Mean structure and seasonal variability of the poleward undercurrent off southern California



[1] The magnitude, location, and extent of the California Undercurrent (CUC) off southern California is investigated using cruises of the California Cooperative Oceanic Fisheries Investigations (CalCOFI) from 1993 to 2003 which provide hydrographic, biochemical, and acoustic Doppler current profiler (ADCP) velocity data on a quarterly basis. This study is the first use of the decade-long ADCP time series; it improves on prior geostrophic calculations by providing an absolute velocity reference for estimates of currents and transport. The long-term mean reveals two undercurrent cores in the region south of Point Conception and north of Baja, California: one in the region of the continental slope within the Southern California Bight (SCB) and a second off the Santa Rosa Ridge (SRR). A single core is observed off Point Conception. Spiciness is found to be a good indicator of the presence of the CUC; however, direct velocity observations or a deep reference level are required to resolve the full strength of the CUC cores. In particular, the core off the SRR would be almost entirely missed by geostrophic calculations relative to 500 m, the maximum depth sampled by CalCOFI. The undercurrent transport off Point Conception is estimated to be about 1.7 ± 0.1 Sv, slightly less than the sum of that estimated for the SCB (1.0 ± 0.1 Sv) and that estimated off the SRR (1.1 ± 0.1 Sv) and consistent with some of the flow turning offshore at Point Conception. The CUC in the SCB is strongest in summer, while that off the SRR is strongest in fall. The CUC off Point Conception is strong in both summer and fall, reflecting the confluence of these two branches. Interannual variability is also present, and the velocity and spiciness of the CUC appear to peak during El Niño periods.

1. Introduction

[2] The California Current System (CCS) comprises a year-round surface equatorward California Current (CC) offshore of the poleward California Undercurrent (CUC) near the continental slope and a seasonal surface poleward flow near the coast in winter, in addition to upwelling-related jets and filaments and mesoscale eddies. The relative strength of the CC and CUC, together with upwelling and stratification, controls the nutrient delivery to the region and hence its productivity [Bograd et al., 2001; Di Lorenzo et al., 2005; Chhak and Di Lorenzo, 2007]. In addition, the climate of the continental regions adjacent to the California coast are strongly influenced by sea surface temperature, which depends not only on patterns of upwelling but also on the strength of the circulation and variability in the characteristics of the source waters feeding into the CC and CUC, i.e., the subarctic, central and equatorial waters of the North Pacific Ocean. The character and relative contribution of these waters is determined by a competition between the relative influence of the Aleutian Low and the North Pacific High [e.g., Bond et al., 2003; Bograd and Lynn, 2003a; Chhak and Di Lorenzo, 2007] and ENSO events [e.g., Bograd et al., 2001; Bograd and Lynn, 2003b]. Thus the strength of the CCS and mixing of the CC and CUC contribute to exchanges important in climate variability and biogeochemical transport rates.

[3] Eastern boundary current (EBC) systems such as the CCS are responsible for a large part of the world fish catch [Wooster and Reid, 1963; Hill et al., 1998], and this ecological and economic importance has motivated the longest running and most comprehensive survey of an EBC, by the California Cooperative Oceanic Fisheries Investigations (CalCOFI). The CalCOFI survey began in 1949 and has been redefined several times. From 1984 to the present, CalCOFI sampling has occurred quarterly along six lines oriented orthogonally to the coast and extending up to 700 km offshore, from San Diego to north of Point Conception (Figure 1), a distance of roughly 400 km. Although not part of the official survey, shipboard acoustic Doppler current profiler (ADCP) velocity measurements have been made on most CalCOFI cruises since October 1993 [Chereskin and Trunnell, 1996] and motivate the present study.

Figure 1.

CalCOFI cruise tracks and bathymetry of the Southern California Bight. The 5-, 400-, 500-, 2000-, and 2200-m isobaths are shown. Station numbers along the six lines are also shown. Features labeled with acronyms are Santa Rosa Island (SRI), San Nicolas Island (SNI), San Clemente Island (SCI), Santa Catalina Island (StCI), San Pedro Basin (SPB), and Santa Monica Basin (SMB). The Santa Rosa Ridge can be seen, outlined by the 400- and 500-m isobaths, extending southeast from Santa Rosa Island. The three regions analyzed in this study are labeled PtC (the region off Point Conception), SRR (the region off the SRR), and SCB (the region of the SCB excluding the Santa Barbara Channel region) and are outlined (dashed line for PtC, dotted line for SRR, and solid line for SCB).

[4] Much of what is known of the CCS off southern and central California is based on hydrography from the repeat surveys of CalCOFI. Hickey [1979, 1998] and Lynn and Simpson [1987] used these measurements to describe the complex geographic pattern of the CCS general circulation. Bograd et al. [2001] give a more detailed view of the region from Point Conception to Mexico where the southern California eddy (SCE) causes onshore geostrophic flow near the Mexican border and poleward geostrophic flow to be particularly persistent along the southern California coast [Chereskin and Niiler, 1994; Chereskin and Trunnell, 1996]. Most studies of the CUC, however, have examined the coastline north of Point Conception or south along Baja California; less is known about either the spatial or the temporal variability of the CUC off southern California and through the complex bathymetry of the Southern California Bight (SCB) consisting of several parallel channels separated by submerged ridges and islands (Figure 1).

[5] A benchmark study by Lynn and Simpson [1990] based on a single summer survey determined the location of the CUC off southern California using distributions of spiciness, dissolved oxygen and geostrophic velocity (relative to 1000 m off the SRR and 500 m in most of the SCB). They found poleward flow at depth on both sides of Santa Catalina Island and offshore of the Santa Rosa Ridge (SRR), the latter emerging from gaps in the ridge as well as flowing around its southern end. Significant poleward flow extended below 500 m depth off the SRR. They found that while some of the undercurrent water in the SCB flowed through the Santa Barbara Channel, the bulk flowed out through the Outer Passage between San Nicolas and Santa Rosa Islands. They also found that the undercurrent water flowing around the southern end of the SRR became entrained in eddies. Identification of waters of CUC origins within the SCB on the basis of their properties [Lynn and Simpson, 1987; Chereskin and Niiler, 1994] suggests multiple pathways, either via multiple cores of the CUC or via eddies. Observations of the latter off central California [Garfield et al., 1999] found scales of 15 to 30 km and suggested they are submesoscale coherent vortices. Narrow cores and small eddies will be missed by the CalCOFI station spacing (∼70 km) and underestimated by geostrophic velocities relative to 500 m, the deepest depth normally sampled at CalCOFI stations.

[6] The focus of the present study is to use the relatively unexploited direct velocity observations from the CalCOFI surveys to describe and quantify the spatial structure and variability of the CUC off southern California on seasonal time scales. We use observations from 34 shipboard ADCP surveys collected over the period 1993–2003 in conjunction with CalCOFI hydrography to analyze the flow and water properties of the CUC in three regions: off Point Conception, off the SRR and in the SCB. In the following section, the observational and theoretical understanding of the CUC is reviewed. In section 3 we discuss the data used in this study. The methods by which the CUC is quantified and its mean structure and seasonal variability are analyzed are presented in section 4. We present the results in section 5, and summarize and discuss our findings in section 6.

