Coupled PhysicalBioOptical Model Experiments at LEO15

1
Coupled Physical/Bio-Optical Model Experiments at
LEO-15
Hernan G. Arango1, Paul Bissett2, Scott M.
Glenn1, Oscar Schofield1 1Institute of Marine and
Coastal Sciences, Rutgers,University, New
Brunswick, N.J. 2 Florida Environmental Research
Institute, Tampa, FL
Bio-Optical Model
Downwelling Event
Bio-Optical Forecasts
Conclusions
Atmosphere-Ocean Models
Ocean color is a function of the mass of optical
constituents in the water column, the inherent
optical properties of those constituents, and the
apparent illumination of the ocean. One of the
goals of ONRs Environmental Optics HyCODE
program is to develop the forecasting ability to
predict the depth-dependent optical constituents
and their impact on water-leaving radiance.
One of the most intense events during the coastal
predictive skill experiment occurred in forecast
cycle 3, July 18-20. Strong NE winds from a
Northeastern resulted in a very intense
downwelling event along the New Jersey Coast.
Figure 7 shows the high resolution COAMPS
forecast for July 20, 1230 with surface winds
ranging between 25-30 m/s. The ocean response
is immediate, as illustrated in ROMS forecast
(Figures 8-10).
Figures 16-22 show seleted EcoSim predicted
ecological fields for July 20 along the A-line
section. These figures focus on the distribution
of pigmented biomass in response to the strong
downwelling event between July 18-20.
The fourth and final Coastal Predictive Skill
Experiment (CPSE) at the Rutgers University
Longterm Ecosystem Observatory (LEO) was
conducted from July 11 through August 7, 2001.
Ensembles of atmosphere and ocean forecasts were
generated twice per week for four consecutive
weeks. The Navy Operational Global Atmospheric
Prediction System (NOGAPS), and the atmospheric
component of the Coupled Ocean Atmospheric
Mesoscale Prediction System (COAMPS) was used to
drive the Regional Ocean Modeling System (ROMS)
to generate real-time forecasts of 3D circulation
in the Mid-Atlantic Bight (Figure 1). These
ocean forecasts were used to plan the daily ship
and aircraft operations during the CPSE and the
Hyperspectral Coastal Ocean Dynamics Experiment
(HyCODE).
Unfortunately, strong storm conditions usually
mean limited remote sensing data. The sky cleared
by the 21st and showed the impacts of the
downwelling conditions that drove the warm,
biomass rich waters in towards the shore (Figure
22). This can be seen as well in the predictions
of chlorophyll as well (Figure 23).
Figure 5 Depth dependent optical constituents.
Figure 1 Ocean Model bathymetry.
In pursuit of this goal, a previously developed
ecological model (Ecological Simulation, EcoSim)
that incorporated the Inherent and Apparent
Optical Properties (IOPs and AOPs) of the
water-column as a means of developing niche
separation between phytoplankton species in open
ocean environments was expanded and improved for
coastal ocean applications. The EcoSim model that
is coupled to the ROMS model in the New Jersey
Bight includes 4 functional groups of
phytoplankton that each includes stocks of
Particulate Organic Carbon (POC), Particulate
Organic Nitrogen (PON), Particulate Organic
Phosphorus (POP), and Particulate Organic Iron
(POFe), and for diatom groups, Particulate
Organic Silica (POS). These particulate ratios
of carbon, nitrogen, phosphorus, iron, and silica
are allowed to vary in non-stoichiometric
proportions to consider the impacts of
non-Redfield dynamics in both inorganic and
organic nutrient acquisition and growth. Each
phytoplankton species has a unique set of
photosynthetic and photoprotective pigments that
are allowed to vary as a function of light and
nutrient history. These pigments, when coupled to
the hyperspectral scalar irradiance light field
allow for the direct calculation of
photosynthetic efficiency and photon utilization
in the calculation of light dependent growth.
  The model also includes 2 classes each of
Colored Dissolved Organic Matter (CDOM),
Dissolved Organic Carbon (DOC), Dissolved Organic
Nitrogen (DON), and Dissolved Organic Phosphorus
(DOP). Bacteria in the form of carbon, nitrogen,
and phosphorus (BC, BN, BP, respectively) are
included as remineralizers of the organic
nutrient pools. The fecal forms of carbon,
nitrogen, phosphorus, iron, and silica, (FC, FN,
FP, FF, and FS, respectively) are included as
well. Lastly, stocks of carbon, nitrate,
ammonium, phosphorus, iron, and silica serve as
inorganic nutrient stocks. The total number of
ecological tracers for this version of EcoSim is
61.
Figure 7 MSL pressure and winds at 10 m.
Figure 14 A-line temperature observations.
Figure 15 Observed Chlorophyll Fluorescence.
Figure 22 SeaWiFS Observed Chlorophyll a.
The various data gathered by the observational
network at the LEO-15 is used to initialize,
update, and validate the coastal prediction
system. These included satellite derived sea
surface temperature, CODAR-derived surface
currents, CTD data from an undulating shipboard
tow-body, and an autonomous underwater Glider.
Ocean forecasts from each ROMS ensemble were
evaluated in real-time using ADCP and
remotely-profiled CTD data from the LEO
underwater nodes.
