Sediment transport refers to the movement of particulate organic and inorganic material, known as sediment, by geophysical agents such as water, wind, or ice, playing a fundamental role in shaping Earth's surface through erosion, entrainment, transport, and deposition. This process is central to geomorphology, as it governs the evolution of landscapes, river channels, coastal zones, and other landforms by redistributing sediment from source areas to depositional sinks. In fluvial systems, for instance, sediment is eroded from upland watersheds and delivered downstream via rivers and streams, ultimately depositing in deltas, mudflats, and wetlands.[1]Sediment transport occurs in distinct modes depending on particle size, flow characteristics, and environmental conditions. Bedload involves coarser particles that roll, slide, or saltate along the bed surface, typically comprising 5-20% of total sediment flux in rivers[2] but exerting significant influence on channel morphology. Suspended load, the dominant mode for finer sands, silts, and clays, consists of particles held aloft in the fluid column by turbulence, often accounting for the majority of sediment transport volume in turbulent flows. A third category, wash load, includes very fine particles like clay that remain in suspension even under low flow conditions due to their low settling velocity. These modes collectively determine the total sediment load, which can be quantified as the maximum transport capacity—the upper limit of sediment a flow can carry before deposition occurs—governed by factors such as shear stress, stream power, and particle characteristics.The initiation and rate of sediment transport are controlled by hydrodynamic forces overcoming particle resistance, as described by foundational models like the Shields criterion for bedload entrainment and the Hjulström curve relating flow velocity to erosion and deposition thresholds. Key influencing factors include flow discharge, channel slope, sedimentgrain size and density, and turbulence intensity, with transport capacity often scaling nonlinearly with these variables across fluvial, aeolian, and coastal environments. In coastal settings, wave action and tidal currents drive longshore sediment movement, while in arid regions, wind shear mobilizes dunes via saltation and suspension.[3] Human activities, such as dam construction and land-use changes, can disrupt natural sediment budgets, leading to erosion or aggradation that alters ecosystems and infrastructure.[1]Understanding sediment transport is essential for environmental management, flood hazard assessment, and predicting landscape responses to climate change, as it sustains habitats, maintains water quality, and influences nutrient cycling in aquatic systems.[1] For example, in riverine and coastal contexts, balanced sediment supply supports wetland resilience against sea-level rise and protects against erosion, benefiting biodiversity and human settlements.[1] Advanced monitoring techniques, including acoustic sensors and numerical modeling, enable precise measurement of transport rates, informing restoration efforts and policy decisions.[4]
Fundamentals
Definition and Key Processes
Sediment transport refers to the movement of solid particles, known as clastic sediments such as sand, gravel, and silt, driven by the action of fluids including water, air, and ice.[5] This process encompasses the sequential stages of erosion, where particles are dislodged from their source; entrainment, where they are lifted into the fluid; transport, where they are carried along; and deposition, where they settle out as fluid energy diminishes.[6] These mechanisms collectively redistribute sediments across landscapes, influencing geomorphic evolution over geological timescales.[7]The key processes begin with entrainment, the initial lifting of particles from the bed surface due to fluid forces exceeding the sediment's resistance.[8] During transport, particles move either as bedload, which involves rolling, sliding, or saltating along the bed in close contact with it, or as suspended load, where finer particles are buoyed within the fluid column and travel farther downstream.[9] Deposition occurs when the fluid's velocity or turbulence decreases, allowing particles to settle under gravity, often forming new depositional features like bars or deltas.[10]Early theoretical foundations for sediment transport were laid by Paul Du Boys in 1879, who developed the first predictive model for bedload transport rate as a function of shear stress, marking a pivotal advancement in quantifying these processes.[11] Over geological time, sediment transport shapes Earth's surface by eroding highlands, filling basins, and constructing landforms, with transport rates typically measured in units of mass per unit width per unit time, such as kilograms per meter per second (kg/m/s).[12][13] In fluvial environments like rivers, these processes maintain channel morphology and sediment budgets.[5]
Sediment Properties and Classification
Sediment particles, the fundamental units of transport in geomorphic systems, exhibit a range of physical properties that determine their mobility and interaction with fluid flows. Particle size is a primary attribute, typically classified using the Wentworth scale, which categorizes sediments into classes such as clay (<0.0039 mm), silt (0.0039–0.0625 mm), sand (0.0625–2 mm), gravel (2–64 mm), and larger categories like cobbles and boulders (>64 mm). This logarithmic scale, developed by Chester K. Wentworth in 1922, provides a standardized framework for describing grain size distributions in sedimentary deposits and is widely adopted in geological and engineering contexts for its simplicity and applicability across environments.Shape influences sediment transport by affecting drag and packing efficiency; spherical or rounded particles experience less resistance during entrainment compared to angular or platy ones, which can interlock and increase stability on bed surfaces. Density, often around 2650 kg/m³ for common quartz grains, contrasts with the density of the transporting fluid (e.g., ~1000 kg/m³ for water), generating the submerged weight that resists motion. Fall velocity, the terminal speed at which a particle settles in still fluid, varies with these properties—finer, less dense particles fall more slowly (e.g., ~10^{-4} m/s for fine silt in water) than coarser ones (e.g., ~0.02 m/s for fine sand, up to ~0.2 m/s for coarse sand)—and serves as a key indicator of suspension potential.[14]Classification extends beyond size to include cohesion, particularly in fine-grained sediments. Sands and gravels are generally non-cohesive, behaving as individual granules under shear stress, whereas muds and clays (<0.0625 mm) exhibit cohesion due to electrochemical forces between clay minerals, requiring higher stresses for initial motion and often forming stable aggregates or flocs. This distinction is critical in cohesive sediment transport models, as seen in estuarine and lacustrine settings where clay minerals like kaolinite or montmorillonite dominate.Sediment sorting refers to the degree of size uniformity within a deposit, arising from selective transport where finer particles are more easily entrained and suspended, leading to coarser lags on beds. Compositionally, sediments may be siliciclastic (quartz, feldspar) or carbonate-based, with implications for durability; for instance, softer calcareous grains erode faster during transport. Biogenic sediments, such as shell fragments or diatom frustules, introduce organic components that can enhance cohesion or buoyancy, altering transport dynamics in coastal or lacustrine environments. These properties collectively influence the threshold for motion, with finer, less dense particles initiating transport at lower fluid stresses than coarser, angular ones.