2. The California Undercurrent

[7] A prevalence of poleward flow near the coast is a feature common to all eastern boundary current systems [Wooster and Reid, 1963]. In the CCS, in addition to the year-round subsurface CUC, seasonal surface poleward flow occurs near the coast during winter [Hickey, 1979, 1992; Tsuchiya, 1980; Lynn and Simpson, 1987, 1990]. This nearshore flow is called the Inshore Countercurrent (ICC) or, north of Point Conception, the Davidson Current. The CUC is concentrated over the upper continental slope with strongest flow at depths of 100–300 m (but some signature to at least 1000 m), a cross-shore scale of 10–50 km, peak speeds about 30–50 cm/s and a volume transport of the order of 1 Sv [Reid, 1962; Wooster and Reid, 1963; Wooster and Jones, 1970; Hickey, 1979; Tsuchiya, 1980; Chelton, 1984; Lynn and Simpson, 1990; Collins et al., 1996; Bray et al., 1999; Pierce et al., 2000]. The CUC is observed from southern Baja California to Vancouver Island. Float trajectories during the period 1992–1995 [Garfield et al., 1999] showed the CUC to be continuous from San Francisco to Cape Mendocino (about 500 km), and shipboard ADCP measurements during the summer of 1995 indicated continuity of the CUC over the 300 km from Cape Mendocino to Cape Blanco [Pierce et al., 2000]. Elsewhere, the CUC showed limited alongshore continuity owing to offshore turning and separation from the continental slope associated with coastline irregularities [Pierce et al., 2000].

[8] The CC transports relatively cool, fresh, oxygen- and nutrient-rich water of subarctic origin equatorward. The CUC originates in the northeastern equatorial Pacific and transports relatively warm, salty, oxygen-poor and nutrient-rich water poleward [Wooster and Reid, 1963]. The pycnocline at around 100 m depth separates the relatively cool, fresh CC surface waters from the spicy (warm and salty) CUC waters [Lynn et al., 2003]. Typically, temperature and salinity gradients compensate so that horizontal density gradients are weak [Koblinsky et al., 1984], making spiciness a good tracer of isopycnal water properties since it is orthogonal to the density surfaces [Pierce et al., 2000] which are close to being surfaces of uniform depth. The distribution of properties is therefore a useful indicator of the strength and influence of the principal currents in the system. The core of the CUC has been observed around a depth of 250 m and a density of 26.6 [Lynn and Simpson, 1987].

[9] Models have identified wind stress curl and both cross-shore and alongshore pressure gradients, as well as their alongshore variability, as key elements in the dynamics of model undercurrents. Positive wind stress curl can drive a subsurface poleward flow that opposes the near-surface equatorward current via Sverdrup dynamics [Hurlburt and Thompson, 1973; Pedlosky, 1974; McCreary et al., 1987]. Using a primitive-equation ocean model (ROMS) with terrain-following coordinates, Marchesiello et al. [2003] find that the CC and CUC are consistent with Sverdrup transport owing to negative wind stress curl offshore and positive curl within 200 km of the coast. Bray et al. [1999] found that wind stress curl could explain about 75% of the poleward transport in the SCB but not the equatorward transport offshore of the bight.

[10] In the absence of wind stress curl, an alongshore pressure gradient field together with vertical mixing of heat and momentum can drive a poleward undercurrent in a stratified linear model [McCreary, 1981; Huthnance, 1984]. The pressure gradient may be due to remotely forced coastal trapped waves, sea level setup by coastal winds or density gradients. Poleward cooling leads to large-scale along-shore dynamic height gradients which drive onshore flow whose downwelling results in geostrophically balanced poleward flow [Coelho et al., 1999]. Isopycnals are found to slope upward (downward) toward the coast above (below) the undercurrent core, consistent with upwelling (downwelling) [see McCreary, 1981]. The influence of the bathymetric slope on the alongshore density gradient can result in poleward current [Coelho et al., 1999; Huthnance, 1984] through the JEBAR (Joint Effect of Baroclinicity and Relief) effect. Bathymetric roughness may also play a role, and may act to enhance rather than inhibit the undercurrent through the interaction of eddies with bathymetry [Zou and Holloway, 1996; Holloway, 1992; Haidvogel et al., 1991].

[11] Observations, obtained primarily from historical CalCOFI hydrographic data, indicate that a semiannual seasonal cycle is the dominant temporal variability of the CUC at most locations along the west coast, with a maximum of poleward flow in summer–fall and a secondary peak in winter in association with the ICC [Hickey, 1979; Chelton, 1984; Lynn and Simpson, 1987; Bray et al., 1999]. Point Sur is a notable exception, with an annual cycle that has a poleward maximum in December. Observational efforts to relate CUC variability to forcing found that, off central California, CUC variations followed the alongshore pressure gradient with phase lags of 2–3 months [Chelton, 1984], consistent with the model of Philander and Yoon [1982]. The CUC and the pressure gradient were observed to vary semiannually off Point Conception and annually off Point Sur, but the cause of the pressure gradient was unclear.

3. Data

[12] Since 1984, CalCOFI sampling has occurred quarterly along six core lines (Figure 1) spaced 74 km apart from San Diego to north of Point Conception, with 74-km along-track spacing of hydrographic stations at offshore locations and closer spacing inshore to within 15 km of the coast. Sampling at each station extends to 500 m depth in water up to 5000 m deep. Stations are designated by a line and a station number (e.g., station 77.60 is station 60 on line 77), and the 6 lines from south to north are numbered 93, 90, 87, 83, 80, and 77 (Figure 1). The data from each cruise, lasting from 15 to 20 days, will be treated as a synoptic measure of the monthly mean and will be grouped into four seasons (December–February, March–May, June–August and September–November) to investigate seasonal variability. Table 1 shows the temporal distribution of the 34 CalCOFI cruises with ADCP data used in this study (year, month and season). The temporal sampling is heavily biased toward four months: January, April, July, and October.

Table 1. Cruise Identification and Timing for Data Used in This Studya
  • a

    The cruise identification follows the convention that the two letters designate the ship (dj, David Starr Jordan; nh,New Horizon; rr, Roger Revelle), the first two numbers give the year, and the last two numbers give the month.


[13] Water samples were collected on the CalCOFI cruises and analyzed for salinity, temperature, dissolved oxygen (DO) and nutrients and are provided at 14 interpolated standard levels: 0, 10, 20, 30, 40, 50, 75, 100, 150, 200, 250, 300, 400, and 500 m [e.g., Scripps Institution of Oceanography, 2000]. We averaged the salinity, temperature, DO and nutrients to a 20- by 20-km horizontal grid at each of the depth levels and then smoothed and gridded the binned data by objective mapping, as described in the following section.

