1. Introduction
The seasonally varying wind stress is the primary forcing mechanism for the circulation along much of the California Current System (CCS). Prevailing equatorward, upwelling-favorable winds during spring and summer off the Oregon coast, for example, lead to the formation of a strong along-shelf coastal jet that is in geostrophic balance with the upwelled isopycnals. In regions of along-shelf uniform topography (both bottom bathymetry and coastline orientation), the upwelling circulation is well described by standard two-dimensional models (e.g., Allen 1980).
The presence of along-shelf variations in topography substantially influences the coastal circulation. Studies over the last decade or so have established that the upwelling jet often separates from the coast near Cape Blanco, Oregon, to become an oceanic jet to the south (Barth and Smith 1998; Barth et al. 2000, 2005b). Off Newport (upstream of the cape), the summer-averaged upwelling jet lies about 20–30 km offshore, over the shelf. South of the cape, however, the jet is located far from shore (∼120 km from the coast), with a weak secondary jet near the shelf break (Huyer et al. 2005). Jet separation at Cape Blanco is further confirmed by satellite observations (Strub and James 1995, 2000; Barth et al. 2000; Castelao et al. 2006; among others), which frequently show a broader area influenced by the upwelling circulation to the south of the cape when compared with regions to the north.
The seasonal evolution of the flow in the Cape Blanco region was depicted during the Global Ocean Ecosystems Dynamics (GLOBEC) Northeast Pacific Program, which provided some of the few mesoscale-resolving, in situ, large-area surveys of the circulation and hydrographic structure in this area. Those observations revealed an upwelling front and jet that was located close to the coast, south of the cape early in the upwelling season (late May–early June 2000), but displaced about 80 km offshore 2 months later (late July–early August 2000; Barth et al. 2005b). The seasonal evolution of the upwelling front and jet from satellite observations is also consistent with that observation. Using 6 yr of altimeter and tide gauge data, Strub and James (2000) found a concentrated equatorward flow close to the coast during May and June. The jet remains closer to the coast off Oregon and Washington, but meanders far offshore (∼100 km) off the California coast during July and August.
The offshore displacement of the upwelling jet provides an important mechanism for exporting material from the highly productive continental shelf toward the deep ocean (e.g., Barth et al. 2002), and thus has important consequences for both the coastal and adjacent deep-ocean ecosystems. In situ observations show that the domain to the south of the cape has a more saline, cooler, denser, and thicker surface mixed layer, a wider coastal zone inshore of the upwelling front and jet, higher nutrient concentrations in the photic zone, and higher phytoplankton biomass when compared with the region north of the cape (Huyer et al. 2005). Huyer et al. attributed these differences to stronger mean northerly wind stress in the south, strong small-scale wind stress curl in the lee of Cape Blanco, and the reduced influence of the Columbia River discharge in the coastal region south of the cape. The different characteristics in the south are consistent with more vigorous winds and enhanced upwelling there. Enhanced upwelling in the lee of topographic perturbations can be attributed to several factors, including intensification in the wind stress resulting from orographic effects [e.g., airflow accelerating within an expansion fan; Dorman et al. (2000)], variations in bottom bathymetry (Peffley and O’Brien 1976; Castelao and Barth 2006) and coastline orientation (Gan and Allen 2002), and variations in vorticity resulting from curvature of the trajectory as the flow passes the cape (Arthur 1965).
The dynamics of the separation process at Cape Blanco are still not understood. Numerical (e.g., Haidvogel et al. 1991; Batteen 1997) and laboratory (e.g., Narimousa and Maxworthy 1987) modeling studies have shown that irregularities in the coastline geometry are important in the formation of meanders and filaments. Previous studies have shown that bottom topography and coastline curvature are important in the separation process (Barth and Smith 1998; Barth et al. 2000). Using an analytic 1.5-layer model of coastal hydraulics with constant potential vorticity in each layer, Dale and Barth (2001) found time-dependent solutions predicting the flow field at critical transitions, in the sense of hydraulic control, consisting of a narrow upwelling jet upstream of the cape that moves offshore and broadens at the cape.
More recently, Samelson et al. (2002) suggested that the wind intensification south of the cape could also influence the local circulation. Because the stress magnitude south of Cape Blanco is 3–4 times that along the northern Oregon coast (see also Perlin et al. 2004; Chelton et al. 2007), Samelson et al. conclude that coastal upwelling must be similarly intensified toward the south, consistent with the observed offshore displacement of the jet and the observed presence of systematically colder ocean temperatures in that region (Barth et al. 2000; Huyer et al. 2005). Samelson et al. also point out that the wind stress curl evident in both atmospheric models and scatterometer data could also be dynamically significant, by causing local Ekman transport divergences and convergences. Using the U.S. Navy’s high-resolution atmospheric model [Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS)], Pickett and Paduan (2003) showed that the vertical transport resulting from Ekman pumping from wind stress curl is as important as the vertical transport resulting from coastal divergence in the Ekman transport in causing upwelling in the central CCS. The main goal of the present study is, therefore, to explore the importance of spatial variability in the wind forcing in the separation of a coastal upwelling jet at an idealized cape. We are particularly interested in identifying the specific roles of the wind intensification by itself, which enhances vertical transport very near the coast, and the associated wind stress curl, which drives weaker upwelling velocities that are spread over a much larger offshore extent.
Although this study is motivated by jet separation at Cape Blanco, results should be relevant for other regions where jet separation occurs at a cape. Off Chile, for example, the upwelling jet frequently separates from the coast at Punta Lavapie (e.g., Mesias et al. 2001, 2003). Off the Iberian Peninsula, jet separation and filament formation occurs associated with Cape Finisterre and Cabo Roca (e.g., Haynes et al. 1993), while in the southern Benguela upwelling system, jet separation frequently occurs at Cape Columbine (e.g., Penven et al. 2000). The wind stress curl downstream of these major capes is intensified (positive in the Northern Hemisphere, negative in the Southern Hemisphere), favoring offshore upwelling (see Chelton et al. 2004), which is a situation analogous to that of Cape Blanco.
