This study presents observations of the cross-sectional structure of resonant response to sea/land breezes (SLBs) off Huntington Beach (HB) in the Southern California Bight (SCB). A resonant response to local diurnal wind stress fluctuations associated with SLB forcing occurs intermittently and produces strong diurnal oscillations of flow and temperature resulting from enhanced work of the diurnal local wind on the sea surface. At nighttime (daytime), a coherent cross-sectional circulation with offshore (onshore) currents in the surface layer (upper 15 m) and onshore (offshore) currents in the intermediate layer around 20 m are generated, with a three-layered vertical structure on the outer shelf. The authors find a net cross-shore eddy heat flux (net cooling of nearshore water) during the period of strong response to SLB, that is, a rectified mean heat flux and steeper isotherms resulting from the diurnal SLB fluctuations. The steepened mean isotherms are also found to be in thermal–wind balance with intensified mean equatorward flow, which thus can also be generated by the resonant SLB dynamics. Similar rectified onshore transport of other quantities is expected, relevant for biogeochemical processes. The distribution of maximum diurnal kinetic energy in time and across the shelf supports the concept that subinertial shears create the sufficient condition for resonant response to SLB forcing.
In the vicinity of 30°N or 30°S (diurnal critical latitude), where the local inertial frequency coincides with that of diurnal forcing, resonant ocean response to a diurnal sea/land breeze (SLB) regime can occur since the daily pulses of wind energy input will become synchronized with the ocean currents resulting from the wind forcing. This can lead to large amplification of diurnal fluctuations in the ocean and has been observed on the Namibian shelf (Hyder et al. 2011; Simpson et al. 2002), the Catalonian shelf (Rippeth et al. 2002), the New York Bight (Hunter et al. 2007), the Georgia Bight (Edwards 2008), and the northwest Gulf of Mexico (Jarosz et al. 2007; Zhang et al. 2009). A similar response is also expected in the Southern California Bight (SCB) located slightly poleward of the critical latitude (Fig. 1) since significant diurnal wind forcing associated with an SLB exists near the coast (Gille et al. 2005; Pidgeon and Winant 2005). The inertial period resulting from the planetary vorticity only is approximately 22 h.
Diurnal flow oscillations in the southern area of the SCB were previously observed using moored instruments during 1996 and 1997 off Mission Beach (MB, 32.75°N) by Lerczak et al. (2001) where it was suggested that the energetic baroclinic motions are driven by SLB winds, with the presence/absence of resonance being regulated by the vorticity of the local subinertial currents. Using observations in the northern area of the SCB (>34.0°N), Pidgeon and Winant (2005) found that clockwise-rotating diurnal oscillations near the surface were coherent over the moored array, coherent with the clockwise-rotating wind stress, and coherent with the surface-enhanced temperature changes. Both studies lead to the conclusion that there exist SLB-driven diurnal oscillations in the SCB and potential resonance in the surface ocean.
However, a recent study by Beckenbach and Terrill (2008) suggests that strong diurnal oscillations over a ridge (32.6°N) in the southern SCB are a response to tidal forcing rather than diurnal wind forcing. In addition, observations in the region off Huntington Beach (HB, 33.6°N, Fig. 1) have demonstrated nonlinearly distorted oscillations of tidal origin (rather than wind), related to shoaling internal tides as a cause of diurnal oscillations and occasional transport of subthermocline water, dissolved and/or suspended material into the surfzone (Nam and Send 2011; Noble et al. 2009; Noble et al. 2006; Boehm et al. 2004). Thus, the occurrence, structure, and implications of SLB-driven diurnal oscillations in the SCB are not fully understood, in particular with respect to resonant amplification and interaction with subinertial currents. Clarification of this process in the SCB, but also for other sites, is of interest considering its possible impacts on nearshore circulation, stratification, upwelling, and local biota as found in the northern Monterey Bay (Woodson et al. 2007).
The present study clearly documents resonant SLB energy flux into the ocean in the SCB, provides observations of a three-layered cross-sectional structure of resonant ocean response to SLB as a result of interaction between subinertial current shears and diurnal/near-inertial waves, and suggests rectification of SLB forcing into mean gradients and alongshore flow (leading to amplified temperature/density fluctuations). These conclusions are documented primarily by contrasting of observations during periods of strong versus weak response to SLB forcing off HB.
