Dynamics of Sandy Shorelines and Their Response to Wave Climate Change in the East of Hainan Island, China

Abstract

Beach erosion and shoreline dynamics are strongly affected by alterations in nearshore wave intensity and energy, especially in the context of global climate change. However, existing works do not thoroughly study the evolution of the sandy coasts of eastern Hainan Island, China, nor their responses to wave climate change driven by climate variability. This study focuses on the open sandy coast and assesses shoreline evolutionary dynamics in response to wave climate variability over a 30-year period from 1994 to 2023, using an open-source software toolkit that semi-automatically identify the shorelines (CoastSat v2.4) and reanalysis wave datasets (ERA5). The shorelines of the study area were extracted from CoastSat, and then tidal correction and outlier correction were performed for clearer shorelines. Combining the shoreline changes and wave conditions derived from ERA5, the dynamics of the shorelines and their response to wave climate change were further studied. The findings reveal that the average long-term shoreline change rate along the eastern coast of Hainan Island is 0.03 m/year, with 44.8% of transects experiencing erosion and 55.2% showing long-term accretion. And distinct evolutionary patterns emerge across different sections. Interannual variability is marked by alternating erosion and siltation cycles, while most sections of the coast experiences clear seasonal fluctuations, with accretion typically occurring during summer and erosion occurring in winter. El Niño–Southern Oscillation (ENSO) cycles drive changes in parameters including significant wave height, mean wave period, wave energy flux, and mean wave direction, leading to long-term changes in wave climate. The multi-scale behavior of the sandy shoreline responds distinctly to the ongoing changes in wave climate triggered by ENSO viability, with El Niño events typically resulting in accretion and La Niña periods causing erosion. Notably, mean wave direction is the metric most closely linked to changes in the shoreline among all the others. In conclusion, the interplay of escalating anthropogenic activities, natural processes, and climate change contributes to the long-term evolution of sandy shorelines. We believe this study can offer a scientific reference for erosion prevention and management strategies of sandy beaches, based on the analysis presented above.

1. Introduction

Sandy coasts are complex and dynamic natural systems, accounting for roughly 31% of the world’s ice-free coasts [1]. Their geomorphology and spatial–temporal dynamics are heavily influenced by waves and tidal actions [2,3]. However, the risk of coastal flooding and beach erosion caused by climate change and escalating anthropogenic activities is increasing in many regions [4,5,6]. Shoreline position is the most commonly used metric for assessing erosion and accretion on sandy beaches [7,8]. Sediment transport induced by wave and tidal processes in the nearshore zone results in continual changes to the sandy shoreline. The dynamics of sandy shorelines vary across different temporal scales. On short time scales, beach erosion is typically driven by storms or extreme wave events [9,10,11,12]. Over annual to inter-decadal scales, climate variability, such as the El Niño–Southern Oscillation (ENSO), significantly influences shoreline evolution [13,14,15,16,17]. On longer time scales, shoreline changes are linked to factors like sea level rise [18,19,20,21]. Spatially, coastal geomorphic features, including the distribution of headlands, can influence sediment transport processes and subsequently control the shoreline evolutionary pattern [22,23]. Understanding the long-term evolutionary patterns of shorelines at a regional scale is crucial for identifying trends in beach geomorphology, which is essential for effective coastal management and planning [24].

Nearshore waves play a crucial role in shaping beach morphology and driving sediment transport, thereby influencing shoreline dynamics [2]. Changes in wave climate, characterized by variations in wave parameters such as wave height, period, and direction, are closely linked to ocean–atmosphere interactions, with climate change capable of inducing long-term shifts by altering atmospheric circulation patterns [25]. Research indicates that the global mean annual wave climate has undergone significant changes due to climate change, with wave energy increasing by approximately 0.4% per year worldwide [26]. The dynamics of deep-water wave energy drive variations in nearshore breaking wave energy, which subsequently influence beach dynamics and geomorphology [27,28]. Consequently, changes in wave climate may have profound impacts on sandy beaches globally in the future.

