1 Introduction

Intertidal mudflats are an essential part of the global coastal system (Amos 1995; Gao 2019; Wang 1983) and a natural transition between ocean and land (Xu and Liu 2021) where physical, chemical, and biological processes are under constant change (Grabowski et al. 2011; Khojasteh et al. 2021). These mudflats provide environmental services such as carbon absorption and burial (Lin et al. 2020) and pollutant filtration (Goodwin et al. 2001) and ecological services such as benthic habitats (Barbier 2013), spawning grounds for offshore zooplankton and nursery grounds for young organisms (Andersen et al. 2006), and wave attenuation and coastal protection (Bale et al. 2006). Therefore, intertidal mudflats occupy a prominent position in marine resources and the natural environment (Kirwan and Megonigal 2013), and their development and protection are of high importance to the sustainable development of society (Costanza et al. 1997; Temmerman et al. 2013). However, the trend of increasing storms as a component of global climate change in many regions of the world has become an indisputable fact (Knutson et al. 2010; Sobel et al. 2016), and mudflats are being increasingly exposed to storms, typhoons, hurricanes, and other marine hazards (Emanuel 2005; Leonardi et al. 2016), in turn leading to greater amounts and rates of mudflat erosion (Lin and Emanuel 2016). A deeper understanding of the processes and mechanisms operating in these areas during storms is required, to reasonably protect, utilize, and manage intertidal mudflats.

Storms are one of the most destructive short-term hydrometeorological extreme events along coastal areas and generate large waves, storm surges, and strong winds. Storm-related processes can lead to a sudden increase in hydrodynamic intensity, rapid changes in topography and bathymetry, and a sharp change in erosion and accretion patterns on intertidal mudflats (Dai et al. 2018; Liu et al. 2015; Wang et al. 2014). Third, the shallow-water performance of waves can cause the waves to produce inclined and asymmetric orbital velocities. This process drives sediment transport and generates a variety of flows (such as wave setup, longshore current, or rip current) by nonlinear interactions with currents (Song et al. 2020), causing more complex sediment transport (Sherwood et al. 2022). Fourth, intertidal mudflats in bays and estuaries are generally turbid and commonly have high near-bed SSCs or even fluid mud layers. Shallow-water bodies carry large amounts of sediment coupled with the scour and settling lags of the sediment itself, which involves complex three-dimensional dynamics (Ge et al. 2020; Zhu et al. 2014). These four reasons can lead to simulation results that are less accurate than those for deep-water environments. In addition, shallow-water environments have been a focus of human activity, and activities such as dam construction and reclamation have added perturbations to numerical modeling, increasing the uncertainty involved in simulations (**ng et al. 2012). Therefore, numerical simulation of shallow-water environments requires many considerations, and it is not easy to balance the accuracy of the simulation with computational efficiency. Using numerical simulation alone to study morphodynamic processes over intertidal mudflats is prone to model divergence and deviation from reality. Combining the in situ observations with numerical simulation can take advantage of both, which can achieve a fine and quantitative study and ensure the accuracy of the model. However, only some studies have combined measured data with numerical simulations.

In this study, a process-based modeling approach using the Delft3D model is applied to simulate the hydrodynamic response processes and erosion–accretion patterns of the Nanhui intertidal mudflats in the Yangtze estuary during the 2014 storm Fung-wong. The numerical model is validated quantitatively by in situ observations of hydrodynamics and SSCs. On this basis, we also investigate changes in the hydrodynamic and geomorphic response of the Nanhui intertidal mudflats under different wind speeds and discuss the influencing factors and mechanisms of intertidal mudflat erosion and accretion patterns during storms.

