Antecedent composition-morphology states shape storm-driven adjustment in an idealized mixed-sediment estuary

Storms can rapidly reorganize estuarine beds, but field records rarely isolate the interacting roles of waves, tides, sediment availability and antecedent morphology. We used a 2D, coupled wave-current-morphodynamic model of an idealized estuary inspired by the Changjiang Estuary to examine storm adjustment from three composition-morphology equilibrium states. The states developed during otherwise identical 1,000-year spin-ups with initially uniform cohesive-mud fractions of 10%, 20% and 30%. We then imposed 3-, 4- and 5-m storms and tested one matched spring-neap timing contrast. Storm forcing concentrated erosion at the exposed delta front, where wave-energy flux declined sharply towards the mouth. By the end of the six-day forcing, seaward delta area had decreased by 6.1%, 4.2% and 4.7% relative to the respective pre-storm areas in the 10%, 20% and 30% states. The mud-richer antecedent states developed higher and more extensive suspended-mud concentrations, whereas the sand-richer state produced the strongest local sand-transport peaks near the mouth. A fixed-bathymetry sensitivity pair further showed that, under a common starting morphology, integrated mud-transport intensity in the 30% case was 3.52 times that in the 10% case, while sand-transport intensity was approximately 9% lower. Model SSC enhancement was 54.2-fold at the seaward station, 18.4-fold at the mouth and 7.3-fold landward, providing an order-of-magnitude and spatial-gradient comparison with reported Changjiang storm responses. These results show that storm response is conditioned by the coupled antecedent composition-morphology state and identify exposed delta-front sectors as priority locations for event-scale monitoring and sediment management.

Estuaries are dynamic interfaces where river discharge, tides, waves and sediment supply jointly shape channels, mouth bars and delta fronts. Their morphology reflects nonlinear feedbacks among hydrodynamic forcing, sediment transport and evolving basin geometry (de Swart and Zimmerman, 2009; Friedrichs and Aubrey, 1994; Coco et al., 2013; Geyer and MacCready, 2014; Uncles, 2002). Under fair-weather conditions, recurrent tidal and fluvial transport drives much of this adjustment. Storms superimpose energetic waves, wind-driven circulation and surge, increasing bed shear stress and sediment resuspension over event time scales (Green et al., 1997; Green and Coco, 2014; Amoudry and Souza, 2011). Quantifying the resulting reorganization is necessary to assess the vulnerability of channels, delta fronts and intertidal landforms to compound coastal hazards.

The response to storm forcing depends not only on hydrodynamic intensity, but also on bed composition and on the morphology that develops under those sediment conditions. Sand and cohesive mud differ in their entrainment, settling and consolidation behaviour, while their coexistence can alter bed erodibility and hydraulic roughness (Winterwerp and van Kesteren, 2004; Van Ledden et al., 2004; Manning et al., 2010). Over longer time scales, these contrasts can generate different channel-flat and delta configurations under the same external forcing (Jiang C.H. et al., 2024). Consequently, the pathways and fate of suspended material vary with sediment composition as well as with tidal, fluvial and topographic controls (Burchard et al., 2018; Sherwood et al., 2018; Son and Hsu, 2011; Tarpley et al., 2019). Recent morphodynamic modelling also shows that sand-mud interaction can strongly affect intertidal morphology and bed composition, so treating the two fractions as independent may obscure important feedbacks (Zhou et al., 2015; Wang et al., 2019; Alonso et al., 2023).

Observations from the Changjiang Estuary demonstrate that storm waves can generate large but spatially heterogeneous sediment responses. At an outer-estuary station, tropical-cyclone swell increased SSC from 0.056 to 0.473 kg m -3 (approximately 8.4-fold; Huang et al., 2022). During Typhoon Hinnamnor, in-situ SSC measurements reached 2.8 kg m -3 and increased 5.6-fold, while the associated modelling indicated contrasting erosion and deposition across delta shoals, fronts and channels (Yao et al., 2024). Daily surveys following a major storm also documented erosion of unvegetated flats and local channel deposition in the outer river mouth (Yang et al., 2003). Such records are essential benchmarks, but their limited spatial coverage and event-specific forcing cannot separately resolve the effects of storm magnitude, tidal timing, bed composition and antecedent morphology (Chu et al., 2015). Process-based models are therefore needed to test how those controls act together across the estuarine system.

