Ocean modeling and prediction for LaTTE from Rutgers ROMS group

The forecast posted yesterday evening at 1815 has not changed. The model showsa plume that was previously filamentous due to primarily shear dispersion will be rapidly dispersed horizontally by the onset of strong winds beginning late Saturday evening (local time).

Recent comparisons to the LEO Node profile data show a severe bias in the model salinity, but reasonable simulation of temperature variability. Stratification increases rapidly on May 5 with the arrival of warm water in the coastal plume. Both model and data show an associated freshening, but the too salty bias of the model persists.

Comparison of surface temperature from the model with surface underway data from RV Cape Hatteras shows the model gets the variability pattern correct but with a consistent cool bias of 1^oC. The salinity patterns are also similar, but again with a severe salty bias compared to observed surface salinity from the ship averaging 28 compared to model of 31.

We suspect that these biases are due in part to poor initial conditions, but comparisons with the gliders also implicate an erroneously strong onshore flow in the southern half of the region where the ships are operating, that would bring overly salty water into the model in this region. Further ship observations between 40^oN and Barnegat Inlet would help constrain the future hindcast simulations with data assimilation.

2006 May 5 1815
      Dye situation:   The model simulated dye patch is being drawn out around the perimeter of the warm circulating feature adjacent to the coast. Comparisons to the drifters being used to follow the dye show encouraging results for the model validity (white squares show drifter locations). At 2006-05-05 1900 UT vertical slices though the modeled dye patch show the dye is deepest on the northern side of the patch, possibly consistent with some subduction of the water at the frontal boundary of the warm water feature. The dye is at most 8 m deep. It is less deep (2 to 4 m) at the trailing edge. About half the dye is remaining against the coast where it has been mixed to the bottom in depths less than 7 m. There is some evidence for offshore movement beneath the mixed layer immediately adjacent to the coast. The outlook for the dye movement is that the dye in the offshore patch will encircle the eddy by late Saturday (local time). The forecast onset of strong northerly winds at around 10 pm EDT Saturday will move the entire pattern southward and accelerate the lateral dispersion of the patch. However, comparisons of the model stratification with RU07 glider sections near the coast indicate that the circular warm feature in the model may be unrealistic, and that the flow may in reality be continuing offshore. The trajectories of the drifters and dye from noon local time on Saturday will provide a significant test of the model forecast skill in this respect.
     Ocean situation: WNW winds at Ambrose assist the exit of freshwater from NY Harbor on the ebb tide today. This freshwater input establishes a plume along Sandy Hook http://marine.rutgers.edu/po/latte/2006forecasts/wrf/s2m/s2m-njcoast.20060506.0500.wrf.html    The plume pushes down coast during Saturday, possibly sustained by its own buoyancy forcing, carrying low salinity water into the warm eddy that has entrained the Latte dye release. Strengthening northerly winds Saturday night into Sunday morning (local time) push the flow further south, but it splits into two trajectories. Part separates from the coast and enters the circulation around the eddy, following the dye, while the rest continues southward in the coastal current reaching 40^oN at the end of the present forecast cycle (2006-05-07 1100 UT) http://marine.rutgers.edu/po/latte/2006forecasts/wrf/s2m/s2m-njcoast.20060507.1100.wrf.html

Previous ocean forecast bulletins:

2006 May 4 2045 EDT The dye released in the coastal current this morning is predicted to quickly move southward, but to stall at about the latitude of the C mooring line in late evening on Thursday May 4 (local time) due to the onset of southerly winds late in the day. This could cause the plume water transporting the dye to disperse with some of it moving offshore in the surface layer, while part remains against the coast. The subsequent switch to winds from the north on Friday morning and variable winds thereafter will further disperse and mix the plume water but keep it in the general vicinity of the C mooring line and original release site.

