The coastal ocean, from the shoreline to the mid–continental shelf with water depths ranging from 0 to about O(100) m, spans regions with circulation patterns driven by distinct processes. Coastal ocean circulation regulates the transport of tracers like nutrients, pathogens, and pollutants critical to maintaining healthy ecosystems (e.g., Grant et al. 2005; Boehm et al. 2017), and controls lateral movement of heat, sediment, and entrained gases (e.g., Fewings and Lentz 2011; Sinnett and Feddersen 2019). Bottom sediment resuspension and the advection and mixing of particles, both organic and inorganic, contributes to variable optical clarity of coastal waters. Fluctuations in coastal ocean temperature modify the local stratification, sound speed, and shallow-water acoustics (e.g., Badiey et al. 2002).
Within the coastal ocean, the surfzone extends from the shoreline to the offshore extent of depth-limited wave breaking, while the midshelf region is categorized by nonoverlapping surface and bottom boundary layers separated by a distinct interior. The inner shelf (e.g., Lentz 1994; Lentz 1995b) is a transition region between the surfzone and the midshelf where the boundary layers can overlap. The dynamics within and immediately outside the inner shelf are complicated as surface waves, internal waves, wind, barotropic tidal processes, buoyancy, submesoscale eddies, and boundary layer–driven processes all contribute to changing the circulation pattern and local stratification on frictional, rotational, and longer time scales.
Previous studies targeting the inner shelf have well documented the wind-driven and surface gravity wave–driven dynamics on simple coastlines and bathymetry (e.g., Lentz and Fewings 2012), yet the role of complex, along-shelf-varying coastlines in modifying inner-shelf dynamics on subtidal and shorter time scales is not well understood. In addition, the importance of other physical mechanisms like the role of turbulence forced by high-frequency processes (e.g., nonlinear internal waves) is both poorly understood and undersampled. Moreover, prior studies generally treated processes like subtidal wind-driven circulation in isolation from other inner-shelf physical processes. Nonlinear interactions between wind, surface gravity waves, internal waves, surface heat fluxes, turbulence, and rip currents are yet to be quantified.
Consequently, in order to understand and predict the exchange of water properties (heat, gases, sediment, pollutants, biota) across the inner shelf over a range of temporal and spatial scales the Office of Naval Research Inner-Shelf Dynamics Departmental Research Initiative coordinated field observations (in situ and remote sensing) and numerical modeling efforts on a 50-km section of coast off of central California, in the vicinity of Point Sal, California. Hereinafter, we refer to this experiment as the Inner-Shelf Dynamics Experiment (ISDE).
The principal goals of the ISDE are to (i) diagnose interactions between physical processes in the time-varying inner-shelf circulation, (ii) quantify the importance of along-coastline variability in creating complex circulation patterns on length scales of order 1–10 km, (iii) determine the role of turbulence in mixing tracer fluxes at subtidal and shorter time scales and in constraining momentum and energy transports on all time scales, and (iv) improve the predictive capability of numerical ocean and wave propagation models to simulate regional dynamics.
Here we report on the major findings and data products from the ISDE. A background describing the major physical processes in the inner shelf is considered in the second section. The field experiment and numerical model applications are considered in the third and fourth sections, respectively. Various important physical processes observed during the field experiment, complemented by numerical modeling efforts, are presented in the fifth section. In the sixth section we discuss the spatial heterogeneity of inner-shelf processes observed during the experiment, and the nonlinear interaction between multiple dynamical drivers of circulation and mixing. Findings from this work and suggested future directions for investigation are summarized in the last section.
