Coastal ocean response to summer upwelling favorable winds in a region of alongshore bottom topography variations off Oregon

Abstract

[1] Repeated mesoscale surveys of waters over the shelf and slope off Oregon were conducted during spring and summer of 2001 to study the spatial structure of the velocity and hydrographic fields. The ocean response to wind forcing is compared between a region of relatively simple topography with alongshore uniformity and a region of complex topography including a shallow submarine bank (Heceta Bank). In the simple topography region the upper water column is influenced by upwelling and fresh water from the Columbia River during spring, with the low-salinity water located farther offshore during summer. Variability in the fields is mostly confined within 30 km from the coast. Over Heceta Bank the region of higher variability is broader, spanning most of the shelf. The coastal upwelling jet is located inshore of the pinnacle of the bank during spring, moving offshore during summer. The region inshore of the bank is characterized by low velocities and flow recirculation. Near-surface fields show that the circulation evolves considerably between seasons. In both regions the area influenced by upwelled water is much broader during summer. Dense waters found over midshelf off Newport are upwelled to the north and advected south. Dense waters inshore of Heceta Bank are substantially influenced by water from the north during spring, but their source is mostly from the south during summer. South of the bank, the separation of the jet significantly increases the cross-isobath transport, constituting an efficient mechanism for transport of material from the shelf into deeper waters.

1. Introduction

[2] Continental shelves located at the eastern boundaries of the oceans have long been recognized as highly productive zones due to coastal upwelling. Off the Oregon coast, strong equatorward winds during the summer drive a net offshore transport in the surface Ekman layer, leading to upwelling of cold, saline, nutrient-rich water near the coast. A strong alongshore southward coastal upwelling jet is formed in geostrophic balance with the upwelled isopycnals [Huyer, 1983]. The density structure within the surface Ekman layer is highly variable as a result of changes in the wind stress [Huyer et al., 1974; Holladay and O'Brien, 1975], but the upward sloping isopycnals in the geostrophic interior persist through the upwelling season [Huyer, 1977; Austin and Barth, 2002]. Offshore waters are generally fresher, in part due to southward advection of Subarctic waters through the California Current [Reid et al., 1958]. This general description is basically two-dimensional, describing no variations in the alongshore direction. Flow at the margins of ocean basins is highly complex, due in part to interactions of the flow with topography.

[3] Most of shelf studies to date have been in regions of simple topography. Off Oregon, most of the previous field experiments were concentrated off and to the north of Newport, where local isobaths and the shape of the coastline vary only slowly with along-shelf distance. Examples of such experiments are the Wisp and the Coastal Upwelling Experiments (CUE-I and CUE-II), which have been extensively discussed in the literature [Kundu et al., 1975; Kundu and Allen, 1976; Huyer et al., 1978, 1979, and references therein]. One exception was the Coastal Jet Separation experiment, which revealed complex, highly three-dimensional flow patterns farther south near Cape Blanco due to interactions with coastal topography [Barth et al., 2000].

[4] There exists a major submarine bank south of Newport (Heceta Bank, 44.2°N). Although its physical oceanography has not been studied in detail prior to 1999 [Oke et al., 2002a, 2002b; Barth et al., 2005b], indirect observations (e.g., low temperature and high chlorophyll concentration in satellite images, fishing success) suggest that upwelling, and hence productivity, are probably enhanced in its vicinity. Satellite images indicate that the interaction of the southward upwelling jet with Heceta Bank can produce complex flow patterns, increasing the offshore excursion of cold water [Barth et al., 2000]. Numerical studies of the area [Oke et al., 2002b] suggest that north of Newport the upwelling circulation is more consistent with standard conceptual models for two-dimensional across-shore circulation, while over Heceta Bank the circulation is complicated, with weaker direct coupling to the wind forcing over most of the shelf.

[5] In a companion paper, Barth et al. [2005a] discuss time-dependent effects, circulation during wind relaxation and reversal, and the nature of a deep suspended material pool over Heceta Bank during the 2001 upwelling season. In the present study, the aim is to describe the spatial structure of the temperature, salinity, density, and velocity fields during the 2001 upwelling season off Oregon, focusing on the differences between the region north of Newport (simple topography) and over Heceta Bank.

2. Methods

[6] Repeated mesoscale surveys of waters over the shelf and slope off Oregon were conducted on two hydrographic cruises during the upwelling season of 2001, as part of the Coastal Ocean Advances in Shelf Transport (COAST) experiment [Barth and Wheeler, 2005]. The first cruise was conducted in late spring (23 May to 13 June), and the second during summer (6–25 August). The surveys were composed of zonal sections 60–100 km long, starting at the 45 m isobath, the shallowest depth in which towed vehicle operations could be conducted safely. Sections were separated by 24 km on average in the north-south direction, covering a 165-km-long region from 43.75° to 45.25°N. The survey area includes the region north of Newport, with relatively straight topographic contours, and the region around Heceta Bank (Figure 1).

