
tion minimum in the turbidity
maximum suggests a minimum in
algal biomass rather than a minimum
in biomass-specific production. The
model faithfully reproduces the
observed spatial trend as well as the
range of the observations. The model
indicates a factor-of-two difference
between maximum long-term mean
fixation in the upper bay (km 240) and
minimum fixation in the lower bay
(km 60).
Gross Production. Observed gross production exhibits a minimum in the
turbidity maximum (Figure 10-11). Observations in the mid- and lower bay exhibit
the same range of values although the mean in mid-bay is higher than in the lower
5
4
TJ
^E 3
o
O)
2
1
n
- '' J,
M ^ (J
. Model f i! N; -1 '
_ — — _ >_
• Observed /\ (
,
^
/ y/
^ '* * t
-. — / §
— \ -^v
_
:x__— ^- — "
;^,.... .„ , ,_ j_ _,. _
°0 '00
^^
"""
F H , ,_
200
J
!'[ '
A"/
" '
-^,
_,
\
1
300
Kilometers
Figure 10-10. Observed and modeled carbon
fixation (mean and range) along bay axis. Model
and observations represent a ten-year interval |
1985-1994.
-199-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
100
200
Kilometers
Figure 10-11. Observed and modeled gross
primary production (mean and range) along bay axis.
Observations from 1989-1992 are compared to
model results from 1985-1994.
bay. The model reproduces the
observed spatial trend although the
model mean and range exceed
observed in the turbidity maximum
station. A factor-of- two separates the
long-term mean maximum production
at km 230 from production upstream
in the turbidity maximum.
Net Production. At first glance, the
observed net production exhibits a
peculiar spatial distribution (Figure
10-12). The apparent spatial distribu-
tion is affected by non-uniform
temporal distribution of the sampling.
At some transects, samples represent
up to 9 months of the calendar year
while at other locations, samples
represent as few two months. Modeled
mean and range provide good agree-
ment with the frequently-sampled
transects. The spatial distribution of
computed long-term mean net produc-
tion matches the pattern of gross
production. The maximum occurs
around km 230 and is double produc-
tion further upstream in the turbidity
maximum.
Temporal Distribution
of Production
The temporal distribution of produc-
tion parameters was examined at three
locations—one in the upper bay, one in
the mid-bay, and one in the lower bay.
For carbon fixation, computations were
drawn from single model cells corre-
sponding to the locations of stations
CB2.2 (upper bay) and CB4.3C (mid-
bay). For gross production,
computations were drawn from single
model cells corresponding to the loca-
tions of stations NB (upper bay), MB
(mid-bay), and SB (lower bay). For net production, daily spatial averages over
regions CB2 (upper bay), CB4 (mid-bay), and CB7 (lower bay) were compared to
all observations within these regions.
Carbon Fixation. Inspection of the observed and computed carbon fixation
(Figures 10-13, 10-14) indicates a distinct seasonal pattern in which minimum
fixation, near zero, takes place in winter and maximum fixation takes place in
300
E
u
200
Kilometers
Figure 10-12. Observed and modeled net primary
production (mean and range) along bay axis.
Observations from 1987-1992 are compared to
model results from 1985-1994.
-200-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
B5 B6 B7 B6 SS 90 91 92 93 S4- 35
Figure 10-13. Observed and computed carbon fixation at Station CB2.2
(upper bay).
85 B6 B7 B&
SO 91 92 93 S4- 35
Figure 10-14. Observed and computed carbon fixation at Station CB4.3C
(mid-bay).
-201 -
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
mid-summer. Neither observations or model show trends or inter-annual variability.
Neither is there a clear distinction between the two stations.
Gross Production. Seasonal trends are difficult to discern in the observed gross
production observations (Figures 10-15 to 10-17). In the mid- and lower bay,
lowest observations were collected in winter but intervals of nearly-zero gross
production were observed in spring and autumn in the upper bay. Observations in
the mid- and lower bay suggest production was diminished in 1992 relative to the
two preceding years. Modeled production exhibits a regular seasonal pattern and, as
with carbon fixation, shows not consistent inter-annual differences. Computations
agree well with observations in the upper and mid-bay. In the lower bay, the
computations show best agreement with the lower observations collected in 1992.
Net Production. Net production observations exhibit characteristics similar to
the gross production observations. Seasonal patterns and inter-annual differences
are difficult to discern (Figures 10-18 to 10-20). The difficulty is at least partially
attributable to the uneven temporal distribution of the sampling. The observations
suggest net production in the mid- and lower bay was lower in 1987 and 1988 than
in subsequent years. Computed net production exhibits a distinct seasonal pattern.
Computations indicate peak production in mid-bay in 1989 and 1990 but inter-
annual variations are not strong.
E .
o
91
Years
Figure 10-15. Observed and computed gross primary production at station NB
(upper bay).
-202-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
91
Years
Figure 10-16. Observed and computed gross primary production at station MB
(mid-bay).
E H-
o
91
Years
Figure 10-17. Observed and computed gross primary production at station SB
(lower bay).
-203-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
67 SB B9 BO
91
Yean:
Si 94 35
Figure 10-18. Observed and computed net primary production in segment CB2
(upper bay).
B7 SB B9 90 91 92 93 94
Figure 10-19. Observed and computed net primary production in segment CB4
(mid-bay).
-204-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Figure 10-20. Observed and computed net primary production in segment CB7
(lower bay).
Seasonal Summaries
The sporadic nature of the observations makes confounds interpretation of
temporal characteristics. To clarify temporal behavior of the model and observa-
tions, seasonal summaries are provided for primary production and other key
parameters. Model results, as seasonal averages, are compared to observations from
three regions of the bay. The seasons are designated winter (December-February),
spring (March-May), summer (June-August), and Fall (September-November).
The northernmost region, designated CB2, is within the estuarine turbidity
maximum. In this region, nutrients are abundant and light is considered to be the
primary limit to algal production (Fisher et al. 1999). Both nutrient concentrations
and light attenuation diminish with distance down the bay axis. In the segment
designated CB4, limits to production are mixed. Limiting factors exhibit a seasonal
progression from light (winter), to phosphorus (spring), to nitrogen (summer)
(Fisher et al. 1999). In the lower bay segment CB7, nitrogen is the primary limit to
algal production (Fisher et al. 1999).
Primary Production. Observations of carbon fixation (Figure 10-21) indicate a
summer maximum, consistent with numerous studies (Malone et al. 1988, Smith
and Kemp 1995, Malone et al. 1996, Harding et al. 2002). The model shows the
same seasonal trends in carbon fixation as the observations. The model also shows
a spatial trend that corresponds to the trend in nutrient availability. Highest fixation
is in the upper bay, close to the Susquehanna River nutrient source. Lowest fixation
is at the station most distant from the Susquehanna. On a seasonal basis, differ-
ences in mean carbon fixed range from +227% (model greater than observed) to
-52% (model less than observed). Median difference is +15%.
-205-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Observed net production demonstrates a significant summer maximum and
substantial differences between the turbidity maximum and stations further down-
stream (Figure 10-22). Computations are consistent with the observed trends.
Computed net production peaks in summer at all locations and is highest in mid-
bay. In the turbidity maximum, production is diminished due to light limitation
while nutrient limitation limits net production in the lower bay. The magnitude of
the light limitation may be under-computed since mean net production exceeds
observed by 31% in the turbidity maximum. System-wide differences between
seasonal mean computed and observed net production range from 316% to -49%.
Median difference is -3%.
Observed trends and model behavior for gross production largely correspond to
net production. Observations indicate a significant summer maximum and substan-
tial differences between the turbidity maximum and stations further downstream
(Figure 10-23). Computed gross production peaks in summer at all locations and is
highest in mid-bay. System-wide differences between seasonal mean computed and
observed gross production range from 221% to -50%. Median difference is 3%.
„
_
E
~O K1~
s
""
o
Q-
in-
T T)--
T3
E rt~
IN -
0
CB2.2
^
L ¥
i
i
1
I
i
i
\
W S 5 F
CB4.3C
1
n-
T
ti
i
E
o ,-.
01
CB6.4
ll '
i
i
i
1
i
1
1
W S S F
t J
^
W S S F
Season
(N -i
T
^
i
£^-
o
01
CB2
1
1
'1
1
1
<•
W S S F
""
n -
^
V
£«-
a
CB4
I" 1
i
1
, 1
4
W S S F
*'
"7"
~D
CM
'E «-
a
EJ1 ,_ _
CB7
i. '
i i
i
1
1
1
r
W S S F
Season
Figure 10-21. Observed (circles) and Figure 10-22. Observed (circles) and
computed (triangles) seasonal (mean computed (triangles) seasonal (mean
and range) carbon fixation in the upper, and range) net primary production in the
mid-, and lower bay. upper, mid-, and lower bay.
-206-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Algal Biomass. Observations in the upper bay indicate the spring bloom barely
intrudes into the turbidity maximum while in mid-bay, mean algal carbon in spring
is double the biomass during other seasons (Figure 10-24). The model shows little
seasonal variability in mean algal biomass in the turbidity maximum but a substan-
tial spring bloom elsewhere. Observations also indicate biomass is lower in the
turbidity maximum than in mid-bay, a trend which is replicated in the model.
System-wide differences between seasonal mean computed and observed algal
biomass range from 15% to -48%. Median difference is -27%.
The spring bloom is not as evident in the observed surface chlorophyll concen-
trations as in the carbonaceous biomass (Figure 10-25). Consistent with the
biomass observations, observed chlorophyll within the turbidity maximum is lower
than further downstream. Computed chlorophyll concentrations are greatest in the
mid-bay and a spring bloom is evident in the mid and lower bay. System-wide
differences between seasonal mean computed and observed surface chlorophyll
range from 58% to -27%. Median difference is 15%.
Limiting Factors. Observed light attenuation shows a strong spatial trend but no
seasonal behavior (Figure 10-26). Attenuation in the turbidity maximum exceeds
attenuation downstream by a factor of three or more. The model replicates the
"1
T
i
£
ai
o-
NB
1
Ji
1 If
1
t '
i
i
W 5 S F
in -i
T "+ ~
T5
^ n -
£
ai
H
TJ
'E""
a "-
en
MB
1
|. ^
."
1 ^
W S S F
SB
ii
1. It 1
4
W S S F
Season
tn -
Tt -
ft
£«-
a
o-
M -
1
£
fM -
O -
H-
1
E
o •"
CT1
CB2.2
11 I'
T
w ;
i i
> S F
CB4.3C
iJ »'
i
W S S F
CB6.4
MM
W S S F
Season
Figure 10-23. Observed (circles) and Figure 10-24. Observed (circles) and
computed (triangles) seasonal (mean computed (triangles) seasonal (mean
and range) gross primary production in and range) algal biomass in the upper,
the upper, mid-, and lower bay.
mid-, and lower bay.
-207-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
O
CT
o-
CB2.2
•J
1 '
^
i
1
• H
W 5 S F
o
o _
01
13
0_
?-
O
K]
Q_
O-
CB4.3C
•1
1 '
i
1 *
I
4 f
W $ S F
CB6.4
1
'
• i
' 1'
' ll
W S S F
Season
"1
o _
E
in-
o-
n-
^_ r\~
£
0-
tN-
T
E •—
o-
CB2.2
i
-1
i
1
' It II
W S 5 F
CB4.3C
^
H
i
• 11 it
W S S F
C86,4
J
<
'
' i
• i" 1
t T
W S S F
Season
Figure 10-25. Observed (circles) and
computed (triangles) seasonal (mean
and range) surface chlorophyll in the
upper, mid-, and lower bay.
Figure 10-26. Observed (circles) and
computed (triangles) seasonal (mean
and range) light attenuation in the upper,
mid-, and lower bay.
observed pattern and magnitude. Differences of 31% to -27% exist between
observed and computed seasonal means. Median difference is -6%.
Nutrient limitations within the bay exhibit a pattern which is dependent on loca-
tion, season, and runoff (Fisher et al. 1999, Malone et al. 1996, Fisher et al. 1992).
During the spring bloom, phosphorus and, perhaps, silica play a role while nitrogen
is commonly considered to be the most limiting nutrient in the mainstem bay
during summer. We emphasize nitrogen here since summer is the period of peak
production (Harding et al 2002), and since summer production has been linked to
nitrogen loading (Malone et al. 1988).
The Chesapeake Bay Program provided an enormous data base of ammonium,
nitrate, and nitrite observations concurrent with the model application. A less
comprehensive data base of urea observations was also available (Lomas et al.
2002) but few observations coincided with the model application period. For
comparison with the model, we pooled ammonium, nitrate and nitrite observations
from the application period with mean observed surface urea concentrations
(0.0056 to 0.0076 gm m~3), averaged within spatial regions of the bay, from later
years. The sum is referred to here as "available nitrogen."
-208-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Observed available nitrogen (Figure
10-27) shows a strong trend down bay
away from the primary source at the
Susquehanna River. Temporally,
minimum concentrations occur in
summer, coincident with minimum
runoff and maximum algal uptake.
Computed available nitrogen conforms
with observed spatial and temporal
behavior. System-wide differences
between seasonal mean computed and
observed surface nitrogen range from
155% to -53%. Median difference is
13%. The greatest relative differences
occur in the lower bay and reflect the
difficulty in assigning boundary condi-
tions at the mouth of the bay.
Discussion
Specific Growth Rate
Three primary features distinguish
the present model from earlier versions
and from models employed in similar
applications. The first is the determina-
tion of specific growth rate as the
quotient of photosynthetic rate and
carbon-to-chlorophyll ratio (Equation
10-7). In the first version of the Chesa-
peake Bay model (Cerco and Cole
1993) and in several investigative
models (McGillicuddy et al. 1995; Fasham et al. 1990; Moll 1998; Doney et al.
1996), specific growth rate was specified directly.
Instantaneous, carbon-specific growth rates that result from the model relation-
ship are high relative to commonly accepted values. Employing the maximum
photosynthetic rate for the summer group, 350 g C g"1 Chi d"1 (14.6 jig C ug'1 Chi
hr1), and a typical carbon-to-chlorophyll ratio of 50 yields an instantaneous
specific growth rate of 7 d"1. By contrast, the classic work of Eppley (1972) indi-
cates maximum specific growth rate is roughly 2 d"1 at 20 °C. Alternate
investigations of primary production (McGillicuddy et al. 1995; Fasham et al.
1990; Moll 1998; Doney et al. 1996) employ growth rates of 0.66 to 2.9 d'1.
Care must be employed in the definition and comparison of growth rates. The
instantaneous growth rate is not the daily average growth rate. The daily average
growth rate can be obtained:
tD -
'E"
o-
«-
r?
£h
o -
n-
KS
'E
z "~"
CT
O -
CB2.2
-4 ,
•f It Ij
iiii
W S S F
CB4.3C
l| ^
•1 Ji (l
W S S F
CB5.4
I* ll x. if
| «,_! ^_y L^
W S S F
Season
Figure 10-27. Observed (circles) and
computed (triangles) seasonal (mean
and range) available nitrogen in the
upper, mid-, and lower bay.
G--L .
94 J
pBm
CChl
dt
Ik
(10-10)
-209-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
in which
Gbar = daily average growth rate (d"1)
I(t) = instantaneous irradiance (E m~2 hr1)
Employing a total daily irradiance of 60 E m~2, a 12-hour daylight period, and a
sinusoidal time series of irradiance yields a daily average specific growth rate of 3
d"1 for the summer group. This rate is still not realized within the water column,
however. At a depth of 1 m (mid-point of model surface layer), with Ke = 1 nr1,
daily average specific growth rate is reduced to 2.24 d"1. This growth rate is within
conventional ranges and is subject to further reduction as a function of nutrient
limitation and sub-optimal temperature. This analysis indicates that conventional
growth rates result in-situ when maximum photosynthetic rate is influenced by
light attenuation and averaged over a day.
Models that initially specify maximum specific growth rate 2d-1 and attenuate
this rate due to the influence of daylength and other factors likely employ daily
average growth that is much less than actual in-situ growth. We examined the effect
of our growth rate formulation on computed production. Maximum instantaneous
specific growth rate was forced to 2 d"1 by fixing carbon-to-chlorophyll ratio at 50
g C g"1 Chi and adjusting maximum photosynthetic rate to give a quotient 2 d"1 at
optimal temperature. Baywide annual production was reduced by 15% when lower,
conventional growth rates were imposed.
Urea
A second distinctive feature of the present model is the use of urea (CO(NH2)2)
as an algal nutrient. The role of urea as a nitrogenous nutrient in Chesapeake Bay
has been recognized for over 25 years. McCarthy et al. (1977) found that urea
comprised an average of 20% of algal nitrogen uptake. More recent work (Lomas
and Glibert 1999) indicates, at times, urea comprises as much as 40% of algal
nitrogen uptake in the bay.
Despite the importance of urea as a nutrient, this compound is omitted from
popular management models (Cerco and Cole 1993) as well as primary production
models (McGillicuddy et al. 1995; Fasham et al. 1990; Moll 1998; Doney et al.
1996). One reason for the omission may be the complication of adding an addi-
tional state variable to the models. The approach taken here of combining
ammonium and urea into a single state variable is a reasonable compromise. Both
are reduced nitrogen forms and share common origins including zooplankton
excretion (Miller and Glibert 1998) and benthic sediment regeneration (Lomas et al
2002). Ammonium and urea are both "preferred" over nitrate as algal nutrients
(McCarthy et al. 1977).
Urea observations within the bay did not permit a one-to-one calibration of the
model to contemporaneous observations. Our approach to incorporating urea was to
adjust model parameters such that model computations of reduced nitrogen agreed
with contemporaneous ammonium observations plus regional, temporal average
urea concentrations observed in later years.
To test the effect of incorporating urea in the model, we ran the model with urea
removed from the reduced nitrogen pool. The amount removed, 0.0082 g N m"3,
was determined as the long-term mean concentration in bay surface waters. Bay-
wide annual production was reduced by less than 2% when urea was removed. This
-210-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
result was surprising in view of the cited importance of urea in algal nutrition. We
attribute this result largely to the nature of the sensitivity run. Urea concentrations
vary with season, location, hydrology, and other factors (Lomas et al. 2002).
Removal of a constant amount does not represent the true role of urea at various
times and locations. Available observations did not allow a more detailed sensitivity
analysis, however.
We believe urea is important as a nutrient and that the substance should not be
ignored in eutrophication modeling. Our present knowledge base, however, does
not allow a detailed representation of urea. Extension of the application period of
the present model into a more data-rich period may allow more detailed modeling.
Additional monitoring of concentrations and loads is also required before urea can
be represented at a level of realism concurrent with other model state variables.
Predation by Higher Trophic Levels
A third distinctive feature of the model is the use of a quadratic term (Equation
10-14) to represent predation by higher trophic levels not included in the model.
Doney et al. (1996) incorporated a similar quadratic term that represented aggrega-
tion. A key difference between their formulation and the present model is that
nutrients incorporated in aggregated algae sink from the photic zone. The quadratic
predation term recycles algal nutrients at the location where predation occurs.
Computed primary production is sensitive to the specification of the predation
term, Phtl (Figure 10-28). Maximum production occurs at Phtl 0.3 although algal
biomass declines monotonically from lower to high values of Phtl. Consequently,
00
Dissolved Inorganic F'
Phtl= Phtl- Phtl- Phtl= Phtl -
0 01 0 1 0 3 t) 5 I 0
Available Nitrogen
0.10 -
0.08 -
0.06
0.04 •
0.02 •
0.00
Figure 10-28. Sensitivity of primary production to quadratic predation term. Panels
indicate summer average gross production, surface algal biomass, surface dis-
solved inorganic phosphorus, and surface available nitrogen at the mid-bay (MB)
location.
-211 -
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
maximum production occurs at moderate algal biomass rather than at low or high
extremes. This behavior is explained by examining concentrations of dissolved
inorganic phosphorus and available nitrogen. At the lowest predation rate exam-
ined, algal biomass was maximized but dissolved inorganic phosphorus was largely
depleted. Due to the stringent nutrient limitation, growth was minimal. Production,
which is the product of biomass and growth (Equation 10-16), was low. As preda-
tion increased, biomass decreased but the nutrient limitation was relaxed as
phosphorus bound up in algal biomass was released, through predation, to the
water column. At moderate predation levels, the limiting nutrient in mid-bay tended
towards nitrogen rather than phosphorus. Although biomass was diminished, relax-
ation of the stringent nutrient limitation allow higher growth. Production was
maximized. As predation approached the maximum rate examined, nutrient limita-
tions to growth were eliminated but algal biomass was diminished so that the
production was less than at lower predation rates.
Predation by non-specific higher trophic levels is the dominant predation term in
the model, consuming 60% to 75% of annual net production (Table 10-2). As origi-
nally conceived, the higher-trophic-level predation term was intended to simulate
activity by menhaden. Attribution of this level of predation to menhaden seems
unreasonable. A bioenergetics model (Luo et al. 2001) that estimated the carrying
capacity of Chesapeake Bay for menhaden assumed menhaden consume 10% of
production. A second argument against menhaden is that our predation term imme-
diately recycles algal biomass as nutrients. The assumption that predator biomass is
minimal and that turnover is rapid is implicit in our formulation.
Our model provides good representation of the distribution of mesozooplankton
biomass (Figure 10-29). Observations collected near our mid-bay station indicated
predation by mesozooplankton consumed 12% to 103% of net production in four
samples from March to October (White and Roman 1992). Our model average over
101
10-
o
ID
10*
Percent Less than
100
Figure 10-29. Cumulative distribution of observed and
modeled mesozooplankton biomass
-212-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
the same months indicates mesozooplankton consume 19% of net production.
Interestingly, both the observations and our model indicate consumption can exceed
net production, suggesting an import of phytoplankton to the mid-bay station.
While the observations do not permit an exact determination of model performance
versus actual mesozooplankton grazing, comparison to available measures of
biomass and grazing indicates performance is reasonable, at least.
Measures of microzooplankton grazing using the dilution technique (Gallegos
1989) indicate microzoplankton potentially consume 45% to 105% of production in
the Rhode River, a Chesapeake Bay tributary. The same methodology applied to
Chesapeake Bay (McManus and Ederington-Cantrell 1992) indicated microzoo-
plankton grazing rates averaged 57% of phytoplankton growth rates. Our annual
estimates of microzooplankton consumption, 6% to 16% of net production, are low
by comparison to the measures. The computed distribution of microzooplankton
biomass (Figure 10-30) is also low relative to observations. At the median,
computed microzooplankton biomass is about half observed. Improvements to our
microzooplankton model to double the biomass might raise computed consumption
to 12 to 32% of net production.
10''
1
o
10*
10"
M«M
missing
SO
Percent Less thin
1W
Figure 10-30. Cumulative distribution of observed and
modeled microzooplankton biomass
Doubling the computed microzooplankton consumption still leaves a substantial
fraction of non-specific predation. We attribute the remaining predation to
heterotrophs not quantified in the microzooplankton observations. Microzoo-
plankton sampling in the bay was conducted with a 44 jim net. Collection of
microzooplankton with nets is no longer recommended since soft-bodied
zooplankton may be destroyed by plankton mesh (Harris et al. 2000).
Heterotrophic flagellates of size less than 44 um will pass through the net
entirely. In future applications, we may wish to re-parameterize the microzoo-
plankton algorithm to increase computed biomass and compare results with
improved microzooplankton observations presently being collected.
-213-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Annual Integrated Production and Nutrient Loads
The nature of the model facilitates computation of annual integrated production.
The model computes daily production for each of the 1215 cells that represent the
surface of the bay. These are readily combined into bay-wide daily areal-average
production and into areal-average annual production. Our computations indicate the
ten-year mean bay-wide areal-average daily gross production is 1.02 g C m~2 d ~l.
Corresponding net production is 0.70 g C m~2 d ~l.
Areal-average production in any year shows no relationship to nutrient loading
from the Susquehanna River, the major source to the mainstem bay (Figure 10-31).
An obvious explanation is the limitation of the calendar in defining years. A late-
November storm will influence annual loads but can have limited influence on
production in the same year. However, attempts by us to relate production to
nutrient loads in shorter, more significant periods (i.e. spring months) produced no
apparent relationship.
Absence of tight coupling between loads and production is partially attributable
to residence time of nutrients in bottom sediments. Numerical experiments with the
model (Cerco 1995) indicate two years are required for sediment nutrient release to
largely adjust to load reductions. Some "memory" of loads persists for ten years or
more. Parameters in the sediment model (DiToro 2001) are based on decomposition
experiments (Westrich and Berner 1984) that showed 65% of organic matter
remained after two years incubation. As a consequence of sediment residence and
recycling, production in any year is a function of contemporary loading as well as
loading in previous years.
A second factor in damping response of production to loads is nutrient exchange
with the continental shelf. Transport across the mouth of the bay results in net
-50
D Total N
• Total P
nNetP
Figure 10-31. Annual nutrient loads from the Susquehanna River and bay-wide
annual net production relative to ten-year mean.
-214-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
export of nitrogen and net import of phosphorus (Boynton et al. 1995). The net
fluxes represent differences in nutrients transported in both directions across the
bay mouth as a result of tides, meteorological forcing, and other factors. Mass-
balance modeling at the mouth of the bay (Cerco 1995) indicates the unidirectional
transport of new nitrogen into the bay, imported from the shelf reservoir, is equiva-
lent to 38% of system loads from other sources. New phosphorus is by far the
dominant source, equivalent to 81% of system loads from other sources. Analyses
that relate production solely to loads from upland sources omit a large fraction of
the total loads to the system.
Our computations of annual integrated production over the ten-year simulation
period range from 0.66 to 0.95 g C m~2 d"1 net and 0.95 to 1.07 g C m~2 d"1 gross.
Our computations represent 50% to 85% of net production and 60% to 100% of
gross production estimated from the same data base by alternate means (Harding et
al. 2001). One immediate explanation for the discrepancy lies in the averaging
periods. Our model represents 1985-1994 while the entire data base extends from
1982 to 1998. We attribute the differences, however, not to the averaging period but
to the averaging methods.
The observations were collected at discrete locations and times. Arbitrary inter-
polations are required to average the discrete observations into system-wide annual
averages. The model uses a different technique in which production system-wide is
continuously computed as a function of nutrient concentrations, light attenuation,
temperature, and other factors. Inevitably, the employment of the model to compute
production in locations and time for which observations are missing will result in
different estimates than using interpolation or other methods to fill in gaps in the
data base.
In Conclusion
Computation of concentrations and processes within the CBEMP is a demanding
task that requires accurate computations and linkages between models as well as an
extensive data base. Daily nutrient and solids loads are generated and distributed
along the bay perimeter by the CBP Watershed Model, a modified version of HSPF
(Donigian et al. 1991). Nutrients and solids are transported throughout the bay
from their input locations based on computations from the hydrodynamic model
(Johnson et al. 1993, Wang and Johnson, 2000). As they are transported, nutrients
are cycled within the water column and between the water column and sediments
(DiToro 2001). We believe we have constructed a model that accurately represents
primary production rates (Table 10-3) while maintaining reasonable agreement with
observed properties including algal biomass, chlorophyll, and light and nutrient
limitation. Discrepancies between observed and computed properties are attribut-
able to inevitable limitations in our ability to quantify loads, boundary conditions,
and forcing functions within multiple models rather than to shortcomings in model
formulation.
-215-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Table 10-3
Statistical Summary of Modeled and Observed Production
Mean
Median
Minimum
maximum
Standard
Deviation
#
Observed
Carbon Fixation,
g C m^ d 1
Observed
1.08
0.93
0.00
4.44
0.84
876
Model
0.91
0.79
0.00
4.6
0.64
Gross Production,
g C m^ d"1
Observed
1.30
0.91
0.02
4.22
1.09
64
Model
1.19
1.01
0.10
3.26
0.73
Net production,
g C m2 d '
Observed
0.72
0.41
0.01
3.34
0.77
166
Model
0.87
0.82
0
2.72
0.50
References
Behrenfield, M., and Falkowski, P. (1977). "A consumer's guide to phytoplankton primary
production models," Limnology and Oceanography, 42(7), 1479-1491.
Bienfang, P., Harrison, P., and Quarmby, L. (1982). "Sinking rate response to depletion of
nitrate, phosphate, and silicate in four marine diatoms," Marine Biology, 67, 295-302.
Boynton, W., Garber, J., Summers, R., and Kemp, W. (1995). "Inputs, transformations, and
transport of nitrogen and phosphorus in Chesapeake Bay and selected tributaries,"
Estuaries, 18,285-314.
Cerco, C., and Cole, T. (1993). "Three-dimensional eutrophication model of Chesapeake
Bay," Journal of Environmental Engineering, 119(6), 1006-10025.
Cerco, C. (1995). "Response of Chesapeake Bay to nutrient load reductions," Journal of
Environmental Engineering, 121(8), 549-557.
Cerco, C. (2000). "Phytoplankton kinetics in the Chesapeake Bay Eutrophication Model,"
Journal of Water Quality and Ecosystem Modeling, 1, 5-49.
Cerco, C., and Meyers, M. (2000). "Tributary refinements to the Chesapeake Bay Model,"
Journal of Environmental Engineering, 126(2), 164-174.
Chan, A. (1980). "Comparative physiological study of marine diatoms and dinoflagellates
in relation to irradiance and cell size. 2. Relationship between photosynthesis, growth, and
carbon/chlorophyll ratio. Journal ofPhycology, 16, 428-432.
Cloern, J., Grenz, C., and Vidergar-Lucas, L. (1995). "An empirical model of the phyto-
plankton chlorophyll: carbon ratio—the conversion factor between productivity and growth
rate," Limnology and Oceanography, 40(7), 1313-1321.
Davis, C., Breitner, N., and Harrison, P. (1978). "Continuous culture of marine diatoms
under silicon limitation. 3. a model of Si-limited diatom growth," Limnology and Oceanog-
raphy, 23, 41-52.
-216-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
D'Elia, C., Nelson, D., and Boynton, W. 1983. "Chesapeake Bay Nutrient and Plankton
Dynamics: III. The Annual Cycle of Silicon," Geochimica et Cosmochimica Acta, 47:1945-
1955.
DiToro, D. (2001). Sediment Flux Modeling, John Wiley and Sons, New York.
Doney, S., Glover, D., and Najjar, R. (1996). "A new coupled, one-dimensional biological-
physical model for the upper ocean: Applications to the JGOFS Bermuda Atlantic
Time-series Study (BATS) site," Deep Sea Research II, 43(2- 3), 591-624.
Donigian, A., Bicknell, B., Patwardhan, A., Linker, L., Alegre, D., Chang, C., and
Reynolds, R. (1991). "Watershed model application to calculate bay nutrient loadings,"
Chesapeake Bay Program Office, US Environmental Protection Agency, Annapolis MD.
Durbin, A., and Durgin, E. (1975). "Grazing rates of the Atlantic menhaden Brevoortia
tyrannus as a function of particle size and concentration," Marine Biology, 33, 265-277.
Eppley, R., Rogers, J., and McCarthy, J. (1969). "Half-saturation constants for uptake of
nitrate and ammonium by marine phytoplankton," Limnology and Oceanography, 14(6),
912-920.
Eppley, R. (1972). "Temperature and phytoplankton growth in the sea," Fishery Bulletin,
70(4), 1063-1085.
Fasham, M., Ducklow, H., and McKelvie, S. (1990). "A nitrogen-based model of plankton
dynamics in the oceanic mixed layer," Journal of Marine Research, 48, 591-639.
Fisher, T, Gustafson, A., Sellner, K., Lacouture, R., Haas, L., Wetzel, R., Magnien, R.,
Everitt, D., Michaels, B., and Karrh, R. (1999). "Spatial and temporal variation of resource
limitation in Chesapeake Bay," Marine Biology, 133, 763-778.
Fisher, T, Peele, E., Ammerman, J., and harding, L. (1992). "Nutrient limitation of phyto-
plankton in Chesapeake Bay," Marine Ecology Progress Series, 82, 51- 63.
Gallegos, C. (1989). "Microzooplankton grazing on phytoplankton in the Rhode River,
maryland: nonlinear feeding kinetics," Marine Ecology Progress Series, 57, 23-33.
Gilmore, J., Glendening, P., Ridgeway, T, Williams, A., Browner, C., and Bullaway, M.
(2000). "Chesapeake 2000," Chesapeake Bay Program, US Environmental Protection
Agency, Annapolis MD.