Background
For decades, the inner shelf has been recognized as an important transition region between the mid–continental shelf and the surfzone. The inshore boundary of the inner shelf is generally agreed to be the surfzone (Garvine 2004; Lentz and Fewings 2012); however, the offshore extent and the dynamical definition of the inner shelf remain somewhat ambiguous and dependent on the dominant processes driving the circulation at a particular coastal region (Fig. 1). Mitchum and Clarke (1986) and Lentz (1994) define the inner shelf in the context of coastal wind-driven dynamics. In this context, the outer boundary of the inner shelf begins where the boundary layers overlap, causing a divergence and eventual shutdown in Ekman transport (Weisberg et al. 2001; Austin and Lentz 2002). Here, we briefly discuss the primary physical processes responsible for inner-shelf circulation and their role in cross-shelf material movement.
Schematic of inner-shelf processes in the coastal ocean considered to be a part of the inner-shelf experiment. Surface waves propagating over the shelf become nonlinear and break generating circulation in the surfzone like rip currents, which eject onto the shelf. Wind and tidally driven along-shelf flows separate from the coastline and generate shelf eddies. Onshore-propagating internal waves also become nonlinear and lead to overturning and mixing in the water column. Additional mixing occurs at the bottom-boundary layer.
Citation: Bulletin of the American Meteorological Society 102, 5; 10.1175/BAMS-D-19-0281.1
Seminal studies of the wind-driven inner shelf include those from the Coastal Ocean Dynamics Experiment (CODE; Beardsley and Lentz 1987; Lentz 1995a), west Florida continental shelf measurements (Weisberg et al. 2001, 2009), Coastal Ocean Processes Study (CoOP; Butman 1994), the Partnership for Interdisciplinary Studies of Coastal Oceans (PISCO; Kirincich et al. 2005), and from the Martha’s Vineyard Coastal Observatory (Fewings et al. 2008). At subtidal time scales, these and other studies demonstrate the sensitivity of cross-shore transport to upwelling-favorable, downwelling-favorable, and cross-shore winds (Austin and Lentz 2002; Fewings and Lentz 2011; Horwitz and Lentz 2016), strength of stratification and structure of mixing (Lentz 1995b), alongshore pressure gradients (Lentz 1995b; Kirincich et al. 2013), and the presence of cross-shore buoyancy gradients (Fewings et al. 2008; Horwitz and Lentz 2014). Previous studies have also demonstrated the role of Stokes drift by surface gravity waves outside the surfzone as a dominant mechanism for cross-shore transport (Lentz et al. 2008; Lentz and Fewings 2012), and radiation stress gradients as a potential leading-order term in the inner-shelf cross-shore momentum budget (Lentz et al. 1999; Fewings and Lentz 2010).
Internal waves are important to the inner shelf over a range of time and spatial scales. Highly nonlinear internal tides and high-frequency internal waves occur in the inner shelf and are responsible for large changes in stratification and strong cross-shore currents over short time scales (Lee 1961; Cairns 1967; Winant 1974; Scotti and Pineda 2004; Sinnett et al. 2018; McSweeney et al. 2020b; Feddersen et al. 2020), and also transport heat from the inner shelf to the surfzone (e.g., Sinnett and Feddersen 2019). These waves also transport plankton and nutrients across the inner shelf (Shanks and Wright 1987; Pineda 1991; Leichter et al. 1998; Lennert-Cody and Franks 1999; Shroyer et al. 2010) and can resuspend and transport sediment (Butman et al. 2006).
Turbulence and mixing generated by internal wave dissipation leads to vertical fluxes of tracers and material (Fig. 1; Bourgault et al. 2008; Walter et al. 2014; Woodson 2018), and enhance stresses which may augment surface and bottom boundary layer overlap (e.g., Palóczy et al. 2021). Processes on a daily time scale such as diurnal heating and diurnal winds can also change the inner-shelf currents and stratification (Lerczak et al. 2003; Cudaback and McPhee-Shaw 2009; Molina et al. 2014; Aristizábal et al. 2017; Walter et al. 2017; Feddersen et al. 2020). In addition, buoyancy-driven flows can dominate the inner-shelf circulation; for example, by river discharge (Allen et al. 1983; Mazzini et al. 2014) or a buoyant response to changes in alongshore winds (Woodson et al. 2009; Washburn and McPhee-Shaw 2013; Suanda et al. 2016).