2. Methods
The Regional Ocean Modeling System (ROMS) used here is a hydrostatic primitive-equation numerical circulation model with terrain-following vertical coordinates (Shchepetkin and McWilliams 2005), based on the S-Coordinate Rutgers University Model (SCRUM) described by Song and Haidvogel (1994). The model incorporates the Mellor and Yamada (1982) 2.5-level turbulence closure scheme as modified in Galperin et al. (1988). The pressure gradient scheme is a spline density Jacobian (Shchepetkin and McWilliams 2003), which minimizes the errors associated with computing horizontal pressure gradients with terrain-following coordinates.
The model domain extends 300 km offshore and 600 km in the along-shelf direction (Fig. 1). The grid resolution is 2 km in the along-shelf direction, and between 1.3 and 4.4 km in the cross-shelf direction, with the highest resolution near the coast. In the vertical, 30 s levels, where s denotes a generalized vertical coordinate, are utilized with grid spacing that varies so that there is higher resolution both near the surface and the bottom in order to resolve the respective boundary layers. The horizontal velocity v has components (u, υ) corresponding to the cross- and along-shelf velocities in the (x, y) directions, so that u is positive onshore and υ is positive toward the north. The depth average of the velocity components are denoted by V and (U, V). We use the notation η for the surface elevation above the undisturbed free surface, T for temperature, S for salinity, t for time, and D for the water depth.
The model has three open boundaries. In the north, we use a zero-gradient condition for η, and the condition proposed by Gan and Allen (2005a) for the remaining variables. The local solution is obtained from calculations utilizing a local two-dimensional (x, s, t) submodel at the boundary. At the southern boundary, η satisfies an implicit gravity wave radiation condition (Chapman 1985), while the depth-averaged velocities satisfy a Flather radiation scheme (Flather 1976). The specified values needed in the Flather condition are obtained from a local two-dimensional solution. For the depth-dependent variables, an oblique radiation condition (Marchesiello et al. 2001) is used. At the offshore boundary, ηx = Vx = Tx = Sx = 0, and we use a radiation condition for U, u, and υ. Shelf wave tests, like those performed by Gan and Allen (2005a), produced results nearly identical to theirs, giving us confidence in the set of the open boundary conditions used. Tests, using a domain doubled in size, produce results very similar to those obtained with the basic domain, including the time and spatial scales of jet separation at the cape, further increasing our confidence in the boundary conditions used. No surface heat flux was used in the simulations presented here. Tests with a surface heat flux showed no significant difference regarding jet separation.
The axis of the wind intensification used in the simulations has an east–west orientation (e.g., Fig. 2). In reality, the axis of the wind intensification off Cape Blanco is tilted by approximately 45° toward the southwest, away from the cape (Perlin et al. 2004). Simulations forced by winds similar to the wind fields used here, but with the axis of the wind intensification tilted by 45°, produce results consistent with those described in this study. We choose not to use the rotated winds, however, because the interpretation of the results becomes more difficult, because more parameters vary among the different cases shown in Table 1. For example, if the wind intensification axis is rotated by 45°, moving the location of the intensification offshore (as in the “curl far” experiment, Table 1) not only increases the width of the region of positive wind stress curl close to the coast, but also changes the north–south extent of the region with positive curl.
To isolate the effects of the wind fields from topographic effects, simulations were also run using a two-dimensional model (x, s, t), with the same bottom slope as that of the regular domain.
Coastal jet separation is quantified by measuring the offshore displacement experienced by a line of constant near-surface along-shelf transport (∫jetcoast ∫0−5m υ dz dx, where z is the vertical coordinate and “jet” is the jet core) in the vicinity of the cape. We first choose a transport isoline that is located along the core of the jet upstream of the cape. The offshore jet displacement Sx is determined by differencing the maximum offshore distance reached by the isoline south of the cape and the offshore distance of the isobath over which the isoline lies upstream of the cape (Fig. 1). Note that similar results are obtained if an isotherm, instead of a line of constant near-surface along-shelf transport, is used in the calculation.
For some simulations, we found it useful to analyze thermal balances. Details of the computation are given in the appendix.
3. Two-dimensional results
To more clearly isolate the effects of the wind intensification and associated wind stress curl from topographic effects, two-dimensional (x, s, t) simulations were pursued. Three different wind scenarios are considered. In the first run, the wind stress is considered constant in the cross-shelf direction and is equal to −0.04 Pa (upwelling favorable), except very near the coast, where a slight decrease in the southward wind magnitude occurs. This is close to the Quick Scatterometer (QuikSCAT) summer-averaged wind stress magnitude north of Cape Blanco (Perlin et al. 2004). Note that QuikSCAT observations are not available in a thin band (∼30 km) close to the coast. In the second experiment, the intensity of the along-shelf component of the wind stress is increased to −0.1 Pa, again with only a weak cross-shelf variation close to the coast (Fig. 3a). Because in both cases there is only a thin, weak band of positive curl close to the coast, these winds will be referred to as having no curl. The goal is to identify the effects of the wind intensification by itself. In the third experiment, the maximum in the southward wind stress (−0.14 Pa) is imposed at about 120 km from the coast. The wind intensity decreases toward the coast (to about −0.05 Pa) and in the offshore direction. Therefore, the region close to the coast is characterized by positive wind stress curl, while the region farther offshore has negative wind curl (Fig. 3b). This cross-shelf wind profile is similar to a profile through the center of the wind intensification in the observed summer-averaged wind stress field south of Cape Blanco (Perlin et al. 2004), with the exception that the maximum in the wind stress magnitude (and hence the curl pattern) is shifted slightly offshore. By comparing results from this simulation with the other two, the effect of the wind curl can be assessed. Only results after 80 days of simulation are shown (Fig. 3). Although 80 days of constant winds is not realistic for the region, long runs are needed to differentiate clearly between the simulations. In addition, the wind magnitudes used are similar to the observations averaged over the upwelling season, which is considerably longer than 80 days.
