The summer 2015 modeling effort coincided with the time period of the pilot experiment. See the modeling group presentation (.pdf) from the November, 2015 meeting in Seattle, detailing the model-data comparison from PSIEX.

Coastal upwelling around a small headland


The Pt. Sal Inner Shelf Experiment (PSIEX) was a 43-day field experiment aimed at measuring the temporal and spatial variability near the Pt. Sal headland on the central California coast during the summer of 2015.  The subtidal temperature and currents were measured with 32 thermistor strings and 6 acoustic Doppler current profilers in 5 to 50 m water depth covering 4 km alongshore and 6 km cross-shore.  The flow was wind-driven with periods of upwelling favorable winds and episodes of wind relaxations.  During upwelling favorable conditions, cold upwelled water was transported to the edge of the surf zone supported by a weakly stratified water column.  During wind relaxations, warm water was advected north to Pt. Sal.  The alongshore temperature anomaly shows that during upwelling favorable winds a temperature gradient is set-up with warmer temperatures north of Mussel Pt., a small (350m) coastal promontory and colder temperatures south of Pt. Sal.  These observations are consistent with recent studies that examined the role of upwelling and cross-shore winds on the inner shelf.  This study expands upon the flow around large headlands, and shows that small coastline irregularities like Mussel Pt. can modify the local upwelling response.  This hypothesis is supported with results from numerical simulations where Mussel Pt. was removed from the system.

PSIEX Observations

The Pt. Sal Inner Shelf Experiment (PSIEX) was a 43-day field experiment from June to August 2015 that measured winds, temperature, and currents to study the spatial and temporal variability near a coastal headland.  Winds were measured at NOAA National Data Buoy Center (NDBC) station 46011, which is located approximately 33 km offshore of Pt. Sal.  For this study, the temperature was measured using 32 bottom-moored thermistor strings (T-strings) arranged in five arrays: four alongshore arrays and one cross-shore array (Figure 1).  The cross-shore array consisted of 12 moorings that extended from 5 m to 50 m water depth.  The alongshore arrays had the following arrangement:  (1) an inner array with nine moorings along the 7 m isobath (Yi array), (2) an outer array with five moorings positioned ~350 m from the coast (Yo array), (4) three moorings immediately south of Pt. Sal (PS array), and (4) two moorings in the lee of Pt. Sal (PSB array).  There were 29 T-strings in water depths less than 25 m.  These moorings used a total of 257 RBR SoloT thermistors with a 0.002°C accuracy and 1 Hz data rate.  They also included HOBO Tilt Sensors at the mid-water depth in order to detect any occurrences of blowover.  The four moorings in water depths from 30-50 m used SBE39 thermistors with a 0.002°C accuracy and 0.03 Hz data rate.

Currents were measured at six locations: four in a cross-shore array at 6, 11, 15, and 20 m water depth, and two on the 8 m isobath immediately north of Pt. Sal (8A) and Mussel Pt. (8B) (Figure 1b).  The currents were measured using bottom mounted, upward looking acoustic Doppler current profilers (ADCPs).  The ADCPs were 1 (water depths greater than 11 meters) and 2 MHz (water depths less than 11 m) Nortek Aquadopps, and collected 1 Hz data.  The 2 MHz systems had 0.5 m vertical bins, and the 1 MHz system had 1 m bins.  Each ADCP had a co-located pressure sensor.


Figure 1.  Map the of the PSIEX field site.  Circles indicate the location of T-stings and are color-coded by array.  ADCP moorings (red diamonds) are labeled as 6, 11, 15, 20, 8A, and 8B.



The model used is Regional Ocean Modeling System (ROMS) module from the open-source Coupled Ocean Atmosphere Wave and Sediment Transport (COAWST) modeling system [Warner et al., 2010]. ROMS is a three-dimensional, free surface, terrain following numerical model that solves finite-difference approximations of the Reynolds-averaged Navier-Stokes equations using hydrostatic and Boussinesq approximations [Shchepetkin and McWilliams, 2005, and Haidvogel et al., 2008].