2. Data and processing
We use time series data of wind, current, and water temperature collected in the southern coastal area of the SCB. An extensive mooring array on the coastal shelf off HB (C2 and MA-MG, Fig. 1b) and a mooring off Del Mar (DM, Fig. 1a) provide overlapping time series data for fall 2006.1 Here, we use time series data from the C2 (water depth: 15 m) and DM (water depth: 100 m) moorings, and moorings across the shelf (MA-MG, water depth ranging from 58 to 8 m) (red crosses in Fig. 1) with an emphasis on the periods of simultaneous measurements of current and water temperature. Details of mooring measurements are described in Table 1. All the data underwent the Science Applications International Corporation (SAIC) standard Quality Assurance/Quality Control (QA/QC) process (http://www.saicocean.com/SAICdocs/ts_QA.html) and we are using hourly averages of the resulting raw data product. We decompose the flow into alongshore υ and cross-shore u components of horizontal currents at the sites off HB and DM by rotating the coordinates 60° and 0° counterclockwise from the north, respectively, for example, u = u0 cos(60°) − υ0 sin(60°) off HB and u = u0 off DM where u0 and υ0 are eastward and northward currents (Fig. 1a).
Time series of wind W measured at two local sites near HB (NA and P2, blue circle and green diamond in Fig. 1b), six National Data Buoy Center (NDBC) buoys (blue circle in Fig. 1a), and five other pier/land stations (green diamond in Fig. 1a) are used to estimate wind stress τ using the quadratic law τ = ρaCD|W|W where air density ρa = 1.25 kg m−3 with a drag coefficient CD according to Anderson (1993). Low-pass and bandpass filters are used to extract subinertial (periods longer than 30 h) and diurnal (15–35 h) fluctuations from horizontal currents u = u + iυ and wind stress τ = τx + iτy in a complex form although unfiltered, hourly-ensemble data are also used in the analysis.
The unfiltered data are used to estimate the rate of work done by the local wind on the sea surface or surface energy flux τ · u = τxu + τyυ (W m−2) and kinetic energy per unit volume KE = (½)ρ0|u|2 (J m−3) to assess the relative importance of diurnal-band work in a broad spectrum of fluctuations. Mean local day–night cycles (as opposed to diurnal tidal cycles) of horizontal currents are also calculated from the unfiltered data to highlight a period of dominant SLB-driven signals among fluctuations at other frequencies. Note that tide-driven internal oscillations at diurnal frequency can be significant in the area and difficult to remove with simple linear methods, for example, harmonic analysis, because of varying phases and nonlinearity, for example, Nam and Send (2011) and Beckenbach and Terrill (2008). The relative importance of winds and tides vary in time depending on many factors including subinertial current shears and interaction with complex bathymetry. This is the reason why we construct an average diurnal cycle following the local day–night cycle (rather than following a diurnal tidal cycle as in Nam and Send 2011; Beckenbach and Terrill 2008). Although not all diurnal tides might be completely removed (or be aliased), it is highly unlikely that the clear contrasts in mean local day–night cycle we will show in the following section come from diurnal tides. We would not expect to see the strong diurnal cycle as observed unless the SLB-driven oscillations dominate.
The bandpass-filtered data are also used to characterize diurnal wind W, to calculate directions of diurnal current θC = arctan(ud/υd) and of diurnal wind stress θτ = arctan(τxd/τyd), and their difference Δθ = θC − θτ. The diurnal-band time series data are used to decompose diurnal current ud = ud + iυd and diurnal wind stress τd = τxd + iτyd into clockwise- (CW) and counterclockwise- (CCW) rotating components as follows:
Here, t and ω are time and constant diurnal frequency (=1.00 cpd); UC, UCC, τC, and τCC, are magnitudes; and ΦC, ΦCC, ϕC, and ϕCC are phases of CW and CCW rotary components of diurnal current and diurnal wind stress, respectively. The CW component of diurnal currents is particularly useful for investigating the inertial response to wind in the Northern Hemisphere (e.g., Jarosz et al. 2007). The phase differences between CW current and CW wind stress Δψ = ΦC − ϕC are compared to check whether and when resonant response occurs as in Jarosz et al. (2007). It is another method we use here for separately validating the near-inertial motions of nontidal origin.