Goldsworthy, A. (1970). "Photorespiration," The Botanical Review, 36(4), 321- 340.
Groeger, A., and Kimmel, B. (1989). "Relationship between photosynthetic and respiratory
carbon metabolism in freshwater plankton," Hydrobiologia, 173, 107-117.
Harding, L., Mallonee, M., and Perry, E. (2002). "Toward a predictive understanding of
primary productivity in a temperature, partially stratified estuary," Estuarine, Coastal and
Shelf Science, in press.
Harding, L., Meeson, B., and Fisher, T. (1986). "Phytoplankton in two east coast estuaries:
photosynthesis-light curves and patterns of carbon assimilation," Estuarine, Coastal and
Shelf Science, 23, 773-806.
Harris, R., Wiebe, P., Lenz, J., Skjodal, H., and Huntley, M. (2000). "Sampling, preserva-
tion, enumeration and biomass of marine protozooplankton." ICES zooplankton
methodology manual. Academic Press, San Diego, 193-221.
Hutchinson, G. (1967). A treatise on limnology, Volume II, John Wiley and Sons, New York,
245-305.
Jassby, A., and Platt, T. (1976). "Mathematical formulation of the relationship between
photosynthesis and light for phytoplankton," Limnology and Oceanography, 21, 540-547.
-217-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Johnson, B., Kim, K., Heath, R., Hsieh, B., and Butler, L. (1993). "Validation of a three-
dimensional hydrodynamic model of Chesapeake Bay," Journal of Hydraulic Engineering,
199(1), 2-20.
Kemp, W., Smith, E., DiPasquale, M., and Boynton, W. (1997). "Organic carbon balance
and net ecosystem metabolism in Chesapeake Bay," Marine Ecology Progress Series, 150,
229-248.
Kiddon, J., Bender, M., and Marra, J. (1995). "Production and respiration in the 1989 North
Atlantic spring bloom: an analysis of irradiance-dependent changes," Deep-Sea Research,
42(4), 553-576.
Laws, E., and Chalup, M. (1990). "A microalgal growth model," Limnology and Oceanog-
raphy, 35(3), 597-608.
Lomas, M., and Glibert, P. (1999). "Temperature regulation of nitrate uptake: A novel
hypothesis about nitrate uptake and reduction in cool-water diatoms," Limnology and
Oceanography, 44(3), 556-572.
Lomas, M., Trice, T., Glibert, P., Bronk, D., and McCarthy, J. (2002) "Temporal and spatial
dynamics of urea concentrations in Chesapeake Bay: Biological versus physical forcing,"
Estuaries, submitted.
Luo, J., Hartman, K., Brandt, S., Cerco, C., and Rippetoe, T. (2001). "A spatially-explicit
approach for estimating carrying capacity: an application for the Atlantic menhaden
(Brevoortia tyrannus) in Chesapeake Bay," Estuaries, 24(4), 545-556.
Malone, T., Crocker, L., Pike, S., and Wendler, B. (1988). "Influences of river flow on the
dynamics of phytoplankton production in a partially stratified estuary," Marine Ecology
Progress Series, 48, 235-249.
Malone, T., Conley, D., Fisher, T., Glibert, P., Harding, and Sellner, K. (1996). "Scales of
nutrient-limited phytoplankton productivity in Chesapeake Bay," Estuaries, 19, 371-385.
McCarthy, J., Taylor, W., and Taft, J. (1977). "Nitrogenous nutrition of the plankton in the
Chesapeake Bay. 1. Nutrient availability and phytoplankton preferences," Limnology and
Oceanography, 22(6), 996-1011.
McGillicuddy, D., McCarthy, J., and Robinson, A. (1995). "Coupled physical and biological
modeling of the spring bloom in the North Atlantic (I): model formulation and one dimen-
sional bloom processes," Deep Sea Research I, 42(8), 1313-1357.
McManus, G., and Ederington-Cantrell, M. (1992). "Phytoplankton pigments and growth
rates, and microzooplankton grazing in a large temperate estuary," Marine Ecology Progress
Series, 87, 77-85.
Miller, C., and Glibert, P. (1998). "Nitrogen excretion by the calanoid copepod Acartia
tonsa: results of mesocosm experiments," Journal of Plankton Research, 20(9), 1767-1780.
Moll, A. (1998). "Regional distribution of primary production in the North Sea simulated
by a three-dimensional model," Journal of Marine Systems, 16, 151- 170.
Monod, J. (1949). "The growth of bacterial cultures," Annual Review of Microbiology, 3,
371-394.
Pritchard, D. (1967). "Observations of circulation in coastal plain estuaries." Estuaries. G.
Lauff ed., American Association for the Advancement of Science, Washington D.C., 37-44.
Parsons, T., Takahashi, M., and Hargrave, B. (1984). Biological Oceanographic Processes,
Pergamon Press, Oxford.
Redfield, A., Ketchum, B., and Richards, F. (1966). "The influence of organisms on the
composition of sea-water." The Sea Volume II. Interscience Publishers, New York, 26-48.
-218-
-------
Chapter 10 • Process-Based Primary Production Modeling in Chesapeake Bay
Richardson, T., and Cullen, J. (1995). "Changes in buoyancy and chemical composition
during growth of a coastal marine diatom: ecological and biogeochemical consequences,"
Marine Ecology Progress Series, 128, 77-90.
Riebesell, U. (1989). "Comparison of sinking and sedimentation rate measurements in a
diatom winter/spring bloom," Marine Ecology Progress Series, 54, 109-119.
Rippetoe, T. (1993). "Production and energetics of Atlantic menhaden in Chesapeake Bay,"
M.S. Thesis, University of Maryland, College Park MD.
Smith, E., and Kemp, W. (1995). "Seasonal and regional variations in plankton community
production and respiration for Chesapeake Bay," Marine Ecology Progress Series, 116, 217-
231.
Strickland, J., and Parsons, T. (1972). "A practical handbook of seawater analysis," Bulletin
of the Fisheries Research Board of Canada" 167, 310 pp.
Taft, J., Taylor, W., and McCarthy, J. (1975). "Uptake and release of phosphorus by phyto-
plankton in the Chesapeake Bay estuary, USA," Marine Biology, 33, 21- 32.
Thomann, R., and Fitzpatrick, J. (1982). "Calibration and verification of a mathematical
model of the eutrophication of the Potomac Estuary," HydroQual Inc., Mahwah, NJ.
Waite, A., Thompson, P., and Harrison, P. (1992). "Does energy control the sinking rates of
marine diatoms?." Limnology and Oceanography, 37(3), 468- 477.
Wang, H., and Johnson, B. (2000). "Validation and application of the second-generation,
three-dimensional hydrodynamic model of Chesapeake Bay," Journal of Water Quality and
Ecosystem Modeling, 1, 51-90.
Westrich, J., and Berner, R. (1984). "The role of sedimentary organic matter in bacterial
sulfate reduction: The G model tested," Limnology and Oceanography, 29(2), 236-249.
Wheeler, P., Gilbert, P., and McCarthy, J. (1982). "Ammonium uptake and incorporation by
Chesapeake Bay phytoplankton: short-term uptake kinetics," Limnology and Oceanography,
27, 1113-1128.
White, J., and Roman, M. (1992). "Seasonal study of grazing by metazoan zooplankton in
the mesohaline Chesapeake Bay," Marine Ecology Progress Series, 86, 251-261.
-219-
-------
Suspended Solids
and Light Attenuation
Data Bases
Solids
Suspended solids observations in the monitoring data base included total
suspended solids, fixed solids, volatile solids, and particulate organic carbon. (For
practical purposes, fixed and volatile solids correspond to inorganic and organic
solids.) Direct analyses of organic solids were limited to stations in the Virginia
tributaries (Table 11-1, Figure 11-1). The preponderance of organic solids observa-
tions was in the form of particulate organic carbon. We converted particulate
organic carbon to organic solids through the relationship:
VSS = 2.5 • POC (11-1)
in which:
VSS = organic solids concentration (g m'3)
POC = particulate organic carbon concentration (g C nr3)
This commonly-employed relationship assumes organic matter is comprised of a
simple carbohydrate CH2O. The total mass of this compound is 2.5 times the
carbonaceous mass. Organic solids were subtracted from total suspended solids to
obtain inorganic solids concentration.
Examination of the system-wide solids distribution indicates that organic solids
represent a small fraction of the total suspended solids (Figures 11-2 to 11-4). Only
in regions characterized by two stations, CB4.2C and CBS.2, do organic solids
comprise as much a half the total suspended solids. The observed solids com-
position has significant implications for management activities in the bay system.
Attempts to reduce organic solids concentrations, and thereby light attenuation,
through nutrient controls address only a small fraction of the total suspended solids
in the water column. Analyses using the submerged aquatic vegetation model
(Cerco and Moore 2001) indicate system-wide restoration of submerged aquatic
vegetation through nutrient controls alone is impossible.
-220-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Table 11-1
Number of Solids Analyses in Surface Samples 1985-1994
Station
CB1.1
CB2.2
CB3.3C
CB4.2C
CB5.2
CB6.1
CB7.3
CB7.4
CB7.4N
CB8.1E
EE1.1
EE2.1
EE3.1
EE3.2
ET1.1
ET2.3
ET4.2
ET5.2
ET6.2
ET9.1
LE1.3
LE2.2
LE3.2
LE4.2
LE5.3
RET1.1
RET2.4
RET3.2
RET4.2
RET4.3
RET5.2
TF1.7
TF2.1
TF3.3
TF4.2
TF4.4
TF5.5
WE4.2
WT1.1
WT2.1
WT5.1
WT8.1
Total
Suspended
Solids
186
201
293
206
208
194
204
204
176
182
197
268
380
344
176
102
204
225
108
102
194
422
151
114
138
194
391
116
148
113
113
415
376
111
116
176
119
187
205
204
420
113
Total Volatile
Solids
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
95
60
88
0
0
60
93
59
60
0
0
58
63
86
65
0
0
0
0
0
Fixed Solids
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
151
114
138
0
0
116
148
113
113
0
0
111
116
176
119
0
0
0
0
0
Particulate
Organic
Carbon
181
197
227
200
196
193
208
204
186
181
182
254
288
253
155
90
191
207
101
93
181
123
12
12
11
179
20
12
12
12
11
388
23
12
12
17
12
186
190
186
322
103
-221 -
-------
Chapter 11 • Suspended Solids and Light Attenuation
Figure 11-1. Monitoring stations used in analysis of suspended solids and light
attenuation.
20 -i
18
A A
\i\
tA
O
CO g
01
4
D Total Suspended Solids
• Organic Solids
n
r
r
•
\\
i
CD en m
0 0 CQ
O
i
o
S
o
\
s i
0 0
1
n
m
0
1
1
1
1
,
i
•
l~- ^ T T- CN CO CO
03 r- OD LJJ LJJ LU LU
O CD CO LLJ LLJ QJ LU
00
Figure 11-2. Median surface total suspended solids and organic solids
concentrations at mainstem bay monitoring stations, 1985-1994.
-222-
-------
Chapter 11 • Suspended Solids and Light Attenuation
40
oc _
E
10
^ 20
o
w
o> 15
•i n
5 -
D Total S
• Organi
uspended Solids
c Solids
L1
tr
t
1
CO CM CNl O) CO T-;
T- rsi co -a- iri -r^
LU LU LU LU LJJ h-
_l _l _l _l _l LU
QL
—
—
r
-
•
r
I
r
—
I
rf
HHHHHLLLLLLLT
U LJJ LU LJ LU 1— 1— 1— 1—
o: o: ct: o: cc
=
r
Ll_ Ll-
1- 1-
Figure 11-3. Median surface total suspended solids and organic solids
concentrations at western tributary monitoring stations, 1985-1994.
40
35
30
« 25
E
tn
1 20
o
w
03 15
10
5
D Total Suspended Solids
• Organic Solids
•«- CO
r £
LU LU
LU
LU
p
LU
10
00
LU
Figure 11-4. Median surface total suspended solids and organic solids
concentrations are minor tributary monitoring stations, 1985-1994.
-223-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Light Attenuation
Two measurements of light attenuation were in the monitoring data base. The
first was disk visibility or Secchi depth. The second was measurement of down-
welling irradiance at multiple depths. Measures of disk visibility were almost
universal from the inception of the monitoring program while measures of irradi-
ance were phased in at various dates and locations. Analysis of light attenuation
over the model simulation period therefore relied on disk visibility as the primary
measure of light attenuation.
Attenuation can be obtained from disk visibility through the relationship:
Ke • DV = K (H-2)
in which:
Ke = coefficient of diffuse light attenuation (m"1)
DV = Secchi depth (m)
K = empirical constant
The value of usually ranges between 1 and 2 although extremes from 0.5 to
nearly 4 have been observed (Koenings and Edmundson 1991). The median value
in clear water is 1.9. Higher values prevail in colored water while lower values are
prevalent in turbid systems.
We obtained from Richard Lacouture, of the Academy of Natural Sciences, a
data base consisting of simultaneous measures of disk visibility and downwelling
irradiance. Measures were collected in the Maryland portion of the bay and in
Maryland tributaries from 1985 to 1996. For each sampling, we computed light
attenuation by fitting an exponential relationship to the irradiance observations:
in which:
I(z) = irradiance at depth z (fiE m"2 s"1)
lo = surface irradiance (|iE m"2 s"1)
z =depth (m)
Ke was determined by linear regression on log-transformed data. The value of
R-squared for the individual regressions usually exceeded 0.9.
A set of paired observations of attenuation and disk visibility resulted after the
determination of attenuation at each sampling. We again employed regression to
obtain an estimate of parameter K. We attempted to relate K to location, solids
concentration, and other variables but found no significant trends. Variance in the
observations and the probabilistic nature of regression analysis prevented determi-
nation of an absolute, precise value for K. We settled on the value K = 1.3. Our
value is slightly lower but comparable to a median of 1.44 (range 1.15 to 2.31)
determined for turbid waters (Holmes 1970) and the value 1.43 determined for the
Patuxent River (Keefe et al. 1976).
-224-
-------
Chapter 11 • Suspended Solids and Light Attenuation
The Light Attenuation Model
We proposed a model that related light attenuation to attenuation from water,
from inorganic suspended solids, from organic suspended solids, from dissolved
organic matter, and from chlorophyll:
Ke = ai + a2 • ISS + a3 • VSS + a4 • DOC + a5 • Chi (11-4)
in which:
aj to a5 = empirical constants
ISS = inorganic suspended solids concentration (g m"3)
VSS = organic suspended solids concentration (g m"3)
DOC = dissolved organic carbon (g C m"3)
Chi = chlorophyll (mg m"3)
We adopted dissolved organic carbon as a surrogate for the dissolved organic
matter that provides color in water.
Parameters in the model were determined through linear regression. Regression
was performed for each sample station to allow for regional variations in parame-
ters. We found no significant relationship of attenuation to dissolved organic
carbon. Various explanations for absence of a relationship may hold. The first is
that dissolved organic carbon is a poor surrogate for colored organic matter. The
second is that no significant spatial gradients of dissolved organic carbon exist. The
third is that enormous variance is present in the analysis of dissolved organic
carbon and masks any relationship to attenuation. Likely all of these explanations
have validity.
We also found we could not distinguish attenuation from both chlorophyll and
organic solids. The two variables were too closely correlated for their properties to
be isolated by regression. We decided to drop chlorophyll from the model and
retain organic solids. We reasoned we would like the model to distinguish potential
differences in attenuation between inorganic and organic solids. Since our parame-
ters were determined by regression, attenuation by chlorophyll was included as part
of the total attenuation by organic solids.
The resulting attenuation relationship had the form:
..3 • VSS (^-3)
in which:
aj = background attenuation (m"1)
82 = attenuation by inorganic suspended solids (m2 g"1)
a3 = attenuation by organic suspended solids (m2 g"1)
The "background" attenuation term included attenuation from both water and
dissolved organic matter.
Individual parameters were determined for each Chesapeake Bay Program
Segment represented in the model domain (Figure 11-5). These were refined to
provide reasonable consistency in the parameterization. In a few cases, background
attenuation was adjusted to counter problems with the model solids computation.
The result (Table 11-2) was a model in which background attenuation was highest
-225-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Figure 11-5. Chesapeake Bay Program Segments.
-226-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Table 11 -2
Coefficients for Light Attenuation Model
Segment
CB1
CB2
CBS
CB4
CBS
CB6
CB7
CBS
EE1
EE2
EE3
ET1
ET2
ET3
ET4
ET5
ET6
ET7
ET8
ET9
ET10
LE1
LE2
LE3
LE4
LE5
RET1
RET2
RETS
RET4
RETS
TF1
TF2
TF3
TF4
TF5
WE4
WT1
WT2
WT3
WT4
WT5
WT6
WT7
WT8
a1,nrT1
0.7
0.47
0.3
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.2
0.47
0.47
0.47
0.3
0.2
0.2
0.2
0.2
0.2
0.2
0.3
0.3
0.3
0.3
0.3
0.47
0.47
0.47
0.47
0.47
0.05
0.47
0.47
0.47
0.47
0.2
0.47
0.3
0.3
0.3
0.2
0.3
0.2
0.2
a2, m2 g-1
0.117
0.117
0.117
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.117
0.117
0.117
0.117
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.117
0.117
0.117
0.117
0.117
0.117
0.083
0.083
a3, m2 g"1
0.117
0.117
0.117
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.117
0.117
0.117
0.117
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.083
0.117
0.117
0.117
0.117
0.117
0.117
0.083
0.083
-227 -
-------
Chapter 11 • Suspended Solids and Light Attenuation
near the freshwater sources and lowest near the ocean interface. This conceptual-
ization was consistent with the observation that freshwater is more colored than
seawater (Kirk 1994). During the refinement, the distinction of coefficients for
inorganic and organic solids was dropped but solids near the fall lines were charac-
terized as having higher attenuation than solids in regions distant from sources in
upland watersheds.
Comparison with the SAV Technical Synthesis II
An additive light attenuation model similar to our own was created as part of the
second technical synthesis on submerged aquatic vegetation and water quality
(Chesapeake Bay Program 2000). The model incorporated background attenuation,
attenuation from total suspended solids, and attenuation from chlorophyll. As with
our model, parameters were initially obtained from regression and then refined to
improve agreement with observations.
Regressions produced background attenuation in range 0.393 to 3.32 m"1. The
value 0.32 m"1 was recommended for use system-wide. This value is centrally
located within the values 0.05 to 0.7 m"1 used in our model and is close to our
predominant value of 0.2 m"1.
Regressions indicated the coefficient that related total suspended solids contribu-
tion to light attenuation was in the range 0.013 to 0.101 m2 g"1. The coefficient
recommended for use system-wide was 0.094 m2 g"1. This value is nearly identical
to the average of the two values, 0.083 and 0.117 m2 g"1, used in our own model.
Fewer than half of the regressions conducted as part of the "Tech Syn II"
resulted in significant relationships between chlorophyll and light attenuation.
Several of the statistically significant relationships were negative, indicating a
physically-impossible relationship. These results echoed our own difficulty in
isolating the effect of chlorophyll via regression. The attenuation coefficient of
chlorophyll is well-known, however (e.g. Pennock 1985). Consequently, the Tech
Syn II investigators elected to include chlorophyll attenuation, 0.016 m2 mg"1, in
their model. If we assume a carbon-to-chlorophyll ratio of 75 and utilize the ratio
of 2.5 between solids and organic carbon, we find the chlorophyll attenuation used
in Tech Syn II is equivalent to 0.085 m2 g"1 organic solids, remarkably close to the
prevailing value used in our own model.
We are relieved and reassured that the light attenuation relationships obtained in
independent investigations are equivalent. Managers and other users of the two
models, ours and Tech Syn II, should be confident that guidance obtained from the
two models will be consistent.
Solids Settling Velocities
The principal suspended solids variable in the CBEMP was inorganic (fixed)
suspended solids. Organic solids were derived from particulate carbon variables
(phytoplankton, zooplankton, detritus) and added to inorganic solids for compar-
ison to observed total suspended solids.
No internal sources or sinks of inorganic solids were considered in the model.
Aside from loads, the distribution of inorganic solids in the water column was
determined by two settling velocities. One represented settling through the water
-228-
-------
Chapter 11 • Suspended Solids and Light Attenuation
column and the other represented net settling to the bed sediments. Within the
water column, the transport equation for inorganic solids was identical to the equa-
tion for all other variables. A modification applied in cells that interfaced the bed
sediments:
§C ,,1, W Wnet ni ,,
— = [transport] + — • Cu • C (H-6)
§t Az Az
in which:
C = solids concentration in cell adjoining bed sediments (g m"3)
Cu = solids concentration two cells above bed sediments (g m"3)
W = settling velocity in water column (m d"1)
Wnet = settling velocity from water to bed sediments (m d"1)
z = cell thickness (m)
t = time (d)
Net settling was always less than or equal to settling through the water column.
The reduced magnitude of net settling represented the net effect of resuspension.
The ratio of net settling to settling through the water column may be viewed as the
fraction of material deposited on the bed that remains in the bed. The employment
of net settling is a primary distinction between our own suspended solids model
and a true sediment transport model. Our model included no resuspension mecha-
nism. Once a particle was deposited on the bottom, it remained there.
Settling and net settling were evaluated through a recursive calibration process
based on visual fitting of computed to observed solids. Settling velocities obtained
in this fashion were highest near the fall lines of several western tributaries and in a
few minor Eastern Shore tributaries (Table 11-3). The high settling velocities in the
western tributaries diminished with distance from the head of tide, perhaps indi-
cating sorting out of large particles carried over the fall line. Our estimates of
loading from bank erosion were highly uncertain in the Eastern Shore embayments.
The high settling velocities assigned in some tributaries there may reflect deposi-
tional environments or may be a compensation for overestimated bank loads.
Settling velocities assigned in the mainstem bay (Table 11-3) also diminished away
from the source in the Susquehanna River but were less than in several western
tributaries. The velocities assigned suggested particles with high settling velocities
settled out in Conowingo Reservoir before reaching the bay.
Net settling velocities were five to ten percent of settling in the water column.
(Table 11-3).
Net settling was enhanced in the presence of submerged aquatic vegetation
(Cerco and Moore 2001) to represent the damping effect of vegetation on waves
and on resuspension.
Our model did not consider solids size classes. We found a modification to
assigned settling velocities was required during major storm events. Storm events
were defined as intervals in which inorganic solids concentration exceeded
100 g m"3. During storm events, the inorganic solids settling rate was specified
as 5 m d"1 and no resuspension was allowed. This assignment simulated rapid
settling of large particles carried over the fall line during floods. If we did not
assign rapid settling during storms, solids loads associated with storms remained in
the water column perpetually.
-229-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Table 11 -3
Inorganic Solids Settling Rates
Segment
CB1
CB2
CBS
CB4
CBS
CB6
CB7
CBS
EE1
EE2
EE3
ET1
ET2
ET3
ET4
ET5
ET6
ET7
ET8
ET9
ET10
LE1
LE2
LE3
LE4
LE5
RET1
RET2
RETS
RET4
RETS
TF1
TF2
TF3
TF4L
TF4R
TF5
WE4
WT1
WT2
WT3
WT4
WT5
WT6
WT7
WT8
Settling,
md1
1.25
1.25
1.25
1
1
1
1
1
1.5
1.5
1
1.5
1
1.5
2.5
2.5
1.5
2
2
2.5
2
1.75
1.5
2
2
1.5
2.5
1.5
2
2
2
2.5
1
3
4
3.5
3
1
1.5
1.5
1.5
1.5
2
1.5
1.5
1.5
Net Settling,
md1
0.1
0.1
0.05
0.05
0.05
0.05
0.05
0.05
0.15
0.15
0.025
0.15
0.025
0.15
0.25
0.25
0.15
0.2
0.2
0.25
0.2
0.175
0.15
0.2
0.2
0.1
0.25
0.15
0.1
0.2
0.1
0.25
0.05
0.3
0.4
0.35
0.3
0.1
0.15
0.15
0.15
0.15
0.2
0.15
0.15
0.15
-230-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Model Results
The time series, spatial plots, and summary plots of suspended solids and light
attenuation produced during the model calibration comprise over 350 figures.
Obviously, not all can be pulled into this chapter. We present a sampling here. A
complete set of plots is provided on the CD-ROM that accompanies this report.
Time Series
Time series of computed and observed total suspended solids and light attenua-
tion are presented for Station CB2.2 in the upper bay (Figures 11-6 to 11-8),
Station CB5.2 at mid-bay (Figures 11-9 to 11-11), and Station CB7.3 in the lower
bay (Figures 11-12 to 11-14). The high degree of short-term variability in both
observations and model confounds interpretation of results although several proper-
ties are apparent. Certainly, the model represents the central tendency and extremes
of the observations although one-to-one agreement between model and instanta-
neous observations is not always present. Higher computed values of solids
concentrations and light attenuation occur in winter and spring, the periods of
maximum runoff and solids loading from the watershed. In the upper bay, concen-
tration and attenuation maxima are "flashy," exhibiting large excursions over short
periods. In the mid-bay, the excursions are spread out in time and damped in ampli-
tude while they are barely noticeable in the lower bay. The model exhibits a large
impact from the wet years of 1993 and 1994 although this impact is not so
pronounced in the observations. Observed suspended solids concentrations show a
greater range at the bottom than at the surface, no doubt due to sediment resuspen-
sion. The model demonstrates higher sediment concentrations at the bottom that at
the surface but the observed range is under-estimated. The specification of net
settling less than settling through the water column results in higher computed
concentrations at the bottom than at the surface. A true resuspension algorithm is
required for the model to reproduce the full range of observed concentrations.
Final Calibration - SENS 136
Total Solids CB2.2 Surface
• I* • ^ * » *,
Final Calibration - SENS 136
Total Solids CB2.2 Bottom
Figure 11-6. Computed and observed total sus-
pended solids at Station CB2.2 (surface).
Figure 11-7. Computed and observed total sus-
pended solids at Station CB2.2 (bottom).
-231 -
-------
Chapter 11 • Suspended Solids and Light Attenuation
Final Calibration - SENS 136
Light Extinction CB2.2 Surface
Final Calibration - SENS 136
Total Solids CBS.2 Surface
Figure 11-8. Computed and observed light
attenuation at Station CB2.2.
Final Calibration - SENS 136
Total Solids CB5.2 Bottom
tftewfe
: /.. ..' v. .-'*.' «... • **.. ".* •.» V.*
, I , : . I , . . 1 . , : I t , , 1 . , I . , . 1 . - , I , , , I . , , I
123456789 10
Years
Figure 11-10. Computed and observed total
suspended solids at Station CB5.2 (bottom).
Final Calibration - SENS 136
Total Solids CB7.3 Surface
Figure 11-12. Computed and observed total
suspended solids at Station CB7.3 (surface).
Figure 11-9. Computed and observed total
suspended solids at Station CBS.2 (surface).
Final Calibration - SENS 136
Light Extinction CB5.2 Surface
Figure 11-11. Computed and observed light
attenuation at Station CB5.2.
Final Calibration - SENS 136
Total Solids CB7.3 Bottom
012
3456
Years
8 9 10
Figure 11-13. Computed and observed total
suspended solids at Station CB7.3 (bottom).
-232-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Final Calibration - SENS 136
Light Extinction CB7.3 Surface
Figure 11-14. Computed and observed light
attenuation at Station CB7.3.
Longitudinal Distributions
The spatial distribution of suspended solids and light attenuation were examined,
on a seasonal basis, along the axes of the bay and major tributaries. Details of the
averaging process and transect maps were presented in the chapter entitled "Intro-
duction to the Calibration."
Chesapeake Bay. The most interesting feature of the observed solids distribu-
tion is the presence of two solids maxima in Chesapeake Bay (Figures 11-15 to
11-26). One maximum, at km 280 to 300, is the classic turbidity maximum noted
by Schubel (1968) and others. The second maximum occurs roughly 70 km above
the mouth of the bay. Hood et al. (1999) noted regions of high particle concentra-
tions in the lower bay. They attributed these concentrations to a persistent eddy and
to processes including convergence and downwelling. The concentrations noted by
Hood et al. were downstream (km 25 on our axis) of the maximum observed here
and primarily to the east of our transect. We cannot be certain if the high observed
concentrations around km 70 are a result of the processes noted by Hood et al. or
not. Our model preforms reasonably in reproducing the solids maximum at km 280
while the maximum at km 70 is not reproduced at all. Hood et al. used a particle-
tracking algorithm to reproduce the maxima in the lower bay. Particle tracking is
not implemented in our model nor is this envisioned as a reasonable improvement
for future model versions.
The solids maximum in the lower bay is not accompanied by a corresponding
peak in light attenuation. As part of our modeling scheme, we assigned more color
to water in the upper bay and assigned higher attenuation properties to solids in the
upper bay. The absence of an observed attenuation maximum associated with the
observed solids maximum in the lower bay suggests our interpretation of color and
attenuation is correct. The solids in the lower bay do not attenuate light to the same
degree as solids in the upper bay. The bay exhibits one turbidity maximum associ-
ated with the solids maximum at km 280 to 300. Our model provides a reasonable
representation of attenuation in the turbidity maximum.
-233-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Mainstem Bay Final Calibration - SENS 136
Surface Total Suspended Solids Winter 1990
Kilometers
Mainstem Bay Final Calibration - SENS 136
Bottom Total Suspended Solids Winter 1990
Kilometers
Figure 11-15. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, winter 1990 (surface).
Figure 11-16. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, winter 1990 (surface).
35
3
h. 2.5
I ?
l.h
Mainstem Bay Final Calibration - SENS 136
Surface Light Extinction Winter 1990
Figure 11-17. Computed (mean and range) and
observed (mean and range) light attenuation along
Chesapeake Bay axis, winter 1990.
Mainstem Bay Final Calibration - SENS 136
Surface Total Suspended Solids Spring 1990
Kilometers
Figure 11-18. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, spring 1990 (surface).
Mainstem Bay Final Calibration - SENS 136
Bottom Total Suspended Solids Spring 1990
Figure 11-19. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, spring 1990 (bottom).
Mainstem Bay Final Calibration - SENS 136
Surface Light Extinction Spring 1990
ICO 2<»
Kilometers
Figure 11-20. Computed (mean and range) and
observed (mean and range) light attenuation along
Chesapeake Bay axis, spring 1990.
-234-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Mainstem Bay Final Calibration - SENS 136
Surface Total Suspended Solids Summer 1990
tOO TO.
Kilometers
Mainstem Bay Final Calibration - SENS 136
Bottom Total Suspended Solids Summer 1990
Figure 11-21. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, summer 1990 (surface).
Figure 11-22. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, summer 1990 (bottom).
Mainstem Bay Final Calibration - SENS 136
Surface Light Extinction Summer 1990
Figure 11-23. Computed (mean and range) and
observed (mean and range) light attenuation along
Chesapeake Bay axis, summer 1990.
ISO
100
90
ao
10
1 60
bO
.10
30
Mainstem Bay Final Calibration - SENS 136
Surface Total Suspended Solids Fall 1990
100 200
Kilometers
Figure 11-24. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, fall 1990 (surface).
Mainstem Bay Final Calibration - SENS 136
Bottom Total Suspended Solids Fall 1990
Kilometers
Figure 11-25. Computed (mean and range) and
observed (mean and range) total suspended solids
along Chesapeake Bay axis, fall 1990 (bottom).
Mainstem Bay Final Calibration - SENS 136
Surface tight Extinction Fall 1990
,,*'*
Kilometers
Figure 11-26. Computed (mean and range) and
observed (mean and range) light attenuation along
Chesapeake Bay axis, fall 1990.
-235-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Western Tributaries. A conspicuous feature of the solids and attenuation
computations in the western tributaries (Figures 11-27 to 11-41) is enormous vari-
ance in computations near the fall lines. The variance corresponds to variance in
loadings from the watershed. We employed "cruise averaging" so that model
results are averaged only over sampling intervals. Still, in view of the variance in
loads, ideal agreement between computed and observed means near the fall lines
should not be expected. Improved comparisons should be expected downstream of
the fall lines as the impact of loads is damped with distance from the source.
Both the location and magnitude of the turbidity maximum in the James River
are well represented (Figures 11-27 to 11-29). Above the maximum, the computed
surface solids and attenuation distributions are flat to the fall line. We found it diffi-
cult to replicate the observed increase in surface solids concentration between the
fall line and the turbidity maximum. Computed solids concentration along the
James River Final Calibration - SENS 136
Surface Total Suspended Solids Summer 1990
no
100
30
James River Final Calibration - SENS 136
Bottom Total Suspended Solids Summer 1990
Kilometers
50 100
Kilometers
Figure 11-27. Computed (mean and range) and
observed (mean and range) total suspended solids
along James River axis, summer 1990 (surface).