Storm impacts on coastal morphodynamics have been investigated in sandy and muddy environments, and numerical studies have begun to resolve the sedimentary signatures of storm events (Hoyal and Sheets, 2009; Green and Coco, 2007; Friedrichs, 2011; Leonardi et al., 2016; Masselink et al., 2016). Storm waves can remobilize relatively coarse material and alter the horizontal and vertical sorting patterns of tidal flats (Zhou et al., 2022). Nevertheless, the event-scale response of mixed-sediment estuaries that first developed under contrasting sediment compositions remains insufficiently constrained. In particular, it is unclear how composition-conditioned equilibrium states influence (i) the partitioning of sand and mud transport, (ii) erosion and deposition around the mouth and delta front, and (iii) the spatial sorting that persists after a storm. The modulation of these responses by prescribed compound storm forcing and tidal phase also remains poorly resolved at the scale of the whole estuarine system.

Long-period wave groups and resonant basin motions provide an additional pathway for redistributing and amplifying storm-wave energy in semi-enclosed coastal systems. Focused and irregular wave groups can excite harbour or bay oscillations, while Bragg reflection over periodic topography can alter the transmission and retention of resonant energy (Gao et al., 2020, 2021, 2024; Hou and Gao, 2026). These processes may modify the timing and location of sediment resuspension near estuary mouths and delta fronts, but they are not resolved by the phase-averaged short-wave forcing used in the present model.

This study therefore aims to investigate (1) how storm intensity and tidal timing control the short-term hydrodynamic and morphodynamic response of a mixed-sediment estuary; (2) how equilibrium states generated from contrasting initial mud fractions differ in storm-driven sand and mud transport and in their erosion-deposition patterns; and (3) how these processes reorganize mud and sand across seaward, mouth and landward sectors. We address these aims with an idealized, process-based model and evaluate its response magnitude and spatial ordering against published Changjiang storm observations.

Based on the open-source morphodynamic model framework (Delft3D), we construct a depth-averaged 2D model, referencing the Yangtze Estuary for the key geometric parameters of our idealized configuration. The model setup and parametrizations are detailed below.

Antecedent model configuration

The storm simulations were initialized from dynamically equilibrated morphologies generated with the two-dimensional, depth-averaged model of Jiang C.H. et al. (2024). The model represents an idealized, river-dominated estuary referenced to the Yangtze Estuary. Its 800-km-long, exponentially convergent channel is connected to a broad offshore basin that permits ebb-delta development. During the antecedent simulations, a 50-m-thick mobile bed was prescribed throughout the domain. Bathymetry deepened from 10 m at the river boundary to 50 m at the seaward boundary. A constant river discharge of 15,000 m³ s -1 was imposed at the upstream boundary. Astronomical forcing at the active seaward boundary combined M2 and S2 constituents, with amplitudes of 1.20 and 0.80 m, respectively. This constituent pair generated the modelled spring-neap modulation. Cross-shore ocean boundaries used zero alongshore water-level gradients. The curvilinear grid was refined inside the estuary (approximately 50–200 m) and coarsened offshore (up to approximately 2 km).

The initial-condition model represents cohesive mud and non-cohesive sand. Following Jiang C.H. et al. (2024), mud transport used the Partheniades-Krone formulation and sand transport used the Engelund and Hansen (1967) transport formulation during the long-term spin-up. Equilibrium sand-concentration boundaries were applied at the fluvial and tidal boundaries to maintain stable boundary-bed conditions; the long river length limits the influence of this boundary supply on mouth morphology. Common physical parameters were a 2-min hydrodynamic time step, a Chezy coefficient of 65 m½ s -1, horizontal eddy viscosity and diffusivity of 1 and 10 m² s -1, respectively, a mud settling velocity of 0.25 mm s -1, a critical mud erosion stress of 0.5 Pa, a mud erosion parameter of 1 × 10 -4 kg m -2 s -1, a mud bulk density of 500 kg m -3, and a sand median diameter of 0.2 mm. Density-driven circulation, flocculation, consolidation, armouring, vegetation and Coriolis effects were not represented. Morphological acceleration was used during the long-term stage, as described by Roelvink (2006). In this approach, the sediment erosion and deposition fluxes are scaled up by a factor (MF) at each hydrodynamic time step. This approach has been extensively used in previous studies, including schematic simulations (Roelvink, 2006; Van der Wegen and Roelvink, 2008) and real-world situations (Dastgheib et al., 2008; Van der Wegen and Roelvink, 2012). It has been suggested that the value of MF should be determined by sensitivity experiments (Zhou et al., 2014). After sensitivity tests, an MF of 200 is considered here, which is similar to the values used in several previous studies (Ni et al., 2022; Gugliotta and Saito, 2019).

