GULF OF MEXICO DISSOLVED OXYGEN
MODEL (GOMDOM) RESEARCH AND
QUALITY ASSURANCE PROJECT PLAN
Gulf of Mexico
Hypoxic Zone
Prepared by James J. Pauer1, Timothy J. Feist2, Amy M. Anstead3, Wilson Melendez4, Russell
G. Kreis, Jr.1, and Kenneth R. Rygwelski1
1USEPA; 2Trinity Engineering Associates, lnc.;3ICF International; 4CSC Corporation;
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects Research Laboratory
Mid-Continent Ecology Division, Duluth, Minnesota
Large Lakes and Rivers Forecasting Research Branch, Grosse lie, Michigan
September 26, 2012
-------
Gulf of Mexico Hypoxia Modeling RQAPP
September 26,2012 \\ Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
FOREWORD
Over-enrichment of waterways by nutrients is a national and global issue and has subsequent
effects on freshwater, brackish, and marine systems. One of the symptoms of nutrient
enrichment is hypoxia, such as that observed in the Gulf of Mexico and is one of the largest
hypoxia zones observed on a worldwide basis. It is incumbent on water quality managers to
protect and to identify appropriate management strategies to mitigate the impacts of nutrient
stressors. In the following research and quality assurance project plan, we provide a modeling
and forecasting approach which will aid managers in the decision-making process for abating
hypoxia impacts to the Gulf of Mexico.
This document has been developed following the U.S. Environmental Protection Agency
(USEPA) Guidance for Quality Assurance Project Plans, EPA QA/G-5 (USEPA, 2002a) and
USEPA Guidance for Quality Assurance Project Plans for Modeling, EPA QA/G-5M (USEPA,
2002b). However, the document is a joint research plan and quality assurance project plan that
also incorporates elements of the USEPA, Office of Research and Development, National
Health and Environmental Effects Research Laboratory, Mid-Continent Ecology Division (MED),
Guidelines for the Preparation of MED Research Plans (USEPA, 2000). Additionally, other
modeling quality assurance guidance documents have been consulted (USEPA, 1991; ASTM,
1992; Richardson et a/., 2004; National Research Council, 2007; USEPA, 2008a, 2009).
Beyond being a prototypical, combined research plan and quality assurance project plan, the
emphasis is on mathematical modeling.
September 26,2012 iii Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
NOTICE
The information in this document has been obtained primarily through funding by the United
States Environmental Protection Agency (USEPA) under the auspices of the Office of Research
and Development (ORD). The report has been subjected to the Agency's peer and
administrative review, and it has been approved for publication as a USEPA document.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
September 26,2012 iv Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
ABSTRACT
An integrated high resolution mathematical modeling framework is being developed that will link
hydrodynamic, atmospheric, and water quality models for the northern Gulf of Mexico. This
Research and Quality Assurance Project Plan primarily focuses on the deterministic Gulf of
Mexico Dissolved Oxygen Model (GoMDOM). The GoMDOM models are similar in that they all
are derived from the LM3 Eutrophication model developed for Lake Michigan, but they differ in
spatial resolution and/or application. The other models are described only for the purposes of
understanding their inputs and linkages to the GoMDOM models. The GoMDOM models are
based on mass-balance principles and integrates multimedia nutrient inputs (primarily from the
atmosphere and the Mississippi and Atchafalaya Rivers) and ecosystem dynamics to establish
a forecasting capability for exploring management options to reduce the hypoxia zone. The
GoMDOM models consist of a coupled (eutrophication/dissolved oxygen (DO) and sediment)
water quality model that is linked to an atmospheric model (Community Multi-scale Air Quality
(CMAQ)) model and are driven by a linked hydrodynamics model (EPACOM). The GoMDOM
model framework will be calibrated and confirmed using cruise data (2003 - 2007) specifically
collected for the modeling effort along with other evaluated project and non-project data.
Uncertainty, sensitivity, and other statistical analyses will be performed to estimate the accuracy
of the water quality model predictions. Finally, the 6km x 6 km gridded GoMDOM model will be
applied to estimate the impact of several nutrient reduction scenarios on Gulf hypoxia, including
the allowable nutrient loads that would reduce the five-year running average areal extent of the
hypoxic zone to less than 5,000 km2 by 2015. This effort will assist managers in formulating a
strategy to achieve the goals specified in the Gulf of Mexico Action Plan.
September 26,2012 V Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
ACKNOWLEDGMENTS
The USEPA, Mid-Continent Ecology Division, Large Lakes and Rivers Forecasting Research
Branch wishes to acknowledge its partners and collaborators.
We are grateful to Bryon O. Griffith and Melanie Magee of the USEPA Gulf of Mexico Program
Office for encouragement and support during the planning phases of this project. We thank the
USEPA Office of Water/ Office of Wetlands, Oceans, and Watersheds for shipboard sampling
allocations aboard the OSV Peter W. Anderson and the OSV Bold; as well as the crews of both
vessels. Instrumental in project sampling, project planning, data analysis and interpretation
have been the staff of the USEPA, ORD Gulf Ecology Division including Richard M. Greene,
James Hagy, Janis Kurtz, John Lehrter, Michael Murrell, and Diane Yates. We also wish to
thank staff from the USEPA Office of Water, USEPA Regions, and State personnel, as well as
Leroy Anderson and Samuel Miller of the USEPA, ORD, Mid-Continent Ecology Division for
their support during shipboard activities.
We would like to thank Robin Dennis and Ellen Cooter of the USEPA, ORD, National Exposure
Research Laboratory for their contributions in atmospheric modeling. We wish to acknowledge
Robert Arnone and Dong-Shan Ko of the US Navy, Naval Research Laboratory for their efforts
in hydrodynamic modeling. We wish to posthumously recognize the contributions of Peter M.
Eldridge of the USEPA, ORD, Western Ecology Division in advancing our understanding of Gulf
of Mexico hypoxia and process modeling research. Also, we wish to thank the EPA staff and
their support contractor (Lockheed Martin) at the USEPA Environmental Modeling and
Visualization Laboratory for their contributions to model development and visualization.
We also wish to show our appreciation for other staff at the USEPA, ORD, MED, Large Lakes
Research Station who have supported the project in numerous ways including: David Miller
(USEPA), Ronald Rossmann (USEPA), Mark Rowe (USEPA), Xiaomi Zhang (TEA), David
Griesmer (CSC), Kay Morrison (CSC), and Debra Caudill (ASRC).
September 26,2012 vi Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
EXECUTIVE SUMMARY
Hypoxia and anoxia (low oxygen and oxygen depletion, respectively) are observed worldwide in
freshwater, brackish, and saltwater systems. These so called "dead zones" have been
observed and studied for well over one-half of a century. They are primarily attributed to human
activity and land use which have increased nutrient inputs and advanced the onset of
eutrophication. Nutrient stimulation of algal and plant growth produces large quantities of
organic matter, and it is the subsequent bacterial decomposition of the organic carbon that
imposes oxygen demand and depletion on the water column and underlying sediment interface.
When hypoxia becomes extensive, vital socio-economic factors such as recreation, food,
energy, transportation and industry can become impaired.
Hypoxia in the northern Gulf of Mexico is the largest such zone in the U.S. and second largest in
the world. Concern regarding the hypoxic zone size, duration and frequency centers around
habitat alteration and impacts to various Gulf fisheries. Since the mid-1980s, hypoxia has been
documented and tracked in the Gulf, where it has been seasonally observed to be as large as
22,000 km2. Approximately 40% of the contiguous U.S., encompassing 31 States, is drained
through the Mississippi River basin and enters the Gulf of Mexico through the Mississippi-
Atchafalaya complex. Point and non-point sources contribute to high nutrient loads originating
from population centers, farms, and industry. To encourage nutrient loads reduction, the
Mississippi River Watershed/Gulf of Mexico Nutrient Task Force, through the Gulf Hypoxia
Action Plan, has promoted an adaptive management approach together with a dual approach
for nitrogen and phosphorus reductions. However, with the many steps taken by Federal, State,
and local agencies, as well as landowners, these activities have not resulted in a significant
reduction in the hypoxic zone.
The Gulf hypoxia modeling framework is designed to integrate monitoring, condition
assessment, diagnosis, and experimentation within a mathematical modeling construct that
incorporates multimedia inputs, environmental data, and ecosystem dynamics to establish a
forecasting capability. The goal of this collaborative effort is to develop a state-of-the-science,
mathematical modeling framework that will aid water resource managers in making scientifically
defensible nutrient restoration decisions. Specifically, the model will be applied to estimate
several nutrient load reduction scenarios, including the nutrient loads that decrease the 5-year
year running average size of the zone to less than 5,000 km2 by the year 2015, a target
specified in the Gulf of Mexico Action Plan. With nutrient caps established through an
integrated, multimedia modeling approach, it is anticipated that the size of the hypoxic zone can
be reduced and associated improvements will be realized in habitat and toward biological
resources that are balanced and productive.
This Research and Quality Assurance Project Plan focuses on the following components of the
modeling project:
Project management, objectives, and description; quality objectives; special training
needs; and documents and records management;
Data generation and acquisition;
Model construct, coding, inputs, confirmation and corroboration, sensitivity/uncertainty
analysis, and application;
September 26,2012 vii Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Assessment and oversight;
Data validation and usability.
September 26,2012 viii Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
LIST OF ABBREVIATIONS
ASRC
ASTM
CDF
CE-QUAL-ICM
CMAQ
CSC
CREM
CTM
1-D
3-D
DO
DQO
ECOM
EPA
EPACOM
EMVL
FIPS
FOIA
FRA
CIS
FRC
GoMDOM
GED
IAS
IASNFS
LLRFRB
LLRS
LM3
LM3-Eutro
LUMCON
MED
MM5
MODIS Aqua
NARA
NASA
NCAR
NCOM
NERL
NetCDF
NH3
NHEERL
NRC
NRL
NOAA
ORD
PCB
PDF
PDOM
POM
PRISM
QA
Artie Slope Regional Corporation
American Society for Testing and Materials
Common Data Format
US Army Corps of Engineers Three-Dimensional Water Quality model
Community Multi-scale Air Quality
Computer Sciences Corporation
Council for Regulatory Environmental Modeling
Chemical Transport Model (a global atmospheric chemical transport model)
One Dimensional
Three Dimensional
Dissolved Oxygen
Data Quality Objective
Estuarine, Coastal, and Ocean Model
Environmental Protection Agency
Coastal Ocean Model for the Northern Gulf of Mexico developed for EPA
USEPA Environmental Modeling and Visualization Laboratory
Federal Information Processing Standards
Freedom of Information Act
Federal Records Act
Geographical Information System
Federal Records Center
Gulf of Mexico Dissolved Oxygen Model
Gulf Ecology Division
Intra-Americas Sea Model
Intra-Americas Sea Ocean Nowcast/Forecast System
Large Lakes and Rivers Forecasting Research Branch
Large Lakes Research Station
Lake Michigan Level 3 water quality model
Lake Michigan Level 3 water quality model - Eutrophication
Louisiana Universities Marine Consortium
Mid-Continent Ecology Division
A regional model for creating weather forecasts and climate projections
Moderate Resolution Imaging Spectroradiometer on the Aqua satellite
National Archives and Records Administration
National Aeronautical and Space Administration
National Center for Atmospheric Research
Navy Coastal Ocean Model
National Environmental Research Laboratory
Network Common Data Form
Ammonia
National Health and Environmental Effects Research Laboratory
National Research Council
Naval Research Laboratory
National Oceanic and Atmospheric Administration
Office of Research and Development
Polychlorinated biphenyls
Portable Document Format file
Princeton Dynalysis Ocean Model
Princeton Ocean Model
Parameter Elevation Regression on Independent Slopes Model
Quality Assurance
September 26, 2012
IX
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
QAPP Quality Assurance Project Plan
RQAPP Research and Quality Assurance Project Plan
ROMS Regional Ocean Model
QC Quality Control
RCS Revision Control System
SEAWIFS Sea-viewing Wide Field -of-view Sensor
SOD Sediment Oxygen Demand
SSWR Safe and Sustainable Water Resources
TEA Trinity Engineering Associates
URL Uniform Resource Locator
USEPA U.S. Environmental Protection Agency
USGS U.S Geological Survey
September 26,2012 X Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
GROUP A: PROJECT MANAGEMENT
Signature
Date
Signature
Date
A.1 Title and Approval Sheet
Research and Quality Assurance Project Plan Approvals - Mid-Continent Ecology
Division
Principal Investigator:
Russell G. Kreis, Jr.
Large Lakes and Rivers Forecasting
Research Branch
Branch Chief:
Russell G. Kreis, Jr.
Large Lakes and Rivers Forecasting
Research Branch
Administrative:
Dave Bolgrien, Chair
Quality of Science Committee
Barbara Sheedy
Quality Assurance Manager
Janet R. Keough
Associate Director for Science
Carl Richards, Director
Mid-Continent Ecology Division
Health and Safety:
Eric S. Mead
Safety, Health and Environmental
Management
Animal Care and Use:
Michael D. Kan!, Chair
Animal Care and Use Committee
Signature
Date
Signature
Date
September 26, 2012
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
A.2 Contents
TABLE OF CONTENTS
FOREWORD iii
NOTICE iv
ABSTRACT v
ACKNOWLEDGMENTS vi
EXECUTIVE SUMMARY vii
LIST OF ABBREVIATIONS ix
GROUP A: PROJECT MANAGEMENT 1
A. 1 Title and Approval Sheet 1
A.2 Contents 2
A.3 Distribution List 4
A.4 Project/Task Organization 5
A.5 Problem Definition/Background 6
A.6 Project/Task Description 9
A.7 Quality Objectives and Criteria for Measurement Data 13
A.8 Special Training Needs/Certification/Expertise 15
A.9 Documents and Records 15
GROUPS: DATA GENERATION AND ACQUISITION 18
B. 1 Model Formulation 18
B.2 Model Coding 23
B.3 Model Inputs 24
B.4 Model Confirmation 26
B.5 Model Calibration and Corroboration 27
B.6 Model Sensitivity/Uncertainty Analysis 28
B.7 Model Application 28
B.8 Data Management 29
GROUP C: ASSESSMENT AND OVERSIGHT 31
C. 1 Assessments and Response Actions 31
C.2 Reports to Management 31
GROUP D: VALIDATION AND USABILITY 32
D.1 Model Review 32
D.2 Verification and Validation Methods 32
D.3 Reconciliation with User Requirements 33
REFERENCES 34
FIGURES 41
TABLES 49
APPENDIX 1: Conceptual Equations for Dissolved Oxygen and Sediment Diagenesis 52
September 26,2012 2 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
LIST OF FIGURES
Figure 1. MED Gulf of Mexico Hypoxia Modeling Organization 41
Figure 2. Overview of Project Clients and Collaborators (with lines of communication only) 42
Figure 3. Gulf of Mexico Study Area 43
Figure 4. Areal Extent of 2007 Hypoxic Zone 44
Figure 5. Changes in Areal Extent of 1985-2008 Hypoxic Zone 45
Figure 6. Annual Nitrate Load to the Gulf of Mexico 46
Figure 7. Total Annual Phosphorus Load to the Gulf of Mexico 47
Figure 8. Integrated, Multimedia Gulf of Mexico Modeling Framework 48
Figure 9. Sediment-Water Interactions 48
LIST OF TABLES
Table 1. Overall Project Schedule 49
Table 2. List of Desired Field Measurements 50
September 26,2012 3 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
A.3 Distribution List
The USEPA Gulf of Mexico Program Office in partnership with the USEPA Office of Research
and Development and USEPA Office of Water are building upon past efforts and have initiated
design plans for a framework that guides the science needed to address the hypoxia problem in
the Gulf of Mexico to meet the objectives of the Gulf Hypoxia Action Plan. The distribution list
consists of those listed below and others to be determined.
Administrative/Management
Russell G. Kreis, Jr., Chief, Large Lakes and Rivers Forecasting Research Branch
Dave Bolgrien, Chair, Quality of Science Committee
Barbara Sheedy, Quality Assurance Manager
Janet R. Keough, Associate Director for Science
Carl Richards, Director, Mid-Continent Ecology Division
Eric S. Mead, Safety, Health and Environmental Management
Michael D. Kahl, Chair, Animal Care and Use Committee
Suzanne van Drunick, SSWR National Program Director
Michael McDonald, SSWR Deputy National Program Director
Walt Nelson, SSWR Project Lead
Richard M. Greene, Chief, Ecosystem Dynamics and Effects Branch, Supervisory Research
Biologist, Gulf Ecology Division
William Benson, Director, Gulf Ecology Division
Ben Scaggs, Director, Gulf of Mexico Program Office
Darrel Brown, Office of Water, OWOW
Michael J. Shapiro, Assistant Administrator, Office of Water
Harold Zenick, Laboratory Director, National Health and Environmental Effects Research
Laboratory
Jennifer Orme-Zavelata, Director, National Exposure Research Laboratory
S. T. Rao, Division Director, Atmospheric Modeling Division
Robin Dennis, Senior Scientist, Atmospheric Modeling Division
USEPA - GED Personnel
Janis Kurtz, Nutrients Team Leader
Michael Murrell, DO Task Coordinator
James Hagy
Diane Yates
John Lehrter, SSWR Task Lead
Jeanne Scott, GED Quality Assurance Officer
Matthew Harwell, Chief, Ecosystem Assessment Branch
Mace Barren, Chief, Biological Effects and Population Response Branch
J. Kevin Summers, Associate Director for Science
USEPA - MED - LLRFRB Personnel
David Miller, USEPA, ORD
Mark Rowe, USEPA ORD
Kenneth Rygwelski, USEPA, ORD
James Pauer, USEPA, ORD
Amy Anstead, ICF International
Phillip DePetro, ICF International
Timothy Feist, Trinity Engineering Associates, Inc.
September 26,2012 4 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Xiaomi Zhang, Trinity Engineering Associates, Inc.