2 Study area and storm overview

The Nanhui intertidal mudflat, which is located on the front of the southern Yangtze Delta and the northern coast of Hangzhou Bay, is an open meso–macrotidal mudflat facing the East China Sea. The Yangtze and Qiantang rivers converge here, meaning that the mudflat is part of a region with complex hydrosedimentary dynamics and frequent exchange of water and sediment (Chen 2004). The study area is located near Luchaogang on the Nanhui intertidal mudflat (Fig. 1). The tidal type of the mudflat is irregular semidiurnal dominated by M2, with a multi-year mean tidal range at the Luchaogang tide gauge station of 3.2 m and a maximum tidal range of up to 4 m (Yang et al. 2019). The tide outside the mouth of Hangzhou Bay is dominated by rotational flow, and the coastal area is dominated by the reciprocal flow. The bed sediment of the mudflat is generally silty and muddy. Surface SSCs have been reported as ranging from < 0.1 to > 2 kg/m3 (Zhu et al. 2014).

Fig. 1
figure 1

Overview of the study area. a Storm Fung-wong track and storm level (2014); b Nanhui intertidal mudflats, model calculation area and grid, tide gauge stations (S1 and S2), and in situ observation station (S3); c locations of Luchaogang, the analysis points N1, N2, and N3 and the analysis cross-section; d bed level profile of cross-section

Full size image

The study area is located in the subtropical monsoon climate zone. According to meteorological statistics, an average of 3.5 typhoons per year affected the area during 1949–2005 (Meng 2008). The mean wind speed near the study area during 2000–2021 was 5.54 m/s, and the maximum wind speed was 22.1 m/s, of which speeds of < 10 m/s accounted for 96.289% of wind time, and 10–15, 15–20, and ≥ 20 m/s accounted for 3.565%, 0.141%, and 0.005%, respectively. The two most frequent wind directions were SSE and NNE, with frequencies of 12.0% and 9.5%, respectively (European Centre for Medium-Range Weather Forecasts, ECMWF). The waves affecting the area are wind waves and swell controlled by the monsoon, with dominant wind wave action, and the most frequent wave directions reflect those of wind (Yang et al. 2008).

Storm Fung-wong, the 16th tropical storm in 2014, was formed on the ocean surface near the eastern Philippines around 14:00 17 September, and it was upgraded to a strong tropical storm on 21 September in the sea southwest of Taiwan. The storm made landfall in Shanghai at 10:00 on 23 September, with a maximum wind speed of 28 m/s at the center of the storm. The path of Fung-wong is shown in Fig. 1a. According to the Chinese National Standard for wind scale (GB/T 28,591–2012), a “strong breeze” (10.8–13.8 m/s) can be regarded as strong wind. Accordingly, we define the beginning and end of wind speeds greater than 10 m/s in the study area as the beginning and end of this storm event, respectively, defining the storm period in the study area as extending from 13:00 on 21 September to 8:00 on 24 September. The wind direction during this period changed gradually clockwise from NE to NW and blew mainly from the ENE, accounting for 40% of the whole period, with an approximate maximum wind speed of 15 m/s in the study area (Fig. 2b).

Fig. 2
figure 2

Wind roses (a) during 2000–2021 and (b) during 2014 storm Fung-wong near Luchaogang (from ECMWF)

Full size image

3 Field campaign

In situ observations were conducted from 22 to 24 September 2014 at station S3 on the Nanhui intertidal mudflat (Fig. 3), with measurements of water depths, wave heights, and SSCs, whereas stations S1 and S2 were used to obtain water levels from their tide gauge stations (Fig. 1b). An HR-Profiler (Nortek, USA) was used to obtain current velocity profiles on the mudflat, with a downward-facing probe height of 0.60 m above the sediment surface and a burst interval of 5 min. Each velocity profile was collected at a frequency of 1 Hz over a duration of 18 s. The cell size was set to 0.03 m, with a blanking distance of 0.10 m. The measurement distance was 0–0.50 m from the sediment surface. Wave observations were performed using an SBE-26plus Seagauge (Sea-Bird Electronics, USA), mainly to obtain water depths and wave heights, with the instrument being placed horizontally on the sediment surface. The sensor was positioned 0.08 m above the sediment surface, and the measuring burst interval was 10 min. Data were collected at a frequency of 4 Hz over a duration of 128 s, and water depths were corrected to mean sea level (MSL) to obtain water levels. Turbidity, which was measured through profiles within 0.40–0.90 m above the sediment surface, was obtained using an ASM turbidity profiling instrument (ARGUS, Germany). The sampling frequency was 1 Hz over a duration of 30 s, and the measuring burst interval was 2.5 min. Turbidity data were converted into SSCs values via calibration with in situ sediment samples (see Sect. 3.2 of Yang 2017 for the calibration equation).