Transport Environments
Fluvial Environments
In fluvial environments, sediment transport occurs primarily in rivers and streams, where flowing water exerts shear stress on the channel bed and banks, mobilizing and displacing particles ranging from fine silt to coarse gravel. The dominant processes vary with channel gradient and sediment size: in high-gradient mountain streams, bedload transport prevails, with coarser particles (typically gravel and larger) moving along the bed through rolling, sliding, and saltation at velocities lower than the surrounding fluid.[15] In contrast, lowland rivers favor suspended load transport, where turbulence lifts finer particles (sand, silt, and clay) into the water column, allowing them to be carried at near-fluid velocities over long distances.[9] Bed-material load, comprising the coarser fraction that interacts directly with the channel form, often moves as bedload in steeper settings, while wash load—finer material not part of the bed—predominates in suspension across both environments.[16]Floods play a critical role in amplifying sediment transport, often accounting for the majority of annual flux during short, high-magnitude events that increase discharge and shear stress, mobilizing large volumes of both bedload and suspended material. These events can elevate suspended sediment concentrations dramatically, leading to aggradation downstream and influencing long-term channel evolution.[17]Channel morphology significantly influences transport patterns, with meandering rivers—characterized by high sinuosity (>1.5), narrow and deep cross-sections, and low gradients (slope ≤ 10⁻⁴)—promoting suspension-dominated transport of fine sands and silts, resulting in vertical accretion on floodplains and point bars through overbank deposition.[16] Braided rivers, conversely, feature low sinuosity, wide and shallow channels, and steeper gradients, where bedload of coarse gravel and cobbles forms mid-channel bars and islands via frequent avulsions and lateral shifts, with deposition occurring primarily within the active channel.[16] Sediment deposition in bars and floodplains thus reflects these dynamics: bars accrete through bedload sheets in braided systems, while floodplains build via suspended fines in meandering ones, often stabilized by vegetation.[16]Key factors modulating transport include discharge variability, which drives episodic erosion and deposition, and bank erosion, contributing substantially to the total sediment supply in unstable channels.[18] For instance, the Mississippi River exemplifies these processes, with an annual suspended sediment load of approximately 126.7 million tonnes at upper reaches, declining to 67.2 million tonnes downstream due to deposition, while bed-material load decreases from 14.1 million tonnes to 4.0 million tonnes amid high discharge variability (average 14,795 m³/s, peaking at 45,845 m³/s).[18] Bank caving and floodplain exchanges further shape its sediment budget, trapping about 70% of sand within the lowermost reaches and causing up to 10 m of bed aggradation in depositional zones.[18]Measurement techniques in fluvial settings often employ acoustic Doppler current profilers (ADCPs) to quantify flow-sediment interactions, using acoustic backscatter intensity to estimate suspended sediment concentrations across the water column via calibrated sonar equations.[19] ADCPs provide continuous vertical profiles of velocity and sediment, offering advantages over discrete sampling by capturing turbulence-driven suspension and bedload proxies through bottom-track data, though calibration against physical samples is essential for accuracy.[19]
Aeolian Environments
Aeolian sediment transport refers to the movement of sediment particles, primarily sand and dust, by wind in subaerial environments such as deserts, coastal dunes, and arid regions. This process is driven by aerodynamic forces where wind shear exceeds the threshold for particle entrainment, typically occurring in areas with sparse vegetation and loose, unconsolidated surfaces. Unlike denser fluid media, air's lower density limits transport capacity but enables long-distance dispersal of fine particles.The primary mechanisms of aeolian transport include saltation, reptation, and suspension. In saltation, sand grains (typically 0.06–2 mm in diameter) are lifted by wind gusts and follow ballistic trajectories, bouncing along the surface and contributing up to 50–75% of total sediment flux through impacts that dislodge additional particles. Reptation involves short, rolling or sliding motions of grains induced by saltating particle collisions, while suspension occurs when finer particles (e.g., silt or clay <0.06 mm) are carried high into the atmosphere, often forming dust plumes that can travel thousands of kilometers. The initiation of these processes requires a threshold wind speed of approximately 5–6 m/s at 10 m height for typical sand grains, beyond which transport rates increase nonlinearly with wind velocity cubed.Characteristic landforms resulting from aeolian transport include dunes and loess deposits. Dunes form through organized patterns of sand accumulation, such as crescent-shaped barchan dunes that migrate in unidirectional winds or linear transverse dunes aligned perpendicular to prevailing winds, with migration rates of 10–30 m per year in active fields. Loess, wind-deposited silt, blankets vast areas like the Chinese Loess Plateau, accumulating over millennia from distant glacial or desert sources. Globally, the Sahara Desert exemplifies extensive aeolian systems, where dune fields cover over 20% of the region and contribute significantly to trans-Saharan dust transport to the Americas.Key factors influencing aeolian transport encompass surface roughness, vegetation cover, and deflation processes. Surface roughness from pebble armoring or crusts can increase the threshold shear velocity by 20–50%, reducing entrainment, while sparse vegetation stabilizes sediment through root networks and traps, limiting transport in semi-arid zones. Deflation hollows, or blowouts, arise where wind erodes unconsolidated material, forming depressions that can expand rapidly during high winds. These factors modulate transport efficiency, with vegetation removal—often due to overgrazing or drought—exacerbating erosion.Aeolian transport rates are generally lower than in fluvial systems due to air's reduced carrying capacity, with sand fluxes often ranging from 0.1–10 kg m⁻¹ h⁻¹ under moderate winds, but they pose significant risks for soil degradation. Historical events like the Dust Bowl in the 1930s American Great Plains illustrate this impact, where poor land management led to dust storms mobilizing millions of tons of topsoil annually, causing widespread ecological and economic damage. In modern contexts, aeolian processes contribute to global dust emissions of about 1–3 billion tons per year, affecting air quality and nutrient cycling.
Coastal Environments
In coastal environments, sediment transport is primarily driven by the interaction of waves, tides, and currents along shorelines, shaping dynamic features such as beaches and coastal landforms. Wave breaking serves as the main energy source, generating turbulence and shear stresses that mobilize sediments in both longshore and cross-shore directions. Littoral drift, or longshore sediment transport, occurs when obliquely approaching waves induce currents parallel to the shore, moving sand and gravel along the coast at rates that can reach millions of cubic meters per year in high-energy settings.[20] Cross-shore transport, meanwhile, involves onshore-offshore movement through swash and backwash processes, where waves carry sediment up the beach during the advancing swash phase and gravity pulls it seaward during backwash, often resulting in net onshore transport during fair weather and offshore during storms.[21][22]Key coastal features influenced by these processes include beaches, spits, and barrier islands, which form and evolve through the accumulation and redistribution of sediments. Beaches act as buffers, with sediment sorting creating coarser grains near the waterline and finer ones higher up the profile due to varying wave energies. Spits and barrier islands develop where longshore drift deposits sediment in elongated forms, often enclosing lagoons or bays, as seen in the recurved spits along the U.S. East Coast. Sediment bypassing around headlands allows transport to continue past rocky protrusions via wave refraction and rip currents, maintaining connectivity in littoral cells despite topographic barriers.[23][24]Several factors modulate sediment transport in these settings, including tidal regimes, storm events, and sea-level rise. Tides influence transport by altering water depths and current strengths, with stronger tidal currents enhancing resuspension in macrotidal coasts. Storms generate high waves and surges that accelerate erosion and offshore transport, redistributing large sediment volumes episodically. Sea-level rise exacerbates these dynamics by increasing inundation and wave energy reach, leading to shoreline retreat and reduced sediment retention on beaches. In deltaic systems like the Nile Delta, historical fluvial sediment inputs have built extensive coastal plains, but reduced river delivery due to upstream dams combined with sea-level rise has shifted transport toward net erosion, threatening barrier formation and wetland stability.[3][25][26]Human interventions, such as the construction of groins, significantly alter natural sediment flows by trapping littoral drift on the updrift side, leading to accretion there but inducing erosion downdrift as the sediment supply is interrupted. These structures, often built to protect infrastructure, disrupt the balance of littoral cells, sometimes necessitating further interventions like beach nourishment to mitigate downdrift impacts. In areas with multiple groins forming "groin fields," cumulative effects can exacerbate regional erosion, highlighting the need for integrated coastal management to preserve sediment budgets.[27][28][29]