Three otherwise identical spin-up simulations were conducted with initially uniform bed mud fractions of 10%, 20% and 30% by volume; the remaining bed material was sand. All hydrodynamic forcing, boundary conditions, grid, sediment parameters and numerical settings described above were held constant across these runs. Each was run for 1,000 morphodynamic years, after which its equilibrated bed elevation and bed composition were transferred to the event-scale model (Figure 1). We refer to these coupled, co-evolved bathymetry-composition configurations as composition-conditioned equilibrium states. They retain long-term composition-morphology feedback and do not independently identify a direct mud-fraction effect at fixed initial bathymetry. In the 20%-mud reference panel, red markers identify the seaward (m_out), mouth (m), and landward (m_in) stations used for time-series diagnostics.

Storm-forcing and wave-current coupling setup

Storm response was simulated with two-way FLOW-SWAN coupling, which couples water levels and depth-averaged currents to wave characteristics and wave-induced forcing. The event-scale model retained the 2-min FLOW time step and updated bathymetry without morphological acceleration (MorFac = 1). Morphological updating began after a 720-min hydrodynamic spin-up (MorStt = 720 min), and four further fair-weather days separated the start of morphological updating from storm onset. Sand transport used the Soulsby-Van Rijn (Soulsby, 1997; Van Rijn et al., 2007) combined tide-wave formulation (IFORM = 11; Acal = 1.0, D90/D50 = 2.0 and z0 = 0.0007 m). The formulation differs from the Engelund-Hansen predictor used during the 1,000-year spin-up because the event stage must represent combined wave-current forcing; Engelund-Hansen was retained in the antecedent stage for continuity with the long-term equilibrium model of Jiang C.H. et al. (2024). Event responses were referenced to the matched pre-storm state or no-storm control so that any residual adjustment after the formulation switch was separated from the storm signal. Soulsby-Van Rijn output was analysed as depth-integrated total sand transport and was not partitioned into bed load and suspended load. Cohesive mud was represented only as suspended load through the retained Partheniades-Krone erosion and deposition formulation. The coupled configuration follows established wave-current and morphodynamic applications in tidal inlets and estuaries (Wang et al., 1995; Van Rijn et al., 2007; Olabarrieta et al., 2011; Elias et al., 2012; Williams and Esteves, 2017; Yin et al., 2016; Chauchat et al., 2017). The wave model used a JONSWAP spectrum imposed at the western sea boundary, a 270-degree wave direction, depth-induced breaking, wave refraction, whitecapping and JONSWAP bed friction (0.067 m² s -3). The SWAN configuration resolved 36 directional bins and 24 frequency bins.

The M2-S2 constituent pair generated the spring-neap cycle used for the tidal-timing comparison. Under fair-weather conditions, offshore significant wave height was 0.7 m, peak period was 5.7 s, wind speed was 3 m s -1, and storm surge was zero. Here, compound storm denotes a single prescribed event in which waves, surge and enhanced wind forcing occur simultaneously and reach a common peak; it does not denote a statistical joint-probability analysis. Each prescribed event lasted six days. Significant wave height and wind speed increased linearly over the first three days to their prescribed maxima and then declined linearly to fair-weather values over the following three days. A time-varying storm-surge water-level boundary increased from 0 to 0.5 m at the storm peak and returned to 0 m over the same cycle. Wave peak period increased from 5.7 s under fair weather to 20 s at the storm maximum, then decreased as the storm subsided; wave direction remained 270 degrees. The 3-, 4- and 5-m wave cases were paired with peak wind speeds of 30, 40 and 50 m s -1, respectively. This ordered forcing series is an idealized sensitivity envelope rather than a reconstruction of a specific typhoon, and the fixed 0.5-m surge is not intended as a wind-specific surge estimate. No upstream sediment inflow was prescribed during the storm simulations; at the sea boundary, mud and sand concentrations were fixed at 0.05 kg m -3. The storm maximum was aligned with high water during either the spring or neap part of the M2-S2 cycle (Figure 2), consistent with the recognized importance of compound wave, water-level and tidal forcing for coastal sediment response (Green and Coco, 2014; Masselink et al., 2016; Yang et al., 2012).