Figure 11-28. Computed (mean and range) and
observed (mean and range) total suspended solids
along James River axis, summer 1990 (bottom).
James River Final Calibration - SENS 136
Surface Light Extinction Summer 1990
*. 5
s
50 100
Kilometers
Figure 11-29. Computed (mean and range) and
observed (mean and range) light attenuation along
James River axis, summer 1990.
-236-
-------
Chapter 11 • Suspended Solids and Light Attenuation
bottom provides a better approximation of the increase from the fall line to the
turbidity maximum around km 70. Downstream of the turbidity maximum, the
model provides good representation of the solids concentration and excellent repre-
sentation of light attenuation.
The observed turbidity maximum in the York River occurs around km 70
(Figures 11-30 to 11-32). Computed solids concentration climbs from the intersec-
tion with the bay (km 0) to around km 40. From km 40 to the fall line, computed
solids distribution is flat. Computed solids concentration in the turbidity maximum
is underestimated. Computed light attenuation agrees well with observed through to
km 70. Above km 70, observations are sparse but the model tends to overestimate
light attenuation.
130
170
110
100
90
80
York River Final Calibration - SENS 136
Surface Total Suspended Solids Summer 1993
50
Kilometers
York River Final Calibration - SENS 136
Bottom Total Suspended Solids Summer 1993
50
Kilometers
Figure 11-30. Computed (mean and range) and
observed (mean and range) total suspended solids
along York River axis, summer 1993 (surface).
Figure 11-31. Computed (mean and range) and
observed (mean and range) total suspended solids
along York River axis, summer 1993 (bottom).
York River Final Calibration - SENS 136
Surface Light Extinction Summer 1993
50
Kilometers
Figure 11-32. Computed (mean and range) and
observed (mean and range) light attenuation along York
River axis, summer 1993
-237-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Observations from 1990 indicate a turbidity maximum in the Rappahannock
around km 80 (Figures 11-33 to 11-35). The model computes a maximum of the
correct magnitude at km 70. Owing to the spatial distribution of observations, the
location of the computed maximum cannot be disputed. In this year, the solids
concentration in the bottom is well represented but solids are underestimated in the
surface layer below km 70. In alternate years e.g. 1993 (see CD-ROM), computed
solids are high in the bottom but match the surface observations well. Computed
solids concentrations exhibit a consistent secondary peak around km 130. The
observations are spaced too far apart to determine if the secondary peak is real or
an artifact of the model. Computed light attenuation follows the solids distribution.
Peak computed attenuation occurs in the turbidity maximum around km 70 and in
the second maximum upstream around km 130. The model provides reasonable
agreement with observed attenuation magnitude and distribution.
Rappahannock River Final Calibration - SENS 136
Surface Total Suspended Solids Summer 1990
0 50 100
Kilometers
100
90
Rappahannock River Final Calibration - SENS 136
Bottom Total Suspended Solids Summer 1990
Figure 11-33. Computed (mean and range) and
observed (mean and range) total suspended solids
along Rapahannock River axis, summer 1990
(surface).
Figure 11-34. Computed (mean and range) and
observed (mean and range) total suspended solids
along Rappahannock River axis, summer 1990
(bottom).
Rappahannock River Final Calibration - SENS 136
Surface Light Extinction Summer 1990
50 100
Kilometers
Figure 11-35. Computed (mean and range) and
observed (mean and range) light attenuation along
Rappahannock River axis, summer 1990.
-238-
-------
Chapter 11 • Suspended Solids and Light Attenuation
The predominant characteristic of the Potomac River solids observations is an
enormous concentration peak in the bottom waters around km 100 (Figures 11-36
to 11-38). The peak barely penetrates into the surface waters, however, so that
pronounced peaks in surface solids and in light attenuation are not apparent at the
location of the subsurface solids maximum. In 1985 and in other seasons (see CD-
ROM), surface solids and attenuation are as likely to peak further upstream, around
km 140, as they are to peak coincident with the bottom solids maximum. Interest-
ingly, in 1985, the observed upstream limit of salt intrusion was around km 130
(Figure 11-39). The location of the salt intrusion suggests the solids peak around
km 130 represents a classic turbidity maximum located at the convergence of
upstream and downstream circulation. The subsurface peak at km 100 may result
from an alternate process such as a concentration of solids in a deep hole or in a
location of large shear stress. In summer, modeled solids and attenuation peak still
further upstream and are associated with the freshwater algal bloom.
Potomac River Final Calibration - SENS 136
Surface Total Suspended Solids Summer 1985
50 100
Kilometers
, 10C!
80
80
Potomac River Final Calibration - SENS 136
Bottom Total Suspended Solids Summer 1985
50 100
Kilometers
Figure 11-36. Computed (mean and range) and
observed (mean and range) total suspended solids
along Potomac River axis, summer 1985 (surface).
Figure 11-37. Computed (mean and range) and
observed (mean and range) total suspended solids
along Potomac River axis, summer 1985 (bottom).
Potomac River Final Calibration - SENS 136
Surface Light Extinction Summer 1985
Potomac River Final Calibration - SENS 136
Bottom Salinity Summer 1985
"\
too
Kilometers
50 100
Kilometers
Figure 11-38. Computed (mean and range) and
observed (mean and range) light attenuation along
Potomac River axis, summer 1985.
Figure 11-39. Computed (mean and range) and
observed (mean and range) salinity along Potomac
River axis, summer 1985 (bottom).
-239-
-------
Chapter 11 • Suspended Solids and Light Attenuation
The turbidity maximum in the Patuxent occurs around km 60 (Figures 11-40 to
11-42). The model provides a reasonable representation of solids concentration
from the mouth of the Patuxent up to km 60. Near the upstream limit of the model,
computed solids tend to exceed observed during 1985. The computed concentra-
tions at the upper end of the Patuxent exhibit enormous variance in this year and in
others (see CD-ROM). Calculated concentrations reflect the loading and can be
affected to only a minimum extent by assignment of parameters with the model.
Except for one improbable observation, the model provides a reasonable represen-
tation of light attenuation in the Patuxent. As with solids, computed attenuation at
the upper end of the river reflects enormous variance in the loads and can be
affected only minimally through adjustment of the model.
Patuxent River Final Calibration - SENS 136
Surface Total Suspended Solids Summer 1985
400
J50
300
250
Kilometers
Patuxent River Final Calibration - SENS 136
Bottom Total Suspended Solids Summer 1985
wo
250
_j
f 200
150
100
SO
-Vf,~
20 40
Kilometers
Figure 11-40. Computed (mean and range) and
observed (mean and range) total suspended solids
along Patuxent River axis, summer 1985 (surface).
Figure 11-41. Computed (mean and range) and
observed (mean and range) total suspended solids
along Patuxent River axis, summer 1985 (bottom).
Patuxent River Final Calibration - SENS 136
Surface Light Extinction Summer 1985
i 15
40
Kilometers
Figure 11-42. Computed (mean and range) and
observed (mean and range) light attenuation along
Patuxent River axis, summer 1985.
-240-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Distribution Plots
Observations and computations from our time series stations (Figure 11-1) were
compared on cumulative distribution plots. Observations were paired with daily-
average computations on sample days so that the plots compared distributions of
observations and computations matched in location and time. Each station
contributed = 150 to 250 attenuation observations and double that quantity (surface
plus bottom) of suspended solids observations. Observations from only our time
series stations were considered and these were pooled into distributions for the bay,
major tributaries, and minor tributaries. More extensive comparisons of selected
parameters were prepared by the Chesapeake Bay Program Office, on a station-by-
station basis, for use in TMDL analyses.
The upper percentiles of the observations (> =98%) are frequently dominated
by a few extreme observations. These extremes are often improbable e.g. light
attenuation of 25 m"1. The use of automated data processing and plotting routines
on enormous numbers of observations makes detection and elimination of bad data
almost impossible. The viewer is cautioned against placing great emphasis on the
extreme observations.
Since disk visibility was recorded at fixed intervals, the distribution of attenua-
tion observations appears as a series of steps.
The distributions of computed and observed solids and attenuation for the bay
(Figures 11-43 to 11-44) show remarkable agreement. The solids distributions
agree at both extremes and are offset by roughly 1 gm m'3 at the median.
Computed and observed attenuation are coincident through most of the distribution
and depart only in the lower 10% to 15%. In this range, the model computes atten-
uation that exceeds the observations by =0.1 m"1.
10'
_J
1,0'
10*
Final Calibration - SENS 1 36 I
Total Solids
Mainstem Bay
i
i
i
j
- I . L I I L J i L • J 1 i 1 1. 1 i
26 50 75 100 J
Percent Less tti an
If
'1
7
6
t. 3
s ?
1
Final Calibration - SENS 1 36
Light Extinction
[ Mainstem Bay
f
" J
,---'"'''
25 60 75 100
Percent Less Hi an
Figure 11-43. Cumulative distribution of
computed and observed total suspended
solids in Chesapeake Bay.
Figure 11-44. Cumulative distribution of
computed and observed light attenuation in
Chesapeake Bay.
-241 -
-------
Chapter 11 • Suspended Solids and Light Attenuation
Computed and observed solids in the James River show reasonable agreement
though the lower half of the distribution but observed solids exceed computed in
the upper half (Figure 11-45). Inspection of the time-series plots indicates inability
of the model to match extreme solids concentrations in the bottom waters of the
turbidity maximum contributes substantially to the mis-match. The attenuation
distributions (Figure 11-46), which are based solely on surface solids, agree well.
The distributions of computed and observed solids and attenuation for the York
River show good agreement (Figure 11-47) through the upper half of the distribu-
tion. In the lower half, computed solids exceed observed by less than 5 g m"3.
Computed and observed attenuation follow a similar pattern (Figure 11-48). Agree-
ment is good in the upper half of the distribution but the model overestimates by
=0.2 m"1 in the lower half.
Final Calibration - SENS 136
Total Solids
James River
Model
_L
Percent Lessttipn
Figure 11-45. Cumulative distribution of
computed and observed total suspended
solids in James River.
Final Calibration - SENS 1 3§
35
30
25
2®
10
I *
Light Extinction
- James River
I
7
:
.
25 50 75 100
Percent Less thin
Figure 11-46. Cumulative distribution of
computed and observed light attenuation
in James River.
"id"
E
10'
Final Calibration - SENS 1 36
Total Solids
York River
'
/•"
..-^
j'jiLi • J. j i . . 1 i i i i I
2S 50 75 100
Percent Less than
1
s
Final Calibration - SENS 1 36
Light Extinction
14
!•*
11
9
8
J
a
5
4
3
Z
: York River
• Model
-
-
,."•'"""
. . , J , I I , !_ . 1 . , . ,
26 50 ?5 1(
Percent Less ttian
Figure 11-47. Cumulative distribution of
computed and observed total suspended
solids in York River.
Figure 11-48. Cumulative distribution of
computed and observed light attenuation
in York River.
-242-
-------
Chapter 11 • Suspended Solids and Light Attenuation
The computed and observed solids distributions in the Rappahannock demon-
strate different shapes (Figure 11-49). The model demonstrates a bulge at the
median that exceeds the observed median by =10 gm m'3. Interpretation of the
bulge is difficult but inspection of the time-series plots indicates computed solids in
the bottom waters of the mid and lower estuary are biased on the high side. The
attenuation distributions (Figure 11-50), which are based solely on surface solids,
agree well. The greatest discrepancy between computed and observed attenuation is
=0.2 m"1 in the lowest 5% of the observations. The upper 75% of the distributions
are virtually coincident.
ir!
i
Final Calibration - SENS 1 36
Total Solids
Rappahannock River
_i__^= . J i_l____ .Ll^^i^l j ; l^i^l I
26 50 74 100
Percent Less than
Final Calibration - SENS 1 36
Light Extinction
1
8
6
ft
4
i. 3
1
F Rappahannock River
:
r
I ,'
'- <
- _...•'''
- ....-•-•""
25 50 75 100
Percent Less* an
Figure 11-49. Cumulative distribution of
computed and observed total suspended
solids in Rappahannock River.
Figure 11-50. Cumulative distribution of
computed and observed light attenuation
in Rappahannock River.
Computed and observed solids in the Potomac River show reasonable agreement
though the lower half of the distribution but computed solids exceed observed in
the upper half (Figure 11-51). Inspection of the time series indicates at least a
portion of the discrepancy is attributable to over-computation of solids concentra-
tions at the bottom of Station RET2.2. Computed and observed attenuation show
good agreement throughout the distributions (Figure 11-52).
The computed and observed solids distributions in the Patuxent demonstrate
different shapes (Figure 11-53). The distributions agree well at the median and at
the upper end but the model demonstrates bulges at the 25th and 75th percentiles.
Interpretation of the bulges is difficult. Inspection of the time- series plots indicates
several breaks in the observations. At Station TF1.7, the observations for the first
two years are higher than observations in the succeeding years. At Station LE2.3,
observations in years 5 and 6 exceed observations in earlier and succeeding years.
While the observations are discontinuous, the model computes relatively constant
values for ten years. The discontinuities in the observations may be the cause of the
different shapes in the computed and observed solids distributions. The computed
and observed attenuation distributions (Figure 11-54) agree well through most of
the range but the minimum attenuation computed by the model exceeds observed
by =0.5 m'1.
-243-
-------
Chapter 11 • Suspended Solids and Light Attenuation
Final Calibration - SENS 1 36
Total Solids
10*
_J
E 1°
10'
Potomac River
Model
• --"""'""
- ,/-"""
-
25 50 .'5 100
Percent Less ti an
13
K
's
8
7
6
5
4
t. 3
41
E z
1
Final Calibration - SENS 1 36
Light Extinction
Potomac River
i
25 SO 75 1*0
Percent Less than
Figure 11-51. Cumulative distribution of
computed and observed total suspended
solids in Potomac River.
Figure 11-52. Cumulative distribution of
computed and observed light attenuation
in Potomac River.
Final Calibration - SENS 136
Total Solids
_»
Patuxent River •
: /
10 25 50 75 1*0
Percent Lessttiin
Figure 11-53. Cumulative distribution of
computed and observed total suspended
solids in Patuxent River.
Final Calibration - SENS 1 36
Light Extinction
25
20
15
to
L. 5
S
a
f
: Patuxent River
'
-
i
_-.--""""'
j'jjjij i j i j . j i i i i j
25 50 76 100
Percent Less ttisn
Figure 11-54. Cumulative distribution of
computed and observed light attenuation
in Patuxent River.
Computed and observed solids distributions in the Eastern Shore minor tribu-
taries (Figure 11-55) agree well at the upper and lower extremes but at the median,
the model underestimates solids concentrations by 3 to 5 g m"3. Inspection of the
time series indicates the model consistently underestimates solids at Stations EE2.1
and ET9.1. Comparisons at these stations may influence the distributions near the
median. The distributions of computed and observed attenuation agree well
throughout the range (Figure 11-56). At the median, the difference between
computed and observed light attenuation is =0.1 m'1
-244-
-------
Chapter 11 • Suspended Solids and Light Attenuation
10'
10'
Final Calibration - SENS 136
Total Solids
Eastern Mrior Stations
Mode!
- .-----"""""""
25 50 75 100
Percent Less than
Final Calibration - SENS 1 36
Light Extinction
13
a
7
6
5
3
1
E Eastern M inor Stations
f i
r
p. i
~
"-
.---"""'"
< , , 1 1 i . ^___i_i i i . j • i j i i
25 50 75 101)
Percent Lessttian
Figure 11-55. Cumulative distribution of
computed and observed total suspended
solids in Eastern Shore minor tributaries.
Figure 11-56. Cumulative distribution of
computed and observed light attenuation
in Eastern Shore minor tributaries.
Computed and observed solids in the Western Shore minor tributaries show
reasonable agreement though the lower three-fourths of the distribution but
computed solids exceed observed in the remainder (Figure 11-57). Inspection of the
time series identifies WT2.1 as a station at which computed solids greatly exceed
observed. Comparison of the attenuation distributions shows a tendency opposite to
the solids (Figure 11-58). The computed and observed distributions show good
agreement in the upper 25% but the observations exceed the computations for most
of the distribution. Computations at stations WT5.1 and WT8.1 appear to have
most influence on the computed distribution relative to observed.
Final Calibration - SENS 1 36
Total Solids
10-
CR
Western Minor Stations j
1
__1 ' -L 1 1 _L 1 ± L 1 1 .1 ± L 1
26 50 75 100
Percent Lessttian
Figure 11-57. Cumulative distribution of
computed and observed total suspended
solids in Western Shore minor tributaries.
w'
1
0*
&
10'
Final Calibration - SENS 1 36
Light Extinction
Western M in or Stations
-
-
"_j L . . J j i . J_i_i_L___i_! ^ _i J 1
26 50 75 100
Percent Lessttian
Figure 11-58. CCumulative distribution of
computed and observed light attenuation
in Western Shore minor tributaries.
-245-
-------
Chapter 11 • Suspended Solids and Light Attenuation
References
Cerco, C., and Moore, K. (2001). "System-wide submerged aquatic vegetation model for
Chesapeake Bay," Estuaries, 24(4), 522-534.
Chesapeake Bay Program, United States Environmental Protection Agency. (2000).
"Submerged aquatic vegetation water quality and habitat-based requirements for restoration
targets: A second technical synthesis," CBP/TRS 245/00, EPA 903-R-00-014, Annapolis
MD.
Holmes, R. (1970). "The Secchi disk in turbid coastal waters," Limnology and Oceanog-
raphy, 15, 688-694.
Hood, R., Wang, H., Purcell, I, Houde, E., and Harding, L. (1999). "Modeling particles and
pelagic organisms in Chesapeake Bay: Convergent features control plankton distributions,"
Journal of Geophysical Research, 104(C1), 1223-1243.
Keefe, C., Flemer, D., and Hamilton, D. (1976). "Seston distribution in the Patuxent River
Estuary," Chesapeake Science, 17, 56-59.
Kirk, J. (1994). Light and photosynthesis in aquatic ecosystems. 2nd edition., Cambridge
University Press, Cambridge
Koenings, J., and Edmundson, J. (1991). "Secchi disk and photometer estimates of light
regimes in Alaskan lakes: Effects of yellow color and turbidity," Limnology and Oceanog-
raphy, 36(1), 91-105.
Pennock, J. (1985). "Chlorophyll distributions in the Delaware Estuary: Regulation by light
limitation," Estuarine, Coastal and Shelf Science, 21, 711- 725.
Schubel, J. (1968). "Turbidity maximum of the northern Chesapeake Bay," Science, 161,
1013-1015.
-246-
-------
Tributary Dissolved 1 J
Oxygen ±M
Introduction
Analysis and modeling of tributary dissolved oxygen has largely focused on
persistent or intermittent anoxia that occurs in bottom waters near the mouths of
four major western tributaries: the York, Rappahannock, Potomac, and Patuxent.
The occurrence of surface dissolved oxygen concentrations substantially below
saturation seems to have escaped notice of the scientific and management commu-
nities. In fact, the tributaries that exhibit bottom-water anoxia commonly exhibit
surface dissolved oxygen less than 5 g m~3 in regions overlying the anoxic bottom
waters (Figures 12-1 to 12-4). Surface dissolved oxygen in the York and Patuxent
falls below 5 g m~3 in multiple locations along the river axes (Figures 12-1, 12-4).
York River Final Calibration - SENS 1 36
10
9
8
7
_l 6
£ 5
4
3
2
1
0
Surface Dissolved Oxygen Summer 1990
-
- •
r\ .
- •. I ^"Xx-\
• ^ * \
- ' .* vL/2--— ~'~x, ,
L ' . X- — ^^
: - \
\
: \
i" . \
r •• \
: , i , , , , i , , , j i
0 50 100
Kilometers
Figure 12-1. Computed and observed surface dissolved oxygen in the
York River, summer 1990. Computed mean and range shown as lines,
observed mean and range shown as solid circles and vertical bars.
-247-
-------
Chapter 12 • Tributary Dissolved Oxygen
12
11
10
9
8
_, 7
| 6
6
4
3
2
1
0
Rappahannock River Final Calibration - SENS 1 36
Surface Dissolved Oxygen Summer 1993
-
-
- • . |
r • • I -A - /^\_,--/ \
- ""^ — ^K/i \ /'i'*'* *
? !' I ^x/7"?^7 t ' '
I ' .I ' ' '
-
i-
-
_
-
0 50 100 150
Kilometers
Figure 12-2. Computed and observed surface dissolved oxygen in
the Rappahannock River, summer 1993. Computed mean and range
shown as lines, observed mean and range shown as solid circles and
vertical bars.
Potomac River Final Calibration - SENS 136
13
12
11
10
9
_l
E
= 6
5
4
3
2
1
0
Surface Dissolved Oxygen Summer 1985
"-
-
-
I- • / ^-^'^
^ ^~^^^^^y . *^' f\
i » » ' V"*
: f
:. . • *
- t !>
-_
'r
'?
^
: 1 , , , , 1 , , , , 1 , , r , i
0 50 100 150
Kilometers
Figure 12-3. Computed and observed surface dissolved oxygen in
the Potomac River, summer 1985. Computed mean and range shown
as lines, observed mean and range shown as solid circles and
vertical bars.
-248-
-------
Chapter 12 • Tributary Dissolved Oxygen
Patuxent River Final Calibration - SENS 1 36
11
10
9
8
7
•J
"63 6
E
5
4
3
2
1
A
Surface Dissolved Oxygen Summer 1985
-
-
y— V ' *
- \^ / \ ' *
- * "•\K i "^V \
: ' • T I ' \
~ T * ~'\
* » !~ \
" \ •
: \
- " \
\
\ ' ,
\ <
• \ /
- \ /
: \ /
- ^\ /
'"- , I , , r , I , i , , I r . , , 1 , ,
0 20 40 60
Kilometers
Figure 12-4. Computed and observed surface dissolved oxygen in
the Patuxent River, summer 1985. Computed mean and range
shown as lines, observed mean and range shown as solid circles
and vertical bars.
A tendency present in the model since the earliest application is the over-predic-
tion of surface dissolved oxygen in the tributaries, especially in the tidal freshwater
portions (Figures 12-5, 12-6). This performance characteristic was overlooked when
attention was focused on bottom waters. During the present study, sponsors noted the
discrepancies between computed and observed surface dissolved oxygen concentra-
tions and asked for improved model performance. An extensive number of calibration
and sensitivity runs were performed while attempting to improve the model. The
present chapter documents a portion of these runs. As with other sensitivity compar-
isons, the model runs are not always sequential and may incorporate multiple
changes. Still, the comparisons provide insight into model sensitivities and illustrate
plausible and implausible mechanisms for observed and computed properties.
O
16.0-
14.0-
12.0-
10.0-
8.0-
6.0-
4.0-
2.0-
0.0
150 125
100
75
r
50
25
Kilometers from Mouth
Figure 12-5.
Computed and
observed surface
dissolved oxygen in
the James River,
summer 1986, from
the original 4,000
cell model (Cerco
and Cole 1993).
Computed mean and
minimum shown as
lines, observed
mean and range
shown as circles
and vertical bars.
-249-
-------
Chapter 12 • Tributary Dissolved Oxygen
K)
*
*
0
16,0-1
14.0-
12. fl-
IC. 0-
8.0-
6.0-
4.0-
2.0-
0.0-
^^^-^^
JulH f>Ti- •••"•• °- ••••••
Ifjf fr T I T ^-..-^ •-.. . ..i
75 150 125 100 75 50 25 0
Kilometers from Mouth
Figure 12-6.
Computed and
observed surface
dissolved oxygen in
the Potomac River,
summer 1986, from
the original 4,000 cell
model (Cerco and Cole
1993). Computed mean
and minimum shown as
lines, observed mean
and range shown as
circles and vertical
bars.
Dissolved Organic Carbon Mineralization Rate
The dissolved organic carbon mineralization (or respiration) rate is a key deter-
minant of oxygen consumption in the water column. As a sensitivity test, we
multiplied the respiration rate by an order of magnitude, from 0.01 d"1 to 0.1 d"1 at
20 °C. Major reaches of the tributaries were insensitive to an order-of- magnitude
increase in the first-order rate (Figures 12-7, 12-8). Only reaches in the immediate
vicinity of a carbon source, e.g. the Potomac below the District of Columbia,
showed significant response to an increase in respiration rate. Lack of general
response indicated that oxygen consumption could not be increased without an
increase in oxygen-demanding material. An increase in first-order rate alone would
not suffice. In addition, the rate increase produced unacceptably low computations
of total organic carbon (Figures 12-9, 12-10).
RAPPAHANNOCK RIVER, SENS75 (Cruise) ON NEW
Dissolved Oxygen
DO Rap Surface
Julian Day 2086
SENS 75
KDC = 0.01
iHANNOCK RIVER, SENS74 (Cruise) ON NEW
'Sd Oxygon
3 Surface . ••"'•
Jy«an Day 20§S
Figure 12-7. Sensitivity of Rappahannock River surface dissolved oxygen
to dissolved organic carbon mineralization rate, summer 1990.
-250-
-------
Chapter 12 • Tributary Dissolved Oxygen
11
e-
&
d 6
2
0
POTOMAC RIVER, SENS75 (Cruise) ON NEW GRID
Dissolved Oxygen
•m DO Potomac Surface
Julian Day 2066
'" • """ """-,
~ *
* # »
y &o TOO
KEtom^ters
SENS 74 —+
KDC = 0.1
9
t "
4
S
0
KDC = 0.01
VER, SENS74 (Cruise) ON NEW GRID
igen
Syrffac©
J6S
. . _.. .. - ' -x
\J
* " *
.
-
Kilometers
Figure 12-8. Sensitivity of Potomac River surface dissolved oxygen to
dissolved organic carbon mineralization rate, summer 1990.
RAPPAHANNOCK RIVER, SENS75 (Cruise) ON NEW
Total Organic Carbon
TOC Rap Surface
Jyfen Day 2086
SENS 75
KDC = 0.01
.PPAHANNOCK RIVER, SENS74 (Crylse) ON NEW
yl Opgaiilc Carbon
. *jC Rrfp Surface
,|u«an Day 2088
Figure 12-9. Sensitivity of Rappahannock River surface total organic carbon
to dissolved organic carbon mineralization rate, summer 1990.
-251 -
-------
Chapter 12 • Tributary Dissolved Oxygen
1
I
CR
E
POTOMAC RIVER, SENS75 (Cruise) ON NEW GRID
Tola! Organic Carbon
TOC Potomac Surface
Julian Day 2068
* * !i
0 30 1(30 150
KllOfneters
SENS 75
KDC = 0.01
VC RIVER, SENS74 (Cruise) ON NEW GRID
janic Carbon
omae Surface
iy
SENS 74
KDC = 0.
Figure 12-10. Sensitivity of Potomac River surface total organic carbon to
dissolved organic carbon mineralization rate, summer 1990.
Our conclusion from this run was that computation of high surface dissolved
oxygen could not be corrected with an increase in respiration rate. The run did
suggest, however, that surface dissolved oxygen could be depressed locally by
altering the respiration rate.
Wetland Dissolved Oxygen Uptake
The sensitivity of surface dissolved oxygen to wetland oxygen uptake was exam-
ined on a time-series basis in the chapter entitled "Hydrology and Loads." When
wetlands impact is examined on a spatial scale, in the York River for example, we
see that the effect is confined to the upper reaches of the estuary (Figure 12-11).
Wetlands impact is greatest where the estuary is narrow and the ratio of wetlands
area to water area is large. In the wide lower estuary, fringing wetlands can do little
to affect dissolved oxygen in open water. In the York, for example, the low surface
dissolved oxygen 20 km above the mouth cannot be due to wetland uptake.
Nonpoint-Source Carbon Loads
Carbon loads to the tributaries are not well-known. Organic carbon is not a state
variable in the watershed model nor is it monitored at the Virginia tributary fall
lines. In the Potomac and Patuxent, total organic carbon measures are collected but
no information is available to partition the total into particulate and dissolved frac-
tions. For our model, we obtained nonpoint-source organic carbon loads by ratio to
organic nitrogen computed by the watershed model (see the chapter "Hydrology
and Loads"). We multiplied carbon loads by a factor often to examine sensitivity in
the face of uncertainty. While the order-of-magnitude load increase certainly
-252-
-------
Chapter 12 • Tributary Dissolved Oxygen
YORK RIVER, SENS110 (Cruise)
Dissolved Oxygen
DO To* Surfaca ,
Julian Day 20iS
SENS 110
•WETLANDS
SENS 104 i
NO WETLANDS
RIVER, SENS104 (Cruise) ON NEW GRID
tfed Oxygen
- rk Surface
Day 2066 '•• - • --.
Figure 12-11. Sensitivity of York River surface dissolved oxygen to wetlands
oxygen uptake, summer 1990.
improved computed surface dissolved oxygen, even this increase could not account
for the low dissolved oxygen occurring at the surface of the lower Potomac and
James Rivers (Figures 12-12, 12-13). Moreover, the computed organic carbon
increased to unrealistic levels (Figures 12-14, 12-15). Our conclusion is that the
modeler has freedom to adjust loads within reasonable levels but uncertainty in the
loads can not account for the over-computation of surface dissolved oxygen.
POTOMAC RIVER, SENS114 (Cruise)
Dissolved Oxygw
OO Potomac Surface
Julian Day 3153
SENS 114
NFS LOADS* 10
IAC RIVER, SENS1 12 (Cruise)
SENS112
BASE
ornac Surface
lay 3153
Figure 12-12. Sensitivity of Potomac River surface dissolved oxygen to
nonpoint-source organic carbon load, summer 1990.
-253-
-------
Chapter 12 • Tributary Dissolved Oxygen
•t
"<»
E •*
0
JAMES RIVER, SENS114 (Cruise)
Dissolved Oxygen
DO James Surface
Julian Dai/
-• *
-
-
o so mo iM
SENS 112 — *»*
BASE
10
ft
ft
_j /
f 6
4
1
^— SENS114
NPS LOADS* 10
, SENS112 (Cruise)
face
66
-
..--•s ,..- --.
- - -"*"" • •-. • ^ .,^
=-. 4
* * + * " * !
~
-
*' 'Kilometsrs
Figure 12-13. Sensitivity of James River surface dissolved oxygen to
nonpoint-source organic carbon load, summer 1990.
POTOMAC RIVER, SENS114 {Cruise)
Total Organic Cartnon
2':
yo
"5) 15
E
10
s
TOC Potomac Surface
Julian Day 3153 / '.
~
'__ /
- _.....--
* * I
* » »
1 j ,!,!•, 1
U 0 50 100 150
SENS 11 2 3-
BASE
=
E j
2
1
^^ SENS 114
NPS LOADS* 10
'arbor"8112'0™'56'
Surface /';
53
_---" • ~ * "--. ./ * * \
''_ >•''•"""" * | \
* * *
•
-
° 0 50 100 150
Kilometers
Figure 12-14. Sensitivity of Potomac River surface total organic carbon to
nonpoint-source organic carbon load, summer 1990.
-254-
-------
Chapter 12 • Tributary Dissolved Oxygen
40
»
f 20
10
5.
0
JAMES RIVER, SENS114 (Cruise)
Total Organic Carbon
TOC James Surface A
Julian Day
-
_„,.-'""
';•''.' .... * * ' • •
0 50
Kilometers
oEH^m t-
BASE
100
I* 4
3
0
150
MPS LOADS* 10
SENS112 (Cruise)
arbon
"face
>6 ,,-'''
_
,-"'' * *
~m / * * * *
r * * I
-
o so KID 150
Kilometers
Figure 12-15. Sensitivity of James River surface total organic carbon to
nonpoint-source organic carbon load, summer 1990.
Spatially-Varying Dissolved Organic Carbon
Mineralization Rate
The initial sensitivity examination of dissolved organic carbon mineralization
rate indicated this parameter could affect dissolved oxygen locally but not globally.
In an attempt to improve surface dissolved oxygen computations in the tidal fresh
portions of the tributaries, we imposed spatially-varying mineralization rates. The
first-order rate was increased from a global value of 0.011 d"1 at 20 °C to 0.1 d"1 in
tidal fresh water (TF segments); to 0.05 d"1 in the river-estuarine transition regions
(RET segments); and to 0.025 d"1 in the lower estuaries (LE segments). All other
portions of the system were left at their original value.