Results from the second experiment (−0.10 Pa, no curl; Fig. 3a) are very similar to those from the first experiment (−0.04 Pa, no curl; not shown here), with the obvious exception that upwelling is enhanced in the case forced by stronger winds. The upwelling front and jet are found farther from the coast (by about 15 km), and along-shelf velocities are stronger. In both cases, the circulation is consistent with two-dimensional models of coastal upwelling (e.g., Allen 1980). The southward wind stress drives an offshore flow in the surface Ekman layer, which is balanced by the upwelling of cold water from below. A vertically sheared along-shelf jet develops in geostrophic balance with the tilted isopycnals. Upwelling, in this case, is restricted to regions close to the coast. At distances greater than ∼60–65 km from the coast, outside the surface mixed layer, isotherms are nearly flat in both experiments 1 and 2, presenting no significant depth changes from the initial condition.
The offshore curl-driven upwelling in the third experiment creates a strong horizontal temperature (and density) gradient, which is enough to sustain a strong intensification of the along-shelf velocities offshore via the thermal wind balance. By computing the difference between the along-shelf velocities in the third (with curl; Fig. 3b) and second (no curl; Fig. 3a) experiments, we verify that the presence of the positive wind curl and weaker winds near the coast strongly intensifies the southward velocities in the offshore region (greater than 80 km from the coast), and strongly weakens the southward velocities in the region close to the coast (30–70 km offshore) (Fig. 3c). There is yet another region of intensified southward flow over the shelf at about 25 km from the coast. Because the winds close to the coast are weaker relative to the no-curl scenario, the cross-shore advection of the upwelling front close to the coast is reduced, and the inshore jet remains closer to the coast when compared with the upwelling jet in the no-curl simulation. Curl-driven upwelling near the shelf break may also contribute to the intensification.
Other simulations (not shown here) reveal that decreasing the offshore extent of the region with positive wind stress curl causes the offshore jet to occur closer to the coast, while increasing the width with positive curl leads to the formation of the offshore jet farther from the coast. In a consistent manner, increasing the magnitude of the positive curl (i.e., increasing the curl-driven upwelling) causes the offshore jet to be established earlier, while decreasing the magnitude of the positive curl leads to a delay in the formation of the offshore jet.
The effect of the wind stress curl is further illustrated by comparing terms in the temperature equation (see the appendix for details) between the second and third experiments (Fig. 4), averaged over the first 80 days of simulation. In the no-curl scenario, the temperature cooling (∂T/∂t < 0) is primarily balanced by cross-shelf advection (here, horizontal advection is synonymous of cross-shelf advection, because the simulation is two-dimensional). Positive vertical advection is only significant within a few kilometers from the coast. Vertical diffusion is most important in the surface and bottom boundary layers, and particularly within 20–30 km from the coast. Away from the coast, ∂T/∂t ∼0, except in the surface layer, where the cooling is due to Ekman transport advecting upwelled waters offshore.
In the wind stress curl scenario, the results are considerably different. Cross-shelf advection is reduced (see difference panel) as a consequence of reduced winds close to the coast. Vertical advection, on the other hand, is now important within 120 km from the coast, which is the same region where the wind stress curl is positive. A local intensification is also found over the shelf. This leads to a significantly larger area with ∂T/∂t < 0, consistent with Fig. 3b. Offshore of 120 km from the coast, the deep water actually warms over the period simulated as surface water is downwelled (negative vertical advection). As in the constant wind case, vertical diffusion of temperature is most important in the boundary layers.
4. Importance of geometry and the basic case
Before exploring the flow dependence on the spatial structure and intensity of the wind forcing, we investigate if, with the setup used, jet separation can be obtained in the absence of the cape. Previous studies have shown that the presence of a cape can play an important role in jet separation (Batteen 1997; Barth and Smith 1998; Barth et al. 2000; Dale and Barth 2001). We compare results from two simulations forced by the same wind stress (and hence the same wind stress curl) representative of the Oregon coastal ocean during summer (Fig. 2), differing only in the geometry of the topography. The basic case topography mimics the coastline geometry in the Cape Blanco region (Fig. 1); that is compared with a domain with a straight coast.
Surface velocity vectors overlain on the surface temperature field for the basic case at days 70, 85, and 110 are shown in Fig. 5. Vertical sections of temperature and alongshore velocity at days 70 and 110 are shown in Fig. 6. Results at day 70 are consistent with the general picture of coastal upwelling, with isotherms tilting upward toward the coast leaving a band of cold water at the surface close to shore, and the jet roughly following the topography. By day 85, the jet has significantly moved offshore south of the cape, advecting large amounts of cold, upwelled waters offshore. Around 80 km south of the cape, the increase in the local water depth causes the flow to turn cyclonically back onshore in order to conserve potential vorticity, leading to reattachment. By day 110, the temperature front and jet are further separated, reaching about 80–90 km from the coast south of the cape. Inshore of the separated jet, a strong recirculation region is established. The northward flow in this recirculation has weak vertical shear. Over the shelf, a secondary, weaker southward upwelling jet is formed, associated with the inshore temperature front. Although the present simulation is highly idealized, results bear close resemblance to in situ observations just south of Cape Blanco (Barth et al. 2000; Huyer et al. 2005). Temperature and alongshore velocity observations along 41.9°N (south of Cape Blanco) during the summer of 1995, for example, are very similar regarding scales, patterns, and direction of the flow to those at day 110, km 300 (see Fig. 6 of Barth et al. 2000).