Scenario design and diagnostic variables

The numerical experiments compare storm intensity, tidal timing and composition-conditioned equilibrium states (Table 1). A no-storm simulation provided a fair-weather control. The spring-tide suite combined three equilibrium states with three peak wave heights, while the 4-m storm provided the standard condition for the between-state comparison. A further 20%-mud, 4-m experiment was aligned with neap rather than spring tide to test one matched tidal-timing contrast. This is a comparative design rather than a fully factorial separation of composition and geometry, because equilibrium bathymetry and bed composition co-vary across the three antecedent states. Differences among those states are therefore interpreted jointly as outcomes of antecedent sediment composition and morphology, which are also coupled in natural estuaries (Fagherazzi et al., 2015; Guo et al., 2014; Nienhuis et al., 2016).

To isolate the event-scale sensitivity to initial bed composition from starting morphology, we performed a fixed-bathymetry sensitivity experiment under the 4-m storm. The 10% and 30% mud cases were both initialized from the 20%-mud equilibrium bathymetry; all hydrodynamic and wave forcing, boundary conditions, sediment parameters and numerical settings were otherwise identical. Morphodynamic updating began at storm onset, so the two cases had exactly the same bed elevation at that time. Bed-level change and area-weighted transport intensity were evaluated from storm onset to two days after storm termination over the estuary-mouth analysis region.

Model output comprised water level, depth-averaged velocity, significant wave height, suspended-sediment concentration (SSC), fraction-specific sediment transport, bed level and bed mud fraction. Event-scale bed-level change was calculated from post-storm minus pre-storm bathymetry, and volume change was obtained by summing cell-wise bed-level change multiplied by cell area. Figure 3 reports hourly volume-change rates and their cumulative integral over the 16-day response window for the seaward delta-front and mid-channel-bar masks. Figure 4 uses a common initial elevation mask (bed level > -15 m) to compare cumulative net volume change among forcing cases over the same 16-day window. To remove initial numerical adjustment from the sorting analysis, changes in bed mud fraction were referenced to the state one day before storm onset. Spatial diagnostics were evaluated separately for seaward, mouth and landward sectors and at the three monitoring stations shown in Figure 1. The simulations continued beyond the six-day imposed forcing with MorFac = 1 to resolve delayed morphodynamic adjustment, and the resulting time series therefore extend beyond the forcing interval. For the regional observation comparison, model event means and maxima were calculated over the six-day forcing interval. Model SSC enhancement was the event-mean storm SSC divided by the mean SSC over the matched interval in the no-storm control; field enhancement factors were used as reported. No model parameter was tuned to the compared observations. The comparison provides regional, observation-informed context rather than site- and event-specific validation.

Hydrodynamic and profile change

The reference 4-m storm strongly modified the local hydrodynamic and sedimentary signals while the tidal water-level pattern remained coherent across the three monitoring stations (Figure 5a). Current velocity responded most strongly at the seaward and mouth stations during the storm window (Figure 5b). Significant wave height was greatest at the seaward station and was markedly reduced at the mouth and landward stations, showing pronounced attenuation across the delta-front and mouth region (Figure 5c).

Suspended-sediment concentration increased sharply during the storm, with a sustained seaward peak of about 1 kg m -3 and episodic peaks of similar magnitude at the mouth (Figure 5d). The landward signal was markedly weaker. At the mouth, both mud and sand concentrations increased during the high-energy phase; mud concentration generally remained higher than sand concentration, although individual sand peaks became pronounced near the storm maximum (Figure 5e).

Under fair weather, along-estuary elevation change remained small and was generally confined within approximately ±0.04 m (Figure 6a). Prescribed compound-storm forcing increased both the magnitude and spatial variability of bed-level change, including local erosion approaching 0.14 m in the mouth-adjacent reach (Figure 6b). Station records show rapid erosion at the seaward station and net deposition at the mouth and landward stations (Figure 6c).