This test indicated computed surface dissolved oxygen could be lowered by 1 to
2 g m~3 in the upper portions of the James River (Figure 12-16) while improving
total organic carbon computations (Figure 12-17). Similar dissolved oxygen
improvements were obtained in the upper 50 km of the Potomac (Figure 12-18)
although total organic carbon computations suffered (Figure 12-19).
This sensitivity run guided assignment of the dissolved organic carbon mineral-
ization rate in the final calibration. We assigned a rate of 0.011 d"1 at 20 °C
everywhere except in the tidal fresh portions of the James (0.075 d"1), Rappahan-
nock (0.05 d"1) and Potomac (0.05 d"1). We reasoned that the higher values were
appropriate due to the input of organic matter from point sources, combined sewer
overflows, and urban runoff near the heads of these three estuaries.
-255-
-------
Chapter 12 • Tributary Dissolved Oxygen
JAMES RIVER, SENS115 (Cruise)
Dissolved Oxygen
DO James Surface
J i' - r ' 2 '
SENS 115
MULTIPLY KDC
BY 5 TO 10
R, SENS112 (Cruise)
tgen
rfaee /'"\
366
SENS 112
BASE
Figure 12-16. Sensitivity of James River surface dissolved oxygen to
spatially-varying dissolved organic carbon mineralization rate, summer 1990.
JAMES RIVER, SENS115 (Cruise)
Total Organic Carbon
TOG James Surface
Julian Day 2066
SENS 115
MULTIPLY KDC
BY 5 TO 10
R, SENS112 (Cruise)
: Carbon
Surface
:06i ,.--'''
SENS 112
BASE
Figure 12-17. Sensitivity of James River surface total organic carbon to
spatially-varying dissolved organic carbon mineralization rate, summer 1990.
-256-
-------
Chapter 12 • Tributary Dissolved Oxygen
POTOMAC RIVER, SENS11 5 (Cruise)
Dissolved Oxygen
10
a
_• '<
la*
E 5
•3
0
I DO Potomac Surface
- Julian Day 3153
<=i SENS 115
L * * . MULTIPLY KDC
BY 5 TO 10
r
-
r OMAC RIVER, SENS1 12 (Cruise)
o t.^ njo IMS ofved Oxygen
Kllom»lers Potomac Surface
SENS 112 c=>
BASE
i u • ' , S3
# * *
*
e s -
4 -
0 50 1LKJ 150
Kilometers
Figure 12-18 Sensitivity of Potomac River surface dissolved oxygen to
spatially-varying dissolved organic carbon mineralization rate, summer 1990.
1 "
________________
Total Organic Carbon
TOC Potomac Surface
Julian Day 3153
SENS 115
MULTIPLY KDC
BY 5 TO 10
: RIVER, SENS112(Cnufee)
nic Carbon
nac Surface A
3153 / I
SENS 112
BASE
Figure 12-19. Sensitivity of Potomac River surface total organic carbon to
spatially-varying dissolved organic carbon mineralization rate, summer 1990.
-257 -
-------
Chapter 12 • Tributary Dissolved Oxygen
Reaeration Rate
The reaeration rate may be viewed as the speed with which the surface dissolved
oxygen concentration achieves equilibrium with the atmosphere. Adjustment in the
reaeration rate is useful as a tool to reduce computed surface dissolved oxygen only
when dissolved oxygen is below equilibrium concentration. In that case, reduced
reaeration will retard the return of surface dissolved oxygen to equilibrium. If
computed surface dissolved oxygen exceeds equilibrium, then a reduction in
reaeration rate will exacerbate any over-computation by retarding the escape of
excess oxygen to the atmosphere. Since the sensitivity run with spatially-varying
dissolved organic carbon mineralization rate indicated the possibility of pulling
surface dissolved oxygen below saturation, we decide to combine that run with a
reduction in reaeration rate.
For Chesapeake Bay, a relationship for wind-driven gas exchange (Hartman and
Hammond 1985) was employed:
Kr = Arear • Rv • Wms15 (12-1)
in which:
Kr = reaeration coefficient (m d"1)
Arear = empirical constant (0.1)
Rv = ratio of kinematic viscosity of pure water at 20 °C to kinematic viscosity of
water at specified temperature and salinity
Wms = wind speed measured at 10 m above water surface (m s"1)
Hartman and Hammond (1985) indicate Arear takes the value 0.157. We halved
that value for the sensitivity run. Results indicated halving reaeration could depress
surface dissolved oxygen by up to 1 g m"3 in the upper Potomac, where dissolved
oxygen was below saturation (Figure 12-20). At other locations in the Potomac,
dissolved oxygen increased slightly. The results were similar in other estuaries.
Halving reaeration decreased dissolved oxygen in the upper York, where dissolved
oxygen was drawn down by wetlands (Figure 12-21). Elsewhere in the York, effects
were unnoticeable or else dissolved oxygen increased slightly.
Upon assessing the results of these and other sensitivity runs, we opted to retain
the value of Arear = 0.078. Use of this value is readily justified. Fetch in the estu-
aries is reduced by embankments and channel sinuosity. Consequently, reaeration is
expected to be less than over open water.
Sensitivity to G3 Carbon
Organic matter deposited to the sediments is routed into three G fractions. The
G3 fraction is inert and does not contribute to diagenetic processes including sedi-
ment oxygen demand. To examine the sensitivity of tributary surface dissolved
oxygen to the specification of the inert fraction, we set the G3 fraction to zero. For
phytoplankton carbon, the 5% of organic matter previously routed to G3 was
instead routed to G2 (refractory) making a total of 35% refractory. For the refrac-
tory particulate organic carbon state variable, the 27% previously routed to G3 was
routed to G2. As a result of these revisions, all deposited organic carbon
contributed to sediment oxygen demand at a fast (labile) or slow (refractory) rate.
-258-
-------
Chapter 12 • Tributary Dissolved Oxygen
11
10
§
1» &
5
3
POTOMAC RIVER, SENS1 16 (Cruise)
DissG'livecl Oxygen
DO Potomac Surface
Julian Day
""•-......- ,/"""" '"••.
* \,. /
• * \..S
*
0 ^0 1CK.1 1SO
Kilomolors
SENS 115 ^™
MULTIPLY KDC
BY 5 TO 10
SENS116
WITH REAERATION
HALVED
»AC RIVER, SENS115 (Cruise)
•0cj Oxygen
:omac Surface
l^y 20BB
Figure 12-20. Sensitivity of Potomac River surface dissolved oxygen to reaeration
rate, summer 1990.
YORK RIVER, SENS116 (Cruise!
Dissolved O'sygert
DO York Surface ••
Julian Day 2066
MULTIPLY KDC
BY 5 TO 10
SENS 116
WTH REAERATION
HALVED
IRK RIVER, SENS115(Crulse)
wo3v&d Oxygen
—I York Surface
Julian Da»> 2066 '"••-... . _.
Figure 12-21. Sensitivity of York River surface dissolved oxygen to reaeration rate,
summer 1990.
-259-
-------
Chapter 12 • Tributary Dissolved Oxygen
The effect of G3 specification on surface dissolved oxygen was barely discern-
able (Figure 12-22). At best 0.1 to 0.2 g DO m~3 additional could be consumed by
eliminated the G3 fraction completely.
JAMES RIVER, SENS118 (Ctuise)
Dissolved Oxygen
DO James Surface
Julian Day
SENS 118
NO G3 CARBON
IVER, SENS112 (Cruise)
Oxygen
s Surface ,-'"
/ 2066
SENS 112
BASE
Figure 12-22. Sensitivity of James River surface dissolved oxygen to specification
of G3 organic fraction, summer 1990.
Sensitivity to Algal Predation Rate
The source of excess dissolved oxygen in the tidal fresh portions of the tribu-
taries is algal production. During the calibration process, we altered the predation
rate on algae from BPR = 0.4 (SENS 120) to BPR = 0.5 (SENS 122) in the tidal
fresh James River. The additional predation dropped computed chlorophyll at km
115 by ~6 mg m~3 (Figure 12-23) and reduced computed dissolved oxygen at the
same location by =1 g m~3 (Figure 12-24). Computed chlorophyll moved away
from the observed mean while computed dissolved oxygen moved closer to the
observed mean. In the lower estuary, we increased predation from BPR =
0.9 (SENS 120) to BPR = 1.5 (SENS 122). At km 50, computed chlorophyll
dropped by 6 mg m~3 and computed surface dissolved oxygen dropped by
= 1 g m~3. In this case, both computed chlorophyll and dissolved oxygen moved
closer to observed means.
This run illustrated the sensitivity of surface dissolved oxygen to algal produc-
tion. This run also illustrates the dilemma faced by the modeler when trying to
obtain optimal calibration of multiple variables.
-260-
-------
Chapter 12 • Tributary Dissolved Oxygen
_j 30
I
2-3
JAMES RIVER, SENS120 (Cruise)
Chlorophyll
Chlorophyll James Surface
Julian Day 2068
^^™ SENS 120
BASE PREDATION RATE
SENS122 (Cruise)
es Surface
SENS 122 ^~*
ALTERED
PREDATION RATE
Figure 12-23. Sensitivity of James River chlorophyll to specification of base preda-
tion rate, summer 1990.
JAMES RIVER, SENS120 (Cruise)
Dissolved Oxygen
DO James Surface ./""'••
Julian Day 2066
1^™ SENS 120
BASE PREDATION BATE
i/ER, SENS122 (Cruise)
Oxygen
Surfdce •-""'•
SENS 122 ^™
ALTERED
PREDATION RATE
Figure 12-24. Sensitivity of James River surface dissolved oxygen to specification
of base predation rate, summer 1990.
-261 -
-------
Chapter 12 • Tributary Dissolved Oxygen
Sensitivity to Vertical Diffusion Coefficient
In the lower portions of the estuaries, the occurrence of depressed surface
dissolved oxygen immediately over the location of bottom-water anoxia suggests
the transfer of oxygen demand or of anoxic water from bottom to surface. The
vertical diffusion coefficient is computed in the hydrodynamic model and is not a
"tuning" parameter available to the water quality modeler. Still, we thought exam-
ining the effect of increased diffusion would be illustrative so we performed a run
with vertical diffusion doubled. This doubling was performed in the water quality
model code and did not feed back upon computed circulation.
The results of this run were mixed. In the lower James, bottom dissolved oxygen
increased by 0.5 to 1 g m~3 (Figure 12-25) while surface dissolved oxygen
decreased by a lesser amount (Figure 12-26). In both cases, the changes brought
computed dissolved oxygen closer to observed. We previously noted (see chapter
entitled "Coupling with the Hydrodynamic Model") our preference for greater
vertical diffusion in the James. Still, bottom-water dissolved oxygen was under-
computed even with the higher diffusion.
In the York, bottom-water dissolved oxygen increased by up to 1 g m"3, arguably
an improvement (Figure 12-27), while surface dissolved oxygen barely moved
(Figure 12-28). Geometry appears to play a role in this result. The deep channel in
the York is narrow while the surface is wide. The volume of anoxic water mixed up
from the bottom is insufficient to affect the surface dissolved oxygen concentration.
James River Final Calibration - SENS 136
Bottom Dissolved Oxygen Symmer 1990
SENS 136
BASE
sr SENS 151
solved Oxygen Summ»r 1990
SENS 151 E==
VERTICAL DIFFUSION
DOUBLED
Figure 12-25. Sensitivity of James River bottom dissolved oxygen to vertical
diffusion coefficient, summer 1990.
-262-
-------
Chapter 12 • Tributary Dissolved Oxygen
James River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1990
SENS 151 f==Ct
VERTICAL DIFFUSION
DOUBLED
SENS 136
BASE
SENS 151
olved Oxygen Summer 1990
Figure 12-26. Sensitivity of James River surface dissolved oxygen to vertical
diffusion coefficient, summer 1990.
York R ivcr F in a I C alibration - S E N S 136
Bottom Dissolved Oxygen Summer 1990
SENS 151 fc
VERTICAL DIFFUSION
DOUBLED
I
SENS 136
•rSENS 151
issolved Oxygen Summer 1990
Figure 12-27. Sensitivity of York River bottom dissolved oxygen to vertical
diffusion coefficient, summer 1990.
-263-
-------
Chapter 12 • Tributary Dissolved Oxygen
B
i:
2
0
York River Final Calibration - SENS 1 36
Surface Dis$ofc/ed Oxygen Summer 1990
-
* ' -..I ...... J •• • . . .
i.
Kilometers
SENS1M _|»
VERTICAL DIFFUSION
DOUBLED
;ti
»
.i
«— SENS136
BASE
SENS 151
&un&ce uosolved Oxygen Summer 1990
I • - ....
-
KUometers
Figure 12-28. Sensitivity of York River surface dissolved oxygen to vertical
diffusion coefficient, summer 1990.
In the Potomac, computed bottom-water dissolved oxygen increased by up to
1 g m~3 (Figure 12-29). At two stations, kms. 0 to 30, agreement with observed
means diminished while at a third station (km 80) agreement improved. In the
11
10
8
w 6
5
-1
2
1
0
Potomac River Final Calibration - SENS 1 36
Bottom Dissolved Oxygen Summer 1990
-
•
/ f f
' * **
: "!' * i
9 "••...••'
0 50 100 i&O
Kilometers
SENS 151 - ^
VERTICAL DIFFUSION
DOUBLED
ES *5
E
5
,s
2
i
0
<— • SENS 136
BASE
liver SENS 151
tsolved Oxygen Summer 1990
,-.
I ' *****
- '^. * /
r "V X-J/
. *
0 50 100 150
Kiksm^ters
Figure 12-29. Sensitivity of Potomac River bottom dissolved oxygen to vertical
diffusion coefficient, summer 1990.
-264-
-------
Chapter 12 • Tributary Dissolved Oxygen
surface waters, computed dissolved oxygen decreased by up to 1 g m~3 resulting in
improved agreement with observed means from kms. 0 to 30 (Figure 12-30). At km
80, surface dissolved oxygen barely moved despite the increase in bottom dissolved
oxygen. As with the York, we believe geometry influenced the results at this loca-
tion. The volume of anoxic water at the bottom is insufficient to significantly affect
the surface through mixing.
These runs indicate that exchange between the surface and bottom can influence
surface dissolved oxygen but that no simple, universal "fix" such as changing
vertical diffusion exists.
Potomac River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1990
SENS 136
BASE
itomac River SENS 151
irface Dissolved Oxygen Summer 1990
SENS 151
VERTICAL DIFFUSION
DOUBLED
t
Figure 12-30. Sensitivity of Potomac River surface dissolved oxygen to vertical
diffusion coefficient, summer 1990.
Discussion
Two phenomenon vex the computation of dissolved oxygen in the western tribu-
taries. The first is the computation of excess dissolved oxygen in the tidal fresh
portions of the James (Figures 12-31 to 12-36) and Potomac Rivers (Figures 12-49
to 12-54). The excess dissolved oxygen is the result of an excess of computed
production over consumption. Computed production can be readily adjusted but a
constraint exists to provide reasonable agreement with observed chlorophyll
concentrations. We believe the problem lies more on the side of consumption.
Riverine organic carbon loads to the Virginia tributaries are virtually unknown as
are point-source carbon loads to all tributaries. In addition, the James and Potomac
receive loads from combined-sewer overflows and urban runoff. Improved
dissolved oxygen computations require improved information on loading. Ideally,
measures of respiration or of biochemical oxygen demand in the water column
should also be conducted.
-265-
-------
Chapter 12 • Tributary Dissolved Oxygen
James River Final Calibration - SENS 1 36
Surface Dissolved Oxygen Summer 1985
18
16
14
12
—l 10
D>
E
8
6
4
2
c\
-
-
-
/' N\
' /
' " / • -,. .S \ ^
•*_, * - / " I \
; " "" "~ "~ — •-•-.-. v'~ ---..„„.• ••""" * ~" * * i LV-
* " * " '•
-
-
0 50 100 150
Kilometers
James River Final Calibration - SENS 136
Bottom Dissolved Oxygen Summer 1985
18
16
14
12
=d 10
E
8
6
4
2
r\
-
r
-
-
/ \s t
- f \
/'' "^' ' \
^ :A... * ' * */ ' *
- " '~'"\, ._,/"""'
-
: t , , . , i
0 50 100 150
Kilometers
Figure 12-31. Computed and observed surface dissolved
oxygen in the James River, summer 1985. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-32. Computed and observed bottom dissolved
oxygen in the James River, summer 1985. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
|
0)
E
James River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1990
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
n
r
j-
-
r
r
=•
'--
•• J/ \
=• * / " v i •
: " - V^.,. ,., , / . \
-* - ^ ,- --•' i\ ,
: « ••"' a * T ^",./
- • • * 1
-
i
'•-
'-
'.
- i I < , i i I ( , , i I i , < , I
0 50 100 150
Kilometers
j
0,
E
James River Final Calibration - SENS 136
Bottom Dissolved Oxygen Summer 1990
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
r
-
-
-
-
-
~~ -'"" \
\. / \
1 '"'"*' '' \
L - - - / ^ • X\A /
k . * ' * '/
!••*""/
- ~ , J
'r v'--,--'
'--
-
'- , 1 , , , , I , t , , 1 , , , , 1
0 50 100 150
Kilometers
Figure 12-33. Computed and observed surface dissolved
oxygen in the James River, summer 1990. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-34. Computed and observed bottom dissolved
oxygen in the James River, summer 1990. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
-266-
-------
Chapter 12 • Tributary Dissolved Oxygen
James River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1993
15
14
13
12
11
10
9
=! 8
E 7
6
5
4
3
2
1
n
-
-
-
-
r " / \
Y ^-x ... / I \ •
s • "N -X/Y ' . • • • K
'.' • * * *
-
-
'-
-
-
= 1 1 , , 1 , 1 r , , I 1 , , , , 1
0 50 100 150
Kilometers
15
14
13
12
11
10
9
=! 8
E 7
6
5
4
3
2
1
Q
James River Final Calibration - SENS 136
Bottom Dissolved Oxygen Summer 1993
-
-
-
r
" / \
~~ • \
'- ' "V \
'-. t \
r' ' ' / ' ' ' I v
'• ^-.. • - -
r V """ ""- s /
i
-
i.
: , 1 , , , , 1 s s , , 1 , , , , 1
0 50 100 150
Kilometers
Figure 12-35. Computed and observed surface dissolved
oxygen in the James River, summer 1993. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-36. Computed and observed bottom dissolved
oxygen in the James River, summer 1993. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
The second vexing process is the occurrence of depressed surface dissolved
oxygen in the lower estuaries, notably the York circa km 25 (Figures 12-37 to 12-
42), the Rappahannock at km 40 (Figures 12-43 to 12-48), and the Potomac circa
kms 30 to 80 (Figures 12-49 to 12-54). Our best explanation of the phenomenon is
transfer of oxygen demand and/or oxygen-depleted water from the bottom to the
surface. The phenomena cannot be represented by simple adjustments vertical
mixing, however. At several locations, the volume of anoxic bottom water is
insufficient to significantly alter the surface concentration when mixed with surface
water. Perhaps a larger volume of anoxic water intrudes from adjacent Chesapeake
Bay. The phenomenon requires additional study and may be beyond modeling
without process-based field observations.
York River Final Calibration - SENS 1 36
Surface Dissolved Oxygen Summer 1985
8
7
6
5
I 4
3
2
1
0
-\.
"Y,
*
* \ * -•''' \
\
\
: \
i , , , , i , , , , i ,
0 50 100
Kilometers
York River Final Calibration - SENS 1 36
Bottom Dissolved Oxygen Summer 1985
8
1
6
5
1
3
1
1
A
-
:
"\ *
• \ „ " • • * * ,"-y/ \ YA
\ ,-A/'"-\J x-; \ / \
V " \
\
:
0 50 100
Kilometers
Figure 12-37. Computed and observed surface dissolved
oxygen in the York River, summer 1985. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-38. Computed and observed bottom
dissolved oxygen in the York River, summer 1985.
Computed mean and range shown as lines, observed
mean and range shown as solid circles and vertical bars.
-267 -
-------
Chapter 12 • Tributary Dissolved Oxygen
York River Final Calibration - SENS 1 36
10
9
8
7
_l 6
E 5
4
3
2
1
0
Surface Dissolved Oxygen Summer 1990
-
-
- \,...., .
- • N---x
- * \
- ' "'-.., A- '" x.. .
: - * \
' \
- \
-
I
0 50 100
Kilometers
York River Final Calibration -SENS 136
8
7
6
5
E 4
3
2
1
Q
Bottom Dissolved Oxygen Summer 1990
-
-
-
- \ . t / '\,. v *
- \ J . • _ X \r \
: ' "\ ^, -( "V ' \
• \ I V~X;' \
-
: ,
0 50 100
Kilometers
Figure 12-39. Computed and observed surface dissolved
oxygen in the York River, summer 1990. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-40. Computed and observed bottom dissolved
oxygen in the York River, summer 1990. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
York River Final Calibration - SENS 136
9
8
7
6
| j-
E"
4
3
2
1
0
Surface Dissolved Oxygen Summer 1993
™ '''•-, I
: ' ~T -\
* - v'\
! \
- • * • \
" ';
;
\
-
1 I , , I 1 i < , , t
0 50 100
Kilometers
York River Final Calibration - SENS 136
7
6
5
-i 4
0,
E
3
2
1
Q
Bottom Dissolved Oxygen Summer 1993
:
-^ _ : /-^ , x
- \ . rfy V
- \ . -• • i
\ I * ; *
\ I • • ,' I
- ' '''• / '
''> '"\ i ^
• \j "\ /' -,.,._/ " i
i \ * / ^
: j \i\J"'~--j '
-
i , , , , i , , , , i
0 50 100
Kilometers
Figure 12-41. Computed and observed surface dissolved
oxygen in the York River, summer 1993. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-42. Computed and observed bottom dissolved
oxygen in the York River, summer 1993. Computed mean
and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
-268-
-------
Chapter 12 • Tributary Dissolved Oxygen
Rappahannock River Final Calibration - SENS 1 36
Surface Dissolved Oxygen Summer 1985
13
12
11
10
9
8
^J 7
E 6
5
4
3
2
1
A
-
-
r
:-
- •• /"S, / \
- ~- -^ ,,-x / i\ ^./
~~ * ' T • [N'x- -'j/' I
- . T x--''""*
r
-
-
-
-
i , 1 , , , i E i , , , 1 , i , , E
0 50 100 150
Kilometers
Rappahannock River Final Calibration - SENS 1 36
11
10
9
8
7
_!
"3> ^
E
5
4
3
2
1
A
Bottom Dissolved Oxygen Summer 1985
-
-
- r, /A
/- •' ! \ / V'y
/v i\ ,v
: i \ ""' \>
'- / 'i
- . i ¥ -'
\ •"•• T
rV/'' \ "
-
:. *
- " 'T":
r
0 50 100 150
Kilometers
Figure 12-43. Computed and observed surface dissolved
oxygen in the Rappahannock River, summer 1985.
Computed mean and range shown as lines, observed mean
and range shown as solid circles and vertical bars.
Figure 12-44. Computed and observed bottom dissolved
oxygen in the Rappahannock River, summer 1985.
Computed mean and range shown as lines, observed
mean and range shown as solid circles and vertical bars.
Rappahannock River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1990
16
15
14
13
12
11
10
_, 9
1 8
7
6
5
4
3
2
1
A
-
I-
-
-
i-
i
"
\- -., . / \ / \
- ' •" " >'"' "\
!• ' I ,"\ A
r I 1 '^
'- 1 « 1
-
-
:-
'- 1 1 1 !
0 50 100 150
Kilometers
Rappahannock River Final Calibration - SENS 1 36
Bottom Dissolved Oxygen Summer 1990
14
13
12
11
10
9
j 8
E 7
6
5
4
3
2
1
A
-
:-
-
-
:-
:-
'-
-_
'-
'- '
k /\« *
: \,x/ v^ *
r » / '""•"''
- ' *
- .-"v
i , 1 , , , i , 1 i , s , 1 i , , , 1
0 50 100 150
Kilometers
Figure 12-45. Computed and observed surface dissolved
oxygen in the Rappahannock River, summer 1990.
Computed mean and range shown as lines, observed
mean and range shown as solid circles and vertical bars.
Figure 12-46. Computed and observed bottom dissolved
oxygen in the Rappahannock River, summer 1990.
Computed mean and range shown as lines, observed mean
and range shown as solid circles and vertical bars.
-269-
-------
Chapter 12 • Tributary Dissolved Oxygen
-J
ra
E
Rappahannock River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1993
12
11
10
9
8
7
6
5
4
3
2
1
-
-
-
- /~\ " / \
' "* " ' \ / i ' ' •
• * , x\/~r '
\ i • i -
• t
-
-
-
-
-
; , t , l , , i , , , , i , , , , i
0 50 100 150
Kilometers
1
0,
E
Rappahannock River Final Calibration - SENS 136
Bottom Dissolved Oxygen Summer 1993
12
11
10
9
8
7
6
5
4
3
2
1
A
r
-
-
" / 7 • * *
- /"'"''' '' *
- I ' ^
- " I '
~ V "'"A /
- \ - /
- \..k /
- . ,\l,x/X
0 50 100 150
Kilometers
Figure 12-47. Computed and observed surface dissolved
oxygen in the Rappahannock River, summer 1993.
Computed mean and range shown as lines, observed mean
and range shown as solid circles and vertical bars.
Figure 12-48. Computed and observed bottom dissolved
oxygen in the Rappahannock River, summer 1993.
Computed mean and range shown as lines, observed
mean and range shown as solid circles and vertical bars.
Potomac River Final Calibration - SENS 136
13
12
11
10
9
8
B> 7
E 6
5
4
3
2
1
0
Surface Dissolved Oxygen Summer 1985
-
-
-
- - /' V
1 „ -. / '" i\
i> / " ~ ^ / ^ V ,'"\
i • • V*
- * '
=•
-
-
• I : I , I 1 < , : , 1 , : , I 1
0 50 100 150
Kilometers
Potomac River Final Calibration - SENS 136
13
12
11
10
9
8
i
6
5
4
3
2
1
Q
Bottom Dissolved Oxygen Summer 1985
-
-
-
-
'- / \
'• / " * \ f\.
- / v --'
: •'' *
- / ' . -
K J
~ \ * /
- ""\x ,.. .^ /
I * V " ^--/
- I , , , , i , , -, , i , , , , i
0 50 100 150
Kilometers
Figure 12-49. Computed and observed surface dissolved
oxygen in the Rappahannock River, summer 1993.
Computed mean and range shown as lines, observed mean
and range shown as solid circles and vertical bars.
Figure 12-50. Computed and observed bottom dissolved
oxygen in the Potomac River, summer 1985. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
-270-
-------
Chapter 12 • Tributary Dissolved Oxygen
Potomac River Final Calibration - SENS 136
12
11
10
9
8
B 7
E 6
5
4
3
2
1
0
Surface Dissolved Oxygen Summer 1990
-
-
-
" • "~\ y' , ••-., / '
t ."I . . > •
-
-
-
I-
1 1 , I , , 1 < i , , 1 , ! ! I 1
0 50 100 150
Kilometers
Potomac River Final Calibration - SENS 1 36
11
10
9
8
7
B) 6
E
5
4
3
2
1
Q
Bottom Dissolved Oxygen Summer 1990
-
-
- /"\
~- / ""~ '"""\ - .
- / \ T f /
- / "H../
r / I . I •
: ; « •*
! '
K ' • l
L \ )
~ \ ,x f"" '\ j
- ' X*' ""\x'/'
; 1 • ! I , I 1 , , , i 1 i , , , 1 I ,
0 50 100 150
Kilometers
Figure 12-51. Computed and observed surface dissolved
oxygen in the Potomac River, summer 1990. Computed
mean and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-52. Computed and observed bottom dissolved
oxygen in the Potomac River, summer 1990. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
B>
E
Potomac River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1993
11
10
9
8
7
6
5
4
3
2
1
n
-
-
.-""""X /' ^~"~-: ' k
:- . .-••'' \ s"""11 """ '' '''' \
'- \.~-, 'f \ \ /
* * * *!
• i |
: * i
» [
r
-
-
-
~
1 i , , , , i > , .. , i , , , , i
0 50 100 150
Kilometers
B,
E
Potomac River Final Calibration - SENS 136
Bottom Dissolved Oxygen Summer 1993
11
10
9
8
6
5
4
3
2
1
-
-
- /"""- ,. / x\_x\
/ • I ^
: / I •
•' » *
- I [
- " / I
A * /
- \...N _, -, /
- - v"'\/ v""\/'
: !• , , A , 1 , , ", , 1 , , ( , i
0 50 100 150
Kilometers
Figure 12-53. Computed and observed surface dissolved
oxygen in the Potomac River, summer 1993. Computed
mean and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-54. Computed and observed bottom dissolved
oxygen in the Potomac River, summer 1993. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
-271 -
-------
Chapter 12 • Tributary Dissolved Oxygen
Our model consistently under-computes dissolved oxygen in the bottom water of
the lower James estuary (Figures 12-31 to 12-36). Here, an increase in vertical
mixing is a fruitful approach to improving the computation. We did not adapt this
approach since we were required to use a single formulation for vertical mixing
system-wide. An alternative approach to improving dissolved oxygen in the James
may be to increase longitudinal net circulation. Kuo and Neilson (1987) determined
dissolved oxygen concentration in James River bottom water was maintained by
strong circulation; residence time in the bottom water was too short for anoxia to
develop. Increasing net circulation requires adjustments in the hydrodynamic model
representations of viscosity and diffusivity. Whether simple adjustments in vertical
mixing are made or whether more complex adjustments in circulation are
attempted, refinement of the hydrodynamic model of the James is required before
improvements in the water quality model can result.
We have said little about the dissolved oxygen computation in the Patuxent River
(Figures 12-55 to 12-60). Despite two increments in resolution since the original
model application (Cerco and Cole 1994), this tributary seems too coarsely gridded
to provide more than a first-order representation of dissolved oxygen. Problems are
especially apparent in the upper river. The channel there is tortuous, the limit of
tidal intrusion is indefinite, and extensive wetlands influence water quality. Surface
dissolved oxygen in the lower Patuxent is depressed, possibly for the same reasons
as depression in the lower York, Rappahannock, and Potomac.
Each western tributary possesses unique properties and each alone could
consume as much study as the entire Chesapeake Bay. We note that an individual
model of the Patuxent is underway and a model of the Potomac is in the planning
stages.
Gridding of each western tributary should be sufficient to provide detailed repre-
sentation of the complex geometry. This requirement calls for roughly the same
number of cells as in the present representation of Chesapeake Bay. The tributary
cells will, of course, be much smaller than the current bay cells. Once a detailed
grid is completed, careful model (and perhaps field) examination of circulation and
mixing processes is required, with emphasis on the lower estuaries.
The York, Rappahannock, and Patuxent Rivers adjoin extensive tidal wetlands
which appear to influence water quality. Nutrients, as well as dissolved oxygen,
are, no doubt, exchanged between wetlands and channel. To represent the wetlands
physically, addition of wetting-and-drying to the hydrodynamic model is required.
A wetland biogeochemical module should be added to the water quality model.
And, as with so many processes, field investigations may also be necessary.
Measures of loads of carbonaceous oxygen-demanding material to the tributaries
are nearly unknown. Large-scale eutrophication modeling presupposes that algal
production is the predominant source of oxygen demand. System-wide, this
premise is certainly true but in confined portions of the tributaries, in the presence
of inputs from multiple sources, external loads of oxygen-demanding material
cannot be ignored. Monitoring of inputs and, perhaps, addition of organic carbon to
the watershed model are recommended.