3. Observational results and discussions
a. SLB forcing and resonant response
The climatology (over six years from 2006 to 2011) of diurnal band-passed winds has a large cross-shore component mostly regardless of seasons and locations over the SCB, for example, the principal axis direction is mostly 50°–80° rotated clockwise from the north and the magnitudes are typically 2.0 m s−1 (1.5 m s−1) at northern (southern) stations (Figs. 2a and 2c). Such large-scale SLB forcing even far from the coast has also been found in other areas, for example, at least 250 km offshore in the Georgia Bight, and is nearly an order of magnitude greater than the anticipated offshore scale (Edwards 2008). Patterns of diurnal SLB forcing during the periods of simultaneous moored current measurements, which will be analyzed subsequently, are not significantly different from the climatology, yielding dominant cross-shore components with only slight variations among the locations. All the spectra of winds have clear peaks at the diurnal frequency (1.00 cpd) (not shown) as found previously (Beckenbach and Terrill 2008; Pidgeon and Winant 2005). Based on these results, we assume that the diurnal wind is mainly driven by SLB forcing hereafter (although in general this need not be true as wind variability from frontal passages can fall into the diurnal band). Here, winds at five locations are selected as a proxy for local HB (NA and P2), remote HB (46025), local DM (LJPC1), and remote DM (46086) winds.
Time series of cross-shore wind stresses τx at HB (NA, P2 and 46025) show diurnal bursts of positive (onshore) stress during local daytime, consistent with diurnal SLB forcing (Fig. 3a). From the observed wind stress τ = τx + iτy and surface current u = u + iυ, the rate of work done by the local wind on the surface water τ · u = τxu + τyυ, or surface energy flux is calculated in units of watts per square meter. The estimates of τ · u using the local HB wind measurements (NA and P2) and surface (4.2 m) current measurements at C2 in Fig. 3b demonstrate that the energy flux is mostly positive with diurnal bursts in early October (wind stress enhances currents) with some modulation over the periods of high winds. Using winds from the more distant NDBC buoy 46025 separated by 98 km from the C2 mooring (Fig. 1) does not lead to the same wind work amplitudes (Fig. 3c), implying a resonant response of the surface current to the “local” winds (Simpson et al. 2002). Time series of surface kinetic energy per unit volume KE = (½)ρ0|u|2 (J m−3) at C2 (Fig. 3d) have correlated envelope variations with those of τ · u estimated using the local HB wind measurements (NA and P2). Here, the density ρ0 is set to 1024.0 kg m−3 based on surface density at C2 averaged over the periods A and B (period of simultaneous moored measurements). The strong response to SLB during period A contrasts with the weak response during period B (periods shaded in Fig. 3), and thus our focus here is on the comparison between the two periods. The two periods chosen for this detailed investigation are the only periods when simultaneous data were collected at various moorings (C2 and MA-MG) across the shelf off HB (Fig. 1b).
The diurnal peaks of KE off HB are strongly enhanced during period A resulting from the positive surface energy flux during the period of strong response to SLB particularly in contrast with those during the period of weak response (period B). Note that diurnal SLB forcing itself is not much weaker during period B (Fig. 2b versus Fig. 2d, Fig. 3a), implying that strong SLB forcing is not a sufficient but only a necessary condition for SLB-driven diurnal oscillations. Off DM, some 100 km to the south alongshore (Fig. 1a), the surface energy flux and KE do not show such strong diurnal bursts during both periods A and B but during other periods, for example, early November (Figs. 4b, 4c, and 4d).
The reasons for the presence (period A at HB) and absence (period B at HB and both periods at DM) of a resonant response are analyzed in Fig. 5, which shows that diurnal wind stress and diurnal surface current are often out of phase but in-phase during period A off HB. During period A, diurnal surface currents at moorings across the shelf off HB are enhanced in magnitudes of the CW components [i.e., UC > UCC, see Eq. (1) and Eq. (2) for notation] (thick versus thin lines in Fig. 5a), and had directions θC and CW phases ΦC matching with those of the wind stress, that is, Δψ = ΦC − ϕC is nearly constant in time, and so is Δθ = θC − θτ (Fig. 5c). During period B, the magnitude of the CW components of diurnal surface currents decreased in spite of similar magnitudes of wind stress, and the differences in CW phases Δψ and diurnal directions Δθ vary significantly (Figs. 5b and 5c). Off DM, the CW components of diurnal surface currents do not show resonant enhancement during both periods in spite of strong local (LJPC1) diurnal wind stress caused by the nonconstant Δψ and Δθ (Figs. 5e and 5f). It will be shown below that the resonant response to SLB forcing during period A is limited to the area off HB because of shears from the subinertial alongshore flow, in spite of the wide extent of the SLB forcing over the SCB (Fig. 2b).