Glacial Environments
In glacial environments, sediment transport occurs primarily through ice-mediated processes in polar and mountainous regions, where glaciers erode, entrain, and deposit vast quantities of rocky material. Glaciers act as powerful agents of erosion via abrasion and plucking at their bases, incorporating sediment that is then transported across landscapes. This transport is divided into supraglacial, englacial, and subglacial pathways, each contributing to distinct depositional patterns. Supraglacial transport involves debris accumulating on the glacier surface from sources such as rockfalls, landslides, and wind-blown material, often remaining angular and minimally sorted due to limited glacial modification.[30] Englacial transport embeds this debris within the ice mass, typically comprising less than 15% of the glacier's total volume but forming layered deposits up to 5 meters thick in some cases, which are released as the ice melts.[30] Subglacial transport, occurring beneath the ice via basal meltwater and traction, processes sediment through intense abrasion and fracturing, producing finer, rounder particles and facilitating high-concentration flows in channels.[30]Key landforms resulting from these processes include moraines and eskers, which record the history of glacial advance and retreat. Moraines form from accumulations of supraglacial and englacial till pushed into ridges at the glacier margins, such as lateral moraines along valley sides or terminal moraines at ice fronts, marking former glacier extents.[30] Eskers, by contrast, are sinuous ridges of sorted sand and gravel deposited by subglacial meltwater flowing through tunnels or braided channels at the ice base, often extending tens of kilometers and indicating former drainage pathways.[30] Outwash plains, or sandurs, develop beyond glacier termini where subglacial meltwater emerges, depositing vast sheets of stratified gravel and sand in braided river systems due to the high sediment concentrations—often exceeding 1-2 kg/L in meltwater streams—that overwhelm the water's carrying capacity.[31] For instance, the Greenland Ice Sheet exemplifies these dynamics, with its marine-terminating glaciers discharging approximately 0.911 Gt of sediment annually (1950–2009 average), contributing about 8% to the global ocean sediment flux and influencing coastal ecosystems through nutrient delivery tied to surface melt.[32]Influencing factors include basal sliding, where the glacier moves over its bed lubricated by meltwater, enhancing subglacial erosion and sediment entrainment rates, and variable melt rates driven by climate.[30] Seasonal variability is pronounced, with peak transport during summer melt seasons when increased temperatures accelerate ice ablation and runoff, leading to episodic sediment pulses.[32] Sediment transport rates in glacial settings can surpass those in fluvial environments during short bursts, particularly in jökulhlaups—sudden glacier outburst floods—where suspended sediment fluxes reach up to 4650 kg/s, accounting for over 50% of annual loads in just days, as observed in Iceland's Skaftá River.[33] These events underscore the pulsed nature of glacial sediment delivery, with implications for downstream landform evolution and, in regions like Greenland, indirect contributions to sea-level rise dynamics through enhanced glacial instability.[32]
Hillslope Processes
Hillslope processes encompass the gravity-induced downslope movement of soil and regolith on slopes, primarily through mechanisms such as soil creep, solifluction, and landslides, which collectively contribute to landscape evolution and sediment supply to fluvial systems.[34] These processes operate on non-channelized terrain, where sediment mobilization occurs diffusely across vegetated or bare slopes, often at rates ranging from millimeters to meters per year depending on environmental conditions.[35] Rainfall serves as a primary trigger by infiltrating soil, increasing pore pressure and reducing shear strength, thereby facilitating particle detachment and downslope transport.[36]Soil creep represents the slowest and most pervasive hillslope process, involving the gradual downslope displacement of soil particles through mechanisms like frost heaving, needle ice formation, bioturbation by roots and animals, and expansion-contraction due to wetting and drying cycles.[37] Rates of soil creep typically vary from 0.1 to 10 mm/year, with nonlinear dependence on slope angle, where steeper gradients accelerate transport by enhancing gravitational forces while limiting soil depth.[38] Vegetation cover significantly mitigates creep by anchoring soil through root systems and intercepting rainfall, reducing erosion efficiency by up to 80% in densely vegetated areas compared to bare slopes.[39] Soil saturation further influences creep, as increased moisture content lowers frictional resistance and promotes particle mobilization during episodic wetting events.[40]Solifluction, a form of creep dominant in periglacial environments, involves the slow flow of saturated soil layers above permafrost due to freeze-thaw cycles that generate high pore pressures and reduce soil cohesion.[37] This process forms characteristic landforms like solifluction lobes and sheets, with global rates averaging 5-50 mm/year, controlled by slope angle (optimal at 10-20°) and seasonal thawing depth.[41] In contrast, landslides occur as more rapid mass movements on steeper slopes (>25°), where intense rainfall rapidly saturates soil, elevating pore pressures and triggering shallow translational slides or deeper rotational failures.[42] These events can mobilize 10-100 m³ of sediment per landslide, with vegetation loss exacerbating susceptibility by diminishing root reinforcement.[43]Hillslope processes play a critical role in supplying sediment to river channels, though delivery is inefficient due to deposition in concave depressions or footslopes, with sediment delivery ratios typically ranging from 10-50% in vegetated landscapes.[44] In badlands, such as those in arid regions like the American Southwest, sparse vegetation and steep, unconsolidated slopes (>30°) promote accelerated creep and landsliding, yielding erosion rates of 1-10 mm/year and forming intricate networks of gullies.[45] Post-fire debris slides exemplify disturbance-enhanced transport, where wildfire removes vegetation cover, increasing soil hydrophobicity and saturation during subsequent rains, leading to slides that deliver pulses of fine sediment to channels at rates 10-100 times baseline levels.[46] Overall, these processes ensure a steady, though variable, flux of hillslope-derived sediment to fluvial environments, influencing downstream morphology and habitat.[35]
Debris Flows
Debris flows represent a distinct mode of sediment transport characterized by rapid, high-density mixtures of water and sediment moving down steep channels or slopes, typically with sediment concentrations exceeding 40% by volume. These flows differ from more dilute fluvial processes by their slurry-like consistency, where sediment particles ranging from clay to boulders interact to create a non-Newtonian rheology. The high sediment load imparts significant destructive potential, as the mixture can entrain additional material along its path, amplifying volume and momentum.[47]The dynamics of debris flows are governed by their Bingham fluid behavior, in which a yield stress must be exceeded for flow to initiate, arising from intergranular friction and matrix viscosity. This rheology enables the flows to propagate as discrete surges, with abrupt fronts that advance at speeds up to 10-40 m/s on slopes greater than 10 degrees. Common triggers include the sudden mobilization of landslide debris by intense rainfall or the breaching of temporary dams formed by rockfalls or glacial outbursts, leading to rapid transformation from static masses into flowing slurries. Surge propagation often involves front-heavy boulder concentrations that plow through obstacles, while the tail consists of finer material.[48][49]Debris flows pose severe hazards in mountainous regions, necessitating accurate path prediction and runout modeling to mitigate risks to infrastructure and communities. Numerical models such as RAMMS and DAN3D simulate flow propagation by incorporating Bingham rheology and topographic effects, allowing estimation of inundation extents up to several kilometers. In Alpine torrents like the Illgraben in Switzerland, recurrent debris flows have demonstrated runouts influenced by channel confinement and sediment supply, with historical events burying valleys under meters-thick deposits. These models highlight the role of surge dynamics in overtopping levees and diverting flows unpredictably.[50][51]Upon deceleration in gentler terrain, debris flows deposit as characteristic lobes at the flow front and levees along the margins, formed by segregation of coarser particles that ride superelevated at flow edges. These features result from frictional freezing at the flow boundary, where yield stress halts motion and builds transverse ridges up to 5-10 m high. The resulting landforms, such as fan-shaped accumulations, record the flow's volume and composition, with boulder-rich levees contrasting finer-grained interiors.[47]
