FY 2006 PI Report

HYCOM Coastal Ocean Hindcasts and Predictions: Impact of Nesting in HYCOM
GODAE Assimilative Hindcasts
George R. Halliwell
MPO/RSMAS, University of Miami, 4600 Rickenbacker Causeway, Miami, FL, 33149
phone: (305) 421-4621 fax: (305) 421-4696 email: [email protected]
Lynn K. Shay
MPO/RSMAS, University of Miami, 4600 Rickenbacker Causeway, Miami, FL, 33149
phone: (305) 421-4075 fax: (305) 421-4696 email: [email protected]
Villy Kourafalou
MPO/RSMAS, University of Miami, 4600 Rickenbacker Causeway, Miami, FL, 33149
phone: (305) 421-4905 fax: (305) 421-4696 email: [email protected]
Eric P. Chassignet
COAPS, Florida State University, 2035 E. Paul Dirac Drive, Suite 200, Tallahassee, FL, 32310
phone: (850) 644-4581 fax: (850) 644-4841 email: [email protected]
Robert H. Weisberg
College of Marine Science, University of South Florida, 140 7th Avenue South,
St. Petersburg, FL, 33701
phone: (727) 553-1568 fax: (727) 553-1189 email: [email protected]
Alexander Barth
College of Marine Science, University of South Florida, 140 7th Avenue South,
St. Petersburg, FL, 33701
phone: (727) 553-3508 fax: (727) 553-1189 email: [email protected]
Harley E. Hurlburt
Naval Research Laboratory, Stennis Space Center, MS, 39529
phone: (228) 688-4626 fax: (228) 688-4759 email: [email protected]
Patrick J. Hogan
Naval Research Laboratory, Stennis Space Center, MS, 39529
phone: (228) 688-4537 fax: (228) 688-4759 email: [email protected]
Ole Martin Smedstad
Planning Systems, Incorporated, Stennis Space Center, MS, 39529
phone: (228) 688-4365 fax: (228) 688-4759 email: [email protected]
James A. Cummings
Naval Research Laboratory, 7 Grace Hopper Avenue, Stop 2, Monterey, CA, 93943
phone: (831) 656-5021 fax: (831) 656-4769 email: [email protected]
Award Number: N000140510892
The overarching goal is to improve our capability to map, understand, and predict changes in currents
and water properties in the coastal ocean. This capability is important for a wide range of purposes,
including naval operations, search and rescue operations, commercial marine operations (including the
influence of ocean currents on shipping and oil rigs), prediction of coastal hazards such as storm surge,
prediction of pollution dispersion, and the influence of currents and water properties on coastal
fisheries and ecosystems. Providing real-time (nowcast) and future (forecast) coastal ocean fields for
these purposes require a coastal ocean nowcast/forecast system that consists of several components:
1. a high-quality ocean general circulation model;
2. accurate surface flux (atmospheric forcing) fields to drive the ocean model;
3. accurate estimates of coastal ocean fields at the start of a model run and along the lateral
boundaries of the coastal domain during model runs;
4. accurate estimates of freshwater input from rivers and estuaries; and
5. high-quality ocean observations.
The observations are necessary to provide the accurate initial and boundary fields required by
component (3) and to evaluate the performance of the nowcast/forecast system. The central focus of
this project is component (3), specifically to quantify and understand the impact of initial and
boundary fields on coastal ocean nowcasts and forecasts, and to provide feedback that will motivate
improvements in generating these fields.
Although the coastal ocean is strongly influenced by surface atmospheric forcing and coastal
freshwater runoff, offshore ocean variability exerts a very significant influence in many regions due to
a wide range of processes such as basin-scale climate variability, boundary current meanders, and
offshore ocean eddies. To accurately represent the influence of this offshore variability on a coastal
ocean model, the model must be nested within fields that accurately represent (1) the initial state of the
coastal ocean throughout the model domain and (2) currents and water properties along the nested
model boundaries with the adjacent ocean. We are evaluating initial and boundary fields provided by a
nowcast/forecast system based on the HYbrid Coordinate Ocean Model (HYCOM) developed at the
Naval Research Laboratory (NRL-Stennis) as part of the Global Ocean Data Assimilation Experiment
(GODAE). Nowcast/forecast systems provide relatively accurate ocean fields by combining ocean
observations and ocean model output using data assimilation. The influence of the initial and boundary
fields on the performance of a coastal model are being thoroughly evaluated using HYCOM as the
nested coastal model. Results are being communicated to NRL to guide improvement strategies for
their nowcast/forecast system. The overall regional focus of this project encompasses the coastal
northern and eastern Gulf of Mexico along with the Florida Straits and Florida Bay, which represent a
broad range of shelf geometries, river runoff, seasonal atmospheric forcing changes, and both weak
and strong offshore forcing. Three regions are emphasized: the South Florida Coastal Region (SoFLA),
including the Florida Straits, Florida Keys, Florida Bay, and the adjacent southwest Florida shelf; the
West Florida Shelf (WFS); and the Northern Gulf Coastal Region (NGCR).