Mesoscale and submesoscale variability due to eddies, filaments, and fronts is often large on continental shelves and slopes and these are dominant mechanisms for exchange between the continental shelf and open ocean (Huyer and Kosro 1987; Barth 1994; Capet et al. 2008; Gula et al. 2016a). Recent observational and numerical studies have demonstrated the significance of submesoscale processes in causing variability in circulation and driving transport and dispersion in the inner shelf (Nidzieko and Largier 2013; Kirincich 2016; Kirincich and Lentz 2017; Dauhajre et al. 2019) and delivering nutrients to coastal ecosystems (Bassin et al. 2005).
Flow separation past abrupt bathymetry like headlands, capes, and coastal promontories leads to vorticity generation and complex three-dimensional circulation patterns (e.g., Fig. 1). Developed eddies can potentially pair, interact, or coalesce to form a wide variety of tidally rectified and residual flows (Signell and Geyer 1991; Pawlak and MacCready 2002; Callendar et al. 2011). The pressure anomalies associated with topographic eddy generation may play an order one role in the form drag that removes momentum from low-frequency alongshore flows (Warner et al. 2013). These eddies also control the dispersion of dissolved pollutants, floating organisms, and sediment (e.g., Pawlak et al. 2003; Doglioli et al. 2004; Roughan et al. 2005; Jones et al. 2006), enhance biological productivity and larval retention (Karnauskas et al. 2011; Gove et al. 2016), and influence local water-column mixing and dissipation (Canals et al. 2009; White and Helfrich 2013; Dewar et al. 2015; Gula et al. 2016b; MacKinnon et al. 2019).
Material movement also occurs between the shoreline and the inner shelf through the surfzone (Fig. 1). Rip currents and surfzone eddies are the primary known mechanism for cross-shore transport (e.g., MacMahan et al. 2006; Dalrymple et al. 2011; Castelle et al. 2016). Bathymetrically controlled rip currents result from wave breaking on alongshore-variable bathymetry and are typically strongest for shore-normal waves, larger wave height, and lower tidal elevation (Haller et al. 2002; MacMahan et al. 2010; Bruneau et al. 2011; Austin et al. 2013, 2014; Moulton et al. 2017). Stochastic surfzone eddy processes including wave-group vortices (e.g., Long and Özkan-Haller 2009), shear instability of alongshore currents (e.g., Özkan-Haller and Kirby 1999; Noyes et al. 2004), and transient rip currents resulting from short-crested wave breaking (e.g., Peregrine 1998; Spydell and Feddersen 2009; Clark et al. 2012; Feddersen 2014) are also known to be a major driver of cross-shore exchange. Infrared and X-band radar imaging of ejection events (e.g., Marmorino et al. 2013; Haller et al. 2014), dye releases (e.g., Hally-Rosendahl et al. 2014, 2015; Hally-Rosendahl and Feddersen 2016), and recent numerical modeling studies (e.g., Suanda and Feddersen 2015; Kumar and Feddersen 2017b; O’Dea et al. 2021) suggest that transient rip currents lead to cross-shore exchange up to several surfzone widths from the coastline.
The list of physical processes discussed here is not exhaustive as other processes may control the circulation and stratification in the inner shelf. The relative role of aforementioned physical processes has been previously considered (e.g., Fewings et al. 2008), yet additional investigation is required to further constrain the implications for exchange through the inner shelf, especially for locations with complex overlap of processes and variable along-shelf coastline and bathymetry.