Initiation of Motion
Fluid Forces and Stress Balance
The initiation of sediment motion requires a balance between fluid-induced forces that tend to dislodge particles and the resistive forces that hold them in place on the bed. The primary fluid forces acting on a sediment particle include drag and lift, which arise from the interaction between the flowing fluid and the particle surface. Drag force acts in the direction of the fluid flow and is proportional to the square of the relative velocity between the fluid and the particle, expressed as F_d = \frac{1}{2} C_d \rho_f A (u - u_p)^2, where C_d is the drag coefficient, \rho_f is fluid density, A is the projected area of the particle, u is fluid velocity, and u_p is particle velocity (typically zero for a resting particle).[52] Lift force, perpendicular to the flow direction, results from asymmetric pressure distributions around the particle, often enhanced by turbulent structures such as vortices in the boundary layer.[53] These forces are countered by the submerged weight of the particle, W' = (\rho_s - \rho_f) g V, where \rho_s is sediment density, g is gravitational acceleration, and V is particle volume, which acts downward and provides stability through contact with the bed.[54]In the turbulent boundary layer near the bed, shear stress generated by fluid turbulence transmits momentum to the particles, exerting a tangential force that promotes rolling or sliding. This bed shear stress, \tau_b, represents the average force per unit area due to both viscous and turbulent fluctuations, and it must exceed the particle's frictional resistance, which depends on the submerged weight and inter-particle contacts.[55] The balance is achieved when the component of fluid forces parallel to the bed overcomes the resistive friction, while the vertical lift component reduces the effective normal force, lowering the threshold for motion.[56] For non-spherical or partially buried particles, these forces interact with neighboring grains, distributing the stress across the bed surface.[53]In unsteady flows, such as those with accelerating or decelerating currents, additional inertial effects come into play through the virtual mass (or added mass) force. This force, F_{vm} = C_m \rho_f V \frac{D(u - u_p)}{Dt}, accounts for the acceleration of the surrounding fluid displaced by the particle, where C_m is the added mass coefficient (typically around 0.5 for spheres), effectively increasing the particle's inertia during rapid velocity changes.[57] Such effects are particularly relevant in turbulent suspensions, where particle acceleration lags behind fluid motion, amplifying the virtual mass contribution to the overall force balance.[58]A schematic illustration of the force vectors on a resting sediment particle typically depicts the submerged weight as a vertical downward arrow from the particle center, drag as a horizontal arrow aligned with the mean flow direction, and lift as an upward arrow perpendicular to the bed. These vectors sum to determine the net force, with equilibrium maintained until turbulent bursts tip the balance toward entrainment.[59]
Critical Shear Stress and Shields Parameter
The critical shear stress, \tau_{cr}, is defined as the minimum value of bed shear stress at which sediment particles begin to move, marking the threshold for initiation of motion in sediment transport processes.[60] This threshold arises from the balance between fluid-induced forces and particle resistance, as explored in early experimental work on granular beds.[61]To express this threshold in a dimensionless form applicable across varying flow and particle conditions, the Shields parameter \theta_{cr} was introduced by Albert F. Shields in 1936.[62] It is calculated as\theta_{cr} = \frac{\tau_{cr}}{(\rho_s - \rho) g d},where \rho_s is the density of the sediment particles, \rho is the density of the fluid, g is the acceleration due to gravity, and d is the characteristic diameter of the sediment grains.[62] For uniform, non-cohesive sand grains in turbulent flows, the Shields parameter typically adopts a value of approximately 0.047, representing the constant plateau on the Shields curve for high particle Reynolds numbers.The derivation of the Shields parameter stems from a force balance on individual particles at the bed surface, equating the destabilizing forces from fluiddrag and lift to the stabilizing forces of submerged particle weight and inter-particle friction.[61] Shields' original experiments involved systematic flume tests with varying grain sizes and flow velocities, plotting \theta_{cr} against the particle Reynolds number Re_* = \frac{u_* d}{\nu} (where u_* is the shear velocity and \nu is the kinematic viscosity) to yield the empirical Shields curve, which decreases from high values in laminar flow regimes to a near-constant level in turbulent conditions.[62] This curve encapsulates the transition from viscous-dominated to inertia-dominated entrainment mechanics.Variations in the Shields parameter account for deviations from ideal uniform, non-cohesive conditions. For coarser gravel particles, \theta_{cr} often exceeds 0.06 due to enhanced frictional resistance and reduced protrusion into the flow, as observed in steep, turbulent channels.[63] On slopes, experiments show that \theta_{cr} increases with bed inclination due to alterations in turbulent flow structures and increased particle stability from enhanced relative roughness and reduced near-bed velocities; this effect is quantified in adjustments to the Shields curve for gradients up to 0.1.[64] In mixed-size beds, hiding and exposure effects alter \theta_{cr} such that smaller grains are more easily entrained when exposed on the surface, while larger grains are sheltered within the bed matrix, leading to size-selective transport; these dynamics are commonly modeled using relative exposure factors.[65] For cohesive sediments like silt or clay, recent studies from the 2020s have proposed adjustments to the Shields framework, incorporating aggregate fractal structure and consolidation time, which elevate \theta_{cr} by factors of 2–10 compared to non-cohesive equivalents due to inter-particle bonding.[66]
Bed Shear Stress Components
Bed shear stress, denoted as \tau_b, represents the tangential force per unit area exerted by the fluid on the channel bed, driving sediment motion in alluvial systems. In steady, uniform open-channel flows, \tau_b is commonly estimated using the depth-slope product, given by \tau_b = \rho g h S, where \rho is the fluid density, g is gravitational acceleration, h is the flow depth, and S is the bed slope.[67] This formulation, originally proposed by DuBoys in 1879 and widely adopted in subsequent hydraulic engineering, assumes a hydrostatic pressure distribution and neglects secondary currents or non-uniformity effects, making it suitable for wide channels with mild slopes.[68]The shear velocity, u_*, provides a characteristic velocity scale for turbulent boundary layers and is defined as u_* = \sqrt{\tau_b / \rho}, linking the shear stress directly to flow dynamics near the bed.[9] This parameter is crucial for scaling turbulence and entrainment processes. In practice, u_* relates to the mean flow velocity U through the friction factor f, expressed as u_* = U \sqrt{f/8}, where f is the Darcy-Weisbach friction factor; this relation arises from integrating the momentum equation over the flow depth and equating it to the bedshear.[69]For unsteady flows, such as those during floods with accelerating or decelerating phases, the simple depth-slope product overestimates or underestimates \tau_b by ignoring inertial terms from the momentum equation. Adjustments incorporate local and convective accelerations, yielding a total effective shear stress \tau_b = \rho g h S - \rho \left( \frac{\partial h}{\partial t} + \frac{\partial (U h)}{\partial x} U \right), which accounts for hydrograph variability and can significantly alter transport rates during rising or falling limbs.[70] Laboratory and field studies during flood events demonstrate that these corrections are essential, as unsteadiness can increase bedload transport by up to 50% compared to steady assumptions.[71]The friction factor f itself is estimated differently for rough and smooth beds. For rough beds typical in gravel-bed rivers, logarithmic velocity profiles apply, where the law of the wall gives U / u_* = (1/\kappa) \ln (h / k_s) + 8.5, with \kappa \approx 0.4 as von Kármán's constant and k_s as the roughness length; solving yields f \approx [2.03 \ln (h / k_s)]^{-2}.[72] In contrast, Manning's roughness coefficient n, empirically derived for natural channels, relates to f via f = 8 g n^2 / R^{1/3}, where R is the hydraulic radius approximating h for wide flows; values of n range from 0.025 to 0.040 for gravel beds, increasing with bed material size and irregularity.[73] For smooth beds, such as those with fine sediments, the friction factor shifts to a viscous sublayer-dominated regime, following f \approx 0.316 / \mathrm{Re}^{1/4} from the Blasius correlation, where \mathrm{Re} = U h / \nu and \nu is kinematic viscosity.[9] These methods enable practical computation of \tau_b for Shields parameter applications in threshold predictions.