Wilson Melendez, CSC
David Griesmer, CSC
US Navy - Naval Research Laboratory (Stennis Space Center)
Robert Arnone
A.4 Project/Task Organization
This project is being conducted within the ORD Safe and Sustainable Water Research Program
(SSWR) and is described in the SSWR Strategic Research Action Plan (USEPA, 2012). The
research is being conducted within Theme 1: Sustainable Water Resources; Topic 2:
minimizing the environmental impacts of land use practices that lead to the sustainability of
surface and subsurface water resources; Task 2.3D: modeling the linkage between discharge
and nutrients from the Mississippi River basin to the Gulf of Mexico hypoxia. The Project
Leader is Walt Nelson and the Task Leader is John Lehrter.
U.S. EPA Mid-Continent Ecology Division (MED) Role
This MED Research and Quality Assurance Plan (RQAPP) describes only those Gulf of Mexico
modeling activities that are conducted within the Large Lakes and Forecasting Research Branch
of MED. Dr. Russell G. Kreis, Jr. (MED Branch Chief) is the principal investigator for this
project. Dr. Kreis is responsible for developing and maintaining the official copy of this RQAPP.
MED is responsible for the development, calibration, confirmation, corroboration, sensitivity
analysis, and forecasting of a suite of Gulf of Mexico Dissolved Oxygen Models (GoMDOM)
including a screening-level one-dimensional (1-D) GoMDOM and three-dimensional (3-D)
GoMDOM models on scales of 6 km x 6 km and 2 km x 2 km grid sizes. See Figure 1 for
details on MED members of the modeling and support teams.
Project Collaborators
This large project requires products and expertise from parties external to MED (see Figure 2).
Partnerships have been established with the following:
The U.S. EPA Gulf Ecology Division (GED) is a primary collaborator and partner in the project.
John Lehrter of GED serves as the project Task Lead for the Safe and Sustainable Water
Resources (SSWR) program, Task 2.3D, Modeling the linkage between discharge and nutrients
from the Mississippi River basin to Gulf of Mexico hypoxia. The work described in this MED
Gulf of Mexico Modeling RQAPP is one of the projects under SSWR Task 2.3D. GED has been
providing analytical chemistry data from the Gulf and serves as a critical expert advising MED
on ocean chemical, biological, and physical processes related to modeling Gulf eutrophication
and hypoxia. Results of the modeling work will result in peer reviewed journal articles
coauthored among MED and GED scientists and engineers.
U.S. Navy Naval Research Laboratory (located at Stennis Space Center), through agreements
with the U.S. EPA Gulf Ecology Division (GED), has been providing hydrodynamic model
transport fields from their Environmental Protection Agency Coastal Ocean Model (EPACOM)
developed for northern Gulf of Mexico water.
September 26,2012 5 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
The U.S. EPA Office of Environmental Information, Environmental Modeling and Visualization
Laboratory (EMVL) has been providing support to the project in the areas of specialized sub-
modeling, visualization, and improvement of modeling run times.
U.S. EPA National Exposure Research Laboratory (NERL) Atmospheric Modeling Division has
been providing atmospheric fluxes of nitrogen compounds to the Gulf waters from their
Atmospheric/Air deposition model (CMAQ, CTM, MM5).
An informal collaboration has been established with the National Aeronautics and Space
Administration (NASA) at the Goddard Space Flight Center, Greenbelt, MD. NASA has refined
algorithms for chlorophyll, total suspended solids, and particulate organic carbon based on Gulf
data from this project that are used in combination with MODIS AQUA and SEAWIFS remote
imagery for comparisons to GoMDOM model output.
Project Clients
Gulf of Mexico hypoxia has been a concern and a priority focus for the USEPA for several
years. The study efforts described here are anticipated to support decision-making by the
USEPA and many other management groups identified on Figure 2 as our clients. This figure
shows relationships and lines of communication among these clients and project collaborators.
The Office of Water is the lead among the USEPA clients with authorities regarding the Gulf of
Mexico and the Assistant Administrator for the Office of Water is Chair of the Interagency
Mississippi/Gulf of Mexico Nutrient Task Force. The Office of Water is directly supported by the
USEPA Office of Wetlands, Oceans, and Watersheds and USEPA Gulf of Mexico Program
Office, Region 4 and Region 6 as they have jurisdictional interests in the Gulf.
A.5 Problem Definition/Background
Investigations of the Gulf of Mexico's inner shelf (Figure 3) in the coastal waters of Louisiana
and Texas have documented seasonal oxygen depletion in this zone during the past several
decades (Rabalais et a/., 1999, 2001, 2002). Hypoxia, defined as dissolved oxygen
concentrations of less than 2 mg/L, has increased in intensity, size, and duration during the past
several decades, averaging an area of impact of approximately 15,000 km2. The areal extent of
the hypoxic zone (Figure 4 and Figure 5) in the past decade has been observed to be as great
as 22,000 km2 and appears to be the largest known hypoxic zone in the waters of the
conterminous U.S (Pew Oceans Commission, 2003; U.S. Commission on Ocean Policy, 2004;
World Resources Institute, 2008). The Mississippi-Atchafalaya River Basin appears to be the
dominant source of macro- and micronutrients which affect the observed hypoxia (Goolsby et
a/., 1999; Mitsch et a/., 1999) through the over-production of phytoplankton and the subsequent
decomposition of the organic carbon that imposes oxygen demand and depletion on the water
column and underlying sediment interface.
The primary environmental problem is the size, duration, frequency, and intensity of hypoxia in
the Gulf of Mexico. Hypoxic bottom waters of the Gulf of Mexico are a detriment to the overall
ecological health of this system and have had chronic and acute effects on marine life. The
hypoxic zone inhibits the occurrence of marine life, degrades the habitat for many aquatic
organisms, and negatively impacts desired aquatic production. The impact on immobile species
such as benthos and shellfish is initially a restriction of range and loss of habitat followed by
mortality. Mobile species, such as fishes and shrimp, may be able to avoid the hypoxic zone,
but their movement and habitat become restricted.
September 26,2012 6 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Marine systems are typically nitrogen-limited in contrast to freshwater systems which exhibit
phosphorus limitation. In each case, the other nutrient, together with silica, may become
secondarily limiting when the primary nutrient is over-enriched. Many coastal areas of the U.S.
have been enriched with nitrogen and are showing signs of secondary phosphorus and silica
limitation, as observed in the case of the Gulf of Mexico. The available historical records
indicate that nitrogen loading has increased more dramatically than that of phosphorus (Figure 6
and Figure 7, respectively). The USEPA has adopted a dual management approach for both
nitrogen and phosphorus as it relates to the freshwater resources of the Mississippi Basin and
the Gulf of Mexico (USEPA, 2008b). The relationships among nitrogen, phosphorus, carbon,
and solids loads and concentrations with algal production, algal biomass, and oxygen demand
are critical to the understanding of hypoxia in the Gulf of Mexico.
Previously applied water quality models and approaches (Bierman et a/., 1994; Greene et a/.,
2009, Justic et a/., 2003; Scavia et a/., 2003, 2004; Hetland and DiMarco, 2007; Morse and
Eldridge, 2007; Scavia and Donnelly, 2007) have yielded insights to Gulf hypoxia but questions
are being raised as to the suitability of their resolution and degree of uncertainty with respect to
confidence related to nutrient reduction forecasts and the final target to be established. These
models used relatively coarse segmentation schemes with limited spatial resolution; simplistic or
limited kinetics; very approximate hydrodynamics, including the flow direction of the Mississippi
River plume; and simplistic sediment and dissolved oxygen interactions. Due to these
limitations, consensus on loading reduction targets have been very difficult to reach when
confronted with a range of 30-65% for nitrogen and/or phosphorus based upon modeling and
empirical approaches (Mississippi River/Gulf of Mexico Watershed Nutrient Task Force, 2001,
2004, 2008a, 20008b; USEPA, 2008b; NRC, 2009).
The body of investigative and mathematical modeling studies during the past decade has
provided considerable insight into the Gulf hypoxia issue and its relationship to the Mississippi
River Basin; however, a number of recommendations for future work and improvements have
been outlined (Committee on Environment and Natural Resources, 2010; Mississippi River/Gulf
of Mexico Watershed Nutrient Task Force 2001, 2004, 2008a, 20008b; Justic et al., 2007;
USEPA, 2008b; NRC, 2009). Selected recommendations, gaps, and issues to promote a
consensus modeling framework with supporting data are presented below:
1) A sampling design is needed to support the development of an integrated, multi-media mass
balance modeling framework.
2) The sampling program should be specifically-designed to reduce the uncertainty associated
with the empirical data and modeled nutrient-reduction forecasts.
3) The sampling program should be seasonally-driven to create at least a full 2-year period
dataset and supplement the existing summer sampling program by directed overlap.
4) The sampling program should be statistically-based with random transects and stations that
include multiple resource classes: embayment/near-coastal, inner shelf, outer shelf, and
bluewater. The offshore boundary should be sufficiently sampled to delineate the boundary
condition.
5) Determine phytoplankton species and carbon flux seasonality.
6) Further define sediment diagenesis and sediment nutrient flux factors.
September 26,2012 7 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
7) Account for water column oxygen demand.
8) Further quantify the relationships among loads, ambient concentrations, chlorophyll, and
dissolved oxygen, using data and models.
9) Establish a multimedia (air, water, sediment), mathematical modeling framework which
builds upon past efforts and includes a hydrodynamic model, atmospheric model, sediment
transport and fate model, and water quality-eutrophication model.
In order to address these factors, a high resolution, multi-media modeling suite is being applied
to address the atmospheric, hydrodynamic, water quality, and sediment interactions as well as
spatial resolution and improved kinetics. The modeling framework is also being supported by a
monitoring and laboratory program, specifically designed for the modeling. The cornerstone of
the multimedia construct being applied is the Gulf of Mexico Dissolved Oxygen Model
(GoMDOM), a version of LM3-Eutro, which includes water quality chemical, physical, and
biological interactions and kinetics with linkage capabilities to other modeling components. The
modeling framework required primary productivity, dissolved oxygen (DO) and other kinetic
equations to realistically represent processes within the Gulf. A sediment-water component is
also necessary to account for this important process in the Gulf of Mexico. Nutrient transport is
driven by hydrodynamic output from the U.S. Navy Naval Research Laboratory's (NRL)
EPACOM model (Ko et a/., 2003). Atmospheric loads of nitrogen compounds are being
provided by EPA's Community Multi-scale Air Quality Model (CMAQ).
The Gulf Hypoxia modeling framework is being designed to integrate monitoring, condition
assessment, diagnosis, and experimentation within a mathematical modeling construct that
incorporates multimedia inputs, environmental data, and ecosystem dynamics to establish a
forecasting capability. Since a wealth of information is available to formulate the many transport
and kinetic processes and to estimate model parameters, it is believed that this deterministic
model will provide a better predictive estimate than using an empirically established relationship.
The goal of this collaborative effort is to develop a mathematical modeling framework that will
aid water resource managers in making scientifically defensible nutrient restoration decisions to
reduce the hypoxia problem. By reducing the size of the hypoxic zone, it is suspected that
habitat and food web assemblages along the Louisiana-Texas (LA/TX) coast will benefit.
Specifically, the model will be applied to estimate dissolved oxygen concentrations and hypoxia
area in the northern Gulf of Mexico under several nutrient load reduction scenarios. Other
major model outputs include the duration of hypoxia, nutrient concentrations, and phytoplankton
concentrations. This modeling effort will assist managers in helping them to understand options
available to achieve a goal of a five-year running average hypoxia zone of 5,000 km2 as
specified by the 2001 and 2008 Gulf of Mexico Action Plan and supporting documents
(Mississippi River/Gulf of Mexico Watershed Nutrient Task Force 2001, 2004, 2008a, 20008b;
USEPA, 2008b).
This Research and Quality Assurance Project Plan focuses on the following components of the
modeling project:
• Project management, objectives, and description; quality objectives; special training
needs; and documents and records management.
• Data generation and acquisition
September 26,2012 8 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
• Model construct, coding, inputs, confirmation and corroboration, sensitivity/uncertainty
analysis, and application
• Assessment and oversight
• Data validation and usability
A.6 Project/Task Description
The modeling framework will build upon earlier models that were developed and applied to the
Gulf of Mexico and is most similar to the efforts of Bierman et a/. (1994). The Gulf of Mexico
Eutrophication and Dissolved Oxygen Model (GoMDOM) is based on the LM3-Eutro water
quality model (Pauer et a/., 2006, 2008, 2011; Melendez et a/., 2009) that was developed for
and applied to Lake Michigan. GoMDOM, in its present form, has salinity, two phytoplankton
state variables, one zooplankton state variable, and several dissolved and particulate nutrient
state variables. The model uses standard eutrophication kinetics to describe the many
biochemical reactions such as: Monod kinetics to describe phytoplankton growth, first- order
nutrient mineralization kinetics, and a temperature dependency function for the biochemical
reactions. The Jassby and Platt equation was used to estimate the limitation of primary
production by available light (Jassby and Platt, 1976; Lehrter et a/., 2009). Light attenuation
was calculated using a site-specific relationship between light attenuation and chlorophyll,
particulate carbon, and salinity. It also has simple user-defined sediment-to-water nutrient
fluxes. To prepare LM3-Eutro for its application to the Gulf of Mexico, the model was modified
to use output from the Navy hydrodynamics model (EPACOM) and a dissolved oxygen
subroutine was included. The model receives loadings primarily from the Mississippi and
Atchafalaya Rivers but also from minor tributaries and an atmospheric model (Community Multi-
scale Air Quality (CMAQ)). See Figure 8 and Figure 9 for information on the integrated
multimedia model interactions. This modeling framework will integrate multimedia inputs (from
statistically-based monitoring programs) and ecosystem dynamics to establish a model that will
have the forecasting capability for exploring alternative futures and/or remedial options.
GoMDOM will be calibrated and corroborated using cruise data collected from 2003 to 2007.
Uncertainty, sensitivity, and other statistical analyses will be performed using the model to
estimate the accuracy of the model predictions. Finally, the model will be applied to estimate
the dissolved oxygen concentration and hypoxic area in the northern Gulf of Mexico under
several nutrient reduction scenarios, including the allowable nitrogen load that would limit the
hypoxic area to a maximum of 5,000 km2.
An overall project timeline is provided in Table 1. In 2007, advanced general project planning
took place. In 2008, database development, which harmonized the various field measurements,
laboratory analyses, and research results by media, site, and time, began along with model
development. Considerable model calibration runs, sensitivity runs, and journal article
preparation took place in FY2011 and FY2012. This timeline is based upon current
understanding of the science affecting hypoxia on the coastal shelf and management objectives.
The schedule should be considered preliminary and may require adjustment if management
priorities change because of future events, if scientific findings during the project indicate a need
to change the project scope, or if deliverables from project partners are not received in a timely
manner.
September 26,2012 9 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Model Development Strategy
The modeling framework is being developed following a two phase strategy (see Section B.1 for
more details). In the first phase, the model eutrophication, DO, and sediment-water
interactions are being developed and tested using a one dimensional (1-D) GoMDOM screening
model. This screening-level model will provide for easier development and testing of water
quality kinetic equations. It is being applied at selected stations in the Gulf of Mexico where
data for calibration purposes are plentiful and where horizontal gradients of model state
variables are minimal. The model is run for a relatively short duration and consists of four water
layers and includes interaction with the sediment. This screening model will not be used for
model scenario forecasting, although it should provide insight into biological and chemical
interactions on the coastal shelf. Calibrated model coefficients/parameters from the screening
model will be used to provide some initial estimates for similar coefficients/parameters in the
higher resolution GoMDOM models.
During the second phase, the model framework developed in the first phase will be applied to
an intermediate resolution model grid (approximately 6 km x 6 km and 26 sigma layers). The
intermediate resolution model will be calibrated and corroborated to existing coastal shelf cruise
data. The model will then be applied as a diagnostic tool to assist in evaluating biochemical
interactions on the coastal shelf and applied to selected management scenarios. If needed, a
high resolution (2 km x 2 km) GoMDOM model will be implemented.
1-D GoMDOM Screening Model
The development of the screening model will include three tasks: a review of previous modeling
efforts and available data, the application of the selected model to the Louisiana coastal shelf,
and the modification of the model to more appropriately represent physical, chemical, and
biological processes on the Louisiana coastal shelf. The proposed schedule for completing
these tasks is outlined in Table 1.
Data/Model Review
This task included reviewing previous modeling efforts, available data, and other
recommendations and identified gaps to determine which modeling frameworks are suitable for
use in the proposed modeling framework. An initial review of the GED cruise data and other
available data was conducted to assist in determining the most suitable extent for the proposed
model grid and to determine what time periods and kinetic processes have sufficient data to
support modeling efforts. Published studies were reviewed to help determine appropriate water
quality processes to include in the model framework. Previous water quality models, both from
the study area and those suitable to be applied to the study area, were reviewed for possible
use in the modeling framework.
Two mathematical models were considered as frameworks for developing the hypoxia model for
the Gulf. CE-QUAL-ICM was developed by the US Army Corps of Engineers (Cerco and Cole,
1995) and applied to Chesapeake Bay (Cerco and Cole, 1994). LM3-Eutro is another high
resolution framework that was developed and applied to Lake Michigan (Pauer et a/., 2006,
2008, 2011; Melendez et a/., 2009). Both models had many positive attributes and very suitable
building block for the next the Gulf model. After careful review and consideration, the LM3
framework was selected as a base for developing a new Gulf model. The results of the model
review found that LM3-Eutro has most of the features to address the hypoxia problem in the
northern Gulf of Mexico. Since this model was developed in-house and staff are familiar with
September 26, 2012 10 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
the modeling framework, it can be relatively easily modified and applied to the Gulf study area.
Modifications include code modification to utilize output from the Navy hydrodynamics model
(EPACOM) and the addition of dissolved oxygen and sediment subroutines.