The model system here is composed of four nested model domains with one-way nesting and off-line techniques [Mason et al. 2010] shown in Figure 2.  The outermost grid (L1, resolution ∆ = 600m, 546× 386 cells) resolves outer-shelf and encompasses the Southern California Bight. The grid system downscales from L1 grid, to L2 grid (∆ = 200m, 194×362 cells) resolving the inner-shelf, to L3 grid (∆ = 66m, 218×194 cells), and finally to L4 grid (∆ = 22m, 302×230 cells). The L4 grid is the primary focus in the present study. It resolves the inner shelf all the way to the surf-zone region, resolving the rocky outcrop, Point Sal and Mussel Point. The model bathymetry is interpolated using 1-arc second bathymetry from NOAA NGDC coastal relief database. These domains have 42 (L1, L2, and L3) or 15 (L4) topography-following vertical levels.


Figure 2.  Bathymetric contours of the nested four grids with nominal grid spacing of L1-600m, L2-200m, L3-66m and L4=22m. The yellow box shows the placement of the nested child grid within the parent grid


The simulation starts from June to July, 2015 corresponding to the time range of the PSIEX. The solution of a simulation with a larger grid (∆ = 1 km) provides the initial and boundary conditions for L1 grid. Subsequently, the solutions of L1, L2, and L3 grids are interpolated to the next child grid at each time step to provide the boundary and initial conditions. Hourly atmospheric forcings are obtained from the Coupled Ocean/Atmosphere Mesoscale Prediction System (COAMPS) model [Hodur et al., 2002, Doyle et al., 2009]. In addition, barotropic tidal elevation and velocities of eight astronomical tidal constituents (K2, S2, M2, N2, K1, P1, O1, and Q1) are projected onto the lateral boundaries of L1 grid from the ADCIRC tidal model [Mark et al., 2004]. The interaction of tidal forcing within L1 grid propagates throughout the numerical domain, which can be transmitted to the higher-resolution grids by the lateral boundary conditions.

The ROMS simulation in L4 grid was conducted with baroclinic time step of 5 s, while the barotropic time step is one-twentieth of the baroclinic time step. A  turbulence closure model is applied to close the momentum balance equation [Warner et al., 2005]. Radiation boundary condition with nudging is used for baroclinic fields [Marchesiello et al., 2001], while radiation boundary conditions are used for barotropic fields to allow outgoing energy [Chapman, 1985; Mason et al., 2010]. Passive tracer fields (temperature and salinity) and baroclinic velocities are strongly nudged to incoming flow (∆T = 1 day) and weakly nudged to outgoing flows (∆T = 365 days). A horizontal eddy viscosity of 0.15  is used.

Because the model does not utilize data assimilation and therefore does not capture large scale stochastic type fluctuations in temperature, the simulated temperatures in the model exhibits a time varying bias (both positive and negative for different time periods) relative to the observed temperatures. Therefore, a bias correction is computed as the overall mean temperature difference between the model and observations. This spatially uniform, but temporally varying correction (~ 1-2o c) is removed from the model temperature each hour to produce similar magnitudes in the temperature while still preserving all temperature gradients as computed by the model.


Cross-shelf Temperature Response

In order to investigate the cross-shore temperature distribution, a mean, cross-shore isotherm was computed.  The mean isotherm is defined as the temporal average of the water temperature from X01 to X12.  During a period of upwelling (yd 173.5 – 175.5), the mean water temperature was 11.7 °C.  The 11.7 °C isotherm sloped from h=15 m at the offshore mooring to h=5 m at the edge of the surf zone (Figure 3a).  The up-slope of the isotherm toward the coast is indicative of upwelling circulation during alongshore, equatorward winds.  When the winds relaxed and warm water arrived from the SBC (yd 176.5 – 178.5), the mean water temperature was 12.4 °C, and the mean isotherm shoaled to h=5m from offshore to the edge of the surf zone (Figure 3b).  The experimental averaged cross-shore isotherm was 13.1 °C, and sloped toward the shore (Figure 3c).  The experimental average isotherm is similar to the upwelling condition suggesting upwelling was the mean circulation state during PSIEX.