-272-
-------
Chapter 12 • Tributary Dissolved Oxygen
B,
E
Patuxent River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1985
11
10
9
8
6
5
4
3
2
1
Q
-
-
i - / x, •
' \ / \ ~ *
: . V. ' •>-,,"/ \
'- * * \
: * . * \
\
- \
\
\
- '''-*. 1
: , | ,,,,!,,,,!, ,"""',
0 20 40 60
Kilometers
Patuxent River Final Calibration - SENS 136
Bottom Dissolved Oxygen Summer 1985
11
10
9
0
7
_, 6
E 5
4
3
2
1
Q
-
-
;
_
- *
'"•- A* \
- / '•"" \ J \
: /' >v> * • /
• '
- - '"•--, i
'- , I , , , , I , , , , 1 , , , , i
0 20 40 60
Kilometers
Figure 12-55. Computed and observed surface dissolved
oxygen in the Patuxent River, summer 1985. Computed
mean and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-56. Computed and observed bottom dissolved
oxygen in the Patuxent River, summer 1985. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
Patuxent River Final Calibration - SENS 136
12
11
10
9
8
_j ^
E 6
5
4
3
2
1
0
Surface Dissolved Oxygen Summer 1990
-
-
-
'-
- ^ /' \ *
- ' V \"
- • ' * . X.. - '
* \-
'•-
- \ /'
-
-
: , 1 , , , . 1 , , , ,
0 20 40 60
Kilometers
Patuxent River Final Calibration - SENS 1 36
10
9
8
7
6
I 5
4
3
2
1
0
Bottom Dissolved Oxygen Summer 1990
-
-
-
*
- N ''"\ r- \
~ , \
„ / \ * \ j \ ,
". •" '< * \ /
'- > - \ \
: . .--"' \ * '
\s i *
: * *
: , i , , , , i , , , , i , , , , i
0 20 40 60
Kilometers
Figure 12-57. Computed and observed surface dissolved
oxygen in the Patuxent River, summer 1990. Computed
mean and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-58. Computed and observed bottom dissolved
oxygen in the Patuxent River, summer 1990. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
-273-
-------
Chapter 12 • Tributary Dissolved Oxygen
Patuxent River Final Calibration - SENS 136
Surface Dissolved Oxygen Summer 1993
11
10
8
7
=! 6
™ 5
4
3
2
1
Q
-
r _
[~X - ' \ x . * •
-*.•** \
" - * \
• *
"'- \ /
- '••//
-
-
0 20 40 60
Kilometers
Patuxent River Final Calibration - SENS 136
Bottom Dissolved Oxygen Summer 1993
11
10
9
8
7
=! 6
£ 5
4
3
2
1
Q
-
-
-
: / "A.
- ;' ""\ _,/ \
L ; \
: f, \ /
~ 1 V • \//
: / ' 1
'/ • .
m *
0 20 40 60
Kilometers
Figure 12-59. Computed and observed surface dissolved
oxygen in the Patuxent River, summer 1993. Computed
mean and range shown as lines, observed mean and range
shown as solid circles and vertical bars.
Figure 12-60. Computed and observed bottom dissolved
oxygen in the Patuxent River, summer 1993. Computed
mean and range shown as lines, observed mean and
range shown as solid circles and vertical bars.
References
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report EL-94-4, US Army Engineer Waterways Experiment Station, Vicks-
burg, MS.
Hartman, B., and Hammond, D. (1985). "Gas exchange in San Francisco Bay," Hydrobi-
ologia 129, 59-68.
Kuo, A., and Neilson, B. (1987). "Hypoxia and salinity in Virginia estuaries," Estuaries,
10(4), 277-283.
-274-
-------
Modeling Processes
at the Sediment-Water
Interface
Introduction
The Sediment Diagenesis Model
The need for a predictive benthic sediment model was revealed in a steady-state
model study (HydroQual 1987) that preceded Corps' modeling activity. The study
indicated sediments were the dominant sources of phosphorus and ammonium
during the summer period of minimum dissolved oxygen. Simultaneously, basic
scientific investigations were illustrating the importance of sediment-water
exchange processes in Chesapeake Bay and other estuarine systems (Boynton and
Kemp 1985; Seitzinger, Nixon, and Pilson 1984; Fisher, Carlson, and Barber 1982).
For management purposes, a model was required with two fundamental capabilities:
• Predict effects of management actions on sediment-water exchange processes,
and
• Predict time scale for alterations in sediment-water exchange processes.
A sediment model to meet these requirements was created for the initial three-
dimensional coupled hydrodynamic-eutrophication model (Cerco and Cole 1993).
The model (Figure 13-1, Table 13-1) is driven by net settling of organic matter
from the water column to the sediments. In the sediments, the model simulates the
diagenesis (decay) of the organic matter. Diagenesis produces oxygen demand and
inorganic nutrients. Oxygen demand, as sulfide (in saltwater) or methane (in fresh-
water), takes three paths out of the sediments: export to the water column as
chemical oxygen demand, oxidation at the sediment-water interface as sediment
oxygen demand, or burial to deep, inactive sediments. Inorganic nutrients produced
by diagenesis take two paths out of the sediments: release to the water column, or
burial to deep, inactive sediments.
-275-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
SEDIMENT-WATER
INTERFACE
DIAGENESIS (DECAY)
OF ORGANIC MATTER
EXERTED
Figure 13-1. Sediment model schematic
Table 13-1
Sediment Model State Variables and Fluxes
State Variable
Temperature
Particulate Organic Carbon
Sulfide/Methane
Particulate Organic Nitrogen
Ammonium
Nitrate
Particulate Organic Phosphorus
Phosphate
Particulate Biogenic Silica
Available Silica
Sediment-Water Flux
Sediment Oxygen Demand
Release of Chemical Oxygen Demand
Ammonium Flux
Nitrate Flux
Phosphate Flux
Silica Flux
Additional details of the sediment model, required to understand the coupling of
the sediment submodel to the model of the water column, are provided in this
chapter. Complete model documentation was provided by DiToro and Fitzpatrick
(1993). More accessible documentation may presently be found in DiToro (2001).
-276-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
The Benthos Model
The ultimate aim of eutrophication modeling is to preserve precious living
resources. Usually, the modeling process involves the simulation of living-resource
parameters such as dissolved oxygen. For the "Virginia Tributary Refinements"
phase of the model activities (Cerco et al. 2002), a decision was made to initiate
direct interactive simulation of three living resource groups: zooplankton, benthos,
and SAV.
Benthos were included in the model because they are an important food source for
crabs, fmfish, and other economically and ecologically significant biota. In addition,
benthos can exert a substantial influence on water quality through their filtering of
overlying water (Cohen et al. 1984; Newell 1988). Benthos within the model were
divided into two groups: deposit feeders and filter feeders (Figure 13-2). The deposit-
feeding group represents benthos which live within bottom sediments and feed on
deposited material. The filter-feeding group represents benthos which live at the
sediment surface and feed by filtering overlying water.
The primary reference for the benthos model is HydroQual (2000). This report is
available on-line at http://www.chesapeakebay.net/modsc.htm. Less comprehensive
documentation may be found in Cerco and Meyers (2000) and in Meyers at al. (2000).
Benthic Algae
In shallow coastal systems, benthic algae can markedly reduce the release of
nutrients from sediments by assimilating some or all of the nutrients as they diffuse
across the surface. At times, the algae can induce substantial flux of nutrients from
the overlying water into the sediments.
Figure 13-2. Benthos model schematic.
-277 -
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
The benthic algal module in the Chesapeake Bay model was originally devel-
oped for the Delaware Inland Bays (Cerco and Seitzinger 1997). Benthic algae are
considered to occupy a thin layer between the water column and benthic sediments
(Figure 13-3). Biomass within the layer is determined by the balance between
production, respiration, and losses to predation. Nutrients from diagenetic sediment
production and from the overlying water are both available to the benthic algae.
Formulation of the benthic algal module is detailed in Cerco and Seitzinger (1997).
Several parameter changes were implemented since the original model develop-
ment. A complete parameter suite is presented here.
Light
ij
Pai
Dep
«
Bi
Water
Column Dissolved 1 to :
, Nutrient
Exchange
1 t
f ! BentwT" f 1K)5
I- - •- ' '
1 ^ 1
Active Dissolved
r * Sediments Nutrient
osition Debitus
' [Cteep Isolated
jri*! } Sediments
Om
mm
I-
;
m
Figure 13-3. Benthic algal model schematic
Submerged Aquatic Vegetation (SAV)
SAV provides habitat for numerous living resources of economic importance and
partly supports the estuarine food chain as well. Major portions of the SAV beds in
Chesapeake Bay disappeared between 1970 and 1975 (Orth and Moore 1984).
Restoration of the beds is a priority goal in management of the system.
An SAV submodel (Figure 13-4), which interacted with the main model of the
water column and with the sediment diagenesis submodel, was created for the
"Virginia Tributary Refinements" phase of the model activities (Cerco et al. 2002).
Three state variables were modeled: shoots (above-ground biomass), roots (below-
ground biomass), and epiphytes (attached growth). Three dominant SAV
communities were identified in the bay based largely on salinity regimes (Moore et
al. 2000). Within each community, a target species was selected: Vallisneria ameri-
cana, Ruppia maritima, or Zostera marina. Each community was modeled using
the same relationships but with parameter values selected for the target species. The
primary reference for the SAV submodel is Cerco and Moore (2001).
-278-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Figure 13-4. Submerged aquatic vegetation model schematic
Coupling With the Sediment Diagenesis Model
Benthic sediments are represented as two layers with a total depth of 10 cm
(Figure 13-5). The upper layer, in contact with the water column, may be oxic or
anoxic depending on dissolved oxygen concentration in the water. The lower layer
is permanently anoxic. The thickness of the upper layer is determined by the pene-
tration of oxygen into the sediments. At its maximum thickness, the oxic layer
depth is only a small fraction of the total.
The sediment model consists of three basic processes. The first is deposition of
particulate organic matter from the water column to the sediments. Due to the
negligible thickness of the upper layer, deposition proceeds from the water column
directly to the lower, anoxic layer. Within the lower layer, organic matter is subject
to the second basic process, diagenesis (or decay). The third basic process is flux of
substances produced by diagenesis to the upper sediment layer, to the water
column, and to deep, inactive sediments. The flux portion of the model is the most
complex. Computation of flux requires consideration of reactions in both sediment
layers, of partitioning between particulate and dissolved fractions in both layers, of
sedimentation from the upper to lower layer and from the lower layer to deep inac-
tive sediments, of particle mixing between layers, of diffusion between layers, and
of mass transfer between the upper layer and the water column.
The water quality and sediment models interact on a time scale equal to the inte-
gration time step of the water quality model. After each integration, predicted
particle deposition, temperature, nutrient and dissolved oxygen concentrations are
passed from the water quality model to the sediment model. The sediment model
computes sediment-water fluxes of dissolved nutrients and oxygen based on
predicted diagenesis and concentrations in the sediments and water. The computed
-279-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
WATER COLUMN
DEPOSITION
Q
UJ
(/I
cc
LU
V
SURFACE MASS TRANSFER: s fd.
PARTITIONING: fd, -* fc~ fp1
REACTIONS:
SEDIMENTATION PARTICLE DIFFUSION
W MIXING KL ,
DIAGENESIS:
PARTITIONING:
REACTIONS:
SEDIMENTATION
W
fd.
fp
BURIAL
Figure 13-5. Sediment model elevation.
sediment-water fluxes are incorporated by the water quality model into appropriate
mass balances and kinetic reactions.
Deposition
Deposition is one process which couples the model of the water column with the
model of the sediments. Consequently, deposition is represented in both the sedi-
ment and water-column models. In the water column, deposition is represented
with a modification of the mass-balance equation applied only to cells that interface
the sediments:
8C
WS
— = [transport] + [kinetics] + -^^ • Cup -
8t Az
Wnet
Az
• C
(13-1)
in which:
C concentration of particulate constituent in cell above sediments (g m~3)
WS = settling velocity in water column (m d"1)
Cup = constituent concentration two cells above sediments (g m"3)
Wnet = net settling to sediments (m d"1)
Az = cell thickness (m)
-280-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Net settling to the sediments may be less than or equal to settling in the water
column. Sediment resuspension is implied when settling to the sediments is less
than settling through the water column.
Diagenesis
Organic matter in the sediments is divided into three G classes or fractions, in
accordance with principles established by Westrich and Berner (1984). Division
into G classes accounts for differential decay rates of organic matter fractions. The
Gl, labile, fraction has a half life of 20 days. The G2, refractory, fraction has a half
life of one year. The G3, inert, fraction undergoes no significant decay before
burial into deep, inactive sediments. Each G class has its own mass-conservation
equation:
H • - = Wnet • fi • C-W • Gi-H • Ki • Gi • 0r20)
ot
in which:
H = total thickness of sediment layer (m)
Gi = concentration organic matter in G class i (g m"3)
f; = fraction of deposited organic matter assigned to G class i
W = burial rate (m d"1)
K; = decay rate of G class i (d"1)
#i = constant that expresses effect of temperature on decay of G class i
Since the G3 class is inert, K3 = 0.
Sediment-Water Flux
The exchange of dissolved substances between the sediments and water column
is driven by the concentration difference between the surface sediment layer and
the overlying water. Flux may be in either direction across the sediment-water
interface, depending on concentration gradient. Sediment-water flux is computed
within the diagenesis model as the product of concentration difference and an inter-
nally-computed mass-transfer coefficient. In the water column, sediment-water
exchange of dissolved substances is represented with a modification of the mass-
balance equation applied only to cells that interface with bottom sediments:
§C r n n . . n BENFLX
— = [transport] + [kinetics] + - (13-3)
§t Az
in which:
BENFLX = sediment-water flux of dissolved substance (g m"2 d"1)
By convention, positive fluxes are from sediment to water. Negative fluxes,
including sediment oxygen demand, are from water to sediments.
Parameter Specification
Coupling with the sediment model requires specification of net settling rates, of
the G splits of organic matter, and of burial rates.
-281 -
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Net Settling Rates
Net settling was specified identical to particle settling velocities in the water
column. These rates were 1 m d"1 for participate organic matter and 0.1 m d"1 for
the spring and summer algal groups. Net settling of zero was specified for the blue-
green algae, reflecting their buoyant properties. Net settling of particulate biogenic
silica was specified as 0.01 m d"1.
Assignment to G Classes
Upon deposition in the sediments, state variables representing particulate organic
matter in the water quality model required conversion into sediment model state
variables. The water quality model considered two classes of particulate organic
matter: labile and refractory. The sediment model was based on three classes of
organic particles: labile (Gl), refractory (G2), and inert (G3). Labile particles from
the water quality model were transferred directly into the Gl class in the sediment
model. Refractory particles from the water quality model had to be split into G2
and G3 fractions upon entering the sediments. Initial guidance for the splits was
obtained from experiments (Westrich and Berner 1984) in which roughly even
distribution between refractory and inert particulate organic carbon was noted. The
final distribution (Table 13-2) was obtained from model calibration. Carbon and
nitrogen were considered slightly more reactive than phosphorus. This treatment
possibly reflected the presence of refractory mineral forms within the water column
particulate phosphorus.
Algae settling directly to the sediments also required routing into sediment
model state variables (Table 13-2). The algal fraction routed into Gl particles was
adopted from the initial model calibration (Cerco and Cole 1994). Routing of
refractory algae into G2 and G3 classes was equivalent to the split employed for
refractory organic particles.
Table 13-2
Routing Organic Particles into Sediment Classes
WQM Variable
Labile Particles
Refractory Particles
Algae
Carbon, Nitrogen
%G1
100
65
%G2
86
30
%G3
14
5
Phosphorus
%G1
100
65
%G2
73
25.5
%G3
27
9.5
Burial Rates
As part of the initial model effort (Cerco and Cole 1994), sedimentation rates
were measured via a pollen-dating method (Brush 1989) at fifteen locations, prima-
rily in the mainstem of the bay. Model sedimentation rates were specified based on
these measures and on an additional assemblage (Officer et al. 1984). Sedimenta-
tion rates were 0.5 cm yr1 near the fall line of the bay and tributaries, 0.25 cm yr1
in the central bay and central and lower tributaries, and 0.37 cm yr"1 near the mouth
-282-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
of the bay. The sedimentation rates from the initial study were retained in the
present effort.
Sediment Model Parameters
With one exception, all parameters within the diagenesis model were exactly as
derived for the original model application (DiToro and Fitzpatrick 1993). The
exception was the phosphate partition coefficient in the surficial sediments. This
value was increased from 300 L kg"1 to 3000 L kg"1 in freshwater sediments
(salinity < 1 ppt). In applying the diagenesis model to multiple systems, DiToro
(2001) noted the phosphorus partition coefficients frequently required revision.
Low ionic strength of freshwater was suggested as the reason for employment of a
higher partition coefficient in Lake Champlain. This explanation may hold, as well,
for the use of a higher partition coefficient in the tidal fresh waters of the Chesa-
peake Bay system.
Sediment Model Results
Data Base
Sediment-water flux measures employed in
performance evaluation were obtained
from the SONE (sediment oxygen and
nutrient exchange) program (Boynton et al.
1986). Observations of sediment oxygen
demand, and ammonium, nitrate, phos-
phate, and silica exchange were conducted
at ten stations in the upper Bay and Mary-
land tributaries (Figure 13-6). Measures
were usually collected four times per year.
The methodology employed intact sedi-
ment cores incubated in triplicate
immediately upon collection. SONE data
were obtained from the principal investiga-
tors by HydroQual Inc. and supplied to
WES in March 2001.
Carbon, nitrogen, and phosphorus
analyses of surficial sediments were
conducted concurrent with the SONE
measures. Mean values from the period
of record, usually 10 years, were computed
for comparison with the model.
model development, calibration, and
Figure 13-6. SONE observations stations.
-283-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Time Series
Time-series comparisons are presented for three stations: Still Pond, R- 64,
and Point-No-Point (Figure 13-6). These stations are selected for their different
bottom environments as characterized by dissolved oxygen concentration.
Bottom dissolved oxygen at Still Pond rarely falls below 2 g m~3 (Figure 13-7).
Complete anoxia is a regular occurrence during summer in bottom waters at R-64
(Figure 13-8). At Point-No-Point, intermittent bottom-water anoxia occurs but the
duration is not so severe as at R-64 (Figure 13-9).
Peak sediment oxygen demand of 2 to 2.5 g m~2 d"1 is observed at Still Pond
(Figure 13-10). The model matches these observations well and occasionally
computes sediment oxygen demand as high as 3 g m"2 d"1. Computed and observed
SENS136 ON NEWGRID
Dissolved Oxygen
Dissolved Qxygen C62.2 Bottom
—-. (l. v Y i
Figure 13-7. Computed and observed bottom
dissolved oxygen at Station CB2.2, near Still Pond.
U
SENS136ONNEWGRID
Dissolved Oxygen
Dissolved Oxygen CB4.2C Bottom
2
4 I
1 f1 i /• 5 I ;i I1 , .I
^ J; ?; 'j :i « ji M VHi •
Figure 13-8. Computed and observed bottom
dissolved oxygen at Station CB4.2C, near R-64.
SENS136 ON NEWGRID
Dissolved Oxygen
13 f Dissolved Oxygen CB5.2 Bottom
12 | Level 2
?• j J
Figure 13-9. Computed and observed bottom
dissolved oxygen at Station CBS.2, near
Point-No-Point.
MAINSTEM BAY, - Final Calibration
Sediment Oxygen Demand (SLPD)
01334
Figure 13-10. Computed (daily) and observed
(instantaneous) sediment oxygen demand at
Still Pond.
-284-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
sediment oxygen demand in excess of 2 g m~2 d"1 occurs at R-64 but the preponder-
ance of observations and computations are less than 1 g m~2 d"1 (Figure 13-11). At
this location, sediment oxygen consumption is constrained by bottom-water anoxia
and oxygen demand escapes from the sediments in the form of chemical oxygen
demand (Figure 13-12). Observed sediment oxygen demand is also constrained by
bottom-water anoxia at Point-No-Point (Figure 13-13). Computed sediment oxygen
demand is higher, however. Careful examination of the computed bottom dissolved
oxygen concentration indicates the model computes the presence of oxygen when
none is observed (Figure 13-9). The availability of oxygen in the model allows
sediment oxygen consumption when little actually occurs. This phenomenon does
not imply more oxygen is consumed in the model than in reality at Point-No-Point.
In the model, oxygen consumption occurs in surficial sediments. In reality, the
sediments export chemical oxygen demand that consumes oxygen within the water
column.
O
O) }
MAINSTEM BAY, - Final Calibration
Sediment Oxygen Demand (R-64)
. .
MAINSTEM BAY, - Final Calibration
Sediment-water COD Flux (R-64)
E 1.5
Si
O
01 I
0 1 2
Yoars
0 1 2 3 4 5 6 ?
Years
Figure 13-11. Computed (daily) and observed
(instantaneous) sediment oxygen demand at R-64.
Figure 13-12. Computed sediment chemical oxygen
demand release at R-64.
I
O
E
MAINSTEM BAY, - Final Calibration
Sediment Oxygen Demand (PNPT)
MAINSTEM BAY, - Final Calibration
Sediment-water NH4 Flux(SLPD)
Ti
z
as
Figure 13-13. Computed (daily) and observed
(instantaneous) sediment oxygen demand at Point-
No-Point.
Figure 13-14. Computed (daily) and observed
(instantaneous) sediment-water ammonium flux
at Still Pond.
-285-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Of the three stations, observed sediment ammonium release is greatest at R-64,
often in excess of 150 mg m~2 d"1 (Figure 13-15). Lesser releases occur at Point-
No-Point (Figure 13-16) and Still Pond (Figure 13-17) respectively. The sequence
of ammonium release corresponds to the occurrence of bottom-water anoxia. In the
absence of oxygen, ammonium freely escapes from the sediments. In the presence
of oxygen, ammonium is nitrified to nitrate which is denitrified in anoxic zones
within the sediments (Jenkins and Kemp 1984). The model sequence of ammonium
release is opposite the observed trend, however. Greatest releases occur in the
aerobic waters at Still Pond. The nitrification/denitrification processes were care-
fully calibrated in the development of the sediment model (DiToro and Fitzpatrick
1993). The most likely explanation for the model behavior is that excess organic
matter is deposited at Still Pond while a shortfall in deposition occurs at R-64.
MAINSTEM BAY, - Final Calibration
Sediment-water NH4 Flux (R-64)
Years
150
um
'E '•"
Z r,ll
O>
£ -in
MAINSTEM BAY, - Final Calibration
Sediment-water NH4 Flux (PNPT)
123456
Years
Figure 13-15. Computed (daily) and observed
(instantaneous) sediment-water ammonium flux at
R-64.
Figure 13-16. Computed (daily) and observed
(instantaneous) sediment-water ammonium flux at
Point-No-Point.
I -30 |ll
O)
E -50
MAINSTEM BAY, - Final Calibration
Sediment-water NO3 Flux (SLPD)
*
* * 1
** * ' i
•• ^ i
rFi f%, y -i /, u 1 i if ,A |
-* i,i. L^' t^ ^'A '^ ' / iA/ »" n,
![ »r "! W «,
^ ' I
j 1
-•:.: t .... i ..-,, I ,,., f
[i'lj'i'i •>• i ')? ''jjy
.i i ' fj (it
i f | ' i f
, * 1 , . . i ,. ,. , i . , i , - : i :, , , i
5 6
Years
MAINSTEM BAY, - Final Calibration
Sediment-water NO3 Flux (R-64)
Figure 13-17. Computed (daily) and observed
(instantaneous) sediment-water nitrate flux at
Still Pond.
Figure 13-18. Computed (daily) and observed
(instantaneous) sediment-water nitrate flux at R-64.
-286-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
The observations show nitrate moves in both directions across the sediment-
water interface (Figures 13-17 to 13-19). Sediment uptake predominates. Release is
limited to 10 mg m~2 d"1 while uptake of 20 to 30 mg m~2 d"1 is common. In the
model, uptake predominates at Still Pond (Figure 13-17). This uptake represents
diffusive flux of nitrate-rich water from the Susquehanna River. Farther down the
bay, bottom-water nitrate concentrations are much less and computed sediment-
water nitrate flux is about evenly balanced between uptake and release.
Observations show some extreme uptake rates that are not matched by the model.
Otherwise, computed flux rates are in reasonable agreement with observations.
Observed sediment-water phosphate flux hovers in the interval ±5 mg nr 2 d"1
when bottom dissolved oxygen remains above a critical concentration. (Figures
13-20 to 13-22). When dissolved oxygen falls below the critical concentration,
phosphate release spikes up as high as 60 mg nr2 d"1. The spiking is attributed to
the reduction of iron oxides in the surficial sediments. Under oxygenated condi-
tions, the oxides are in particulate form and sorb phosphorus liberated by
diagenetic processes deeper in the sediments. As oxygen in the water is depleted,
the iron oxides are reduced to soluble form and phosphate freely diffuses to the
overlying water. The critical dissolved oxygen concentration is specified as 2 g nr3
in the model. Computed dissolved oxygen level seldom falls this low at Still Pond
so release events are infrequent and of short duration. At Point-No-Point, release
events occur more frequently although duration remains short. At R-64, which
experiences lengthy intervals of complete anoxia, sediment phosphate release is a
regular occurrence of substantial duration. The model correctly differentiates phos-
phate release among the stations and provides reasonable agreement with the
magnitude of observed releases.
MA1NSTEM BAY, - Final Calibration
Sediment-water NO3 Flux(PNPT)
Figure 13-19. Computed (daily) and observed
(instantaneous) sediment-water nitrate flux
at Point-No-Point.
MAINSTEM BAY, - Final Calibration
Sediment-water PO4 Flux (SLPD)
f*y
5 6
Years
Figure 13-20. Computed (daily) and observed
(instantaneous) sediment-water phosphate flux
at Still Pond.
-287-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
MAINSTEM BAY, - Final Calibration
Sediment-water PO4 Flux (R-64)
. ! , 1 , t * 1 i . , , i i I I . . I , , I 1 . , I , , , I , , , t
5
Years
MAINSTEM BAY, - Final Calibration
Sediment-water PO4 Fiux (PNPT)
44-
5 5
Yoars
Figure 13-21. Computed (daily) and observed
(instantaneous) sediment-water phosphate flux
at R-64.
Figure 13-22. Computed (daily) and observed
(instantaneous) sediment-water phosphate flux at
Point-No-Point.
Observed silica flux is exclusively from sediments to water and regularly
exceeds 300 mg m~2 d"1 (Figures 13-23 to 13-25). Computed silica fluxes occur in
both directions across the sediment-water interface. Maximum release is 200 mg nr
2 d"1 and most releases are much less. Shortfalls in computed silica release have
been a property of the coupled model since its first application. In the initial appli-
cation, the shortfall was attributed to omission of particulate biogenic silica from
system-wide silica loads. Diagenesis of 100 mg nr2 d"1 was added to the sediments
to account for the deficit in deposition. In the present model, biogenic silica was
explicitly included in loads so the added sediment diagenesis was eliminated.
Apparently, the added loading did not compensate for the diagenesis removed. The
computed occurrence of sediment silica uptake suggests, also, that the modeled
silica partition coefficients should be revisited. In view of management emphasis
on nitrogen and phosphorus, minimal attention was devoted to silica in the present
model effort. With a good deal of effort, the present silica results can, no doubt, be
improved.
350
300
250
I" ?<™
,E ir.o
cfi
g 100
iO
0
•if)
MAINSTEM BAY, - Final Calibration
Sediment-water Silica Flux(SLPD)
i \YI •/
^y/^;f\
MAINSTEM BAY, - Final Calibration
Sediment-water Silica Flux (R-64)
Years
Figure 13-23. Computed (daily) and observed
(instantaneous) sediment-water silica flux at
Still Pond.
Figure 13-24. Computed (daily) and observed
(instantaneous) sediment-water silica flux at R-64.
-288-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
MAINSTEM BAY, - Final Calibration
Sediment-water Silica Flux (PNPT)
» \"
i i-
0 1 'I 3 4 5 6 7
Years
Figure 13-25. Computed (daily) and observed
(instantaneous) sediment-water silica flux at
Point-No-Point.
Cumulative Distribution
Cumulative distributions were created for the population of Chesapeake Bay
sediment-water flux observations and for corresponding model computations.
Computed sediment oxygen demand exceeded observed throughout the distribution
(Figure 13-26). Median computed demand exceeded observed by more than 0.5 g
m~2 d"1. Observed sediment ammonium release exceeded computed throughout the
distribution (Figure 13-27). Median observed release exceeded modeled by 10 mg
m~2 d"1. A number of explanations can be offered for these results. The excess of
sediment oxygen demand may be attributed to computed bottom dissolved oxygen.
Computed bottom water dissolved oxygen does not match the lowest observations
(e.g. Figure 13-9) at all locations. As a result, computed sediment oxygen demand
MAINSTEM BAY, - Final Calibration
Sediment Oxygen Demand
SONE
Ml
Percent Less than
Figure 13-26. Cumulative distributions of
observed and computed sediment oxygen
demand at all Chesapeake Bay stations.
MAINSTEM BAY, - Final Calibration
Sediment-water NH4 Flux
SONE
Model
Ml fb
Percent Less than
Figure 13-27. Cumulative distributions of
observed and computed sediment-water
ammonium flux at all Chesapeake Bay
stations.
-289-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
exceeds observed because the model allows consumption of oxygen where no
oxygen is present in the bay. The excess of observed over computed ammonium
release may indicate more nitrogen should be deposited on the bottom. Or the occa-
sional excess of computed dissolved oxygen may be allowing more nitrification to
take place in the modeled sediments than in the observations.
The preponderance of observed and computed sediment-water nitrate fluxes are
essentially zero (Figure 13-28). At the lower end of the distribution, observed sedi-
ment uptake is greater than modeled although the magnitude of the excess is not
great.
Half the observed and computed sediment-water phosphate fluxes are less than
or effectively zero (Figure 13-29). In the upper half of the distribution, the observa-
tions show a gradual transition to sediment phosphorus release while the model
shows a much steeper gradient. The model behavior may be attributable to the
specification of a critical dissolved oxygen concentration and to the prescribed
response of the sorption coefficient to dissolved oxygen less than the critical value.
Maximum phosphorus releases agree in both model and observations.
I
J:
f-60
60
-m
MAINSTEM BAY, - Final Calibration
Sediment-water NO3 Flux
SONE
MAINSTEM BAY, - Final Calibration
Sediment-water PO4 Flux
SONE
Percent Less than
Percent Less than
Figure 13-28. Cumulative distributions of
observed and computed sediment-water nitrate
flux at all Chesapeake Bay stations.
Figure 13-29. Cumulative distributions of
observed and computed sediment-water
phosphate flux at all Chesapeake Bay
As noted in the time series, observed
sediment silica release greatly exceeds modeled (Figure 13-30). At the median,
modeled release is essentially zero while median observed release is 175 mg m~2 d~
l. Maximum observed release exceeds 600 mg m~2 d"1 while the maximum modeled
release is 100 mg m-2 d"1.
One-to-one matching of observations with the model is an extremely demanding
test. For highly-varying phenomenon such as sediment-water phosphorus flux, a
small phase shift in the model can produce widely disparate pairs of observations
and computations. We also produced distribution plots that considered the popula-
tion of observations and the population of computations. In those comparisons, the
distributions of nitrate flux matched almost perfectly. For the remaining substances,
the computed distributions changed but the trends and conclusions noted above
remained valid.
-290-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
MAINSTEM BAY, - Final Calibration
Sediment-water Silica Flux
SONE
Model
Percent Less than
Figure 13-30. Cumulative distributions of
observed and computed sediment-water
silica flux at all Chesapeake Bay stations.
Sediment Composition
The author of the sediment diagenesis model describes particulate organic matter
in the sediment as the "ashes" of the diagenesis process. That is, the majority of
sediment organic matter is refractory or inert even though the majority of material
deposited is labile (Figure 13-31). The labile material rapidly decays away leaving
mostly ashes. Particulate carbon and nitrogen in the sediments is almost exclusively
organic; the end products of diagenesis escape to the water in dissolved or gaseous
form. For phosphorus, however, a substantial amount of diagenetically-produced
phosphate is trapped in the sediments in mineral form. Consequently, sediment
particulate phosphorus is a mixture of organic and mineral substances.
10000
to""""
E 10®®
o>
c
o
Concentrat
P3 I
— G1
— G2
G3
35 0.15 0.25 0.35 0.45 0.55 0.65 0.75
Sedimentation (cm yr )
Figure 13-31. Effect of sedimentation (burial) rate on
sediment organic matter. Calculations based on deposition
of 1 g nr2 d'1 and G1, G2, G3 splits of 0.65, 0.3, 0.05
respectively.
-291 -
-------
Chapter 13 * Modeling Processes at the Sediment-Water Interface
Computed sediment composition is determined by computed deposition, by the
specified splits into Gl, G2, and G3 organic matter, and by the specified burial rate.
Comparison of computed and observed composition provides a check that the
modeled combination of deposition and burial are in balance with actual processes.