Hydrodynamic models were also reviewed. The models examined included the Princeton
Dynalysis Ocean Model (PDOM), Environmental Fluid Dynamics Code (EFDC), Estuarine,
Coastal and Ocean Model (ECOM), Regional Ocean Model (ROMS), and the Navy's EPACOM
model. These were generally regional models with various spatial and depth operational limits.
The EPACOM model is based on the Navy's Intra-Americas Sea Ocean Nowcast/Forecast
System (IASNFS) and was selected to provide the hydrodynamic transport fields. IASNFS is
based on the Navy Coastal Ocean Model (NCOM) (Martin, 2000; Martin et a/., 2009).
The Community Multi-scale Air Quality Model (CMAQ) was selected to provide atmospheric
nitrogen compound loads to the water quality model. It is the premier national deposition model,
operated by EPA, has nitrate deposition over the Gulf of Mexico, and is being run with finer-
resolution deposition for the purposes of this study (Byun and Ching, 1999; Byun and Schere,
2006; Dennis etal., 2007, 2008.
The initial model development task involved applying LM3-Eutro equations to the 1-D model grid
(four water layers with interactions with the sediment) to evaluate and test the application.
Simplification included limiting the model such that all phytoplankton were represented as a
single model state variable along with fewer particulate nutrient and carbon state variables.
GoMDOM-1D used site-specific measurements and empirical relationships to determine nutrient
and oxygen demand sediment fluxes. MATLAB was used in this model development.
After the initial testing, the model will be modified as needed to appropriately simulate important
processes affecting hypoxia on the coastal shelf. The model then will undergo further testing
and evaluation to ensure that physical, chemical, and biological processes are being suitably
simulated. The 1-D GoMDOM will be used to allow for easier testing as the model framework is
being developed. The majority of model confirmation/corroboration activities (Section B.4)
related to kinetic processes have been completed. The screening model will not be formally
applied to management scenarios. The screening model will be compared to field data and may
provide insight into processes in the study area that may need to be further evaluated. The
calibrated model coefficients from the 1-D GoMDOM will be used as initial estimates of similar
coefficients for the 3-D GoMDOM models. Any deficiencies identified in the model framework
will be addressed, and the model re-confirmed, before application of the model to the
intermediate resolution GoMDOM.
3-D GoMDOM 6 km x 6 km Intermediate Resolution Model
Finalize Intermediate Resolution Model Grid
The model grid for the 3-D GoMDOM 6 km x 6 km intermediate resolution model was based
upon the review of available data. The grid extends from the shoreline southward to
approximately the 80-100 m contour and from east of the Mississippi River Delta westward to
93° W longitude. This grid extent contains the area of hypoxia during most years and provides
sampling stations outside the grid for use as boundary conditions. The model is being applied
to an approximately 6 km x 6 km model grid, with hydrodynamic transport provided by the NRL's
EPACOM model output aggregated to this size (the original scale is approximately 2 km x 2
km). EPACOM output provided for this project contains 26 vertical sigma layers which should
provide suitable resolution of surface, pycnocline, and hypoxia zone layers. This grid size
September 26, 2012 11 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
provides a compromise between higher spatial resolution and faster model run times. The
extent of the grid may be expanded to include additional areas of the coastal shelf if the initial
model results show it would be useful and if hydrodynamic data are available to support an
expansion. The vertical resolution of the grid is identical to that of EPACOM.
Due to the hydrodynamic and biochemical processes in the study area that occur at relatively
small scales because of the shallowness of the shelf and the mixing and stratification of the
freshwater outflow from the Mississippi and Atchafalaya Rivers, a model with a resolution similar
to the intermediate resolution GoMDOM model is recommended for properly simulating
processes affecting hypoxia in this area. In the second phase of model development, the
knowledge gained from the first phase will be applied to the intermediate resolution GoMDOM
model. The model will then be calibrated and corroborated against cruise and process data and
used for diagnostic evaluation of biochemical processes and for management scenarios. The
proposed schedule for completing these tasks is included in Table 1. This schedule should be
considered preliminary and may need to be adjusted if findings from the screening model
suggest a change in project scope, if management priorities change, or if products from project
partners are not received in a timely manner.
Create Input and Linkage Files
This task included developing the model grid and geometry files, developing software to convert
the NRL hydrodynamic model output into a format that the water quality model can use as input,
creating input decks (model input files describing oxygen and nutrient initial conditions and
estimates for the model parameters), modifying the original source code to read input data and
to write simulation results to output using NetCDF, and running and testing the model. Software
developed in this phase, for example to generate mapping and linkage files, was designed so
the model framework can easily be applied to higher resolution model grids.
Calibration and Corroboration
The intermediate resolution model will be tested to confirm that it is working properly (Section
B.4) and then calibrated and corroborated against GED cruise survey data following procedures
outlined in Section B.5. Data sets for these procedures will be selected from databases
completed by the time of the procedure.
An evaluation of model sensitivity and uncertainty (Section B.6) will be conducted concurrently
with the calibration and corroboration of the intermediate resolution model.
Model Diagnostic Testing and Scenarios
In this task the calibrated and corroborated model will be applied in a diagnostic mode for
scientific evaluation of shelf processes and in scenarios to assist in evaluating management
options following procedures described in Section B.7. Additional input files and hydrodynamic
inputs may need to be created depending upon the scenarios selected.
3-D GoMDOM 2 km x 2 km High Resolution Model
If the modeling results and analysis from the 3-D GoMDOM 6 km x 6 km model indicate that a
higher resolution model is warranted, and if time permits, a 3-D 2 km x 2km high resolution
model will be utilized. The modeling framework from the intermediate resolution GoMDOM
model is directly transferable to the 2 km x 2 km model. Also, once the intermediate resolution
September 26,2012 12 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
model is calibrated, those calibrations (when applicable) can be transferred to the high
resolution model. This will be a time saver when calibrating the 2 km x 2 km model. However,
due to the large number of model cells in the high resolution model, the run times will be much
longer than that in the intermediate model. Consequently, an effort is being planned in FY2013
(see Table 1) to parallelize the code to significantly reduce the model run times. This model will
likely be run and tested on EPA's supercomputer in RTP, NC in FY2013.
A.7 Quality Objectives and Criteria for Measurement Data
Quality objectives and criteria will be established to ensure that the model output addresses the
management questions with the accuracy required by the user. This can be achieved by
establishing statistical criteria to determine, during the model evaluation stage, if the overall
accuracy of the model is acceptable and if the model uncertainty is acceptable.
Level of accuracy and precision of model output
Before a model is used for remedial guidance and/or regulatory purposes, agreement between
the expectations of the managers who will be using the model and the model developers is
needed. Managers need to be generally well versed in the science of modeling natural
systems. Modelers have the responsibility of not only attempting to make the models reliable
but also to state unequivocally their assumptions and uncertainties. This is usually done by
providing the most probable answer(s) along with uncertainty brackets which provide a range
that is very likely to contain the actual answer. The decision-maker must determine whether to
use the model with the uncertainties and caveats provided or to provide additional resources to
refine the results. If refinement is needed, the modeler can advise management on what needs
improvement because of their knowledge gained in determining model sensitivity to various
model-controlling forcing functions or processes.
Modeling quality objectives continue to be discussed regarding the Gulf of Mexico and will
depend upon the certainty required by managers and the importance of the modeling tool in
developing nutrient loading targets to reduce the extent of hypoxia. With respect to these
concerns, a preliminary data quality objective (DQO) is for the model to simulate the average
water quality within plus or minus two standard errors of the mean of the field measurements,
meaning there is approximately 95% confidence that the actual model-predicted result falls
within this range. It is likely that the model fit to data will be much better than this criteria for
many of the model-predicted state variables. The data means and standard errors will be
computed using appropriate spatial and temporal statistical averaging and interpolation
techniques.
Obviously, the range of plus or minus two standard errors of the mean of the measurements is
(in part) a function of measurement (including both sampling and instrument) precision. Most of
the field data used in model calibration and confirmation will originate from the U.S. EPA Gulf
Ecology Division. The quality objectives and criteria for these data are described in the Gulf of
Mexico Hypoxia Quality Assurance Project Plan (Greene, 2007). In this document, most
parameters have an instrument accuracy target of 10% and an instrument precision target of
30% Relative Standard Deviation, also known as the Coefficient of Variation.
Prediction bias will be minimized by calibration, the process of parameter optimization seeking
to minimize residuals (the difference between model calculated and measured concentrations),
without violating constraints imposed by scientific observations and principles. Modelers
commonly plot field observations vs. model output for a given model state variable (Pineiro et
September 26, 2012 13 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
a/., 2008). This method provides both qualitative and quantitative feedback to the modeler on
how well the model compares to field observations. If the model predictions match field
observations, then it is expected that the residuals (difference between the plotted points and
the 1:1 line) fall randomly about the 1:1 line and are relatively close to that line. Model biases
can be noted using this method when a majority of points lie either above or below the 1:1 line.
If a majority of the points fall either above or below the 1:1 line, then a serious model bias exists
and will be explored further to determine the cause.
An R-squared of the correlation described in the previous paragraph provides information on the
"goodness of fit" of the model to observations. In a calibration exercise, the modeler will try to
maximize the R-squared. However, no target R-squared can be established because this can
vary from state variable to state variable. For example, modeling a conservative substance like
salinity may yield a maximum achievable R-squared of 0.8; however, for a much more complex
state variable involved in a multitude of kinetic reactions such as nitrogen, an R-squared of 0.5
may be the best that can be achieved.
The model will be considered calibrated when the results for important model state variables fall
within the 95% confidence intervals of the majority of the data cruise means and the results
have a highest achievable R-squared when correlating model output to field observations,
stratified appropriately in time and space. In addition, model simulations will attempt to
reproduce the statistical distribution properties of the data. This will be evaluated by comparing
cumulative frequency distribution plots of data to frequency distribution plots from comparable
model predictions.
Once calibrated to field data, the model will be valid within the error constraints specified for the
calibration period. However, for forecast predictions, it is not possible to know the uncertainty of
predicted forcing functions and boundary conditions. Therefore, the model will be run for
various forecast scenarios with inputs bracketed in terms of extreme expectations and
probability distributions, and the results will be provided in terms of prediction means and
exceedance limits.
Criteria for using secondary data (literature values, etc.)
Data generated specifically for this project will be used for model development and calibration;
however, where no project data are available, data from the literature and other modeling
studies will be used. The majority of data to be used as model inputs originate from the Gulf of
Mexico Hypoxia Study Project and samples are being collected and analyzed following the U.S.
EPA Gulf Ecology Division's "Gulf of Mexico Hypoxia Monitoring Survey Quality Assurance
Project Plan" (Greene, 2003; Greene, 2007). The monitoring QA plan describes the QA
program and process, organizational structure, data quality objectives, implementation of the
QA program, and information management guidelines for the data collection activities of the
study. All GED's analytical data for the model's target analytes and most supporting data will
have been verified through their QA program's process and will have met the performance
criteria established before release to modelers. Data will undergo an additional screening by
project modelers to ensure suitability for modeling purposes
Data generated through other projects or studies may be obtained from either published or
unpublished sources. The published data (including those from gray literature) will have had
some degree of QA review, although there is a wide range of review quality among possible
sources. Unpublished databases may be obtained directly from authors or from on-line
databases.
September 26,2012 14 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
When possible, all data used by the modelers will be checked for bias, comparability, outliers,
normality, completeness, precision, accuracy, validity of station names and sample identification
codes, and units errors. Modelers will also review any documentation or data qualifiers
accompanying data sets. As questions arise, we will contact the data generator if possible.
Negative consequences of making inappropriate decisions due to poor model prediction
ability
The chances of making inappropriate decisions due to poor model prediction ability will be
minimized through the quantification and evaluation of the accuracy and reliability of model
predictions (Sections A.7, B.5, and B.6) and through the reconciliation of model prediction
reliability with user requirements (Section D.3). In addition, the Gulf of Mexico hypoxia model
proposed for this study is only one of several tools that will be used by managers when
determining management and regulatory options for the Mississippi River/Atchafalaya
River/Gulf of Mexico system. Other tools include previous modeling studies, on-going
monitoring efforts, and summary reports by scientific panels such as the Science Advisory
Board Hypoxia Panel. As one of many tools available, the proposed model will provide
additional weight of evidence to proposed nutrient management options and provide additional
insight into ecological processes affecting hypoxia, but it will not be the sole determining factor
in management decision-making. Model results will include estimates of reliability provided by
modelers and reviewers that will guide managers in how much weight to place on model results.
A.8 Special Training Needs/Certification/Expertise
Two primary categories of specialized training and certification are envisioned. Typically an
environmental engineering degree or environmental science degree with training in systems
science is suggested for mathematical modelers. With the Agency's emphasis on integrated,
multimedia, modeling, it is valuable to have a broad background that includes the aquatic
sciences (chemistry, biology, and physical processes). A degree in computer sciences is
recommended for model programmers and database specialists. In both cases, degrees may
be in other primary disciplines that enable each to conduct the respective job skill. In addition,
strong backgrounds in mathematical sciences and statistical analyses are typically necessary.
Additional specialized (such as geographical information systems) training will be provided if
needed on an individual basis and will be documented by the project leader. The project leader
will be responsible for assuring that the modeling staff have the training necessary to complete
the project.
All modeling staff (both federal and contracting staff) will be required to have had training in
NHEERL/MED's Quality Assurance program. This QA training course covers the following
topics for new hires: QA Orientation, Laboratory Recordkeeping, QA Planning Documents,
Operating Procedures and Technical Systems Audits. Every three years, all scientific staff will
be required to attend a QA Refresher Course.
A.9 Documents and Records
A PDF copy of the Gulf of Mexico Research and Quality Assurance Project Plan (this document)
with all signature approvals will be made available via the Internet. The notification of
accessibility of the approved plan will be sent to those individuals and organizations listed in
Section A.3 of this plan. Any modifications resulting from an annual review of the plan will also
be posted on the designated web site as addendums to the plan.
September 26, 2012 15 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
A Study File (project records at the completion of a project) will be prepared by the Principal
Investigator at the termination of the project. The Study File will contain all necessary
information to substantiate any project findings and will include both paper and electronic
records. Any relevant electronic model records not physically contained in the Study File (such
as very large model files) will be stored within a model archive located on a local server. An
index to the materials in the Study File will be included. The contents of the Study File may
change as needed and directed by the Principal Investigator. The Study File contents for the
Gulf of Mexico hypoxia modeling project will contain:
• Research and Quality Assurance Project Plan
• Applicable Operating Procedures related to modeling
• Study-related correspondence including Gulf of Mexico modeling meeting minutes
between MED-Grosse lie and Gulf Ecology Division; MED-Grosse lie and the RTP, NC
Environmental Modeling and Visualization Laboratory.
• Model archive describing where input and output files are located, source code, and any
other files related to running the Gulf models
• Electronic media with the field data used in the project will be placed in the study file
• Any peer-reviewed journal articles related to the project
Principal model documentation will be provided within the electronic model archive.
Documentation of the models will include a description of the model construct (including the
governing equations), model calibration and validation runs, model input and output, and
"readme" files. Sensitivity and uncertainty analysis results will be archived along with the model
computer code (both source and executable files). Internal documentation is also maintained in
the header comments of each program subroutine. A summary of field, literature, and external
data sources used in the model input, calibration, and validation process will be documented.
A complete description of the model equations, underlying assumptions, and numerical methods
can be found in several user manuals including CE-QUAL-ICM (Cerco and Cole, 1995), the
LM3-Eutro model (Paueref a/., 2006, 2008, 2011), and the LM3 model manual (Melendez et a/.,
2009). All functional changes made to the model program will be documented along with the
new code within the electronic Revision Control System (RCS) that maintains all versions of
modeling code used at LLRS and serves as the model code archive.
Various model products will be prepared throughout the project. These will include interim
reports, and at the request of management, oral presentations will also be given periodically.
Presentations at scientific meetings will be encouraged on any aspect of Gulf modeling. The
Gulf of Mexico modeling project would likely be classified as a QA Category II, or research of
high programmatic relevance which, in conjunction with other ongoing or planned studies, is
expected to provide complementary support of Agency rule-making, regulatory, or policy
decisions (USEPA, 2005). Because of this designation, significant findings from the study must
be published in peer-reviewed scientific/engineering journals. If publication does not occur, then
a formal review of the project and results will be required through a formal peer panel review
process.
September 26,2012 16 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
As a QA Category II project, the retention and disposition of project records must be in
accordance with the Agency's National Records Management Program led by the Office of
Environmental Information. Records retention schedule under Function Code 501 (Function
Number 316-258) would likely be applicable to the Gulf of Mexico project. These records are
first stored at the office that generated them for three years after the files have been closed;
then transferred to the National Archives and Records Administration (NARA); Federal Records
Center (FRC) for 20 years; and then a final transfer to the National Archives for permanent
archive. The MED Technical Information Officer will manage the transfer of the records to the
appropriate archival entity. Details can be found at EPA Records Schedules by Function Code
established in 2/20/2007 at http://www.epa.gov/records/policy/schedule/function.htm.
September 26, 2012 17 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
GROUP B: DATA GENERATION AND ACQUISITION
B.1 Model Formulation
Study Area
The study area is the Northern Gulf of Mexico and the Louisiana Continental Shelf. It extends
from the Mississippi Delta west to the Louisiana-Texas border, and from the shoreline seaward
to the 60-100 m bathymetric contours (approximately 26°N to 30°N by 88°W to 94°W - see
Figure 3). The average depth of the hypoxic zone is approximately 20 meters. The Mississippi
and Atchafalaya Rivers account for almost all of the freshwater entering this part of the Gulf.
This area is strongly stratified over the April to October period, largely due to salinity gradients.