Fig. 3
figure 3

Schematic representation of the instrument deployment at Nanhui intertidal mudflat

Full size image

4 Model setup and validation

4.1 Model description

The Delft3D model, developed by the Delft Water Conservancy Institute in the Netherlands, integrates several modules for flow, waves, sediment transport, morphological development, and ecology and allows stable simulation of hydrodynamics and ecological processes in two (horizontal or vertical) and three dimensions in coastal, fluvial, and estuarine areas (Hu et al. 2009; Lesser et al. 2004). The control equations of each module and details of their use can be found in the manual on the website of Delft3D (https://oss.deltares.nl/web/delft3d/manuals) and are not repeated here. The WAVE module based on the SWAN model is a third-generation spectral wave model for nearshore applications. This module is based on the discrete spectral action balance equation and is fully spectral and tightly coupled with the storm surge hydrodynamic model. The FLOW module is the sediment transport and morphological development module and is based on the incompressible fluid, shallow water, and Boussinesq assumptions of Navier–Stokes equations to calculate hydrodynamics, sediment transport, and morphological changes simultaneously.

Bathymetric data for the study area was digitized from nautical charts and corrected to MSL. Surface wind fields used to activate the model were obtained (the ECMWF 6 h interval, 10 m height surface wind field dataset: https://cds.climate.copernicus.eu/cdsapp#!/dataset/reanalysis-era5-single-levels?tab=form). The accuracy of this reanalysis dataset has been verified by a large number of measured data (Fan 2019; Yang 2017) and is widely used in model wind field driving (Ge et al. 2020; Kuang et al. 2020a; Zhang et al. 2018).

SSC near Hangzhou Bay is high, but vertical gradients are small (**e et al. 2009), so a 2D model was chosen for simulations. To observe the spatial variation in erosion and accretion with high accuracy, the nearshore grid resolution was increased locally. Similar results were obtained for resolutions of 30 and 10 m, so the 30 m grid was used to improve the computational efficiency. The calculation area covers most of the Nanhui intertidal mudflats. The simulation time range was from 7:00 on 12 September to 7:00 on 26 September 2014, with a time step of 0.2 min. On the open boundaries, water level time-series provided by a large-scale circulation model (Bixuan Tang, unpubl.) were used to drive the local Delft3D model, in which both the astronomical tidal component and the storm surge component are taken into account. In addition, wave boundaries derived from the large model were entered in the WAVE module for online coupling with the FLOW module.

Model calibration focused on regional topographic variability, where parameters such as settling velocity, critical bed shear stress for erosion (τe) and deposition (τd), and erosion parameter (M) are critical for sediment transport and morphological development (Hu et al. 2018). Therefore, these parameters need to be calibrated before they can be used to study intertidal mudflat morphological changes during storms. Based on the results of available sediment core analyses from the study area (Wu 2013, 2014) and sensitivity experiments, the sediments are classified as cohesive and non-cohesive particles. The initial bottom bed was set in two layers, with the first layer being 0.1 m of cohesive particles and the second layer being 4.9 m thick, including 4.4 m of cohesive particles and 0.5 m of non-cohesive particles. The bottom friction coefficient was determined according to the Manning coefficient formula from (**ng et al. 2012). The cohesive particle settling velocity, τd, τe, and M were taken as 1 mm/s, 2 N/m2, 0.15 N/m2, and 2 × 10−5 kg/m2/s, respectively. The median diameter of non-cohesive particles was 100 μm, and other parameters were assigned the default settings in Delft3D. In this study, we did not consider the impact of the harbor dam in Fig. 1c.