Hainan Island is one of the most developed and concentrated sandy beaches in China and is famous for its high-quality beach resources. This area is surrounded by the sea; the wave climate is more active because of the long-term impact of extreme marine forcing. However, in recent decades, the dynamic balance of these beaches has been disrupted by frequent extreme events and increased anthropogenic activities, resulting in an elevated risk of beach erosion [29,30]. Therefore, it is necessary to investigate the response relationship of the sandy coast of eastern Hainan to climate variability in order to better assess the coastal vulnerability and erosion risk and develop protective countermeasures.

Previous studies on beach geomorphological evolution have primarily focused on short-term beach profile observations [31,32] and storm responses [12,33]. Yuan et al. [12] extracted and analyzed the coastline and beach slope data of 20 representative beaches in Hainan Island, revealing the significant seasonal changes in the location of the coastline. Based on measured beach profiles and surface sediments, Li et al. [31] analyzed the topographic variation and sedimentary dynamic characteristics of Haikou Bay beach. Liu et al. [32] clarified the natural and human factors driving the change in beach geomorphology through interdecadal monitoring of the coastline and profile topography of Sigentan, Hainan Island. Zhong et al. [33] studied the fractal characteristics of the spatial distribution of storm-induced coastline changes in Hainan Island and evaluated the upper limits of shoreline erosion and erosion and deposition. As global climate variability continues to change, identifying the impact of wave climate change on the evolution of sandy coastlines has become increasingly challenging and important. However, less attention has been drawn to the potential impacts of long-term changes in wave climate on the evolution of regional shorelines. As global climate variability continues to intensify, understanding the effects of wave climate changes on sandy shoreline evolution becomes increasingly important.

Building on the aforementioned background, this study utilizes open-source satellite imagery from the past 30 years to extract long time series and high-frequency shoreline data from the sandy beaches in eastern Hainan Island. Combined with long-term wave reanalysis datasets, we explore the response of sandy shorelines to ENSO-driven changes in wave climate across various time scales. The findings aim to provide a scientific basis for the protection of sandy beaches against erosion and to inform their integrated management.

2. Study Area and Data Sources

2.1. Study Area

The study area is located on the east coast of Hainan Island, China, within the cities of Wanning and Qionghai, adjacent to the South China Sea. It encompasses an open sandy coast approximately 53 km in length, stretching from Tanmen Harbor in the north to Dahuajiao Headland in the south, with geographic coordinates of 110°53′–110°63′ E and 18°79′–19°24′ N (Figure 1). The sandy coast predominantly runs north to south and can be divided into four sections based on regional geomorphological characteristics: Xiaohai, Hele-Zhengmenhai (HZ), Yudaitan, and Boao (Figure 1b). Figure 2 presents site photographs of the beaches within each of these sections.

Geologically, the study area is part of the Southeast Hainan Depression Belt, characterized by strata ranging from the Mesoproterozoic to the Cenozoic, with granite formations resulting from multiple phases of magmatic intrusion. The coastal strata primarily consist of Quaternary marine loose sediments. The surface sediments of the beaches are predominantly sand, with some areas containing biological debris, such as coral fragments and shells. The coastal region features the Xiaohai Lagoon and Boao Lagoon, forming a typical barrier–lagoon–tidal channel geomorphic system [34] (Figure 1d–e). In particular, the Boao coast features a beach–coral reef system, predominantly characterized by fringe reefs [35]. The coast is situated in a low-latitude tropical monsoon climate zone, with northeasterly winds prevailing in winter and southeasterly and southwesterly winds dominating in summer. It is frequently impacted by typhoons during the summer and autumn months, making it one of the most typhoon-prone regions on Hainan Island. River discharge, particularly from the Wanquan River, plays a significant role in coastal sedimentation by transporting inland sediments to the sea, often altering the geomorphology of the Wanquan River mouth [36].