Approximately 50% of the autochthonous material produced in this area settles to the sediment,
resulting in carbon and nutrient rich sediments. A description of the system and causes of the
hypoxia in the northern Gulf of Mexico was described in Section A.5. Details can be found
elsewhere (Rabalais and Turner, 2001; Rabalais et a/., 2002; Dagg et a/., 2007)
Modeling Framework
The model design for the Gulf of Mexico is based on the linked sub-model approach as was
used in the Lake Michigan Mass Balance Project (Pauer et a/., 2006). It consists of a water
quality model that includes eutrophication and DO kinetics that is driven by output from a
hydrodynamic model and a coupled sediment model. At this time, however, the sediment model
has not yet been incorporated into the model. The water quality model receives tributary
loading inputs directly and atmospheric nitrogen compound loads from an atmospheric fate and
transport model (CMAQ) developed and run by our collaborator, U.S.
EPA/ORD/NERL/Atmospheric Modeling Division. A schematic representation of the overall
mass balance design is shown in Figure 8.
Hydrodynamic Model
Hydrodynamic models developed and maintained by the Naval Research Laboratory (NRL) in
Stennis, Mississippi are being used to describe the hydrodynamics of the Gulf of Mexico for the
modeling framework. The screening model and the high resolution model use output from the
NRL hydrodynamic model EPACOM (Northern Gulf of Mexico Coastal Ocean Model for EPA)
(http://www7320.nrlssc.navy.mil/IASNFS_WWW/EPANFS_WWW/). This model covers the
coastal areas of Louisiana, Texas, Mississippi, Alabama, and part of Florida. For purposes of
the hypoxia modeling, only output from the study area is being used. The model uses a high
resolution grid that has an approximate size of 2 km x 2 km and 34-40 vertical layers, consisting
of 26 proportional-depth sigma layers on the shelf and 14 fixed-depth layers beneath the sigma
layers in deeper Gulf waters. The intermediate resolution model utilizes EPACOM vertical
mixing coefficients that has been aggregated into a 6 km x 6 km horizontal grid. Water
temperature and salinity are taken directly from measurements. The high resolution 2 km x 2
km version of the model (if needed) will use the original 2 km x 2 km output from EPACOM.
The main goal of the hydrodynamic model will be to generate three-dimensional fields of
currents and temperature in the Gulf. Currents are very important for the transport simulation of
state variables, while water temperature is a critical forcing function of algal growth. Other
parameters that the hydrodynamics model provides to the water quality model are
September 26, 2012 18 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
hydrodynamic vertical diffusion coefficients and salinity and sea surface heights. Cell volumes
are calculated using the sea surface heights, undisturbed water depths, and the sigma layer
percentages. Horizontal diffusion coefficients are not archived by the EPACOM model but have
been calculated from EPACOM model output. Diffusion coefficients are needed by the transport
part of the simulation. Diffusive and advective transport are non-trivial components of the
overall movement of particles in the water column. Grid cell volumes are also needed in the
transport calculation of the water quality model and in the calculation of state-variable
concentrations.
The aforementioned parameters will be calculated by the hydrodynamic model and averaged
over an appropriate time span for the selected periods for model calibration, corroboration, and
scenarios. The averaging interval used for the intermediate resolution model is one-hour for
flows and sea surface elevations and three-hour for temperature and vertical mixing coefficients.
Water Quality Model
The transport algorithm is based on the CE-QUAL-ICM and LM3-Eutro modeling frameworks
(Cerco and Cole, 1994; Pauer et al., 2006, 2008, 2011) that were applied in Chesapeake Bay,
Lake Michigan, and other systems. This algorithm describes the movement of nutrients,
phytoplankton and other constituents in the system.
The one-dimensional Gulf of Mexico hypoxia modeling framework, GoMDOM-1D, is largely
based on the three-dimensional Lake Michigan Eutrophication Model "LM3-Eutro" (Pauer et al.
2006, 2008, 2011; Melendez et al., 2009). However, a number of simplifications were made
which include a one-dimensional single vertical water column, 6km x 6km scheme, a single
phytoplankton state variable, and fewer nutrient state variables. However, the model uses a
revised light limitation formulation and simulates dissolved oxygen in the system. GoMDOM-1D
uses site-specific measurements and empirical relationships to determine nutrient and oxygen
demand sediment fluxes.
The one-dimensional approach is based on the assumption that horizontal advective flows and
diffusion across the boundaries of the column are negligible during the time scale of model
simulation (-100 hours), and thus the model is defined as an isolated, layered batch reactor.
This isolation allows for the parameterization of kinetic and vertical processes in the Gulf of
Mexico exclusive of the effects of horizontal transport. Defining process kinetics through
GoMDOM-1D assists in the calibration of the three-dimensional model, GoMDOM-3D model,
which is being developed concurrently with this work.
For the 3-D intermediate resolution GoMDOM model, the study area has a grid structure of 6 km
x 6 km horizontal segments and 26 vertical sigma levels. For the intermediate resolution model,
thickness, and thus volumes, of individual cells will vary significantly from relatively small cells in
the nearshore regions to much larger cells in the deeper areas of the Gulf. The hydrodynamic
transport and eutrophication kinetic equations will be incorporated into this high resolution grid.
The transport is based on the integrated compartment method or box model methodology which
is a loose extension of the WASP model (Ambrose et al., 1993). The box model concept will be
retained in order to allow the coupling, via map files, of the eutrophication/DO model with
hydrodynamic models of different dimensions and degrees of complexity. The transport will be
performed as a one-dimensional exchange between two adjacent cells through an individual cell
face, irrespective of the dimensionality of the model. The model will handle horizontal and
vertical transport during separate operations. The constituent transport equation can be written
as follows:
September 26, 2012 19 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
- — (Dx —) - — (Dz —) = 0 (B.1)
dt dx dz dx x dx dz z dz
F = constituent concentration (mass volume"1)
U = horizontal cell face velocities (length time"1)
W = vertical cell face velocities(length time"1)
Dx = horizontal cell face hydrodynamic diffusion coefficient (area time"1)
Dz = vertical cell face hydrodynamic diffusion coefficient (area time"1)
x = horizontal dimension
z = vertical dimension
This transport equation is solved using the third-order accurate Non-Uniform Grid ULTIMATE
QUICKEST algorithm (Leonard, 1991; Chapman et a/., 1997) in the horizontal and second-order
implicit Crank-Nicholson scheme in the vertical. A detailed discussion can be found elsewhere
(Melendez et a/., 2009).
Like the 1-D GoMDOM model, the kinetic equations for the 3-D GoMDOM are based on the
LM3-Eutro modeling framework (Pauer et a/., 2006, 2008, 2011; Melendez et a/., 2009). A
schematic diagram of the state variables and transformation reactions is shown in Figure 8.
General equations for phytoplankton (chlorophyll-a) and dissolved oxygen are shown below.
Detailed equations of the other variables and transformation equations can be found in
Appendix 1 (equations which were not described in LM3-Eutro) and elsewhere (Pauer et a/.,
2006, 2008, 2011; Melendez etal., 2009).
General phytoplankton equation
The kinetic change in phytoplankton concentration can be written as:
V dP - v k
~dt = 9~ d 9Z (B'2)
where
V = volume
P = phyt
t = time
kg = phytoplankton growth rate constant (time"1)
kd = phytoplankton mortality/respiration rate constant (time"1)
kgz = predation rate (time"1)
Z = zooplankton concentration (mass volume"1)
The growth rate can be written as:
kg=kgmaxf Nf Tf I (B.3)
where
= optimum growth rate constant (time"1)
P = phytoplankton concentration (mass volume"1)
September 26,2012 20 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
f(N) = nutrient growth dependency
f(l) = light growth dependency
f(T) = temperature growth dependency
General DO equations
The general dissolved oxygen equation, often expressed as the Enhanced Streeter-Phelps
equation, can be expressed as follows:
V—-—-= Reaeration - Carbon oxidation - Nitrification
dt (B.4)
- Respiration + Photosynthesis - SOD
where
V = volume
[DO] = dissolved oxygen concentration (mass volume"1)
t = time
Reaeration = oxygen exchange across the air-water interface (mass oxygen time"1)
Carbon oxidation = oxygen consumed due to organic carbon oxidation (mass oxygen time"1)
Nitrification = oxygen consumed due to ammonia oxidation (mass oxygen time"1)
Respiration = oxygen consumed due to algal respiration (mass oxygen time"1)
Photosynthesis = oxygen produced due to algal photosynthesis (mass oxygen time"1)
SOD = oxygen consumed due to sediment processes (mass oxygen time"1)
Detailed oxygen equations can be found in Appendix 1.
Sediment Diaqenesis Model
It is well known that the sediment is a major oxygen sink and an important contributor to the
problem of summer hypoxia in the Gulf of Mexico. Algae and detrital material settle to the
sediment bed and subsequent diagenesis of organic material occurs. This diagenesis process
results in nutrient and reduced carbon (oxygen demand) fluxes from the sediment to the water
column. A good understanding of sediment processes and formulation of a predictive sediment
diagenesis model is necessary to describe and predict nutrient fluxes and oxygen consuming
processes. Initially, nutrient and oxygen fluxes between the water column and sediment will be
described in the model using user-defined fluxes or as empirically-derived relationships based
on recent studies performed in the northern Gulf of Mexico (Murrell and Lehrter, 2011; Lehrter et
al, 2012). The empirical equation below (Murrell and Lehrter, 2011) represents the sediment
oxygen demand (consumption). It calculates the amount of dissolved oxygen per unit time per
unit area (kg O2/m2/s) that gets consumed or removed from the bottom layer of the water
column by the sediments.
SOD= 0.094xCDOx 106/32 -1.35 x3.7xl(T10 (B.4)
where
September 26,2012 21 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
SOD = sediment oxygen demand (kg O2/m2/s)
CDO = dissolved oxygen concentration of water column bottom layer (kg/m3)
A sediment diagenesis model will be developed when field data and process studies are
sufficient to support it. Figure 9 illustrates the sediment diagenesis model. The diagenesis
model will be based on the sediment model developed for and applied to Chesapeake Bay (Di
Toro and Fitzpatrick, 1993) and will be incorporated into the water quality model with the detrital
particles settling out of the water column onto the sediments. The sediments are represented
as two layers. The upper layer is in contact with the water and may be oxic or anoxic depending
on dissolved oxygen concentration in the overlaying water. The lower layer is permanently
anoxic. The depth of the upper layer is variable while the depth of the lower layer is fixed. A
general mass balance equation for the two layers can be written as follows:
= J + KL12 c(2) - c(1) - co2c(1)
(B.5)
H2 = co2c(1) + KL12 c(1) - c(2) - co2c(2)
where
H! = depth of surface layer (length)
H2 = depth of bottom layer (length)
c(1) = concentration in surface layer (mass volume"1)
c(2) = concentration in bottom layer (mass volume"1)
t = time
J = flux (mass area"1 time"1)
KL12 = mass transfer coefficient between layers (length time"1)
oo2 = sedimentation velocity (length time"1)
Detailed equations can be found in Appendix 1. Because the sediment model is a coupled
model (incorporated into the water quality model), it will be updated at the same time as the
water quality model. Although it can be difficult to obtain accurate values for sediment model
parameters, the full sediment diagenesis model has a major advantage over using user-defined
sediment fluxes for predictive capability.
Atmospheric Model
The atmospheric component of nitrogen load to the surface water of the Gulf is generally
estimated at 2% of the total nitrogen load to the Gulf. CMAQ (Community Multi-scale Air Quality
model), or Models3/CMAQ (http://www.epa.gov/asmdnerl/CMAQ/cmaq model.html), will
provide MED with atmospheric fluxes to the surface water segments of GoMDOM of both
reduced and oxidized nitrogen compounds. CMAQ and GoMDOM are run independently of
each other. Fluxes will be provided for both wet and dry (gaseous and particulate) deposition.
The CMAQ grid will be overlaid onto the GoMDOM grid. GIS tools will be used by the MED staff
to estimate a surface area-weighted flux to all surface water cells of GoMDOM to yield a load
September 26,2012 22 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
(mass/time). The aggregation of CMAQ fluxes into loads for GoMDOM is not expected to be a
major effort for the MED staff. CMAQ does not make estimates of atmospheric phosphorus
loads.
CMAQ is a grid-based chemical transport model (fixed, regular grid) that can be nested from a
continental/sub-continental scale (at 36-km and 32-km grids) down to finer scales over multi-
state geographic regions. Nitrogen deposition will be input from a 12 x 12 km grid for the
purposes of this project. It is driven by a mesoscale meteorological model (a weather model),
currently MM5 from Penn State/NCAR (National Center for Atmospheric Research).
Precipitation volumes are adjusted by the Parameter-elevation Regressions on Independent
Slopes Model (PRISM). CMAQ also incorporates the effects of lightning on the generation of
nitrates. CMAQ is not a calibrated model, per se, but tries to work as much as possible from
basic scientific theories. It outputs on an hourly time-step and requires significant computer
resources to run.
CMAQ computes the gas- and particle-phase concentrations of the inorganic N nutrients of
ammonia (reduced nitrogen) and nitric acid (oxidized nitrogen). In the eastern U.S., a majority
of oxidized-N air concentration is gaseous nitric acid, and a majority of reduced-N air
concentration is particulate ammonium. An aqueous chemistry and cloud module is used to
derive rainwater nutrient and pollutant concentrations for the computation of wet deposition,
given the precipitation predictions from PRISM. Dry deposition algorithms are parameterized for
different land use categories for the gases and particles to determine the dry deposition rates by
grid cell.
B.2 Model Coding
The water quality model is written using FORTRAN 90/95 programming language. This
programming language is suitable for models that require very intensive numerical
computations, such as the GoMDOM models. FORTRAN 90/95 has all the features that are
important to scientific programming and most of the features of an object oriented language.
The language is designed to generate executable codes that are highly optimized and, thus, run
extremely fast. FORTRAN 90/95 also supports parallel programming, making it an ideal
language for implementation of the water quality model on parallel computers if the hardware is
available.
The 3-D GoMDOM models are being run on high-end Linux-based computers. These
computers are relatively inexpensive and easy to maintain and update. The Linux operating
system offers the advantages of low cost, high stability, high performance, easy networking,
multitasking, compatibility with UNIX software packages, and high security.
The model source code will use external libraries which will be needed to handle input and
output tasks in addition to what FORTRAN 90/95 provides. The application will be reading large
sets of input data and at the same time writing a large amount of model calculations to output;
thus the use of a library to store and document the data in binary format will be required. The
library chosen to handle those tasks is known as the Network Common Data Form (NetCDF).
This library, in addition to using a binary format, allows the modeler to create, access, and share
array-oriented data in a form that is self-describing and portable. Self-describing means that a
dataset includes information defining the data it contains. Portable means that the data in a
dataset is represented in a form that can be accessed by computers with different ways of
storing integers, characters, and floating-point numbers. NetCDF will be implemented within the
water quality model by using a Fortran interface. This interface consists of a number of routine
September 26,2012 23 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
calls that will be made within the program to the NetCDF library. The library will be linked to the
water quality model at compilation time.
Libraries will also be needed to handle reporting messages that convey some kind of
information, warnings, and/or errors that might occur during the execution of the source code.
Reporting messages are important when performing simulations because they can let the
modeler know if the model is running smoothly or if bugs are present in the source code.
The systematic development of a model requires the use of a source code revision tracking
system. Thus the model will be archived and source code changes will be tracked using the
source-control system known as Revision Control System (RCS). RCS offers the ability to
record source code file revisions, retrieve previous file revisions, control new revision creation,
record description of changes made to a revision, control who can make source code
modifications, and specify user-access to source code files. RCS uses a separate archive file to
hold all the revisions of a given source file. Each revision of a file that is put into an archive file
is assigned a revision number. The archive files will be stored and maintained under the
computer account of the person in charge of managing RCS.
Input and output files related to the model will be archived under a designated directory in the
Linux system. Depending on the size of a given file, it will be stored in binary or ASCII format.
Large input and output files are better stored using a binary format to save disk space which
calls for the use of a library such as CDF or NetCDF.
B.3 Model Inputs
Parameter estimation
The Eutrophication/DO and sediment diagenesis sub-models consist of many biochemical
transformation reactions which require estimates for a large number of model parameters. For
this study, a number of coefficients will be obtained from Gulf of Mexico in-situ and laboratory
measurements, including phytoplankton growth parameters based on primary production
measurements and SOD coefficients based on laboratory studies. However, the majority of the
model parameters (similar to most other eutrophication/DO studies) will be based on values
from similar modeling studies and measurements reported in the literature. Several of these
model coefficients will be adjusted during the model calibration process (see Model Calibration
and Corroboration section) in order to obtain a final value. Adjusting of model coefficients will
be done within a reasonable range of reported literature values. Model parameter uncertainty is
discussed in Section B.6.
Initial conditions
Initial conditions for a number of variables (Table 2) in the water column and sediments are
required for the model. These include values for dissolved oxygen, dissolved and particulate
nutrient species, phytoplankton and zooplankton densities, organic carbon, and salinity in the
water column. Initial values for the sediment diagenesis variables include organic carbon and
nutrients. Data from the first and perhaps second field surveys of a specific year of interest will
provide the majority of the values for initial conditions for the model. However, it is possible that
values for some of the initial conditions cannot be calculated (directly or indirectly) from the field
surveys. In these cases, values will be determined using peer-reviewed literature, technical
reports, similar modeling studies, or unpublished databases. Data from peer-reviewed journals
have been subjected to a certain amount of review, but these data will be examined by the
September 26,2012 24 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
modelers for QA. Unpublished data from reliable sources will be examined thoroughly and
analyzed with respect to QA. Section A.7 details QA procedures for data obtained from outside
USEPA.