Time series of Sx, a measure of separation (see section 2 for definition), for both the basic case and the straight-coast simulation are shown in Fig. 7. As seen in Figs. 5 and 6, the jet in the basic case is still attached until day 70, experiencing only a small (∼15 km) offshore deflection. From day 70 onward, the jet moves away from the coast, being displaced about 50 km offshore by the end of the simulation. Instabilities in the jet create smaller-scale oscillations around the mean position of the jet, which are clearly seen in the time series of Sx. In the straight-coast case, the jet stays attached for the entire duration of the simulation. This shows that the presence of a cape plays an important role in jet separation, consistent with previous results (Haidvogel et al. 1991; Batteen 1997; Barth and Smith 1998; Barth et al. 2000; Dale and Barth 2001). The wind intensification with a strong positive curl is not enough, by itself, to lead to jet separation.
5. Response to different wind forcing
Several simulations were run using the cape geometry, but varying the intensity and/or spatial structure of the wind forcing. The wind patterns represent variations about the basic case (Table 1; Fig. 2), which are chosen to highlight the relative importance, if any, of the wind intensification versus the wind stress curl in the separation of the upwelling jet at the cape. The along-shelf component of the wind stress for all cases considered is shown on the upper panels of Fig. 8. The cross-shelf component of the wind is set to zero. The basic case is repeated for reference. In the lower panels of Fig. 8, the associated wind stress curl (∂τys/∂x) field for each case is shown. As described in section 2, all cases considered here have a thin band of weak, positive curl close to the coast. Therefore, when a simulation is referred to as being forced by a spatially uniform wind, or by a wind field with no curl, what is really meant is that there is no intensification in the wind stress and/or wind stress curl south of the cape. Wind patterns only differ in the region of the intensification, being identical in the northern and southern parts of the domain. Time series of Sx are then compared between different wind scenarios (Figs. 9 –12). In all groups of comparisons, the time series of Sx from the basic case is shown as the solid black line.
Results from the basic case (Fig. 8a) are first compared with results from simulations forced by one-half of the wind stress curl (Fig. 8b) and one and one-half of the wind stress curl (Fig. 8c). The wind stress shown in Fig. 8b is characterized by a weaker intensification as compared with the basic case. As a consequence, the wind stress curl is also weaker. In Fig. 8c, on the other hand, winds are characterized by a stronger intensification with stronger curl. Note, however, that although the intensity is different, the spatial structure of the wind stress and wind stress curl fields is identical to that of the basic case. Time series of Sx (Fig. 9) reveal that decreasing the magnitude of the wind stress and wind stress curl leads to later jet separation when compared with the basic case, while increasing their magnitudes leads to earlier separation. This is consistent with the hypothesis that enhanced upwelling resulting from stronger winds and stronger curl would favor separation. Note that although the timing of separation is affected, the offshore deflection of the jet after about 95 days is similar in the three cases.
The previous comparison, although suggestive that the structure of the wind fields is important in controlling details of separation, does not shed light onto the specific roles of the wind stress and the wind stress curl intensification in the process, because both varied between the cases. In the second group of comparisons, results from the basic case are compared with results forced by wind with no curl (Fig. 8d) and by a spatially uniform wind (Fig. 8e). In Fig. 8d, the wind is characterized by an intensification south of the cape, but with no curl. The magnitude of the intensification is equal to the averaged winds within 15 km (∼1 internal Rossby radius of deformation) from the coast in the basic case. Thus, both cases produce roughly the same amount of Ekman transport and upwelling at the coast. The wind is spatially uniform in Fig. 8e, with no intensification or curl. Time series of Sx (Fig. 10) reveal that the wind intensification with no curl produces approximately the same amount of separation as that of spatially uniform winds. A similar result is obtained if the model is forced by winds with the same spatial structure as that of the wind with no curl (Fig. 8d), but with a stronger intensification (reaching −0.14 Pa). If left long enough, the interaction of the flow with the cape causes jet separation, independent of details of the wind structure. The wind intensification with the curl (basic case; Fig. 8a), on the other hand, causes the jet to separate considerably earlier (by about 20–25 days). These results suggest the wind stress intensification by itself does not aid in jet separation, and that the cause of the observed variability in the time of separation in Fig. 9 is the variability in the wind stress curl between the simulations.
To further confirm this conclusion, results from the basic case are compared with results from simulations forced by winds with the strength decreased by half (Fig. 8f) and increased by one and one-half (Fig. 8g). The offset is spatially constant in the intensification region, leading to the same wind stress curl in all three cases. Therefore, even though the amount of upwelling at the coast changes, the curl-driven upwelling is the same in all cases. The time series of Sx for the simulation forced by winds with half the strength is very similar to the time series for the basic case (Fig. 11). In the simulation forced by stronger winds (“one-and-one-half strength”), there seems to be a slight delay in the timing of the separation. This may be influenced by the way Sx is computed. Because winds close to the coast are stronger when compared with the basic case, locally driven near-surface currents are also stronger. This means that even if the jet moves offshore at the same time in both cases, the near-surface transport in the basic case can be matched in the case forced by stronger winds by integrating to a distance closer to the coast. The agreement in the timing of separation in this case is indeed closer if an isotherm is used in the calculation (see section 2). Once the jet starts to move offshore in the simulation forced by stronger winds, Sx quickly becomes very similar to the time series from the basic case. This is consistent with the idea that simulations forced by the same wind stress curl field produce roughly similar timing and amounts of separation.
Last, results from the simulation forced by the basic case are compared with results from simulations in which the offshore extent of the region with positive curl close to the coast varies. In this comparison, the same intensity and spatial structure for the wind stress field are used, except that the center of the intensification is shifted either inshore (Fig. 8h) or offshore (Figs. 8i,j). This leads to a narrower extent of the region of positive wind stress curl in Fig. 8h (zero contour at 30 km from the coast, when compared with 80 km in the basic case), and a wider band of positive curl in Figs. 8i,j (zero contour at 130 and 200 km from coast, respectively). Note that because the area of positive curl in Fig. 8h is considerably smaller, curl-driven upwelling is significantly reduced in that case (see Table 1).