The area-based diagnostics localized this response at the exposed delta front. In the 20%-mud reference case, seaward delta area decreased by roughly 100 km² during the storm and remained below its pre-storm value at the end of the simulated response window (Figure 7c). In contrast, landward delta area changed by only a few square kilometres and ended slightly above its pre-storm value (Figure 7d).

Volume-change rates confirmed that the delta front underwent the dominant storm-driven adjustment. During normal weather, erosion and deposition oscillated at rates of only tens of cubic metres per second. In the 4-m reference case, the seaward delta-front mask reached an hourly erosional peak magnitude of 2.77 × 10³ m³ s -1 and a cumulative loss of 0.29 × 10 9 m³ over the 16-day response window (Figures 3a, b). Mid-channel bars behaved differently: they accumulated sediment during the storm and gained 0.10 × 10 9 m³ over the same window (Figures 3c, d).

Sediment transport partitioning and compositional adjustment

Storm-integrated transport at the delta front was strongly partitioned by both sediment fraction and elevation zone (Figure 8). In the intertidal zone, the sand transport magnitude (1.8 × 10 8 m³) exceeded the mud magnitude (2.9 × 10 7 m³). In the subtidal zone, both fractions were much more mobile, with mud (2.5 × 10 9 m³) exceeding sand (1.7 × 10 9 m³). The hourly time series show substantially larger subtidal transport throughout the storm-response window, while the cumulative curves indicate that sand transport advanced earlier than subtidal mud transport.

The response of bed composition depended on the initial mud fraction and on elevation zone (Figure 9). During the storm, subtidal bed mud fraction exhibited a coherent rise followed by a decline toward the pre-storm level, whereas the intertidal records remained more variable. Event-scale changes were positive in both zones for the 10% and 30% initial-mud cases. The 20% case was distinctive: its intertidal mean change was weakly negative and small relative to the temporal variability across the active cells, while its subtidal mean change was positive. The opposite signs indicate local intertidal sand enrichment alongside broader subtidal mud enrichment; the present diagnostic does not identify a unique source-to-sink pathway. The 30% case showed the largest positive change in both zones.

The fixed-bathymetry test confirmed an event-scale sensitivity to initial composition (Supplementary Figure S1). Starting bathymetry was identical (maximum difference, 0 m), and the spatial bed-change patterns remained similar (r = 0.954; RMSE = 0.00143 m); nevertheless, the 30% mud case produced integrated mud-transport intensity 3.52-fold that of the 10% case during both the storm and recovery intervals, while its sand-transport intensity was approximately 9% lower. Thus, the original three-state comparison is a combined composition-morphology signal, but part of the fraction-specific transport contrast persists when initial bathymetry is held fixed.

Dependence on prescribed compound-storm forcing and wave-energy attenuation

The prescribed compound-storm forcing suite controlled the cumulative sediment-volume response, but the delta front and channel followed contrasting trajectories (Figure 4). In the 20%-mud reference state, final delta-front volume loss in the common initial delta mask (bed level > -15 m) increased from approximately 0.30 × 10 9 m³ for the 3-m event to 0.52 × 10 9 m³ for the 5-m event. The channel first eroded during forcing and then became depositional; final deposited volume increased across the forcing suite. The no-storm control remained close to zero in both regions.

The wave-energy fields identify the delta front as the main zone of wave attenuation (Figure 10). Under normal weather, cross-section mean wave-energy flux decreased from about 1.6 × 10³ W m -1 offshore to less than 0.5 × 10³ W m -1 near the mouth. Storm forcing increased offshore flux by more than two orders of magnitude, but flux decreased steeply across the delta front in every storm case. Consequently, the large offshore differences among the 3-, 4- and 5-m wave conditions were substantially reduced toward the mouth.

Storm response across composition-conditioned equilibrium states

The three composition-morphology equilibrium states differed in delta geometry and storm response (Figures 11a–c). Before storm onset, seaward delta area was 2083.3, 2333.2 and 2365.4 km² in the 10%, 20% and 30% states, respectively. By the end of the six-day forcing, the corresponding absolute losses were 127.4, 98.1 and 110.9 km², equal to 6.11%, 4.20% and 4.69% of each state’s own pre-storm seaward area (Figure 11d). Landward-area adjustments were small by comparison (Figure 11e). The 20% state therefore showed the smallest proportional loss, demonstrating that the response is not monotonic with antecedent mud fraction across the co-evolved states.