Loadings
Nutrient loadings from atmospheric and tributary sources are essential to the calibration of the
Gulf of Mexico model. Current loading estimates are also important when performing load
reduction scenarios to meet management objectives. Atmospheric nutrient loadings are being
provided by the atmospheric model (CMAQ) as was described in Section B.1. The USGS
calculates monthly nutrient loads in the Mississippi River from water quality data collected near
St. Francisville, LA and from flow data from Tarbert Landing, MS. Monthly loads for the
Atchafalaya River are calculated based upon water quality data from Melville, LA and flow data
from Simmesport, LA. Loads for nitrite plus nitrate and silica for the mainstem Mississippi River
extend as far back as October 1967. From October 1981 to September 2007 monthly load
values for both rivers are available for nitrite plus nitrate, total Kjeldahl nitrogen (organic nitrogen
plus ammonia), ammonia, total phosphorus, ortho phosphorus, and silica (Aulenbach et a/.,
2007; recent loads at: http://toxics.usgs.gov/hypoxia/mississippi/flux_ests/delivery/index.html).
In addition to monthly loads, nitrite plus nitrate and ortho phosphorus daily concentration and
flux values are available for the Mississippi River at Baton Rouge, LA and nitrite plus nitrate
daily concentration and flux values are available for the Atchafalaya River at Morgan City
(USGS, 2006). Water quality concentrations and flows from the USGS (USGS, 2006;
http://waterdata.usgs.gov/nwis/) are used to calculate loads for parameters not provided by the
USGS and for smaller tributaries that are not included in the USGS loading estimates.
Other forcing function estimations
Water temperature and advective and vertical dispersive flows are obtained from the output
from the hydrodynamics models as described in Section B.1. The NRL also provided solar
radiation and wind velocity data that were used in EPACOM. Horizontal dispersive flows are
calculated in-house using data from EPACOM and algorithms provided by EMVL. Flows from
the Mississippi and Atchafalaya Rivers will be obtained from the US Geological Survey
upstream stations (USGS, 2006). The study area modeled has a large boundary with Gulf of
Mexico open waters and is strongly affected by movement across this boundary. Boundary
concentrations are estimated from a number of cruise measurements made at stations adjacent
to the study area, while exchanges are determined from the hydrodynamics model.
Field data
Accurate and reliable field data (Table 2) are essential for model development, estimation of
model coefficients, and model confirmation. In support of this project and the modeling study,
sampling was undertaken by the USEPA Gulf Ecology Division. Multiple sampling cruises were
conducted during several years from 2003 to 2007. The Gulf was sampled for dissolved
inorganic nitrite plus nitrate, dissolved ammonium, particulate nitrogen, total dissolved nitrogen,
total nitrogen, physical parameters, biological parameters, and other chemical parameters.
Other field data have been collected by researchers associated with the Louisiana Universities
Marine Consortium (LUMCON) and Gulf Coast research institutions and universities which have
been sampling the Gulf of Mexico for many years. Data from the Gulf of Mexico Hypoxia
Monitoring Survey QAPP (Greene, 2003; Greene, 2007) will be subject to the QA procedures
specified in those documents. Data received from other reliable sources will be subject to QA
September 26,2012 25 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
procedures similar to those required for the Monitoring Survey. Details of QA can be found in
Section A.7.
B.4 Model Confirmation
Model confirmation is the process of reviewing the physical aspects of the model to ensure they
match the proposed processes. Model confirmation includes reviewing the model equations for
appropriateness to the physical, chemical, and biological attributes of the system under study
and for conformity to established theory; reviewing model computer code to verify that model
equations have been accurately implemented; and testing of completed individual code modules
for functionality.
A mathematical model consists of differential equations representing physical, chemical, or
biological processes in the system of interest. When a model is selected or designed for a
natural system, the proposed equations should be reviewed to make sure they appropriately
represent the system. The proposed model for this project is based upon models using
accepted formulations for eutrophication kinetics that have been successfully applied to
simulate eutrophication in estuarine (Cerco and Cole, 1994), saltwater (Hall and Dortch, 1994),
and freshwater systems (Pauer et a/., 2006). Scientific studies from the Gulf of Mexico
Louisiana-Texas shelf area will be reviewed to confirm that the proposed chemical and
biological kinetics are appropriate for describing processes in this region. If new equations are
required to describe processes not presently included in this model, they will be developed
based upon peer-reviewed scientific studies. The model construction, including the equations,
will also undergo an informal internal and external peer review process by scientific experts with
experience in the Gulf of Mexico to confirm its appropriateness for this project. Further peer
review is obtained during the publication process in a peer reviewed scientific/engineering
journal.
The Gulf of Mexico hypoxia model will be based upon the LM3-Eutro model, which has
undergone extensive testing, code review, and formal peer review (Pauer et a/., 2006; Melendez
et a/., 2009). This original code will not require further review.
New equations or changes to existing model code will undergo a rigorous review process. The
programmer responsible for translating model equations into code will provide an initial review.
The originating personnel providing the initial equations will also conduct a review of the code to
confirm that proposed equations or changes have been correctly implemented. Finally, model
code will be available for review during any peer review process. All changes to model code will
be documented and tracked through the RCS system (see Section B.8).
Any revised or new code will be tested to ensure it correctly calculates the embedded
equations. The output of any module that is revised or newly added will be tested before
including the module in the overall computer model. Output will be compared to hand-
calculated and/or spreadsheet derived analytic solutions and to results from previous versions.
Sensitivity analyses will be used to confirm that the module is calculating correctly.
The revised model will be tested to ensure that fundamental operations, such as continuity and
mass conservation, are verified. Tests will include checking of numerical stability and
convergence properties of model code algorithms, if appropriate. Model results will be checked
by comparing results to those obtained by other models and by comparison to manual
calculations. Visualization of model results and statistical correlations to field data will assist in
determining whether model simulations are realistic.
September 26,2012 26 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
B.5 Model Calibration and Corroboration
Calibration is "the process of adjusting model parameters within physically defensible ranges
until the resulting predictions give the best possible fit to observed data" (USEPA, 2009). The
model calibration will be accepted when the model simulates the majority of the cruise data
within the 95% confidence intervals (+/- two standard errors) for important measures of
constituents such as minimum dissolved oxygen concentrations, area and duration of hypoxia,
and concentrations of nutrients and phytoplankton. It is expected that the model fit to data will
be better than this criteria in many instances. For important constituents, the calibrated model
should also have a significant correlation at the 95% confidence level to field data stratified
appropriately in time and space.
Calibration of the proposed Gulf of Mexico hypoxia model will be conducted following a
systematic procedure. Initial parameterization will be accomplished as described in Section B.3.
Model parameters will be individually adjusted to determine the sensitivity of the model
simulation results to each parameter. Parameters will only be adjusted within ranges obtained
through project studies or published in the literature. Model calibration will start with the most
conservative constituents and proceed through constituents that depend on previously
calibrated constituents. Calibration efforts will focus on parameters with the largest uncertainty
and upon which model results have the largest sensitivity. The model will be calibrated against
a one-year data set from the 2003 to 2007 monitoring program data collected by EPA's Office of
Research and Development Gulf Breeze laboratory. Methods of calculating or estimating
loadings or other forcing functions may be refined, if necessary, but no calibration of forcing
functions will be allowed. The calibration will proceed until an optimal fit to data is achieved for
all important constituents. Goodness-of-fit will be assessed by qualitative comparison of model
results to data plots as well as by the quantitative statistical tests described in the preceding
paragraph.
There is an attempt within this document to help managers determine the degree to which the
models will be calibrated to field data. This constitutes the project acceptance criteria and
reflects what can practically be done with the resource commitments. The criteria for accepting
the modeling results lies in the ability to simulate measured concentrations of materials in water,
sediment, and biota during the field collection period. If this is done within the statistical range
required, then the model(s) can be used to extrapolate these concentrations in space and time.
Model corroboration, also called validation, includes the "quantitative and qualitative methods
for evaluating the degree to which a model corresponds to reality" (USEPA, 2009). The
calibrated model will be considered corroborated if model results for important constituents,
generated using a second independent set of inputs, fall within the 95% confidence intervals of
most of the data cruise means from the second data set. Corroboration will focus on important
measures of constituents such as minimum dissolved oxygen concentrations, area and duration
of hypoxia, nutrient concentrations, and phytoplankton concentrations.
For corroborating the model, the calibrated model will be compared against one year of data
from the 2003-2007 GED monitoring program independent of the calibration data set. Initial
conditions and external loadings will be obtained from the same independent data set. The
model will be run using these inputs, and model results will be compared against data means
and confidence intervals. The data from the calibration and corroboration years will be
compared to available long-term data sets to determine if these years were representative of
typical conditions. Goodness-of-fit will be assessed by qualitative comparison of model results
to data plots as well as by quantitative statistical tests.
September 26,2012 27 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
B.6 Model Sensitivity/Uncertainty Analysis
The sensitivity of the model to individual model parameters will be evaluated informally as part
of model calibration. Parameters will be adjusted one at a time and model results reviewed to
determine the sensitivity of the model to the adjustment. Model calibration will focus on the
most sensitive and uncertain parameter values.
After the model is calibrated and corroborated, a more formal sensitivity analysis will be
conducted. Individual parameters will be varied by a specified percentage as a measure of the
sensitivity of the model to each parameter and the change in model results recorded.
Parameterization error can be a significant source of model prediction uncertainty. Statistical
measures of uncertainty in the input data will be reviewed. This review will not only include data
used for model initialization, calibration, and corroboration but also the uncertainty in tributary
and atmospheric loading estimates and in the hydrodynamic predictions.
To evaluate and quantify the effects of parameterization error, uncertainty analysis will be
performed for selected model simulations. A statistical procedure to estimate uncertainty, such
as the parameter variance-covariance estimation procedure of Di Toro and Parkerton (1993),
and/or Bayesian Monte Carlo, propagation of error, and other statistical techniques will be
applied to estimate data, parameter, and model error components.
The uncertainty in forecast predictions is higher than in simulations of present conditions due to
the higher uncertainty in predicted inputs. For forecast predictions, the model will be run with
inputs, boundary conditions, and process rates bracketed in terms of extreme expectations and
probability distributions. The results will be provided in terms of prediction means and
exceedance limits.
Model results will also be qualified according to any explicit and implied assumptions made in
developing or applying the model. Managers will have to decide whether or not to use the
model results and whether or not to conduct additional research to improve the models. The
modelers can advise management on areas of input that have high uncertainty and to which the
model is very sensitive to. Of course, cost will also be taken into consideration. This is a
continuing process.
B.7 Model Application
After the model is calibrated and corroborated, it will be applied to assist in answering
management questions and to provide insight into physical, biological, and chemical processes
affecting hypoxia in the Gulf of Mexico. The model will be run using inputs developed according
to selected nutrient loading scenarios. Management questions will include running the model to
review the reduction in nutrient loads required to reduce the area of hypoxia to 5000 km2, a goal
specified in the 2001 Hypoxia Action Plan (Mississippi River/Gulf of Mexico Watershed Nutrient
Task Force, 2001). Subsequent scientific reviews have suggested that phosphorus reductions
may also be necessary to reduce the area of hypoxia, and additional loading scenarios may be
run to determine the model response to varying loads of both phosphorus and nitrogen.
Additional scenarios may be specified by project managers or peer review members during the
project. The model is not designed to make predictions on the effects of intensive, short
duration events, such as a hurricane.
Model output will be presented both graphically and in tabular form. The area and duration of
hypoxia will be important measures as well as nutrient and phytoplankton concentrations.
September 26,2012 28 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Scenario results will include comparisons to base line, target, and other scenarios. An
animation tool will be developed to present time series results of model scenarios for the area of
interest.
B.8 Data Management
All records, including modelers' electronic files, will be maintained according to Agency
standards as defined by the USEPA Office of Information Resources Management Federal
Information Processing Standards (FIPS). Paper notebooks will be issued to modelers that
meet ORD paper notebook guidelines. Notes in these books are of secondary importance to
documentation that will exist in electronic form. Minimum requirements for documenting and
maintaining ORD paper notebooks are covered in Chapter 13.2 - Paper Laboratory Records
(12/01/2006) of the ORD Policies and Procedures Manual at:
http://dcordhqapps1.epa. gov:9876/orma/policies.nsf/webPolicy?OpenView. Many of the
records associated with the project will be in the form of electronic mail (email). Email and
electronic records that are subject to both the Federal Records Act (FRA) and the Freedom of
Information Act (FOIA) will be preserved within EPA's Enterprise Content Management System
(ECMS). ECMS is a NARA - approved electronic recordkeeping system. These laboratory
notebooks and electronic files will be maintained by each modeler and turned over to the
Principal Investigator upon completion of the project. Similarly, electronic files containing
documentation of model testing, calibration, and validation will be maintained by each modeler
and transferred to a central project archive as designated by the Principal Investigator.
The primary water quality and process studies data used to support modeling activities will be
obtained from the GED Gulf of Mexico Hypoxia Study which has been subjected to an EPA QA
process. Secondary data from other studies, published and unpublished sources; other data of
opportunity; and equations, kinetics, process rates, and coefficients routinely used in modeling
applications can vary in their extent of QA examinations. All data, to the extent possible, will
undergo review as specified in Section A.7. A database tracking system has been instituted by
the LLRFRB (Large Lakes and Rivers Forecasting Research Branch), Grosse lie for modeling
systems. This database system will be used to provide data entry, storage, access, and
analysis capabilities to meet the needs of the modelers and other potential users of Gulf of
Mexico data. The system employs a single contact person for data being received. The contact
person logs in routine information about the data and coordinates its use. The process provides
updated versions if changes occur within the database. The second component of tracking
involves versions which have been assessed and completed for modeling purposes. Currently,
this position is provided through the Computer Sciences Corporation contract.
Development and production of software code is maintained at the LLFRB in the Revision
Control System (RCS). RCS forces strict revision control, supports check-out, locking, and
check-in of individual program files for development, and maintains a history and documentation
on all changes made to each program file. RCS allows the user to recover specific versions of
files so that they can be tested and re-used in the model. Documentation associated with
modifications made to files is stored in RCS and within the program file itself. This
documentation helps the user/modeler recall or understand why changes were made to files
over the course of the source code development. In order to facilitate code maintenance and
readability, standard programming style and code documentation will be followed (Melendez
and Griesmer, 2002; also see Section B.2).
An Operating Procedure will be made available to modelers of the project that will outline types
of model-related files that will be archived at the end of the project. At a minimum, sample input
September 26,2012 29 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
and output files, key calibration and final runs, model source code, compiled source code with
compiler identified, and any principal pre- and post-processing programs that were used to
support the models will be archived. Readme files will be included to describe any critical
information on running the model. The guiding principle in model archiving is to save whatever
is necessary to recreate the supportive files and model runs that have been selected. For
extremely large output files, saving only the first part of the run and last part of the run would be
permissible in order to conserve disk space. This would be explained in an accompanying
readme file. The directory structure for the model archive should be hierarchical with
meaningful names given to folders. A contiguous archival directory should be used for all of the
models applied to the project by a given organization. Please refer to Rygwelski (2005) for a
draft operating procedure for archiving models.
September 26,2012 30 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
GROUP C: ASSESSMENT AND OVERSIGHT
C.1 Assessments and Response Actions
There will be internal and external assessment of the model throughout development, including
the evaluation stage. Assessments will be conducted internally during the project by the
modeling team and project management staff. Review of model structures and implementation
are described in Section B.4. Reviews will be documented in project notebooks and internal
memoranda with responses documented accordingly. Project activities will also be reviewed
and assessed during bi-weekly meetings of the modeling staff, with assessments and
responses recorded in meeting minutes.
The external review will be conducted through a continuing informal review process. Review by
one's peers is an essential component to any successful and credible scientific/modeling
endeavor. Model development and application is a very complex process with many important
issues with multiple approaches available to address them. The external reviews provide an
objective means to arrive at a scientific consensus as well as provide judgment on scientific
credibility. Reviewers will include staff from the ORD/NHEERL Gulf Ecology Division and the
Office of Environmental Information, Environmental Modeling and Visualization Laboratory.
Assessments will be performed on the model structure and code (including the numerical
schemes used), estimations of model coefficients and forcing functions, test runs on other
computer systems, and reasonableness of model results.
External assessments will also include submission of articles to peer-reviewed
scientific/engineering journals. Because this project is likely to be assigned a QA Category
Level II status, publication in journals is a requirement. Otherwise, a formal external peer
review is required.
C.2 Reports to Management
The Principal Investigator will meet periodically with the modeling team and provide periodic
email and verbal communication to EPA management on model progress, questions, and
changes to the original modeling plan.
September 26,2012 31 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
GROUP D: VALIDATION AND USABILITY
D.1 Model Review
The mathematical model will be evaluated to determine how well the model meets the specified
data objectives and acceptance criteria (see Section A.7). These criteria include framework
evaluation, code verification, numerical methods accuracy, validation of input data, calibration
and corroboration results, and appropriateness of model scenario results. This section contains
a summary of the criteria that will be used for checking and accepting data and model output.
The theoretical and mathematical basis of model processes developed as part of this project
must be consistent with established scientific theories and modeling practices. Any new model
code must accurately describe the theoretical processes and be free from typographical errors.
New code must be consistent with the numerical solution technique and not cause significant
numerical dispersion or rounding errors.
Project-generated data used for the model inputs or for comparison to model output must have
passed the monitoring QAPP and an additional modeler review. Data from external sources
should preferentially have undergone a quality assurance process and will also be required to
pass a review by project modelers.
The model will be considered calibrated and corroborated when the model results are within the
95% confidence intervals of the majority of the data cruise means for important measures of
constituents for the respective data sets. For important constituents, the calibrated model
should also have a significant correlation at the 95% confidence level to field data, stratified
appropriately in time and space. Important measures of constituents include minimum
dissolved oxygen concentrations, area and duration of hypoxia, and concentrations of nutrients
and phytoplankton.
The model will be applied to selected loading scenarios to evaluate the effects of load on the
extent and duration of hypoxia in the Gulf and the effects of reducing both nitrogen and
phosphorus. There is no comparative data set for these model runs because the input data are
hypothetical predictions. However, to be considered valid, all model scenarios and results must
be approved by expert elicitation provided by the project peer review process.
D.2 Verification and Validation Methods
This section summarizes the methods that will be used to assess and verify that criteria
summarized in D.1 are met. Full descriptions of these procedures are contained in Section B.