Experiment description
The field component of ISDE brought together a novel and synergistic suite of measurement strategies, sensors, and platforms to study a heterogeneous coastal region from the outer shelf to the nearshore within the Santa Maria basin, off of central California, spanning alongshore from south of Purisima Point to Pismo Beach and centered on Point Sal (Fig. 2 and Table 1; the offshore tip of Point Sal is located at 34.9030°N, 120.6721°W). Point Sal is a prominent 2-km-wide asymmetric headland, where the coast is rocky and the coastline bends approximately 120°. Bathymetry offshore off Point Sal is steep and rocky to ≈15 m depth, with several shoals and outcrops within 500 m. South from Point Sal, the rocky coastline extends eastward for 2.5 km before bending to the south where bathymetry contours are alongshore uniform and bottom slopes are less steep. To the north of Point Sal, offshore of Oceano, the shoreline is generally straight, and the offshore bathymetry is roughly alongshore uniform but with sandy crescentic bars close to shore. South of Oceano, a rocky outcrop extends several kilometers offshore at Mussel Point.
Map of the Inner-Shelf Dynamics Experiment study site, showing locations of moorings and bottom landers and measurement footprints of coastal X-band and coherent radar systems. Contour lines represent water depth in meters.
Citation: Bulletin of the American Meteorological Society 102, 5; 10.1175/BAMS-D-19-0281.1
Inner-Shelf Dynamics Experiment observational and modeling efforts, along with the names and affiliations of the principal investigators. Applied Physics Laboratory (APL), Georgia Tech University (GTA), Naval Postgraduate School (NPS), Naval Research Laboratory (NRL), Oregon State University (OSU), Scripps Institution of Oceanography (SIO), Sofar Ocean Technologies (Sofar), University of California Santa Cruz (UCSC), University of Miami (U Miami), University of Otago (UO), University of Washington (UW).
Following preliminary field measurements in 2015 to help define conditions and refine measurement strategy (Allen et al. 2018; Colosi et al. 2018), the main experiment took place from late August to early November 2017, and the field site covered about 50 km along the coast and 15 km across shore in water depths ranging from 5 to 150 m (Fig. 2). The field campaign included moored and bottom time series measurements, ship and small boat surveys, surface drifters, and remote sensing from land, airplanes, and space.
Time series from moored and bottom-mounted sensors
A total of 173 moorings and bottom landers were deployed during the experiment to measure time series of temperature, salinity, current velocity, turbulence, surface gravity waves, and suspended sediment (Fig. 2). Water-column temperature and salinity were measured along vertical mooring lines at 95 locations. Temperature sensors spanned the water column with vertical spacing between sensors of 1–5 m at the deeper locations and 1 m or less in shallow water. Limited salinity measurements were also made. However, temperature was the dominant parameter controlling density, with salinity varying by less than 0.3 psu across the study site during the entire experiment consistent with previous studies (Washburn et al. 2011). Sensor sample intervals varied from 0.5 to 30 s (McSweeney et al. 2020b).
At 52 locations, moorings were paired with bottom landers, separated horizontally by about one water depth. Each lander was equipped with an upward-looking acoustic Doppler current profiler (ADCP) to measure current velocities. The vertical resolution was dependent on instrument frequency and water depth, but ranged between 0.25 and 3 m. Ping averaging resulted in temporal intervals of 0.5 to 30 s, and some of the ADCPs were five-beam instruments with sample frequencies of 1, 2, or 8 Hz allowing for the measurement of currents of surface gravity waves and calculations of turbulent stresses. Most landers were also equipped with temperature sensors, programmed to sample at the same rate as the mooring sensors. In addition, high-precision pressure sensors (Ppods; Moum and Nash 2008; Thomas et al. 2016) were deployed on landers at five locations. Two of the landers were instrumented to observe near bed currents, turbulence, and seabed roughness using a suite of acoustic Doppler velocimeters (ADVs), high-resolution ADCPs (2 MHz), and high-frequency seabed imaging sonars.
Time series measurements of turbulence were acquired by several methods. As noted above, some five-beam ADCPs resolved turbulent stresses (Guerra and Thomson 2017). Temperature microstructure was measured along mooring lines and landers using χ pods (Moum and Nash 2009). A newly developed instrument for this experiment, the GusT (Becherer et al. 2020), was equipped to measure temperature and velocity microstructure as well as pressure and instrument orientation, pitch, roll, and acceleration. Approximately 80 GusT instruments were broadly deployed on moored and shipboard platforms, providing greatly enhanced coverage of turbulence over the inner shelf.