A vertical cross-section of the modeled temperature from the 66m grid for a transect passing through the instrument array is shown in Figure 4. For comparisons with the observations, the mean isotherm from the measurements is included. During the period of upwelling shown in Figure 4(a), the mean isotherm shows the signature of upwelling with a strong upward slope. Offshore, the modeled mean isotherm has more of a slope than the observations indicate. For the relaxation period shown in Figure 4(b), the modeled mean isotherm completely flattens similar to the observations. Finally, as seen in the observations, the overall mean isotherm from the model shown in Figure 4(c) indicates an overall upwelling condition.


Figure 3.  Vertical cross-section of observed temperature (a) averaged during upwelling favorable conditions (yd 173.5 – 175.5), (b) during wind relaxation (yd 176.5 – 178.5), and (c) averaged over the entire experiment.  The black-dashed contour indicates the mean isotherm during each time period, and the white dots are thermistor locations.



Figure 4.  Vertical cross-section of modeled temperature (a) averaged during upwelling favorable conditions (yd 173.5 – 175.5), (b) during wind relaxation (yd 176.5 – 178.5), and (c) averaged over the entire experiment.  The dashed contour is the mean isotherm from the observations during the same period.


Alongshore Temperature Response

The alongshore, mean isotherm was computed similarly to the cross-shore isotherm (described above).  To observe any influence from Pt. Sal, the PS array was included with the Yi array for the alongshore computations.  During an upwelling period (yd 173.5 – 175.5), the alongshore, mean isotherm (12.2 °C) intersected the sea surface near Pt. Sal and the sea bottom in the lee of Mussel Pt (Figure 5a).  The coldest water was in the lee of Pt. Sal, and the warmest water was north of Mussel Pt.  When the winds relaxed (yd 176.5 – 178.5), the mean isotherm (13.0 °C) was relatively uniform from south of Pt. Sal to north of Mussel Pt, although the warm surface layer tended to be thicker north of Mussel Pt. (Figure 5b).  The alongshore, experimental mean isotherm (13.1 °C) is similar to the upwelling condition (Figure 5c), again suggesting a mean upwelling state during PSIEX.

The alongshore mean isotherm based on the model temperatures is shown in Figure 6. During the upwelling period, the alongshore structure is similar to the observations with the warm water north of Mussel Pt and cool water south. During the relaxation the mean isotherm flattens with essentially alongshore uniform temperature. Overall, the model fully captures the alongshore structure of the temperature.


Figure 5.  Similar to Figure 3, but for the alongshore direction.  Arrows indicate the locations of Pt. Sal and Mussel Pt.  Note the color scale for (a) and (b) is different than for (c).


Figure 5.  Similar to Figure 5, but for the model.


In general, temperatures were warmer north of Mussel Pt. than south of Mussel Pt.  In order to study the temporal evolution of the alongshore temperature gradient, the alongshore, surface temperature anomaly was calculated.  The temperature anomaly is calculated by subtracting the 1-hr temporal- and spatial-averaged alongshore, surface temperature of the combined PS and Yi arrays from the time-averaged surface temperature at each mooring in the PS and Yi arrays.  During periods of upwelling favorable winds, the surface temperatures were anomalously higher north of Mussel Pt. (Figure 6(a)).  Consistent with the Gan and Allen [2002] findings, the coldest water was in the lee of Pt. Sal.  When the alongshore winds relaxed to ~0 m s-1, the alongshore temperature gradient also reduced.  The trends in the alongshore temperature anomaly correspond with the trends in the cross-shore isotherm depth anomaly.  When the mean isotherms shoaled toward the coast, the temperatures north of Mussel Pt. were anomalously higher; when the mean isotherms flattened, the temperature gradient relaxed.