[14] ADCP measurements have been made on most CalCOFI cruises since October 1993 [Chereskin and Trunnell, 1996]. The ADCP data are not part of the core CalCOFI observations; for the most part, they were collected as data of opportunity and analyzed on an ad hoc basis. An analysis grant from the National Science Foundation supported processing the data to a uniform standard and putting them into the public domain via the Joint Archive for Shipboard ADCP data (JASADCP). This analysis is the first use of the complete ADCP data set. Typically, two cruises per year are carried out on the R/V New Horizon and two on the R/V David Starr Jordan; one cruise (January 1999) was done from the R/V Roger Revelle. In all cases, the ADCP was an RD Instruments 150 kHz narrowband unit configured to sample with an 8-m pulse and an 8-m vertical bin. Sampling interval (1-, 3- and 5-min averages), transducer depth (4–5 m), and “blank-before-transmit” (4–12 m) varied between ships and cruises. The velocities were calibrated for in situ temperature and transducer misalignment and navigated from ship relative to absolute currents using Global Positioning System (GPS) position measurements. ADCP instrumental errors are quite small, less than 1 cm/s for a 10-min average, and instrumental bias was minimized through selection of profiling parameters [Chereskin and Harding, 1993; Chereskin and Trunnell, 1996].

[15] The largest errors in absolute currents are due to errors in navigation: position and heading. During the 1990s, the Department of Defense intentionally degraded GPS positions (SA or selective availability), and position accuracy was estimated at 100 m. However, military grade receivers (P-code) were not affected by SA. SA was discontinued in May 2000. Between 1995 and 2003, P-code quality GPS was available on 26 of the cruises. P-code position accuracy is about 3 m. The uncertainty in absolute current owing to errors in ship position over a 15-min average is estimated to be 5 cm/s in the case of SA GPS measurements and less than 1 cm/s in the case of P-code measurements [Chereskin and Harris, 1997]. Heading bias errors are problematic for transport calculations, since even a small bias can result in a large transport error when integrated over long distances. Heading bias errors have been largely eliminated by GPS attitude measurements [King and Cooper, 1993; Griffiths, 1994; Chereskin and Harris, 1997], but only the most recent cruises have GPS heading corrections. An overall mean bias in heading is removed by the calibration procedure [Chereskin and Harris, 1997], and over short intervals heading errors are coherent ensemble-to-ensemble and so largely cancel in short-term estimates of horizontal shear. Comparing cumulative transport on pairs of adjacent lines, which are traversed in opposite directions a few days apart, suggests that bias effects do appear to be significant for cruises without GPS heading corrections. Transport integrated over short distances, such as the width of the undercurrent, is probably not dominated by error. Finally, the data were edited, primarily to remove data at the end of the profile contaminated by bottom interference in the SCB region where there is complex and shallow topography within range of the ADCP.

[16] In order to create a uniform time series, the ADCP data from each cruise were averaged over 15-min intervals (about 5 km along-track at a steaming speed of 5 m/s) and interpolated to fifty 8-m-depth bins, from 16 m to 408 m. The surface and deepest data are less reliable, so data from 32 m to 400 m are used here, and in investigating the undercurrent we will focus on waters below the pycnocline between 100 and 400 m. For each cruise, the subset of the ADCP data along each line were converted from latitude/longitude coordinates to a Cartesian grid in meters, with axes along and across the cruise tracks, and then averaged into 20-km bins along-track and 20-m depth bins. The cross-line component is oriented approximately 30 degrees counterclockwise relative to true north. Data from locations offset from the cruise track by greater than 5 km, owing to deviations of the ship from the line for various reasons, are excluded. The binned velocity data were then objectively mapped to a 10-km by 20-m grid in a vertical section along each line as described in the next section. Note that the cross-shore (along-track) resolution for the ADCP data is an order of magnitude better than alongshore.

4. Methods

4.1. Objective Mapping

[17] Observations were mapped using a Gaussian covariance of the form

equation image

with the normalized distance, r, given by r2 = (Δx/Lx)2 + (Δy/Ly)2, where Δx, Δy are two-dimensional separations between data and mapping locations and Lx, Ly are decorrelation length scales in these directions. For horizontal maps of hydrographic observations, we used Lx = Ly = 120 km, and the noise was set at 0.05 on the basis of the variance in the binned observations. For vertical sections of horizontal currents from the ADCP observations, Lx = 120 km (along-track distance) and Ly = 50 m (depth), and the error was set at 0.1 on the basis of the variance in the binned observations. Horizontal mapping of the ADCP data to a 20- by 20-km grid, after first binning the data to a 20-km grid, followed the approach of Chereskin and Trunnell [1996], using a two-parameter exponential covariance function of the form

equation image

where now r2 = (Δx)2 + (Δy)2 and the fitted parameters, a and b, were estimated as 120 km and 130 km by Chereskin and Trunnell [1996]. The noise level was set at 0.1, as in the vertical section mapping. The use of a stream function ensures that the velocity field is nondivergent (consistent with geostrophy) and reduces unwanted noise. Below the Ekman layer, the meridional flow is primarily in geostrophic balance. Using ADCP measurements from a single CalCOFI cruise, Chereskin and Trunnell [1996] found ageostrophic velocity to account for less than a third of measured velocities, and this fraction should decrease with increasing temporal and spatial scales and with depth. For the cruises analyzed here, there is fairly good agreement in shear between the 1993–2003 mean cross-track (along-shore) velocity observations and the mean along-shore geostrophic velocity calculated from dynamic height, relative to 500 m, on line 90 (Figure 2), but note the significant offsets at 400 m.

Figure 2.

Mean (1993–2003) vertical profiles of cross-track velocity at four locations on line 90 evaluated using geostrophic velocity determined from the hydrography (dashed line) referenced to 500 m, and from the ADCP data (solid line). Profile locations are indicated on the mean vertical section shown in Figure 3b. Positive values are poleward.

4.2. CUC Parameter Fitting

[18] In order to characterize the CUC we fit a very simple model to the subsurface (100–400 m) cross-track velocities on each line for the regions in which a mean CUC was identified in the observations. The model characterizes the CUC by a transport magnitude, width and location. The location of the hypothetical CUC along each line is determined for each cruise by identifying the location where the depth-integrated cross-track velocity (from 100 to 400 m depth) is a maximum. The width of the CUC for each cruise and on each line is then determined by increasing the width on each side of the bin of maximum depth-integrated poleward velocity so long as the transport in the added bins is greater than half this maximum. This is roughly comparable to taking the width to be defined as where the velocity is half the maximum, as in the work of Pierce et al. [2000]. The location of the model CUC for each cruise on each line is then taken as the center of this region. A least squares fitting procedure is used to estimate the long-term mean and seasonal CUC parameters and provide uncertainties by weighting the least squares using estimated observational uncertainties (Gauss-Markov estimation; see Wunsch [2006]).

4.3. Regional Averages

[19] The region off Point Conception is simpler than that off the coastline between San Diego and Santa Barbara, which has both the region off the SRR and the SCB with its channels and islands (Figure 1). In order to simplify the analysis while exploiting all available data, we have averaged the three northern lines (77, 80, and 83) and the three southern lines (87, 90, and 93) separately, treating the former as characterizing the Point Conception region and the latter as characterizing the SCB and the region off the SRR. We expect the Point Conception region to exhibit the classic single-core CUC near the continental slope, while the SCB/SRR region has been observed to have multiple CUC cores both in the SCB and off the SRR in addition to a subsurface eddy field offshore of the SRR with which the CUC interacts [Lynn and Simpson, 1990]. In the subsequent analysis, we average over the three lines in each of the two regions, and then examine velocity and hydrographic properties as a function of distance offshore from either the tip of Point Conception where line 80 ends (station 80.50), or off Los Angeles where line 90 ends (station 90.30). As a result of this averaging, the delineation of the CUC will be blurred somewhat but a correlation between the three lines will be emphasized so that features characterizing the entire region can be analyzed.