b. Structure of SLB-driven diurnal oscillations
Mean daily cycles of cross-shore u and alongshore υ currents were constructed at MA (off HB, Table 1 and Fig. 1b) as shown in Fig. 6 (top). This mooring has only a short record relative to C2 but covers the periods A and B and is the deepest mooring in the area. Time series of τ · u and KE at MA (and MB-ME across the shelf) during periods A and B look similar to those at C2 shown in Figs. 3b and 3d. Figure 6a shows the resulting enhancement of diurnal baroclinic oscillations in the upper layers in response to the SLB during period A in contrast with those during period B (Fig. 6b). During period A, in the upper 15 m, offshore currents (u < 0) prevailed at nighttime whereas onshore currents (u > 0) are found at daytime (Fig. 6a). Additionally, in the vicinity of 20 m depth, the cross-shore currents are in opposite direction to those in the upper layer, yielding onshore (offshore) currents at nighttime (daytime) during period A (Fig. 6a). The cross-shore currents are weak (or smeared out in the averaging over 11 days) below 30 m during period A, and also throughout the water column during period B (Figs. 6a and 6b).
Alongshore currents in the upper half (~30 m) of the water column covered by MA also show strong differences between periods A and B, with intense mean equatorward currents (υ < 0) during period A relative to those during period B (Figs. 6c and 6d). Opposite currents (υ > 0) in the lower half of the MA mooring during period A and over nearly whole water column during period B show a maximum twice a day (around 0600 and 1800 UTC) during both periods, indicating that strong semidiurnal tidal oscillations (Noble et al. 2009; Lerczak et al. 2003) were not completely removed in the 24-h “stacking.” While we cannot be sure that the averaging into a diurnal cycle phase-locked to day–night has removed all diurnal tides, the large difference between periods A and B in Figs. 6a–d must be caused by the presence and absence of SLB response based on the calculations of wind work shown above.
The vertical structure of the SLB-driven cross-shore current response shown in the upper panels of Fig. 6 is not the first baroclinic mode as commonly observed on other shelves (Simpson et al. 2002; Rippeth et al. 2002; Zhang et al. 2009) but has a three-layer structure: 1) a surface layer of cross-shore currents in the same direction as the wind stress, 2) an intermediate layer of cross-shore currents in the opposite direction from the surface layer, and 3) a deep layer of weak cross-shore flow, as also found in the Georgia Bight (Edwards 2008). The equatorward current is strong in the two upper (surface and intermediate) layers. The thickness of the high KE in the upper layers during period A is discussed in the next sections in connection with horizontal and vertical propagation of diurnal/near-inertial energy and with quantitative estimates of horizontal ∂υ/∂x and vertical ∂υ/∂z shears of the subinertial alongshore current.
Similar contrasts between periods A and B are discernible at other moorings across the shelf (see the vertical u profiles in Figs. 6e–h). At all moorings off HB, the offshore (onshore) surface currents and onshore (offshore) intermediate currents occur at nighttime (daytime) with offshore (onshore) wind stress during period A, in contrast with the weak cross-shore currents during period B. These enhanced cross-shore currents during period A cause a horizontally coherent cross-sectional nighttime (daytime) circulation (vertical u profiles; heavy lines in Fig. 6e and 6g). In the nearshore zone (MD-ME, 5–7 km separated from MA where the water depth is 15–20 m), the vertical structure becomes more similar to the first baroclinic mode generating significant near-bottom cross-shore currents (Fig. 6e and 6g).
c. Associated temperature changes and equatorward baroclinic jet
The cross-sectional structure of the rate of diurnal temperature changes (ΔT over 6 h) (color in Figs. 6e–h) is also in clear contrast between periods A and B. During period A, an enhanced and coherent cross-sectional circulation at nighttime (daytime) cools (warms) waters in the upper layers to a degree much exceeding those during period B (color in Figs. 6e and 6g versus Figs. 6f and 6h). During period B, the cooling at nighttime is generally weak except in the nearshore zone (Fig. 6f) where the near-bottom shoaling and breaking of internal tides is able to cool the nearshore water regardless of SLB forcing (Nam and Send 2011; Noble et al. 2009; Pineda 1995). Such cooling by the interaction of diurnal internal tides with topography (not matching the local daily cycle) is presented in Nam and Send (2011) using the same data (i.e., C2 for different period) used here. The SLB-related warming at daytime is also weaker during period B with the strongest response confined to the near-surface (Fig. 6h). The differences between periods A and B in cross-sectional structure of the temporal temperature change ΔT demonstrate that the SLB-driven diurnal oscillations can have a pronounced effect on thermal structure on the shelf as detailed below.