Although our central focus is on component (3), we also consider the other four components of the
coastal nowcast/forecast system. We are striving to improve ocean model performance, evaluate model
sensitivity to different atmospheric forcing products, study sensitivity to river runoff, and assess the
adequacy of available coastal ocean observations. Coastal ocean simulations are also being analyzed to
improve our scientific understanding of ocean variability observed in our region of interest.
The objectives are split into two categories: operational and scientific. Operational objectives include:
1. determining how changes in the initial and boundary conditions provided by nowcast/forecast
products (e.g. resolution, free-running versus assimilative, assimilation methodology)
influence the capability of nested models to simulate and predict the coastal ocean
2. evaluating the coastal hindcasts and predictions against observations that include existing
elements of the Coastal Ocean Observing System (e.g. SEACOOS, GCOOS);
3. identifying the most useful observations for evaluating and improving coastal ocean models
that should be maintained as part of a coastal observation network;
4. evaluating HYCOM development as a coastal ocean model against observations and against
simulations by other model types (ROMS and POM); and
5. offering feedback to guide improvement in the nowcast/forecast products that provide initial
and boundary information to nested coastal (and regional) models.
Specific scientific objectives include:
1. determining the impact of offshore currents (Loop Current and eddies, Florida Current; basinscale wind-driven gyre circulation) on coastal ocean circulation;
2. determining the impact of higher-frequency offshore variability (rectification), if any, on the
annual cycle of coastal circulation;
3. studying the impact of vertical mixing and friction (including the surface and bottom boundary
layers) on simulated coastal circulation, taking advantage of the multiple vertical mixing
choices (Halliwell, 2004) in HYCOM;
4. quantifying and understanding shelf-slope exchange processes;
5. studying the dynamics of transient eddies, including the Tortugas eddy in the Florida Straits in
the SoFLA domain;
6. determining the impact of Everglades runoff in the WFS and SoFLA domains;
7. understanding impact of the Mississippi river plume locally in the NCGR domain and
remotely in the WFS and SoFLA domains; and
8. studying the coastal ocean response to hurricanes.
We are providing physical fields from the West Florida Shelf and the South Florida domains to other
researchers interested in biochemical processes such as larval recruitment, red tide, etc (e.g., Olascoaga
et al., 2006).. We are also providing SoFLA fields to scientists and managers associated with the
Florida Bay and Florida Keys Feasibility Study of the Comprehensive Everglades Restoration Project.
The initial effort involves running and evaluating nested coastal ocean hindcasts, then using all
available ocean observations to evaluate coastal model performance. Sensitivity to initial and boundary
conditions is now being assessed by nesting the coastal models in fields provided by three products:
the original NRL experimental Atlantic Ocean product that used optimum interpolation to assimilate
ocean observations into model fields (HYCOM-OI), the next-generation NRL product being tested
within a Gulf of Mexico domain that uses multi-variate optimum interpolation to assimilate all
available ocean observations (HYCOM-NCODA), and a free-running Gulf of Mexico simulation
where no observations are assimilated (HYCOM-FREE). Other nowcast-forecast products generated
using more advanced techniques to assimilate ocean observations will be evaluated as they become
available. For evaluation, these runs are being compared to each other, to the fields that provided
initial and boundary conditions, and to in-situ observations. Evaluation of coastal models will also
determine how often boundary condition fields should be provided (daily or sub-daily), where the
boundaries of the nested domains should be situated, and how wide the boundary zones should be.
Model evaluation in each domain will take advantage of existing observing system programs. For the
SoFLA domain, this includes coastal radar (WERA) and an extensive observational network centered
in Florida Bay supported by NOAA for the Everglades Restoration Project. For the WFS domain, this
includes the extensive observational network maintained by the University of South Florida by R.
Weisberg. For NGCR, this includes observations acquired by NRL and other regional research
institutions. The most optimal runs identified by the evaluation effort will be used to conduct studies
that address the scientific objectives listed above. The initial evaluation and scientific analysis will
focus on 2004-2005. Longer time intervals will be analyzed later to study the annual cycle and
interannual variability. Following evaluation of coastal ocean hindcasts, the sensitivity of coastal
ocean forecasts to different initial/boundary products will be evaluated. The evolving boundary
conditions during coastal forecast runs will be provided by forecasts produced by NRL using their
nowcast/forecast systems.
The initial work naturally emphasized setting up the nowcast/forecast systems and generating the
ocean fields within which the coastal models are being nested. The original NRL HYCOM-OI
assimilation system routinely produced ocean fields during 2006. Initial development of the nextgeneration HYCOM-NCODA system was completed during 2006. For its initial evaluation, this new
system was run from September 2003 to the present within a new 0.04° Gulf of Mexico domain. Also,
a free-running simulation (HYCOM-FREE) was run for the same time interval in the same Gulf of
Mexico domain. These three products are the focus of our initial evaluation effort, which will be
conducted for a two-year time interval (2004-2005).