5. Results

5.1. The 1993–2003 Mean

[20] The 10-year-mean alongshore velocity averaged over the northern region reveals a single core of poleward subsurface flow off Point Conception, centered on station 60 (Figure 3a), that we identify as the CUC. Averaged over the southern region, two CUC cores of poleward subsurface flow are evident in the long-term mean: one offshore of the SRR and centered on station 65 and the other within the SCB and centered on station 40 (Figure 3b). The poleward flow in the SCB appears to extend all the way to the surface (Figure 3b); however, near the coast only the subsurface flow is poleward year-round as the surface flow reverses in spring (see Figure 11b). The core of poleward flow off Point Conception and the core within the SCB are both considerably sheared in the vertical, in marked contrast to the core offshore of the SRR. If one assumed a level of no motion at 500 m (the usual reference for CalCOFI), the poleward flow offshore of the SRR would be almost entirely missed because it extends to 500 m and is weakly sheared. Instead there would appear stronger equatorward flow at the surface and negligible flow beneath. This core requires the direct velocity observations or the use of a deep (>1000 m) reference level for its detection, though its presence has been inferred from property distributions (high temperature and salinity and low dissolved oxygen). Other features of the CCS are also apparent in the long-term mean sections. Off Point Conception the inshore edge of the equatorward CC lies above the CUC at station 60; offshore of station 60 the CC appears as a broad, 200-km-wide band of equatorward flow in the upper 200 m (Figure 3a). Further south, offshore of the SRR, there are 2 equatorward jets of the CC, one centered on station 60 and the other on station 95 (Figure 3b). In the SCB, there is an equatorward surface jet against the coast, likely related to coastal upwelling.

Figure 3.

The 1993–2003 mean (a) and (b) cross-track velocity and (c) and (d) spiciness averaged over the northern lines (77, 80, and 83) off Point Conception (Figures 3a and 3c) and over the southern lines (87, 90, and 93) off the SRR and in the SCB (Figures 3b and 3d). Also shown are the 25.0, 26.2, 26.6, and 26.8 isopycnals. The vertical dashed line in Figure 3b indicates the location of the SRR. The top axis in Figures 3a and 3b is labeled with CalCOFI station numbers, and the inverted triangles along the top of Figure 3b mark the locations of the profiles shown in Figure 2. The vertical axes are depth (m), and the horizontal axes are distance offshore (km).

[21] So far we have identified CUC currents as cores of poleward flow at depths between 100 m and 400 m located near the SRR or the coast, but for unambiguous identification we need to consider water mass properties as well. On the basis of the 3 core locations, we identified 3 subregions for fitting model parameters and calculating average water properties. The region denoted PtC (Figure 1) is bounded by stations 70 and 52 on lines 77, 80 and 83 and is expected to be similar to the CCS further north. The region denoted SRR (Figure 1) is bounded by stations 70 and 52 on lines 87, 90 and 93 and is expected to be in the SCE and to be influenced by flow through the gaps in the SRR to the south of San Nicolas Island (Figure 1). The region denoted SCB (Figure 1) lies inshore of station 52 on lines 87, 90 and 93. This region notably excludes the Santa Barbara Channel, which is more strongly affected by the influence of Point Conception on the wind field [Hsu et al., 2007] but is restrictive in its capacity to carry the CUC owing to a 200-m-deep sill at the eastern end. Distance offshore is measured relative to the inshore end of line 80 (station 80.50) for the northern three lines, and the inshore end of line 90 (station 90.30) for the southern three lines. The depth range is restricted to 100 to 400 m.

[22] Fitting a model CUC to the observations as described in section 4.2, we estimate the CUC transport off Point Conception to be about 1.7 ± 0.1 Sv (Table 2). This transport is slightly less than the sum of that estimated for the SCB (1.0 ± 0.1 Sv, in agreement with the transport estimates for the SCB of Bray et al. [1999]) and that estimated off the SRR (1.1 ± 0.1 Sv), consistent with some of the flow turning offshore at Point Conception. The model mean CUC width is slightly narrower (54 ± 3 km) in the SCB and off Point Conception (61 ± 3 km) than off the SRR (89 ± 4 km). The CUC is located about 60 ± 10 km offshore of both the SRR and Point Conception as well as off the southern California coastline in the SCB. The mean alongshore velocity of the CUC off Point Conception (between 100 and 400 m) is about 56% larger than the mean CUC velocity in the SCB and off SRR. The transport and location estimates of the model are consistent with other estimates [e.g., Hickey, 1998]; however, the CUC widths fit from the model are somewhat wider, especially off the SRR. Collins et al. [2003] suggest that the midstream, maximum velocity, part of the CUC does not exceed 20 km. Hickey [1998] gives the CUC width as 10–40 km but finds that seaward of the shelves in the CCS, filament jet widths (down to deeper than 200 m) are 50–75 km. Huyer et al. [1989] suggest the inshore poleward flow in the SCB can be as wide as 100 km but further north it is only 10–20 km wide. Because the data to which the model CUC is fit were binned to 20 km and mapped using a 120-km decorrelation scale, and because times of weaker (and broader) poleward flow are included, it is reasonable to expect that our width estimates should be biased high. The core in the SRR region is the widest, and it is possible that offshore features associated with eddies, which should not be treated as part of the CUC, are being included more in this region than off Point Conception.

Table 2. The 1993–2003 Mean Subsurface Property Distributions and Velocity Components Averaged Over the Subregions Where the CUC Is Observed for Each Cruise and Depths of Maximum Velocity and Maximum Spicinessa
PropertySCBSRRPoint Conception
  • a

    Mean subsurface of 100–400 m depth; subregions are PtC, SCB, and SRR (see Figure 1). Also shown are the estimated transport, width, and location (distance offshore relative to the inshore ends of lines 80 for Point Conception, and 90 for SRR and SCB) of the CUC. Uncertainties shown are standard errors.