To check whether SLB-driven cross-shore heat advection u∂T/∂x is consistent with the rate of temperature change ∂T/∂t, orders of magnitudes of each part are estimated. The large diurnal temperature amplitudes can result from oscillating advection by the cross-shore flow u (shown in Fig. 6a) of the steepened temperature gradients during period A (thin contour lines in Figs. 6e and 6g). This steepening of isotherms makes the cross-shore temperature gradient ∂T/∂x increase up to O(10−4) °C m−1, resulting in enhanced O(10−5) °C s−1 cross-shore heat advection u∂T/∂x with the increased cross-shore current u ~ O(10−1) m s−1 during period A. The resulting cross-shore heat advection compares well to the observed range of temporal change, i.e., 0.5 ~ 1.0°C in 6 h during period A, yielding 2.3 ~ 4.6 × 10−5 °C s−1.
The more steeply sloped isotherms during period A are postulated here to be a result of a net cross-shore eddy heat flux which cools the nearshore water. We have calculated the depth-integrated, cross-shore eddy heat flux (where means cooling inshore) directly from the C2 mooring data (after subtracting the mean u, over the water column, such that , here angle brackets denote time average over the periods A and B). Time series (Fig. 7) and power spectra of show much larger diurnal fluctuations, that is, larger mean daily cycles (green line), during period A (than period B). Importantly, the positive and negative phases of the cross-shore eddy flux are far from symmetric and more severe cooling often occurs at local nighttime than warming at daytime during period A. The net heat flux averaged over the period A is −0.15°C m2 s−1 (red line in Fig. 7), which corresponds to more than 1°C cooling in one day of the volume of water inshore of the 15-m isobath (location of C2 mooring), approximately 11 250 m3 per unit alongshore distance. This is more than double of that over the period B (−0.06°C m2 s−1, blue line in Fig. 7) and sufficient to produce and maintain, in only one day, a mean cross-shore temperature gradient of 1°C over 2.7 km, comparable to our observations during period A (contours in Figs. 6e and 6g). Such large cross-shore eddy heat flux (cooling) during period A (when the flow is dominated by diurnal oscillations) must result from mixing of the cold water which gets advected onshore (in the bottom layer) during night-time, since the upper onshore flow (at the top layer) during day-time is warmer. That means that the SLB-driven oscillations cause a rectified heat flux and a mean inshore cooling and steepening of isotherms up toward the coast.
In addition, these enhanced mean thermal gradients during period A are in thermal–wind balance (i.e., ∂T/∂x ~ ∂υ/∂z) with the strong mean (subinertial) equatorward currents observed in the upper layers during period A (Fig. 6c). The vertical shear of the alongshore current ∂υ/∂z of O(10−2) s−1 corresponds to a cross-shore temperature gradient ∂T/∂x of 3.7 × 10−4 °C m−1 or 1°C change over the cross-shore distance of 2.7 km resulting from the thermal–wind relationship, which is comparable to our observations during period A as shown in Figs. 6e and 6g. Note that the diurnal change in thermal gradient (difference between contours in Fig. 6e versus Fig. 6g) is much less than the subinertial change (difference between contours in Fig. 6e versus Fig. 6f or Fig. 6g versus Fig. 6h). This is indicative of a tight thermal–wind balance of the vertical shear of “subinertial” (not diurnal) alongshore currents ∂T/∂x ~ ∂υ/∂z ≠ ∂υd/∂z, that is, isotherms do not slope down but are kept up toward the coast even during daytime (contours in Fig. 6g) in spite of daytime cross-sectional circulation with onshore flow at the surface layer (vertical profiles of cross-shore current in Fig. 6g), maintaining the strong equatorward currents at the upper layers during period A (Fig. 6c). This cross-shore geostrophic balance at subinertial scale contrasts with a mainly inertial balance ∂ud/∂t = fυd at diurnal time scale, which does not modulate the thermal gradient ∂T/∂x ≠ ∂υd/∂z. The strong (negative) vertical shear of subinertial currents must be induced by nearshore cooling, a slower process accumulating over several days, and geostrophic adjustment of the mean thermal gradient. Note that no significant equatorward jet was observed during period B off HB and both periods A and B off DM (not shown).