4.2 Model validation

The following skill model (Willmott 1981) was used to assess model performance:

$$skill=1-\frac{\sum_{i=1}^{n}{|S-D|}^{2}}{\sum_{i=1}^{n}{(\left|S-\overline{D}\right|+|D-\overline{D}|)}^{2}}$$
(1)

where S and D represent the simulated and observed values, respectively; \(\overline{D }\) is the mean value of the observed data; and n is the number of observed data. The model performance is perfect for a skill value of 1.0, excellent for skill values between 0.65 and 1.0, very good for skill values between 0.5 and 0.65, good for skill values in the range of 0.2–0.5, and poor for skill values below 0.2.

The validation results compared with the observed data are shown in Fig. 4. The observed values of the three water level observation stations (Fig. 4a, b, and c) are consistent with the simulated values, and the skill values exceed 0.95 (S1 = 0.9609, S2 = 0.9595, and S3 = 0.9730). The skill values for current direction, current velocity, and wave height in S3 are 0.6431, 0.6109, and 0.8849 (Fig. 4d, e, and f), respectively. The simulated data are more stable than the observed data, probably because S3 is located near the supratidal zone, and there are additional influencing factors in the field (such as biological activity and human activities) that are not considered in the model. More factors affect SSCs than are considered in the model. Kuang et al. (2020b) thought that the skill value of SSCs above 0.2 indicate the model can capture the first-order variation in SSCs. So in this study, the skill value of SSCs at station S3 is 0.5752, signifying a good level of model performance, and the time-series comparison between simulated and observed SSCs also demonstrates that the model closely reproduces the change in SSC with time. All of the above indicate that the established model suitably describes the current field and suspended sediment transport during the storm period in the study area.

Fig. 4
figure 4

Comparisons of modeled and observed water level (a, b, and c) at stations S1-S3, current velocity (S3) (d), current direction (S3) (e), wave height (S3) (f), and SSC (S3) (g) for the period of storm Fung-wong

Full size image

5 Results

5.1 Physical processes during the storm

5.1.1 Hydrodynamic variation

During normal weather, the current field shows a reciprocating pattern (Fig. 5a and b). The maximum flood velocity is 0.49 m/s in the study area, directed to the west. The maximum ebb velocity is 0.46 m/s, directed to the east. The wind speed in the study area is 4.6–6.6 m/s, and the wind direction is easterly. In the western part of the study area, the direction of the flood and ebb currents are both parallel to the dike, owing to the shelter provided by the dike on the shore. During storm conditions (Fig. 5c and d), strong NE winds of 15 m/s increase the flood velocity to 0.6 m/s, whereas the flood velocity over most of the intertidal mudflats drops to 0.2 m/s at ebb. Strong winds change the direction of the current on the low mudflat in the near-offshore area to be aligned with the wind direction, but the direction of the current on the high mudflat near the shore is less affected by wind. Two tidal cycles (T4 and T9) with similar water depths during normal weather and storm conditions were selected to analyze the hydrodynamic changes at the three points (N1, N2, and N3 in Fig. 1c; Table 1) from the high to low mudflats. The mean current velocities at the three points increase from 0.11, 0.14, and 0.21 m/s during T4 to 0.14, 0.20, and 0.21 m/s during T9, respectively.

