Dispersal of the Hudson River Plume in the New York Bight: Synthesis of Observational and Numerical Studies During LaTTE

characterized the variability of the Hudson River discharge and identified several freshwater transport pathways that lead to cross-shelf mixing of the Hudson plume. The plume's variability is comprised of several different outflow configurations that are related to wind forcing, river discharge, and shelf circulation. The modes are characterized by coastal current formation and unsteady bulge recirculation. Coastal currents are favored during low-discharge conditions and downwelling winds, and represent a rapid downshelf transport pathway. Bulge formation is favored during high-discharge conditions and upwelling winds. The bulge is characterized by clockwise rotating fluid and results in freshwater transport that is to the left of the outflow and opposed to classical coastal current theory. Upwelling winds augment this eastward flow and rapidly drive the freshwater along the Long Island coast. Upwelling winds also favor a midshelf transport pathway that advects fluid from the bulge region rapidly across the shelf on the inshore side of the Hudson Shelf Valley. A clockwise bulgelike recirculation also occurs along the New Jersey coast, to the south of the river mouth, and is characterized by an offshore veering of the coastal current. Modeling results indicate that the coastal transport pathways dominate during the winter months while the midshelf transport pathway dominates during summer months. Finally, because the time scales of biogeochemical transformations in the plume range from hours to weeks or longer, the details of both the near-and far-field plume dynamics play a central role in the fate of material transported from terrestrial to marine ecosystems.

aBSTR aCT. Observations and modeling during the Lagrangian Transport and Transformation Experiment (LaTTE) characterized the variability of the Hudson River discharge and identified several freshwater transport pathways that lead to cross-shelf mixing of the Hudson plume. The plume's variability is comprised of several different outflow configurations that are related to wind forcing, river discharge, and shelf circulation. The modes are characterized by coastal current formation and unsteady bulge recirculation. Coastal currents are favored during low-discharge conditions and downwelling winds, and represent a rapid downshelf transport pathway. Bulge formation is favored during high-discharge conditions and upwelling winds. The bulge is characterized by clockwise rotating fluid and results in freshwater transport that is to the left of the outflow and opposed to classical coastal current theory. Upwelling winds augment this eastward flow and rapidly drive the freshwater along the Long Island coast. Upwelling winds also favor a midshelf transport pathway that advects fluid from the bulge region rapidly across the shelf on the inshore side of the Hudson Shelf Valley. A clockwise bulgelike recirculation also occurs along the New Jersey coast, to the south of the river mouth, and is characterized by an offshore veering of the coastal current. Modeling results indicate that the coastal transport pathways dominate during the winter months while the midshelf transport pathway dominates during summer months. Finally, because the time scales of biogeochemical transformations in the plume range from hours to weeks or longer, the details of both the near-and far-field plume dynamics play a central role in the fate of material transported from terrestrial to marine ecosystems.
iNTRODuCTiON River discharge into the coastal ocean represents a major link between terrestrial and marine systems. Moreover, with over half of the world's human population located within coastal watersheds, this discharge is an important pathway that extends anthropogenic impacts into the ocean. Although biogeochemical processes can significantly modify this transport pathway, understanding the processes that determine freshwater transport pathways is essential in determining the fate and transport of material fluxing across the land-sea interface.
River outflows are less dense than the saline ocean waters and this density difference produces a buoyancy force that drives the plume's circulation. The classic model of plume dynamics balances this buoyancy force with the Coriolis force that causes the outflow to turn to the right (in the northern hemisphere) and form a narrow coastal current trapped within a few internal Rossby radii (the ratio of internal wave speed to the local Coriolis frequency) of the coast (Garvine, 1987(Garvine, , 1999. The coastal current may be confined to a thin surface layer or may be attached to the bottom (Yankovsky and Chapman, 1997). In general, the classic model emphasizes that, in the absence of winds, coastal current dynamics severely limit the crossshelf transport of river plumes.
Buoyant outflows may also contain a bulge-like region in the vicinity of the outflow, and the cross-shelf extent of these bulges can be many times the width of the downstream coastal current. Yankovsky and Chapman (1997) incorporated a bulge in a steady-state model that they closed by equating the freshwater flux in the coastal current to the freshwater flux exiting the estuary. With this steady-state assumption, they developed an elegant theory that related coastal current structure to the estuarine discharge and the cross-shore slope of the seafloor.
Recent modeling and laboratory studies of buoyant outflows provide a more detailed characterization of bulge structure (Fong and Geyer, 2002;Avicola and Huq, 2003;Horner-Devine et al., 2006) and emphasize that a bulge may be unsteady and grow in time.
Consequently, the freshwater flux out of the estuary may be greater than the freshwater flux in the coastal current, with the remainder going into bulge formation. Based on laboratory experiments in a rotating tank, Avicola and Huq (2003) reported that approximately one-third of the outflow became incorporated in a coastal current. More detailed analysis afforded by numerical modeling indicated that the fraction of freshwater from the river that is incorporated in the coastal current depends on the outflow parameters. Specifically, as the flow becomes increasingly nonlinear, less of the discharge goes into the coastal current and more into bulge growth.
The nonlinearity is characterized by the Rossby number, which is the ratio of inertial to rotational forces. Fong and Geyer (2002) discuss the mechanisms by which the bulge feeds the coastal current by invoking a model by Nof (1988) Fong and Geyer (2002) note that bulge formation in models is more pronounced than in nature.
Wind forcing also plays a critical role in the cross-shelf transport of river plumes (Whitney and Garvine, 2005).
Modeling studies reveal that upwelling winds are effective in both transporting river plumes offshore and mixing the plume with the coastal ocean (Fong and Geyer, 2001). Observational studies of coastal currents reveal that the structure of the flow and salt fields (Rennie and Lentz, 1999) and of the diapycnal fluxes (Houghton et al., 2004) appear to be consistent with numerical studies (Fong and Geyer, 2001). Despite this consistency, there has been little research on the effect of wind forcing on bulge dynamics, with the notable exception of Choi and Wilkin (2007).
In this paper, we discuss the character of the Hudson River's discharge into the coastal ocean based on observations and modeling efforts during the Lagrangian Finally, LaTTE included physical and biogeochemical modeling (Choi and Wilkin, 2007;Cahill et al., 2008;Zhang et al., in review-a, b), which was essential in providing a coherent framework to characterize annual variability in the plume's structure and transport pathways. Modeling efforts used the Regional Ocean Modeling System (ROMS; http:// www.myroms.org) that was forced by tides, winds, and remotely forced flows at the offshore boundaries as specified by Lentz (2008) The bulge's structure was also modified by shelf circulation as suggested by Fong and Geyer (2002). In particular, during upwelling-favorable winds, a jet develops that transfers freshwater from the bulge toward the shelf break along the inshore side of the HSV. This rapid cross-shelf advection of the plume was documented following a second "10-year flood event" during summer 2006 (Castelao et al., 2008a). Glider data revealed that following the flood event, freshwater was transported over 100 km from the coast in fewer than two weeks. Chlorophyll-a imagery suggests that the jet entrained biomass from the bulge region and advected it cross shelf along the inshore side of the jet (Figure 5b).
Late spring/early summer shelfwide freshening was also observed in glider data from previous years (Castelao et al., 2008a;Chant et al., 2008); this seasonal shelfwide freshening is driven by the seasonal transition from downwelling-to upwelling-favorable winds (Castelao et al., 2008b)   Color is chlorophyll a on a relative scale, with red representing high values; the bulge is characterized by high chlorophyll a values. The old plume over the hudson Shelf Valley is devoid of plankton but less than 29. Note that the single isohaline on the offshore side of the last four sections is the 29 isohaline.
Consequently, the midshelf freshwater pathway likely represents a robust mechanism that rapidly transports the spring freshet water across the continental shelf to the shelfbreak. We note that climate models are sensitive to the details of how freshwater mixes into the deep ocean (Garvine and Whitney, 2006) it into the coastal ocean as suggested by the modeling studies of Fong and Geyer (2001) and Choi and Wilkin (2007). A second mode of outflow was characterized by bulge formation, which occurred during moderate to high discharge and weak or upwelling-favorable wind forcing. With upwelling winds, the bulge became compressed along the Long Island coast and extended eastward.
Once formed, the bulge's structure was strongly modified by wind forcing and shelf circulation with a particularly rapid cross-shelf transport pathway associated with a midshelf jet. Finally, a third mode was observed that consisted of a coastal current with a downstream recirculating region. Next, we present numerical simulations of the Hudson outflow to provide a more detailed characterization of the spatial and temporal structure of the plume, of the processes that control this structure, and ultimately the freshwater transport pathways.