The Gulf of Mexico hypoxia model will be based upon existing models. The majority of
theoretical basis, process formulation, numerical solution technique, and model code have been
previously verified, passed a formal peer panel review process, and published in peer-reviewed
scientific journals. These portions of the model will not require additional review. Any new
processes added to the model will be based upon established scientific studies and must pass a
peer review process. New model code will be documented, verified, and checked for numerical
accuracy as described in Section B.2 Model Coding and Section B.4 Model Confirmation.
September 26,2012 32 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Project-generated data will only be used after it has passed the monitoring QAPP acceptance
criteria (Greene, 2003, 2007) and an additional review of suitability for modeling by the project
modeling team. Data from other sources will preferentially be taken from sources that have
followed a QA-QC process or have undergone peer review. These data will undergo an
additional review by the project modeling team. Additional details of the data review and
acceptance procedure is included in Section A.7 Quality Objectives and Criteria for
Measurement Data and Section B.8 Data Management
The model will be calibrated and corroborated following the procedures listed in Section B.5
Model Calibration and Corroboration. The calibration and corroboration will be considered
acceptable when criteria listed in Sections D.land A.7 are met. Standard statistical tests will be
applied to compare model output with appropriately averaged field data to determine if the
criteria have been met.
Applying the model to run forecast scenarios is described in Section A.7 Quality Objectives and
Criteria for Measurement Data and in Section B.7 Model Application. Scenarios will include
those suggested by the management team to help evaluate restoration goals for the Mississippi
River/Atchafalaya River/Gulf of Mexico system. Because the scenarios involve predicted inputs,
there will not be field data for comparison to outputs. Inputs will be bracketed by developing
through expert elicitation likely minimum and maximum loads, and the results will be reviewed
through the peer review process to ensure the reasonableness of predicted outputs.
D.3 Reconciliation with User Requirements
The accuracy and reliability of models were quantified in Sections A.7 (Quality Objectives and
Criteria for Measurement Data), B.5 (Model Calibration and Corroboration) and B.6 (Model
Sensitivity/Uncertainty Analysis). The specific Data Quality Objective for this study was to
develop a modeling suite capable of simulating nutrient concentrations in the Gulf of Mexico to
within two standard errors of the means of observed concentrations in the water column. These
estimates were based on other modeling studies and the reported performance statistics of
equivalent modeling frameworks when applied to similar systems. The modeling project leader
will meet regularly with the users and managers to communicate the status of model
development including the model accuracy. The project leader will also inform the users how
changes in model accuracy will impact the application and interpretation of the model results
when performing load reduction scenarios.
September 26,2012 33 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
REFERENCES
Ambrose, R.B., Jr., J.L Martin, and T.A. Wool. 1993. WASP4, a Hydrodynamic and Water
Quality Model - Model Theory, User's Manual and Programmer's Guide. U.S.
Environmental Protection Agency, Office of Research and Development, Environmental
Research Laboratory, Athens, Georgia. EPA/600/3-87/039, 297 pp.
American Society for Testing and Materials (ASTM). 1992. Standard Practice for Evaluating
Mathematical Models for the Environmental Fate of Chemicals. American Society for
Testing and Materials, West Conshohocken, Pennsylvania. ASTM Standard E978-92, 8 pp.
Aulenbach, B.T., H.T. Buxton, W.T. Battaglin, and R.H. Coupe. 2007. Streamflow and Nutrient
Fluxes of the Mississippi-Atchafalaya River Basin and Sub-basins for the Period of Record
through 2005: U.S. Geological Survey Open-File Report 2007-1080 (with updated data at
http://toxics.usgs.gov/hypoxia/mississippi/flux_ests/delivery/index.html)
Bierman, V.J., Jr., S.C. Hinz, D. Zhu, W.J. Wiseman, Jr., N.N. Rabalais, and R.E. Turner. 1994.
A Preliminary Mass Balance Model of Primary Productivity and Dissolved Oxygen in the
Mississippi River Plume/Inner Gulf Shelf Region. Estuaries. 17(4):886-899.
Byun, D.W. and J.K.S. Ching. 1999. Science Algorithms of the EPA Models-3 Community
Multi-scale Air Quality (CMAQ) Modeling System. U.S. EPA/600/R-99/030, 611 pp.
Byun, D.W. and K.L. Schere. 2006. Review of the Governing Equations, Computational
Algorithms, and other Components of the Models-3 Community Multi-scale Air Quality
(CMAQ) Modeling System. Appl. Mech. Rev. 59, 51-77.
Cerco, C. and T. Cole. 1994. Three-Dimensional Eutrophication Model of Chesapeake Bay.
U.S. Army Corps of Engineers, U.S. Army Engineer Waterways Experiment Station,
Vicksburg, Mississippi. Technical Report EL-94-4, 658 pp.
Cerco, C. and T. Cole. 1995. User's Guide to the CE-QUAL-ICM Three-Dimensional
Eutrophication Model. U.S. Army Corps of Engineers, U.S. Army Engineer Waterways
Experiment Station, Vicksburg, Mississippi. Technical Report EL-95-15, 2, 420 pp.
Chapman, R.S., T.M. Cole, and T.K. Gerald. 1997. Development of Hydrodynamic/Water
Quality (POM-IPXMT) Linkage for the Lake Michigan Mass Balance Project. U.S.
Environmental Protection Agency, Office of Research and Development, National Health
and Environmental Effects Research Laboratory, Mid-Continent Ecology Division-Duluth,
Large Lakes Research Station, Grosse lie, Michigan, 51 pp.
Committee on Environment and Natural Resources. 2010. Scientific Assessment of Hypoxia in
U.S. Coastal Waters. Interagency Working Group on Harmful Algal Blooms, Hypoxia, and
Human Health of the Joint Subcommittee on Ocean Science and Technology, Washington,
D.C., 153pp.
Dagg, M., J. Ammerman, R. Amon, W. Gardner, R. Green and S. Lohrenz. 2007. Review of
water column processes influencing hypoxia in the northern Gulf of Mexico. Estuaries and
Coasts, 30:735-752.
September 26,2012 34 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Dennis, R.L., R. Haeuber, T. Blett, J. Cosby, C. Driscoll, J. Sickles, and J.M. Johnston. 2007.
Sulfur and Nitrogen Deposition on Ecosystems in the United States. EM: Air and Waste
Management Associations Magazine for Environmental Managers. Air and Waste
Management Association, Pittsburgh, PA, 12-17.
Dennis, R.L., P. Shave, and R.W. Pinder. 2008. Observable Indicators of the Sensitivity of
PM2.5 Nitrate to Emission Reductions, Part II: Sensitivity to Errors in Total Ammonia and
Total Nitrate of the CMAQ-Predicted Nonlinear Effect of SO2. Atmos. Environ. 42(6): 1287-
1300.
Di Toro, D.M., and J. Fitzpatrick. 1993. Chesapeake Bay Sediment Flux Model. U.S. Army
Engineer Waterways Experiment Station, Vicksburg, Mississippi. Contract Report Number
EL-93-2, 337 pp.
Di Toro, D.M. and T.F. Parkerton. 1993. Uncertainty Analysis of the Green Bay Mass Balance
Models - Abstract. Presented at the 36th Conference on Great Lakes Research,
International Association for Great Lakes Research, St. Norbert College, DePere,
Wisconsin, June 4-10, 1993.
Goolsby, D.A., W.A. Battaglin, G.B. Lawrence, R.S. Artz, B.T. Aulenback, R.P. Hooper, D.R.
Keeney, and G.J. Stensland. 1999. Flux and Sources of Nutrients in the Mississippi-
Atchafalaya River Basin. Topic 3: Report for the Integrated Assessment on Hypoxia in the
Gulf of Mexico. U.S. Department of Commerce, National Oceanic and Atmospheric
Administration, Coastal Ocean Program, Silver Springs, Maryland. NOAA Coastal Ocean
Program Decision Analysis Series Number 17, 130 pp.
Greene, R.M. 2003. Gulf of Mexico Hypoxia Monitoring Survey 2003-2005 Quality Assurance
Project Plan. U.S. Environmental Protection Agency, Office of Research and Development,
National Health and Environmental Effects Research Laboratory, Gulf Ecology Division, Gulf
Breeze, Florida.
Greene, R.M. 2007. Gulf of Mexico Hypoxia Quality Assurance Project Plan 2007-2010. U.S.
Environmental Protection Agency, Office of Research and Development, National Health
and Environmental Effects Research Laboratory, Gulf Ecology Division, Gulf Breeze,
Florida.
Greene, R., J. Lehrter, and J. Hagy. 2009. Multiple regression models for hindcasting and
forecasting midsummer hypoxia in the Gulf of Mexico. Ecol. Appl. 19:1161-1175.
Hall, R.W. and M.S. Dortch. 1994. New York Bight Study; Report 2; Development and
Application of a Eutrophication/General Water Quality Model. U.S. Army Engineer
Waterways Experiment Station, Vicksburg, Mississippi. Technical Report CERC-94-4, 308
pp.
Hetland, R.D. and S.F. DeMarco. 2007. How Does the Character of Oxygen Demand Control
the Structure of Hypoxia on the Texas-Louisiana Continental Shelf. J. Mar. Systems.
doi: 10.1016/j.jarsys. 2007.03.002.
Jassby, A. D., T. Platt. 1976. Mathematical formulation of the relationship between
photosynthesis and light for phytoplankton. Limnol. Oceanogr. 21:540-547
September 26,2012 35 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Justic, D., N.N. Rabalais, and R.E. Turner. 2003. Simulated Response of the Gulf of Mexico
Hypoxia to Variations in Climate and Anthropogenic Nutrient Loading. J. Mar. Systems.
42(3/4): 115-126.
Justic, D., V.J. Bierman Jr., D. Scavia, and R.D. Hetland. 2007. Forecasting Gulfs Hypoxia:
The Next 50 Years? Estuar. Coasts. 30(5): 791-801.
Ko, Dong S., R.H. Preller, and P.J. Martin. 2003. An Experimental Real-Time Intra Americas
Sea Ocean Nowcast/Forecast System for Coastal Prediction, Proceedings, AMS 5th
Conference on Coastal Atmospheric & Oceanic Prediction & Processes.
http://www7320.nrlssc.navy.mil/IASNFS_WWW/IASNFS_intro.html
Lehrter, J.C., M.M. Murrell and J.C. Kurtz. 2009. Interactions between Freshwater Input, Light,
and Phytoplankton Dynamics on the Louisiana Continental Shelf. Cont. Shelf Res.
29(15):1861-1872.
Lehrter, J. C., R. Devereux, D.L Beddick, D.F. Yates, and M.C. Murrell. 2012. Sediment-Water
Fluxes of Dissolved Inorganic Carbon, O2, Nutrients, and N2 from the Hypoxic Region of the
Louisiana Continental Shelf. Biogeochemistry. Springer, New York, NY, 109:233-252.
Leonard, B. 1991. The ULTIMATE Conservative Difference Scheme Applied to Unsteady One-
Dimensional Advection. Comp. Methods. Appl. Mechan. Engin. 88(1): 17-74.
Martin, P.J. 2000. Description of the Navy Coastal Ocean Model Version 1.0. Ocean
Dynamics and Prediction Branch, Oceanography Division, Naval Research Laboratory,
Stennis Space Center, MS. NRL/FR/7322-00-9962, 45 pp.
Martin, P.J., C.N. Barren, LF. Smedstad, T.J. Campbell, A.J. Wallcraft, R.C. Rhodes, C.
Rowley, T.L. Townsend, S.N. Carroll. 2009. User's Manual for the Navy Coastal Ocean
Model (NCOM) Version 4.0. Ocean Dynamics and Prediction Branch, Oceanography
Division, Naval Research Laboratory, Stennis Space Center, MS. NRL/MR/7320-09-9151,
73pp.
Melendez, W. and D. Griesmer. 2002. Standard Operating Procedure for EPA-Grosse lie
Programming Style and Documentation Guidelines. U.S. Environmental Protection Agency,
Office of Research and Development, National Health and Environmental Effects Research
Laboratory, Mid-Continent Ecology Division-Duluth, Large Lakes Research Station, Grosse
lie, Michigan. LLRS-ADP-SOP-002, 6 pp.
Melendez, W., M. Settles, and J. Pauer. 2009. LM3: A High Resolution Lake Michigan Mass
Balance Water Quality Model. U.S. Environmental Protection Agency, Office of Research
and Development, National Health and Environmental Effects Research Laboratory, Mid-
Continent Ecology Division-Duluth, Large Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-09/020, 285 pp.
Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. 2001. Action Plan for
Reducing, Mitigating, and Controlling Hypoxia in the Northern Gulf of Mexico. U.S.
Environmental Protection Agency, Office of Wetlands, Oceans, and Watersheds,
Washington, DC., 36pp.
September 26,2012 36 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. 2004. A Science Strategy to
Support Management Decisions Related to Hypoxia in the Northern Gulf of Mexico and
Excess Nutrients in the Mississippi River Basin: prepared by the Monitoring, Modeling, and
Research Workgroup of the Mississippi River/Watershed Nutrient Task Force. U.S.
Geological Survey Circular 1270, 58 pp.
Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. 2008a. Gulf Hypoxia Action
Plan 2008 for Reducing, Mitigating, and Controlling Hypoxia in the Northern Gulf of Mexico.
U.S. Environmental Protection Agency, Office of Wetlands, Oceans, and Watersheds,
Washington, DC., 61 pp.
Mississippi River/Gulf of Mexico Watershed Nutrient Task Force. 2008b. FY 2008 Operating
Plan. A Compilation of Actions to Implement the Gulf Hypoxia Action Plan 2008. U.S.
Environmental Protection Agency, Office of Wetlands, Oceans, and Watersheds,
Washington, DC., 41 pp.
Mitsch, W.J., J.W. Day, Jr., J.W. Gilliam, P.M. Groffman, D.L. Hey, G.W. Randall, and N. Wang.
1999. Reducing Nutrient Loads, Especially Nitrate-Nitrogen, to Surface Water,
Groundwater, and the Gulf of Mexico. Topic 5: Report for the Integrated Assessment on
Hypoxia in the Gulf of Mexico. U.S. Department of Commerce, National Oceanic and
Atmospheric Administration, Coastal Ocean Program, Silver Springs, Maryland. NOAA
Coastal Ocean Program Decision Analysis Series Number 19, 134 pp.
Murrell, M. C. and J.C. Lehrter. 2011. Sediment and Lower Water Column Oxygen
Consumption in the Seasonally Hypoxic Region of the Louisiana Continental Shelf. Estuar.
Coasts. 34(5):912-924.
Morse, J.W., and P.M. Eldridge. 2007. A Non-steady State Diagenetic Model for Changes in
Sediment Biogeochemistry in Response to Seasonally Hypoxic/Anoxic Conditions in the
"Dead Zone" of the Louisiana Shelf. Mar. Chem. 106(1-2):239-255.
National Research Council (NRC). 2007. Models in Environmental Regulatory Decision
Making. National Research Council of the National Academies, National Academic Press,
Washington, D.C., 286pp.
National Research Council (NRC). 2009. Nutrient Control Actions for Improving Water Quality
in the Mississippi River Basin and the Northern Gulf of Mexico. National Research Council
of the National Academies, National Academic Press, Washington, D.C., 79 pp.
Pauer, J.J., K.W. Taunt, and W. Melendez. 2006. LM3-Eutro. In: Rossmann, R. (Ed.), Results
of the Lake Michigan Mass Balance Study: PCBs Modeling Report, Part 2, pp. 120-182.
U.S. Environmental Protection Agency, Office of Research and Development, National
Health and Environmental Effects Research Laboratory, Mid-Continent Ecology Division,
Large Lakes Research Station, Grosse lie, Michigan. EPA-600/R-04/167, 579 pp.
Pauer, J.J., A.M. Anstead, W. Melendez, R. Rossmann , K.W. Taunt, and R.G. Kreis, Jr. 2008.
The Lake Michigan Eutrophication Model, LM3-Eutro: Model Development and Calibration.
Water Environ. Res. 80(9): 853-861
September 26,2012 37 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Pauer, J.J., A.M. Anstead, W. Melendez, K.W. Taunt, and R.G. Kreis, Jr. 2011. Revisiting the
Great Lakes Water Quality Agreement Phosphorus Targets and Predicting the Trophic
Status of Lake Michigan. J. Great Lakes Res. 37, pp. 26-32.
Pew Oceans Commission. 2003. America's Living Oceans - Charting a Course for Sea
Change. A Report to the Nation - Recommendations for a New Ocean Policy. Pew
Oceans Commission, Arlington, Virginia.
Pineiro, G., S. Perelman, J.P. Guerschman, and J.M. Paruelo. 2008. How to Evaluate Models:
Observed vs. Predicted or Predicted vs. Observed? Ecol. Model. 216: 316-322.
Rabalais, N.N., R.E. Turner, D. Justic, Q. Dortch, and W.J. Wiseman, Jr. 1999.
Characterization of Hypoxia. Topic 1: Report for the Integrated Assessment on Hypoxia in
the Gulf of Mexico. U.S. Department of Commerce, National Oceanic and Atmospheric
Administration, Coastal Ocean Program, Silver Springs, Maryland. NOAA Coastal Ocean
Program Decision Analysis Series Number 15, 203 pp.
Rabalais, N.N., R.E. Turner, and W.J. Wiseman, Jr. 2001. Hypoxia in the Gulf of Mexico. J.
Environ. Qual. 30(2):320-329.
Rabalais, N.N. and R. Turner (Eds.). 2001. Coastal Hypoxia: Consequences for Living
Resources and Ecosystems. American. Geophysical Union, Washington, DC., 454 pp.
Rabalais, N.N., R.E. Turner, Q. Dortch, D. Justic, V.J. Bierman, Jr. and W.J. Wiseman, Jr.
2002. Nutrient-Enhanced Productivity in the Northern Gulf of Mexico: Past, Present and
Future. Hydrobiologia. 475/476(1 ):39-63.