Particle Reynolds Number
The particle Reynolds number, denoted as \mathrm{Re}_*, characterizes the relative importance of inertial to viscous forces in the flow around individual sediment particles resting on a bed surface. It is defined as\mathrm{Re}_* = \frac{u_* d}{\nu},where u_* is the shear velocity of the fluid, d is the diameter of the particle, and \nu is the kinematic viscosity of the fluid.[74] This parameter is crucial for understanding how flow conditions affect particle stability and motion initiation, as it determines the dominant mechanisms of fluid-particle interaction.[61]The value of \mathrm{Re}_* delineates distinct flow regimes around the particle. In the low-\mathrm{Re}_* regime (typically \mathrm{Re}_* < 2), viscous forces dominate, resulting in the Stokes regime where drag arises primarily from laminar viscous shear layers enveloping the particle.[60] For intermediate \mathrm{Re}_* (roughly 2 to 500), the flow transitions to a regime where both viscous and inertial effects contribute significantly to drag. At higher intermediate values (500 < \mathrm{Re}_* < $10^5), the Newtonian regime prevails, with inertial forces controlling a separated wake and form drag becoming prominent. For very high \mathrm{Re}_* (> $10^5), turbulence in the wake further modifies the flow structure.[75] These regimes influence the overall force balance on particles, transitioning from viscosity-dominated stability at low speeds to inertia-driven entrainment at higher flows.The particle Reynolds number directly impacts the drag coefficient C_D, which quantifies the fluid drag force relative to dynamic pressure and particle area. For smooth spheres, C_D varies systematically with \mathrm{Re}_*: in the Stokes regime, C_D \approx 24 / \mathrm{Re}_*, decreasing sharply with increasing \mathrm{Re}_*; in the transitional regime, it follows a steeper decline; and in the Newtonian regime, it stabilizes at approximately 0.4 to 0.5, reflecting constant form drag dominance.[75] Empirical curves plotting C_D against \mathrm{Re}_* for spheres provide a foundational reference, though natural sediment grains exhibit higher C_D values due to angularity and roughness.[76]In applications to sediment transport, \mathrm{Re}_* adjusts the Shields curve, which relates the critical Shields parameter (dimensionless shear stress) to flow and particle properties for predicting motion initiation. Larger particles or lower-viscosity fluids (e.g., air versus water) yield higher \mathrm{Re}_*, shifting along the curve to account for regime-specific thresholds and enabling predictions across diverse conditions like gravel in rivers or sand in winds.[77]
Entrainment Modes
Bedload Entrainment
Bedload entrainment involves the dislodgement and near-bed movement of sediment particles through traction, where particles maintain frequent contact with the bed surface. This process is dominated by three primary mechanisms: rolling, in which particles rotate along the bed; sliding, where particles shift laterally under shear forces; and saltation, characterized by intermittent hops typically on the order of several particle diameters in height.[78][79] These motions occur within a thin "bed layer" approximately two particle diameters thick, distinguishing bedload from higher suspension.[78]Bedload entrainment predominates for coarse particles such as sand and gravel, which exhibit high Rouse numbers greater than 2.5, indicating that settling velocities greatly exceed turbulent diffusion rates near the bed and limit vertical suspension.[80]Entrainment initiates when fluid forces overcome particle resistance, often referenced to critical shear stress thresholds established in foundational studies.[78] For particles larger than 10 mm, motion remains confined to the bed layer due to their weight overpowering hydrodynamic lift.[78]Particle trajectories in saltation follow curved paths with lengths and heights scaling proportionally to the shear velocity, ensuring intermittent bed contact that sustains transport without full suspension.[81] Experimental observations from flume studies using pebble tracking reveal size-selective movement, where larger pebbles exhibit longer resting periods interrupted by bursts of rolling or hopping, influenced by flow turbulence.[82] Additionally, interactions with bedforms such as ripples modulate entrainment by altering local shear and creating preferential pathways for particle hops over ripple crests.[83] These seminal insights, building on Einstein's probabilistic entrainment concepts and Bagnold's saltation dynamics, underscore the stochastic yet mechanically driven nature of bedload motion.[78][81]
Suspended Load Entrainment
Suspended load entrainment involves the lifting of fine sediment particles into the water column by turbulent fluctuations in the flow, where coherent structures such as bursts and sweeps near the bed provide the necessary upward forces to overcome gravitational settling. These turbulent eddies, generated by shear in the boundary layer, disrupt the quiescent viscous sublayer adjacent to the bed, allowing particles to be ejected into the flow and subsequently diffused throughout the water column by ongoing turbulence. This mechanism is particularly effective for particles where the turbulent kinetic energy exceeds the particle's settling energy, enabling prolonged suspension without frequent recontact with the bed.Entrainment of suspended load requires conditions of elevated turbulence intensity, typically associated with high flow velocities and shear stresses, combined with low particle settling velocities characteristic of silt- and clay-sized sediments (generally <0.0625 mm). In such flows, the ratio of turbulent diffusion to settling—often quantified through dimensionless parameters like the Rouse number—favors suspension when turbulence dominates, preventing rapid deposition. Wash load, a subset of suspended load, consists of ultra-fine particles (e.g., clays <0.002 mm) that remain perpetually suspended due to their negligible settling velocity relative to the flow's turbulent mixing, deriving primarily from upstream sources rather than local bed entrainment.[84]The vertical distribution of suspended sediment concentration follows a power-law profile with increasing height above the bed, reflecting the diminishing influence of near-bed turbulence and the increasing dominance of settling farther upstream in the flow. Near the bed, concentrations are highest, transitioning to more uniform profiles in the outer flow region for very fine particles. The reference concentration at a nominal height (typically 0.5 to 1 times the particle diameter above the bed) anchors this distribution, serving as a boundary condition derived from incipient motion criteria or bedload flux estimates under the given hydrodynamic conditions.[85]Turbulent mixing in suspended load transport is parameterized through the eddy diffusion coefficient, which quantifies the vertical spreading of particles by eddies and is approximately 0.15 times the product of the friction velocity and flow depth (ε_s ≈ 0.15 u_* h) in steady, uniform flows. This coefficient is often equated to the fluid's eddy viscosity in simplified models, balancing diffusive flux against settling in the governing advection-diffusion equation, though adjustments account for particle inertia and flow stratification effects.[85]
Rouse Number and Settling Velocity
The Rouse number, denoted as P, is a dimensionless parameter that characterizes the mode of sediment transport by comparing the settling velocity of particles to the upward turbulent diffusion in the flow. It is defined as P = \frac{w_s}{\kappa u_*}, where w_s is the particle settling velocity, \kappa is the von Kármán constant (approximately 0.4), and u_* is the shear velocity of the flow.[80] This parameter, originally derived from considerations of turbulent diffusion and gravitational settling in steady flows, provides criteria for distinguishing transport modes: values of P < 0.8 indicate wash load (full suspension); $0.8 \leq P < 1.2 for fully suspended load; $1.2 \leq P \leq 2.5 for mixed suspended and bedload regimes; and P > 2.5 for dominant bedload transport.[80] Note that thresholds may vary slightly across studies, but these are standard for fluvial contexts.The settling velocity w_s, a key input to the Rouse number, represents the terminal fall speed of a sediment particle under gravity in still fluid and is essential for assessing suspension potential. For fine particles where the particle Reynolds number \text{Re}_p = \frac{w_s d}{\nu} < 1 (with d as particle diameter and \nu as kinematic viscosity), Stokes' law applies, givingw_s = \frac{g (\rho_s - \rho) d^2}{18 \nu},where g is gravitational acceleration, \rho_s is particle density, and \rho is fluid density; this assumes spherical particles in laminar flow conditions.[86] For coarser grains across a wider range of Reynolds numbers, the Ferguson-Church equation provides a general explicit form:w_s = \frac{\sqrt{2 R g d}}{C_1 \sqrt{\nu} + \sqrt{0.75 C_2 R g d}},with submerged specific gravity R = (\rho_s - \rho)/\rho, and empirical constants C_1 \approx 20 and C_2 \approx 1 for natural grains; this bridges viscous (Stokes) and turbulent drag regimes.[87] Recent updates for non-spherical particles, such as those in carbonate sands, incorporate shape-dependent drag coefficients to refine w_s estimates, improving accuracy for irregular coastal sediments.[76]In suspended transport, the Rouse number governs the vertical distribution of sediment concentration, assuming equilibrium between turbulent diffusion and settling. The classic Rouse profile for concentration C(z) at height z above the bed is\frac{C(z)}{C_a} = \left( \frac{a}{z} \right)^P,where C_a is the reference concentration at a reference height a (typically near the bed), and P determines the profile's steepness—lower P yields more uniform suspension throughout the water column.[88] This logarithmic form arises from solving the advection-diffusion equation under steady, uniform turbulent flow.The Rouse framework assumes steady, uniform open-channel flow with constant eddy diffusivity, which limits its direct application to non-steady or stratified conditions where turbulence is suppressed.[89] In coastal environments with waves, recent models from the 2020s adjust the Rouse number by incorporating oscillatory boundary layers and wave-induced turbulence, enabling better prediction of suspension under combined wave-current flows.