Figure 1

Study area showing cruise tracks (bold line). Sections are named line 1 (L1) to line 8 (L8). Also shown are the locations of the moorings (crosses, position offset 0.04° north for clarity), NOAA National Data Buoy Center (NDBC) buoy 46050 (circle), and conventional conductivity-temperature-depth (CTD) stations (triangles). During spring, conventional CTD stations were measured only along 44.22° and 45°N. All spring CTD stations were occupied twice, while during summer, only stations along 44.22°N were occupied twice. Topographic contours shown are 50, 70, 90, 100, 200 (dashed), 500, 1000, and 2000 m.

[7] Hydrographic data were collected using a towed, undulating vehicle, SeaSoar [Pollard, 1986], which cycles rapidly from the surface to depth (typically ∼110 m) while being towed at 4 m s⁻¹ (8 knots) behind the R/V Wecoma. The SeaSoar vehicle was equipped with a Sea-Bird 911-plus conductivity-temperature-depth (CTD) instrument with dual temperature/conductivity sensors and a WET Labs Wetstar fluorometer. Phytoplankton fluorescence was calibrated against fluorometric determination of discrete samples analyzed according to Strickland and Parsons [1972]. The surveys were intended to provide nearly synoptic coverage with mesoscale horizontal resolution, being completed in 2–2.5 days. Typical along-track resolution was 1200 m for surface points and 600 m at middepth, improving to 600 m (surface) and 300 m (middepth) over the shallow continental shelf. Some conventional CTD stations were also made, covering the entire water column. Data processing is described in detail by O'Malley et al. [2002] and Barth et al. [2003] (both available at http://damp.coas.oregonstate.edu/coast/seasoar). After processing, measured and derived quantities (temperature, salinity, density anomaly σ_t, computed following Fofonoff and Millard [1983], and chlorophyll), obtained in a sawtooth-shaped pattern, were gridded into vertical sections using a local polynomial method [Pierce and Barth, 2000]. A second-order polynomial surface is fit via least squares to the data points in a neighborhood surrounding each output grid point location. The neighborhood is defined by using a tricube window with width equal to 1.5 km. The resulting vertical profiles were separated horizontally by 500 m, with 2 dbar vertical bins.

[8] Geopotential anomalies were computed using a reference level of 50 dbar, since the presence of Heceta Bank complicates the use of a deeper reference level. On sections where the SeaSoar profiles were shallower that 50 dbar, geopotential anomaly was calculated using the extrapolation technique described by Reid and Mantyla [1976]. Tests using a reference level of 90 dbar wherever possible produced results qualitatively similar to those reported here.

[9] Velocity profiles along the ship track were obtained with an RD Instruments hull-mounted 153.6 kHz narrowband acoustic Doppler current profiler (ADCP). Data were collected nearly continuously using an ensemble averaging interval of 1 minute and a vertical bin length of 8 m. The inherent short-term random uncertainty is estimated to be ±0.02 m s⁻¹ [Pierce and Barth, 2002a, 2002b] (available at http://damp.coas.oregonstate.edu/coast/adcp). The deepest measurement was generally deeper than 250 m, and the shallowest reliable data were from 17 m. Tides were removed from the data by using an estimate of the tidal motion from the Erofeeva et al. [2003] tidal model. The maximum magnitudes of the computed tidal velocities were everywhere less than 0.08 m s⁻¹, except over Heceta Bank, where they reached 0.12 and 0.17 m s⁻¹ at 44.25° and 44°N, respectively. The velocity data were gridded using the same method described for the hydrographic data, but with vertical grid spacing equal to 8 dbar. Velocities were rotated into a modified principal axis system, following Kosro [1987]. For that definition, 〈〉 was found along each line (, are means of the east-west and north-south components of velocity, respectively, and angle brackets indicate averaging over the across-shore and the vertical directions). The sections were rotated to a coordinate system for which 〈〉 = 0, that is, one for which the spatial structure in u is forced to be uncorrelated with the spatial structure in v [Kosro, 1987]. Angles of rotation for each section are given in Table 1.

Table 1. Number of Surveys and Modified Principal Axis (Measured Clockwise From North) for Each Line of COAST 2001 Experiment

	Spring			Summer
	Number of Surveys	Number of Surveys Used for Mean Calc.	Major Principal Axis, deg	Number of Surveys	Number of Surveys Used for Mean Calc.	Major Principal Axis, deg
Line 1	5	5	−2	4	2	−2
Line 2	15	15	7	7	4	18
Line 3	5	5	7	4	2	38
Line 4	5	5	11	5	3	22
Line 5	5	4	14	4	3	−2
Line 6	5	4	27	5	3	−4
Line 7	4	3	3	4	3	12
Line 8	5	4	5	4	3	−7

[10] To calculate transport through each section, southward ADCP velocities were integrated down to 200 m, assuming uniformity between the surface and the velocity measured closest to the surface (17 m). Note that this will result in an underestimation of the transports because the jets off the Oregon coast during this time of the year are vertically sheared.