Richardson, W.L, D.D. Endicott, R.G. Kreis, Jr., and K.R. Rygwelski. 2004. The Lake Michigan
Mass Balance Project: Quality Assurance Plan for Mathematical Modeling. U.S.
Environmental Protection Agency, Office of Research and Development, National Health
and Environmental Effects Research Laboratory, Mid-Continent Ecology Division-Duluth,
Large Lakes Research Station, Michigan. EPA-600/R-04/018, 233 pp.
Rygwelski, K.R. 2005. Draft, Operating Procedure for U.S. EPA-Grosse lie Model Archiving
and Operating Guidelines. U.S. Environmental Protection Agency, Office of Research and
Development, National Health and Environmental Effects Research Laboratory, Mid-
Continent Ecology Division-Duluth, Large Lakes Research Station, Grosse lie, Michigan.
LLRS-MCM-OP-001,6pp.
Scavia, D., N.N. Rabalais, R.E. Turner, D. Justic, and W.J. Wiseman Jr. 2003. Predicting the
Response of Gulf of Mexico Hypoxia to Variations in Mississippi River Nitrogen Load.
Limnol. and Oceanogr. 48(3):951-956.
Scavia, D., D. Justic, and V.J. Bierman, Jr. 2004. Reducing Hypoxia in the Gulf of Mexico:
Advice from Three Models. Estuaries. 27:419-425.
Scavia, D.J. and K.A. Donnelly. 2007. Reassessing Hypoxia Forecasts for the Gulf of Mexico.
Environ. Sci. Technol. 41:8111-8117.
U.S. Commission on Ocean Policy. 2004. An Ocean Blueprint for the 21st Century.
Washington, DC.
September 26,2012 38 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
U.S. Environmental Protection Agency. 1991. Quality Assurance Guidelines for Modeling
Development and Application Projects: A Policy Statement. U.S. Environmental Protection
Agency, Office of Research and Development, Environmental Research Laboratory, Duluth,
Minnesota.
U.S. Environmental Protection Agency. 2000. Guidelines for the Preparation of MED Research
Plans. U.S. Environmental Protection Agency, Office of Research and Development,
National Health and Environmental Effects Laboratory, Mid-Continent Ecology Division,
Duluth, Minnesota.
U.S. Environmental Protection Agency. 2002a. Guidance for Quality Assurance Project Plans.
EPA QA/G-5. U.S. Environmental Protection Agency, Office of Environmental Information,
Washington, D.C. EPA/240/R-02/009, 111 pp.
U.S. Environmental Protection Agency. 2002b. Guidance for Quality Assurance Project Plans
for Modeling, EPA QA/G-5M. U.S. Environmental Protection Agency, Office of
Environmental Information, Washington, D.C. EPA/240/R-02/007, 121 pp.
U.S. Environmental Protection Agency. 2005. The Graded Approach to Quality Assurance and
Definitions of Research Categories. In: Quality Management Plan for the National Health
and Environmental Effects Research Laboratory (NHEERL), Attachment D. U.S.
Environmental Protection Agency, Office of Research and Development, Research Triangle
Park, North Carolina, QMP-NHEERL/99-01-001, 55 pp. URL:
http://www.nheerl.epa.gov/administration/qa/files/qmp05_attachments.pdf
U.S. Environmental Protection Agency. 2008a. Integrated Modeling for Integrated
Environmental Decision-Making. A White Paper by U.S. Environmental Protection Agency,
Washington, D.C. EPA100/R-08/010, 58 pp.
U.S. Environmental Protection Agency Science Advisory Board. 2008b. Hypoxia in the
Northern Gulf of Mexico. An Update by the EPA Science Advisory Board. Washington, DC.
EPA Science Advisory Board, EPA-SAB-08-003.
U.S. Environmental Protection Agency. 2009. Guidance on the Development, Evaluation, and
Application of Environmental Models. U.S. Environmental Protection Agency, Office of the
Science Advisor, Council for Regulatory Environmental Modeling, Washington, D.C.
EPA/1 OO/K-09/003, 90 pp.
U.S. Environmental Protection Agency. 2012. Safe and Sustainable Water Resources
Strategic Action Plan 2012-2016. U.S. Environmental Protection Agency, Office of
Research and Development, Washington, D.C. EPA 601/R-12/004, 38 pp.
U.S. Geological Survey. 2006. Real-time Streamflow and Water Quality (Mississippi River
Basin Discharge to the Gulf). U.S. Department of the Interior, U.S. Geological Survey,
Lafayette, Louisiana. URL: http://toxics.usgs.gov/hypoxia/mississippi/real time.html.
World Resources Institute. 2008. Eutrophication and Hypoxia in Coastal Areas: A Global
Assessment of the State of Knowledge. World Resources Institute Policy Note: Water
Quality: Eutrophication and Hypoxia, No 1. Washington, DC.
September 26,2012 39 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
September 26,2012 40 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
FIGURES
MED Gulf of Mexico Hypoxia
Modeling. Principal Investigator
Russell G. Kreis, Jr., EPA
MED Modeling Team
Kenneth Rygwelski, EPA-
Modeling Facilitator/Coordinator
Amy Anstead, ICF International
Phillip DePetro, ICF International
Timothy Feist .Trinity Engineering
Wilson Melendez, CSC
James Pauer, EPA
Mark Rowe, EPA
Xiaomi Zhang, Trinity Engineering
Hypoxia Modeling Project
Collaborators (See Figure 2)
Model Programming
Wilson Melendez, CSC
Data Management for Modeling
David Griesmer, CSC
Xiangsheng Xia, CSC
Figure 1. MED Gulf of Mexico Hypoxia Modeling Organization
September 26, 2012
41
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
_L
USER A Region 4
USER A Office of
Water
M ississi ppirtS ulf of M e*i co
Nutrient Task Force
USER A Office of
Wetlands, Oceans
and Watersheds
USEPAGulfof
Mesdco Program
Office
USER A Region 6
USEPA
ORD.NHEERL
Gulf Ecology
Division
USEPA
ORD.NHEERL
Mid-Continent
Ecology Division
USEPA
Office of Environmental Information
Environmental Modeling and Visualization
Laboratory
USEPA
ORD.NERL
Atmospheric Modeling
Division
Figure 2. Overview of Project Clients and Collaborators (with lines of communication
only)
September 26, 2012
42
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Louisiana Inner Shelf
Study Area
Figure 3. Gulf of Mexico Study Area
September 26, 2012
43
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
30WN
29'0'O-N-
MWN-
Bottom Oxygen (mg/L)
IMM
• H>9h 9
_ Low 0
• Surions
9«WW
93'0'0-W
92'0'0"W
9ro'0"W
Areal Extent of 2007 Hypoxic Zone
Data courtesy of N. Rabalais and A. Sapp
Figure 4. Areal Extent of 2007 Hypoxic Zone
90*0'O'W
September 26, 2012
44
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
z:a,uuu -
on nnn -
J*~* 11^ nnn
ID.UUU
I
ra
(D
c -m nnn -
5 nnn
, UUU •
0 •
1
fl n i
•
1-1
-
||.
i-i
101^-051-00101^-05
OOOOOOO5O5O5O5O5
(35(35(35(35(35(35(35(35
•
^^^^^^H
-
-
t- CO IO Is-
o o o o
o o o o
(N (N (N (N
1 1 Annual Hypoxic Zone Size — - Action Plan Goal
Data from Louisiana Universities Marine Consortium:
5-Yr Average (2004-2008) nttp://w w w gu,f hyp0xia.net/Overview /
Figure 5. Changes in Areal Extent of 1985-2008 Hypoxic Zone
September 26, 2012
45
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
i,ouu,uuu -
•\ Rnn nnn
1 ,DUU,UUU
"^
re
1 ?nn nnn
o
** 1 nnn nnn
•c
0>
Ennn nnn
ouu,uuu
•o
Spnn nnn
S
2 /inn nnn -FT n
4UU,UUU
+j
z n
onn nnn
01 i i
ii
m
m
(35
n n
n
n n
(35
m
(35
1
co
CD
(35
li
I
r^
CD
(35
t- m O5 co r^
r^ r^ r^ oo oo
(35 (35 (35 (35 (35
n n
n n n
n
f]
t- m O5 co r^
(35 (35 (35 O O
(35 (35 (35 O O
T- T- T- CM CM
Calendar Year
1955-1996 loads from Goolsby et al., 1999
1997-2007 loads from Aulenbach et al., 2007
Figure 6. Annual Nitrate Load to the Gulf of Mexico
September 26, 2012
46
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
^uu,uuu
1_
SJ? -i on nnn -
^^i
i_
9? -icn nnn -
M
O -i/in nnn -
+2 I4U,UUU
o
*j -ion nnn
0) IZU.UUU
£
T* -inn nnn -
y luu.uuu
Co
o
sn nnn -
y) OU,UUU
3
i«
O en nnn -
.C DU,UUU
Q.
w
o ,1 n nnn
^ 'HJ,UUU
Q.
5 on nnn
zu.uuu
|2
-
PI
CT)
r-
CT>
i
_ .
i-i
?—
DO
cn
i
i
<
c
(
,— i
*5
»
35
-
p.
S)
DO
35
-
"
r^
DO
ro
-
1
ro
DO
ro
i
-
s—
33
3)
-
1
•*3
3)
3)
L
(
<
n
3>
3>
~
1
rn
» —
3)
3)
1
i-i
3)
3)
3)
1
? —
O
•SI
1
r
^
"
L
(
<
f
-
n
D
M
-
1
(
(
(
n
* —
3
M
Calendar Year
data from Aulenbach et al., 2007
Figure 7. Total Annual Phosphorus Load to the Gulf of Mexico
September 26, 2012
47
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
hydrodynamic
model
(EPACOM)
atmospheric
model
(CMAQ)
other external
loadings and
exchanges
t t t
carbon cycle
zooplankton
A carbon
t |
phytoplankton , f detr
carbon * cart
J
^x ^*
dis
oxyg
1 1
water
column
^Kv
tal nutrient cycle
~ "^~ cycle
soLd ^: n'^een
GH CVCI6 i
1 f
1 1
sediment
diagenesis
j
Figure 8. Integrated, Multimedia Gulf of Mexico Modeling Framework
settling of
organic matter
1
1
nutrients, carbon,
and oxygen fluxes
t
\
sediment
diagenesis
burial \
water
column
sediment
Figure 9. Sediment-Water Interactions
September 26, 2012
48
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
TABLES
Table 1. Overall Project Schedule
MED Major Gulf of Mexico Project Activities
Project planning
Database development/management
Hydrodynamic model (EPACOM) incorporation into 3-D GoMDOM
Incorporate CMAQ atmospheric nitrogen fluxes into GoMDOM models
1-D GoMDOM model development
1-D GoMDOM model calibration
1-D GoMDOM model calibration / sensitivity / scenarios
1-D GoMDOM model journal article preparation
3-D 6km x 6km GoMDOM model development
3-D 6km x 6km GoMDOM model calibration
3-D 6km x 6km GoMDOM model corroboration
3-D 6km x 6km GoMDOM model management scenarios / sensitivity
3-D 6km x 6km GoMDOM model journal article preparation
3-D 6km x 6km GoMDOM model climate change scenarios
3-D 2km x 2km GoMDOM model parallelization
Fiscal Years
07 08 09 10 11
12
13
x x
X X X X
X
X
XXX
X
X
X X
X
X
X X X X
X
X
X
X
X
X
X
X
X
X
X
X
September 26, 2012
49
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Table 2. List of Desired Field Measurements
Total Nitrogen
Dissolved Inorganic Nitrate
Dissolved Inorganic Nitrite
Dissolved Inorganic Ammonium
Particulate Organic Nitrogen
Total Kjeldahl Nitrogen
Total Phosphorus
Dissolved Inorganic Phosphorus
Particulate Organic Phosphorus
Dissolved Inorganic Silica
Total Organic Carbon
Dissolved Organic Carbon
Particulate Organic Carbon
Total Suspended Solids 0.7um
Conductivity
Salinity
Chloride
PH
Alkalinity
Transmissivity
Temperature
Wind Speed / Direction
Dissolved Oxygen
Photosynthetically Active Radiation
(400-700 nm)
Incident PAR
Light Extinction
Chlorophyll Fluorescence
Fast Repetition Rate Fluorometry
(Fluorometry/Productivity)
Phyto Biomass HPLC Pigments
Phyto Biovolume Size Fraction
Atmosphere
Wet Dry Gas
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Gulf
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Tributaries
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Sediment
X
X
X
X
X
X
X
X
September 26, 2012
50
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
Error! Reference source not found, (continued).
C-14 Primary Productivity
Zooplankton Biovolume
Secondary Productivity
Microbial Productivity
Plankton Oxygen Demand
Porosity
% Water
% Solids
Reduction Oxidation Potential
Porewater
Sediment Oxygen Demand
Sediment-Water Nitrogen Flux
Atmosphere
Wet Dry Gas
Gulf
X
X
X
X
X
Tributaries
X
X
X
X
X
Sediment
X
X
X
X
X
X
X
September 26, 2012
51
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
APPENDIX 1: CONCEPTUAL EQUATIONS FOR DISSOLVED
OXYGEN AND SEDIMENT DIAGENESIS
The Gulf of Mexico hypoxia modeling framework, GoMDOM, is based on the eutrophication
model, LM3-Eutro, which was developed for Lake Michigan (Pauer etal. 2006, 2008, 2011).
This document describes equations added-to or changed from those in the LM3-Eutro model.
There were several major changes to the LM3-Eutro model. Dissolved oxygen (DO) was added
as a state variable with its transformation reactions. Denitrification equations are also briefly
described which indirectly affect the DO in the Gulf of Mexico. The GoMDOM model also uses
a somewhat different approach than LM3-Eutro to estimate the impact of solar radiation (light)
on algal production. The Jassby and Platt equation was used to estimate the limitation of
primary production by available light (Jassby and Platt, 1976, Lehrter et al 2009). Light
attenuation was calculated using a site-specific relationship between light attenuation and
chlorophyll, particulate carbon and salinity. A sediment diagenesis and flux sub-model are
planned for the future.
DISSOLVED OXYGEN EQUATIONS (WATER COLUMN)
The sources of DO in the water column include algal photosynthesis and reaeration. DO sinks
in the water column include algal respiration, organic carbon oxidation (bacterial respiration),
and chemical oxygen demand (COD), mainly sulfide oxidation and nitrification. DO sinks in the
sediment will be discussed in the SEDIMENT-WATER INTERACTION section.
Phytoplankton photosynthesis (PHOTO) and respiration (RESP) - see equation B.4 in Section
B.1 of main document.
Phytoplankton generate dissolved oxygen (photosynthesis) when sufficient nutrients, sunlight,
and "warmth" (temperature) are available and consume oxygen as a result of respiration.
Several equations have been proposed to describe these processes ranging form very complex
to rather simplistic approaches. We propose the following equation for these processes, which
is a simplification of the CE-QUAL-ICM (Cerco and Cole, 1995) equation.
= kg-kd P.AOCR (1)
where
P = phytoplankton carbon concentration (mass carbon-volume"1)
DO = dissolved oxygen concentration (mass-volume"1)
kg = phytoplankton growth coefficient (time"1)
kd = respiration rate (time"1)
AOCR = dissolved oxygen-to-carbon ratio in respiration (2.67 gO2-gC"1)
In general, the Arrhenius equation was used to calculate the effect of temperature on the many
reaction rate coefficients.
k(T) = k(opt) 6[T-T(opt)] (2)
where
k(T) = rate coefficient at temperature, T
September 26,2012 52 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
k(opt) = rate coefficient at optimum temperature, T(opt)
0 = temperature correction constant
Dissolved organic carbon oxidation - see equation B.4.
Dissolved organic carbon (DOC), for the most part, is present in the water because of the
decomposition of phytoplankton, zooplankton, and detrital particulate organic carbon. In the
presence of oxygen, DOC can be oxidized to carbon dioxide (CO2), an important oxygen sink.
Similar to the CE-QUAL-ICM model, the equation can be written as
(3)
dt Ksres + DO
where
Ksres = half-saturation cone, of DO required oxidation (mass-volume"1)
Kdoc = DOC oxidation rate (time"1)
Chemical Oxygen Demand (COD) - see equation B.4.
Chemical oxygen demand is the concentration (oxygen equivalents) of reduced species in the
water that can be rapidly oxidized chemically (absence of microorganisms). It is assumed that
the source of COD in the Gulf of Mexico is mainly due to sulfide released from the sediments.
Using an oxygen dependency, we can write the equation similar to CE-QUAL-ICM.
dt KsCOD + DO
(4)
where
COD = chemical oxygen demand (mass O2 equivalents-volume"1)
KSCOD = half-saturation concentration of COD (mass-volume"1)
KCOD = COD oxidation rate (time"1)
Nitrification - See equation B.4.
Due to the nature of the nitrifiers, it is generally accepted that nitrification occurs much faster in
the oxic regions of the sediment-water interface than in the water column. Even at these slow
rates, nitrification can be important in a relatively deep water column. Nitrification is typically
modeled in low oxygen systems as a double Monod equation with a dependency on both
oxygen and ammonia.
dNH4 DO NH4
= * * A 1 ) * kxJTT tC
dt KsDO + DO Ks^ + NH4 4 \°
where
KsDO = half saturation rate for DO (mass-volume"1)
KsNH4 = half saturation rate for NH4 (mass-volume"1)
f(T) = temperature function similar to other equations in the LM3-Eutro model
kNH4 = maximum nitrification rate (mass-volume"1-day"1)
Reaeration - See equation B4
It is generally accepted that reaeration in estuaries is largely dependent on wind effects (e.g.