Decreasing both the offshore extent of the positive wind stress curl to 30 km (Fig. 8h) and its intensity leads to a separation occurring approximately 15–20 days later as compared with that of the basic case (Fig. 12). The amount of separation, that is, how far offshore the jet is deflected, is reduced. Increasing the area of positive curl (zero-curl line at 130 km; Fig. 8i), on the other hand, leads to more separation. The timing of jet separation is similar to the basic case. Further increasing the offshore extent of the area with positive curl to 200 km (Fig. 8j) leads to separation occurring at a later time (similar to the case forced by spatially uniform wind or by intensified wind with no curl; Fig. 10). A longer simulation shows that Sx continues to increase after day 110, reaching almost 65 km by day 120.
6. Vorticity balance
The absolute values of terms in the vorticity equation are spatially averaged around the cape in a region extending 150 km in the offshore direction and 100 km to the north and south of the cape. Time series of terms for the simulation with a straight coast forced by the same winds as in the basic case (discussed in section 4; Fig. 7) are compared with those for the simulation with the cape forced by winds with curl extending 130 km from the coast (curl far; Figs. 8i and 12), the case in which maximum separation occurs (Fig. 13). The labels beneath (5) are used in the figure. In the straight-case scenario (Fig. 13 top), when no jet separation occurs, the largest terms after 35 days of simulation are the nonlinear advection and the tendency; after 50 days, the remaining terms (pressure gradient, vortex stretching, and surface and bottom stresses) are smaller by a factor of 3, making a roughly equal contribution to the balance.
In the simulation with a cape (Fig. 13 bottom), the balance is significantly different from the straight-coast result. The pressure gradient and the vortex-stretching terms are as important as the nonlinear advection and tendency terms. After 20 days, the contribution from the bottom stress to the vorticity balance is small, as is the contribution from the surface stress torque, because the area in which terms are averaged is substantially larger than the area with intensified wind stress curl. These results demonstrate that bottom topography is of increased importance to jet separation in the presence of a cape. There is also a notable increase in both the pressure gradient and the vortex-stretching terms between days 82 and 92, the same period as when the jet rapidly moves offshore (Fig. 12). This is in agreement with results from section 4, which suggested that bottom topography plays an important role in the separation.
7. Discussion and conclusions
By combining information obtained from the two-dimensional simulations (section 3) with results from comparisons of the effect of different intensities and spatial structures of the wind stress and wind stress curl fields on separation (section 5), a conceptual model of how the wind forcing affects separation can be formed.
The two-dimensional simulations showed that the occurrence of offshore upwelling driven by the wind stress curl leads to a significant cross-shelf gradient in the density field, which is in geostrophic balance with a southward offshore jet. The location of the offshore intensification in the current is closely related to the width of the area of positive wind stress curl. The time scale of the spinup of the offshore current intensification, on the other hand, is related to how strong the curl-driven upwelling is. In the three-dimensional cases, the maximum in the wind stress and wind stress curl intensifications occurs just south of the cape. Therefore, whenever the wind field is characterized by a region close to the coast with positive (negative) curl in the Northern (Southern) Hemisphere, as is the case off Cape Blanco, currents offshore and downstream of the cape are intensified. This leads to alongshore divergence in the offshore region, which could be balanced by an offshore flow from the coast converging at the region of the intensification. The separation of the jet at the cape, which would occur independently of details in the wind field (see Fig. 10), is facilitated by this process. A stronger positive wind stress curl (one and one-half curl; Fig. 8c) leads to an earlier establishment of the offshore current intensification (resulting from two-dimensional, curl-driven upwelling), which leads to jet separation at the cape at shorter time scales when compared with the basic case (Fig. 9). A weaker wind stress curl (“half curl” and “curl near”; Figs. 8b,h) has an opposite effect, leading to a later establishment of the offshore current intensification, and later separation (Figs. 9 and 12). If the offshore extent of the region of positive wind stress curl is decreased (curl near; Fig. 8h), the curl-driven offshore current intensification occurs closer to the coast, leading to a smaller offshore deflection of the jet separated at the cape (Fig. 12). If the area of positive curl is increased (curl far; Fig. 8i), the offshore current intensification occurs farther from the coast, leading to more separation (Fig. 12). However, if the width of the region with intensified positive wind stress curl is increased further (“curl very far”; Fig. 8j), the offshore current intensification resulting from curl-driven upwelling occurs too far from the cape, no longer being able to interact with currents near shore and help separation (Fig. 12). Jet separation, then, occurs at the same time scales as when the forcing has no wind stress curl (Figs. 8d,e and 10), solely by interactions with the cape. Once the jet separates at the cape and moves several kilometers offshore, it is then able to interact with the offshore current intensification, leading to further deflection of the jet.
In all simulations, separation occurs after more than 60 days because time is needed for the wind stress curl to affect the density and, as a consequence, velocity fields (see Fig. 3). Before that, the intensification of the offshore flow is not enough to substantially affect the currents near the cape. The response is not linear, however (e.g., doubling the wind stress curl does not decrease the time of separation by half), and other effects are also important in the process (e.g., instabilities of the flow, inertia around the cape).
An estimation of the potential vorticity modification by currents generated by the wind stress curl field can be obtained by a series of two-dimensional model simulations. The alongshore spacing between the two-dimensional sections is 30 km, and the wind forcing cross-sectional profile is the same as the wind forcing cross-sectional profile in the three-dimensional case at that location. Results at day 110 from the simulations are grouped and used to compute Q2D (Q that is computed based on the two-dimensional simulations will be referred to as Q2D), which is then averaged over the top 80 m. The wind stress curl–driven currents change the background relative vorticity south of the cape, creating Q2D contours that favor separation (black contours in Fig. 14). If it were not for the currents generated by the wind intensification south of the cape, contours of Q2D would follow contours of f /D (because the cross-shore bottom topography profile does not vary among the sections). Results are similar if Q2D is computed for each two-dimensional simulation first [note that in that case y derivatives in (6) are zero], and then is interpolated between the cross sections.