Salinity (psu)34.12 ± 0.0134.04 ± 0.0134.08 ± 0.01
Temperature (deg C)8.78 ± 0.058.31 ± 0.048.25 ± 0.04
Dissolved oxygen (mg/L)1.77 ± 0.032.12 ± 0.041.82 ± 0.04
Phosphate (μM/L)2.38 ± 0.012.30 ± 0.022.42 ± 0.01
Alongshore velocity (cm/s)3.9 ± 0.43.9 ± 0.36.1 ± 0.5
Poleward velocity (cm/s)4.5 ± 0.34.7 ± 0.27.3 ± 0.5
Cross-shore velocity (cm/s)−0.1 ± 0.5−0.7 ± 0.4−1.5 ± 0.6
Depth of maximum velocity (m)215 ± 12238 ± 10194 ± 10
Depth of maximum spiciness (m)202 ± 5183 ± 5174 ± 4
CUC transport (Sv)1.0 ± 0.11.1 ± 0.11.7 ± 0.1
CUC width (km)54 ± 389 ± 461 ± 3
CUC location (km)60 ± 10220 ± 1060 ± 10

[23] In Table 2 we summarize the mean hydrographic and biochemical properties of the CUC in the three regions, evaluated for each cruise over the subregions in which the CUC is found to lie on the basis of the model parameters (location and width) for each cruise. Lynn and Simpson [1987] and Chereskin and Niiler [1994] identified the presence of the CUC water on either side of the SRR by mapping spiciness anomalies in several isopycnal ranges. For the location of the model CUC in the three subregions, we examined the distribution of alongshore velocity versus spiciness on the 26.60 ± 0.05 isopycnal (Figure 4). There is no significant trend for any of the three regions shown in Figure 4 as would be expected if the region analyzed contained other waters than those of the CUC, in which case weaker poleward velocity might be associated with the presence of filaments of less spicy CC water, and we do indeed find a uniformly spicy water mass compared with waters further offshore where the CC is observed. The mean spiciness at the core of the CUC on the 26.6 isopycnal is 0.29 ± 0.10 in the SCB, 0.14 ± 0.08 off the SRR and 0.18 ± 0.08 off Point Conception compared with a mean spiciness of 0.00 ± 0.03 in offshore waters around the 26.6 isopycnal.

Figure 4.

The 1993–2003 mean velocity in the region of the CUC estimated for each cruise as a function of spiciness for subsurface waters (100–400 m) with density (sigma-theta) in the range 26.55 to 26.65. The dark lines indicate the mean spiciness and mean velocity.

[24] The mean velocities in Figure 4 are less than those in Table 2 as a result of being restricted to the 26.6 isopycnal for comparison to Lynn and Simpson [1987]. During 1993–2003, however, this isopycnal lay at about 300 m in the region where the CUC is found (Figure 3) whereas Lynn and Simpson [1987] found this isopycnal to lie closer to 250 m. In Table 2 we show the depth of maximum velocity in each of the three regions, evaluated over the region where the CUC is found for each cruise, and we also show the depth of maximum spiciness; in all cases these maxima occur at depths shallower than 300 m, corresponding to a density between 26.4 and 26.5. The depths of the velocity and spiciness maxima are similar in the SCB and off Point Conception with the velocity peak slightly deeper, but off the SRR the peak velocity is considerably deeper than the peak spiciness perhaps reflecting the influence of the SCE as well as mesoscale eddies generated near the SRR. Lynn and Simpson [1987] note that the peak velocity of the CUC lies close to the 26.6 isopycnal in summer but can lie deeper in other seasons; however, in the 1993–2003 data, summer is not consistently the season in which the CUC is shallowest (Table 3). PtC has the largest range of alongshore velocity in this density range, with peak poleward values around 30 cm/s (Figure 4a), in agreement with peak values observed elsewhere [Reid, 1962; Wooster and Jones, 1970]. The mean spiciness values are similar for SRR and PtC with the slightly higher values off Point Conception likely owing to being derived from a mixture of CUC waters from the SCB and off the SRR. The spicier water within the SCB may be the result of reduced influence of offshore waters of the CC as well as a greater impact of heating and evaporation.

Table 3. Statistics of Salinity, Temperature, Dissolved Oxygen, Phosphate, Along- and Cross-Track Velocity, and Poleward Velocity for Waters Between 100 and 400 m Depth Averaged Over Subregions Where the CUC Is Observed for Each Cruise and Depths of Maximum Velocity and Maximum Spicinessa
 S (psu)T (deg C)DO (mg/L)PO4 (μM/L)u (cm/s)v (cm/s)vpoleward (cm/s)Depth of Maximum Velocity (m)Depth of Maximum Spiciness (m)
  • a

    Statistics are seasonal means and standard errors. Poleward velocity is positive cross-track velocity. Subregions are PtC, SCB, and SRR; see Figure 1).

Winter34.09 ± 0.019.00 ± 0.201.96 ± 0.052.29 ± 0.03−1.1 ± 0.64.0 ± 0.64.7 ± 0.5290 ± 16176 ± 12
Spring34.15 ± 0.018.39 ± 0.031.62 ± 0.052.47 ± 0.021.7 ± 0.71.7 ± 0.32.7 ± 0.3245 ± 13187 ± 10
Summer34.16 ± 0.018.75 ± 0.071.60 ± 0.042.45 ± 0.01−0.4 ± 0.95.3 ± 0.75.8 ± 0.6196 ± 18223 ± 11
Fall34.08 ± 0.018.95 ± 0.071.90 ± 0.052.32 ± 0.020.1 ± 1.14.1 ± 0.64.6 ± 0.5170 ± 13210 ± 8
Winter34.03 ± 0.028.64 ± 0.122.17 ± 0.102.25 ± 0.04−1.8 ± 0.62.8 ± 0.53.9 ± 0.3216 ± 19177 ± 10
Spring34.01 ± 0.028.47 ± 0.102.27 ± 0.082.23 ± 0.03−0.2 ± 1.12.5 ± 0.53.8 ± 0.4264 ± 22181 ± 11
Summer34.10 ± 0.018.20 ± 0.101.78 ± 0.042.44 ± 0.02−1.2 ± 0.74.2 ± 0.54.6 ± 0.4245 ± 21161 ± 8
Fall34.04 ± 0.018.41 ± 0.062.11 ± 0.042.30 ± 0.020.1 ± 0.85.6 ± 0.66.1 ± 0.5215 ± 19195 ± 8
Point Conception
Winter34.03 ± 0.028.32 ± 0.092.09 ± 0.112.31 ± 0.04−1.5 ± 1.04.3 ± 0.75.7 ± 0.8200 ± 21161 ± 11
Spring34.10 ± 0.017.95 ± 0.051.65 ± 0.042.51 ± 0.02−1.6 ± 1.34.7 ± 0.95.8 ± 0.8207 ± 21165 ± 6
Summer34.12 ± 0.018.26 ± 0.051.64 ± 0.032.48 ± 0.01−1.7 ± 1.37.7 ± 0.88.7 ± 0.8173 ± 17168 ± 8
Fall34.05 ± 0.018.44 ± 0.061.92 ± 0.042.37 ± 0.02−1.2 ± 1.07.0 ± 1.08.1 ± 1.0201 ± 18195 ± 9

[25] Vertical sections of spiciness over the northern and southern lines (Figures 3c and 3d, respectively) confirm that the region of the CUC is anomalously spicy, with a peak spiciness around 200 m and a spiciness minimum between 100 and 150 m. Clearly our choice to examine the distribution of spiciness for waters below 100 m, as discussed in the previous paragraph, excludes those waters whose spiciness is impacted by surface or Ekman layer properties, but does include waters within the spiciness minimum. This minimum, which is stronger in the SRR/SCB region (Figure 3d) is likely due to a combination of the influence of the CC and the decrease in temperature with depth. Below this the decrease in spiciness is offset by the upper boundary of the core of the spicy CUC waters. The offshore extent of the spicy peak indicating the presence of the CUC coincides nicely with the regions of poleward flow in Figures 3a and 3b. Several isopycnals are also shown in Figures 3c and 3d. The isopycnals are not orthogonal to the spiciness contours since these are averages over a number of cruises. The isopycnals diverge shoreward and slope upward toward the coast above the CUC core and downward toward the coast at depth, consistent with McCreary [1981], over the region where the CUC is found as is expected for geostrophically balanced subsurface poleward flow, with more complicated structure in the SCB. Again, the core of the CUC in the three regions is shallower than the 26.6 isopycnal, suggesting that the waters have become less dense since the observations analyzed by Lynn and Simpson [1987].