[11] Averaged fields were computed using only data collected during upwelling winds, as defined by the sign of a running mean of the hourly wind stress. Austin and Barth [2002] develop a methodology that compares a measure of upwelling intensity (height at the coast of an exponential fit to the 25.8 σ_t contour) and an exponentially weighted running mean of the alongshore component of the wind to relate frontal displacement and wind stress. A similar analysis using the present data set suggested that the front relaxes to a dynamic (geostrophic) equilibrium on a timescale of roughly 5 days. We then use an integral of the wind stress (W_5d), given by

where τ^y is the alongshore component of the wind stress, ρ_○ is a reference density, and k = 5 days is the relaxation timescale, to determine which sections were measured under upwelling conditions. Winds were measured at the NOAA National Data Buoy Center buoy 46050 at Stonewall Bank (37 km west of Newport, see Figure 1 for location), and neutral wind stress was computed following Large and Pond [1981]. The alongshore wind stress, divided by a reference density, and the integrated W_5d are shown in Figure 2. To illustrate the temporal coverage of the sampling, vertical bars on the second and third panels represent each time data were collected along line 2 (45.02°N). Sections marked with a black bar were discarded in the calculation of the mean, since they were measured under positive W_5d conditions, i.e., downwelling favorable integrated winds. The same criteria was used for selecting sections obtained during upwelling conditions for the other lines. The number of surveys at each line is given in Table 1, along with the number of surveys actually used in calculating the mean. Note that for some lines the number of surveys is relatively small (particularly during summer), leading to less certainty in the standard deviation estimates (e.g., lines 4, 6–8 for summer).

Figure 2

(top) Observed hourly alongshore component of the wind stress from the NDBC station 46050. The time of the cruises is denoted by the boxes. (middle, bottom) Wind stress from spring and summer cruises, respectively. The dashed line is the integrated wind (W_5d × 10⁻⁵ m² s⁻¹; see explanation in text). The time of each SeaSoar survey at line 2 (45.02°N), as an example, is denoted by a vertical bar. Surveys during downwelling favorable W_5d are marked by a black vertical bar.

[12] Observations also include time series of velocity and salinity collected from mid-May to late August 2001 at mooring lines along 44.22°N (located at 51, 99, and 132 m isobaths) and 45°N (located at 50, 81, and 130 m isobaths) (Figure 1). Velocity was measured with 300 kHz RDI ADCP and 250 and 500 kHz Sontek acoustic Doppler profiler (ADP) moored near the bottom looking up. Salinity was measured with a Seabird 37-SM MicroCAT. Mooring data collecting and processing is described in detail by Boyd et al. [2002]. The time series were low-pass filtered (half-power point of 40 hours) to suppress tidal and inertial oscillations.

3. Results

3.1. Horizontal Structure and Variability

[13] A near-surface (6 m) map of averaged temperature during spring (Figure 3) shows a strong offshore gradient, with coldest water adjacent to the coast, consistent with upwelling favorable winds. In the northern region, where the topography is relatively simple, the offshore gradient is highest close to the coast, decreasing in the offshore direction. The minimum in the temperature field is found in the south, where cold waters reach a greater distance from the coast. This pattern could be a result of enhanced upwelling in the south. In addition, the warmer water from the Columbia River (mouth located ∼140 km north of the study region) is closer to the coast in the north (see also salinity plot), particularly during wind relaxation events, which could contribute to the warmer average temperature found there. The average field at 25 m shows a similar pattern, with the coldest water displaced slightly to the north compared with 6 m temperatures. At 25 m, the influence of Columbia River waters is small, thus allowing a cold water core to be maintained north of the bank.

Figure 3

Near-surface (left) 6 m and (right) 25 m (top) temperature, (middle) salinity, (bottom) geopotential anomaly (GA), and acoustic Doppler current profiler (ADCP) velocity averaged fields for spring. Shaded lines are the 100 and 200 m isobaths. Contour interval is 0.5°C for temperature, 0.2 for salinity, and 0.1 m² s⁻² for GA.

[14] Surface waters inshore and to the south of the bank (43.75°–44.48°N) are much saltier than inshore waters north of it. In the north, water influenced by the Columbia River (S < 32.5 [Huyer et al., 2002]) reaches all the way to the coast, while in the bank region the fresher water is restricted offshore. In contrast to the temperature fields, which were monotonic in the cross-shore direction almost everywhere, the salinity field presents a local minimum near the shelf break, due to advection in jet of low-salinity waters. Below the fresher surface layer, the salinity field is similar to the temperature field, with a high gradient in a narrow band close to the coast in the northern region, and with the saltier water spanning a larger area of the shelf in the south. The structure of the density field (not shown) is very similar to the salinity field.

[15] The mean velocity data from 25 m for the spring cruise are also shown in Figure 3. Since we do not have ADCP measurements at 6 m (the shallowest ADCP data is from 17 m), we present a geopotential anomaly map near the surface instead. Kundu et al. [1975] found that the temperature fluctuations at 40 m at a mooring located at the 100 m isobath close to 44.85°N were significantly correlated with the first baroclinic and second empirical modes of the alongshore velocity there, which is consistent with the hypothesis that the thermal wind equation holds [Huyer et al., 1978]. The position of the jet in the near surface geopotential anomaly field and the ADCP map at 25 m is very similar, i.e., the jet is located close to the coast in the north, gradually moving offshore as it passes over the bank. At 25 m, maximum velocities occur roughly at locations of maximum salinity (and hence density) gradients, and velocities are relatively weak over the midshelf over Heceta Bank (inshore of 124.5°W at 44°N). There is significant cross-isobath flow throughout the region at that level. Velocities decay roughly exponentially in the offshore direction. The coastal jet separates south of the bank, and it is found much more offshore at 43.75°N (600 m isobath) than in the northern region (inshore of the 100 m isobath), even though the width of the shelf is similar in those areas. The near-surface geopotential anomaly contours show an average positive alongshore pressure gradient between 44° and 45°N, implying an associated offshore flow there.