Observed sediment composition is on a percent basis by weight. Modeled
composition is as mass per unit volume. Model units are readily converted, for
comparison with the data, through division by the model sediment solids concentra-
tion, 0.5 L kg"1. Results indicate the model sediments are largely deficient in
carbon (Figure 13-32) and nitrogen (Figure 13-33). Model sediment phosphorus is
in excess by a small amount (Figure 13-34). The computed carbon-to-nitrogen ratio
in the sediments is correct (Figure 13-35) but the carbon-to-phosphorus ratio is low
(Figure 13-36) indicating the sediments are enriched with phosphorus.
6 ;
11D Obs C |
I Calibration!
Figure 13-32. Observed and computed sediment carbon
at stations in the upper Chesapeake Bay and Maryland
tributaries.
Figure 13-33. Observed and computed sediment nitrogen
at stations in the upper Chesapeake Bay and Maryland
tributaries.
0.5
0.45
0.4
0.25
iObsP I
Calibration
3 0.35 ----
o
€. 0.3 •
Figure 13-34. Observed and computed sediment
phosphorus at stations in the upper Chesapeake Bay
and Maryland tributaries.
^VV^vvvv0
Figure 13-35. Observed and computed sediment carbon-
to-nitrogen ratio at stations in the upper Chesapeake Bay
and Maryland tributaries
-292-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
70
;nObsC:P
60 >• • Calibrahor
50
O)
"30
10 -• - m
1
•
-i
\
I
, _- • - -
,
1 1
* ™ !
1 I 1
-
1
1
1
r
i
i
Figure 13-36. Observed and computed sediment carbon-
to-phosphorus ratio at stations in the upper Chesapeake
Bay and Maryland tributaries.
Computed sediment composition can be altered by three processes: deposition,
burial, and fractionation. Deposition is computed dynamically while burial and
splits into G fractions are specified parameters. Sediment composition is linearly
proportional to deposition (Figure 13-37). Increasing deposition by a factor of two
or three, necessary to bring about a proportional increase in sediment carbon and
nitrogen, is impossible, however, without depleting the water column of these
substances. Moreover, an increase in deposition of all substances would increase
sediment phosphorus beyond observed concentrations. Balancing sediment compo-
sition through deposition requires simultaneous increases in carbon and nitrogen
deposition and a decrease in phosphorus deposition. Deposition is likely not the
origin of discrepancies in computed and observed concentrations nor is alteration
of deposition likely to be the remedy.
1200
1000
r-
'E soo
I 600
g 400
o
200
j — G1'
|^G2
I G3;
0.2
0.4
0.6
0.8
1.2
1.4
Deposition (g m'
Figure 13-37. Effect of deposition rate on sediment organ-
ic matter. Calculations based on burial rate of 0.25 cm yr1
and G1, G2, G3 splits of 0.65, 0.3, 0.05 respectively.
-293-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Sediment composition is inversely proportional to burial rate (Figure 13-31).
Due to relatively rapid decay, concentrations of Gl and G2 organic matter show
little sensitivity to burial rate while large alterations in the G3 fraction are possible.
Uncertainties in specification of burial rate allow wide latitude to alter sediment
composition through this process. Concentrations of carbon, nitrogen, and phos-
phorus will all change by the same percentage, however. Consequently, raising
carbon and nitrogen concentration to observed levels will raise phosphorus well
above the observations.
Concentration of sediment organic matter is roughly linearly proportional to
specification of the G3 fraction of material deposited (Figure 13-38). In the present
calibration, the G3 fraction of phosphorus is double the fraction of carbon and
nitrogen (Table 13-2). Increasing the G3 fraction of carbon and nitrogen to the level
specified for phosphorus appears to be the means to increase sediment carbon and
nitrogen while minimizing change to sediment phosphorus.
3000
2500
'E 2000
1500
1000
500
0.05
0.1
inert (G3) Fraction
0.15
0.2
Figure 13-38. Effect of G3 fraction on sediment organic
matter for a system comprised of labile and inert fractions.
Calculations based on deposition of 1 g nr2 d~1 and burial
of 0.25 cmyr1.
Sensitivity Analyses
More than 140 multi-year model runs were completed during the calibration
process. In some of these runs, parameters were varied with the aim of improving
model-data agreement. Other runs were classic sensitivity analyses aimed at exami-
nation of model behavior in response to parameter alteration. We have paired
several of these runs to provide insight into the effects of parameter selection on
model calibration. Some caution should be used in interpreting these comparisons.
The paired runs are not always sequential and the parameters considered are not
necessarily the only differences between the runs.
Net Settling. Net settling of inorganic solids was specified as 10% of settling
through the water column to mimic the effects of resuspension. Net settling of
organic solids was specified as equal to settling through the water column. In
effect, no resuspension of organic matter was considered. In model run SENS85,
-294-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
net settling of organic solids was set to 20% of settling through the water column.
This specification increased and improved calculations of particulate organic matter
near the bottom (Figure 13-39) but was detrimental in virtually all other respects.
Computed sediment oxygen demand and ammonium release were vastly dimin-
ished and bottom-water dissolved oxygen increased by 1 to 2 g m~3 (Figures 13-40
to 13-42).
This run illustrates a paradox in the present model operation. Matching of
observed suspended solids, inorganic and organic, requires a representation of
resuspension while the diagenesis model requires the direct input of settling
organic matter. Since computation of sediment diagenesis and its effects is far more
important than the concentration of organic solids near the bottom, we have always
specified net settling of organic matter equal to settling through the water column.
The problem may be that our model does not truly represent resuspension. Rather,
we inhibit settling. Inhibited settling prevents the deposition of fresh organic matter
into the sediments and diminishes diagenesis. A possible solution is to incorporate
true resuspension into the sediment model. Solids would settle directly into the
sediments at the same rate as settling through the water column. Subsequently, the
mixture of bed sediments would be resuspended at a specified resuspension
velocity. This resuspension velocity could be calculated from first principles (e.g.
shear velocity) or estimated through the calibration procedure. In this process, fresh
organic matter would freely settle into the sediments while the resuspended mate-
rial would be a mixture of fresh and aged solids. We cannot be certain this process
would produce improved results but we see no other way out of the present
paradox.
SENS85 ON NEW GRID
Paniculate Organic Carbon
POC CBS.2 Bottom
CB5.2 Level 2
SENS85
NIT SETTLING < SETTLING
THROUGH WATER
SENS 136
MET SETTLING = SETTLING
THROUGH WATER
I
SENS1360NNEWGRID
Parteulate Organic Carbon
POC CB5.2 Bottom
CBS,2 Level 2
Figure 13-39. Effect of net settling on computed particulate organic carbon at the
bottom of Station CBS.2 (near Point-No-Point).
-295-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
MAINSTEM BAY, • SENS8S New Grid
Sediment-water NH4 Flux (CB5.2)
•5
1 '
SENS 136
NET SETTLING = SETTLING
THROUGH WATER
MAINSTEM BAY, - SENS136 New Grid
Nitrogen Flux (Ammonium)
Sedirncnt-wa«rrNH4Flu»(CB5.2) i
L-JS: W.i 1,
Figure 13-40. Effect of net settling on computed sediment-water ammonium flux
(seasonally-averaged) at Station CB5.2 (near Point-No-Point).
MAINSTEM BAY, • SENS85 New Grid
Sediment Oxygen Demand (CBS.2)
SENS 136
NET SETTLING = SETTLING
THROUGH WATER
MAINSTEM BAY, • SENS136 New Grid
S0dimi0iil Oxygen Demand
Sedtwnt G^ygi'n Demand (CBS 1}
Figure 13-41. Effect of net settling on computed sediment oxygen demand (sea-
sonally-averaged) at Station CBS.2 (near Point-No-Point).
-296-
-------
Chapter 13 * Modeling Processes at the Sediment-Water Interface
=! ;
Hi
£ b
SENS850NNEWGRID
Dissolved Oxygen
• Dissolved Oiygen CBS, 2 Bottom
CB5.2 Level 2
. .
> • j ' } ' ' * ; * ' I i . *
" • : , \ * * i *• i i
'* .., . V. .'*, .1 .''I. \ .'* i I
SENS85
NET SETTLING < SETTLING
THROUGH WATER
SENS136QNNEWGRID
Dissolved Oxygen
i:if pissolved Oxygen CB5.2 Bottom
13 } d.BS.i Lewi 2
SENS 136
NET SETTLING = SETTLING
THROUGH WATER
Figure 13-42. Effect of net settling on computed dissolved oxygen at the bottom of
Station CBS.2 (near Point-No-Point).
L- Obs C |
f«SENS137 L
Calibration^
Burial Rate. No spatial pattern was evident in the sedimentation rates collected as
part of the original model study. Model sedimentation rates were specified based on
the Brush data set and on other information. In SENS 137, the median rate from the
Brush data, 0.16 cm yr"1, was employed throughout the system. This amounted to a
reduction in burial since the rates employed in the calibration were 0.25 to 0.5 cm yr"1.
As expected, reducing the burial rate increases the concentration of sediment
organic matter although not by an amount sufficient to make up the deficit noted
earlier (Figure 13-43). Reducing
burial also increases sediment oxygen
demand by 0.2 g m"2 d"1 (Figure
13-44), sediment ammonium release
by 5 mg m"2 d"1 (Figure 13-45), and
sediment phosphorus release by up
to 5 mg m"2 d"1 (Figure 13-46). The
impact of the increased sediment
oxygen demand on bottom-water
dissolved oxygen is less than
0.5 g m"3 at mid-bay (Figure 13-47).
The major impact of reduced burial
rate is not shown in any of these
figures. The residence time in the
sediments increases so that the time
for the bay to respond to load changes,
especially phosphorus, is increased.
a. 2
nil
II
M
Figure 13-43. Effect of burial rate on computed sediment
organic carbon at stations in upper Chesapeake Bay and
Maryland tributaries.
-297-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
,.<
!r
i 0
• '
1 "
1 '
0 ' '
£ 1 1
.,
11
MAINSTEMBAY,-SENS136NswGrid i
Sediment Oxygen Demand j
pediment Oxygen Demand |CB52| j
I
i
.IS
' 1 E I I I
i' 4 a 3 10
S '-
| S -i-i
I IK
SENS 137 | ''
i 1 h
BEDUCED BUBJAL > -> ' •
BATE
SENS 136
<=i CALIBRATION
MAINSTEMBAY,-SENS137 New Grid
Sediment Oxygen Demand
Sediment Oxygen Dtfinand (CBS. 2)
-
*
-
-
:( ' [ ! l
1 4 6 H 10
Years
Figure 13-44. Effect of burial rate on computed sediment oxygen demand
(seasonally-averaged) at Station CBS.2 (near Point-No-Point).
M AiNSTEH BAY, - SENS136 New Grid
Nitrogen Flux (Ammonium)
Sedimeqt-water NH4 Fh» (C6S.2) i
SENS 136
CALIBRATION
SENS 137
REDUCED
BUBIAL RATE
MAINSTEM BAY, - SENS137 New Grid
Nitrogen Flux (Ammonium)
Sediment-water NH4 Flux !Cp5.2}
Figure 13-45. Effect of burial rate on computed sediment-water ammonium flux
(seasonally-averaged) at Station CBS.2 (near Point-No-Point).
-298-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
MAINSTEM SAY, - SENS136 NewGrid
Phosphorus flax (PCM)
Sediment-waff P04 Flui(CB5.2)
CBS, 2 Level 1
SENS 136
CALIBRATION
SENS 137
REDUCED
BURIAL RATE
MAINSTEM BAY, -SENS1 37 New Grid
Phosphorus Flux (P04)
Sediment-water P04 Flux (CBS, 2)
C85.2 lew! 1
Figure 13-46. Effect of burial rate on computed sediment-water phosphate flux
(seasonally-averaged) at Station CBS.2 (near Point-No-Point).
SENS1360NNEWGRID
Dissoked Oxygen
" Desojved Oxygen CBS.2 Bottom
Leva', 2
•
I
i! •
SENS 136
CALIBRATION
_L
SENS 137
REDUCED
BURIAL RATE
SENS1370NNEWGRID
Dissolved Oxygen
Dissolved 0
-------
Chapter 13 * Modeling Processes at the Sediment-Water Interface
G3 Fractionation. In SENS 104,
the G3 fraction of settled organic
carbon was increased from 5% to
9.5% at the expense of G2 carbon.
This alteration had a large and
usually beneficial effect on
computed sediment carbon (Figure
13-48) although an even larger inert
fraction is apparently required to
match observations at many loca-
tions. (Or the SENS 104
fractionation and the SENS 137
burial rates.) Oddly, sediment
carbon diminished at Still Pond.
The odd behavior illustrates a
caveat in these sensitivity analyses. Figure 13-48. Effect of G3 fraction on computed
SENS104 proceeded the final calibra- organic carbon at stations in upper Chesapeake
tion by over 30 model runs. Other Maryland tributaries.
parameter changes took place as well
as G3 fractionation. One of the other changes affected computed sediment carbon
at Still Pond.
6 i
r Obsc
5 , BSENS104 ,
Calibration
o, ,
c
O
CL 0 , .
0 •
r-
-
r
i
i 1
> t
1;
i
i
: -
!
t
« I
i
fS3 -f" ^ ^ ^
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
SENS1040NNEWGRIO
Dissolved Oxygen
" Dissolved Oxygen CBS.2 Bottom
| Level 2
SENS 104
•t i! '
' ¥ •* »
i '• i i- ? i r .r* t f
SEHS1200N NEWGRID
Dissolved Qxyygn
? DKN(iiv,->dO*yg«'nCB5.2 Bottom
•• 'i: n nU
SENS 120
REDUCED
G3 FRACTION
< i,
Figure 13-50. Effect of G3 fraction on computed dissolved oxygen at the bottom of
Station CBS.2 (near Point-No-Point).
Benthos Model Results
The initial phase of the Chesapeake Bay model (Cerco and Cole 1994) relied on
living-resource indicators. Living resources, including benthos, were introduced
during the Virginia Tributary Refinements (Cerco et al. 2002) and living-resource
responses to management actions were emphasized in interpreting model results.
The present phase of the model activity has come full-circle. The living-resource
indicators dissolved oxygen, chlorophyll, and clarity are once again the compo-
nents emphasized in model computations. Living resources introduced earlier have
barely been examined.
Complete results of the benthos model are provided in the CD-ROM that accom-
panies this report. Examination of present model results indicates the computation
of filter feeders closely resembles the original application (Figures 13-51, 13-52).
Computed deposit feeders have, perhaps, increased since the Virginia Tributaries
application (Figures 13-53, 13-54). In view of the multiple orders-of-magnitude
variation in observed deposit feeders, factor-of-two differences in computed
biomass don't have major impacts on the calibration status. The role of deposit
feeders in the model is minor; they enhance mixing in the bed sediments. The
increase in deposit feeders is interpreted to have negligible impact on model
computations.
Correct calculation of primary production was of fundamental interest in the
present study. Improvement in the production calculation necessitated revisions in
phytoplankton formulations and parameter values. Following initial improvements
to the production calculation, the authors of the benthos model were invited to map
-301 -
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Filter Feeders
CBS Level 1
LD-1
CM
*
#
,£"-
t>
OJ -
£
o>
Filter Feeders CBS
n
A A /uyvv/1 ^ v Av/\/\Ax
0 123456789 10
Yeors
PRESENT CALIBRATION e==5>
<== TRIBUTARY REFINEMENTS
Final Calibration -SENS 136
Filter Feeders CBS
1 1
i
M
> F,
1 *
u
E t .
en
01
0
I
"' '
-
-
'. '
.
'
.
: . i , i , , _ t . , , i . i i . i . t , , , i , . _ i . . , l i , i .. , , t !
1 2 3. 4 5 fc ? g 9 10 i
Yoare
Figure 13-51. Computed filter feeders in central Chesapeake Bay (CBS): Virginia
Tributary Refinements versus present calibration.
Filter Feeders
TF2 Leve! 1
_, Filter Feeders TF;
t
01234567
Years
i
10
PRESENT CALIBRATION
TRIBUTARY REFINEMENTS
Final Calibration-SENS 136
Filer Feeders TF2
E tj
o
I *
Years
B i 10
Figure 13-52. Computed filter feeders in the tidal fresh Potomac River (TF2):
Virginia Tributary Refinements versus present calibration.
-302-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Deposit Feeders
CBS Level 1
q
CN ~~
*
-1.°-
^S-
q
o
C
Deposit Feeders CBS
A l\ 1 A A A /I /I
J1234S678910
Yeors
TRIBUTARY REFINEMENTS
Final Calibration-SENS 136
Deposit Feeders CBS
PRESENT CALIBRATION
Figure 13-53. Computed deposit feeders in central Chesapeake Bay (CBS):
Virginia Tributary Refinements versus present calibration.
Deposit Feeders
TF2 Level \
q
o-i ~
m
CM -"-
*
-K-
E«
o>2-
o
d.
Deposit Feeders TF2
A
„ A / \ f\
/\/\/\/\ A A A A A
w v v v \y\/V/V/ AyV
n i i i i r i 1 ! \ 1
0 1 23456789 10
Years
PRESENT CALIBRATION c=£>
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
the model to the new grid and to implement parameter revisions, if necessary. Since
that time, no revisions were made to the benthos model. The stability of the
benthos model through more than 100 water quality runs that succeeded the re-
mapping indicates the benthos model is robust. The resemblance of results between
the present model and the Virginia Tributaries model indicates the activity of the
benthos, as originally calibrated, is maintained in the present model.
Results of the Submerged Aquatic Vegetation
(SAV) Model
When the phytoplankton production relationships and parameters in the eutroph-
ication model were revised, corresponding changes were made to the epiphyte
component of the SAV model. One potentially significant alteration was a change
from daily-average irradiance to sinusoidally-varying irradiance. Examination of
the SAV component of the model revealed that these changes, and perhaps others,
had a substantial, detrimental, effect on computed SAV. Computed epiphytes over-
whelmed the vegetation. As a consequence, a re-calibration of the SAV model was
completed. We endeavored to bring epiphytes and SAV back into calibration while
minimizing revisions to the extensive model parameter suite. Changes were
centered on the epiphyte loss terms and on the SAV production-irradiance relation-
ships. Revised parameter sets are presented in Table 13-3 (SAV) and Table 13-4
(epiphytes).
Table 13-3. Parameters in SAV model
Parameter
Acdw
Ada
Fpsr
Ksh
Khnw
Khns
Khpw
Khps
Pmax
Rsh
Rrt
SL
Trs
a
Definition
carbon to dry weight ratio
shoot carbon per unit leaf area
fraction of production transferred from
shoots to roots
light attenuation by shoots
half-saturation concentration for nitrogen
uptake by shoots
half-saturation concentration for nitrogen
uptake by roots
half-saturation concentration for
phosphorus uptake by shoots
half-saturation concentration for
phosphorus uptake by roots
maximum production at optimum
temperature
shoot respiration
root respiration
sloughing
transfer from root to shoot
initial slope of Pvsl curve
Freshwater
0.37
7.5
0.12to1.0
0.045
0.19
0.95
0.028
0.14
0.1
0.02
0.02
0.01 to 0.1
0.0 to 0.05
0.015
Ruppia
0.37
4.0
0.1 to 0.85
0.045
0.19
0.95
0.028
0.14
0.09
0.015
0.015
0.01 to 0.035
0.0
0.0036
Zostera
0.37
4.0
0.1 to 0.85
0.045
0.1
0.4
0.02
0.1
0.06
0.013
0.013
0.01 to 0.035
0.0
0.0068
Potomac
0.37
7.5
0.12to1.0
0.045
0.19
0.95
0.028
0.14
0.1
0.02
0.02
0.01 to 0.1
0.0 to 0.05
0.022
Units
g shoot C rrT2 leaf
area
m2 g"1 C
g N m"3
g N rrf3
gPrrr3
g Pnr3
g C g"1 DW d"1
d'1
d-1
d'1
d-1
(g C g" DW)
(E m-2)-1
-304-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Table 13-4. Parameters in epiphyte model
Parameter
Acchl
Adwcep
Kep
Khep
Khn
Khp
Pep
PR
Rep
a
Definition
carbon to chlorophyll ratio of viable epiphytes
ratio of epiphyte dry weight to viable epiphyte
carbon
light attenuation coefficient
density at which growth is halved
half-saturation concentration for nitrogen uptake
half-saturation concentration for phosphorus
uptake
maximum production at optimum temperature
predation rate
respiration
initial slope of Pvsl curve
Value
75
18
0.06
0.1
0.025
0.0025
350
5.0
0.02
8
Units
g C g"1 Chi
gDWg-1C
m2 leaf surface
g"1 epiphyte C
g epiphyte C
g"1 shoot C
g N m"3
g Pm"3
g C g"1 Chi d"1
g shoot C
g"1 epiphyte C d"1
d"1
(g C g"1 Chi)
(E m1)-1
Complete results of the SAV model are provided in the CD-ROM that accompa-
nies this report. The computations are consistent with, but not identical to, the
original application (e.g. Figures 13-55 to 13-57). The original application obeyed
light attenuation criteria listed in the first SAV Technical Synthesis (Batuik et al.
1992). During the re-calibration we verified that the model remains consistent with
these criteria.
SAV Biomass
CB6 Level
g^SAV Biomass CB6
C o-
o 10
c
O O .
I— o-
0 1
4 5
Years
Present Calibration
9 10
Tributary
Refinements
CHESAPEAKE BAY, SENS 153
SAV Biomass CB6
'/ >f ''
Figure 13-55. Submerged aquatic vegetation in the lower western shore of
Chesapeake Bay (CB6): Virginia Tributary Refinements versus present calibration.
-305-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
SAV Biomoss
WE4 Level 1
^ SAV Biomass WE4
^
O o
_Q to
i_ i~
O
O
c§^
,O co
0123456789 10
Years
Pros out Cali
Tubular}'
Rofin cm cuts
CHESAPEAKE BAY, SENS 153
SAV Biomass WE4
V V '»
Figure 13-56. Submerged aquatic vegetation at the York River mouth (WE4):
Virginia Tributary Refinements versus present calibration.
SAV Biomass
TF2 Level 1
9 10
Present Calibration
1400
120G
1000
c
o
« MO
O
o 600
Tributary
Refinements
CHESAPEAKE BAY, SENS 153
SAV Biomass TF2
n t
Figure 13-57. Submerged aquatic vegetation in the tidal fresh Potomac River
(TF2): Virginia Tributary Refinements versus present calibration.
-306-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Results from the Benthic Algal Model
Benthic algae were introduced to the model as part of the Virginia Tributary
Refinements (Cerco et al. 2002). At the time, their role in the model ecosystem was
perceived as minor and little attention was devoted to them. Concurrent with the
revisions to the phytoplankton kinetics in the present study, production relation-
ships and parameters in the benthic algal model were revised for consistency with
the phytoplankton model. During the model calibration, aberrant behavior of
computed phosphorus was noted and attributed to the benthic algae. The revised
relationships and parameters caused benthic algae to attain impossible biomass and
sequester large quantities of nutrients in the sediments. Consequently, a recalibra-
tion of the benthic algal model was conducted. The primary goal of the
recalibration was calculation of algal biomass comparable to measures in Chesa-
peake Bay and elsewhere. The resulting parameter set is presented in Table 13-5.
Table 13-5
Parameters in Benthic Algal Model
Symbol
PBm
a
CChl
Ancb
Apcb
Khn
Khp
Topt
KTg1
KTg2
BMr
Tr
KTb
Phtl
Kesed
Kebalg
Definition
maximum photosynthetic rate
initial slope of production vs. irradiance
relationship
carbon-to-chlorophyll ratio
nitrogen-to-carbon ratio
phosphorus-to-carbon ratio
half-saturation concentration for nitrogen
uptake
half-saturation concentration for phosphorus
uptake
optimal temperature for algal growth
effect of temperature below Topt on growth
effect of temperature above Topt on growth
metabolic rate at reference temperature
reference temperature for metabolism and
predation
effect of temperature on metabolism and
predation
rate of predation by planktivores
light attenuation by sediment solids
light attenuation by benthic algal self-shading
Value
300.
8.
100
0.167
0.0167
0.01
0.001
25
0.003
0.010
0.02
20
0.032
0.2
0.5
0.20
Units
g C g 1 Chi d 1
g C g'1 Chi
(E nr2)-1
g C g'1 Chi
g N g 1 C
g P g 1 c
g N m 3
g P m 3
°C
°C~2
oQ-2
d1
°C
=c-,
m2 g 1 C d 1
m2 g 1 C
-307-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Computed benthic algal biomass ranges up to 3 g C m~2, in agreement with
measures conducted in a variety of systems (Table 13-6). The primary determinate
of algal density is light. Algal biomass shows an inverse relationship to optical
depth (total depth x light attenuation) at the sediment-water interface (Figure 13-
58). No algae are computed above optical depth 5. The highest densities of
computed benthic algae are found in shallow water near the mouths of the lower
western tributaries, along the lower eastern shore, and in eastern embayments
(Figure 13-59). Lesser densities occur in tidal fresh waters and in other shoal areas.
Table 13-6
Reported Benthic Algal Biomass
Biomass,
gCm2
4
2.1
2 to 4
1.6
0.4 to 7.2
2.2 to 15
System
Delaware
Inland Bays
Goodwin
Islands, York
River mouth
North Inlet SC
Ems-Dollard
Estuary
Laholm Bay
Savin Hill
Cove, near
Boston
Harbor
Citation
Cerco and
Seitzinger
(1997)
Buzzelli
(1998)
Pinckney and
Zingmark
(1993)
Admiraal et
al. (1983)
Sundback
(1986)
Gould and
Gallagher
(1990)
Comments
Computed annual average.
Mean of 1 08 observations. Converted from
mg Chi a m2 using C:Chl ratio of 50 (Gould
and Gallagher 1990).
Range over a year. Converted from mg Chi a
m2 using C:Chl ratio of 50 (Gould and
Gallagher 1990).
Converted from cells cm 2 using 1 g C m 2 =
1 .25 x 1 06 cells cm 2 found in text.
Range over a year observed at 1 4 to 1 6 m.
Converted from cells cm 2 as per Admiraal et
al. (1983).
Report C:Chl of 18.7 to 60.4.
2.5
"E
O 2
v
B 1.5
o
I 1
5 10
Optical Density
Figure 13-58. Summer-average benthic algal bio-
mass versus optical density.
Figure 13-59. Computed summer-average benthic
algal biomass, 1986.
-308-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Several investigators have extrapolated the results of experimental benthic flux
measurements to describe whole ecosystem responses. For example, Rizzo et al.
(1992) suggested an increase in water-column light attenuation would decrease
benthic algal assimilation of nutrients, increase nutrient release to the overlying
water, and thereby increase phytoplankton production. Other studies have suggested
that increased nutrient loading would increase benthic algal production (Sundback
etal. 1991).
To investigate the role of computed
benthic algae in modulating sediment-water
oxygen and nutrient flux, we isolated sedi-
ment-water fluxes in shoal areas likely to
contain benthic algae. Shoals were defined
as cells adjacent to the shoreline. The pres-
ence of 0.5 to 1.5 g algal C m~2 in shoal
areas of CBS (Figure 13-60) had little
apparent effect on sediment-water fluxes
compared to a deep station in the center of
the segment. Peak sediment oxygen demand
exceeded 3 g m~2 d"1 in shoals and deep
water (Figure 13-61). Sediment ammonium
(Figure 13-62) and phosphate (Figure 13-
63) release were arguably higher in the
shoals than in deep water, except for incidents
of bottom-water anoxia at the deep station.
Similar results were noted in other segments.
0
E
Final Calibration - SENS 136
Benthic Algae - CBS
r,
Years
Figure 13-60. Spatially-averaged benthic algal
biomass in shoal areas of Chesapeake Bay Program
Segment CBS.
-0.5
-1
gm O/m2/Day
-35
-4
i
Final Calibration - SENS 1 36
Sediment Oxygen Demand - CBS
, h |; /' (' \ ,'•, h i\ r, /
- i i i i i I
'- 1 ! ' ! 1 1
Mi V • I}. i
p( h IK
E. ''HI ' 1 '
. _, _J ,
1 25456789 10
Years
Shoal Areas
MNSTEM BAY, - Final Calibration
diment Oxygen Demand (CB5.2)
(.'liuiinel Slad(.>n
Figure 13-61. Daily-average sediment oxygen demand in shoal areas of CBS and at
deep station CB5.2.
-309-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Fin
al Calibration -SENS 136
Sediment-water NH4 Flux - CBS
160
140
1?0
i?10u
0
"E so
Z
| 60
40
20
0
r
r
r
i i
i
- 1
" if IS I
yl '
E.f | i
- 1
•> \ '
\f
0 1
t
|
i
J
1 1
2
J
i
I
i
1
r
i
( i
! 1
*
i/'1 U
Y Y
J
I
i
1
1
1
i
/
i . .
j
I
11
j
fl
i
1
V
1
1
1
1 1
1
I '
1 1
I1 1
1 i I '
III
v V( \f
i ..,*,,,,
Years
i
•
I
\
10
( 'ham
el
Slaticti'i
>•
60
50
>i
1 ,0
E
0) 30
E
2<~
10
0
j*
"*lf
i < i . ..i \ . . . .
•Mlual Altas
AINSTEM BAY, - Final Calibration
ediment-water NH4 Flux (CBS. 2)
i
i i
i! i
:« H ,'
• '1 ' ;
- i I
>
-
-
v 1 ^
i , i ' '
> '
' ' , ' ' 1 1
1,1, i
'
i i i
i
,
;
> 6 *> i / ,1 Q !••>
Years
Figure 13-62. Daily-average sediment-water ammonium flux in shoal areas of CBS
and at deep station CB5.2.
Final Calibration - SENS 136
Sediment-water PO4 Flux - CBS
345S78910
Station
70
& 60
•D
~£ 50
£ 40
E 30
20
to
MNSTEM BAY, - Final Calibration
diment-water PO4 Flux (CB5.2)
01234
6 7 8 9 10
Figure 13-63. Daily-average sediment-water phosphate flux in shoal areas of CBS
and at deep station CB5.2.
-310-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
These results appear contrary to numerous studies in which benthic algae were
observed to trap nutrients in bottom sediments. Several factors contribute to our
results. First, our fluxes are computed at model time steps (15 minutes) but
reported as daily averages. Since daylight hours are only half the day, our daily
average results will differ from studies that compare fluxes under light and dark
conditions. Our regional averages include cells in which benthic algae are absent so
that the effect of algal-rich areas is damped in the averaging process. Finally, the
shoals receive direct loads from runoff and bank erosion. Consequently, deposition
in these areas will differ from the channel reaches. Still, we must conclude that
under present conditions, at time and spatial scales of regional significance,
computed benthic algae do not exert major impact on sediment-water oxygen and
nutrient fluxes.
Sensitivity Analysis
During the management scenarios conducted with the present model, positive
responses of dissolved oxygen to reductions in bank loadings were noted. Experi-
ence suggested the dissolved oxygen response was one part of a system response to
diminished light attenuation by solids and increased light availability at the sedi-
ment-water interface. Our hypothesis was that increased light stimulated benthic
algae and enhanced algal nutrient trapping. Sequestration of nutrients in the sedi-
ments reduced nutrients available to phytoplankton, reduced phytoplankton
production, and diminished carbon deposition to bottom waters. Diminished depo-
sition led to diminished sediment oxygen demand and higher dissolved oxygen. To
test this hypothesis, we conducted a sensitivity analysis in which bank solids loads,
but not nutrient loads, were eliminated. The separation of solids and nutrient loads
is not realistic but allows isolation of solids effects in the sensitivity analysis.
As expected, elimination of solids loads diminished suspended solids concentra-
tion (Figure 13-64) and light attenuation (Figure 13-65). Benthic algae increased in
both abundance and distribution (Figure 13-66 compared to Figure 13-59). The
sensitivity run confirmed a decrease in phytoplankton (Figure 13-67) despite the
greater light availability. The diminished phytoplankton were brought about by
decreased phosphate availability (Figure 13-69). Ammonium (Figure 13-70) and
nitrate (Figure 13-71) increased due to diminished algal demand. The reduced
phytoplankton resulted in increases of more than 0.5 g m"3 bottom dissolved
oxygen in portions of the bay (Figure 13-71).
Response of benthic algal biomass to elimination of bank solids loads varied with
location but doubling of computed biomass was characteristic (Figure 13-72).