Table 1 Results of hydrodynamics for the analysis points during normal weather (T4) and storm conditions (T9)
Full size table
Fig. 5
figure 5

Current fields of maximum flood (a, c, e, and g) and maximum ebb (b, d, f, and h) during normal weather (a and b), storm conditions (maximum wind speed in the study area during storm Fung-wong = 15 m/s) (c and d), storm conditions (maximum wind speed in the study area during a storm = 20 m/s) (e and f), and storm conditions (maximum wind speed in the study area during a storm = 25 m/s) (g and h)

Full size image

Waves are another dominant influence on the Nanhui mudflats, and their propagation is influenced by tidal fluctuations and wind speed. As shown in Fig. 6b and c, because of the depth-limited wave-breaking, wave height variation at the three points shows a strong tidal signal, with maxima and minima at high and low tide, respectively. During storm conditions, wave heights at the three points are substantially larger than those during normal weather. For example, the wave heights at N1, N2, and N3 during T9 are 0.34, 0.54, and 0.66 m; i.e., 2.43 (0.14 m), 3.18 (0.17 m), and 3.88 (0.17 m) times larger than during T4 respectively (Table 1). The difference in wave height between normal weather and storm conditions is more pronounced on the low than high mudflat.

Fig. 6
figure 6

Temporal variations in wind velocity (The legend arrow points east) (a), water depth (b), wave height (c), SSC (d), and suspended sediment transport rate (SSTR) (e). Green, brown, and blue lines represent points N1, N2, and N3, respectively. Shading represents the storm period

Full size image

5.1.2 Sediment transport and bed-level changes

The mean SSCs during T4 at N1, N2, and N3 are 0.45, 0.18, and 0.20 kg/m3 during normal weather, respectively (Table 1). The corresponding values during storm conditions are 2.94, 1.90, and 2.02 kg/m3 during T9; i.e., 6.53, 10.56, and 10.1 times greater than those during normal weather, respectively. The mean SSTRs at the three analysis points are ~ 0, 1 × 10−5, and 2 × 10−5 m3/s/m during T4, and they are 1.1 × 10−4, 2 × 10−4, and 2.6 × 10−4 m3/s/m during T9, an increase of one order of magnitude. As can be seen from the temporal variation in cross-section elevation (Fig. 7a), the change in elevation during storm Fung-wong (21–23 September) is substantially greater than that during normal weather (18–20 September). Specifically, the elevation changes at N1, N2, and N3 during 18–20 September are − 0.04, − 0.08, and − 0.10 m, compared with − 0.10, − 0.20, and − 0.27 m during 21–23 September. Changes in elevation at N1, N2, and N3 during storm conditions are 2.50, 2.50, and 2.70 times greater than those during normal weather, respectively.

Fig. 7
figure 7

a–c Bed-level change of the cross-section before, during, and after storm conditions for maximum wind speeds of (a) 15 m/s (storm Fung-wong), (b) 20 m/s, and (c) 25 m/s (Compared to the time the model started running). d Bed-level change of the cross-section at different wind speeds during the storm

Full size image

Figure 8a shows the mean SSTR on the intertidal mudflat during storm Fung-wong, with the main transport direction being to the southwest; i.e., offshore. This direction is consistent with the direction of the prevailing wind (ENE) during the storm, indicating that wind is an important influence on sediment transport. SSTR on the high mudflat is lower than that on the low mudflat during T9 (15 m/s) for N1, N2, and N3 with SSTRs of 1.1 × 10−4, 2 × 10−4, and 2.6 × 10−4 m3/s/m, respectively (Table 1). This difference is probably due to the faster current velocity and stronger wave-breaking processes on the low (cf. high) mudflat. There is a divergence in the elevation change of the mudflat during the storm, and the intensity (amount) of erosion increases gradually over time from the high to the low mudflats (i.e., from shore to sea), with < 0.1 and 0.35 m of erosion occurring on the high and low mudflats, respectively (Fig. 8b). The mean bed shear stress (BSS) on the high mudflat is < 0.1 N/m2, that on the low mudflat reaches 0.7 N/m2, and higher values of BSS are associated with greater amounts of erosion (Fig. 8b).