Delta-front exposure localizes storm-driven morphological change

The simulations show that storm-driven adjustment is not distributed uniformly across the estuary. The exposed delta front is the principal locus of rapid bed erosion and planform contraction, whereas channel and landward sectors undergo smaller net change or local deposition. This contrast is consistent with the sharp decline in wave-energy flux across the delta front (Figure 10): the morphology acts as an energy-dissipating interface, but it is also the first and most strongly forced part of the estuary. Comparable links between wave exposure, morphological setting and erosion have been identified in tidal flats, river mouths and vegetated estuarine margins (Green et al., 1997; Friedrichs, 2011; Fagherazzi et al., 2015; Leonardi et al., 2016). The co-occurrence of delta-front erosion and mid-channel-bar accretion is consistent with spatially differentiated adjustment in the model (Figures 3, 7).

Storm intensity amplifies this spatial contrast. Stronger storms generate larger offshore wave-energy fluxes and larger cumulative delta-front losses, while the shoreward reduction of wave energy limits the direct propagation of the disturbance. The single matched spring-neap comparison provides one case-specific contrast: in this configuration, aligning the storm peak with spring-tide high water produced a more flood-dominant mouth-current pattern than aligning it with neap-tide high water. This is consistent with estuarine-circulation and sediment-trapping theory, in which tidal asymmetry, settling lag and the residual circulation control the net transport of fines (Geyer and MacCready, 2014; Uncles et al., 1985; Chernetsky et al., 2010; Scully and Friedrichs, 2007), but it does not on its own establish a general tidal-phase effect. A systematic timing experiment spanning high water, low water and mid-tide across storm intensities and antecedent states would be required. Together, the results show that storm magnitude alone is not a sufficient predictor of morphological impact; wave forcing and antecedent delta-front morphology determine where event-scale change is concentrated.

Long-period motions could modify this exposure pattern. Focused or irregular wave groups can excite resonant oscillations in semi-enclosed basins, and Bragg reflection by periodic bedforms can either restrict incident energy or modify its retention near an entrance (Gao et al., 2020, 2021, 2024; Hou and Gao, 2026). Such motions may alter resuspension and exchange at the delta front and mouth. Because the present SWAN forcing is phase averaged and does not resolve infragravity-wave or basin-resonance dynamics, the identified attenuation pattern applies to the represented short-wave energy; resolving long-period motions is a priority for event-specific applications.

How sediment composition regulates along-channel transport partitioning

Along-channel transport rates show that the coupled antecedent composition-morphology state controls both the magnitude and the spatial partitioning of storm-driven transport (Figure 12). Under the 4-m storm, mud transport increases from the 10% to 30% states, whereas the largest sand-transport maximum occurs in the 10%-mud state and remains confined to the mouth-proximal reach. The response is therefore not a proportional increase in both fractions with antecedent mud content. Instead, the sand-rich state preserves a highly mobile sand-transport corridor near the mouth, while the mud-rich states favour a stronger mud-transport response. This contrast is consistent with sand-mud mixture theory, in which bed composition modifies cohesion, grain contact and erodibility (Winterwerp and van Kesteren, 2004; Van Ledden et al., 2004; Manning et al., 2010), and with studies showing that sorting, hydrodynamics and morphology co-evolve (Zhou et al., 2015; Wang et al., 2019; Alonso et al., 2023). The corresponding cross-section-averaged mud and sand SSC distributions along the channel are provided in Supplementary Figure S2 and show the same controls of antecedent composition and storm intensity on the concentration field.

At a fixed initial mud content of 20%, stronger storm forcing increases transport of both fractions but sharpens their contrast. From the 4-m to the 5-m storm, the peak mud transport rate increases from 1.58 × 10 -3 to 2.68 × 10 -3 m³ s -1 m -1, whereas the peak sand transport rate increases from 1.83 × 10 -3 to 5.67 × 10 -3 m³ s -1 m -1. The pronounced sand maximum remains concentrated near the mouth, indicating that stronger wave-current forcing preferentially intensifies coarse-sediment mobilization within this dynamically active reach. This behaviour is consistent with wave-current transport formulations and storm-morphodynamic studies showing that stronger storms can mobilize coarser sediment and amplify local sediment sorting (Van Rijn et al., 2007; Zhou et al., 2022). The simulations therefore identify the mouth-proximal zone as the principal location where storm energy is converted into a strong, sand-dominated transport response.

