566
of each forecast. From that time onward climatological mean daily river flow
is specified.
The second difference is the source of meteorological forcing. CBLAST
used a high-resolution COAMPS forecast specially run for the intensive
observing periods. Instead, we used the 72-hour ETA-12 forecast from the
U.S. National Centers for Environmental Prediction (NCEP). This choice of
a standard forecast product in widespread use was also partly motivated by
the wish to evaluate these data for forcing operational coastal ocean models
for any North American coastal observatory. An attractive feature of the
ETA-12 forecast is the availability of hourly data, as soon as each day of the
forecast completes, through an OPeNDAP (http://opendap.org) server at
NCEP. This means we are able to start the ocean forecast promptly and are
relieved of many data management tasks. The ETA-12 product includes
forecast downward shortwave and long-wave data, so that we need not
locate alternate sources of these as was necessary for the CBLAST 2002
hind-casts.
The remaining differences in configuration are minor. The initial
conditions on April 16, 2004, were zero velocity, and temperature and
salinity from an along-shelf average of all historical hydrographic station
data for April within 100 km of Hudson Canyon. The focus on a short-term
forecast for the duration of the dye experiment meant we neglected any
detailed treatment of the open boundary climatology.
The first dye release of the LATTE Pilot Program was on May 3, 2003,
just off Sandy Hook, New Jersey. Rather than proceed in a classic coastally
trapped buoyant plume, the dye promptly headed due east toward Long
Island and 36 hours later was dispersed on the coast. A second dye injection
on May 4 in a weak plume proceeded down the coast but reversed and
dispersed on May 5 with the onset of southerly winds. ROMS qualitatively
captured all of these flow changes (Figure 9).
This example is presented to illustrate that a coastal ocean modeling
system with useful forecast skill can be relatively easily constructed with
ROMS to support a coastal observational program. While the science
objectives of the LATTE program focus on fundamental studies of buoyant
plume transport and biogeochemical processes, an aspect of the program is
to demonstrate the capabilities of a coastal observatory comprised
principally of re-locatable observational systems; namely, gliders, CODAR
and ships. A readily configured model that does not require tuning or in situ
observations for initialization or open boundary conditions, and for which
forcing data (both river and atmosphere) are easily accessible by internet,
contributes to this objective.
JOHN L. WILKIN AND LYON LANEROLLE
of each forecast. From that time onward climatological mean daily river flow
is specified.
The second difference is the source of meteorological forcing. CBLAST
used a high-resolution COAMPS forecast specially run for the intensive
observing periods. Instead, we used the 72-hour ETA-12 forecast from the
U.S. National Centers for Environmental Prediction (NCEP). This choice of
a standard forecast product in widespread use was also partly motivated by
the wish to evaluate these data for forcing operational coastal ocean models
for any North American coastal observatory. An attractive feature of the
ETA-12 forecast is the availability of hourly data, as soon as each day of the
forecast completes, through an OPeNDAP (http://opendap.org) server at
NCEP. This means we are able to start the ocean forecast promptly and are
relieved of many data management tasks. The ETA-12 product includes
forecast downward shortwave and long-wave data, so that we need not
locate alternate sources of these as was necessary for the CBLAST 2002
hind-casts.
The remaining differences in configuration are minor. The initial
conditions on April 16, 2004, were zero velocity, and temperature and
salinity from an along-shelf average of all historical hydrographic station
data for April within 100 km of Hudson Canyon. The focus on a short-term
forecast for the duration of the dye experiment meant we neglected any
detailed treatment of the open boundary climatology.
The first dye release of the LATTE Pilot Program was on May 3, 2003,
just off Sandy Hook, New Jersey. Rather than proceed in a classic coastally
trapped buoyant plume, the dye promptly headed due east toward Long
Island and 36 hours later was dispersed on the coast. A second dye injection
on May 4 in a weak plume proceeded down the coast but reversed and
dispersed on May 5 with the onset of southerly winds. ROMS qualitatively
captured all of these flow changes (Figure 9).
This example is presented to illustrate that a coastal ocean modeling
system with useful forecast skill can be relatively easily constructed with
ROMS to support a coastal observational program. While the science
objectives of the LATTE program focus on fundamental studies of buoyant
plume transport and biogeochemical processes, an aspect of the program is
to demonstrate the capabilities of a coastal observatory comprised
principally of re-locatable observational systems; namely, gliders, CODAR
and ships. A readily configured model that does not require tuning or in situ
observations for initialization or open boundary conditions, and for which
forcing data (both river and atmosphere) are easily accessible by internet,
contributes to this objective.
JOHN L. WILKIN AND LYON LANEROLLE