d. Diurnal/near-inertial waves modulated by subinertial current shears
The kinetic energy KE and cross-shore current u at moorings across the shelf and different depths during period A are shown in Fig. 8. High KE lasted longer at offshore than nearshore locations, that is, the shutdown of KE during 6–9 October occurred earlier at nearshore than offshore locations (blue vector in Fig. 8a). The thickness of the high KE layer at MA also varies in time, getting thicker from 2 to 6 October and thinner thereafter. The thickening of the high KE would be reasonable when considering the upward phase (thus downward energy) propagation of diurnal/near-inertial waves which is clear in the time–depth contours of cross-shore currents at MA shown in Fig. 8d (and also at other locations off HB), supporting the notion that the surface is the energy source. This is consistent with the resonant response to SLB during the period A and also with previous observations in the other areas of the SCB, that is, a surface source of diurnal kinetic energy (Beckenbach and Terrill 2008; Pidgeon and Winant 2005; Lerczak et al. 2001). There is no significant phase shift in the cross-shore direction (Figs. 8c and 8e) indicating a long ratio of horizontal to vertical wavelength with nearly vertical phase and thus nearly horizontal energy propagations (phase and energy vectors are orthogonal). The vertical group velocity cgz of near-inertial waves roughly estimated using the dispersion relation under subinertial shears as below (Federiuk and Allen 1996; Kunze 1985) is O(10−4) m s−1 which is consistent with our observations at MA from 2 to 6 October 2 [i.e., O(10) m day−1; red vector in Fig. 8b]:
Here, the estimated orders of magnitudes are for buoyancy frequency N2 = −g(dρ/dz)/ρ0 [=O(10−3) s−1], Coriolis frequency f [=O(10−5) s−1], vertical shear of subinertial current ∂υ/∂z [=O(10−2) s−1], squared horizontal wavenumber k2 [=O(10−9) m−2 or wavelength of 30 km], and vertical wavenumber m [=O(10−1) m−1 or wavelength of 10 m] based on the observations (e.g., vertical and horizontal separations of cross-shore currents shown in Figs. 8d and 8e). However, the thinning of the high KE during 6–9 October (blue vector in Fig. 8b) and earlier shutdown of high KE at nearshore locations (blue vector in Fig. 8a) in spite of still upward phase (downward energy) propagation (Fig. 8d) cannot be explained by the surface generation and downward propagation of diurnal/near-inertial energy. Instead, the three-layered structure in cross-shore current during period A (Figs. 6a, 6e, and 6g) is believed to result from insufficient time for energy to propagate into the lowest layer.
We hypothesize that the shutdown of KE results from offshore KE propagation forced by modulation of the diurnal/inertial oscillations by subinertial shear as analyzed in Fig. 9. Figures 9a–c show the subinertial alongshore current and its horizontal and vertical shears. The contour lines in Fig. 9d show the longest allowable period (1/ωm) for free diurnal/inertial oscillations. Free diurnal/near-inertial oscillations would not exist and thus no resonance would be found at this latitude without subinertial current shears (i.e., local inertial frequency f = 1/22 cph > 1/24 cph or 1.0 cpd). However, both horizontal and vertical shears of the subinertial currents significantly alter the lowest frequency ωm allowed for propagating inertia–gravity waves, calculated as (Federiuk and Allen 1996; Kunze 1985).
Here, the effective Coriolis frequency feff = f + (∂υ/∂x)/2. Locations of the longest allowable period 1/ωm shown in Fig. 9d (contours) move offshore during 6–9 October, following strong equatorward currents (Fig. 9a), positive horizontal shear (Fig. 9b), and negative vertical shear (Fig. 9c) of the subinertial alongshore currents. The timing and location of maximum diurnal KE agrees well with the regions of allowed free diurnal/inertial oscillations. The earlier shutdown of KE at nearshore locations is explained by the offshore movement of the “allowed region.” Previous studies have shown that diurnal/near-inertial waves of frequency ω should propagate out of the area where ωm > ω (e.g., Chant 2001; Federiuk and Allen 1996; Kunze 1985). After 7 October when the surface energy input from the work done by SLB stopped, diurnal KE was observed only at the offshore locations since free oscillations at the diurnal/near-inertial frequencies are prohibited in the inshore area. During period B the longest allowable period becomes shorter than 23 h and close to local inertial period (22 h) across the shelf, thus no resonance with diurnal-period winds is possible.