2006 May 3 1620 PDT The offshore drift of the warm eddy off Asbury Park stalls during Wednesday. This causes the coastal current to feed into the eastern flank of the eddy in an almost continuous stream in a pattern with similarities to the recognized ‘highway’ path. This flow is assisted by mild northerly winds until early Thursday (local time). The next ebb tide on Thursday afternoon (local) amplifies the flow but is abruptly arrested by the onset of southerlies late Thursday evening. Southerlies changing to southwesterlies on Friday move surface waters offshore and may lead to a weakening or breakup of the continuous coastal current.

2006 May 3 0800 PDT ROMS model is showing the warm eddy at the C-line separating from the coast during Tuesday night (local). The satellite SST gives some indication of similar behaviour but whether these features in model and satellite correlate well is unknown. If the model is correct, and the separation continues through Wednesday p.m. (local), then the coastal current would have a clear path through to Tuckerton. Gregg Foti's idealized dye release scenario shows dye joining the eddy, but a later release time could give a qualitatively different result with dye remaining close to shore

2006 May 2 1300 PDT Light winds and low flow continue to favor the Hudson River discharge entering a coastal plume along the NJ coast during Wednesday. As the ebb tide wanes and flood commences on each tidal cycle during late Tuesday and Wednesday, the water drawn into Raritan Bay is initially from the north coast of Sandy Hook so that the now familiar bifurcation pattern in the Sandy Hook coastal currents occurs. The warm anomaly that the model shows hovering adjacent to the C mooring line persists but may be breaking up through interaction with currents offshore on the southwestern side of the Hudson Shelf Valley. The short length scales and variability in this feature, and its displacement slightly to the north of the similar feature in satellite SST, leaves uncertainty in the accuracy of this aspect of the ocean forecast. However, the presence of the warm anomaly in satellite SST does suggest that the ocean model is correct that this feature marks the terminus of the coastal plume, and that material advected downstream in the coastal current is likely to abruptly turn seaward upon reaching the northern side of the warm feature.

2006 May 1 1700 PDT While northerlies persist through to 2:00 am EDT May 2 the plume stays trapped against the coast. But when the winds weaken the bifurcation of flow at Sandy Hook occurs on the flood tide at 20060502.1330 UT and 20060503.0130 UT. Decreasing winds late Tuesday show a spreading of the surface signature but plots of the depth integrated freshwater anomaly still show the majority of the freshwater is strongly coastally trapped. By early May 3 the Asbury Park eddy is moving offshore.

2006 Apr 30 1645 EDT Hudson River discharge has fallen to 1000 m³s^-1. Steady northeasterly winds through Monday keep the plume against the NJ coast. During the flood tide late Monday evening (EDT) the bifurcation pattern at Sandy Hook seen in previous years is expected to occur. The anti-cyclonic recirculation feature near Asbury Park persists, and seemingly grows during Monday. The ocean model therefore indicates that any dye injection during Monday should be placed south of Sandy Hook if it is to enter the coastal current and not re-enter the harbor. However, the coastal current may terminate abruptly in the recirculation a short distance down coast.

2006 Apr 28 1500 EDT Northerly winds on Saturday will arrest the outflow of the Hudson and store freshwater in Raritan Bay. A wind relaxation on Saturday afternoon will allow some of this freshwater to escape Raritan bay and it will subsequently be pushed down coast by strengthening northerlies on Sunday.

2006 Apr 27 1530 EDT Alternating weak and northerly winds during the next 2 days will keep the plume against the NJ coast with relatively little dispersal toward the east. Temperature at 2 m views of the entire NJ coast (same lon/lat view and color scale as the Latte satellite SST) are now being plotted (see links at left). The pattern is consistent with the coastal trapping of the plume that shows chlorophyll and RGB visible imagery. The terminus of the plume appears to be associated with a circulating feature near Asbury Park in the forecast.