It is important to realize that the previous simulations were highly idealized and do not reproduce some important characteristics of jet separation at Cape Blanco. Satellite and in situ observations show that once the jet separates at the cape it stays offshore of the shelf break, becoming an oceanic jet. In the numerical simulations presented here, the separated jet reattaches to the shelf about 100 km downstream of the cape. Although differences in the details of the bottom bathymetry between the real ocean and the used idealized domain may be partially responsible for the different response, we believe the boundary conditions that are used also play an important role. As described in section 2, we used a Flather condition (Flather 1976) for the depth-averaged normal velocity at the southern boundary. In the Flather condition, one needs to specify the velocity at the boundary, and we chose to use results from two-dimensional simulations (forced by the same wind profile imposed at the southern boundary in the three-dimensional run) as the source of the boundary velocity data. Of course, the two-dimensional run is consistent with the classical picture of coastal upwelling forced by constant winds (remember, there is only a thin band of weak, positive wind stress curl close to the coast in that case), and the upwelling jet is located over the shelf. The differences between the velocity calculated by the model and the specified two-dimensional velocity are allowed to radiate out of the domain at the speed of the external gravity waves (Palma and Matano 1998). Nonetheless, the two-dimensional velocity “imposed” at the boundary represents a constraint on the circulation via the continuity equation acting to preclude jet separation.
Instead of using a two-dimensional simulation as the source of the boundary data, one could use in situ observations, or even climatological data. That option was not pursued here because other issues would be raised. In the observations (or climatology), for example, the upwelling jet during summer is separated from the coast, being found around 120 km from the coast (Huyer et al. 2005). If that is used as boundary condition for the model, a strong southward jet would be imposed in the offshore region at the southern boundary of the domain. That information would propagate into the domain by coastally trapped waves, affecting the circulation around the cape. In other words, jet separation in the model could occur at the cape not by adjustment to internal dynamics, but via volume conservation to satisfy the boundary condition in the south. Although the two-dimensional results used here as boundary conditions may also affect separation, at least their effect is to preclude (and not enhance) separation. Therefore, any simulated jet separation is due to adjustment to the topography/forcing fields, and not a response to the boundary condition.
Although the boundary conditions probably play an important role in jet reattachment in the simulations, other factors not considered in the simulations are also likely to be important. The north–south variation in the Coriolis parameter (the β effect), the poleward undercurrent, and the time variability in the forcing are possibly the most important.
Several studies have shown that β-plane dynamics are important to allow coastally generated jets and filaments to propagate offshore. The offshore scale of the response of the coastal zone to fluctuations in the wind with period P1, at low frequencies, is the distance Rossby waves travel in time P1 (Philander and Yoon 1982). This is consistent with satellite observations, which show a westward movement of the upwelling jet at speeds consistent with Rossby wave dynamics (Strub and James 2000). The upwelling front off central California (Point Sur) also seems to move offshore at speeds consistent with Rossby wave propagation over long periods (Breaker and Mooers 1986). In the Cape Blanco region, once the jet is off the continental slope and free of the strong topographic constrain, variations in the planetary vorticity could aid to its offshore propagation. The implicit “separation formula” derived by Marshall and Tansley (2001) shows that, for an eastern boundary current, the planetary vorticity gradient acts to decelerate the current and, therefore, always encourages separation. Sensitivity studies pursued in previous numerical studies (e.g., Batteen 1997; Marchesiello et al. 2003) show that the inclusion of the β effect increases the horizontal and vertical shear in the upper layers. Currents can become baroclinically and barotropically unstable, leading to the development of meanders and upwelling filaments. Unfortunately, the boundary conditions used do not allow for the β effect to be included in the simulations.
It is not clear how important the β-plane mechanism is in the region around Cape Blanco. At this latitude (∼43°N), Rossby waves propagate at fairly low speeds. However, the β effect could also favor separation by enhancing convergence at the cape. The area south of the cape is characterized by strong, persistent positive wind stress curl. A positive surface stress curl would drive a depth-averaged northward flow via Sverdrup balance. This would lead to convergence at the cape. A convergence in the along-shelf direction (∂υ/∂y < 0) at the cape would drive a westward flow by continuity (∂u/∂y > 0, with u = 0 at the coast), which could help advect momentum offshore.
Another important factor leading to convergence at the cape is wind relaxation events. Gan and Allen (2002) showed that as the upwelling jet passes around Point Reyes in central California, nonlinear terms became important, changing the cross-shelf momentum balance from geostrophic to gradient wind (Holton 1992). This leads to an additional ageostrophic decrease in pressure at the coast. A negative pressure gradient is established south of the cape, and this scenario is balanced under constant wind forcing. As the wind relaxes, a northward flow is established. Barth et al. (2005a) suggested that a similar balance holds at Heceta Bank, Oregon, and acts to drive northward flow following wind relaxation. Because of similarities between the regions, we expect a similar situation to occur at Cape Blanco. Wind relaxation could lead to northward flow, causing convergence at the cape. As discussed above, that could aid in jet separation. There seems to be some evidence of a link between convergence at Cape Blanco and jet separation. Mooring data suggest that the timing of jet separation, as seen from satellite images, is roughly coincident with times of strong convergence at the cape (S. Ramp 2006, personal communication). However, nearshore convergence south of the cape is expected to occur as a result of separation (Fig. 5), making it difficult to isolate the cause and effect.