Surface gravity wave directional spectra were measured using Sofar Spotter buoys (Raghukumar et al. 2019) at 18 locations at the study site, and one miniature wave buoy near Point Sal. In addition, a suite of meteorological measurements (wind speed and direction, air temperature, atmospheric pressure, relative humidity, and shortwave radiation) were made from a mooring near Point Sal (Fig. 2) as well as from two locations on land near Oceano and Vandenberg.
Ship and small boat sampling
During two intensive operations periods (IOPs) in early September and mid-October (hereinafter IOP1 and IOP2; Fig. 3), as many as three ships (R/Vs Oceanus, Sally Ride, and Robert Gordon Sproul) and three small boats (R/Vs Kalipi, Sally Ann, and Sounder) conducted coordinated surveys within the study site, designed to resolve processes of interest. All vessels were equipped with downward-looking ADCPs and profiling conductivity–temperature–depth (CTD) instruments. Some ships were also equipped with echosounders, turbulence profilers (VMP-250, χ pods, and GusTs), fluorometers, and meteorological sensors. Bow-chains were deployed from the R/V Sally Ride and the R/V R. G. Sproul and measured temperature, salinity, and turbulence with high resolution vertically (1 m) and in time (sampling frequency: 2-8 Hz) in the upper 20 m of the water column. The R/V Sally Ride conducted over 5,100 vertical profiles using the VMP-250, while the R/V Oceanus equipped with a towed CTD installed with a GusT probe, attached at the leading edge of the CTD (Becherer et al. 2020) conducted over 4,200 profiles. The R/V R. G. Sproul conducted over 3900 profiles with a towed CTD.
Summary of conditions during the Inner-Shelf Dynamics Experiment. (a) Significant wave height. (b) Wind velocity rotated into primary (black) and secondary (red) coordinates; negative winds in the primary direction are upwelling favorable. Wave and wind data are from NDBC Buoy 46011 (Santa Maria). (c) Tidal sea level variations pressure measurements at the 50 m lander at Oceano (Fig. 2). (d) Top-to-bottom temperature difference at the 50 m mooring at the Oceano (Fig. 2). Periods of ship, small boat, airplane, and drifter surveys [intensive operations periods (IOPs)] are indicated by green shading.
Citation: Bulletin of the American Meteorological Society 102, 5; 10.1175/BAMS-D-19-0281.1
Surface drifters
During the IOPs, on 14 separate days, approximately 30 drifters were deployed from small boats on daily (or longer) missions (Fig. 4) that were designed to target specific processes—for example, along- and across-shore transport and dispersion (Spydell et al. 2021), flow around the headland (Point Sal), and mapping surface vorticity and divergence (Spydell et al. 2019). All drifters were equipped with GPS and had real-time tracking capability. Most drifters measured surface temperature. Some were designed to directly measure surface vorticity. Others were equipped with sensors to measure surface shear and turbulence, wave statistics, and meteorological fields (Thomson 2012).
Surface drifter tracks on 13 Sep 2017. Thirty-three CODE (black tracks) and five SWIFT (red tracks) drifters were released in 20–30 m water depths at approximately 0800 local time and sampled for approximately 5 h. Release locations are blue and green dots for CODE and SWIFT drifters, respectively. CODE and SWIFT drifters sample the top-1-m horizontal flow. Bathymetry is contoured at 10 m intervals (labeled thick gray contours) and 2.5 m intervals (thin gray contours). Two internal bores cause the two distinct “kinks” in the tracks. Adapted from Spydell et al. (2021).
