Using nonlinear forecasting to learn the magnitude and phasing of time-varying sediment suspension in the surf zone

B. E. Jaffe; D. M. Rubin

doi:10.1029/96jc00495

Home/Archive/USGS

USGS · 70018689

Using nonlinear forecasting to learn the magnitude and phasing of time-varying sediment suspension in the surf zone

B. E. Jaffe·D. M. Rubin

View original source ↗Export metadata ↓

Abstract

The time-dependent response of sediment suspension to flow velocity was explored by modeling field measurements collected in the surf zone during a large storm. Linear and nonlinear models were created and tested using flow velocity as input and suspended-sediment concentration as output. A sequence of past velocities (velocity history), as well as velocity from the same instant as the suspended-sediment concentration, was used as input; this velocity history length was allowed to vary. The models also allowed for a lag between input (instantaneous velocity or end of velocity sequence) and output (suspended-sediment concentration). Predictions of concentration from instantaneous velocity or instantaneous velocity raised to a power (up to 8) using linear models were poor (correlation coefficients between predicted and observed concentrations were less than 0.10). Allowing a lag between velocity and concentration improved linear models (correlation coefficient of 0.30), with optimum lag time increasing with elevation above the seabed (from 1.5 s at 13 cm to 8.5 s at 60 cm). These lags are largely due to the time for an observed flow event to effect the bed and mix sediment upward. Using a velocity history further improved linear models (correlation coefficient of 0.43). The best linear model used 12.5 s of velocity history (approximately one wave period) to predict concentration. Nonlinear models gave better predictions than linear models, and, as with linear models, nonlinear models using a velocity history performed better than models using only instantaneous velocity as input. Including a lag time between the velocity and concentration also improved the predictions. The best model (correlation coefficient of 0.58) used 3 s (approximately a quarter wave period) of the cross-shore velocity squared, starting at 4.5 s before the observed concentration, to predict concentration. Using a velocity history increases the performance of the models by specifying a more complete description of the dynamical forcing of the flow (including accelerations and wave phase and shape) responsible for sediment suspension. Incorporating such a velocity history and a lag time into the formulation of the forcing for time-dependent models for sediment suspension in the surf zone will greatly increase our ability to predict suspended-sediment transport.

Explore related subjects

Journal of Geophysical Research C: Oceans Article

Keep this discovery

Explore connections, maps & timelines

BibTeX RIS

B. E. Jaffe, D. M. Rubin. 1996-06-15. Using nonlinear forecasting to learn the magnitude and phasing of time-varying sediment suspension in the surf zone. https://doi.org/10.1029/96jc00495

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related USGS reports

USGS2020

Surface methane concentrations along the mid-Atlantic bight driven by aerobic subsurface production rather than seafloor gas seeps

Relatively minor amounts of methane, a potent greenhouse gas, are currently emitted from the oceans to the atmosphere, but such methane emissions have been hypothesized to increase as oceans warm. Here, we investigate the source, distribution, and fate of methane released from the upper continental slope of the U.S. Mid-Atlantic Bight, where hundreds of gas seeps have been discovered between the shelf-break and ~1600 m water depth. Using physical, chemical, and isotopic analyses, we identify two main sources of methane in the water column: seafloor gas seeps and in situ aerobic methanogenesis which primarily occurs at 100 – 200 m depth in the water column. Stable isotopic analyses reveal that water samples collected at all depths were significantly impacted by aerobic methane oxidation, the dominant methane sink in this region, with more than 50% of the methane being oxidized, on average. Due to methane oxidation in the deeper water column, below 200 m depth, surface concentrations of methane are influenced more by methane sources found near the surface (0 – 10 m depth) and in the subsurface (10 - 200 m depth), rather than seafloor emissions at greater depths.

Atlantic margin↗

USGS2020

Seasonal variation in sediment delivery across the bay-marsh interface of an estuarine salt marsh

Sediment transport across bay–marsh interfaces depends on wave energy, vegetation, and marsh-edge morphology, and varies over a range of timescales. We investigated these dynamics in a tidal salt marsh with a gently-sloped, vegetated edge adjacent to northern San Francisco Bay. Spartina foliosa (cordgrass) inhabits the lower marsh and Salicornia paciﬁca (pickleweed) predominates on the marsh plain. We measured suspended-sediment concentration (SSC) and hydrodynamics in bay shallows and along a 100-m cross-shore transect in the marsh, during winter and summer. Four-year averaged accretion measured with marker-horizon plots was twice as great along the marsh transect as adjacent to a tidal creek, 50 m from the bay. We estimated deposition and trapping eﬃciency from the time-series data to assess its variation with season and wave energy. At high tide the transition zone (between cordgrass and pickleweed) was usually erosional, the pickleweed zone was depositional, and both erosion and deposition increased with wave energy, as did the landward position of maximum deposition. Erosion from the transition zone accounted for approximately one-third of the sediment ﬂux into the pickleweed. In the pickleweed zone, SSC, the diﬀerence between ﬂood- and ebb-tide SSC and trapping eﬃciency were greater in summer than winter for comparable wave conditions, which we attribute to increased sediment trapping by dense summer cordgrass. Moderate waves in summer (46%) accounted for more annual accretion in the pickleweed zone than larger waves in winter (28%), although the contribution of winter storms was diminished by the dry winter during the study.

California↗

USGS2020

Alkalinity in tidal tributaries of the Chesapeake Bay

Despite the important role of alkalinity in estuarine carbon cycling, the seasonal and decadal variability of alkalinity, particularly within multiple tidal tributaries of the same estuary, is poorly understood. Here we analyze more than 26,000 alkalinity measurements, mostly from the 1980s and 1990s, in the major tidal tributaries of the Chesapeake Bay, a large, coastal-plain estuary of eastern North America. The long-term means of alkalinity in tidal-fresh waters vary by a factor of 6 among seven tidal tributaries, reflecting the alkalinity of non-tidal rivers draining to these estuaries. At 25 stations, mostly in the Potomac River Estuary, we find significant long-term increasing trends that exceed the trends in the non-tidal rivers upstream of those stations. Box model calculations in the Potomac River Estuary indicate that the main cause of the estuarine trends is a declining alkalinity sink. The magnitude of this sink is consistent with a simple model of calcification by the invasive bivalve Corbicula fluminea. More generally, in tidal tributaries fed by high-alkalinity non-tidal rivers, alkalinity is consumed, with sinks ranging from 8 to 27% of the upstream input. In contrast, tidal tributaries that are fed by low-alkalinity non-tidal rivers have sources of alkalinity amounting to 34 to 171% of the upstream input. For a single estuarine system, the Chesapeake Bay has diverse alkalinity dynamics and can thus serve as a laboratory for studying the numerous processes influencing alkalinity among the world’s estuaries.

Maryland, Virginia↗