While the predictions appear to resolve general
features, the total quantity of biomass appears
to be lower than in the satellite-derived
estimates. This discrepancy appears in other
satellite/prediction comparisons, as well as
cross-sections from other days. This appears to
result in part from an over-estimation of the
biomass loss from grazing and mortality.
Figure 9 A-line temperature cross-section.
Figure 8 Temperature and currents at 2 m.
Figure 16 Total Chlorophyll-a (?g/liter).
Figure 17 Large diatoms Chlorphyll-a (?g/liter).
Figure 9 shows a forecast temperature
cross-section along the A-line. The water column
is well mixed nearshore. The thermocline front
is found intercepting the bottom at a depth of
10m, 7 km offshore. This downwelling event lasted
four days and affected the entire New Jersey
coast as shown in Figure 10.
Figure 2 LEO observation network, July 2001
Figure 23 SeaWiFS Observed Chlorophyll-a.
Nowcast and Forecast Cycles
The results shown here demonstrate the coupling
of an high resolution ecological model that
resolves multiple groups of phytoplankton, CDOM,
DOM, and nutrients with a high resolution
circulation forecasting model. Initial analysis
of the results suggest that the phytoplankton
interaction equations are behaving in a
reasonable fashion, with the individual
functional groups of phytoplankton separating
into niches that are best suited for their
mathematical description. One problem that
appears consistent in the comparison of data and
predictions is that the pigment concentrations
are lower than expected, suggesting on
over-estimate of the biomass loss terms. Further
analysis of the 61 ecological tracers, as well as
the IOPs, are expected to yield better
constraints on all of the ecological interaction
equations. The ecological model includes the
prediction of the inherent optical properties of
the water column, and thus will allow us to
directly forecast the water column remote sensing
reflectance, Rrs(?), and water-leaving radiance,
Lw(?). These optical predictions will be
compared to spectral mooring, gliders, ship,
aircraft and satellite data in an attempt to
validate optical measurements to optical
predictions, thereby directly closing the loop
between photon density measurements and numerical
modeling.
The prediction system included two regional
mesoscale atmospheric models (COAMPS) running at
coarse and fine spatial and temporal resolutions.
Both models were use to force the ocean model
ROMS via an atmospheric bulk boundary layer. A
current/wave/sediment boundary layer model is
attached at the bottom. The system is then
coupled to 61-components bio-optical model
(EcoSim).
EcoSim Model Formulation
Figure 10 13C and 18C Isotherms and winds.
Forecast Validation
Figure 19 Dinoflagellate Chlorophyll-a
(10-2?g/liter).
Figure 18 Small diatoms Chlorophyll-a (?g/liter).
Surface Stress
Bottom Stress
Figure 11 Thermistor observations time series at
COOL2.
Figure 12 Model validation mooring array
locations.
The ROMS forecasts were validated against data
collected at the validation array shown in Figure
12. A time series of the predicted temperature
(Fig 13), for the month-long experiment, is
compared with the thermistors data (Fig 12) at
COOL2 mooring. The time series show episodic and
alternating upwelling and downwelling events
typical of summertime conditions over the New
Jersey Coast.
Figure 20 Cyanobacteria Chlorophyll-a
(10-2?g/liter).
Figure 21 Silica concentration (?mol/liter).
Figure 3 Coupled models flowchart.
During July-August 2001, several real-time
atmospheric and oceanic forecast cycles were
carried out to predict the 3D coastal circulation
associated with recurrent summer upwelling and
downwelling events. An ensemble of 72-hour
forecasts were generated twice a week. Forecast
briefings were carried out on each Sunday and
Wednesday to optimize the adaptive sampling over
the next three days involving aircraft, ships and
AUVs.
Acknowledgements
Under downwelling conditions phytoplankton
biomass is pushed towards the coast in waters
that have become increasingly depleted in
nutrients. The predicted distribution of
chlorophyll-a appears to match the fluorometry
observations (Figure 15) on that day, with the
exception of near surface waters. This may be the
results of the model mathematical formulation, or
photo-quenching of fluorescence in the surface
waters. The distribution of phytoplankton shows
relative increases near the coast of non-silica
using dinoflagellates, and at deeper depths,
cyanobacteria.
Figure 6 Bio-optical compartments.
The advantage of the EcoSim model is that it
allows for estimates of the inherent optical
properties (IOPs) so it can be initialized using
the satellite-derived IOP estimates. The goal is
to begin to run ensemble biological forecasts,
which can be validated using the real-time data
from the field assets. This will be used to guide
the evolution of the biological model.
Initialization of the model will be based on
satellite estimates and glider measurements of
the inherent optical properties. The maps of the
inherent optical properties are deconvolved into
constituent end members using the algorithms,
developed through the ONRs HyCODE program,
providing fields for phytoplankton, dissolved
organics, detritus and sediments.  
We would like to thank a bunch of people, and
include the entire HyCODE field team. Naming them
individually would take up more space than we
have for poster. Special thanks, however, goes to
John Wilkin for his good humor and down under
patience. This work is funded by the Office of
Naval Research.
Figure 13 Predicted temperature time series at
COOL2.
For detailed evaluation of both atmospheric and
ocean models forecast skill, see posters by
Bowers et al. and Lichtenwalner et al.
Find this on the web http//marine.rutgers.edu/co
ol/coolresults/agu2002/
Figure 4 Coastal Predictive Skill Experiment
Calendar, July 2001.