Along the east coast of North America from Nova Scotia to Cape Hatteras,
relatively fresh and cool shelf water is separated from warmer, saltier
open ocean water by a region of strong gradients, the shelfbreak front
(Figure 2). 
While the water properties on either
side of the front change seasonally, the shelfbreak front remains a
persistent feature in the lower half of the water column throughout the
year.
Traditionally the shelfbreak front along the East Coast of the United States was viewed as a density front separating shelf water from open ocean water. However, observations indicated that, while there is a sharp contrast in temperature and salinity across the front, at times these two tend to compensate so there is relatively little density contrast (Figure 2, summer). Motivated by this observation, Chapman [1986] used a simple model to show that a tracer front could be generated at the shelfbreak in the absence of a density difference between the shelf and the open ocean water. For a vertically uniform flow with no density variation, both convergence at the shelfbreak of offshore flow associated with bottom friction and diffusion of tracer offshore of the shelfbreak, due to the increasing water depth, resulted in a tracer front at the shelfbreak. While this simple model was provocative, the assumption of no vertical variations was clearly an oversimplification. Gawarkiewicz and Chapman [1991, 1992] continued investigating this mechanism using a three-dimensional numerical model. For the case of unstratified flow, Gawarkiewicz and Chapman [1991] found a shelfbreak tracer front was formed, consistent with the analytic results of Chapman [1986]. The numerical model results revealed that the shelfbreak front in this case did not act as a barrier to cross-shelf exchange. There was a continuous offshore flux of water from the shelf to the slope in the bottom boundary layer, which was inconsistent with observations. Gawarkiewicz and Chapman [1992] subsequently added stratification and found that a shelfbreak front still formed, but the mechanism was different. Offshore transport in the bottom boundary layer pushed lighter water under heavier water, eventually resulting in a well-mixed shelf. A convergence in the bottom boundary layer transport resulted in large cross-shelf density gradients, i.e. a front, near the bottom at the shelfbreak. Particles (drifters) placed in the bottom boundary layer in the numerical model moved offshore and then detached from the bottom at the shelfbreak front, moving upward along the front and then alongshore. The particles did not cross the front, in contrast to the previous results for unstratified flow.
While the numerical model results of Gawarkiewicz and Chapman [1992] suggested that the shelfbreak front is a barrier to exchange between the shelf and open ocean in the absence of other processes, there is considerable evidence that exchange of water across the shelfbreak front does occur. The difference in water properties on either side of the front (Figure 2) make it relatively easy to identify offshore water that has penetrated onto the shelf [e.g., Churchill et al., 1993]. Penetration of offshore water onto the shelf apparently is a common occurrence all along the East Coast of the U.S. [ Churchill and Cornillon, 1991a&b] and has been associated with a variety of processes such as warm-core rings colliding with the slope [ Joyce, 1991] and frontal eddies [ Lee et al., 1991; Churchill et al., 1993; Glenn and Ebbesmeyer, 1994a&b]. An important, unresolved question is how effective these processes are in producing a net exchange between the shelf and slope waters. Is there substantial mixing associated with these processes or does the slope water move onshore and back offshore without interacting with the shelf water?
Recent observations have suggested that intrusions of offshore water onto the shelf can represent a substantial volume of water over the shelf [e.g., Gawarkiewicz et al., 1992]. Flagg et al. [1994] used observations from the second phase of the Shelf Edge Exchange Processes program (SEEP-II) to study intrusions of salty slope water onto the shelf within the seasonal thermocline. They found that at a fixed location such intrusions had short time scales, a day or less, often with large onshore velocities (10--20 cm/s), suggesting onshore displacements of 10--20 km over a day. The intrusions were correlated with the occurrence of upwelling favorable wind stress events, however, the dynamics of the process remain unclear. The onshore velocities were larger than expected from a simple wind-driven response. Rough estimates of the stability of the observed flow based on the vertical shear and stratification suggested mixing with the surrounding shelf water occurred during strong onshore flows.
Houghton et al. [1994] estimated the contribution of eddy heat fluxes to the heat exchange between the shelf and slope water using the SEEP-II data and observed a seasonal variation in the character of the eddy heat fluxes. During winter and spring, the eddy heat flux was largest in the lower water column and was dominated by a few events. During summer, the eddy heat flux was largest just below the thermocline and there was the suggestion of a contribution from frontal instabilities. Houghton et al. [1994] noted that the estimated eddy heat flux was large enough to balance the offshore flux of heat due to advection in the bottom boundary layer and hence maintain a steady state.
Satellite observations have shown meandering and eddy formation along the shelfbreak front suggestive of instabilities [e.g., Figure 1 from Garvine et al., 1988]. The pinching off of eddies as a result of frontal instabilities is a potentially important mechanism for exchange of water across the front. Lee et al. [1991] argued that frontal eddies in the South Atlantic Bight act as a ``nutrient pump'' for the shelf, bringing nutrient rich water up onto the shelf where it is available to phytoplankton. Gawarkiewicz [1991] studied the linear stability of various configurations of the shelfbreak front and found that for both winter and summer conditions the instabilities had characteristics which were generally consistent with the limited observations available [e.g., Garvine et al., 1988], i.e., wavelengths of about 25 km and time scales for growth of 5--10 days. Chao [1990] examined the more general problem of frontal instabilities over a continental margin by considering several different geometries using a numerical model. Chao found that coastal upwelling and Gulf Stream fronts, where the front slopes in the same direction as the bathymetry, tended to be more unstable than fronts associated with density driven coastal currents, where the front slopes in the opposite direction from the bottom bathymetry.