Rapid vertical exchange at fronts in the Northern Gulf of Mexico

Coastal systems face multiple anthropogenic and climate-related stressors. On the Texas–Louisiana shelf, excess nutrients from the Mississippi/Atchafalaya rivers combined with strong stratification from the freshwater plume lead to seasonal bottom hypoxia (“dead zone”). The plume forms density fronts with tilted isopycnals that can connect surface and bottom waters, raising the possibility that slantwise vertical motions along isopycnals could ventilate oxygen-deficient bottom waters without overcoming the stratification barrier by turbulent mixing. During summer, a prominent diurnal land–sea breeze near 30°N resonates with the local inertial frequency, generating strong near-inertial currents while tides are relatively weak. This suggests strong interactions between inertial currents and plume fronts that could facilitate rapid vertical exchange. As part of the SUNRISE project, theoretical, modeling, and observational studies were undertaken to explore how land–sea-breeze–front interactions drive vertical exchange and impact marine environmental conditions over the shelf.

Background work documents Gulf of Mexico hypoxia and shelf dynamics, including near-inertial oscillations on the Texas–Louisiana shelf and relatively weak tides. Prior studies show submesoscale fronts and eddies arise from river plumes and can support strong lateral buoyancy gradients, frontogenesis, and slantwise circulations. Potential vorticity (PV) dynamics and Ekman buoyancy flux (EBF) have been implicated in front destabilization and mixing. Previous research on primary river plume fronts reported significant subduction at leading edges where density contrasts are large; the present work examines similar magnitudes of subduction at submesoscale plume-associated fronts with smaller density contrasts (1–2 kg m⁻³).

Observations: The SUNRISE Campaign (June 19–July 9, 2021) used R/Vs Walton Smith and Pelican to conduct repeated parallel transects (~20 km, ~4 h each), separated by ~1 km, adaptively sampling plume fronts. Instruments included Vertical Microstructure Profilers (VMPs) for hydrography (1 m vertical resolution, ~110 m between casts) and 600/1200 kHz ADCPs (processed with UHDAS+CODAS; 1 m and 0.5 m vertical bins, 2 min averages). Synchronized sampling enabled calculation of buoyancy gradients, divergence, relative vorticity, and PV using a plane-fitting method minimizing spatial-temporal aliasing. Numerical simulations: Triple-nested TXLA configurations using ROMS (hydrostatic) for L1 and L2 and CROCO-NBQ (non-hydrostatic kernel) for L3. Domains: L1 (entire shelf and slope; 650 m–3.7 km horizontal resolution; 30 terrain-following vertical levels), L2 (central shelf; ~300 m), L3 (front-focused; 100 m). Vertical grids concentrated near surface and bottom to resolve boundary layers; same S-coordinate stretching across domains. Forcing: Global HYCOM reanalysis (L1 nesting), ERA-Interim surface fluxes; river discharge from USACE and USGS. Nesting: L1 (1994–2016), L2 (June 1–July 26, 2010), L3 (June 10–16, 2010; wind conditions similar to 2021 field). Turbulence: k–ε GLS closure. Outputs include velocity, temperature, salinity, and dissolved oxygen. L3 is one-way nested into L2; L2 is two-way nested into L1. Oxygen model: Oxygen treated as a passive tracer with surface gas exchange assumed instantaneous at saturation based on temperature and salinity; riverine waters are oxygen-saturated; lateral open boundaries relaxed to parent models. Benthic respiration provides a bottom oxygen sink following established parameterizations; photosynthesis effects are not included. Reduced physics model: Linear mixed-layer momentum equations with Coriolis, vertically averaged pressure-gradient forcing across the front, and vertically averaged frictional body force via turbulent viscosity; nonlinear advection omitted. Forced with diagnosed pressure-gradient and frictional terms from L3 (top 5 m) along an across-front section; integrated for 3 days. Ageostrophic velocity obtained by subtracting geostrophic component. Friction interpreted via surface wind stress and bottom-of-mixed-layer geostrophic and ageostrophic stresses (geostrophic stress relates to thermal-wind shear and turbulent viscosity). Slab mixed-layer model: Idealized inertial response to wind stress in a 5 m-thick mixed layer using linear momentum with Coriolis and wind stress forcing; initialized from L3 ageostrophic velocity; forced by spatially averaged wind stress from L3 region; integrated for 3 days to generate reference inertial phasing on dense side of front. Lagrangian analyses: 3D floats seeded in simulations near temperature streamers and oxygen intrusions; backward tracking to determine origin and forward tracking to map subduction/upwelling pathways and rates. Metrics included float depth changes, along-isopycnal pathways, and Lagrangian oxygen tendency.