[16] Later in the summer, the continuous input of energy from the predominantly upwelling favorable winds have significantly changed the field patterns (Figure 4). The surface averaged temperature field shows that the offshore water has warmed by 2.5°C, probably due to the seasonal surface heating. Climatological values of net summer downward heat flux (∼150 W m⁻²) indicate that surface heating would be sufficient to heat a 25-m-deep surface layer by 2.5°C in less than a month. A tongue of warm water is found inshore of the bank, around ∼124.5°W, and is likely related to the flow recirculation around the bank bringing offshore waters onto the bank (see 25 m ADCP along 44°N). The tongue-like structure is less pronounced in the surface salinity average, suggesting that local surface heating in the region of low flow inshore of Heceta Bank is also important [Barth et al., 2005b]. At 25 m, the temperature presents a much more homogeneous spatial structure, especially inshore of the 200 m isobath. During summer, the mean thermocline is very shallow and horizontal variations are relatively small below it. A large gradient is only found in the very offshore parts of lines 7 (44°N) and 8 (43.75°N), where the mean mixed layer gets considerable thicker.

Figure 4

Same as Figure 3, except for summer.

[17] The salinity fields at both levels show that the area of upwelling is broader in summer than in spring and, as in spring, is broader at and to the south of the bank than north of it. Surface salinities are everywhere higher than in spring, probably a combined effect of persistent upwelling winds and a decrease in the Columbia River discharge. Monthly averaged discharges from 2001 (spring to summer average discharge was lower than normal) from the U.S. Geological Survey station USGS 1424900 in m³ s⁻¹ were as follows: April, 4262; May, 4948; June, 4297; July, 2786; and August, 3011.

[18] The offshore shift of the jet as it approaches the bank is more abrupt during summer. It will be shown later (section 4.3) that this shift contributes to the observed increase in salinity. The geopotential anomaly contours turn sharply offshore at about line 3 (44.83°N), a result consistent with the principal axis calculation from the ADCP data (Table 1). This is expected, since the salinity (and density) front moves abruptly offshore there. As in spring, there is an average positive alongshore pressure gradient north of the bank. The velocity field at 25 m shows a much less clear picture of the jet than in spring. Some of the obvious features are the change in the location of the core of the jet at and to the south of the bank (south of 44.25°N), being found more offshore than during spring, and the increase in the recirculation at line 7 (44°N). This increase in the recirculation can also be seen at the geopotential anomaly map. Off Newport (44.65°N), the mean inshore velocities are weakly southward, but northward flow was measured in some of the individual surveys [Pierce and Barth, 2002b]. This is frequently observed [e.g., Huyer et al., 1974; Kundu and Allen, 1976], and numerical model simulations [Oke et al., 2002b] suggest it to be a response to a negative alongshore pressure gradient following wind relaxation events. A negative alongshore pressure gradient is not apparent off Newport, but is present between 43.75° and 44.25°N, offshore of the 100 m isobath. There are two local maxima of southward velocities at 43.75°N. The offshore core is a result of the separation of the jet south of the bank, similar to what happened in spring. A jet has spun-up closer to the coast (inshore of 200 m isobath), presumably with contributions from both local upwelling and from recirculation around the bank.

[19] Chlorophyll concentrations are enhanced over and inshore of the bank (Figure 5), and it is also evident that the offshore shift of the jet during summer increases the width of the area of high concentrations. A similar result was also found by Barth et al. [2005b]. Along 43.75°N, the offshore extent of high chlorophyll concentrations is maximum, reaching deeper waters compared to the bank transects. This indicates that the separation of the coastal jet in both seasons causes significant cross-isobath transport, increasing the export of material from the shelf to the deep ocean.

Figure 5

Near-surface (6 m) chlorophyll (derived from fluorescence calibrated with discrete samples) averaged fields for (left) spring and (right) summer. Dashed lines are the 100 and 200 m isobaths. Contour interval is 3 mg m⁻³.

