Forecasting the dispersal of the Hudson River Plume John Wilkin Gregg Foti Byoung-Ju Choi Institute of Marine and Coastal Sciences Rutgers, The State University of New Jersey 2005 ROMS Workshop SIO, San Diego October, 2005

Coastal ocean modeling: Analysis and prediction Coastal ocean is a filter –nutrients and sediments from rivers and atmosphere –majority are re-mineralized with little export to the deep ocean U.S. East Coast: –discharge of many rivers is modified in estuaries Chesapeake Bay: –assimilation by plankton and export as inorganic nitrogen New York: –90% nitrogen is exported unassimilated onto the shelf –NY/NJ Harbor sediments are ~4% of MAB load, but carry 90% of PCBs and 70% of mercury Where it goes and how it is transformed depends not biology, geochemistry and physics

Visible image from Aqua 6-Apr :30 UT

Bob Chant, John Wilkin, John Reinfelder, Scott Glenn, Oscar Schofield 1, Bob Houghton 2, Bob Chen 3, Meng Zhou 3, Mark Moline 4, Paul Bissett 5,Tom Frazer 6. Lagrangian Transport and Transformation Experiment 1 Rutgers, 2 Lamont-Doherty, 3 U. Mass Boston, 4 FERI, 5 Calpoly, 6 U. Fla. Lagrangian studies of the transport, transformation and biological impact of nutrients and contaminant metals in a buoyant plume: A process study in an operational ocean observatory.

LaTTE: An interdisciplinary process study in an operational ocean observatory

R/V Cape HatterasR/V Connecticut

ROMS modeling of the Hudson River plume: develop intuition – a Latte playbook (Choi) forecasts for adaptive sampling during field operations (Foti) the shear geeky pleasure of the adjoint model (Zhang, Levin) post-experiment analysis (all of us)

The Hudson River plume and adjacent shelf circulation in the New York Bight were forecast in support of the 2005 LaTTE field program using the Regional Ocean Modeling System (ROMS) The Hudson River plume and adjacent shelf circulation in the New York Bight were forecast in support of the 2005 LaTTE field program using the Regional Ocean Modeling System (ROMS) Operational model configuration: Domain: central Long Island to mouth of Delaware Bay, out to approximately the 100-m isobath 1 km horizontal resolution; 30 vertical s-levels Forcing: NCEP NAM-ETA hourly forecast u 10 winds, T air, q air, cloud, radiation from OPeNDAP server GNU wget Daily mean Hudson River flow from USGS Mohawk and Fort Edward gauges + persistence (for forecast) accessed via web using GNU wget Boundary tides from ADCIRC, tuned by assimilation of local tidal harmonic data No boundary mean flow or pressure gradient at open boundaries Output: Model runs daily to generate 48- to 60-hour forecast Output graphics for 2005 at Output data on OPeNDAP server J. Wilkin

2004 Dye Injections Injection 1  May 2 nd, 2004  14:54 GMT – 15:03 GMT  40° 31.48' 73° 46.6' Injection 2  May 4 th, 2004  13:42 GMT – 14:00 GMT  40° 21.9' 73° 54.8' LaTTE Pilot Experiment 2004

29-Apr :24 – 30-Apr :28 01-May :52 – 02-May :31 RU02 – Hudson River Plume SouthNorth WestEast LaTTE Pilot Experiment 2004

3 May 2004

Freshwater plume that carried first dye injection east and onto the Long Island shore Second dye injection into coastal current Re-circulation feature related to freshwater anomaly and bottom flow steered by bathymetry Wind change reversed coastal current and redirected Hudson outflow to east

LaTTE Experiment 2005

OCM visible RGB and CODAR 09-Apr :90 UT Observations Model Observations Model ROMS salinity and velocity at 2m 09-Apr-2005 OCM visible RGB and CODAR 18-Apr :38 UT ROMS salinity and velocity at 2m 18-Apr-2005 J. Wilkin

1 day average 19-Apr-2005

LaTTE Experiment 2005

Real-time forecast dye trajectory: second injection LaTTE Experiment 2005 G. Foti and J. Wilkin

The horizontal dispersal of the freshwater anomaly due to the plume can be visualized in terms of the equivalent freshwater depth: J. Wilkin

The modeled plume trajectory is dependent on local winds. Modeled dynamical term balances show that direct downstream forcing by the buoyancy pressure gradient force is small compared to surface wind stress. No wind simulation – freshwater ‘ bulge ’ continues to grow: B.-J. Choi and J. Wilkin Fong and Geyer (2002)

Plume response under idealized wind scenarios For high river discharge typical of the spring freshet, a set of idealized wind forcing simulations shows how the plume responds to wind direction. Southerly directs plume north of Hudson Shelf Valley and toward Long Island Northerly amplifies coastal current and drains freshwater bulge B.-J. Choi and J. Wilkin

Westerly wind directs plume toward Hudson Shelf Valley Easterly wind arrests plume For high river discharge typical of the spring freshet, a set of idealized wind forcing simulations shows how the plume responds to wind direction. B.-J. Choi and J. Wilkin Plume response under idealized wind scenarios

April 2-18 average

The sensitivity of sea surface temperature (arbitrary units) in the marked triangular box to SST over the preceding half day for an idealized simulation of low Hudson River flow in the inner New Jersey/New York Bight region, computed using the ROMS Adjoint model. G. Zhang, J. Levin and J. Wilkin

ROMS prediction system –with NCEP/NAM atmospheric forecast –observed river flows Reasonable simulations of the dispersal of the spring freshet during the LaTTE April 2005 observing period –no assimilation –OK for a locally forced regime Flow is responsive to local winds on short time scales Relatively little low salinity water immediately traveled southward as a coastally trapped buoyancy driven current The coastal current only forms with the assistance of favorable winds The Hudson Shelf Valley exerts control on the overlying circulation Much of the freshwater is dispersed eastward along the L.I. coast Some heads offshore along the south flank of the Hudson Shelf Valley Summary