NumERiCaL mODELiNg Of ThE PLumE
The variability we observed in the plume's structure raises the following questions: To what extent is the observed variability representative of its typical behavior? Alternatively expressed: What is the relative importance of each "mode" of outflow and what is the seasonal variability of these modes and of these transport pathways? To address these questions in greater detail, we ran numerical simulations, first in a process study (Choi and Wilkin, 2007) and later with realistic forcings (Zhang et al., in review-a, b). The process study  (Zhang, et al., in review-a, b (Yankovsky et al., 2004) or associated with lateral shears that develop across the HSV (Harris, et al., 2003).
Together, these transport pathways   Figure 8d is characterized by broad features as is the cross-shore structure of the annual mean salinity based on glider data (Castelao et al., 2008b). However, the freshwater pathway that produced the broadly distributed freshwater transport pathway was not solely the result of upwelling winds acting on a coastal current but also was significantly influenced by bulge formation and rapid cross-shelf advection associated with a cross-shelf jet along the 40-50-m isobath (Castelao et al., 2008a). Although the dynamics that underlie this cross-shelf jet remain elusive, it appears to be initiated by persistent upwelling winds (Castelao et al., 2008a). Several other studies have noted frontal systems in this region (Bumpus, 1973;Biscaye et al., 1994;Ullman and Cornillon, 1999), and analysis of longterm hydrographic data from the Mid-Atlantic Bight also revealed a shelfwide freshening that was localized in the New York Bight region (Mountain, 2003).
Finally, the tendency for the Hudson's outflow to recirculate near the apex rather than rapidly advect away in a coastal current has significant implications for biogeochemical pathways. For example, nutrient uptake and primary production was so rapid in this region