3.2. Vertical Structure and Variability

[20] Since there is a relative alongshore uniformity in the property fields in the northern region, we only show vertical sections along lines 2 (45.02°N), 4 (45.65°N), and 6–8 (44.25°–43.75°N). Figure 6 presents mean temperature sections for those locations for both spring and summer. The temperature field during spring shows the results of coastal upwelling, with cold water at the surface near the coast and isotherms tilted upward. At line 2, this tilting is maximum close to the coast, while at the bank the maximum tilting occurs farther offshore. In most sections (except the offshore regions of the southern lines), the temperature vertical gradient reaches the near surface, indicating a relatively thin mixed layer. During summer, the vertical gradient at the thermocline level is much more intense, presumably due to surface heating warming up the upper layer, and the effect of predominant southward winds throughout spring and summer, upwelling cold water from below. At line 2, surface water is warmed by more than 2°C compared to spring, while water below the thermocline is cooled by as much as 1.5°C. The volume of cold water over the shelf is larger in summer. The 8°C isotherm is lifted by 20–40 m over the shelf, the same happening at line 4. Lifting of the 8°C isotherm can also be seen at lines 6 and 7, although a little bit less intense. It is interesting to note that at the bank region the summer subsurface cooling occurs at and offshore of the pinnacle of the bank (Figure 7). Subsurface waters close to the coast (inshore of 124.45°W) experience less temperature variation (<0.5°C). Along line 7 (Figure 6), water close to the bottom inshore of 124.7°W is actually warmer in summer than in spring; the same holds true along line 8. This is probably related to the intensification on the recirculation and to flow from offshore onto the bank during summer, which happens at all depths. It is very clear that the maximum horizontal temperature gradient during summer at lines 6 and 7 is located offshore of the pinnacle of the bank, in contrast to spring conditions, when the front was located inshore of the bank (Figure 6). In both sections, there is a large area of the shelf in which the isotherms are almost level, turning upward again near the coast. The mixed layer depth seemed to increase in summer, particularly in the southern offshore region.

Figure 6

Across-shore sections of the averaged temperature field for (left) spring and (right) summer along (from top to bottom) lines 2, 4, 6, 7, and 8. Contour interval is 0.5°C, with bold contours every 2°C. The dashed line is the 8°C contour.

Figure 7

Across-shore section of the averaged seasonal temperature difference ΔT (summer T minus spring T) along line 6 (44.25°N). Positive (negative) values of ΔT are indicated by solid (dashed) contours. The contour interval is 0.5°C; the bold contour is zero.

[21] The standard deviations (SD) of the temperature sections are shown in Figure 8. During spring, highest variability in the temperature field is confined to the upper 20–30 m. The temperature SD are well correlated to the vertical gradient, suggesting that internal waves and tides are an important source of the variability. At line 2, maximum values are found near the coast, while at lines 6 and 7 they are found just inshore of the pinnacle of the bank. These locations are roughly coincident with the position of the temperature front, indicating that part of the variability is related to offshore displacement of water during upwelling events, and inshore movement during relaxations. During summer, maximum values are found on a subsurface layer, which deepens offshore. This layer coincides with the position of the thermocline, getting deeper in the south, particularly at lines 7 and 8, where the mixed layer is thicker offshore.

Figure 8

Across-shore sections of temperature standard deviations (°C) for (left) spring and (right) summer along (from top to bottom) lines 2, 4, 6, 7, and 8; max, maximum value.

[22] Vertical sections of the averaged salinity show a pattern similar to the temperature field, with the front located closer to the coast in the northern lines, and farther offshore at the bank in both seasons (Figure 9). A strong core of fresh waters (S < 30.5) is located around 20–30 km offshore at line 2 (45.02°N) during spring. The greater influence of Columbia River water (see also Figure 3) creates a strong shallow halocline, which is absent during summer. Similar to the temperature field, the salinity front over the bank moves offshore during the second cruise. Salinities are higher in summer than in spring over most of the water column.

Figure 9

Same as Figure 6, except for salinity. Contour interval is 0.25, with bold contours every 1. The dashed line is the 33 contour.

[23] The spring salinity SD (Figure 10) also show maximum variability at the surface, due to internal waves and cross-shore displacements of the position of the front in response to wind events. At line 2, highest values of SD are found at 20–30 km offshore, where the core of fresher water is located, and higher variability is confined within 30 km from the coast. Over the bank, the area of higher variability is wider and detached from the coast, extending from 13 to 50 km from the coast at line 6, being even broader at line 7. The broadening of the higher-variability region over the bank is also evident in the density SD field (not shown). All spring sections present a deep feature of relatively higher SD over the shelf extending down to 90–100 m. This feature coincides with the position of the permanent halocline, and is presumably a combined effect of internal waves and horizontal movement of the cross-shelf salinity gradient. The jet moves around more frequently in the south, a possible reason for the intensification of the feature at lines 6 and 7. During summer, the salinity variability is much less. The high value at the surface along line 2 occurs due to southward advection of a fresh water lens during one of the surveys [O'Malley et al., 2002].

Figure 10

Same as Figure 8, except for salinity.

[24] Mean alongshore velocities (rotated and detided as described in section 2) during spring are southward everywhere over the shelf at line 2 (Figure 11). The coastal jet is strongly vertically sheared, with maximum velocities found at 12 km from the coast. This result is similar to the findings of Huyer et al. [1978]. The alongshore velocity SD increase toward the coast (Figure 12), but they are smaller than the mean currents over the shelf except in a thin layer close to the bottom, indicating that the flow is persistently southward. Inshore of the 90 m isobath, SD are nearly depth-independent. During summer, the core of the jet on line 2 moves offshore, and the vertical shear is reduced (Figure 11). Southward velocities are weaker, and the mean flow along the bottom is poleward, as reported by Huyer et al. [1978]. Standard deviations near the coast on line 2 are smaller in summer than in spring (Figure 12). The general picture along line 4 (44.65°N) is similar, although the flow close to the bottom over the shelf during summer is weakly southward (considerably weaker than during spring), and high values of SD are also found near the shelf break. The core of the jet approximately follows the 90 m isobath.