[26] The first mode of an EOF analysis (Figure 5) of the cross-track velocity inshore of station 70 on line 80 (off Point Conception) and line 90 (in the SCB and off the SRR) for subsurface waters (100–400 m) accounts for 28% of the variance off Point Conception (Figure 5g), 26% off the SRR (Figure 5d) and 22% in the SCB (Figure 5a). Off Point Conception and the SRR, the first mode is fairly uniform with depth (Figure 6) but slightly stronger near the surface, and is tilted with the core of the CUC occurring at greater depths further offshore. This mode is also uniform in the cross-shore direction except that at Point Conception there is a band of weak flow in the opposite direction below 150 m within about 30 km of the coast (Figure 5g). In the SCB the first mode has uniform flow between 100 and 200 m across the whole width of the bight below which the strongest subsurface flow is in the central 70 km of the SCB (Figure 5a). The first mode is slightly more sheared and surface intensified in the offshore region of the SCB (Figure 6c). The width of this mode of the CUC is about 70 km in all three regions with its core about this same distance offshore from either the coast or the SRR. The variability of this mode (the principal components are shown in Figure 7) appears to represent the strength of the mean CUC in the regions analyzed.

Figure 5.

EOFs of cross-track velocity within 150 km off Point Conception on line 80, within 150 km off the SRR for line 90, and in the SCB on line 90, for subsurface waters between 100 and 400 m depth. The origin of the x axis (0 km) corresponds to station 90.30 (SCB), station 90.50 (SRR), and station 80.50 (PtC). Shown in the title for the image is the variance explained in each mode. Positive (negative) values are red (blue).

Figure 6.

Vertical structure of the first (dark line), second (dash-dot line), and third (thin line) EOF modes of the cross-track velocity (evaluated over the regions shown in Figure 5) at locations around (a) 100 km, (d) 60 km, and (g) 40 km offshore of Point Conception on line 80; (b) 100 km, (e) 60 km, and (h) 40 km off the SRR on line 90; (c) 100 km, (f) 60 km, and (i) 40 km offshore of the southern California coastline on line 90.

Figure 7.

Smoothed time series (1.25-year box-car filter) of principal components of first-mode (a) EOF, (b) spiciness, and (c) velocity in the three regions. Gray line, SCB; dashed line, Point Conception; dotted line, SRR.

[27] The second EOF mode in all three regions, accounting for 17% of the variance off Point Conception (Figure 5h), 16% off the SRR (Figure 5e) and only 14% in the SCB (Figure 5b), consists of flow in one direction within 70 km of the coast (or ridge) and in the other direction between 70 and 150 km. This mode, whose variability appears to represent variability in the offshore location of the CUC, has subsurface maxima off Point Conception (Figure 5h) and in the SCB (Figure 5b) but is surface-intensified off the SRR (Figure 5e). The third mode appears to capture some of the variability of the vertical structure of subsurface along-shore flow in the regions analyzed. The relatively low variance explained by the first three modes in the SCB (46% compared with 55% off Point Conception and 50% off the SRR) suggests a greater variability of flow patterns in the SCB.

[28] A smoothed time series of the principal components associated with the first-mode EOF in the three regions is shown in Figure 7, as well as smoothed time series of mean spiciness and along-shore velocity in these regions. The mean spiciness and velocity have been evaluated between 100 m and 400 m in each of the three regions. There is a clear correlation between all three variables with peaks corresponding with El Niño periods, which confirms that the first-mode EOF relates to the strength of the CUC. The smoothed time series of PCs of the second-mode EOF (not shown) show a slight correlation with the smoothed time series of distance offshore for the three regions, supporting the idea that the second-mode represents the offshore location of the CUC.

[29] Nondivergent horizontal mapping of the long-term mean horizontal flow field from the ADCP data at 250 m and 200 m (Figures 8 and 9, respectively), around the core depth of the poleward undercurrent, shows that the CUC flows primarily through the SCB between San Clemente Island and Santa Catalina Island (Figure 1), then passes near the coast north of Los Angeles through the Santa Monica Basin inshore of a ridge extending into the SCB from the eastern end of the Santa Barbara Channel before flowing out of the SCB through the Outer Passage and then around Point Conception. There is also weak poleward flow throughout the Santa Barbara Channel. In association with a counterclockwise eddy centered south of San Clemente Island, there is weak equatorward flow inshore of the SRR, and the CUC off the SRR is kept offshore near the 2000-m isobath. After the inner branch of the CUC flows out of the Outer Passage and the Santa Barbara Channel and joins this outer branch, the CUC becomes stronger, additionally augmented by onshore flow at the southern edge of a large (∼200 km) counterclockwise eddy centered southwest of Point Conception. Some of the CUC turns offshore past Point Conception to recirculate in this eddy. Some of the flow out of the Outer Passage turns offshore southeast of this eddy after merging with the branch of the CUC off the SRR, subsequently meandering south in the CC. Figures 8 and 9 also show the mean spiciness distribution at each depth and confirm that the path of the CUC is distinguished by anomalously high spiciness, especially in the SCB and more so off Point Conception than off the SRR.

Figure 8.

The 1993–2003 mean nondivergent velocity at 250 m depth, mapped from the ADCP observations using a stream function. The scale for velocity is shown by the arrow at the upper right. The color field represents spiciness. Rectangles indicate the three regions shown in Figure 1.

Figure 9.

The 1993–2003 mean nondivergent velocity at 200 m depth, mapped from the ADCP observations using a stream function. The scale for velocity is shown by the arrow at the upper right. The color field represents spiciness. Rectangles indicate the three regions shown in Figure 1.

5.2. Seasonal Variability

[30] Off Point Conception (Figure 10) the CUC is strongest and most uniform with depth in summer (Figure 10c). Above 100 m, equatorward flow (the CC) lies above the offshore half of the CUC while the strongest poleward flow is near the coast between 50 m and 200 m. There is negligible or weak equatorward flow near the coast below 220 m. The CUC off Point Conception in fall (Figure 10d) is similar to that in summer but weaker and slightly broader at depth while the stronger poleward flow at the coast has shoaled and appears to extend to the surface (the ICC). Winter (Figure 10a) has much weaker poleward flow centered about 60 km offshore between about 160 m and 360 m with a peak around 300 m, and slightly increased flow above this shoaling inshore. There is equatorward flow against the coast below about 200 m, increasing in width to about 40 km at 220 m and then increasing further below 300 m. The equatorward flow above the offshore edge of the CUC off Point Conception remains throughout the year but in spring (Figure 10b) reaches all the way to the coast down to about 50 m. Below this is a weak poleward current between 100 m and 400 m, confined to a narrow region near the coast but extending further offshore with depth to a maximum width of about 60 km at 300 m. Below this it becomes weaker and separates from the coast. This springtime CUC off Point Conception has a very different character from the other seasons and is similar to model undercurrents related to coastal along-shore winds [McCreary et al., 1987].