2006 Apr 25: 1745 EDT Strong Hudson River transport and northerly winds will form a significant NJ coastal current with equivalent freshwater anomalies of up to 1.5 m extending to the C mooring line through early Apr 26. A rotation of the winds to southerlies around 1600 UT Apr 26, and then westerly, will then disperse the plume eastward. This could mark the onset of a freshwater bulge forming around 1030 UT Apr 27 with an accompanying bifurcation of the coastal flow at Sandy Hook (with strong northward current on the flood tide).

2006 Apr 24: 2000 EDT On Sunday the Hudson River discharge increased dramatically to it highest value (over 2000 m³s^-1) this spring, but the increased freshwater discharge is not forecast to impact the shelf waters before Tuesday night. Strengthening northwesterly winds coinciding with the ebb tide at 0130 UT on 26-Apr (or 2130 EDT 25-Apr) should favor the formation of a low salinity coastal current along Sandy Hook.

2006 Apr 23: 1600 EDT Strong easterly winds on Sunday April 23 drive westward flow along the south coast of Long Island. This tends to trap the very weak Hudson River freshwater input (little more than 300 m³s^-1) in Raritan Bay and NY Harbor. The westward Long Island flow turns to the south to form a New Jersey coastal current but the coastal current carries relatively little discharge from NY Harbor. A wind shift to SSE late on Sunday will allow some of the trapped freshwater to exit the harbor and join the coastal current, sending a freshwater pulse down coast. But this pulse ends on the next flood tide. Weak winds and a shift to SSW on Monday and Tuesday will restore the harbor outflow but move the coastal current offshore to join the highway route immediately west of the Hudson Shelf Valley.

Forecast model configuration summary:

The Regional Ocean Modeling System (ROMS) LaTTE model covers the Mid-Atlantic Bight from the center of Long Island southward to south of the mouth of the Delaware River.

10-meter winds, surface air pressure, 2-m temperature and 2-m relative humidity, net shortwave and downward longwave radiation from the NCEP-NAM forecast system (and soon Louis Bowers’ RU WRF local nested forecast) are used. These inputs and model SST determine the air-sea heat and momentum fluxes.

For dates prior to current, the forcing data used are the 24-hour forecast from the cycle that commenced on the immediately preceding day at 0000UT.

In forecast mode, we automatically acquire 60-hour forecast fields from the NCEP and RU-WRF Opendap servers and commence a new ROMS run shortly after 0000UT.

Each day the 48-hour forecast is overwritten with the newer 24-hour forecast, etc., to provide a rolling best-estimate model-based analysis that is plotted at the links above.

Further details on the ROMS LaTTE configuration

The model horizontal resolution is approximately 1 km with 30 vertical levels. The domain is limited to the continental shelf. The model bathymetry was taken from the 15 arc second (~460 m) data of the NGDC Coastal Relief Model. The maximum depth is 100 m at the end of the Hudson Shelf Valley. Most of the domain has depths shallower than 60 m.

Initial temperature-salinity conditions were estimated via a simple distance-from-shore weighting applied to springtime hydrographic observations in the region. The hydro data used were a combination of stations in the NOAA/NMFS archive provided by Maureen Taylor of NMFS Woods Hole, data in the NODC archive, and other observations by the Rutgers University Long-term Ecosystem Observatory (LEO) programs.

Open boundary conditions are Orlanski-type radiation augmented with tidal harmonic forcing taken from an ADCIRC model of the western Atlantic (Luettich et al. 1992). The weak mean southward flow of the mid-Atlantic Bight inner shelf is not imposed in the boundary conditions.

Air-sea heat and momentum fluxes are calculated by the bulk formulae of Fairall et al. (1996,2002) using the model sea surface temperature and sea level air temperature, pressure, relative humidity, and 10-meter winds.

The inflow of the Hudson River is specified using a simple estimate computed as 2 times the sum of the daily mean observations at the Mohawk and Fort Edward USGS gauges. The Delaware River is included with a monthly climatological flow rate and the model solution for the Delaware should not be trusted.