Chapra, 1997). The general equation can be written as:
September 26,2012 53 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
D0sat-D0 (6)
Ql ^ '
where
K| = oxygen mass-transfer velocity (length-time"1)
A = surface area of the water body (area)
The Wanninkhof equation was used to calculate the oxygen mass-transfer velocity as follows:
( Sc Y-5
K1=0.108Ur l||J (7a)
where
Sc = Schmidt number (dimensionless number used to characterize fluid flows)
Uw = wind speed 10 meters above surface (length-time"1)
The equation can be simplified if a Schmidt number of 500 is used to represent the value of
oxygen in water.
Kj = 0.0986U"4 (7b)
Denitrification
In the absence of oxygen, nitrate can act as an electron acceptor during the oxidation of organic
matter. This process, known as denitrification, affects nitrogen and carbon concentrations.
Similar to nitrification, it can be modeled via a double Monod function.
• Kdoc /«\
,
Ksres+D0 Ksdemt.N03
where
Ratedenit = denitrification rate (time"1)
Ksres = half saturation cone, of DO for oxic respiration (mass.volume"1)
Ksdenit = half-saturation cone, of nitrate for denitrification (mass.volume"1)
ANOX = ratio of denitrification to oxic carbon respiration
Kdoc = respiration rate of DOC (time"1)
The subsequent mass balances terms for carbon and nitrogen are as follows:
dt
dNO3
dt
where
= - Ratedenit • ANDC • DOC
ANDC = mass nitrate-N reduced per mass DOC oxidized (mass N-mass C"1)
September 26,2012 54 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
SEDIMENT-WATER INTERACTION
A common approach used to describe the interaction between the sediment and the water
column is to assume a net sedimentation (difference between sedimentation and resuspension)
of phytoplankton and detrital material and to describe sediment kinetic reactions and transport
of nutrients, oxygen and carbon within and across the sediments. This is known as sediment
diagenesis (defined as all chemical, physical, and biological modifications undergone by a
sediment after its initial deposition) or sediment flux modeling. The approach will be in large
part based (at least as a first approach) on the work by Dl Toro and co-workers as described in
the CE-QUAL-ICM users guide (Cerco and Cole, 1995), the Chesapeake Bay project report
(Cerco and Cole, 1994), The Chesapeake Bay Sediment Flux model report (Dl Toro and
Fitzpatrick, 1993), and the textbook "Sediment Flux Modeling" (Dl Toro, 2001). The sediment
flux model includes the following state variables: phosphorus, ammonia, nitrate, silica, and
sulfide. Below is a discussion of each state variable and equations to describe their transport
and chemical conversions in the sediments and at the sediment-water interface.
Phosphorus flux
A schematic diagram of the transport and kinetic reactions is shown in Figure A-1. To simplify
the model, a partition coefficient is used to account for phosphate (dissolved form) in the pore
water and attached to the solids. Several other assumptions are also made:
1 . The sediment consists of two layers, a thin oxic layer (< 1 mm) and a much thicker
anoxic layer (-10 cm)
2. Because the aerobic layer is much smaller than the anaerobic layer, the particulate
organic phosphorus, POP, settles directly to the anaerobic layer
3. Thus no diagenesis occurs in this thin oxic layer
4. The anoxic layer has a fixed (user specified) thickness
5. The solids concentration (TSS) is constant within the sediments
6. The particulate organic phosphorus consist of several reactivity classes ranging from a
labile (highly reactive) to a refractory form
The phosphate mass balance in the oxic, H! and anoxic layers, H2 can be written as follows:
d[P04 1 1
H! L dt JT = s [P04 0 ]-fdl[P04 1 ]T (11a)
+ co12 fp2[P04 2 ]T-fpl[P04 1 ]T
+ KL12 fd2[P04 2 ]T-fdl[P04
d[po' 2 IT , f rPO 2-| _f
— \jj-i f-t JL o JL V_/ A ^ JL
^ 12 P2 L 4 JT i
- KL12 fd2 [P04 2 ]T - fdl [P04
[P04 1 ] - [P04 2
pl
o>2
September 26,2012 55 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
where
H! and H2 = thicknesses of sediment layers 1 and 2, respectively (length)
[PO4(0)]T, [PO4(1)]T and [PO4(2)]T = total PO4 concentration (which
includes the dissolved and solid sorbed fractions) of the overlaying
water, for layers 1 and 2, respectively (mass-area"1)
fd1 and fd2 = dissolved fractions in layers 1 and 2, respectively
fp1 and fp2 = particulate fractions in layers 1 and 2, respectively
These fractions can be calculated by the following equations (assuming the porosity is
approximately one)
fdi = — - fR = 1 - fdi (12a)
1 + nij • TTj v '
fd2 = - - - - fp2 = 1 - fd2 (12b)
1 + m2 '712
where
rrH, m2 = the solids (mainly Fe3+) concentration in layers 1 and 2, respectively
(mass-volume"1)
TTL TT2 = partition coefficients in layer 1 and 2, respectively (volume- mass"1)
KL12 = mass transfer coefficient between the aerobic and anaerobic
layers (length-time"1)
oo12 = particle mixing velocity between the aerobic and anaerobic
layers (length-time"1)
oo2 = burial velocity (length-time"1)
JP = source of phosphate from diagenesis of particulate organic
phosphorus, POP (mass-area"1 -time"1) and it can be estimated as follows
T -Vk a(T20).
Jp - /1/JKPOP,itl
i = l
POPi = concentration of particulate organic phosphorus in reactivity
class i (mass-area1)
kpopj = first-order reaeration rate coefficient (time"1)
0 = temperature correction constant
s = surface mass transfer coefficient between sediment and water (length. time"1)
The depth of the aerobic layer, \\-\ can be determined by the following equation:
where
DDO = diffusion coefficient in the aerobic layer (length2. time"1)
September 26,2012 56 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
For the participate organic phosphorus (POP) the mass balance equations can be written as
follows:
H2 — — - = -kpopjGpop; POP;H2 - (fl2POP; + fPOp,iJpop (15)
where
POPj = concentration of particulate organic phosphorus in reactivity
class i (mass-area"1)
kpopj = first-order reaction rate coefficient (time"1)
0 = temperature coefficient
T = temperature (°C)
oo2 = sedimentation velocity (length-. time"1)
JPOP = depositional flux of POP from the overlying water to the
sediment (mass-area"1 -time"1)
= fraction of JPOp that is in the ith G class
It is well known that the phosphate sediment fluxes (across the sediment-water interface) are
strongly affected by the water column DO concentration. In most systems this is caused by the
Fe3/Fe2 redox reaction. In moderate and high DO concentrations, the phosphate forms a
precipitate with iron, and thus a very small sediment phosphate release rate. In contrast, below
a critical DO concentration, phosphate is released to the overlying water. This can be described
mathematically:
it, = K2 ATTPO DO > DO pn (16)
1 / FlJ4,l cnt, PO4
where
[DO]crit, po4 = critical DO concentration (mass. volume"1)
ATT = enhanced sorption (in the oxic sediments)
Sulfide flux
A schematic diagram of the sulfide behavior in the sediments is shown in Figure A-2. In a
manner similar to phosphorus, we can write equations for sulfide. For dissolved sulfide, the only
difference is a reaction (oxidation) term in the oxic layer. The mass balance equation for sulfide
in the oxic layer (1) and anoxic layer (2), respectively, can be written as follows:
T =
0 ]-fdl[s2- 1 ] (17a)
KL12 fds2- 2
September 26,2012 57 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
2 ' ~fPiL5 A JT (17b)
-KL12 fd2[s2- 2 1 -fdl[s2- 1 1
LI/ a/ |_ JT ai |_ J
+ co2 [S2~ 1 1 - [S2~ 2 1 + J
2 L JT L JT
where
T
2
sz~
[S2"(0)]T, [S"2(1)]T and [S2"(2)]T = total sulfide concentration (which
includes the dissolved and solid sorbed fractions) of the overlaying
water, layer 1 and layer 2, respectively (g/m2)
The reaction term, kHi, and oxygen flux term Js can be written as follows:
klHl = kH2S,dlfdl + kH2S,plfpl 0H2S
'•M,H,S,O,
J - r, T _ r, Tl M rr I 0^)
S2~
The carbon diagenesis term, Jc, can be expressed as:
2 (20)
where
kh2s,di = reaction rate constant for dissolved oxidation (time"1)
kH2s,pi = reaction rate constant for particulate oxidation (time"1)
KM,H2s,o2 = scaling factor
ao2,H2s = stoichiometric coefficient (a value of 2.67 mgO2-mgC"1) is used
ac,N2 = stoichiometric coefficient (a value of 1.071 mgC-mgN) is used
Jc = carbon diagenetic flux (massC-area"1-time"1)
J[N2(9)] = nitrogen gas flux (massN-area"1-time"1)
kpoc,i= reaction rate (time"1)
Silica flux
A schematic diagram of the silica transport and kinetic reactions in the sediments are shown in
Figure A-3. The equations are similar to phosphorus and sulfide sediment diagenesis. The
mineralization of particulate silica is believed to be a chemical as opposed to a biochemical
(mediated by bacteria) reaction. It has been determined that the rate of biogenic silica
dissolution is proportional to the silica solubility deficit [Si]sat - [Si]aq, where [Si]aq is the dissolved
silica concentration .
The rate of dissolved silica production , SSi can be written as follows:
SSI = ksi esr PSI S!sat-[Sl aq]
September 26, 2012 58 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
From this we can calculate JT2 and k2 and can be written as:
JT2 = ksi ej-20 Si sat H2 (22)
k _ k A T-20 f (23)
K - K ° Z
2 - Si Si d2
The dissolved silica mass balance in the oxic layer (1) and the anoxic layer (2), respectively,
can be written as follows:
i 0]-fdl[Si 1]T (24a)
co12 fp2[Si 2]T-fpl[Si
KL12 fd2[Si 2 ]T-fdl[Si 1) ]T
[Si 1 ]T
- o>2
dfSi 2 1
H2 L JT=-k2H2[Si 2]T (24b)
-o)12 fp2[Si 2 ]T-fpl[Si 1 ]T
-KL12 fd2[Si 2 ]T-fdl[Si 1 ]
- co2 [Si 1 ] - [Si 2 ] + JT2
where
kSj = rate coefficient for silica dissolution rate (time"1)
k2 = rate coefficient (time"1)
PSI = particulate biogenic silica (mass. volume"1)
Ammonia and nitrate fluxes
Different from the phosphorus, silica and sulfide, the dissolved nitrogen species are present only
in the pore water (no solids partitioning). Other than that, the equations are very similar as
before. A schematic of the transport and kinetic reactions are shown in Figure A-4 for ammonia
and Figure A-5 for nitrate. The mass balance equations for ammonia and nitrate can be written
for the oxic layer (1) and anoxic layer (2) as:
d[NH4 1 1 (25)
H! L dt J = - k^ [NH4 l^-K™ [NH4 l]-[NH4 ( ( ]
+ KL12 [NH4 2 ]-[NH4 1 ]
d[NH4 2 1 r ., r ..
H2 J = -KL12 [NH4 2]-[NH4 l] + JN2
September 26,2012 59 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
d[N03 1 1
_L - =< = - kN03)1 [NO3 1 ]H! - KL01 [NO3 1 ] - [NO3 01 1 (26)
+ KL12 [NO3 2 ] - [NO3 1 ] + S NO3
d[N03 2 r n
H2 -L_J J = _ kN03,2 [N03 2 ]H2 - KL12 [N03 2 ] - [NO3
where
S[NO3] = is the source of nitrate from nitrification
The mass balance for the particulate nitrogen can be written as follows:
(27)
H2 = - kPOHi 9P0; PON.H, - co.PON, + fPON^ JPON
where
The nitrogen diagenetic flux can be written as:
JN = IXoN, epS? PON,H2 (28)
i=l
kNH4 = nitrification rate (time"1)
S[NO3]= nitrate generated due to nitrification (mass-area"1 -time"1)
Finally we can expressed the Sediment Oxygen Demand Flux (SOD - see equation B.4.) in
marine systems as the sum total of the oxygen demand due to the oxidation of sulfide (CSOD)
and nitrification (NSOD) within the sediments
SOD = CSOD + NSOD (29)
where
CSOD and NSOD can be defined as:
CSOD = a0^s J0 - aow J [N2 g ] ^ $ +^ S$ + ^ $ (30)
NSOD = a0i>NH. % 9-20 [NH4 1 ] (3D
o
where
frox(s) = fraction of sulfide being oxidized
fraq(s) = fraction of sulfide being loss due to mixing with the overlying water
frbr(s) = fraction of sulfide being removed due to burial
K2NH4,i = nitrification velocity (length. time"1)
September 26,2012 60 Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
WATER COLUMN
o r
S DC
2
HI
0
UJ
CO
UJ
SURFACE MASS TRANSFER: K|_Qf
PARTTnQNJNQ:
PO4 + FeQQH
PARTICLE MIXING
*12
DIAGENESIS:
POP
PO4 + FeQOH
SEOiME^fTATlON
w2
F«OOH=PO4
DIFFUSION
PO,
FeOOH«PO4
V
BURIAL
Figure A-1: Schematic diagram of the phosphorus transport and kinetic reactions in the
sediments.
September 26, 2012
61
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
o T*
CD CC
O LU
CC >
m 3
LU
S
5
LU
CO
CC UJ
UJ >'
< 5
WATER COLUMN
1
H2SCO)
SURFACE MASS TRANSFER: KLQ1 f
PARTntONiNG:
FiS
REACTIONS: H2S + 2 O2 -> H2SO4
FeS + 9/4 62 -> 1/2 Fe2Oa + H2SO4 • H2O
PARTICLE MIXING
DIfTUSlON
KL12
DIAGENESIS:
2 CH2O + H2SO4 -> 2 CO2 + H2S + 2 H2O
PARTITIONING: H2S
SEDIMENTATION
FeS
BURIAL
Figure A-2: Transport and kinetic reactions of sulfide in the sediments
September 26, 2012
62
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
o *-
LU
S
O
o cc
EC IJJ
WATER COLUMN
Si(0)
SURFACE MASS TRANSF1H:
PARTmONINQ:
SI H-
PARTICli MIXING
w12
KL12
JPSI
DIASENESS: PSi
SI
SOLU1IUTY:
Si
PARTmONINQ:
SI + FeOOH
SEDIMENTATION
BURIAL
Figure A-3: Silica transport and kinetics in the sediments
September 26, 2012
63
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
o *-
5 OC
o ui
ff >
til <
WATER COLUMN
NH4{0}
SURFACE MASS TRANSFER: KL01
JN1
|NH4(1)
DIAGENESIS:
REACTION;
DIAGENESIS:
REACTION:
PON
NH,
KNH4,1
NH4 - - ^ NO3
DIFRJSION: K|_12
PON
NONE
NH,
Figure A-4. Ammonia transport and kinetic reactions in the sediments
September 26, 2012
64
Revision 1
-------
Gulf of Mexico Hypoxia Modeling RQAPP
g *-
CD C
o uj
cc >
UJ <
< -J
LU
1
O
UJ
CD
o cc
CC 111
uu >
WATER COLUMN
NQ3(Q)
SURFACE MASS TRANSFER: «L01
KNH4,1
SOURCE:
REACTION:
SOURCE:
REACTION:
NHj
•N03<1)
N03
KNO3,1
NO3 *- N2(g)
DIFFUSION:
»
-------
Gulf of Mexico Hypoxia Modeling RQAPP
REFERENCES
Cerco, C. F. and T.M. Cole. 1994. Three-Dimensional Eutrophication Model of Chesapeake
Bay; Technical Report Number EL-94-4; U.S. Army Engineer Waterways Experiment
Station, U.S. Army Corps of Engineers: Vicksburg, Mississippi.
Cerco, C.F. and T.M. Cole. 1995. User's Guide to the CE-QUAL-ICM Three-Dimensional
Eutrophication Model (version 1); EL-95-15; U.S. Army Engineer Waterways Experiment
Station: Vicksburg, Mississippi.
Chapra, S.C. 1997. Surface Water-Quality Modeling, McGraw-Hill, New York, NY, pp844
Dl Toro, D.M. and J. Fitzpatrick. 1993. Chesapeake Bay Sediment Flux Model, HydroQual Inc.
Mahwah, NJ Prepared for the US Army Engineer Waterways Experiment Satiation,
Vicksburg, MS
Dl Toro, D.M. 2001. Sediment Flux Modeling, John Wiley & Sons, Inc. New York, NY pp 624
Jassby, A. D. and T. Platt. 1976. Mathematical formulation of the relationship between
photosynthesis and light for phytoplankton. Limnol. Oceanogr. 21: 540-547
Lehrter, J. C., M.M. Murrell, and J.C. Kurtz. 2009. Interactions between Freshwater Input,
Light, and Phytoplankton Dynamics on the Louisiana Continental Shelf. Cont. Shelf Res.
29(15):1861-1872.
Pauer, J. J., K.W. Taunt, and W. Melendez. 2006. In: Results of the Lake Michigan Mass
Balance Study: PCBs Modeling Report, Rossmann, R., Ed., Report EPA/600/R-04/167;
United States Environmental Protection Agency, Office of Research and Development,
National Health and Environmental Effects Research Laboratory, Mid-Continent Ecology
Division, Large Lakes and Rivers Forecasting Research Branch, Large Lakes Research
Station: Grosse lie, Ml, 2006; pp 120-182.
URL: http://www.epa.gov/medatwrk/grosseile site/LMMBP/pcb-report.html
Pauer, J. J., A.M. Anstead, W. Melendez, R. Rossmann, K.W. Taunt, and R.G. Kreis, Jr. 2008.
The Lake Michigan Eutrophication Model, LM3-Eutro: Model Development and Calibration.
Water Environ. Res. 80(9): 853-861
Pauer, J.J., A.M. Anstead, W. Melendez, K.W. Taunt, and R.G. Kreis, Jr. 2011. Revisiting the
Great Lakes Water Quality Agreement Phosphorus Targets and Predicting the Trophic
Status of Lake Michigan. J. Great Lakes Res. 37, pp. 26-32.
September 26,2012 66 Revision 1
------- |