Beaches within estuaries and bays (BEBs) are dynamic systems that are often neglected in research due to a perception that they are simply scaled down versions of open-ocean systems (Jackson et al., 2017; Nordstrom and Jackson, 2012; Vila-Concejo et al., 2020; Vila-Concejo et al., 2024). While their energy environment is often lower than on coasts directly exposed to deep-water ocean waves, modelling their response to sea-level rise is difficult as their morphology and dynamics are strongly related to local-scale boundary conditions. For example, many may only be active during high-energy events rather than modal conditions, especially deep inside the estuarine basin away from the open-ocean entrance (Goodfellow and Stephenson, 2005; Kennedy, 2002; Gallop et al., 2020; Costas et al., 2005). Those connected to major sediment sources such as flood-tide deltas (Fellowes et al., 2021; Gallop et al., 2020; Bittencourt et al., 2001) are more able to adjust their profile to changing energy conditions compared with those dependant on highly localised sediment sources such as cliff erosion (Jackson et al., 2002; Vila-Concejo et al., 2020).

Human development on the margins of estuaries and bays also adds additional complexity to understanding future behaviour (Nordstrom, 1992). The shorelines of urbanised beaches are often highly developed with sea walls, groynes, pipes and other hard structures dominating the intertidal zone (Nordstrom, 1989; Bird and Lewis, 2014; Kennedy et al., 2025). In many cases, beaches are constructed, or nourished, as ‘nature-based’ coastal defence (Morris et al., 2019; IUCN, 2020; Chen et al., 2022) as well as to create new recreational resources (Hanson et al., 2002). The result is many BEB’s can be considered to be artificial, with their morphology a product of deliberate or unintended management actions.

As urban areas grow, the degree of management intervention into the littoral zone scales through time with development. For understanding contemporary beach dynamics, it therefore becomes necessary to view morphologic behaviour in the context of a history of human modification, both deliberate and unintentional. In this study, we explore the morphodynamics of a highly managed temperate sandy beach in order to understand the tipping points of these vulnerable geomorphic systems through time. Using a fetch-limited system in Port Phillip Bay, southeastern Australia, we aim to understand how the contemporary beach movement is determined by present and past management activity. The aim is to assess the impact of increased human intervention on shoreline behaviour in order to determine how changing management priorities have affected shoreline movement. We use a case-study approach as an example of one pathway of beach behaviour. Such information is critical in the context of greater uncertainty with climate change and population growth which is expected to impact estuarine environments to a greater degree than open-ocean coasts in the future (Vila-Concejo et al., 2024).

Sandringham Beach is a 3 km long sandy system (centred at 37°57′25.47″S, 145° 0′23.61″E) located on the northeastern edge of Port Phillip Bay, Victoria, Australia (Figure 1a). The Sandringham embayment is composed of two tertiary-scale sediment compartments, Sandringham Beach and Half Moon Bay, separated by a rocky promontory (Red Bluff, see Figure 1b) (Kennedy et al., 2025). The beach is found at the base of cliffs (up to 30 m high) formed in sandstones and conglomerates of late Miocene to early Pliocene age (VandenBerg, 2016).

Port Phillip Bay is an enclosed estuary 54 km long (north-south) and 32 km wide, extending to 56 km width through the Geelong Arm (Figure 1a). The bay floor is on average 9–24 m deep and shallows to <5 m depth within 12 km of its bedrock-defined entrance at its southern edge (Dalby et al., 2024). Tides are microtidal, with a mean spring tide range of 1.4 m at the bedrock-defined entrance, decreasing to 0.9 m further inside the bay (PoM, 2013). Waves near the entrance have a mean significant height (H _s ) of 1.0–1.5 m, but they do not propagate far inside the estuary past the limit of the floodtide delta. The majority of the bay is fetch-limited dominated by seasonal wind patterns (Goodfellow and Stephenson, 2005; Bird, 2011; Kennedy et al., 2023; Lowe and Kennedy, 2016) with mean annual H_s of <0.5 m in the Sandringham region (Tran et al., 2021). Winds are predominantly southerly in the austral summer and northerly in winter. The 100-year return period storm tidal elevation at Sandringham is 1.07 m (Tran et al., 2021).