Fig. 8
figure 8

a, c, and e SSTR and b, d, and f mean BSS and erosion–accretion (bed-level change) results for maximum wind velocities of 15 m/s (during storm Fung-wong) (a and b), 20 m/s (c and d), and 25 m/s (e and f)

Full size image

5.2 Intertidal mudflat erosion and accretion at different wind speeds

The maximum wind speed during the past 20 years near the study area is 22.1 m/s (from ECWMF), and the maximum wind speed during storm Fung-wong was 15 m/s (Fig. 2b). To investigate the effect of wind speed on intertidal mudflat erosion and accretion, the maximum wind speed during a storm in the study area was set in the modeling to 20 and 25 m/s, respectively, with wind direction kept constant (without considering the effect of air pressure). Results for current field and bed-level change of the mudflats under the two scenarios are shown in Figs. 5e–h, 7b and c, and 8d and f.

The enhancement of wind speed causes an increase in flood current velocity and changes the direction of the ebb current to be aligned with the wind direction. When the maximum wind speed in the study area is adjusted in the model from 15 m/s to 20 and 25 m/s, the maximum flood current velocity of the intertidal mudflat increases from 0.6 m/s to 0.8 and 1.1 m/s, respectively (Fig. 5e and g). The maximum ebb velocity also increases from 0.55 m/s to 0.7 and 0.95 m/s (Fig. 5f and h). The difference in current direction between the high mudflat and low mudflat becomes smaller with increasing wind speed. SSTR increases with increasing wind speed (Fig. 8a, c, and e). The direction of transport remains the same owing to the consistent wind direction, and SSTR increases more on the low mudflat compared with the high mudflat (i.e., in the offshore direction), with a maximum increase of one order of magnitude compared with the wind speed of 15 m/s of storm Fung-wong. The maximum BSSs on the intertidal mudflat are 0.7 and 1 N/m2 at maximum wind speeds of 20 and 25 m/s, respectively. The maximum erosion magnitudes for maximum wind speeds of 20 and 25 m/s are 0.43 and 0.52 m, respectively, and both are larger than 0.35 m of 15 m/s. The stronger the wind, the more uneven the slope of the mudflat and the steeper the profile (Fig. 7a-c). The overall spatial pattern for erosion–accretion for the higher wind speeds is consistent with that of storm Fung-wong, with an increase in erosion from the high to the low mudflats (Fig. 8b, d and f). With increasing wind speed, the analysis points also show an increase in current velocity, wave height, SSC, SSTR, BSS, and section elevation for the same tidal period (T9), as revealed in Table 1 and Fig. 7b and c.

6 Discussion

6.1 Spatial variation in intertidal mudflat erosion and accretion patterns during storms

Hydrodynamic and erosion–accretion responses varied in different parts of the studied intertidal mudflat during storm Fung-wong. A comparison of two tidal cycles (T4 and T9) with similar water depth between normal weather and storm conditions shows that the increase in wave height, mean SSC, SSTR, and bed-level change are all more pronounced on the low mudflat than the high mudflat, implying that the response of the former in the study area is more sensitive to storms than the latter. Within the Nanhui mudflat under storm conditions, almost the entire mudflat acts as a sediment source, transporting sediment to the ocean, but the low mudflat supplies more. Hu et al. (2018) found that the marshes in the western part of the bay where erosion occurred were the sediment source, while the accretive eastern marshes were the sediment sink during the storm. In addition, the spatial distribution of erosion and accretion at different wind speeds shows the same pattern, as shown by the bed-level change of the cross-section presented in Fig. 7d.