Figure 10.

(a–d) Seasonal cross-track velocity and (e–h) spiciness off Point Conception (averaged over lines 77, 80, and 83). Also shown are the 25.0, 26.2, 26.6, and 26.8 isopycnals. These sections are the seasonal means that correspond to the long-term shown in Figures 3a and 3c. The vertical axes are depth (m), and the horizontal axes are distance offshore (km).

[31] The CUC in the SCB (Figure 11) is strongest in summer (Figure 11c) and fairly uniform throughout the upper 400 m but does diminish somewhat at 400 m. A band of more intense poleward flow above 70 m lies just inshore of the SRR, seaward of the core of the CUC, while a narrow (20 km) band of equatorward flow lies above 70 m along the southern California coastline. Offshore of the SRR in summer, a slightly weaker CUC of uniform width (∼100 km) between 120 m and 400 m, lies below the inshore branch of the CC. The CUC on either side of the SRR is about 100 km wide but in the SCB it widens toward the surface to fill the entire width of the bight between 100 m and 200 m. By fall (Figure 11d), the CUC in the SCB has weakened considerably and moved offshore at depth so that the inshore half below 200 m is replaced with weak equatorward flow increasing in width down to 400 m. In the inner bight, a thin band of equatorward flow lies above a narrow poleward current between 60 m and 200 m which appears to be separate from the main part of the CUC in the central SCB. The CUC off the SRR is still strong and extends above 50 m. Except for a thin band of negligible or weak equatorward flow at the SRR, the inshore branch of the CC has disappeared or moved offshore to merge with the main part of the CC. In winter (Figure 11a) the CUC off the SRR has weakened considerably and the inshore branch of the CC has returned, extending below 150 m. In the SCB, a narrow and weak CUC extends down to 300 m just inshore of the SRR, with a stronger surface-intensified poleward jet in the upper 100 m slightly closer to the SRR. This surface poleward jet inshore of the SRR disappears by spring (Figure 11b) and a weak CUC has reappeared both inshore and offshore of the SRR with the offshore branch lying between (and below) the two branches of the CC. In the immediate vicinity of the SRR there is equatorward flow in spring at intermediate depths.

Figure 11.

(a–d) Seasonal cross-track velocity and (e–h) spiciness in the SCB and off the SRR (averaged over lines 87, 90, and 93). Also shown are the 25.0, 26.2, 26.6, and 26.8 isopycnals. These sections are the seasonal means that correspond to the long term shown in Figures 3b and 3d. The vertical dashed lines in Figures 11a–11d indicate the location of the SRR. The vertical axes are depth (m), and the horizontal axes are distance offshore (km).

[32] Evaluating seasonal averages of the cross-track velocity over the cruise-by-cruise location of the CUC (Tables 3 and 4) we see that the mean alongshore velocity in the CUC peaks in summer in the SCB and off Point Conception but peaks in fall off the SRR. Summer and fall peaks off Point Conception are roughly equal, suggesting that the summer peak in the SCB and the fall peak off the SRR both appear off Point Conception. Poleward velocity remains fairly strong throughout the year off Point Conception but weakens considerably in spring in the SCB and in both winter and spring off the SRR. Table 3 also gives mean hydrographic properties for the three regions, evaluated for each cruise over the region where the CUC is found and subsequently averaged in each season, as well as the depths of maximum velocity and maximum spiciness. Off SRR and Point Conception, the peak spiciness is deepest in fall and is consistently shallower than the peak velocity which is deepest in spring, likely owing to the influence of upwelling on upper-layer circulation. In the SCB the peak spiciness becomes deeper than the peak velocity in summer and fall, perhaps reflecting the influence of the SCE on surface waters in the SCB, as well as the influence of the SRR on isolating and retaining the deeper, spicier waters brought into the SCB by the CUC. The peak velocity is deepest in winter, likely reflecting the earlier influence of upwelling in the SCB than off the SRR or off Point Conception.

Table 4. Transport, Width, and Distance Offshore of the Model CUC in the SCB, off the SRR, and off Point Conception
 Transport (Sv)Width (km)Distance Offshore (km)
Winter0.6 ± 0.153 ± 560 ± 20
Spring0.8 ± 0.152 ± 550 ± 20
Summer1.7 ± 0.271 ± 550 ± 20
Fall0.7 ± 0.152 ± 570 ± 20
Winter0.3 ± 0.143 ± 6220 ± 20
Spring0.6 ± 0.149 ± 5250 ± 20
Summer1.2 ± 0.263 ± 5210 ± 20
Fall1.8 ± 0.272 ± 6220 ± 20
Point Conception
Winter1.3 ± 0.163 ± 650 ± 20
Spring1.0 ± 0.139 ± 420 ± 20
Summer3.0 ± 0.277 ± 540 ± 20
Fall2.0 ± 0.267 ± 640 ± 20

[33] The seasonal variability of the CUC should also be evident in the distributions of spiciness. Vertical sections of spiciness averaged over the northern three lines (77, 80 and 83) are shown in Figures 10e10h and comparable sections are shown for the southern lines (87, 90 and 93) in Figures 11e11h. The increase in summer in the region where the CUC is found is evident for both the Point Conception and SRR/SCB sections. The increased slope of the constant spiciness lines in summer (compared with fall) for the Point Conception region agrees with the greater depth of the CUC in this season. The influence of upwelling-related equatorward flow can be seen by comparing the winter and spring 3.5 spiciness contours (Figures 10e and 10f) which slope upward toward the coast in winter but reverse in spring, gradually leveling off through the rest of the year. A similar reversal of the slope of the upper 3.5 spiciness contour between winter and spring can be seen in the SRR/SCB section (Figures 11e11h) but only inshore of the SRR. Spiciness contours in the SCB are again more vertical in summer than fall reflecting the increased strength and depth in this season. The increased strength in fall of the CUC off the SRR is not readily evident in the spiciness plots. Several isopycnals are also shown in Figures 10e10h and 11e11h. The isopycnals diverge toward the coast, in the region where the CUC is found, most strongly in those seasons when the CUC is strongest and again there is more complicated structure in the SCB. Generally, the regions of poleward (equatorward) flow coincide with isopycnals that slope downward (upward) toward the coast and diverge toward the coast more (less) strongly.

[34] EOF analysis for the data in each season (not shown) gives first modes with similar structure to the 1993–2003 mean (see previous subsection) with variances explained which peak in summer off Point Conception at 51% and in the SCB at 37%, but peak in fall off the SRR at 38%. The variance explained in the second mode peaks in summer in the SCB at 26%, so that the variance explained by the first two modes in the SCB in summer is 53%. The variance explained by the second mode off the SRR also peaks in summer at 26% but that off Point Conception peaks in winter at 21%.