Taken together, these results show that storm-driven transport partitioning is controlled by a coupled antecedent state rather than by storm intensity alone. The fixed-bathymetry pair confirms that a fraction-specific contrast also persists under a common starting morphology: the 30% case transports 3.52 times more mud than the 10% case during both forcing and recovery, while transporting approximately 9% less sand (Supplementary Figure S1). The original three-state comparison remains a joint composition-morphology response, whereas the fixed-bathymetry result demonstrates an event-scale sensitivity to prescribed bed composition within the tested pair.

Regional comparison with storm observations

The 20%-mud, 4-m reference simulation was compared with published Changjiang storm observations not to test site-specific predictive skill, but to assess whether the model’s response magnitude and spatial gradient are regionally consistent with known natural behavior (Figure 13). Despite inevitable mismatches in site geometry, storm tracks, and water depths, the comparison provides useful context for interpreting the simulated exposure gradient.

In terms of suspended-sediment concentration (SSC), the model captures the observed seaward-to-landward ordering of storm enhancement. Model enhancement was 54.2-fold at the seaward station, 18.4-fold at the mouth and 7.3-fold at the landward station (Figure 13b). The landward value lies within the available field range of 5.0-7.5-fold, whereas the seaward and mouth values are larger than the reported field factors. Model event-mean SSC values of 0.10-0.56 kg m -3 are of the same order as reported storm concentrations of approximately 0.47-2.8 kg m -3 (Huang et al., 2022; Yao et al., 2024; Jiang J. et al., 2024), although the field records represent different storms, depths and sampling elevations. The comparison therefore supports the simulated spatial gradient and order of magnitude, not site-specific predictive skill.

For bed-level changes, the model again reproduces the pronounced contrast between major channel deposition and muted adjustment in sheltered areas. Yang et al. (2003) documented ~0.5 m of channel-axis deposition and ~0.04 m of marsh accretion. The model yields similar order-of-magnitude values (~0.59 m at the mouth and ~0.06 m landward). The slightly larger modeled channel response may reflect the absence of bank vegetation and three-dimensional flow structures in our idealized 2D setup, whereas the landward comparison, although not a direct geomorphic analogue (given the vegetated marsh setting), nonetheless confirms that protected sectors experience an order-of-magnitude smaller change than the exposed delta front.

Collectively, these consistency checks lend observational support to our primary finding that antecedent composition-morphology states and storm timing jointly determine the location and intensity of event-scale impacts, rather than validating the model’s absolute predictive accuracy.

A coupled mechanism for storm-induced sediment sorting and morphological adjustment

Figure 14 summarizes the simulated response as a common seaward-to-landward sequence that is expressed differently by the three composition-conditioned equilibrium states. In all cases, rising waves, wind and surge increase bed stress and resuspend sediment at the exposed delta front. The delta-front-mouth configuration then acts as an energy-dissipating interface. Wave forcing is strongest offshore and decreases rapidly through the front and funnel-shaped mouth, limiting its direct landward propagation. This exposure gradient localizes the principal morphological adjustment at the seaward delta front, where erosion and planform contraction occur. The mouth and channel region undergo local deposition, whereas the landward reach adjusts more weakly.