Citation: Bulletin of the American Meteorological Society 102, 5; 10.1175/BAMS-D-19-0281.1
Remote sensing
During ISDE, a wide range of remote sensing observations were collected from research aircraft, small unmanned aircraft systems (sUAS), land- and ship-based platforms, and satellites. During the IOPs, two airplanes conducted surveys of the entire study site. Sensors included thermal infrared cameras, optical cameras, hyperspectral cameras, interferometric synthetic aperture radar (InSAR), Modular Aerial Sensing System (MASS; Lenain and Melville 2017), and light detection and ranging (lidar). During these periods, sUAS deployed from land and from ships were equipped with optical cameras to focus on small-scale processes such as rip currents, and interactions between rip currents and internal bores. Surface winds and mean-square slope were estimated using the MASS (Lenain et al. 2019) along the track of the aircraft.
Four marine radars and a coherent imaging radar (all X band) were deployed from towers along the coast during the entire field experiment, with footprints that covered the entire study site (Fig. 2). A coherent marine radar was also operated from R/V Oceanus, and another marine radar was on board the R/V Sally Ride. Data from these systems were processed to focus on surface gravity waves as well as longer-time-scale processes, such as tracking internal waves and bores propagating to the coast, rip currents, and buoyant fronts and eddies and instabilities. In addition, spaceborne X-band and C-band SAR and optical satellite images of the study site were collected during the study period.
Conditions during the experiment
Waves were small during the first IOP (significant wave height, Hsig < 2 m; Fig. 3a). Wave heights were variable in October, with Hsig exceeding 3 m during the second IOP and exceeding 5 m on 21 October. Winds were principally upwelling favorable during the experiment with several relaxation events (Fig. 3b). Strong diurnal wind variability was also apparent. Tides were mixed semidiurnal with a peak range of about 1.8 m (Fig. 3c). Stratification was strongest during the first half of September and was reduced and variable during the remainder of the experiment (Fig. 3d). At a water depth of 50 m, the associated mode-one, linear internal wave speed ranged from 0.3 to 0.15 m s−1 (McSweeney et al. 2020b).
Modeling program
The ISDE modeling program centered on realistic and process-based hydrodynamics spanning a range of model hindcast, forecast, and sensitivity studies. The overall approach is briefly described here with more complete methodology provided elsewhere (Suanda et al. 2016; Kumar et al. 2019). The core modeling program consists of the open-source Rutgers Regional Ocean Modeling System (ROMS) and Simulating Waves Nearshore (SWAN) models, integrated in the Coupled Ocean–Atmosphere–Wave–Sediment Transport Modeling System (COAWST; Warner et al. 2010; Kumar et al. 2012). ROMS is a three-dimensional, bathymetry-following, hydrostatic numerical model (Shchepetkin and McWilliams 2005, 2009) with a long history of coastal applications (Olabarrieta et al. 2011; Kumar et al. 2015, 2016; Wu et al. 2020). SWAN is a third-generation, spectral wave model, which simulates shoaling, refraction, energy input from winds, and energy loss from whitecapping, bottom friction, and depth-limited breaking (Booij et al. 1999).
The model system is configured as a series of one-way, offline, nested grids. The outermost parent grid has resolution of 1/30° (Veneziani et al. 2009) covering the eastern Pacific Ocean from the Baja Peninsula, Mexico to Vancouver, Canada (L0). Subsequent child simulations have resolutions of 1 km (L1), 600 m (L2), 200 m (L3), 66 m (L4), and 22 m (L5), resolving processes from the continental slope and shelf break through the inner shelf and a bulk representation of the surfzone transition region (Fig. 5). Surface atmospheric forcing is taken from a nested Coupled Ocean–Atmosphere Mesoscale Prediction System (COAMPS) model (Hodur et al. 2002; Doyle et al. 2009). Tides are applied as boundary forcing on the L2 domain through harmonic sea level and barotropic velocities from the Advanced Circulation (ADCIRC) tidal model (Mark et al. 2004).