Sandringham Beach has been subjected to major management interventions for over a century (Figure 1b). At the northern and southern ends of the compartment, harbours, seawalls and breakwaters were constructed from the 1880’s with a major phase of seawall construction occurring in the 1930’s and 1940’s (Figures 1b, 2a). Further to this, the bluffs behind the beach have been buried in building waste presumably to infill old gullies likely similar in form to the current face of Red Bluff or the rear of the Middle sub-compartment (Figures 2b,c). The result of these actions is the eroding cliffs can no longer be considered a significant source of beach sediment.

Intentional interference with longshore sediment transport occurred in 1990 (opposite Royal Ave) in the form of a 30 m long rock groyne which was extended to 40 m between 1990 and 2010 (Figure 1b). A second rock groyne 80 m long was constructed in 2006 at Southy St, later shortened to 50 m in 2017 (Figure 2c). The hard structures of groynes and rocky outcrops define five sub-compartments within the embayment (North, Middle, South-1, South-2 and Half Moon Bay (HMB)) (Figure 1b). Nourishment projects have occurred concurrently with groyne construction, focussed on specific sub-compartments, rather than the entire beach at one time (Table 1). The first recorded was in 1986, with further projects in 1993, 2009, 2018 and 2021.

Shoreline change was analysed from aerial imagery using a vegetated and wetted lines proxy. The vegetation line, defined as the seaward limit of plant growth, was considered to represent the longer-term position of the shoreline and exclude short-term tidal and storm surge dynamics (Burningham and Fernandez-Nunez, 2020), however it is acknowledged that the active management at the site means a natural equilibrium between plant growth and sedimentation is unlikely to occur. This means the vegetation lines can often correspond to the position of artificial structures such as seawalls. The wetted line, defined by the change in colour of sand resulting from wave and tidal inundation, represents an instantaneous shoreline at the time of image acquisition (see Boak and Turner (2005); Pajak and Leatherman (2002) for a review of these proxy methods). A vertical elevation of 0.5 m AHD which is approximately the location of mean high water spring (MHWS) tide elevation has previously been used to delineate the shoreline by the Victorian Government Department of Energy, Environment and Climate Action, to streamline shoreline analyses. This study also tested the reliability of this delineation using Digital Surface Models (DSM’s) produced through Unoccupied Aerial Vehicle (UAV) surveys.

Historical aerial images were obtained from the Victorian Government, spanning 93 years from 1930 to 2023. Vegetation and instantaneous waterlines were manually extracted from the georectified imagery. No tidal correction was applied. The minimum horizontal resolution of older images (pre-2000) is 0.5 m, with a maximum of 0.1 m for newer images (2010 onward). An independent georectification (Carvalho et al., 2020) of images from the same dataset determined a root mean square error (RMSE) of ≤1 m for images from 1960 onwards and due to the same workflows being conducted in this study we assume that older images have a similar RMSE. Based on a comparison with that dataset, we conservatively estimate an uncertainty of ±3 m for the extracted vegetation lines based on vectorisation and image errors. All shorelines measured are referenced to the 2019 a tidally-adjusted annual mean wet-dry line extracted from Digital Earth Australia (DEA) (DEA, 2022; Bishop-Taylor et al., 2021). The 2019 shoreline was selected as shadows from the cliffs behind the beach, meant many of the shorelines in the DEA database were incorrect. The DEA shorelines have a Mean Absolute Error (MAE) of 7.3 m and RMSE of 10.3 m (DEA, 2025).

UAV surveys, using a DJI Phantom 4 Pro and RTK models, with RTK-GPS ground control points, were undertaken every 6–8 weeks from April 2020 until May 2023 with a vertical and horizontal accuracy (RMS error) of 0.11 and 0.02 m respectively (see Pucino et al. (2021); Ierodiaconou et al. (2022) for methodological and quality assurance details). All imagery was processed into digital surface models using Structure-from-Motion techniques (Westoby et al., 2012). For all aerial datasets, transects at 30 m spacing were cast across all time steps to analyse shoreline change. Rates of change in shoreline position were calculated based on linear regression of the entire dataset. Further details on the method of integration of UAV and satellite datasets for coastal management can be found in McCarroll et al. (2024).

Concurrent with the UAV surveys, a SOFAR Spotter wave buoy was deployed in approximately 10 m water depth, 1.6 km offshore southwest of Black Rock Jetty (Figure 1b). The buoys use high-resolution GPS to measure its vertical and horizontal motion. This 3D motion time series was processed using the Fast Fourier Transform (FFT) to generate the wave energy spectrum, from which significant wave height, wave period, and wave direction are derived. These wave parameters are transmitted to remote servers and visualized via a user dashboard. In this study, the bulk wave parameters were directly obtained from AUSWAVES (https://auswaves.org/), a national platform for wave observation and data sharing.