While waiting for these outer model runs, progress was made in setting up and running HYCOM in the
coastal domains. For the SoFLA domain, an initial experiment spanning 1999-2002 nested in an old
free-running model simulation was evaluated to provide benchmark results for the evaluation of the
three products during 2004-2005. A new SoFLA domain was set up for the 2004-2005 evaluation
effort by extending the northern and eastern boundaries to allow a better representation of coastal to
deep sea interactions. Meanwhile, HYCOM was configured to run in the WFS domain using the same
curvilinear horizontal grid used at the University of South Florida to run other model types. Coastal
simulations in the SoFLA and WFS domains using the three choices for initial/boundary conditions are
now underway and will be completed during January 2007. Setup of the third (NGCR) coastal domain
has just been completed and simulations within this domain will commence shortly. Preliminary
results were presented during November 2006 at the HYCOM Principal Investigators meeting.
Figure 1. SoFLA domain sea surface height (SSH) on 5/29/04 from twin experiments with initial
and boundary conditions from the HYCOM-FREE (left) and HYCOM-NCODA Gulf of Mexico
simulations (right).
Initial 2004-2005 results in the SoFLA domain demonstrate that the nesting procedure performed very
well, as currents and water properties in the vicinity of the open boundaries exhibited a smooth
transition between the outer and nested domains. Preliminary analysis of sea surface height (SSH)
fields for the simulations with boundary conditions obtained from HYCOM-FREE and HYCOMNCODA Gulf of Mexico simulations revealed noticeable changes in the location of the Loop Current
and Florida Current fronts along with the associated eddies (Figure 1). Comparison with satellitederived sea surface temperature (SST) and ocean color maps (not shown) demonstrate that the data
assimilation used in HYCOM-NCODA produced initial/boundary conditions that more accurately
reproduced the location of these ocean features.
At the University of South Florida, a test WFS simulation for 2005 using the ROMS (Regional Ocean
Modeling System) coastal model nested within the HYCOM-OI product was evaluated against water
column currents from Acoustic Doppler Current Profiler measurements, surface currents from highfrequency radar measurements, and water column temperature-salinity from profiling floats. The
coastal simulation showed reasonable agreement with the ocean observations. In particular the model
was able to simulate the strong vertical temperature-salinity stratification observed prior to Hurricane
Katrina that was associated with a benthic die-off due to the trapping of red tide organisms (K. brevis)
beneath the pycnocline. The ROMS simulation performed better when nested in the HYCOM-OI
product compared to nesting ROMS within climatological ocean fields.
Figure 2. Time series of temperature over the first 10 months of 2004 at two locations over the West
Florida Shelf west of Tampa Bay. Time series are shown for two depths at station C12 (middle
continental shelf) and at one depth at station C16 (outer shelf). Black lines are observed
temperature, red lines are simulated temperature using HYCOM-FREE initial/boundary conditions,
and blue lines are simulated temperature using HYCOM-NCODA initial/boundary conditions.
At the University of Miami, the HYCOM coastal model runs on the same model grid used by USF.
Initial comparison of observed and simulated temperature time series at two locations in the WFS
domain (Figure 2) demonstrate that the error in simulated temperature as measured by the root-meansquare difference between time series decreased by 15 to 30% when the more accurate initial and
boundary conditions provided by HYCOM-NCODA were used instead of fields provided by HYCOMFREE. This result is encouraging, and these and other detailed comparisons between observed
temperature, salinity, and current velocity will be performed in different regions to understand the
impact of changing initial and boundary conditions and determining the best source of these ocean
fields for nested coastal ocean simulations.
National Security
This project will improve the capability of HYCOM to hindcast and forecast currents and water
properties in the coastal ocean. HYCOM will be transitioned to the U. S. Navy for operational use by
FY2008, and will be used to provide information on ocean currents and water properties for naval
Quality of Life
Improved performance of coastal ocean models will improve our capabilities in commercial marine
operations (including the influence of ocean currents on shipping and oil rigs), storm surge prediction,
prediction of pollution dispersion, studies of coastal fisheries and ecosystems, and providing ocean
currents for search and rescue operations.
Science Education and Communication
As part of this project and the HYCOM-GODAE project, we are making research results available on
the internet – see http://hycom.rsmas.miami.edu. These projects are supporting graduate students.
National Security
The ocean model used in this study (HYCOM) will become the operational ocean model used by the
U. S. Navy by FY2008. The work performed under this project will lead to model improvements that
will positively impact Navy operations.
The HYCOM-GODAE product is being developed as part of a NOPP project funded by ONR and
NOAA to develop a global ocean data assimilation system that will be used to both nowcast and
forecast the state of the global ocean. This project will provide information that will be used to
improve the HYCOM-GODAE product.
Olascoaga, M. J., I. I. Rypina, M. G. Brown, F. J. Beron-Vera, H. Kocak, L. E. Brand, G. R. Halliwell,
and L. K. Shay, 2006: Persistent transport barrier on the West Florida Shelf. Geophys. Res. Lett., in