[35] The seasonal mean patterns derived from nondivergent mapping of the horizontal flow field from the ADCP data at 250 m and 200 m are shown in Figures 12 and 13, respectively. There is only slight difference between the fields at 250 and 200 m and the following discussion of Figure 12 applies to Figure 13 as well. There is a preference in summer (Figure 12c) for bifurcation off San Diego with the offshore branch flowing around the southern end of the SRR and then meandering poleward, while the inshore branch flows poleward through the SCB. Offshore, there are two dominant clockwise eddies, a larger one off Point Conception and a smaller one off the SRR, while a counterclockwise eddy lies south of San Clemente Island off San Diego. In fall (Figure 12d) the CUC off the SRR is straighter and stronger and comes more directly from the south. In fall there is also the greatest likelihood of flow out of the gap in the SRR to the south of San Nicolas Island, onshore flow seaward of the Outer Passage and flow across the Outer Passage toward the Channel Islands rather than directly out of the channel. The counterclockwise eddy off San Clemente Island appears most frequently in fall, as does a clockwise eddy off Los Angeles. In winter (Figure 12a) only a weak poleward flow out of the Outer Passage and around Point Conception remains. The three eddies observed in fall are weaker and more separated, with the flow out of the Outer Passage branching, some flowing offshore along the northern edge of the central eddy and the rest flowing past Point Conception. Equatorward flow occurs in winter and spring (Figure 12b) over much of the region, with poleward flow confined to the inshore region of the SCB and a clockwise eddy in the Santa Barbara Channel. In spring (Figure 12b) there is flow into the Outer Passage, counterclockwise eddies south and west of Point Conception, onshore flow off San Diego and a counterclockwise eddy off Los Angeles.

Figure 12.

Seasonal mean nondivergent velocity field at 250 m depth, mapped from the ADCP observations using a stream function. The scale for velocity is shown by the arrow at the upper right. The color field represents spiciness.

Figure 13.

Seasonal mean nondivergent velocity field at 200 m depth, mapped from the ADCP observations using a stream function. The scale for velocity is shown by the arrow at the upper right. The color field represents spiciness.

[36] Figures 12 and 13 also show spiciness and support the idea that the CUC has a spiciness signature. Spiciness is least in spring and maximum in summer in the SCB but maximum in fall off the SRR, consistent with our observations regarding poleward velocity. There is evidence of spicier water in anticyclonic eddies suggesting that these may originate in the region of the CUC and subsequently move offshore. Spicier water appears to be advected around the periphery of cyclonic eddies. The spiciness in the SCB is closer to the SRR in fall and winter than in spring and summer, reflecting the equatorward flow along the SRR in spring. The offshore flow through the SCB and Outer Passage in summer is seen in the spiciness maximum moving closer to the SRR along its northern section and the dramatically increased spiciness off Point Conception in summer. It appears that there is strong recirculation of anomalously spicy water in the SCB in spring which is subsequently released to flow past Point Conception in summer and fall.

6. Discussion and Summary

[37] In this paper we have used the ADCP data collected on 34 CalCOFI cruises from fall, 1993 to fall, 2003, together with hydrographic data, to characterize the long-term mean structure of the CUC off southern California and its seasonal variability. We find that the CUC has two branches off southern California which sometimes result from a bifurcation off San Diego (more frequently in summer than fall) and which merge to flow past Point Conception, with both branches carrying about 1 Sv but some turning offshore at Point Conception so that slightly less than the sum of the two branches flows past the point. The CUC is strongest in summer in the SCB and in fall off the SRR, with the SCB carrying 1.7 ± 0.2 Sv at its peak while the peak CUC off Point Conception (in summer) carries 3.0 ± 0.2 Sv. The fall peak off the SRR carries 1.8 ± 0.2 Sv while 2.0 ± 0.2 Sv flow past Point Conception in that season. The CUC off the SRR extends deeper (below 400 m) than off Point Conception or in the SCB, which may explain why it has not been observed in geostrophic velocity sections deduced from the hydrography with the assumption of zero flow at 500 m. Figure 2 shows this for line 90, where the mean ADCP cross-track velocity at 400 m is about 2 cm/s off the SRR. The seasonal peaks in poleward velocity (Table 3) for the three regions parallel the transport variability, with the poleward flow in the SCB (5.8 ± 0.6 cm/s) and PtC (8.7 ± 0.8 cm/s) peaking in summer and the SRR (6.1 ± 0.5 cm/s) peaking in fall. The poleward flow off PtC in fall (8.1 ± 1.0 cm/s) is almost as large as in summer.

[38] The equatorial character of the CUC is evident in its temperature and salinity characteristics. The core of the CUC (maxima in spiciness and poleward velocity) lies on a shallower density, between 26.4 and 26.5, relative to the 26.6 finding of Lynn and Simpson [1987], and the depth range is somewhat shallower, 200–250 m. This is likely due to differences in the time periods analyzed since the upper ocean in this region has experienced a warming trend [Roemmich, 1992]. The range in CUC salinity over all regions and seasons is [34.01–34.16] and the range in temperature is [7.95–9.0] C (Table 3). The depths of peak spiciness in the region of the CUC are around 20 m shallower than depths of peak currents off Pt Conception and in the SCB but off the SRR, the deeper peak velocity is not reflected in the depth of peak spiciness with the depths separated by about 50 m.

[39] Multiple eddies observed offshore of the SRR and Point Conception may exert a strong influence on the CUC. Davis [1985] found that mesoscale eddies off northern California contributed to the along-shore subsurface momentum balance, and Holloway [1992] suggested that undercurrents near the continental slope can be generated by the interaction of mesoscale eddies with bottom topography. Mesoscale eddies may also have some influence on the width or location of along-shore currents driven by large-scale forcing such as along-shore pressure gradients, though these may also be trapped to the bathymetry as with the CUC being found principally at the continental slope. The eddy which appears offshore of the SRR, similar to “eddy E” observed by Lynn and Simpson [1990] could influence the recirculation of the CC through the SCB, contributing significantly to the SCE. The eddy inshore of the SRR, extending to just south of San Clemente Island, could influence the path of the inner branch of the CUC which flows inshore of San Clemente Island.

[40] An interesting feature of the CUC apparent in Figures 2b2d and Figures 3a3b (but less apparent in the EOF analysis (Figures 5 and 6) or the seasonal sections (Figures 10 and 11)) is a subsurface minimum in the alongshore velocity around 300 m depth, below the undercurrent core. We do not have an immediate explanation for this minimum. It may provide evidence for the competition between various mechanisms forcing the CUC, and we hope to look into this in further studies.

[41] This analysis is the first use of a newly available set of direct velocity observations for the southern California Current System. Our results clearly show that geostrophic calculations referenced to 500 m in the southern California Current (and likely in the entire CCS) are inadequate for estimating the current speed, transport and even at times the direction of the flow. There is much more that can be done to utilize the ADCP data; in particular, we plan to extend the present study to investigate the dynamics and interannual variability of the CUC.


[42] This study would not have been possible without the dedication and hard work of the scientists, technicians, and volunteers who participate in the CalCOFI surveys. Sharon Escher reprocessed and edited the ADCP data used in this study. Comments from Emanuele Di Lorenzo and an anonymous reviewer greatly improved the manuscript. We gratefully acknowledge support from the National Science Foundation (OCE-0324360) and Scripps Institution of Oceanography. A Web atlas of the CalCOFI ADCP data set can be found at The ADCP data are available from the National Oceanic Data Center Joint Archive for Shipboard ADCP,