Benthic algal production responded in a non-linear fashion so that the proportional
increase in net production exceeded the increase in biomass (Figure 13-73). Sediment
oxygen demand decreased in shoal areas due to algal oxygen production (Figure 13-
74). The demand was not diminished in all years, however. Years with lower algal
biomass (years 4 to 5 and 8 to 10 on our scale, corresponding to 1989 and 1993-
1994) had sediment oxygen demand comparable to the base run. Benthic algal uptake
reduced sediment ammonium release (Figure 13-75). In the colder portions of the
year, ammonium was removed from the water column. As with oxygen demand,
however, the effect on ammonium was not uniform; years with lower algal biomass
exhibited ammonium release similar to corresponding years in the base run. Except
for two years with low algal biomass, nitrate was predominantly removed from the
water column (Figure 13-76) as a result of reduction in bank solids loads. Under base
conditions, sediment-water nitrate flux was roughly balanced between sediment
uptake and release. The activity of benthic algae stripped phosphate from the water
column in most years (Figure 13-77). In contrast, sediments almost exclusively
released phosphorus under the base conditions.
-311 -
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Mainstem Bay Final Calibration - SENS 136
Surface Total Suspended Solids Summer 1990
so
40
*/- •
100 200 300
Kilometers
Figure 13-64. Surface total suspended solids along Chesapeake Bay axis with and
without bank solids loads, summer 1990.
5
__ 4
1 3
'
1
Mainstem Bay Final Calibration - SENS 136
Surface Light Extinction Summer 1 990
-
-
-
V* * HI
^.. '_ I."-". -\ '-- .--'•-"-"'
100 200 300
Kilometers
ainstem Bay SENS 147
urface Light Extinction Summer 1990
1 3
Figure 13-65. Light attenuation along Chesapeake Bay axis with and without bank
solids loads, summer 1990.
-312-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Figure 13-66. Computed summer-average benthic algal
biomass without bank loads, 986.
Mainstem Bay Final Calibration - SENS 136
45
40
35
30
3 20
10
0
Surface Chlorophyll Summer 1990
.
-
-
-
. * *
(» „ • /•'•• --•-.,
E , ''''^
100 200 K>0
Kilometers
, \ i ) 1 '>cliiJl\ 1 , i Kids —————————— -p*
35
30
_j
3 20
15
10
5
:
* '. alioialiun
iinstem Bay SENS 1 47
iff ace Chlorophyll Summer 1990
-
:
•
• '. I .**.'- ,--•-. »-, %.
--•*—•*—- ~r-'"^ 9-~\
0 100 200 300
Kilometers
Figure 13-67. Surface dissolved chlorophyll along Chesapeake Bay axis with and
without bank solids loads, summer 1990.
-313-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Mainstom Bay Final Calibration - SENS 136
Surface Dissolved Phosphate Summer 1990
ttainstem Bay SENS 147
Surface Dissolved Phosphate Summer 1990
Figure 13-68. Surface dissolved phosphate along Chesapeake Bay axis with and
without bank solids loads, summer 1990.
02
0.15
£ oi
005
°{
Mainstem Bay Final Calibration - SENS 136
Surface Ammonium Summer 1990
; ' l\f\
j'' * * V
: s - ,--""•,'• \ ' '
100 200 300
Kilometers
ainslem Bay SENS 147
urface Ammonium Summer 1990
-,<.> Bank I.O;K!S
Figure 13-69. Surface ammonium along Chesapeake Bay axis with and without
bank solids loads, summer 1990.
-314-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
1.75
1 5
1.25
f
0.75
0.5
0?5
>
Mainstem Bay Final Calibration - SENS 136
Surface Nitrate Summer 1990
r
-
r
r
/' \
/
. ,,/• •
j.-.-V**-* r"*r *,,,i,
100 200 300
Kilometers
1
1.5
^) 1
E
0.5
0.25
0
;
j» /' '' S " 1 i '
'•" ', uilnKlUoii
ainstem Bay SENS 147
urface Nitrate Summer 1990
; /. • I
X »
.X"' *
.,... ..-.--- f'~"t 1. j • . *. . - I , 1 . 1 I
100 200 300
Kilometers
Figure 13-70. Surface nitrate along Chesapeake Bay axis with and without bank
solids loads, summer 1990.
Mainstem Bay Final Calibration - SENS 1 36
9
8
7
E
4
3
2
1
Bottom Dissolved Oxygen Summer 1990
L
-
T I
V v
'K ' }••
~ *
* * * » '
1 00 200 300
Kilometers
\
' - I > „ . I •• ] • 1 ^^^^^^^^^^^
O HiUlK l.TKIilS — ~~~^^
?
6
E
4
3
'
1
.
•*• ( ';ilibr;ition
l/lainstem Bay SENS 147
3ottom Dissolved Oxygen Summer 1990
4 •
''••'"•"""'•'• ^ *
'v'"'\ (
', , i
• | *
*
• •
* *
0 100 200 300
Kilometers
Figure 13-71. Bottom dissolved oxygen along Chesapeake Bay axis with and
without bank solids loads, summer 1990.
-315-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Final Calibration - SENS 136
Benthic Algae - CB4
- ov
O 06
SENS 147
Benthic Algae - CB4
u
I 1 rl
0.75 -
05 -
0.75 -
01234
6 7 8 9 10
Figure 13-72. Spatially-averaged benthic algal biomass in shoal areas of
Chesapeake Bay Program Segment CB4 with and without bank solids loads.
Final Calibration - SENS 136
Benthic Algal Net Production - CB4
ENS 147
•nthic Algal Net Production - CB4
Figure 13-73. Spatially-averaged benthic algal net production in shoal areas of
Chesapeake Bay Program Segment CB4 with and without bank solids loads.
-316-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Final Calibration-SENS 136
Sediment Oxygen Demand - CB4
SENS 147
Sediment Oxygen Demand - CB4
8 9 10
Figure 13-74. Daily-average sediment oxygen demand in shoal areas of
Chesapeake Bay Program Segment CB4 with and without bank solids loads.
180
1SO
140
120
D 100
Z 80
E &
20
0
{
Final Calibration - SENS 1 36
Sediment-water NH4 Flux - CB4
" 1 i
^ 1 M ,
U' •' 1
:| '1 ;, 'I I1 i; J
=! 1 '"' i ' ; i / ' ' 1 ' ' ' 1
I \f tr'^y V.y..V'. t!.V..\ :
123456 7 89 10
Years
:NS147
diment-water NH4 Flux - CB4
0 1 2
678
Figure 13-75. Daily-average sediment oxygen demand in shoal areas of
Chesapeake Bay Program Segment CB4 with and without bank solids loads.
-317-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
Final Calibration - SENS 136
Sediment-water NO3 Flux - CB4
i i
7 8 9 10
ENS 147
ediment-water NO3 Flux - CB4
\ I \
Figure 13-76. Daily-average sediment-water nitrate flux in shoal areas of
Chesapeake Bay Program Segment CB4 with and without bank solids loads.
Final Calibration - SENS 1 36
Sediment-water PO4 Flux - CB4
18
16
14
s "
8 10
"E
0. »
E
4
2
0
r
r
r-
^
^ I I '
•ill '| y « !
n y V V V
1
u
I*
0 1 2 3 4 5 ii
Years
y
I1
\ lff\
7 b y 10
;NS147
diment-water PO4 Flux - CB4
Bank Loads
7 8 9 10
Figure 13-77. Daily-average sediment-water phosphate flux in shoal areas of
Chesapeake Bay Program Segment CB4 with and without bank solids loads.
-318-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
An entire volume could be written about this sensitivity run. Perhaps the best
lesson is that water quality improvements analogous to those resulting from
nutrient load reductions can potentially be gained through solids load reductions. It
is also worthwhile to view the model behavior in light of historical reconstruction
of the bay ecosystem (Cooper and Brush 1993). Analysis of sediment cores indi-
cates sedimentation rates have increased significantly in Chesapeake Bay since
European settlement. Increased sedimentation has been accompanied by a change
from an evenly matched planktonic and benthic community to a predominantly
planktonic community. Model results indicate reduction in solids loading will aid in
restoration of the benthic community and drive the bay towards pre-settlement
conditions.
References
Admiraal, W., Bouwman, L., Hoekstra, L., and Romeyn, K. (1983). "Qualitative and quanti-
tative interactions between microphytobenthos and herbivorous meiofauna on a brackish
intertidal mudflat," Internationale Review ofGesamten Hydrobiologie, 68, 175-101.
Batiuk, R., Orth, R., Moore, K., Dennison, W., Stevenson, J., Staver, L., Carter, V., Rybicki,
N., Hickman, R., Kollar, S., Bieber, S., and Heasly, P. (1992). "Chesapeake Bay submerged
aquatic vegetation habitat requirements and restoration targets: A technical synthesis,"
CBP/TRS 83/92, United States Environmental Protection Agency Chesapeake Bay
Program, Annapolis MD.
Boynton, W., Kemp, W., Garber, J., and Barnes, J. (1986). "Ecosystem processes compo-
nent (EPC) level I data report no. 3," UMCEES CBL Ref. No. 85-86, Univesity of
Maryland System Center for Environmental and Estuarine Studies, Solomons MD.
Brush, G. (1989). "Rates and patterns of estuarine sediment accumulation," Limnology and
Oceanography, 34(7), 1235-1246.
Buzzelli, C., (1998). "Dynamic simulation of littoral zone habitats in lower Chesapeake
Bay. I. Ecosystem characterization related to model development," Estuaries, 21(4B), 659-
672
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report El-94-4, US Army Engineer Waterways Experiment Station, Vicks-
burg, MS.
Cerco, C., and Seitzinger, S. (1997). "Measured and modeled effects of benthic algae on
eutrophication in Indian River-Rehoboth Bay, Delaware," Estuaries, 20(1), 231-248.
Cerco, C., and Meyers, M. (2000). "Tributary refinements to the Chesapeake Bay Model,"
Journal of Environmental Engineering, 126(2), 164-174.
Cerco, C., and Moore, K. (2001). "System-wide submerged aquatic vegetation model for
Chesapeake Bay," Estuaries, 24(4), 522-534.
Cerco, C., Johnson, B., and Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model, ERDC TR-02-4, US Army Engineer Research and Development Center, Vicksburg,
MS.
Cohen, R., Dresler, P., Phillips, E., and Cory, R. (1984). "The effect of the Asiatic clam,
Corbicula Fluminea, on phytoplankton of the Potomac River, Maryland." Limnology and
Oceanography, 29, 170-180.
Cooper, S., and Brush, G. (1993). "A 2,500-year history of anoxia and eutrophication in
Chesapeake Ray" Estuaries, 16(3B), 617-626.
-319-
-------
Chapter 13 • Modeling Processes at the Sediment-Water Interface
DiToro, D., and Fitzpatrick, J. (1993). "Chesapeake Bay sediment flux model," Contract
Report EL-93-2, US Army Engineer Waterways Experiment Station, Vicksburg, MS.
DiToro, D. (2001). Sediment Flux Modeling, John Wiley and Sons, New York.
Fisher, T., Carlson, P., and Barber, R. (1982). "Sediment nutrient regeneration in three
North Carolina estuaries," Estuarine, Coastal and Shelf Science, 4, 101- 116.
Gould, D., and Gallagher, E. (1990). "Field measurements of specific growth rate, biomass,
and primary production of benthic diatoms of Savin Hill Cove, Boston," Limnology and
Oceanography, 35, 1757-1770.
HydroQual. (1987). "A steady-state coupled hydrodynamic/water quality model of the
eutrophication and anoxia process in Chesapeake Bay," Final Report, HydroQual Inc.,
Mahwah, NJ.
HydroQual. (2000). "Development of a suspension feeding and deposit feeding benthos
model for Chesapeake Bay," produced by HydroQual Inc. under contract to the U.S. Army
Engineer Research and Development Center, Vicksburg MS.
Jenkins, M., and Kemp, W., (1984). "The coupling of nitrification and denitrification in two
estuarine sediments," Limnology and Oceanography, 29(3), 609-619.
Meyers, M., DiToro, D., and Lowe, S. (2000). "Coupling suspension feeders to the Chesa-
peake Bay eutrophication model," Water Quality and Ecosystem Modeling, 1, 123-140.
Moore, K., Wilcox, D., Orth, R. (2000). "Analysis and abundance of submersed aquatic
vegetation communities in the Chesapeake Bay," Estuaries, 23, 115-127.
Officer, C., Lynch, D., Setlock, G., and Helz, G. (1984). "Recent sedimentation rates in
Chesapeake Bay," The estuary as a filter. V. Kennedy, ed. Academic Press, Orlando FL,
131-157.
Newell, R. (1988). "Ecological Changes in Chesapeake Bay: Are They the Results of Over-
harvesting the American Oyster (Crassostrea Virginica)?" Understanding the Estuary:
Advances in Chesapeake Bay Research. M Lynch and E Krome Eds., Chesapeake Bay
Research Consortium Publication 129, Gloucester Point VA, 536-546
Orth, R., and Moore, K. (1984). "Distribution and abundance of submerged aquatic vegeta-
tion in Chesapeake Bay: An historical perspective." Estuaries (4B):531-540
Pinckney, J., and Zingmark, R. (1993). "Modeling the annual production of intertidal
benthic microalgae in estuarine ecosystems," Journal of Phycology, 29, 396-407.
Rizzo,, W, Lackey, G., and Christian, R. (1992). "Significance of euphotic subtidal sedi-
ments to oxygen and nutrient cycling in a temperate estuary," Marine Ecology Progress
Series, 85, 51-61.
Seitzinger, S., Nixon, S., and Pilson, M. (1984). "Denitrification and nitrous oxide produc-
tion in a coastal marine ecosystem," Limnology and Oceanography, 29(1), 73-83.
Sundback, K. (1986). "What are the benthic microalgae doing on the bottom of Lanholm
Bay?," Ophelia Supplement, 4, 273-286.
Sundback, K., Enoksson, V., Graneli, W, and Peterson, K. (1991). "Influence of sublittoral
microphytobenthos on the oxygen and nutrient flux between sediment and water: A labora-
tory continuous flow study," Marine Ecology Progress Series, 74, 263-279
Westrich, J., and Berner, R. (1984). "The role of sedimentary organic matter in bacterial
sulfate reduction: The G model tested," Limnology and Oceanography, 29, 236-249.
-320-
-------
Dissolved Phosphate
Introduction
An excess of computed dissolved phosphate, especially during summer, has been
a characteristic of the model since the earliest phase (Cerco and Cole 1994). While
tuning the model to effect an overall reduction in computed dissolved phosphate
presents no problem, reducing phosphate in summer while maintaining sufficient
phosphate to support the spring phytoplankton bloom is precarious.
The first-phase model employed two processes to reduce excess phosphate. The
first was variable algal stoichiometry. During periods of plentiful phosphate, the
algal phosphorus-to-carbon ratio increased so that more phosphate was stored as
algal biomass. During periods of phosphate depletion, the phosphorus-to-carbon
ratio decreased so that less phosphate was required to support algal growth.
The second process was co-precipitation of phosphate with iron and manganese.
Iron and manganese are released in dissolved form by Chesapeake Bay sediments
under anoxic conditions. These metals precipitate at the interface between oxic and
anoxic water and during the autumn aeration event. Dissolved phosphate is
removed from the water by sorption to the fresh precipitates.
The variable algal stoichiometry relied on an iterative bisection algorithm. The
algorithm was time-consuming and numerically unstable. Consequently, we
reverted to constant algal composition in the Virginia Tributary Refinements study
(Cerco et al. 2002). Iron and manganese (together referred to as Total Active Metal,
TAM) occupied a slot in the model code originally intended for fixed solids. Since
fixed solids were not modeled in the initial study, this variable was available for
reprogramming as TAM. Fixed solids were modeled in the Virginia Tributary
Refinements and in the present study so that TAM was dropped from the state
variable suite. As a result, in this study, we reverted to the original problem of an
excess of dissolved phosphate, especially during summer.
-321 -
-------
Chapter 14 • Dissolved Phosphate
The present chapter outlines our efforts to bring dissolved phosphate into cali-
bration. As with other sensitivity analyses, the runs compared are not necessarily
sequential and may have multiple differences. Still, the results are illustrative of the
magnitude and effects of various processes.
Dissolved Organic Phosphorus Mineralization
Dissolved phosphate is produced by multiple processes in the model ecosystem.
A classic mechanism is the first-order mineralization of dissolved organic phos-
phorus to dissolved phosphate. The sensitivity of dissolved phosphate to
mineralization was tested by setting this process to zero. This was a reduction in
first-order rate from 0.15 d"1 to 0.0 d"1. Results indicated mineralization of 0.15 d"1
was responsible for 0.005 g PO4-P m~3 in surface waters of the bay during summer
(Figure 14-1). Elimination of mineralization diminished phosphate uniformly
except at the system boundaries. In the central bay (km 100 to 200), computed
phosphate changed from an excess to agreement with observations. In the upper
bay (> km 250) computed phosphate that agreed with observed decreased to below
computed values. In portions of the lower bay (< km 100) computed phosphate was
high and remained high following elimination of mineralization. This run illus-
trated the freedom the modeler has to alter phosphate via specification of
mineralization. The run also illustrated that mineralization rate alone cannot be
used to tune the model unless spatially-variable rates are employed.
0.02
0.015
1
0.01
0.005
0
MAINSTEM BAY, SENS102 (Cruise) ON NEW GRID
Dissolved Inorganic Phosphorus
PO4 Mainbay Surface
~ Julian Day 2066
-
~i V A
- \ y^'U
: \ ^ /vyxr..' r
: • ..." ' ' ...
0 100 200 300
Kilometers
No Mineralization
STEM BAY, SENS107 (Cruise) ON NEW GRID
Ived Inorganic Phosphorus
dainhay Surface
Day 2068
Figure 14-1. Effect of mineralization rate on surface dissolved phosphate. Results
are shown for summer 1990 along Chesapeake Bay axis.
-322-
-------
Chapter 14 • Dissolved Phosphate
Sulfide Oxidizing Bacteria
Gavis and Grant (1986) noted that dissolved phosphate, transported upwards
from anoxic bottom waters of the bay, was consumed by sulfide-oxidizing bacteria
at the interface between the oxic and anoxic layers. We simulated this process in
the model by relating phosphate uptake to oxidation of chemical oxygen demand:
PUPBACT = PSBMAX • ^ • Kcod • COD H 4 1 "l
KHPSB + PO4 ^ >
in which:
PUPBACT = phosphate uptake by sulfide oxidizing bacteria (g P m"3 d"1)
PSBMAX = Stoichiometric relationship of phosphate uptake to bacterial COD
oxidation (g P g-1 COD)
KHPSB = phosphate concentration at which bacterial COD oxidation is halved
(g P m-3)
PO4 = dissolved phosphate (g P m"3)
Kcod = first-order COD oxidation rate, a function of dissolved oxygen and temper-
ature (d"1)
COD = chemical oxygen demand (g O2 equivalents m"3)
The quantity PUBACT was removed from the model phosphate pool. Since
bacteria were not explicitly modeled, the quantity removed from the phosphate
pool was transferred to the labile particulate organic phosphorus pool.
Parameters in the relationship were derived from Gavis and Grant (1986). The
authors noted that consumption of 1 mol of phosphate requires oxidation of 60 mol
of sulfide. This number was converted to model parameter PSBMAX through the
relationship:
1 mol P mol HS" mol Q2 31 g P _ 0.008 g P
60 mol HS" 2 mol 02 32 g 02 mol P g oxygen exerted
Due to phosphorus limitation on the bacteria, only a small fraction of chemical
oxygen demand exertion is bacterially mediated. The phosphorus limitation is
incorporated through parameter KHPSB. Inspection of Figure 6 in Gavis and Grant
(1986) indicates phosphate is drawn down to 0.25 jimol. This value was taken as
KHPSB, converted to model units as 0.008 g nr3.
Model results (Figure 14-2) indicate sulfide oxidizing bacteria can diminish
summer-average surface phosphate by up to 0.01 g P m"3. The primary effect is at
the upper end of the deep trench (km 250) where anoxia is prevalent and phos-
phorus is abundant. At mid-bay (km 150) phosphate removal is roughly 0.005 g P
m"3 while in the lower bay, phosphate removed is perhaps 0.002 g P m"3.
-323-
-------
Chapter 14 • Dissolved Phosphate
MAINSTEM BAY, SENS50A (Cruise) ON NEW GRID
Dissolved Inorganic Phosphorus
PO4 Mainbay Surface
Julian Day 2066 „/ \
Sullidc < )xitli/'iiit
Bacteria
002
0.015
=5,
E
001
0.005
0
MAINSTEM BAY, SENS50L (Cruise) ON NEW GRID
Dissolved Inorganic Phosphorus
PO4 Mainbay Surface
' Julian Day 206S
• ^
-' V x/xy/'x/ .^
\ / " • I
/
v —/ . . • u
• * •
* *
' * •
0 100 200 300
Kilometers
Figure 14-2. Effect of sulfide oxidizing bacteria on surface dissolved phosphate.
Results are shown for summer 1990 along Chesapeake Bay axis. Note change of
vertical scale.
Precipitation
We invoked precipitation in regions of the bay that experience anoxia (Table
14-1). Phosphorus was routed from the phosphate pool to the refractory particulate
organic pool during the fall aeration period. Our relationship mimicked the results
of precipitation although the approach was less mechanistic than the original
formulation involving TAM:
PRECIP = PO4 • PCPMAX
PO4
KHPCP + PO4
PCPDAY
(14-3)
in which:
PRECIP = precipitation rate (g P nr3 d'1)
PCPMAX = first-order precipitation rate (d"1)
KHPCP = phosphate concentration at which precipitation rate is halved (g P nr3)
PCPDAY = a binary variable indicating the reaeration period (0 or 1)
Experience and inspection suggested PCPMAX = 0.1 to 0.2 d'1, KHPCP =
0.0025 g P nr3, and a reaeration interval extending from mid-September to mid-
December.
Results indicated daily precipitation of 10% to 20% of the phosphate pool
effected a tremendous improvement in late-summer and autumn phosphate
(Figure 14-3). Precipitation in late summer did little to impact summer-average
phosphate concentration, however (Figure 14-4).
-324-
-------
Chapter 14 • Dissolved Phosphate
SENS102 ON NEW GRID
Dissolved Inorganic Phosphorus
DIP CB3.3C Surface
CB3.3 Level 1
Precipitation
Base
04 ON NEW GRID
red Inorganic Phosphorus
53.3C Surface
Level 1
Figure 14-3. Effect of precipitation on surface dissolved phosphate at Chesapeake
Bay station CB3.3C.
0.02
0.015
001
0.005
0
MAINSTEM BAY, SENS102 (Cruise) ON NEW GRID
Dissolved Inorganic Phosphorus
PO4 Wlainbay Surface
" Julian Day 2066
- V A
\ AXV« '"'i
1 Vx _, ~j^t • V
0 100 200 300
Kilometers
recipitation
STEM BAY, SENS104 (Cruise) ON NEW GRID
Ived Inorganic Phosphorus
/lainbay Surface
Day 2086
Figure 14-4. Effect of precipitation on surface dissolved phosphate. Results are
shown for summer 1990 along Chesapeake Bay axis.
-325-
-------
Chapter 14 • Dissolved Phosphate
Table 14-1
Chesapeake Bay Program Segments with Phosphate precipitation
Segment
CBS
CB4
CBS
CB6
CB7
EE1
EE2
EE3
LE1
LE2
PCPMAX, d 1
0.2
0.2
0.2
0.1
0.1
0.1
0.1
0.2
0.2
0.2
Segment
LE3
LE4
LE5
RET1
RET2
RETS
RET4
WE4
WT5
PCPMAX, d 1
0.2
0.2
0.1
0.1
0.1
0.1
0.1
0.2
0.2
Summary
The final model calibration incorporated dissolved organic phosphorus mineral-
ization (KDP = 0.15 d"1), uptake by sulfide oxidizing bacteria, and precipitation.
Introduction of the two uptake mechanisms as well as alterations in multiple
parameter values provided a reasonable representation of summer-average phos-
phorus in the surface of the bay, especially during years of dry (1985, Figure 14-5)
to moderate hydrology (1990, Figure 14-6). Considerable excess of computed
phosphate remained present in a wet year (1993, Figure 14-7).
00?
a
0610
Ma in stem Bay final Calibration - SENS 136
Surface Dissolved Phosphate Summer 1985
Figure 14-5. Surface dissolved phosphate along
Chesapeake Bay axis in summer 1985. Final model
calibration.
_
™0.0
Mainstem Bay Final Calibration - SENS 1 36
Surface Dissolved Phosphate Summer 1 990
Figure 14-6. Surface dissolved phosphate along
Chesapeake Bay axis in summer 1990. Final model
calibration.
-326-
-------
Chapter 14 • Dissolved Phosphate
Mainstem Bay Final Calibration - SENS 136
Surface Dissolved Phosphate Summer 1993
100 300
Kilometers
Figure 14-7. Surface dissolved phosphate along
Chesapeake Bay axis in summer 1993. Final model
calibration.
Phosphorus is a limiting nutrient in the mainstem bay in spring while nitrogen is
the primary limiting nutrient in summer (Fisher et al. 1992, Malone et al. 1996,
Fisher et al. 1999). Consequently, our rough approach to modeling dissolved phos-
phate in summer is acceptable. Still, a new phosphorus model seems appropriate,
especially for management of freshwater segments where phosphorus is the more
important nutrient.
Our model (and every other model we know of) ignores phosphate uptake by
heterotrophic bacteria. A cursory search of extensive literature indicates, however,
that bacteria compete with phytoplankton for phosphate (Thingstad et al. 1993) and
may draw phosphate down to limiting levels (Zweifel et al. 1993). Bacterial utiliza-
tion of phosphate in Chesapeake Bay itself has been recognized for decades (Faust
and Correll, 1976). The first step in a new phosphate model is to explicitly recog-
nize phosphate uptake by heterotrophic bacteria. Bacteria do not necessarily have
to be incorporated into the model as a state variable. One reviewer suggested
relating phosphate uptake to organic carbon respiration, which is a bacterial
process.
Multiple studies have indicated that a substantial fraction of particulate phos-
phorus in the bay is of inorganic form (Keefe, 1994; Conley et al. 1995). The
second step in an improved phosphorus model is to explicitly differentiate particu-
late inorganic phosphorus from particulate organic phosphorus through the addition
of a particulate inorganic phosphorus state variable. Simple, linear partitioning of
phosphate to inorganic solids is insufficient. Studies indicate that particulate matter
in the bay has low capacity to adsorb additional phosphorus and that solids do not
buffer the dissolved phosphate concentration (Conley et al. 1995).
In the Virginia Tributary Refinements, we approached the problem of particulate
inorganic phosphorus by assigning a constant composition, 0.1% phosphorus, to
fixed solids. This fraction was settled and resuspended as fixed solids rather than
organic detritus. The approach showed promise in recognizing the distinct nature of
the inorganic fraction and in reproducing the spatial phosphorus distribution
(Figure 14-8). In the present study, we abandoned this approach when we were
-327-
-------
Chapter 14 • Dissolved Phosphate
Total Solids
Summer 85—88
5_ Total Solids Rap Bottom
-40
40 80
Kilometers
Total Phosphorus
Summer 85—88
Total P Rap Bottom
in
5'
q
8'
o
0.
-40
40 80
Kilometers
120
160
Figure 14-8. Median summer bottom total solids and total phosphorus along
Rappahannock River axis 1985-1988. Results from Virginia Tributary Refinements
with solids consisting of 0.1% phosphorus.
experiencing difficulty calibrating phosphorus. Time was unavailable to fully
explore the implications of a new state variable so we fell back to a reliable,
accepted phosphorus cycle.
A third step is to explore the utilization of dissolved organic phosphorus by
bacteria and phytoplankton. From the start (Cerco and Cole 1994), this study has
recognized that bacteria and phytoplankton enhance the mineralization of dissolved
organic phosphorus to inorganic form (e.g. Chrost and Overbeck 1987). Our
approach has been to enhance mineralization during periods of phosphate limita-
tion. The modeled process increases the pool of dissolved phosphate in the water
column. Although the process by which organisms enhance mineralization does not
seem well documented, discussion with colleagues indicates the spatial scale of the
process is limited. That is, mineralization takes place at the cell boundary and is
immediately followed by uptake. From the model standpoint, this is equivalent to
utilization of dissolved organic phosphorus. An approach worth exploring is to
build in a preference function in which organisms utilize phosphate until it is
exhausted and then utilize dissolved organic phosphorus.
A concluding modification is to implement realistic sediment transport
processes. No doubt, a distinct particulate inorganic phosphorus form exists and is
transported along with the solids with which it is associated. Our ability to simulate
solids transport with the present model is limited, however. Consequently, correct
representation of total phosphorus (Figure 14-9) is impossible when solids distribu-
tions cannot be reproduced (Figure 14-10).
-328-
-------
Chapter 14 • Dissolved Phosphate
50 r --
Figure 14-9. Observed surface total solids and total
phosphorus along Potomac River axis, summer 1985.
Total Solids
Summer 85-88
° Total Solids Potomac
o
a-
a>
3ottom
V
-60 0 60 120 180
Kilometers
Total Phosphorus
Summer 85—88
* Total P Potomac Bottom
CJ* O
£
-60
—i—
60
120 180
Kilometers
Figure 14-10. Median summer bottom total solids and total phosphorus along
Potomac River axis 1985-1988. Results from Virginia Tributary Refinements with
solids consisting of 0.1% phosphorus.
-329-
-------
Chapter 14 • Dissolved Phosphate
References
Cerco, C., Johnson, B., and Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model," ERDC TR-02-4, U.S. Army Engineer Research and Development Center, Vicks-
burg, MS.
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," Technical Report EL-94-4, U.S. Army Engineer Waterways Experiment Station,
Vicksburg, MS.
Chrost, R., and Overbeck, J. (1987). "Kinetics of alkaline phosphatase activity and phos-
phorus availability for phytoplankton and bacterioplankton in Lake Plubsee (north German
eutrophic lake), Microbial Ecology, 13, 229-248.
Conley, D., Smith, W., Cornwell, J., Fisher, T. (1995). "Transformation of particle-bound
phosphorus at the land-sea interface," Estuarine, Coastal and Shelf Science, 40, 161-176.
Faust, M., and Correll, D. (1976). "Comparison of bacterial and algal utilization of
orthophosphate in an estuarine environment," Marine Biology, 34, 151-162.
Fisher, T., Gustafson, A., Sellner, K., Lacouture, R., Haas, L., Wetzel, R., Magnien, R.,
Everitt, D., Michaels, B., and Karrh, R. (1999). "Spatial and temporal variation of resource
limitation in Chesapeake Bay," Marine Biology, 133, 763-778.
Fisher, T., Peele, E., Ammerman, J., and harding, L. (1992). "Nutrient limitation of phyto-
plankton in Chesapeake Bay," Marine Ecology Progress Series, 82, 51- 63.
Gavis, J, and Grant, V. (1986). "Sulfide, iron, manganese, and phosphate in the deep water
of the Chesapeake Bay during anoxia," Estuarine, Coastal and Shelf Science, 23, 451-462.
Keefe, C. (1994). "The contribution of inorganic compounds to the paniculate carbon,
nitrogen, and phosphorus in suspended matter and surface sediments of Chesapeake Bay,"
Estuaries, 17(1B), 122-130.
Malone, T., Conley, D., Fisher, T., Gilbert, P., Harding, and Sellner, K. (1996). "Scales of
nutrient-limited phytoplankton productivity in Chesapeake Bay," Estuaries, 19, 371-385.
Thingstad, T., Skjodal, E., and Bohne, R. (1993). "Phosphorus cycling and algal- bacterial
competition in Sandsfjord, western Norway," Marine Ecology Progress Series, 99, 239-259.
Zweifel, U., Norrman, B., and Hagstrom, A. (1993). "Consumption of dissolved organic
carbon by marine bacteria and demand for inorganic nutrients," Marine Ecology Progress
Series, 101, 23-32.
-330-
-------
Statistical Summary
of Calibration
Introduction
The calibration of the model involved the comparison of hundreds of thousands
of observations with model results in various formats. Comparisons involved
conventional water quality data, process-oriented data, and living-resources obser-
vations. The graphical comparisons produced thousands of plots which cannot be
assimilated in their entirety. Evaluation of model performance requires statistical
and/or graphical summaries of results. We present summaries here for major water
quality constituents in the mainstem bay and western tributaries. Additional graph-
ical comparisons are available on the CD-ROM that accompanies this report.