Figure 11

Across-shore sections of averaged alongshore velocity. Contour interval is 0.05 m s⁻¹. Vertical lines along middle panels (line 6) show the location of the moorings. Coordinate system has been rotated using angles shown in Table 1.

Figure 12

Same as Figure 8, except for alongshore velocity (m s⁻¹).

[25] Farther south (lines 6 and 7), the jet is located inshore of the pinnacle of the bank during spring, moving offshore during summer, a result already expected from the temperature and salinity plots through the thermal wind balance. Standard deviations during spring are higher than the mean in some regions over the bank at lines 6 and 7, suggesting that reversals in the flow direction might be frequent. The summer intensification of the recirculation at line 7 already observed at the 25 m ADCP map occurs at all depths. In both sections, northward flow likely associated with the poleward undercurrent can be observed offshore of the shelf break, getting intensified and reaching shallower depths during summer. At line 8, the southward flow over the shelf is stronger during summer, in contrast to the other lines.

4. Discussion

4.1. Relations to Wind Forcing

[26] The dominant empirical orthogonal functions (EOFs) of the demeaned density field for across-shore sections were computed, to assess relations between σ_t and the wind forcing. Only data from spring were used, since there is a better temporal coverage during the first cruise (Table 1). The amplitude time series was correlated to the north-south component of the wind stress (τ^y). In the northern region (45.02°N), the maximum lagged 95% significant correlation coefficient between the mode 1 (22% of variance) and 2 (14% of variance) amplitude time series and τ^y is 0.65 and 0.53, respectively, with lag time of roughly 1.9 inertial periods. An EOF decomposition of velocity data from an east-west mooring array (45.25°N) was performed by Kundu and Allen [1976]. They found that the first mode explains about 74% of the energy for the v component, and the amplitude time series of mode 1 was correlated to the wind stress. These results show that both the velocity and density fluctuations are directly linked to the wind forcing in the northern region.

[27] Over the bank (44.25°N), the number of section is much smaller and results must be analyzed with caution. There, only the mode 1 EOF (74% of variance) can be distinguished from the product of an EOF analysis of a spatially and temporally uncorrelated random process (following Overland and Preisendorfer [1982]). The mode 1 amplitude time series is not correlated to the wind, suggesting that density fluctuations are not as strongly linked to wind forcing there compared with the northern region. Despite the small number of sections available, this result based on in situ data is in agreement with a similar analysis performed by Oke et al. [2002b] using numerical modeling results.

[28] Kosro [2005], using HF radar measurements off Oregon from summer 2001, showed that the correlation between meridional currents and meridional wind stress varies spatially. The strongest relation occurred near the coast, in the narrow shelf/simple topography region north of Cape Foulweather (44.8°N). He found a notable breakdown of the relationship over Heceta Bank, indicating a weaker direct coupling to wind forcing over the bank.

4.2. Transport Estimates

[29] The jet averaged transport computed as described in section 2 from both seasons is shown in Figure 13. An estimate of the standard error of the mean which assumes that all surveys represent independent samples is also shown in Figure 13. During spring, alongshore variations in the transport north of 44°N are relatively small, showing volume conservation in the jet. Alongshore variability is higher in summer, but that might be related to the lower number of measurements during that season, particularly at lines 1 (45.25°N) and 3 (44.83°N) (Table 1). The spring (summer) averaged transport plus and minus one standard deviation north of 44°N is 0.66 ± 0.20 (0.44 ± 0.20) Sv. There is an increase in the transport at 43.75°N compared to the other sections in both seasons (1.42 ± 0.45 for spring, 1.25 ± 0.56 for summer), perhaps due to entrainment as the jet moves from shallow waters over the shelf into the slope. During summer, the second core of southward velocities inshore of the 200 m isobath also contributes to that increase.

Figure 13

Averaged southward transport for spring (bold line) and summer (thin line) cruises. Horizontal lines are ±1 standard error of the mean, assuming all samples are independent, for spring (circles) and summer (crosses). Position of uncertainty estimates are offset slightly for clarity; 1 Sverdrup (Sv) = 10⁶ m³ s⁻¹.

[30] The mean southward transport south of the bank (43.75°N) is in good agreement with previous estimates present in the literature. The ADCP-derived southward transport above 200 m from sections measured in June 1987 were 1.5, 2.0, and 3.5 Sv at 43.2°, 41.5°, and 40.0°N, respectively [Smith, 1995]. The transport in the meandering eastern boundary current has been observed to increase downstream. Barth et al. [2000] estimate the total southward geostrophic transport relative to 200 m to be 0.93 Sv across 43°N during August 1995, again comparing well with the present estimate. At and to the north of the bank, the computed transport for spring is slightly smaller than the ADCP-derived transport above 200 m using data from spring 2000 (1.0 ± 0.2 Sv [Barth et al., 2005b]). It should be noted that the transport estimates from line 1 to 8 could contain some temporal alias, since it would take 2–2.5 days to complete a survey, and HF radar measurements show that a coastal jet can spin-up in response to wind events on shorter timescales [Kosro et al., 1997].