To calculate sediment budgets, shoreline position (wetted-line) was used to determine active profile volume. This was completed using the assumptions of equilibrium profile shape, depth of closure and active profile height (for full details see (McCarroll et al., 2021; Harley et al., 2022; Dean and Houston, 2016) (See Supplementary Table S1). As a general rule shoreline change (δX) is a function of the volumetric change (δV) and height of the active profile (h_a). The depth of closure was estimated from Hallermeier (1980)) combined with examination of single beam echosounding profiles perpendicular to the field site. Volume may be expressed as per unit alongshore (m³/m) or as a total volume (m³) by integrating alongshore. δ X = δ V   /   h a (1)

Sandringham Beach is a system that has been subjected to significant human intervention in the past 145 years (Cardno, 2016; Bird, 2011). The first phase of intervention prioritised the stability of the subaerial landscape rather than the beach. This occurred most intensively from the 1880’s to 1950’s when harbours/marinas and seawalls were built, the former being constructed at either end of the sediment compartment. Armouring of the cliff occurred concurrently with these activities. All these activities were detrimental to the beach sediment budget, through blocking sand supply from the subaerial cliffs and interrupting longshore transport. Effectively these activities changed the sediment compartment from being open to closed (sensu Thom et al., 2018). This management scheme led to shoreline retreat from 1960 up to the 1990’s for the centre and southern ends of Sandringham Beach, with some accretion at the northern end of the beach at the downdrift end of the longshore sediment transport system. Enhanced shoreline erosion due to seawall construction is common especially in estuarine settings (Nordstrom, 1989). The main period of shoreline recession occurs after 1980, preceded by a period characterised by large shoreline fluctuations (1950–1980). The large fluctuations preceding a period of sustained retreat may imply a sediment system not in equilibrium and therefore in a period of adjustment to the major infrastructure construction activities being undertaken. While the total amount of retreat is low at Sandringham (−0.1 ± 0.1 m/yr based on the vegetation line since 1957), the narrow width of the beach meant it necessitated a change in management approach.

From 1990, groyne construction and nourishment became the primary management tool. This had the effect of progressively subdividing the tertiary compartment into smaller units (sub-compartments) (Figure 1). Seasonal rotation of the beach planform has been noted previously (Bird, 2011; Lowe and Kennedy, 2016), however the subdivision of the beach into smaller units made this adjustment quite pronounced (Figures 6, 7). The creation of the sub-compartments by 2006 saw the shoreline progressively widen until around 2018 where its position remained relatively stable. Groyne construction was accompanied by nourishment which assisted progradation and its impact of management can be observed in comparing the rates of shoreline change after 1990 (Table 2) (Figure 10). Sub-compartments which were previously eroding became accretionary (zone A). Only the North Sub-compartment is continually accreting as it is at the downstream end of the littoral drift system and has greater lateral accommodation space (Lowe and Kennedy, 2016).

Only one sub-compartment (Middle) shows continual erosion, except in the wetted line proxy 1990–2023 (Table 2). Here, the beach rests on top of an intertidal shore platform backed by a rocky cliff. The beach thickness here is low (<1 m), and this location likely operates more as a rocky shoreline with an ephemeral beach veneer (Kennedy and Milkins, 2015; Trenhaile, 2004). As rocky coasts are erosional in nature (Stephenson et al., 2013; Trenhaile, 1987) the long term retreat of this location is therefore not surprising.

As Sandringham Beach has undergone significant modification, it is necessary to differentiate between the natural and artificial sediment budgets to understand its future evolutionary trajectory. The sediment budget for the nourishment activities is based on reported volumes of sand added to the profile, though the accuracy of historical records is less than ideal. For sediment budget calculations, it is assumed that profile volume change is the product of shoreline change and active profile height (Dean and Houston, 2016; Rosati, 2005) (Equation 1) (Supplementary Table S1). This is generally a good assumption over longer time periods where longshore transport gradients are responsible for accretion/erosion. Conversely, this is a poor assumption where shoreline change is due alteration of the profile shape, e.g., soon after storms or nourishments. In the case of Sandringham, storm erosion is known to be minimal; however, total profile volume may be overestimated immediately following nourishments, until the profile shape re-adjusts (∼1-year) (Beetham et al., 2023). Through this approach, the natural sediment budget for Sandringham Beach is slightly positive over the long term, but less than the nourishment volumes. On the short term (last 3 years) the natural volume change is almost the same as the nourishment volume (Table 4). The natural source of sand is most likely alongshore from adjacent compartments (Bird, 2011; Lowe and Kennedy, 2016; Kennedy et al., 2025) although an offshore source, from below 7 m depth, cannot be discounted (Dalby et al., 2024).