Nested Regional Ocean Modeling System grids (L2, L3, L4, L5) and observation locations: (a) second-level nested grid (L2, resolution: 600 m). Color bar is the water depth h in meters; (b) L3 grid (resolution: 200 m), and (c) L4 grid (resolution: 66 m). In (a) dashed black lines are depth contours of 30, 50, 100, 200, 500, 1,000, 1,500, 2,000, 2,500, and 3,000 m. Parent grids L0 and L1 are shown elsewhere (Suanda et al. 2016).
Citation: Bulletin of the American Meteorological Society 102, 5; 10.1175/BAMS-D-19-0281.1
All modeled variables are stored hourly, spanning multiple years of simulation available through the data archive (Table 2). The nested model configuration was used to create a multiscale coastal forecasting system, coincident with the field experiment September–October 2017. Sensitivity tests and adjoint modeling were also conducted to determine the relative importance of initial and boundary conditions from the parent grid, and surface forcing.
Model hindcasts conducted in the ISDE and made available as part of data archive.
Simulated hydrodynamics mirrors the ISDE field campaign focus; the regional transition between summer and fall determined by the prevailing atmospheric conditions (e.g., Dorman and Winant 2000; Melton et al. 2009). A continued emphasis of the model simulations has been comparison against historical and ISDE observations. Model–data comparisons have spanned subtidal water-column temperature and velocity statistics at both outer- and inner-shelf mooring locations (Suanda et al. 2016; Kumar et al. 2019), to semidiurnal tidal band variability on the inner shelf including barotropic (Suanda et al. 2016) and baroclinic tidal oscillations (Suanda et al. 2017; Kumar et al. 2019). These studies document favorable model–data correlation, and reliable simulation of time-mean water-column vertical structure, large-scale sea level gradients, and water-column velocity.
Forecasting during the experiment
An experimental forecasting system for the inner-shelf circulation around Point Sal was operational from 1 September to 30 November 2017 to assist the field experiment of ISDE. Forecast of ocean conditions were conducted daily through the downscaling of several nested COAWST simulation grids (L0 → L1 → L2 → L3), each forced with the COAMPS 2-day atmospheric forecasts at hourly resolution. Comparisons of the field observations to the model forecasts showed that COAWST at 200 m grid resolution has significant skill in simulating near-surface temperature (not shown here).
Ensemble modeling to determine the relative importance of initial, boundary, and surface forcing
To further diagnose the dynamics underlying the model forecast skill, ensemble simulations of the COAWST Point Sal modeling framework were generated with different configurations of boundary and surface forcing conditions. The ensemble was used to quantify the sources of predictability (e.g., deterministic vs internal variability) that originate from knowledge of the open-ocean boundary conditions, surface forcings and initial conditions. The initial conditions result in little skill beyond a few days and that the largest fraction of dynamical skill is associated with knowledge of the surface and open boundary conditions (Sutherland et al. 2011; Giddings et al. 2014). These findings suggest that direct data assimilation to initialize the ocean model state may not be required for an operational ocean forecast at the inner-shelf spatial and temporal scales.
Adjoint modeling to identify the role of physical processes
The efficacy of the ROMS simulations is controlled by model inputs such as initial and boundary conditions, and model parameterizations to be specified a priori. All circulation aspects (e.g., “Background” section) as represented in the model are therefore sensitive to variations in any or all of these factors, and parameters can be quantified using the ROMS adjoint. Specifically, the sensitivity of any scalar function of the circulation to model parameter and input variability can be computed from a single integration of the adjoint model, by utilizing a state-of-the-art four-dimensional variational data assimilation system which forms part of the ROMS framework.
Two specific processes have been the focus of adjoint sensitivity analyses on the L3 grid: the onshore semidiurnal baroclinic energy flux associated with internal waves (e.g., Kumar et al. 2019), and the vertical (over the water column) transfer of horizontal momentum through vertical mixing. For the former, a specific focus has been on the sensitivity to variations in the formulation of bottom drag, bathymetry and vertical mixing parameterizations. In the case of the vertical transfer of momentum, an index based on the gradient Richardson number (i.e., the ratio of the squared buoyancy frequency to the squared velocity shear) is used to explore how each ocean state component controls vertical transfer of momentum. These numerical studies also serve as a useful prelude to the assimilation of the field observations into ROMS since they provide the spatiotemporal sensitivities of scalar functionals to the circulation fields.