- Observations (June 23–28, 2021) captured a plume front oscillating near-inertially, with strong near-surface convergence at the leading (denser) edge and divergence on the lighter side. VMP sections showed depressed isopycnals beneath the convergent leading edge, warm surface water streamers subducting, and intrusions of cooler, higher-turbidity, lower-oxygen water upwelling on the light side. - Simulations reproduced diurnally modulated upwelling of lower-oxygen water into the surface mixed layer at fronts, strongest in the finest-resolution L3 domain (~100 m), consistent with observations. - Near-surface convergence at fronts varied near-inertially with the land–sea breeze, producing narrow convergence lines (δ/f < −5) and strong vertical motions. Convergence occurred primarily in frontal filaments with positive relative vorticity, distinguishable from eddies with negative vorticity. - Slantwise exchange along isopycnals: Warm surface waters subducted ~10 m within a few hours beneath the leading edge; bottom waters with lower oxygen upwelled along tilted isopycnals on the light side, rising ~10 m within ~18 h. - Lagrangian backward/forward float tracking confirmed origins and pathways: surface-released floats uniformly subduct along the front; bottom boundary layer floats uniformly rise on the light side. - Oxygenation: Lagrangian rate of increase in dissolved oxygen near the surface during upwelling pulses estimated at 0.82 mg L⁻¹ day⁻¹, comparable to summertime deoxygenation rates from aerobic respiration in the Gulf of Mexico. - Potential vorticity (PV) structure mirrored temperature and oxygen: high PV associated with strongly stratified bottom waters was drawn upward; low to negative PV subducted from the surface ahead of the leading edge. PV conservation held until reaching the surface boundary layer, where frictional/diabatic processes (positive vertical PV flux) rapidly reduced PV, sometimes changing sign. - Ekman buoyancy flux (EBF), expressed as an effective heat flux, peaked near 20,000 W m⁻² at the front, indicating intense destabilization and boundary-layer mixing; similar magnitudes were inferred from observations. - Mechanism: Diurnally varying winds mix geostrophic momentum at the front (elevated turbulent viscosity), disrupting geostrophic balance and driving ageostrophic cross-front flow reversals. Opposing ageostrophic flows on dense vs. light sides yield strong convergence/divergence patterns that modulate frontogenesis (intensified lateral buoyancy gradients, increased cyclonic vorticity) over the inertial cycle.

The study demonstrates that diurnal land–sea-breeze forcing near the inertial frequency interacts with plume fronts to generate strong, narrow, near-surface convergence/divergence bands that drive slantwise vertical circulations. These circulations subduct warm, oxygenated surface waters on the dense side and upwell cooler, oxygen-poor bottom waters along isopycnals on the light side, providing a ventilation pathway that can bypass the strong stratification of the Mississippi/Atchafalaya plume. The process yields rapid vertical displacements (~10 m over hours) and measurable oxygenation rates comparable to biological oxygen consumption, indicating potential to alleviate hypoxia intermittently. The PV and EBF diagnostics reveal the surface boundary layer at fronts as hotspots of irreversible mixing, with frictional and diabatic processes extracting PV and enhancing turbulence. The findings also suggest that convergence-driven downwelling could subduct buoyant material (e.g., microplastics, oil droplets, phytoplankton, Sargassum) beneath the front, complicating surface accumulation patterns. The magnitude of modeled subduction is comparable to that observed at primary river plume fronts despite smaller density contrasts, implying that such diurnally modulated frontal processes may be widespread in submesoscale-rich coastal regions and important for tracer, heat, and carbon exchange.

High-resolution observations and nested simulations reveal that near-resonant diurnal land–sea-breeze forcing produces inertially modulated convergence/divergence at Mississippi/Atchafalaya plume fronts, driving slantwise vertical exchange along isopycnals. This mechanism subducts surface waters and ventilates bottom waters, elevating dissolved oxygen and modifying PV and stratification, thereby offering a pathway that bypasses the plume’s stratification barrier and potentially influencing the dynamics of the Gulf’s seasonal dead zone. The study isolates the underlying physics via reduced models, highlighting the role of wind-driven mixing of geostrophic momentum and subsequent ageostrophic flow reversals in frontogenesis and vertical circulation. Future work could quantify the cumulative seasonal impact on hypoxia extent, assess variability across different wind regimes and frontal strengths, incorporate full biogeochemical oxygen sources/sinks (e.g., photosynthesis), and evaluate implications for transport of buoyant pollutants and biological material.

- The oxygen model assumes instantaneous surface gas exchange to saturation and excludes photosynthesis, simplifying oxygen sources/sinks; benthic respiration is parameterized, which may affect oxygen tendency estimates. - The finest-resolution dynamics and upwelling/subduction signatures are most evident in L3 (100 m), with weaker signatures in coarser L1/L2 due to resolution limitations. - Simulations analyze June 2010 forcing scenarios analogous but not identical to the 2021 field conditions; year-to-year variability is not fully sampled. - Observational snapshots cover several days and specific fronts; broader spatial-temporal generalization relies on simulations. - PV conservation breaks down in the surface boundary layer due to frictional/diabatic effects; precise partitioning of these processes has uncertainties. - Reduced physics and slab models omit nonlinear advection and other complexities, serving as interpretative tools rather than full dynamical representations.