Transport Quantification
Bedload Transport Rates
Bedload transport rates describe the volumetric flux of coarse sediment particles that roll, slide, or saltate along the near-bed layer without entering significant suspension, typically quantified as q_b, the volume of sediment transported per unit time per unit channel width.[90] This flux depends on flow hydraulics, sediment properties, and bed conditions, with empirical formulas providing the primary means of prediction for gravel and coarser materials in rivers and flumes.[91]One of the most influential empirical relations is the Meyer-Peter and Müller (1948) formula, derived from flume experiments with uniform coarse sand and gravel under plane-bed conditions.[90] The equation expresses the dimensionless bedload transport rate \Phi = q_b / \sqrt{(s-1) g d^3}, where s = \rho_s / \rho is the relative density of sediment to fluid, g is gravitational acceleration, and d is grain diameter, as:\Phi = 8 (\theta - \theta_{cr})^{3/2}where \theta = \tau_b / [(s-1) \rho g d] is the Shields parameter for bed shear stress \tau_b, and \theta_{cr} is the critical value for entrainment initiation.[90] An equivalent dimensional form is q_b^{1/2} = 8 (\theta - \theta_{cr})^{3/2} (s-1)^{-1/2} \sqrt{(s-1) g d^3}.[67] This relation assumes a power-law dependence on excess shear stress and has been widely adopted for its simplicity, though it incorporates a form-drag correction to partition total shear stress between grain and bedform resistance.[90]Einstein's 1950 model introduced a probabilistic framework to bedload transport, treating particle entrainment as a stochastic process where the probability of motion depends on the fluctuating lift forces exceeding submerged particle weight.[78] The approach derives the bedload intensity i_b = q_b / (c_a d \sqrt{(s-1) g d}), where c_a is a roughness layer concentration, from an exponential probability distribution of flow velocities near the bed, validated against flume data for a range of grain sizes.[78] This probabilistic method accounts for intermittency in particle hops but requires integration over velocity profiles, making it computationally intensive compared to deterministic formulas like Meyer-Peter and Müller.[92]For gravel-bed rivers with mixed sediment sizes, the Wilcock and Crowe (2003) surface-based model predicts fractional transport rates relative to the bed surface grain-size distribution, addressing selective entrainment and hiding effects.[93] The dimensionless rate for size fraction i is given by:W_i^* = \begin{cases}
0.002 \left( \frac{\tau^*}{\tau_{ri}^*} \right)^{7.5} & \tau^* < 1.35 \tau_{ri}^* \\
14 \left( 1 - \frac{0.894}{\sqrt{\tau^* / \tau_{ri}^*}} \right)^{4.5} & \tau^* \geq 1.35 \tau_{ri}^*
\end{cases}where W_i^* = (s-1) g q_{bi} / (F_i u_*^3), \tau^* = \tau_b / [(s-1) \rho g D_{sm}] is the Shields parameter referenced to median surface grain size D_{sm}, u_* is shear velocity, F_i is the surface fraction of size i, and \tau_{ri}^* is a size-dependent reference shear stress that varies with sand content.[94] Developed from 48 flume runs with sand-gravel mixtures (6–34% sand), it captures reduced gravel mobility in sand-rich beds and equal mobility under high flows.[93]These formulas have been validated against extensive flume datasets, where Meyer-Peter and Müller predicts transport within a factor of 3–5 of measured rates for uniform sediments under moderate Shields parameters (\theta > 0.1), though it overestimates at low flows and underestimates for non-uniform beds.[95] Einstein's probabilistic predictions align well with flume observations of intermittent transport for sands but deviate at high rates due to unmodeled correlations in particle jumps.[78] Wilcock and Crowe performs robustly in mixed flume experiments, matching 80–90% of data within a factor of 2, particularly for gravel fractions.[96] However, limitations arise on steep slopes (S > 0.02), where Meyer-Peter and Müller underpredicts due to unaccounted turbulence and acceleration effects; recent adjustments incorporate 2D shear stress distributions or slope-dependent corrections to improve accuracy by 20–30% in steep flume tests.
Suspended and Wash Load Rates
Suspended load refers to sediment particles held in the water column by turbulence, transported at velocities comparable to the fluid flow. The volumetric transport rate per unit width, q_s, is computed as the vertical integral of the product of local sediment concentration C(z) and flow velocity u(z) from the reference height above the bed to the water surface:q_s = \int_a^h C(z) \, u(z) \, dzwhere h is the flow depth and a is the reference height (typically 1-2 times the particle diameter). This formulation captures the flux through the water column, distinguishing it from near-bed rolling or saltation.[97]To approximate this integral, the concentration profile is often modeled using the Rouse equation, which assumes a balance between turbulent diffusion and gravitational settling. The Rouse profile expresses C(z) relative to a reference concentration C_a at height a:\frac{C(z)}{C_a} = \left( \frac{h - z}{z} \cdot \frac{a}{h - a} \right)^{P}where P = \frac{w_s}{\kappa u_*} is the Rouse parameter, with w_s as the particle settling velocity, \kappa \approx 0.4 the von Kármán constant, and u_* the bed shear velocity. The velocity profile u(z) is typically assumed to follow the logarithmic law:u(z) = \frac{u_*}{\kappa} \ln \left( \frac{z}{z_0} \right)with z_0 as the roughness length. Analytical integration of these profiles yields q_s \approx C_a \cdot U \cdot h \cdot \beta, where U is the depth-averaged velocity and \beta is a correction factor (often 1.2-2.0 depending on P) accounting for the non-uniform distribution. The reference concentration C_a is estimated from bedload flux or empirical relations based on shear stress excess.[97][98]Wash load constitutes the finest fraction of suspended load, comprising particles smaller than 0.0625 mm (silt and clay sizes) that remain nearly uniformly distributed in the water column due to negligible settling and minimal interaction with the bed. Unlike bed-material suspended load, wash load originates primarily from upstream watershederosion rather than local channelentrainment, and its transport rate depends more on sediment supply than flow capacity. In many lowland rivers with fine sediments, wash load often comprises a large portion of the total suspended sediment flux.[99][100]Computing wash load rates follows the same integral approach but uses a near-uniform concentration profile (low Rouse parameter P < 0.25) and observed or supply-limited C_a, often bypassing Rouse corrections since settling is insignificant. Velocity integration employs the log-law profile, but the flux simplifies to approximately q_w \approx C_w \cdot U \cdot h, with C_w as the nearly constant wash concentration measured via depth-integrated sampling. Post-dam scenarios, such as in the Yangtze River after the 2003 Three Gorges Dam impoundment, have amplified the relative role of wash load; reservoirs trap coarser bed material, increasing the proportion of finer particles in the downstream suspended flux amid an overall sediment reduction of about 76%.[101][97] Recent studies as of 2025 highlight the use of machine learning models to better predict these shifts in suspended load composition under changing flow regimes.[102]
Total Load Models
Total load in sediment transport refers to the combined flux of bedload and suspended load, expressed as q_t = q_b + q_s, where q_t is the total volumetric sediment transport rate per unit width, q_b is the bedload rate, and q_s is the suspended load rate.[103] This integrated approach simplifies prediction by avoiding separate calculations for each mode, relying instead on flow and sediment parameters to capture overall transport capacity. Such models are particularly useful in alluvial streams where both modes contribute significantly, drawing on principles of stream power and energy balance.[103]A foundational total load model is the Engelund-Hansen equation, developed from flume experiments with uniform sands (0.19–0.93 mm).[103] It equates the work done by flow drag to the potential energy of lifted particles, yielding:q_t = 0.05 \, g^{1/2} \, \frac{C^{5/2} S^{3/2}}{\Delta \, d_{50}^{1/2}}where g is gravitational acceleration, C is the Chézy coefficient, S is the bed slope, \Delta = ( \rho_s - \rho ) / \rho is the relative submerged density (with \rho_s and \rho as sediment and fluid densities), and d_{50} is the median grain diameter. This explicit stream power-based formula performs well for sand-bed rivers but assumes equilibrium conditions and uniform sediment.[103]The Ackers-White model (1973) offers another hybrid empirical approach for total load, calibrated on flume data for uniform gradations from sand to fine gravel.[104] It employs regime theory, distinguishing lower regime (bedload dominant), transition, and upper regime (suspended load dominant) based on dimensionless grain size d_{gr} and mobility parameters, with transport intensity G_{gr} scaling the rate relative to critical conditions. The core relation is q_t / (V D) = G_{gr} (S_s - 1) d_{50} / [D (u_*/V)^n], where V is mean velocity, D is flow depth, S_s is specific gravity, u_* is shear velocity, and n is a regime-dependent exponent.[105] This framework accounts for flow regime shifts and is widely applied in gravel-bed contexts.[104]Despite their influence, these models have limitations, notably overpredicting transport for fine sediments due to inadequate treatment of cohesion or wash load dynamics.[106] For non-uniform sediments, they often fail to capture hiding-exposure effects among grain sizes, leading to errors in mixed sand-gravel beds. Recent revisions, as reviewed in 2023 studies, incorporate multi-fraction adjustments and transport length concepts to better handle graded mixtures, improving predictions in natural rivers.[102]
Hjulström–Sundborg Diagram