Similar controls of subinertial current shear on near-inertial waves have been examined previously, for example, Davies and Xing (2002, 2003) and Xing and Davies (2003, 2005). While the above mechanism can explain the three-layer structure, this could in principle also result from strong bottom boundary layer friction (Edwards 2008). Thus, the three-layer structure might be maintained as long as the near-bottom tidal mixing is in action (Nam and Send 2011) even without the control by subinertial current shear. Nonetheless, our observations are consistent with the idea that the shear of the background (subinertial) currents regulates the degree to which surface diurnal energy can be vertically pumped into the interior in the area (Lerczak et al. 2001). The data therefore provide further evidence for SLB together with favorable subinertial shear for a resonant SLB response as a sufficient condition.
e. Resonant response to SLB forcing off Del Mar
Even though the SLB forcing was ineffective to generate a resonant response during both periods A and B off DM (Figs. 4 and 5d–f), there are periods when a resonant response to SLB seems to occur off DM. Surface energy flux or rate of wind work at the surface shows positive diurnal bursts with associated KE fluctuations for example during the period shown in Figs. 10b and 10c (period C). Similar to the case during period A off HB, diurnal surface currents at DM were enhanced in magnitudes of the CW components (UC > UCC), and had directions θC and CW phases ΦC matching to some degree with those of the local wind stress (partially constant Δθ = θC − θτ and Δψ = ΦC − ϕC in time, Figs. 10d and 10e). Thus, resonant response to SLB can occur off DM as well. The resulting mean local day–night cycles of cross-shore and alongshore currents at DM also shows clear daytime onshore flows (u > 0) at the upper layer in response to sea breeze (not shown). Also present are strong equatorward mean currents (υ < 0) in the upper layers, which are different from the climatology of the season (not shown), possibly because of similar nearshore cooling with isotherms sloping up toward the coast with subsequent geostrophic adjustment, like was the case during period A off HB.
Overall, however, the SLB-driven oscillations found at DM in multiyear records are not as clear and strong as those during period A off HB. The diurnal-band wind work and surface KE do not have as regular and pronounced daily peaks, and the agreement in direction between diurnal wind stress and surface currents is more scattered (Fig. 10e), as is their phase difference of the CW component (Fig. 10d). Also, the diurnal baroclinic structure in cross-shore flow, like what is found at MA during period A off HB (Fig. 6a), is not as clear off DM with more complex higher modes. These weaker resonant responses to SLB forcing suggest either less favorable subinertial current shears (needed to create maximum allowed periods near 24 h) and/or other flows at diurnal/near-inertial periods, which confuse the direction/phase relation between the wind forcing and the flow off DM. Thus, the case found during period A off HB cannot be generalized without long-term time series measurements and given that subinertial currents are significantly different between HB and DM. Future studies are necessary to address the generality of what we observed during period A (fall 2006) off HB in a long-term context.
4. Summary and implications
The SLB-driven diurnal oscillations off HB are described by contrasting observations during periods of strong (A) versus weak (B) response to SLB forcing. The resonant response to diurnal wind stress fluctuations associated with SLB forcing produces periods of pronounced and localized diurnal oscillations of flow and temperature during period A (where enhanced diurnal wind work is found with matched diurnal directions of wind and surface currents and with constant relative phases of their clockwise-rotating components). During period A, a coherent cross-sectional circulation with offshore (onshore) currents at the surface layer (upper 15 m) and onshore (offshore) currents at the intermediate layer around 20 m are found at nighttime (daytime), together with significant cooling (warming) caused by enhanced cross-shore heat advection resulting from the strong oscillating currents coupled with steepened temperature gradients (sloped up toward the coast).
The diurnal baroclinic motions observed at the two upper layers during period A are consistent with the mechanism of resonant response to the SLB with a coastal boundary proposed for other shelves (Simpson et al. 2002; Rippeth et al. 2002; Edwards 2008; Zhang et al. 2009). There, it was also found that the directions/phases of surface currents match those of wind stress at diurnal frequency but that the cross-shore currents have a two-layered structure with 180° phase difference between the layers. Our observations at the outer half of the shelf during period A clearly display a three-layered structure (Figs. 6a, 6e, and 6g) as found in the Georgia Bight (Edwards 2008). The three-layer structure is hypothesized to result from the slow downward group velocity and thus the long time it takes energy to be pumped into the lowest layer. The resonant SLB events apparently are not long enough for the energy to penetrate the entire water column.