The weakly negative intertidal and positive subtidal mud-fraction changes in the 20% state are consistent with selective removal of fines from wave-exposed active cells and redistribution toward the broader subtidal domain. Vertical sorting and surface coarsening could reinforce this pattern by reducing subsequent fine-sediment availability, while cohesive sand-mud interaction could suppress erosion once a connected mud matrix develops. These are candidate mechanisms rather than a resolved source-to-sink budget: Figure 9 diagnoses active-zone composition, but the model output used here does not trace sediment exchanged between individual zones. Within this shared spatial framework, the fraction-specific response and the magnitude of front adjustment depend on the inherited composition-morphology state. The sand-richer 10%-mud state shows the strongest localized near-bed sand reworking around the delta front and mouth, together with the largest proportional seaward-area loss. The 20%-mud state provides the reference expression of the coupled response. Mud is broadly suspended and redistributed over the subtidal domain, whereas sand is concentrated in local near-bed wave-current reworking. In the mud-richer 30%-mud state, suspended mud is more extensive and reaches farther landward, while the local sand signal is less dominant. These contrasts should be interpreted as state-dependent responses of co-evolved composition and morphology, rather than as an isolated direct effect of mud fraction. The interpretation accords with conceptual and numerical work on wave-driven estuarine resuspension, channel-scale sediment trapping and sand-mud sorting (Green and Coco, 2014; Traykovski et al., 2004; Burchard et al., 2018; Ralston et al., 2012; Zhou et al., 2015). Figure 14 is a conceptual synthesis of modelled spatial patterns, not an independently resolved sediment trajectory or a general prediction for every tidal setting.

Implications and limitations

The results show that estuarine storm vulnerability cannot be inferred from wave height alone. Differences among composition-morphology states arise from sediment availability and delta-front geometry, which jointly control exposure to and attenuation of storm energy. The regional comparison with Changjiang observations provides an order-of-magnitude and spatial-gradient check rather than direct calibration. The symmetric six-day ramp, fixed wave direction, spatially uniform wind and fixed 0.5-m surge make the forcing an idealized sensitivity envelope. The timing of post-peak deposition is particularly sensitive to the decay of this forcing, and a larger surge could change depth-limited breaking and shift the location of maximum bed stress. Constant river discharge and zero upstream sediment supply isolate the marine storm response; flood-driven discharge and sediment delivery could alter current asymmetry, landward SSC and channel deposition. The depth-averaged model also omits density-driven circulation, flocculation, consolidation, armouring, vegetation and Coriolis effects, and each of these processes has been shown to alter estuarine sediment transport. Density stratification and Coriolis forcing can reorganize residual and lateral transport (Geyer and MacCready, 2014; Ralston et al., 2012; Burchard et al., 2018; Kasai et al., 2000); the depth-averaged configuration used here resolves neither baroclinic nor rotational dynamics (Lesser et al., 2004), so neither the estuarine residual circulation nor the rotational deflection of the residual flow is represented and the associated lateral redistribution is absent by construction. Flocculation and consolidation change mud settling and erodibility (Manning et al., 2010; Son and Hsu, 2011; Le Hir et al., 2007; Tarpley et al., 2019); armouring and vertical sorting change fraction availability (Van Ledden et al., 2004; Traykovski et al., 2004; Zhou et al., 2015, 2022; Alonso et al., 2023); and vegetation alters stress and trapping (Mariotti and Fagherazzi, 2010; Yang et al., 2012; Malarkey et al., 2015; Leonardi et al., 2016). These omissions may change response magnitude and local partitioning, so the results should be interpreted as process-based sensitivities rather than event forecasts. The common localization of adjustment at the exposed front and the persistence of fraction-specific transport contrasts in the fixed-bathymetry test remain the directly supported outcomes. Future tests should vary storm asymmetry, surge, river forcing, long-period motions and the omitted sediment processes, and should use full fraction-specific mass budgets and colocated event observations.

Storm-driven adjustment in this idealized mixed-sediment estuary is organized by the coupled antecedent composition-morphology state. Three findings answer the study aims. First, increasing storm intensity amplified offshore wave-energy flux and cumulative delta-front loss, while one matched spring-neap pair showed that tidal alignment can modify the mouth-current response but does not establish a general tidal-phase effect. Second, the three co-evolved states produced distinct erosion and transport patterns: the exposed delta front was the principal locus of loss, the sand-richer state generated the strongest localized sand-transport peak near the mouth, and the mud-richer states sustained broader mud transport. Third, storm reorganization was spatially partitioned, with strong subtidal mud transport, localized mouth-proximal sand transport and much smaller landward morphological change. The fixed-bathymetry pair confirmed that part of the fraction-specific contrast persists under a common starting morphology. Model SSC enhancement and bed-level change reproduced the observed spatial ordering and order of magnitude in the Changjiang Estuary, providing regional context rather than site-specific validation. These results identify exposed delta-front sectors as priority targets for post-storm monitoring and sediment management.