Processes investigated and preliminary findings
Field measurements from various in situ sensors and remote sensing platforms, combined with numerical model results are used to highlight four important physical processes in the Santa Maria basin: internal wave dynamics from the midshelf to the inner shelf, flow separation and eddy shedding off Point Sal, offshore ejection of surfzone waters from rip currents, and wind-driven subtidal circulation dynamics. In addition, the capability of using a few wave sensors to create a regional wave forecasting system is discussed.
Internal wave dynamics
Observed nonlinear internal waves (NLIWs) included steep bores, undular bores, and high-frequency waves of elevation and depression. Internal bores propagated into the region every 6 h (McSweeney et al. 2020b) and were detectable in both in situ and remote observations. Data from satellite SAR and land- and ship-based radar stations demonstrate that internal bores were alongshore coherent of order tens of kilometers, with additional short-scale horizontal variability (Fig. 6). This alongshore coherence decreased toward shore (see BAMS-RadarAnimation-IW.mp4 in the online supplementary information; https://doi.org/10.1175/BAMS-D-19-0281.2; McSweeney et al. 2020a), contributing to nearshore semidiurnal temperature variability (Feddersen et al. 2020).
(left) COSMO-SkyMed X-band synthetic aperture radar (SAR) satellite image at 0158:49 UTC 9 Sep 2017, showing surface signatures of multiple internal waves. Yellow contours indicate the 25, 50, and 100 m isobaths. (right) Same SAR image overlaid with data from three land-based X-band radar stations and one shipboard X-band radar (outlined in white). Red dots indicate mooring locations. COSMO-SkyMed Product Agenzia Spaziale Italiana (ASI) 2017 processed under license from ASI. All rights reserved. Distributed by e-GEOS.
Citation: Bulletin of the American Meteorological Society 102, 5; 10.1175/BAMS-D-19-0281.1
As internal waves propagated into shallower water, their evolution depended on the shelf stratification and background shear. The pycnocline depth ahead of a bore was found to influence the evolution of frontal steepness (McSweeney et al. 2020a). Cross-shore mooring transects illustrate that an internal bore can maintain a sharp front from the 150 to 9 m isobath, while higher-frequency internal waves evolve over shorter distances (Fig. 7). The high space–time resolution of the continuously sampling shore-based radars provided detailed observations of NLIW transformations, including alongshore variability, speed tracking (Celona et al. 2021), and along-crest scalloping (see BAMS-RadarAnimation-IW.mp4 in supplementary information). Other interesting observations were internal wave–wave merging, wave packet stretching, NLIW reflection, and breaking.
(top to bottom) Data from a cross-shore transect of moorings, ranging from (top) 150 to (bottom) 9 m depth. Data include 16 h high-pass-filtered eastward velocities (colored; positive indicates onshore flow) and 1-min-resolution temperature data (contoured at 1°C intervals; 15°C is bold). Internal wave arrivals are indicated with green triangles. (a) Zoom in on the OC50 mooring data, including temperature contours, (middle) the Richardson number [Ri = the ratio of the squared buoyancy frequency to the squared velocity shear], and (bottom) estimates of E (m2 s−3) from the GusT turbulence probes, as well as (top) the probability density function of E values before and after the internal bore front. (b) Zoom-in on the OC25M mooring data, showing near-bottom (1–6 m above bottom) 10-min-averaged estimates of E from a modified ADCP structure function method (Scannell et al. 2017). Similar figures with internal wave arrivals are discussed extensively in McSweeney et al. (2020b,a).
Citation: Bulletin of the American Meteorological Society 102, 5; 10.1175/BAMS-D-19-0281.1