During resonant SLB events, we find a strong co-occurrence in time and in cross-shelf location between high diurnal KE and the existence of allowed periods for near-inertial oscillations near 24 h. The allowed periods in this range are a result of (both horizontal and vertical) shears of the subinertial currents. We therefore have strong observational support of the concept that resonant response to SLB occurs when the SLB is sufficiently strong and when the subinertial shears allow free near-inertial oscillations near 24-h period (sufficient condition) (Federiuk and Allen 1996; Kunze 1985; Davies and Xing 2002; Davies and Xing 2003; Xing and Davies 2003; Xing and Davies 2005).
The SLB-driven, near-bottom cross-shore currents (in the opposite direction to cross-shore wind stress) in the nearshore zone off HB have large implications for the nearshore thermal structure, cooling/mixing, and the driving of mean currents. Woodson et al. (2007) and Zhang et al. (2010) proposed significant impacts of intermittent SLB-driven upwelling on the nearshore temperature field and an increased local dissipation by vertical mixing near the critical latitude. Our observations indeed demonstrate an enhanced mean (subinertial) cross-shore temperature gradient maintained by net eddy-flux cooling of the nearshore water. This cooling eddy heat flux implies mixing from the enhanced diurnal currents in the inshore bottom layer (water depth less than 20 m) resulting in a nonlinear rectification process for the cross-shelf heat flux. In similar ways, water properties such as nitrate or dissolved oxygen may also be advected and mixed, implying onshore mean transports relevant for biogeochemical processes, for example, plankton blooms or hypoxia events, resulting from the resonant SLB forcing.
The enhanced temperature gradients from the eddy-flux rectification cause a related vertical shear of the alongshore current (strong equatorward current at upper levels) via geostrophic adjustment. The implication is that the resonant SLB oscillations indirectly can cause an intensified mean (subinertial) equatorward flow, clearly seen in our observations, which contrasts alongshore flow during resonant SLB response with that during other periods.
Periods of resonant response are also observed off DM, as judged by enhanced diurnal KE, wind work, and aligned directions of diurnal winds and surface currents, but this did not occur during periods A and B. An SLB response there also leads to daytime onshore flow in the upper layer, and enhanced mean equatorward flow at the upper levels and associated vertical shear supporting the eddy-flux rectification in the nearshore zone. However, the absence of resonant SLB-driven oscillations off DM during period A and the overall weaker resonance at DM implies that either the subinertial current shears (modulating the periods of near-inertial oscillations) and/or other flows confusing the diurnal phase relations between wind and currents are different between the two areas. These results suggest relatively small spatial scales of subinertial currents relative to the wider extent of the SLB forcing over the SCB. Alongshore correlation scales of subinertial alongshore currents on the southern California shelf are on the order of 30 km during summer (Winant 1983) and in excess of 50 km during winter (Lentz and Winant 1986). The alongshore correlation scale of alongshore currents, which is longer than that of cross-shore currents (>30 km), over the northern California shelf was estimated to be 60 km—their maximum mooring separation (Dever 1997). Our observations during period A suggest that the alongshore correlation scale of subinertial currents is less than 100 km (the separation between HB and DM). The results also suggest that resonant response to SLB previously found in the northern area of the SCB (>34.0°N, Pidgeon and Winant 2005) can extend well into the central and southern area, resulting in rectified mean cross-shelf fluxes and mean alongshore flow, with a different and incoherent response over spatial scales of 100 km.
We thank two anonymous reviewers for their careful and detailed reviews of this manuscript. We thank to M. Noble, M. Omand, P. J. S. Franks, and F. Feddersen for many valuable comments and discussions on recent and early versions of this manuscript and G. Robertson and others for support with this and past HB experiments. Thanks also go to M. Buijsman, J. McWilliams, M. Fewings, L. Washburn, S. Gille, J. MacKinnon, A. Boehm, and D. Lucas for their discussions, idea exchanges, and a few suggestions for this study. Mooring data used here were collected and provided by OCSD, SAIC, USGS University of Southern California, and SIO as contributions to the Southern California Coastal Ocean Observing System project. The dedicated effort of all the members of SIO Ocean Time Series Group and other participating groups for technical developments, field work, and data handling is greatly appreciated. S.H. Nam was partially supported by a Joint Institute for Marine Observations, now Cooperative Institute for Marine Ecosystems and Climate (CIMEC) postdoctoral fellowship at SIO.
Durations of the deployments are different between the moorings but an overlapping dataset exists for late September to mid-October 2006.