The observations of the alongshore temperature anomaly (Figures 8,11a) are described well by the model (Figure 6b).  The model tendency shows warmer anomalies immediately north of Pt. Sal, and colder anomalies in the lee of Mussel Pt.  However, the warm anomaly during active coastal upwelling (e.g., yd 165 to yd 175) is clearly evident north of Mussel Pt.   Based on the model results, Mussel Pt. effectively modifies the local upwelling response by enhancing upwelling in the lee of the promontory.


Figure 6.  (a) Observed alongshore temperature anomaly, (b) modeled alongshore temperature anomaly with Mussel Pt., and (c) modeled alongshore temperature anomaly with Mussel Pt. removed.


It is hypothesized that the cooler upwelled water is transported on-shore, and cross-shore winds are decreasing the upwelling north of Pt. Sal.  However, the orientation of the isobaths in the vicinity of Mussel Pt. leads to an increased alongshore component of the wind stress in the lee of the promontory, and effectively enhances the upwelling in the lee of the small promontory.  The region between Mussel Pt. and Pt. Sal is an intermediate area where cooler water in the lee of Mussel Pt. mixes with the warmer water north of Pt. Sal.


In order to better isolate the effect of Mussel Pt. on the circulation, a new model simulation was completed in which Mussel Pt was removed by forcing the depth to be essentially alongshore uniform inside of the 10m isobaths for the 22m grid (L4). The resulting bathymetric contours, SST and depth average velocity without and with Mussel Pt are shown in Figure (7). Despite the fact that the bathymetry retains the alongshore variability from the offshore rocky outcrop, it is apparent that once Mussel Pt is removed from the domain, the SST and currents become much more alongshore uniform. Most importantly, the pocket of cooler water south of Mussel Pt. is absent when Mussel Pt. is removed. The alongshore temperature anomaly for this case is also shown in Figure 6(c). The temperature gradient across the location where Mussel Pt. was present is eliminated whereby the only appreciable gradient is across Pt Sal. This clearly indicates that Mussel Pt. and the bathymetry in the shallow region (< 10m) is the driving force for the cooler water south of Mussel Pt.


Figure 7.  Model coastline (left) with Mussel Pt. removed and (right) including Mussel Pt.  Shading is the time-averaged SST, contours are bathymetry and vectors are depth and time-averaged velocities. Time averaging was during the upwelling period of yd 173.5 – 175.5.


In general, Mussel Pt. appears to modify the circulation pattern such that the upwelling is strengthened leading to a patch of cooler water. It is clear from evaluating the momentum balances that because of the orientation of the isobaths and coastline to the south of Mussel Pt, the relative orientation of the wind stress is in the alongshore direction. This leads to a local maximum in the offshore Ekman transport for this portion of the coast resulting in stronger onshore upwelling for this section bringing more cold water to the nearshore region producing the alongshore anomaly in temperature. Once the wind relaxes, the upwelling shuts down and the alongshore temperature anomaly goes away. When the model is run for the case without Mussel Pt, the relative orientation of the coast is consistent for the full stretch of the coast, thereby no alongshore temperature anomaly is present.


Freismuth, T.M., D. Cai, J.H. MacMahan, K.A. Haas, S. Suanda, J.A. Colosi, N. Kumar, E. Di Lorenzo, A. Miller, and C. Edwards. (2017) Upwelling response to subtidal, alongshore inner shelf flow around a small-scale coastline promontory on the central California coast. To be submitted to JGR [ see AGU poster ]


In collaboration with Dr. John Colosi, we are also preparing an analysis of regional semidiurnal internal tidal energetics from the 2015 experiment.

N. Kumar, S. Suanda, J.A. Colosi, D. Cai, K.A. Haas, E. Di Lorenzo, A. Miller, C. Edwards, and F. Feddersen (2017) Semidiurnal internal tide generation, propagation, and transformation near Pt. Conception, CA: Observations and model simulations. To be submitted to JPO -[ see AGU poster ]