4.3. Seasonal Changes in the Coastal Jet

[31] With the exception of the section south of Heceta Bank (43.75°N), there is a general tendency for decreasing southward and/or increasing northward flow over the shelf during summer (summer v minus spring v is positive almost everywhere). During summer, the near surface geopotential anomaly map indicates the existence of a geostrophically adjusted negative alongshore pressure gradient south of 44.25°N. Time-dependent deviations from the geostrophic balance may contribute to decrease (increase) the southward (northward) flow via the negative alongshore pressure gradient. More details about this can be found in Barth et al. [2005a].

[32] In order to further investigate the change in the position of the coastal jet from inshore of the pinnacle of the bank during spring to offshore during summer, time series of near-surface north-south velocity and salinity from the midshelf and shelf break moorings along 44.22°N are shown in Figure 14. No rotation was applied to the velocity component, since the principal axes were different in spring and summer. Tests using those angles showed no qualitative differences. The midshelf velocity record shows that near-surface velocities are always southward until day 180, being on average −0.25 m s⁻¹. This value is smaller than the maximum averaged southward velocity computed from the ADCP vertical sections (−0.44 m s⁻¹), since the mooring was not located in the core of the jet, but on the inshore edge of it (Figure 11). From day 180 to 187, even though winds are persistently upwelling favorable, southward velocities at midshelf drop by 0.35 m s⁻¹. At the same time, there is an equivalent increase in the velocities at the shelf break mooring, indicating that the jet has moved offshore. After day 190, velocities at the shelf break decrease considerably, but velocities do not increase at midshelf, even though some persistent upwelling favorable wind events occur (e.g., day 203 to 209). The averaged velocity section (Figure 11) shows that strongest southward velocities were located between the two moorings, consistent with the weak velocities observed in the time series. The change in the location of the jet offshore occurred during a single, strong upwelling event. It is not clear, however, that the wind event is the only cause for the observed offshore shift.

Figure 14

Time series of (top) wind velocity from the NDBC station 46050 and north-south velocity (solid line) and salinity (dashed line) from the (middle) midshelf and (bottom) shelf break moorings. Depths of measurements are 13 m for velocity and 10 m (9 m) for salinity at the midshelf (shelf break) mooring. The time series were low-pass filtered (half-power point of 40 hours) to suppress tidal and inertial oscillations.

[33] The effect of this transition can also be seen in the salinity time series. Before the upwelling event on day 180, surface waters were relatively fresh and there was large variability at both moorings. The shelf break mooring shows lower salinity values, consistent with the spring averaged salinity field from the surveys (Figure 3). As the jet moves offshore around day 180, the fresher water was presumably pushed offshore, increasing the surface salinity at the mooring locations. The salinity time series present much less variability after day 190 in both moorings, consistent with the lower standard deviation observed along line 6 during the summer cruise (Figure 10).

4.4. Source of Isolated Dense Water Mass

[34] Oke et al. [2002a, 2002b] used observations and numerical model simulations to show the presence of an isolated mass of dense water, with potential density σ_θ > 26.5 kg m⁻³ over the midshelf off Newport (44.65°N, see their Figure 12 in part a). That feature was also observed during CUE-I [Huyer, 1973]. Their simulations show that the water is upwelled to the north of Newport and advected southward beneath the coastal jet. They also found a pool of dense water inshore of Heceta Bank, and numerical simulations suggest that the water is upwelled to the south of it. The two pools are not always connected, in part due to the presence of a physical barrier (Stonewall Bank, sill depth at approximately 80 m). In this section, we look for observational evidence of the presence and origin of these water pools.

[35] Mean σ_t vertical sections are shown in Figure 15. For the depth range shown, differences in σ_t and σ_θ are not significant (less than 0.003 kg m⁻³). The isolated dense water along 44.65°N can be seen during both spring and summer. It is also evident that the pool of dense water inshore of the pinnacle of the bank (lines 6 and 7) is disconnected from waters offshore with the same density in both seasons.

Figure 15

Across-shore sections of the averaged density anomaly σ_t field for (left) spring and (right) summer along (from top to bottom) lines 4, 6, and 7. Contour interval is 0.25 kg m⁻³, with bold contours every 1 kg m⁻³. The dashed (shaded) line is the 26 (26.5) kg m⁻³ contour.

[36] A map of the deepest averaged density data available is shown in Figure 16 for both spring and summer. Data is only plotted if the depth of the measurement is within 30 m from the bottom. For the majority of the data shown, measurements depth are actually within 15 m from the bottom, being even closer to the bottom along 44.48°N and 44.65°N. The tongue of dense (σ_t > 26.5 kg m⁻³) water present in both seasons in the north inshore of the 120 m isobath between 44.5° and 45.25°N is consistent with the hypothesis of the water being upwelled in the north and advected south. The pool inshore of the bank (south of 44.5°N) can be seen in both seasons. The two pools are very close together, and maps of individual surveys [O'Malley et al., 2002; Barth et al., 2003] reveal that sometimes dense water from the two regions are merged.