Methods
No standard set of model performance statistics exists. We employ summary
statistics that were developed as part of our initial Chesapeake Bay model study
(Cerco and Cole 1994). Use of a consistent set of statistics facilitates comparisons
with earlier model versions and with applications to other systems.
Statistics computed were mean error, absolute mean error, and relative error:
(15-1)
N
N
(15-2)
o
(15-3)
-331 -
-------
Chapter 15 • Statistical Summary of Calibration
in which
ME = mean error
AME = absolute mean error
RE = relative error
O = observation
P = prediction
N = number of observations.
The mean error describes whether the model over-estimates or under-estimates
the observations, on average. The mean error can achieve its ideal value, zero,
while large discrepancies exist between individual observations and computations.
The absolute mean error is a measure of the characteristic difference between indi-
vidual observations and computations. An absolute mean error of zero indicates the
model perfectly reproduces each observation. The relative error is the absolute
mean error normalized by the mean concentration. Relative error provides a
statistic suitable for comparison between different variables or systems.
Quantitative statistics were determined through comparison of model computa-
tions with observations from 7 to 27 stations sampled at approximately monthly
intervals. The number of model-data comparisons ranged from roughly 400 to
17,000 (Table 15-1). For salinity, total nitrogen, and total phosphorus, comparisons
were made with surface and bottom samples. For chlorophyll and light attenuation,
surface samples only were examined. Dissolved oxygen comparisons were
restricted to bottom samples collected in summer. We formed this restriction for
two reasons. First, summer, bottom dissolved oxygen is of primary interest to
management. Second, dissolved oxygen at the surface during summer and
throughout the water column during colder periods is near saturation. The
predictive ability of the model is largely determined by the ability to calculate the
Table 15-1 Number of Stations and Observations in Statistical Summaries
# Stations
Chlorophyll
Dissolved
Oxygen
Light
Attenuation
Salinity
Total Nitrogen
Total
Phosphorus
Mainstem Bay
46
8216
2745
8351
17166
16700
17000
James
11
2087
605
1925
4300
4335
4331
York
12
2188
669
2108
4437
4480
4483
Rappahannoc
k
7
1548
462
1552
3233
3306
3296
Potomac
15
1971
718
2223
4056
4184
4253
Patuxent
7
1277
424
1330
2690
1511
2663
-332-
-------
Chapter 15 • Statistical Summary of Calibration
saturation dissolved oxygen concentration. Computation of summer, bottom
dissolved oxygen requires that the model correctly portray interactions of physical
and biological processes. Statistics based on summer, bottom dissolved oxygen
provide more realistic indication of model performance.
Our graphical summaries are in the form of cumulative distribution plots.
Creation of the plots first requires pairing in space and time of observations and
computations. These were from the same stations used in the statistical summaries.
(The dissolved oxygen and chlorophyll graphical summaries include surface and
bottom samples throughout the year.) Observations were paired with daily average
computations in the cell corresponding to sample location and depth. Next, the
observations and computations were individually sorted from smallest to largest.
The sorted arrays were divided into quantiles and plotted as cumulative distribu-
tions. A point on the line in x-y space indicates the percentage of observations or
computations (y-axis) less than the indicated concentration (x-axis). The 50th
percentile indicates the median value. Perfect correspondence is indicated when the
cumulative distribution of modeled values exactly overlays the cumulative distribu-
tion of observed values.
Statistics of Present Calibration
Examination of statistical summaries (Table 15-2) requires a good deal of judge-
ment and interpretation. Generalizations and distinctions are not always possible.
One clear pattern is that the model overestimates, on average, surface chlorophyll.
The overestimation ranges from less than 1 mg m~3 to more than 2 mg m~3. The
model consistently underestimates salinity although the mean error is always less
than 1 ppt. For all systems except the Potomac, computed mean summer, bottom
dissolved oxygen is within 1 g m~3 of the observed average. In the Potomac,
computed mean summer bottom dissolved oxygen is almost 2 g m~3 higher than
observed. Careful examination of model results indicates the region of greatest
computed excess is in the tidal fresh portion of the river, where observed bottom
dissolved oxygen exceeds 5 g m~3. Excessive computed dissolved oxygen, surface
and bottom, is a characteristics of the present model in most tidal freshwater
regions. Except in the James, the model underestimates mean total phosphorus
concentration. Underestimation of total phosphorus has been a characteristic of the
model since the earliest application (Cerco and Cole 1994). We originally attributed
the shortfall to omission of bankloads. In this version we include bankloads of
phosphorus but they are difficult to estimate accurately. The model also omits
resuspension of particulate phosphorus and has difficulty reproducing the concen-
tration of particulate phosphorus in the turbidity maximums. In the James, we
attribute the excess computed phosphorus to uncertainty in the large point-source
and distributed loads to this tributary.
Examination of relative errors (Table 15-2, Figure 15-1) indicates that chloro-
phyll has the greatest error, salinity the least. Relative error in chlorophyll
prediction is 60% to 80% while relative error in salinity prediction is 10% to 20%.
The chlorophyll error reflects the difficulty in computing this dynamic biological
component which can attain unlimited magnitude. In contrast, salinity is purely
physical and is bounded at the upper end by oceanic concentration. The remaining
components are in the mid-range, 30% to 50%, with total phosphorus, perhaps
exhibiting slightly higher relative error. The higher error in phosphorus reflects the
-333-
-------
Chapter 15 • Statistical Summary of Calibration
Table 15-2 Statistical Summary of Calibration 1985-1994
Surface Chlorophyll, ug/L
Mean Error
Absolute Mean Error
Relative Error
Summer, Bottom Dissolved
Oxygen, mg/L
Mean Error
Absolute Mean Error
Relative Error
Light Attenuation, 1/m
Mean Error
Absolute Mean Error
Relative Error
Salinity, ppt
Mean Error
Absolute Mean Error
Relative Error
Total Nitrogen, mg/L
Mean Error
Absolute Mean Error
Relative Error
Total Phosphorus, mg/L
Mean Error
Absolute Mean Error
Relative Error
Mainstem Bay
-0.53
5.01
58.4
Mainstem Bay
0.32
1.47
35.7
Mainstem Bay
0.02
0.36
35.3
Mainstem Bay
0.71
1.97
11.8
Mainstem Bay
0.04
0.17
24.3
Mainstem Bay
0.005
0.014
37.6
James
-2.05
9.29
75.7
James
-0.09
2.43
36.6
James
-0.21
0.97
43.7
James
0.11
2.01
31.2
James
-0.16
0.42
44.6
James
-0.021
0.069
63.8
York
-1.68
4.71
60.1
York
0.37
1.18
22.8
York
0.09
0.84
41.9
York
0.95
1.84
14.5
York
0.01
0.23
33.1
York
0.012
0.036
49.2
Rappahannock
-2.55
8.22
81.4
Rappahannock
0.62
1.93
35.4
Rappahannock
-0.17
0.89
42.3
Rappahannock
0.01
1.49
18.3
Rappahannock
0.14
0.28
33.8
Rappahannock
0.001
0.036
52.6
Potomac
-1.85
7.45
80.2
Potomac
-1.31
2.13
40.5
Potomac
-0.02
1.03
45.2
Potomac
0.45
0.97
22.5
Potomac
0.32
0.61
31.9
Potomac
0.032
0.053
58.9
Patuxent
-1.53
8.15
65.4
Patuxent
-0.92
1.74
39.3
Patuxent
-0.20
0.84
38.4
Patuxent
0.16
1.69
17.5
Patuxent
-0.13
0.43
41.5
Patuxent
0.041
0.047
47.6
aforementioned difficulties in evaluating loads, in simulating resuspension, and in
representing particulate phosphorus transport.
The mainstem bay is clearly superior in computations of salinity, total nitrogen,
and total phosphorus. The James River stands out as demonstrating the highest rela-
tive error in these components. We partially attribute the greater accuracy in the
mainstem to the relatively dense computational grid in this region. An additional,
and probably more significant influence, is that the mainstem is dominated by
internal processes while the tributaries are strongly influenced by point-source and
distributed loads. The point-source loads are incompletely described, especially in
the early years of the simulation and in the Virginia tributaries. Below-fall-line
distributed loads cannot be measured; they can only be computed by the watershed
-334-
-------
Chapter 15 • Statistical Summary of Calibration
90 -
80 -
70 -
60 -
|50-
0
£40-
30 -
20 -
10 -
0 -
H Mainstem Bay
• James
DYork
D Rappahannock
• Potomac
D Patuxent
I
I
1
f
Chlorophyll Dissolved Light
Oxygen Attenuation
1
1
1
Salinity
r-i
1
|
Total Total
'-i
Nitrogen Phosphorus
Figure 15-1. Relative Error in mainstem bay and western tributaries.
model. We believe the uncertain loads, discharged into the constrained volumes of
the tributaries, are the major reason for higher relative error in the tributaries. The
James River stands out in this regard.
Statistics of Model Improvements
The Chesapeake Bay model has gone through three stages. The first was the
initial 4,000-cell model that was used in the 1992 re-evaluation (Cerco and Cole
1994). The second stage was the tributary refinements in which segmentation was
substantially increased and living resources were introduced (Cerco et al. 2002).
The present 2002 model is the third stage. This stage involved further re-segmenta-
tion and re-calibration. As part of the tributary refinements documentation, we
summarized model performance at several stages in the mainstem bay and Virginia
tributaries. We can now supplement that summary with the performance of the
present model phase (Table 15-3, 15-4). This summary is based on the years 1985-
1986 only. These years are common to the three simulations. Consequently, the
summary statistics presented for the 2002 model differ from those presented earlier.
Statistics for summer, bottom dissolved oxygen are superior for the present
model, especially when compared to the initial model application. Mean error is
closer to the ideal value of zero everywhere and relative error is diminished every-
where except the James River.
Clear improvements are not apparent in the salinity statistics, which is inter-
esting considering the two new grid resegmentations and the high level of attention
devoted to the present effort. If anything, the salinity statistics show a decrease in
performance in the mainstem bay. We suspect a trade-off has occurred between
dissolved oxygen and salinity. In the initial model application, the hydrodynamic
model was calibrated and the results were passed to the water quality team. In the
-335-
-------
Chapter 15 • Statistical Summary of Calibration
Table 15-3
Mean Errors in Three Model Phases
Salinity
4000 Cell
VA Tribs
2002 Model
Mainstem
-0.06
-0.62
1.69
James
-0.95
0.55
0.32
Rappahannock
2.43
-1.64
1.14
York
0.08
-0.94
1.68
Surface
Chlorophyll
4000 Cell
VA Tribs
2002 Model
Mainstem
-1.59
-1.19
-0.32
James
0.24
-4.84
-2.47
Rappahannock
0.54
2.25
-0.14
York
-1.19
-1.66
-1.46
Light
Attenuation
4000 Cell
VA Tribs
2002 Model
Mainstem
-0.146
-0.206
0.065
James
-0.329
-0.425
0.009
Rappahannock
-0.169
-0.599
0.135
York
0.021
0.222
0.125
Total Nitrogen
4000 Cell
VA Tribs
2002 Model
Mainstem
-0.003
0.011
0.025
James
0.108
-0.174
-0.205
Rappahannock
0.197
0.128
0.098
York
0.057
0.092
0.004
Total
Phosphorus
4000 Cell
VA Tribs
2002 Model
Mainstem
0.012
0.0091
0.0115
James
-0.0077
-0.0456
-0.0534
Rappahannock
0.0265
0.0043
-0.0008
York
0.0146
0.0149
-0.0025
Summer,
Bottom,
Dissolved
Oxygen
4000 Cell
VA Tribs
2002 Model
Mainstem
-0.89
0.58
-0.03
James
-1.88
0.26
-0.53
Rappahannock
-2.96
0.81
0.72
York
-0.48
-0.08
0.40
-336-
-------
Chapter 15 • Statistical Summary of Calibration
Table 15-4
Relative Error in Three Model Phases
Salinity
4000 Cell
Step 2
2002 Model
Mainstem
9.1
9.5
13.1
James
19.1
19.4
25.4
Rappahannock
24.0
18.3
16.5
York
17.4
11.6
15.6
Surface
Chlorophyll
4000 Cell
VA Tribs
2002 Model
Mainstem
61.6
57.6
59.2
James
74.1
78.6
66.4
Rappahannock
83.4
64.8
70.8
York
64.0
66.2
49.1
Light
Attenuation
4000 Cell
VA Tribs
2002 Model
Mainstem
36.5
45.2
38.5
James
32.6
59.8
39.7
Rappahannock
32.8
57.6
41.0
York
39.9
46.6
44.0
Total Nitrogen
4000 Cell
VA Tribs
2002 Model
Mainstem
22.1
21.3
21.9
James
31.9
46.0
48.5
Rappahannock
32.5
31.9
28.6
York
20.8
20.4
28.1
Total
Phosphorus
4000 Cell
VA Tribs
2002 Model
Mainstem
42.5
38.5
41.5
James
49.3
67.9
78.3
Rappahannock
53.4
41.2
41.3
York
41.0
43.9
47.0
Summer,
Bottom,
Dissolved
Oxygen
4000 Cell
VA Tribs
2002 Model
Mainstem
44.9
31.6
27.9
James
31.7
38.1
40.5
Rappahannock
63.9
46.6
32.2
York
41.4
25.9
22.7
-337-
-------
Chapter 15 • Statistical Summary of Calibration
present application, the hydrodynamic and water quality models were calibrated
interactively. The salinity computation in the hydrodynamic model was weighed
against the dissolved oxygen computation in the water quality model. We placed
great emphasis on the dissolved oxygen computations. Apparently the devotion to
dissolved oxygen diminished the performance of the model with regard to salinity.
The performance of the model with regard to light attenuation deteriorated from
the initial application to the Virginia Tributary application. In the initial application,
regression was used to specify attenuation as a function of location and runoff.
(Runoff was used as a surrogate for sediment load). This specification was highly
accurate, if not predictive. In the Virginia Tributaries phase, we made the first
attempt to compute attenuation as a function of computed solids concentration.
This model was intended to compute response to management control actions. The
predictive model initially was less accurate than the regression model. In the
present phase, we lavished attention on the solids calculation with the result that
attenuation was computed at levels comparable to or better than a regression model.
Consistent montonic changes in chlorophyll, total nitrogen, and total phosphorus
are difficult to discern. Mean and relative errors have varied but are now essentially
where they were in the earlier model stages.
The statistical analysis that accompanied the Tributary Refinements examined
multiple stages of model calibration. We noted then that clear, monotonic improve-
ments in model representations of nitrogen, phosphorus, and chlorophyll were
difficult to discern. We declared that model revisions contributed realism and confi-
dence rather than quantitative improvements. Our conclusions then remain valid
now. Improvements that have been made to the model over 15 years include:
• Increasing resolution from a grid of 4,000 cells to a grid of 13,000 cells
• Improved vertical resolution in shallow-water areas
• Incorporation of living resources including submerged aquatic vegetation,
zooplankton, and benthos
• Accurate modeling of primary production
• Modeling of suspended solids
• Predictive modeling of light attenuation
We have no doubt the model is of more use to management than it has ever been
although we may have reached a plateau in terms of quantitative model
performance.
Comparison with Other Applications
No standard criteria exist for judging acceptable model performance. One
approach is to compare performance with similar statistics from other model appli-
cations. Statistics comparable to other systems at least indicate the model is in the
performance mainstream. We compared relative error in the 2002 Chesapeake Bay
application with two recent model applications, Florida Bay (Cerco et al. 2000) and
the lower St. Johns River, Florida (Tillman et al. 2003). Florida Bay is a shallow
sub-tropical lagoon. The St. Johns is a partially- to well-mixed estuary with a
substantial tidal freshwater extent.
-338-
-------
Chapter 15 • Statistical Summary of Calibration
All models indicate chlorophyll has the greatest relative error (Table 15-5,
Figure 15-2). This finding is consistent with our earlier indication derived from the
mainstem bay and tributaries (Figure 15-1). The Chesapeake Bay application is
centrally located within the relative error in the other two systems. Relative errors
for Chesapeake Bay total nitrogen and total phosphorus are also comparable to the
other systems.
Table 15-5. Relative Error in Three Estuarine
Applications
Chlorophyll,
ug/L
Dissolved
Oxygen, mg/L
Light
Attenuation, 1/m
Salinity, ppt
Total Nitrogen,
mg/L
Total
Phosphorus,
mg/L
Chesapeake
Bay
58.4
35.7
35.3
11.9
24.3
37.6
Lower St.
Johns River
49.3
9.3
15.6
27.6
29
26.9
Florida Bay
72
7.7
4.7
38.9
31.3
• Chesapeake Bay
• Lower St. Johns River
D Florida Bay
Chlorophyll Dissolved Light Salinity
Oxygen Attenuation
Total Total
Nitrogen Phosphorus
Figure 15-2. Relative error in three model applications.
-339-
-------
Chapter 15 • Statistical Summary of Calibration
Significant differences exist for dissolved oxygen, light attenuation, and salinity.
The St. Johns River and Florida Bay have much lower relative error in computed
dissolved oxygen. The discrepancy is due to the way the statistics were computed.
Chesapeake Bay was restricted to summer, bottom dissolved oxygen while the
other systems were not. Florida Bay is a shallow well-mixed lagoon in which
dissolved oxygen occurs in the range 6 to 9 g m~3. The preponderance of observa-
tions in the St. Johns were from the surface. The estuary demonstrates little
dissolved-oxygen stratification so that the few subsurface observations showed
minor departure from the surface observations. The performance in Florida Bay and
the St. Johns validates our observation that dissolved oxygen statistics that include
surface and cold-weather samples are biased by the tendency for dissolved oxygen
to equilibrate with the atmosphere. The denominator in the relative error computa-
tion, essentially saturation concentration, is also larger than the mean bottom
dissolved oxygen that forms the denominator in the Chesapeake Bay statistics.
The Chesapeake Bay and St. Johns applications use identical models for solids
and light attenuation. The different statistics reflect the components that make up
attenuation. In the St. Johns, the major component is color, represented in the
model as refractory dissolved organic carbon. Computation of attenuation in the St.
Johns relies on accurate observations of carbon loading and boundary conditions.
Once the boundary conditions are specified, the refractory dissolved organic carbon
passes through the system with little alteration except dilution. Consequently,
computation of light attenuation is relatively simple. In Chesapeake Bay, the prin-
ciple components of light attenuation are fixed and volatile solids. These
components are highly dynamic and depend on loading, transport and internal
kinetics. The higher relative error in Chesapeake Bay light attenuation reflects the
greater difficulty in computing solids versus refractory dissolved organic carbon.
The salinity statistics are most interesting. Florida Bay has the least relative error
yet we don't regard this application as particularly successful. The statistics are
biased by a large number of samples in the perimeter of the bay in which salinity is
nearly constant between 35 and 40 ppt. The St. Johns River has a well-validated
hydrodynamic model yet relative error there is apparently the greatest of all. We
cannot ascertain the origin of the higher error relative to Chesapeake Bay. Our
opinion is that the St. Johns application (Sucsy and Morris 2001) is comparable in
quality to the Chesapeake Bay application. The lesson here is that statistics can be
deceptive and must be employed judiciously, in concert with other quantitative and
qualitative performance measures.
Graphical Performance Summaries
Dissolved Oxygen
Computed and observed dissolved oxygen distributions in the mainstem bay
virtually superimpose above 6 g m~3 (Figure 15-3). Computed dissolved oxygen in
the range 2 to 6 g m~3 comprises a greater fraction of the population than observed
while observations below 2 g m~3 out-number corresponding calculations. The
James (Figure 15-4), Rappahannock (Figure 15-6) and Potomac (Figure 15-7)
Rivers show an interesting crossover around 6 g m~3. The crossover indicates the
-340-
-------
Chapter 15 • Statistical Summary of Calibration
fractions of computed and observed populations below 6 g m"3 (25%) and above 6
g m"3 (75%) are identical. When observations are below 6 g m"3, the corresponding
computation tends to be low. When observations are above 6 g m"3, the correspon-
ding computation tends to be high. The model tends to underestimate dissolved
oxygen throughout the distribution in the York River (Figure 15-5). The opposite
phenomenon occurs in the Patuxent; the model overestimates dissolved oxygen
throughout the distribution (Figure 15-8).
Final Calibration - SENS 136
Dissolved Oxygen
Mainstem Bay
Model
Percent Less than
Final Calibration - SENS 136
Dissolved Oxygen
20 h James River
Mode!
y
Percent Less than
Figure 15-3. Cumulative distribution plot of
dissolved oxygen in mainstem bay.
Figure 15-4. Cumulative distribution plot of
dissolved oxygen in James River.
Final Calibration - SENS 136
Dissolved Oxygen
,5t York River
Percent Less than
Final Calibration - SENS 136
Dissolved Oxygen
Rappahannock River
Percent Less than
Figure 15-5. Cumulative distribution plot of
dissolved oxygen in York River.
Figure 15-6. Cumulative distribution plot of
dissolved oxygen in Rappahannock River.
-341 -
-------
Chapter 15 • Statistical Summary of Calibration
Final Calibration - SENS 136
Dissolved Oxygen
Potomac River
Model
Percent Less than
Final Calibration - SENS 136
Dissolved Oxygen
Patuxent River
Model
Percent Less than
Figure 15-7. Cumulative distribution plot of
dissolved oxygen in Potomac River.
Figure 15-8. Cumulative distribution plot of
dissolved oxygen in Patuxent River.
Salinity
The graphical distributions indicate excellent representations of salinity in the
mainstem bay (Figure 15-9) and Potomac River (Figure 15-13). We believe the
larger number of grid elements and major role of gauged runoff versus ungauged
below-fall-line flow contribute to the computed accuracy in these systems. Salinity
computations in the James (Figure 15-10), York (Figure 15-11), Rappahannock
(Figure 15-12) and Patuxent Rivers are less accurate. This characteristic has been
with the model since the earliest stages and remains after two grid re-segmenta-
tions. Further re-segmentation should be of benefit but we have to recognize the
role of local processes and effects that may never be represented in a system-wide
model.
Final Calibration - SENS 136
Salinity
Mainstem Bay
Percent Less than
(_ IS
0.
a.
Final Calibration - SENS 136
Salinity
James River
Model
Percent Less than
Figure 15-9. Cumulative distribution plot of
salinity in mainstem bay.
Figure 15-10. Cumulative distribution plot of
salinity in James River.
-342-
-------
Chapter 15 • Statistical Summary of Calibration
0.
0.
Final Calibration - SENS 136
Salinity
York River
Model
Percent Less than
Final Calibration - SENS 136
Salinity
Rappahannock River
Model
Percent Less than
Figure 15-11. Cumulative distribution plot
of salinity in York River.
Figure 15-12. Cumulative distribution plot of
salinity in Rappahannock River.
Final Calibration - SENS 136
Salinity
Potomac River
Model
Percent Less than
Final Calibration - SENS 136
Salinity
Patuxent River
Model
Percent Less than
Figure 15-13. Cumulative distribution plot
of salinity in Potomac River.
Figure 15-14. Cumulative distribution plot of
salinity in Patuxent River.
Chlorophyll
Most of the chlorophyll distributions indicate a crossover around 10 to 20 mg
m~3. In the mainstem bay (Figure 15-15), James (Figure 15-16), Potomac (Figure
15-19) and Patuxent (Figure 15-20) Rivers, the model tends to underestimate the
very highest chlorophyll concentrations. This property is characteristic of the model
and is attributed to the unlimited extremes that observed chlorophyll observations
can attain. The model is restricted to concentrations averaged over space (cell
volume) and time (time step and averaging interval). The observations are subject
to sub-grid scale processes that create local accumulations and are also affected by
analytical variance. Except for the highest observations, the model tends to over-
compute chlorophyll. In systems such as the mainstem bay and Potomac River, the
overestimation is negligibly small while the overestimation is significant in the
Rappahannock (Figure 15-18) and Patuxent Rivers (Figure 15-20).
-343-
-------
Chapter 15 • Statistical Summary of Calibration
Final Calibration - SENS 136
Chlorophyll
Ma in stem Bay
Mode!
Percent Less than
Figure 15-15. Cumulative distribution plot
of chlorophyll in mainstem bay.
Final Calibration - SENS 136
Chlorophyll
- James River
Model
25 50 75 100
Percent Less than
Figure 15-16. Cumulative distribution plot of
chlorophyll in James River.
Final Calibration - SENS 1 36
10
10'
Chlorophyll
h York River
:
i
, : , , 1 , , ; , 1 , i : 1 , _ 1
'J^ bij /h 100
Percent Less than
Figure 15-17. Cumulative distribution plot
of chlorophyll in York River.
Final Calibration - SENS 136
Chlorophyll
Rappahannock River
Mode!
Percent Less than
Figure 15-18. Cumulative distribution plot of
chlorophyll in Rappahannock River.
Final Calibration - SENS 136
Chlorophyll
Potomac River
Model
Percent Less than
Figure 15-19. Cumulative distribution plot
of chlorophyll in Potomac River.
-344-
Final Calibration - SENS 136
Chlorophyll
t Patuxent River
Model
Percent Less than
Figure 15-20. Cumulative distribution plot of
chlorophyll in Patuxent River.
-------
Chapter 15 • Statistical Summary of Calibration
Light Attenuation
The light attenuation computations are remarkably successful, especially consid-
ering the relatively crude solids and attenuation algorithms in the model. The
computed and observed distributions in the mainstem bay (Figure 15-21), James
(Figure 15-22) and Potomac Rivers (Figure 15-25) are virtually congruent. In the
York (Figure 15-23), Rappahannock (Figure 15-24) and Patuxent (Figure 15-26)
Rivers, the upper halves of the computed and observed distributions tend to super-
impose. In the lower portions of the distributions, below 1 to 2 nr1, the model tends
to overestimate observed attenuation.
Final Calibration - SENS 136
Light Extinction
IS I Mainstem Bay
;
Percent Less than
Final Calibration - SENS 136
Light Extinction
James River
Model
Percent Less than
Figure 15-21. Cumulative distribution plot
of light attenuation in mainstem bay.
Figure 15-22. Cumulative distribution plot of
light attenuation in James River.
9 -
8 -
7 -
B r
Final Calibration - SENS 136
Light Extinction
York River
Model
Percent Less than
Final Calibration - SENS 136
Light Extinction
F RappahannockRiver
Mode!
?
Percent Less than
Figure 15-23. Cumulative distribution plot
of light attenuation in York River
Figure 15-24. Cumulative distribution plot of
light attenuation in Rappahannock River.
-345-
-------
Chapter 15 • Statistical Summary of Calibration
2
e
I 1
Final Calibration - SENS 136
Light Extinction
Potomac River
Model
Percent Less than
Final Calibration - SENS 136
Light Extinction
Patuxent River
Model
Percent Less than
Figure 15-25. Cumulative distribution plot
of light attenuation in Potomac River.
Total Nitrogen
Figure 15-26. Cumulative distribution plot of
light attenuation in Patuxent River.
The James (Figure 15-28), York (Figure 15-29) and Patuxent (Figure 15-32)
Rivers exhibit the familiar crossover pattern. In these tributaries, the model tends to
overestimate the highest nitrogen concentrations and underestimate the lower
concentrations. The crossover occurs around 0.8 g m~3. We suspect this behavior is
attributable to occasional enormous concentrations that are computed near the fall
lines during storm events. The water quality and hydrodynamic models use gauged
flows while the watershed model uses its own computed flow. Discrepancies
between gauged and computed flows can cause wild fluctuations during the large
loadings associated with storm events. The mainstem bay (Figure 15-27) and
Potomac River (Figure 15-31) exhibit two crossovers. As with the preceding three
tributaries, the model exceeds the very highest observed concentrations. The model
also exceeds the lowest observations, less than about 0.6 g m~3. In the portions of
the distributions between the extremes, observations tend to exceed modeled
concentrations. The Rappahannock River (Figure 15-30) stands out as one in which
total nitrogen is under-computed throughout the distribution.
Final Calibration - SENS 136
Total Nitrogen
Mainstem Bay
Model
Percent Less than
Figure 15-27. Cumulative distribution plot
of total nitrogen in mainstem bay.
Final Calibration - SENS 136
Total Nitrogen
James River
Model
Percent Less than
Figure 15-28. Cumulative distribution plot of
total nitrogen in James River.
-346-
-------
Chapter 15 • Statistical Summary of Calibration
I
Final Calibration - SENS 136
Total Nitrogen
York River
Model
Percent Less than
Figure 15-29. Cumulative distribution plot
of total nitrogen in York River.
Final Calibration - SENS 136
Total Nitrogen
Rappahannock River
Model
Percent Less than
Figure 15-30. Cumulative distribution plot of
total nitrogen in Rappahannock River.
|
Final Calibration - SENS 136
Total Nitrogen
Potomac River
Model
Percent Less than
Final Calibration - SENS 136
Total Nitrogen
Patuxent River
Model
Percent Less than
Figure 15-31. Cumulative distribution plot
of total nitrogen in Potomac River.
Figure 15-32. Cumulative distribution plot of
total nitrogen in Patuxent River.
Total Phosphorus
As noted in the statistical summaries, the model usually underestimates total
phosphorus. The graphs indicate the model matches the very highest 1 to 2% of the
observations. The underestimation is almost universal in the lower 98% of the
observed distribution. The mainstem bay (Figure 15-33) and the Rappahannock
River (Figure 15-36) suggest that the model overestimates the lowest total phos-
phorus concentrations but interpretation is clouded in the Rappahannock by
reporting increments and detection levels. We have attributed underestimation of
computed total phosphorus to uncertainty in bank loading, to lack of a model resus-
pension mechanism, and to inability to represent concentration of particulate
phosphorus in turbidity maximums.
-347-
-------
Chapter 15 • Statistical Summary of Calibration
Final Calibration - SENS 136
Total Phosphorus
Mainstem Bay
Model
Percent Less than
Figure 15-33. Cumulative distribution plot
of total phosphorus in mainstem bay.
O)
E{
Final Calibration - SENS 136
Total Phosphorus
James River
Model
Percent Less than
Figure 15-34. Cumulative distribution plot of
total phosphorus in James River.
i10'
Final Calibration - SENS 136
Total Phosphorus
York River
Model
Percent Less than
Figure 15-35. Cumulative distribution plot
of total phosphorus in York River.
o :\
025
07
0.10
0 I
=!
Final Calibration - SENS 136
Total Phosphorus
" Rappahannock River
• '
;'
-
.. > , . 1 , • , : t , , . , 1 , ( 1
25 50 75 100
Percent Less than
Figure 15-36. Cumulative distribution plot of
total phosphorus in Rappahannock River.
I
Final Calibration - SENS 136
Total Phosphorus
Potomac River
Model
Percent Less than
Figure 15-37. Cumulative distribution plot
of total phosphorus in Potomac River.
0.-1
0,35
03
025
0,2
£
Final Calibration - SENS 136
Total Phosphorus
Patuxent River
Model
Percent Less than
Figure 15-38. Cumulative distribution plot of
total phosphorus in Patuxent River.
-348-
-------
Chapter 15 • Statistical Summary of Calibration
References
Cerco, C., and Cole, T. (1994). "Three-dimensional eutrophication model of Chesapeake
Bay," TR EL-94-4, US Army Engineer Waterways Experiment Station, Vicksburg MS.
Cerco, C., Bunch, B., Teeter, A., and Dortch, M. (2000). "Water quality model of Florida
Bay," ERDC TR-00-10, US Army Engineer Research and Development Center, Vicksburg
MS.
Cerco, C., Johnson, B., and Wang, H. (2002). "Tributary refinements to the Chesapeake Bay
model," ERDC TR-02-4, US Army Engineer Research and Development Center, Vicksburg
MS.
Sucsy, P., and Morris, F. (2001). "Salinity intrusion in the St. Johns River, Florida." Estu-
arine and Coastal Modeling, Proceedings of the Seventh International Conference. M.L.
Spaulding ed., American Society of Civil Engineers, Reston VA, 120-139.
Tillman, D., Cerco, C., Noel, M., Martin, J., and Hamrick, J. (2003). "Three-dimensional
eutrophication model of the lower St. Johns River, Florida," ERDC TR-03EL353, US Army
Engineer Research and Development Center, Vicksburg MS.
-349-
------- |