The occurrence of this erosion and accretion pattern on the intertidal mudflat is ascribed to two main controls. The first is the varying submersion duration of different parts of the mudflat. The submersion duration at N1, N2, and N3 during normal weather (T4) and storm conditions (T9) are 7, 7–8, and 12–13 h, respectively, with the submersion duration of the low mudflat being longer in all cases than those of the high mudflat. The increase in storm wind speed from 15 m/s to 20 or 25 m/s does not change the submersion duration in this study area, suggesting that the height of the bottom bed of the mudflat itself may control the submersion duration of the intertidal mudflat. The height of the bottom bed of the high mudflat (e.g., N1) is in all cases higher than that of the low mudflat (e.g., N3), and therefore the tide causes the high (cf. low) mudflat to have a shorter submersion duration and is thus less affected by hydrodynamic processes. Even when the low mudflat is at low tide, so long as it continues to be covered by shallow water, waves should still have sufficient time to act on the bottom bed. The magnitude of bottom-bed change under shallow-water conditions during non-storm periods can be up to 35% of that of the entire tidal cycle (Shi et al. 2017). Given the increased intensity of hydrodynamic processes on tidal flats during storms, the magnitude of bed-level change under shallow-water conditions may be even more substantial than in non-storm periods.

The second control on the erosion and accretion pattern on the intertidal mudflat is the spatial gradient of current velocity on the mudflat, which is characterized by a generally decreasing trend toward land (Zhou et al. 2022); that is, the intensity of hydrodynamic processes on the low mudflat is substantially greater than that on the high mudflat, as is the BSS. Strong wind enhances hydrodynamic power, which increases variation in bed-level change, but the spatial trend of decreasing current velocity toward land remains unchanged.

A spatial pattern of bed-level erosion and accretion during storms similar to that of the Nanhui intertidal mudflat has also been observed in the Kapellebank estuary in the Netherlands (De Vet et al. 2020), where the significant wave height of both the high and low mudflats exceeds 0.60 m. However, because of the longer emersion duration of the high mudflat than the low mudflat at Kapellebank and the difference in hydrodynamic strength (peak current velocity decreases with increasing bottom-bed elevation), the bed-level change of the high mudflat (< 0.5 cm) is much less than that of the lower mudflat (20 cm) (De Vet et al. 2020). Hu et al. (2018) found a negative relationship between bed-level elevation and BSS for Jamaica Bay Estuary in the United States, with the same spatial distribution of more significant erosion of the low (cf. high) mudflat, which was ascribed to differences in hydrodynamic intensity and submersion duration. In summary, erosion of the low (seaward) mudflat is greater than that of the high (nearshore/landward) mudflat in unvegetated meso–macrotidal environments.

6.2 Factors influencing the erosion–accretion pattern of intertidal mudflats under storm conditions

Our model results indicate that the magnitude of erosion–accretion of intertidal mudflats increases with increasing wind speed (Figs. 7 and 8d, f). The spatial and temporal variations in the erosion–accretion of mudflats involves complex processes operating under the combined influence of meteorological, hydrological, and morphological conditions (Kuang et al. 2020a; Xu et al. 2016). Different factors determine the variation in the erosion–accretion of mudflats over different time scales, and storms increase the importance of meteorological factors at short time scales. Wind transfers large amounts of energy to surface waters, generating wind-driven currents, which in turn influence flood and ebb velocities and the strength and direction of currents under normal weather conditions (Fan 2019), even promoting currents on mudflats instead of the tide (De Vet et al. 2018). In addition, wind can generate waves and increase wave heights, thereby enhancing wave energy. According to Fig. 6c, the wave heights at analysis points N1, N2, and N3 during tide cycle T9 (storm conditions) are 0.67, 0.92, and 1.02 m, respectively, compared with those during T4 (normal weather), which are 0.20, 0.22, and 0.23 m, respectively. These results imply that the increase in wave height is related to increased wind speed, as well as the increased wave energy dissipation or wave height attenuation that occurs during storm conditions compared with normal weather. These enhanced hydrodynamic conditions and dissipation of large amounts of energy enhance BSS, enabling more sediment to be entrained and transported, thereby increasing the magnitude of intertidal mudflat erosion.