The Hjulström–Sundborg diagram is an empirical log-log plot of mean flow velocity against sediment grain diameter, illustrating the thresholds for erosion, entrainment, transport, and deposition in fluvial environments. It divides the graph into distinct zones: below the erosion curve, deposition dominates; between the erosion and transport curves, net sediment movement occurs; and above the transport curve, suspension and wash load prevail. The diagram highlights how the minimum velocity for erosion varies inversely with grain size for non-cohesive particles, peaking for very fine and very coarse sediments due to cohesion and inertial effects, respectively.[107][108]Developed by Filip Hjulström in 1935 through flume experiments and field observations on the River Fyris in Sweden, the original curve focused on riverine conditions and emphasized the lowest erosion velocities for medium sands around 0.1 mm diameter, requiring flows of approximately 20–30 cm/s. Hjulström's work established the foundational erosion threshold line, demonstrating that larger gravels demand progressively higher velocities to initiate motion. In 1956, Åke Sundborg refined and expanded the diagram in his study of the River Klarälven, incorporating data for finer particles down to clay sizes and adding an upper boundary for sustained transport as well as a deposition curve set below the erosion line. Sundborg's modifications accounted for cohesive forces in silts and clays, where erosion velocities rise sharply for particles finer than 0.06 mm due to increased interparticle bonding.[107][108]The diagram's interpretation reveals key hydraulic behaviors, such as the hysteresis effect during flood cycles: on the rising limb of a hydrograph, erosion initiates at the lower curve, but on the falling limb, deposition begins at the higher velocity threshold, leading to temporary bed aggradation before full stabilization. This inverse relationship between grain size and critical velocity underscores why mixed-bed rivers often sort sediments, with fines susceptible to suspension at lower flows and coarser fractions requiring turbulent bursts for entrainment. The critical velocities align conceptually with the Shields parameter for non-cohesive thresholds, though the diagram simplifies depth and slope influences.[107][108][109]Recent adaptations have integrated turbulence fluctuations and wave effects for coastal applications, deriving probabilistic thresholds from large-eddy simulations to better predict intermittent motion in oscillatory flows. For instance, updates for nearshore environments adjust the curves upward for turbulence-enhanced erosion in sands under breaking waves. However, the diagram's limitations become evident for cohesive sediments, where it overestimates erosion velocities for clays due to underrepresenting electrochemical and biological bindings, often requiring supplementary models for mixed beds.[109][110]
Modeling Approaches
Empirical Formulas
Empirical formulas provide foundational algebraic expressions for predicting sediment transport rates and associated bed morphodynamics, derived from physical principles and calibrated against experimental data. These semi-empirical relations simplify complex interactions between fluid flow, sediment properties, and bed response, enabling practical applications in river engineering and geomorphology. Unlike more advanced numerical models, they rely on readily measurable parameters such as shear stress, flow energy, and grain characteristics to estimate transport without solving full hydrodynamic equations.[111]A cornerstone of empirical approaches is the Exner equation, which governs the temporal evolution of bed elevation due to sediment continuity. It states that the rate of change in bed elevation \eta is proportional to the spatial gradient in volumetric sediment transport rate q, accounting for bed porosity p:\frac{\partial \eta}{\partial t} = -\frac{1}{1-p} \frac{\partial q}{\partial x}This relation, originally formulated by Felix Exner in 1925, assumes conservation of sediment mass and links local erosion or deposition to upstream-downstream transport imbalances, forming the basis for morphodynamic modeling in alluvial channels.[112]One of the earliest empirical formulas for bedload transport was proposed by Paul Du Boys in 1879, emphasizing the role of bed shear stress in initiating and sustaining sediment movement. The Du Boys equation expresses the volumetric bedload transportrate per unit width q_b as proportional to the excess shear stress beyond the critical threshold for entrainment \tau_{cr}:q_b = C \tau_b (\tau_b - \tau_{cr})Here, \tau_b is the bed shear stress, and C is an empirical coefficient dependent on sediment characteristics. This shear stress-based approach marked a shift toward rational prediction of transport in gravel-bed rivers, influencing subsequent formulas by highlighting the importance of flow-induced tractive forces.[113]In contrast, Ralph Bagnold's 1956 energetics framework views sediment transport as a dissipative process driven by the work rate of the flow, particularly applicable to total load including both bedload and suspended components. Bagnold posited that the immersed-weight transport rate is proportional to the product of flow energy flux and the near-bed concentration of sediment immersed weight:i_b \propto \left( \frac{\rho_s - \rho}{\rho} g \right) c_b u_b h \cdot ewhere i_b is the immersed bedload transport rate, \rho_s and \rho are sediment and fluid densities, g is gravity, c_b is the volumetric concentration at the bed, u_b is the bed shear velocity, h is flow depth, and e represents the efficiency of energy transfer to sediment motion. This model, developed from flume experiments with sands and gravels, underscores the conversion of fluid kinetic energy into sediment work, offering a physically grounded alternative to purely stress-based methods.[114]Despite their simplicity, empirical formulas often require calibration to reconcile discrepancies between laboratory and field conditions, where lab-derived coefficients tend to underpredict real-world transport rates by factors of 2 to 10 due to idealized uniform flows and sediments. Field measurements reveal higher variability from turbulenceintermittency and unsteadiness, necessitating adjustments like increased efficiency factors in Bagnold's model. Applicability to non-uniform beds further complicates predictions, as surface grain hiding and protrusion effects alter critical shear stresses, reducing entrainment of buried finer particles while exposing coarser ones; formulas like Du Boys perform better on uniform substrates but demand modifications, such as hiding factors, for heterogeneous gravel mixtures. For instance, refinements like the Meyer-Peter and Müller equation build on Du Boys-style shear stress relations but incorporate gravel-specific adjustments for better field alignment.[115][116][117]
Numerical and Recent Modeling Advances
Numerical models for sediment transport have evolved to simulate complex hydrodynamic-sediment interactions across scales. One-dimensional (1D) models, such as the HEC-RAS sediment module developed by the U.S. Army Corps of Engineers, solve unsteady flow equations coupled with sediment transport functions to predict bed evolution and deposition in rivers and channels. This module incorporates quasi-unsteady and fully unsteady flow regimes, allowing for the analysis of sediment routing under varying discharge conditions, with applications in flood control and reservoir management. For higher-dimensional simulations, computational fluid dynamics (CFD) approaches like the TELEMAC system provide 2D and 3D capabilities, integrating turbulence models (e.g., k-ε or LES) with sediment transport via modules such as Sisyphe or GAIA.[118] TELEMAC-3D resolves the Navier-Stokes equations under hydrostatic or non-hydrostatic assumptions, enabling detailed coupling of flow turbulence with erosion, deposition, and bedload/suspended load dynamics in coastal and estuarine environments.[119]Recent advances incorporate machine learning (ML) to enhance parameter estimation in traditional models, addressing limitations in empirical formulations. Artificial neural networks (ANNs) have been applied to predict sediment incipient motion by improving on classical curves through nonlinear hydraulic dependencies. A 2023 ANN model for river bedload transport, informed by over 8,000 measurements, demonstrated superior performance in variable flow regimes compared to physics-based predictors.[120] In wave-current interactions, coupled models driven by CMIP6 climate projections forecast reductions in longshore sediment transport, with simulations indicating up to double the decrease compared to CMIP5 scenarios under high-emission pathways, potentially halving transport rates in vulnerable coastal zones by 2100.[121]For cohesive sediments, flocculation processes—where fine particles aggregate under turbulent shear—pose unique modeling challenges, as they alter settling velocities and transport efficiency. A 2023 study parameterized floc yield strength in population balance models to better represent breakup and reformation in turbulent flows, revealing that temporal statistics of floc size distributions significantly influence net deposition rates.[122] These models predict reservoirsiltation with global annual storage losses estimated at approximately 0.35-0.4% of capacity, equivalent to roughly 20-25 km³ of sediment accumulation per year, based on aggregated data from thousands of large dams.[123] Such predictions underscore the need for integrated approaches to mitigate long-term water security risks.[124] In 2025, further advances include the sedInterFoam 1.0 model, which extends Eulerian multiphase simulations to air-water-sediment interactions for improved nearshore predictions, and the AHMS-SED framework for regional-scale erosion and transport integration.[125][126]Despite these progresses, numerical modeling faces persistent challenges, particularly in handling unsteady flows where non-equilibrium transport leads to uncertainties in erosion predictions during floods or tidal cycles.[127] Basin-scale simulations further require seamless integration with geographic information systems (GIS) to incorporate spatially variable inputs like land-use changes and watershedhydrology, though data resolution and computational demands often limit scalability.[128] Ongoing efforts focus on hybrid ML-physical models to reduce these uncertainties while maintaining physical interpretability.