An alternate approach for understanding sediment budgets for the past 3 years can be taken through directly measuring volumetric change (Figure 5b). The UAV data shows that South-2 accreted by 6,433 m³ post-nourishment representing 51% of the volume introduced by the contractor, the latter estimated by a pump delivery rate rather than direct measurement. The nourishment led to an increase of beach volume up to 7,007 m³ when compared with December 2020. This material was then reworked by April 2022, but 93% returned in the summer of 2022–2023. Given the presence of extensive subtidal bars, sediment exchange between the subaerial and submarine components is to be expected (Short and Jackson, 2013). The variation in UAV-measured volume therefore likely relates to sediment exchange above and below mean sea level.

While the volumes measured from the UAV are less than those calculated from standard sediment budgets equations (Dean and Houston, 2016), given the different proxies and assumptions of the methods plus the UAV only measures subaerial change, the calculated longshore drift rates can be considered geomorphically similar. These rates (maximum 8,050 m³/yr) (Table 4) are at the lower end of measured longshore transport rates of 3,600–13,300 m³/yr on estuarine beaches with a 7–9 m wide active beachface exposed to a 15 km fetch at Fire Island New York, USA (Jackson et al., 2017; Nordstrom et al., 2003). The wave height on Fire Island was 0.1–0.3 m. A similar wave height also caused longshore drift rates of 12,000 m³/yr on beaches 20 m wide in the Tagus Estuary Portugal (Freire and Andrade, 1999). The maximum H_s during the past 3 years at Sandringham Beach was 0.41 m and while slightly larger than those measured at Fire Island and Tagus Estuary, it is likely that the strong seasonal rotation drives the lower longshore transport rates at the entire compartment scale measured in this study. The rates at Sandringham Beach are however an order of magnitude less than those of open-ocean beaches such as Collaroy-North Narrabeen in eastern Australia which also undergoes major seasonal rotation of the planform profile (Harley et al., 2015).

The fluctuations in shoreline position and the associated sediment budget show that Sandringham Beach is mobile, with only a small sediment deficit (Figure 11). The recent nourishment program in 2021 had the most significant change on the shoreline dynamics acoss the entire beach, excluding the rocky sub-compartment. Associated with this shoreline dynamic has been a background rise of mean sea level within Port Phillip Bay of 4.8 mm in the past 29 years based on data from the Hovell Pile near the entrance to Port Phillip Bay (PSM, 2022) but the shoreline trends, at least at present, appear to be operating independently of this driver.

Sandringham Beach represents an important beach form, namely, a sandy system in a fetch-limited embayment removed from the influence of deepwater ocean swell. Such systems are found globally in large estuarine embayments and are commonly the location of significant urban development. Sandringham Beach, and much of the shoreline of Port Phillip Bay, is highly managed with beach nourishment having replaced hard armouring as the primary method of management of the sediment budget. The change in management schemes has been partly driven by a change in priorities from protection of the hinterland to maintenance of a functioning sedimentary system for recreation and as nature-based coastal defence.

It is found that management interventions on the beach have been a major driver of shoreline change in the past century. While seawalls initially reduced natural sediment supply, this has been balanced recently by nourishment programs (Figure 11). Of greatest significance to shoreline movement has been the subdivision of the tertiary-scale compartment into several sub-units through the construction of groynes. This subdivision has enhanced the seasonal rotation of the beach, by containing the sediments within small compartments which operate quasi-independently on a seasonal basis. This likely has had the effect of slowing the decadal-scale longshore migration of sand along the northeastern coastline of Port Phillip Bay. As nourishment now replaces the previously supply of sediment sourced from cliff erosion, the beach system can be considered as in an equilibrium state. Our findings show how understanding the sediment dynamics over seasonal to decadal scales can lead to the sustainable management of the coastal environment.