Factors other than wind speed affecting the pattern of intertidal mudflat erosion–accretion are not explored here, and their roles during storms should be studied in the future. For example, tide level modulates wave height by controlling water depth and causing a shift in hydrodynamic intensity and wave-breaking position, thereby triggering changes in the magnitude and pattern of mudflat erosion and accretion (Hu et al. 2017; Kim 2003). In this paper, the level and timing of the tide of the storm period were very close to those of the spring tide. The hydrodynamic intensity and BSS of a spring tide are usually greater than those of a neap tide for similar wind conditions (Shi et al. 2015, 2017). Sediment transport and the magnitude of erosion and accretion are expected to also follow this trend. In addition, the erosion rate of weak storms during spring tides has been shown to be greater than that of strong storms during neap tides (Fan et al. 2006). Therefore, we infer that the overall bed-level variation of an intertidal mudflat at neap tide is less than that at spring tide. Furthermore, wind direction can trigger bed-level change by affecting the current field, sediment transport direction, and water level (Kuang et al. 2020b). In the present study, a NE–SW-oriented erosion zone appears in the study area for all three wind-speed scenarios, which may be related to the prolonged strong ENE winds that occurred during storm Fung-wong. In addition, the geometry of the intertidal mudflat itself, such as the slope of the bottom bed, can affect the spatial hydromorphological response to storms. De Vet et al. (2018) showed that the steeper the bed, the larger the bed-level variability during storm events. Furthermore, human activities and interventions should also be considered, as the relative angles of waves and currents tend to vary spatially near structures such as artificial dikes, which cause a clear contrast between erosion and accretion in their vicinity (Zhang et al. 2021), as in the case of accumulation near the dike in the Nanhui study area.

The difficulty of simultaneously obtaining large-area observational data during high-energy conditions such as those generated by storm events means that quantitative model validation is complex. Failure of the modeling to consider some phenomena, such as the generation of fluid mud on the Nanhui intertidal mudflat during storm conditions, may lead to the deviation between the model and nature. However, the model provides an acceptable representation of the variation in physical processes in a coastal area during storm conditions. The combined effects of fluid mud, salinity, and biology on these processes and patterns of mudflat erosion–accretion could be considered in future models to improve further the understanding of changes in mudflats under extreme weather conditions and provide a reference for the better management of mudflats. Moreover, reasonably increasing the number of validation stations on the mudflats is conducive to improving the accuracy and persuasiveness of model simulation.

7 Conclusions

The hydrodynamic response and erosion–accretion pattern of the Nanhui intertidal mudflats under storm weather conditions and different wind speeds during storm Fung-wong in September 2014 was analyzed using the Delft3D model and observed data. Our main conclusions are as follows:

  1. (1)

    Compared with normal weather, storm Fung-wong substantially increased the intensity of hydrodynamic processes on the mudflats and the magnitude of erosion and accretion. The wave height, SSC, SSTR, and magnitude of erosion–accretion all increased during storm conditions compared with normal weather. Wave heights during the storm were 2.40–3.88 times higher than those of normal weather during tidal cycles with similar water depth, SSC increased by 6.53–10.56 times, mean SSTR increased by order of magnitude, and the magnitude of erosion–accretion increased by 2.28–2.70 times.

  2. (2)

    Owing to the greater submersion duration and hydrodynamic intensity of the high mudflat compared with the low mudflat, spatial differences occurred in the response of the Nanhui intertidal flats to storm Fung-wong. The geomorphological response to the storm increased from the high to low mudflats, and the amount of bed-level change also gradually increased, with the bed-level change at three points from the high to low mudflats being − 0.13, − 0.15, and − 0.17 m during the storm (maximum wind speed of 15 m/s).

  3. (3)

    An increase in wind speed increases the hydrodynamic strength and erosion–accretion magnitude of the intertidal mudflats and even changes the direction of currents. The changes in bed-level of three analysis points from the high to low mudflats are − 0.25, − 0.28, and − 0.32 m for an assumed maximum regional wind speed of 20 m/s and are − 0.36, − 0.43, and − 0.51 m for a wind speed of 25 m/s. However, wind speed has a negligible effect on mudflat submersion duration and does not change the established spatial pattern of greater erosion of the low mudflat and lesser erosion of the high mudflat.