Applications and Mitigation
Geomorphic and Environmental Applications
Sediment transport plays a pivotal role in geomorphic processes that shape landscapes over geological timescales. In river systems, it facilitates the construction of deltas through the deposition of suspended and bedload sediments as fluvial energy diminishes near coastal zones. For instance, the balance between riverine sediment supply, wave reworking, and tidal currents determines the primary morphology of deltas, with high sediment flux leading to progradational bird's-foot or lobate forms.[129] Conversely, in upstream reaches, sediment-laden flows drive canyon incision by abrading bedrock channels, particularly where increased discharge and stream power enhance erosive capacity during tectonic uplift or climatic shifts.[130] Globally, prior to widespread damconstruction, rivers delivered approximately 15–20 × 10^9 tons of sediment annually to the oceans, forming a critical long-term budget that sustains coastal landforms and continental margins.[131]Environmentally, sediment transport vectors nutrients and pollutants across aquatic ecosystems, exacerbating eutrophication and contamination in downstream waters. Fine particles adsorb phosphorus, nitrogen, heavy metals, and organic contaminants, facilitating their redistribution from upland sources to rivers, estuaries, and coastal seas, where they can trigger algal blooms or bioaccumulate in food webs.[132] Habitat alteration arises from excessive fine sediment deposition, which infiltrates gravel beds and reduces interstitial oxygen flow, impairing salmonid embryo survival; for example, egg mortality in redds can reach up to 86% due to siltation.[133] Such changes disrupt spawning grounds, as seen in Pacific Northwestrivers where siltation clogs redds, leading to decreased juvenile recruitment and altered benthic communities.[133]Links to climate change amplify sediment dynamics in polar and coastal regions. Glacierretreat, driven by warming, exposes fresh bedrock and increases subglacial sediment yields through enhanced meltwater flushing, with proglacial fluxes rising by factors of up to 19 in modeled retreating systems.[134] In the Arctic, studies from Svalbard indicate heightened paraglacial redistribution of tills, elevating riverine loads and coastal sedimentation rates.[135] Sea-level rise, projected at 0.3–1.0 m by 2100 under various scenarios, intensifies coastal erosion by elevating wave base and promoting offshore sediment transport, resulting in shoreline retreat rates of 1–10 m/year on low-gradient barriers and altered longshore budgets.[136][137]Monitoring these processes relies on remote sensing techniques, particularly LiDAR, which quantifies bed-level changes with centimeter-scale precision over large river reaches. Airborne and terrestrial LiDAR surveys detect erosion hotspots and aggradation volumes by differencing digital elevation models, revealing sediment transport patterns during floods or seasonal variations. For example, repeat LiDAR acquisitions in gravel-bed rivers have mapped bed scour depths exceeding 0.5 m, informing geomorphic budgets without invasive sampling.[138][139]
Engineering Mitigation Techniques
Engineering mitigation techniques for sediment transport focus on structural and operational strategies to prevent or minimize sediment accumulation at water infrastructure, thereby preserving functionality and reducing maintenance costs. These methods are essential in riverine, reservoir, and coastal settings where sediment ingress can impair operations such as hydropower generation, water supply, and navigation. Key approaches include protective devices at intakes, sediment removal in reservoirs, and coastal stabilization measures, often informed by hydraulic modeling to optimize design and performance.Intake protection systems are designed to exclude bedload and suspended sediments from entering pumps, turbines, or diversion channels, particularly in rivers with high sediment loads. Vortex drop structures, also known as vortex intakes, induce a swirling flow that separates heavier sediments from water by centrifugal force, directing clean water downward through a central shaft while sediments settle in the outer annulus. These structures are effective for managing coarse bedload particles, with designs based on empirical hydraulic criteria to ensure self-cleaning velocities and minimal head loss, as detailed in comprehensive monographs on vortex-flow intakes. Screens, typically coarse bar racks or traveling screens, provide a physical barrier to trap larger debris and bedload, often combined with upstream settling basins to reduce sediment loading; for instance, positive barrier screens in diversion systems can achieve near-total exclusion of particles greater than 1 mm when paired with sediment traps. Bedload traps, such as vortex tubes or sediment diverters, reroute coarse sediments away from intakes by exploiting helical flow patterns, demonstrating up to 90% diversion efficiency in field tests on alpine torrents. These techniques reference bedload transport rates to size traps appropriately, ensuring they handle peak sediment fluxes without overflow.Reservoir management strategies address the global issue of sedimentation, which causes an annual storage capacity loss of 0.8–1% worldwide, threatening water security and infrastructure longevity. Flushing operations involve controlled drawdown to mobilize and expel deposited sediments through bottom outlets, restoring capacity and maintaining downstream sediment supply; successful implementations, such as in Philippine reservoirs, have recovered up to 20% of lost volume per event when timed with high flows. Dredging, using mechanical excavators or hydraulic suction, targets localized deposits near dams or outlets, with economic assessments showing viability for reservoirs losing over 1% capacity annually, as in the case of the Mrica hydroelectric facility where dredging costs were offset by extended operational life. Sediment routing models simulate deposition patterns and predict flushing efficacy, incorporating one-dimensional or two-dimensional hydrodynamic equations to guide operations; recent applications in wide reservoirs have optimized sluicing sequences to achieve sustainable sediment removal rates exceeding 50,000 m³ per event.In coastal environments, engineering techniques mitigate sediment transport driven by waves and currents, focusing on stabilizing shorelines against erosion and accretion imbalances. Breakwaters, constructed as offshore or nearshore barriers, dissipate wave energy and interrupt longshore drift, reducing sediment bypassing to downdrift areas; for example, 85 segmented breakwaters at Holly Beach, Louisiana, have created sheltered pockets that promote local accretion without excessive downdrift erosion. Beach nourishment replenishes eroded sediments by placing sand from offshore or upland sources, countering longshore losses and widening beaches; projects like those by the U.S. Army Corps of Engineers have sustained profiles for 5–10 years, with volumes typically ranging from 100,000 to 1,000,000 m³ per site. Balancing longshore drift often employs groins or jetties to trap updrift sediments while minimizing downdrift deficits, with designs calibrated to match natural transport rates of 10⁴–10⁶ m³/year along many U.S. coasts.Recent advances emphasize integrated and sustainable approaches, including nature-based solutions and treatments for contaminated sediments. Vegetated buffers, such as riparian wetlands or seagrass meadows, act as natural sediment traps by increasing flow resistance and promoting deposition, reducing transport by up to 70% in coastal and riverine settings; these solutions enhance ecosystem services while mitigating erosion, as evidenced in European Interreg projects. For contaminated sediments, ex-situ methods involve dredging followed by off-site processing, such as thermal desorption or bioremediation, to immobilize or degrade pollutants before reuse; a 2023 review highlights stabilization/solidification as a widely adopted technique, achieving over 90% contaminant reduction in treated volumes exceeding 100,000 m³ annually at major sites. As of 2025, emerging applications include AI-enhanced numerical models for real-time sediment transport prediction in restoration projects.[140]