Observations also suggest that the poleward undercurrent, absent in the present simulations, may contribute to separation. Barth et al. (2000) used in situ data from 1995 to show that once the upwelling jet is already off the shelf, it is deep enough to interact with the upper part of the poleward undercurrent, causing a portion of the undercurrent to turn offshore and then southward, augmenting the now-separated equatorward jet. They concluded that the offshore branch of the poleward undercurrent contributed about 20% of the jet transport at the time of the survey.
Despite the limitations discussed above, the present study sheds light on the effects of the wind stress and wind stress curl on the separation of an upwelling jet at a cape. In summary, the results are in agreement with previous studies, in the sense that the presence of the cape is crucial for separation. The pressure gradient and vortex-stretching terms are very important in the area-averaged vorticity balance around the cape, and jet separation occurs regardless of the structure of the wind field. Wind intensification by itself, without wind stress curl, does not aid in jet separation. The wind stress curl, on the other hand, does affect details of the separation process. The intensity of the wind curl affects the timing of the separation, while the spatial structure of the wind stress curl field affects the amount of separation. Cases in which the jet separates earlier are in closer agreement with observations of jet displacements in the Cape Blanco region. These results point out that numerical modeling efforts to simulate the circulation in the Cape Blanco region and to understand its dynamics should include spatially variable wind forcing to represent the wind stress curl.
Last, recently revealed air–sea interactions in the CCS during summer suggest that the use of a coupled ocean–atmospheric model might be advisable. Chelton et al. (2007) showed that surface winds are weaker over cold water, and stronger over warm water. If the wind blows along a SST front with cold water to the left looking downstream, as is the case most of the time in the Cape Blanco region, positive wind stress curl is generated. Jet separation can then be enhanced by this time-evolving interaction. The offshore movement of the jet causes cold water to reach greater distances from the coast. That increases the width of the region with positive wind stress curl via the air–sea interaction described above. Positive wind curl would then cause offshore upwelling, leading to current intensification even farther from the coast, aiding further separation. This feedback mechanism would be most efficiently investigated with a coupled model.
Acknowledgments
The authors are grateful to John Allen, Roger Samelson, and Murray Levine for helpful comments and suggestions. Alexander Kurapov helped with the implementation of the open boundary conditions. This research was supported by the National Science Foundation under Grant OCE-9907854. Author RC acknowledges support by the Brazilian National Research Council (CNPq Grant 200147/01-3). Additional support was provided by the National Oceanic and Atmospheric Administration (NOAA), U.S. Department of Commerce Award NA03NES4400001 to Oregon State University. The statements, findings, conclusions, and recommendations are those of the authors and do not necessarily reflect the views of NOAA or the Department of Commerce.
REFERENCES
Allen, J. S., 1980: Models of wind-driven currents on the continental shelf. Annu. Rev. Fluid Mech., 12 , 389–433.
Arthur, R. S., 1965: On the calculation of vertical motion in eastern boundary currents from determinations of horizontal motion. J. Geophys. Res., 70 , 2799–2803.
Barth, J. A., and R. L. Smith, 1998: Separation of a coastal upwelling jet at Cape Blanco, Oregon, USA. S. Afr. J. Mar. Sci., 19 , 5–14.
Barth, J. A., S. D. Pierce, and R. L. Smith, 2000: A separating coastal upwelling jet at Cape Blanco, Oregon and its connection to the California Current System. Deep-Sea Res. II, 47 , 783–810.
Barth, J. A., T. J. Cowles, P. M. Kosro, R. K. Shearman, A. Huyer, and R. L. Smith, 2002: Injection of carbon from the shelf to offshore beneath the euphotic zone in the California Current. J. Geophys. Res., 107 .3057, doi:10.1029/2001JC000956.
Barth, J. A., S. D. Pierce, and R. M. Castelao, 2005a: Time-dependent, wind-driven flow over a shallow midshelf submarine bank. J. Geophys. Res., 110 .C10S05, doi:10.1029/2004JC002761.
Barth, J. A., S. D. Pierce, and T. J. Cowles, 2005b: Mesoscale structure and its seasonal evolution in the northern California Current System. Deep-Sea Res. II, 52 , 5–28.
Batteen, M. L., 1997: Wind-forced modeling studies of currents, meanders, and eddies in the California Current System. J. Geophys. Res., 102 , 985–1010.
Breaker, L. C., and C. N. K. Mooers, 1986: Oceanic variability off the central California coast. Prog. Oceanogr., 17 , 61–135.
Capet, X. J., P. Marchesiello, and J. C. McWilliams, 2004: Upwelling response to coastal wind profiles. Geophys. Res. Lett., 31 .L13311, doi:10.1029/2004GL020123.
Castelao, R. M., and J. A. Barth, 2006: The relative importance of wind strength and alongshelf bathymetric variations on the separation of a coastal upwelling jet. J. Phys. Oceanogr., 36 , 412–425.
Castelao, R. M., T. P. Mavor, J. A. Barth, and L. C. Breaker, 2006: Sea surface temperature fronts in the California Current System from geostationary satellite observations. J. Geophys. Res., 111 .C09026, doi:10.1029/2006JC003541.
Chapman, D. C., 1985: Numerical treatment of across-shore open boundaries in a barotropic ocean model. J. Phys. Oceanogr., 15 , 1060–1075.
Chelton, D. B., M. G. Schlax, M. H. Freilich, and R. F. Milliff, 2004: Satellite measurements reveal persistent small-scale features in ocean winds. Science, 303 , 978–983.
Chelton, D. B., M. G. Schlax, and R. M. Samelson, 2007: Summertime coupling between sea surface temperature and wind stress in the California Current System. J. Phys. Oceanogr., 37 , 495–517.
Dale, A. C., and J. A. Barth, 2001: The hydraulics of an evolving upwelling jet flowing around a cape. J. Phys. Oceanogr., 31 , 226–243.
Dorman, C. E., T. Holt, D. P. Rogers, and K. Edwards, 2000: Large-scale structure of the June–July 1996 marine boundary layer along California and Oregon. Mon. Wea. Rev., 128 , 1632–1652.