Figure 16

Near-bottom averaged density (kg m⁻³) field for (left) spring and (right) summer cruises. Data are only shown if depth of measurement is within 30 m from the bottom. The black line is the 26.5 kg m⁻³ contour. Shaded lines are the 100 and 200 m isobaths.

[37] In order to help clarify the source of the water inshore of the bank, averaged T-S curves from SeaSoar data are shown in Figure 17. Some deep conventional CTD data were also used to help compose the figure. Mean profiles were calculated by averaging along isopycnals. During spring, the dense water located inshore of the bank (44°N) is less spicy (as defined by Flament [2002]) than dense waters located over the shelf break and slope (conventional CTD stations along 44.22° and 45°N and offshore SeaSoar stations along 43.75°N), and has similar spiciness to shelf waters to the north (Figure 17a). This suggests that early in the upwelling season, waters to the north are an important source for the dense pool inshore of the bank through events during which dense water from the two regions are connected.

Figure 17

Averaged T-S curves for (a) spring and (b) summer cruises. Left panels show the position of each profile, which are shown in the same color in the right panels. Data used are from both SeaSoar (asterisks, solid lines) and from conventional CTD stations (triangles, dashed lines). Dotted curves sloping up to the right are lines of constant density (σ_t), and dotted curves sloping down to the right are lines of constant spiciness (Π) as defined by Flament [2002].

[38] During summer, waters inshore of the bank (44°N) have similar spiciness to waters to the south (CTD stations south of the bank, SeaSoar data along line 8), and are spicier than waters to the north, even those over the slope (Figure 17b). This suggests a stronger influence of waters upwelled to the south there. This stronger influence of southern origin waters during summer is related to the increase in the flow recirculation already observed in sections 3.1 and 3.2.

[39] The fact that the dense waters in the north and inshore of the bank are not always directly connected and can originate in different locations may have significant biological implications [Oke et al., 2002b]. For example, Wheeler et al. [2003] report a significant gradient in nutrient concentrations (nitrate, phosphate, silicate) as a function of salinity for these deep waters. Hence differing nutrient concentrations delivered to the euphotic zone farther inshore on the shelf could result in spatial differences of phytoplankton production (Figure 5).

5. Summary and Conclusions

[40] The adjustment of the flow in a region of bottom topography transition off Oregon, from a region of alongshore uniformity to a region with a shallow submarine bank, was observed using a CTD on a SeaSoar and shipboard ADCP during a late spring and a summer cruise in 2001. The ocean response to upwelling favorable winds in both regions is quite different. In the region of simple topography, the response is most intense within 30 km from the coast. This area is largely influenced by the Columbia River discharge, particularly during spring. Velocities are persistently southward, but the bottom layer is dominated by northward flow during summer, consistent with previous field experiments [Huyer et al., 1978]. Standard deviations of the alongshore velocity inshore of the 90 m isobath are nearly depth-independent during spring. The presence of Heceta Bank disturbs this pattern, moving the coastal upwelling jet offshore. The area influenced by the upwelling circulation is broader compared to north of Newport. During spring, the jet is located inshore of the pinnacle of the bank, moving offshore during summer. Mooring observations show that this transition occurs during a single strong upwelling event. As the jet moves abruptly offshore just upstream of the bank, it pushes the fresh water offshore, increasing the mean salinity over the shelf. Near-surface fields show that during summer the area under influence of the upwelled waters is wider and that the flow recirculation in the southern flank of the bank is intensified. An EOF analysis suggests that the density field is not as strongly linked to wind forcing over Heceta Bank compared with the simple topography region to the north.

[41] The combination of mean T-S diagrams and near-bottom averaged density fields suggests that dense waters found over the shelf off Newport are upwelled to the north and advected south. Inshore of the bank, another pool of dense deep water is found, with its source mostly from the south during summer. During spring, waters to the north are an important source for that pool. Waters from the two sources are not always directly connected. This is consistent with numerical results from Oke et al. [2002b], but represents the first description of the dense pool based on in situ data.

[42] Significant cross-isobath flow is found south of the bank. Averaged chlorophyll concentrations suggest that the separation of the jet is a very efficient mechanism for transport of coastal water and the material it contains off the shelf into deeper waters.

Acknowledgments

[43] We thank the officers and crew of the R/V Wecoma and the OSU Marine Technician group, who were responsible for the highly successful SeaSoar operations. We also would like to thank R. O'Malley, S. Pierce, and C. Wingard for help with data collecting and processing. G. Egbert and S. Erofeeva generously provided their tidal model of the region. The authors acknowledge M. Levine, M. Kosro, and T. Boyd for providing the mooring observations used in this study. This paper has benefited from discussions with A. Huyer and M. Levine as well as from the comments of two anonymous reviewers. This research was supported by the National Science Foundation under grant OCE-9907854. We also acknowledge support by the Brazilian Government (CNPq, Conselho Nacional de Desenvolvimento Científico e Tecnológico, grant 200147/01-3) and by the Inter American Institute for Global Change Research (IAI) through project SACC (CRN-061).