Flather, R. A., 1976: A tidal model of the northwest European continental shelf. Mem. Soc. Roy. Sci. Liege, 6 , 141–164.
Galperin, B., L. H. Kantha, S. Hassid, and A. Rosati, 1988: A quasi-equilibrium turbulent energy model for geophysical flows. J. Atmos. Sci., 45 , 55–62.
Gan, J., and J. S. Allen, 2002: A modeling study of shelf circulation off northern California in the region of the Coastal Ocean Dynamics Experiment: Response to relaxation of upwelling winds. J. Geophys. Res., 107 .3123, doi:10.1029/2000JC000768.
Gan, J., and J. S. Allen, 2005a: On open boundary conditions for a limited-area coastal model off Oregon. Part 1: Response to idealized wind forcing. Ocean Modell., 8 , 115–133.
Gan, J., and J. S. Allen, 2005b: Modeling upwelling circulation off the Oregon coast. J. Geophys. Res., 110 .C10S07, doi:10.1029/2004JC002692.
Haidvogel, D. B., A. Beckman, and K. Hedstrom, 1991: Dynamical simulations of filament formation and evolution in the coastal transition zone. J. Geophys. Res., 96 , 15017–15040.
Haynes, R., E. D. Barton, and I. Pilling, 1993: Development, persistence and variability of upwelling filaments off the Atlantic coast of the Iberian Peninsula. J. Geophys. Res., 98 , 22681–22692.
Holton, J. R., 1992: An Introduction to Dynamic Meteorology. Academic Press, 511 pp.
Huyer, A., J. H. Fleischbein, J. Keister, P. M. Kosro, N. Perlin, R. L. Smith, and P. A. Wheeler, 2005: Two coastal upwelling domains in the northern California Current System. J. Mar. Res., 63 , 901–929.
Marchesiello, P., J. C. McWilliams, and A. Shchepetkin, 2001: Open boundary conditions for long-term integration of regional oceanic models. Ocean Modell., 3 , 1–20.
Marchesiello, P., J. C. McWilliams, and A. Shchepetkin, 2003: Equilibrium structure and dynamics of the California Current System. J. Phys. Oceanogr., 33 , 753–783.
Marshall, D. P., and C. E. Tansley, 2001: An implicit formula for boundary current separation. J. Phys. Oceanogr., 31 , 1633–1638.
Mellor, G. L., and T. Yamada, 1982: Development of a turbulence closure model for geophysical fluid problems. Rev. Geophys., 20 , 851–875.
Mesias, J. M., R. P. Matano, and P. T. Strub, 2001: A numerical study of the upwelling circulation off central Chile. J. Geophys. Res., 106 , 19611–19623.
Mesias, J. M., R. P. Matano, and P. T. Strub, 2003: Dynamical analysis of the upwelling circulation off central Chile. J. Geophys. Res., 108 .3085, doi:10.1029/2001JC001135.
Narimousa, S., and T. Maxworthy, 1987: On the effects of coastline perturbations on coastal currents and fronts. J. Phys. Oceanogr., 17 , 1296–1303.
Palma, E. D., and R. P. Matano, 1998: On the implementation of passive open boundary conditions for a general circulation model: The barotropic mode. J. Geophys. Res., 103 , 1319–1341.
Peffley, M. B., and J. J. O’Brien, 1976: A three-dimensional simulation of coastal upwelling off Oregon. J. Phys. Oceanogr., 6 , 164–180.
Penven, P., C. Roy, A. Colin de Verdière, and J. Largier, 2000: Simulation of a coastal jet retention process using a barotropic model. Oceanol. Acta, 23 , 615–634.
Perlin, N., R. M. Samelson, and D. B. Chelton, 2004: Scatterometer and model wind and wind stress in the Oregon–northern California coastal zone. Mon. Wea. Rev., 132 , 2110–2129.
Philander, S. G. H., and J-H. Yoon, 1982: Eastern boundary currents and coastal upwelling. J. Phys. Oceanogr., 12 , 862–879.
Pickett, M. H., and J. D. Paduan, 2003: Ekman transport and pumping in the California Current based on the U.S. Navy’s high-resolution atmospheric model (COAMPS). J. Geophys. Res., 108 .3327, doi:10.1029/2003JC001902.
Samelson, R., and Coauthors, 2002: Wind stress forcing of the Oregon coastal ocean during the 1999 upwelling season. J. Geophys. Res., 107 .3034, doi:10.1029/2001JC000900.
Shchepetkin, A. F., and J. C. McWilliams, 2003: A method for computing horizontal pressure-gradient force in an oceanic model with a nonaligned vertical coordinate. J. Geophys. Res., 108 .3090, doi:10.1029/2001JC001047.
Shchepetkin, A. F., and J. C. McWilliams, 2005: The Regional Oceanic Modeling System (ROMS): A split-explicit, free-surface, topography-following-coordinate oceanic model. Ocean Modell., 9 , 347–404.
Song, Y. T., and D. B. Haidvogel, 1994: A semi-implicit ocean circulation model using a generalized topography-following coordinate system. J. Comput. Phys., 115 , 228–244.
Strub, P. T., and C. James, 1995: The large-scale summer circulation of the California current. Geophys. Res. Lett., 22 , 207–210.
Strub, P. T., and C. James, 2000: Altimeter-derived variability of surface velocities in the California Current System. 2: Seasonal circulation and eddy statistics. Deep-Sea Res. II, 47 , 831–870.
APPENDIX
Thermal Balance Computation
To evaluate terms in the potential temperature equation, we follow Gan and Allen (2005b). Their approach is repeated here for completeness.
As pointed out by Gan and Allen (2005b), the sum is preserved (ADV = HOR + VER), and the common part of opposite sign that would cancel in the sum is removed. For ADV < 0, the signs of the terms with absolute values in (A3) and (A4) are reversed.
Summary of experiments.