The study area is generally wave-dominated and microtidal, with wave regimes primarily consisting of a mixture of wind waves and swells. The tidal pattern is classified as irregular diurnal, with an average tidal range of approximately 0.6–1 m. Tidal currents display an irregular semi-diurnal and diurnal pattern [34]. In recent decades, sandy coasts have increasingly faced anthropogenic interventions, such as the construction of breakwaters and offshore artificial islands (

Table 1. Information on the typical coastal structures.

Typical Coastal Structure	Location	Construction Time
Coral Artificial Island	Northern Boao (Figure 1c)	2011
Xiaohai Breakwater	South of the Xiaohai Lagoon inlet (Figure 1e)	2013
Xiaohai Artificial Dike	North of the Xiaohai Lagoon inlet (Figure 1e)	1972

). These coastal engineering projects, designed for resource development and natural disaster mitigation, have significantly altered the dynamics of beach morphology [37,38,39,40].

2.2. Data Sources

2.2.1. Satellite Imagery Sources

This study utilizes historical satellite imagery to analyze shoreline evolution in the study area over a 30-year period, from January 1994 to December 2023. Data were collected from Landsat 5, Landsat 7, Landsat 8, and Sentinel-2, which offer revisit intervals ranging from 5 to 16 days and spatial resolutions between 10 and 30 m.

For the shoreline extraction study, the relevant spectral bands across all four satellites include the red (R), green (G), blue (B), near-infrared (NIR), and short-wave infrared (SWIR1) bands. Additionally, Landsat 7 and Landsat 8 imagery contain a panchromatic (Pan) band, which provides high-resolution data for further analysis.

Table 2. Information on the satellite datasets used for analysis.

Satellite	Sensor	Time Coverage	Spatial Resolution	Temporal Resolution	Bands
Landsat 5	TM	1994–2011	30 m	16 days	R, G, B, NIR, SWIR1
Landsat 7	ETM+	1999–2023	30 m	16 days	R, G, B, NIR, SWIR1, Pan
Landsat 8	OLI	2013–2023	30 m	16 days	R, G, B, NIR, SWIR1, Pan
Sentinel-2	MSI	2015–2023	10 m	5 days	R, G, B, NIR, SWIR1

details the satellite datasets used, including their spectral bands, spatial resolutions, and coverage periods.

2.2.2. Wave Climate Datasets

Due to the absence of long-term historical wave measurements in the study area, this research utilizes the fifth-generation wave reanalysis (ERA5) dataset [41] from the European Centre for Medium-Range Weather Forecasts (ECMWF) to extract hourly wave data spanning 30 years to characterize the long-term wave climate of the study area. The ERA5 dataset provides global climate reanalysis data from 1940 to the present, with a spatial resolution of 0.25° × 0.25° and a temporal resolution of 1 h. Liu et al. [42] validated the ERA5 wave data against measurements from nearshore areas of the South China Sea, demonstrating that the ERA5 wave data accurately reflect the region’s wave characteristics. This validation confirms their applicability and reliability for this area.

In this study, wave parameters were extracted from three reanalysis grid points closest to the study coast: E1 (19°15′ N, 111° E), E2 (19° N, 111° E), and E3 (18°45′ N, 111°E) (Figure 1a). All three points are used to analyze the spatial difference in wave conditions. The 30-year wave climate parameters include significant wave height (Hs), mean wave direction (Hd), mean wave period (Tp), and wave energy flux (P, W/m) calculated from Hs and Tp as follows:

$$P = \frac{\rho g^2{Hs}^{2}Tp}{64\pi }$$

(1)

where $\rho$ = 1025 kg m^–3, and g is the gravitational constant. Wave energy flux is generally regarded as a more comprehensive indicator of long-term wave climate behavior than significant wave height and mean wave period alone [43].

2.2.3. Datasets for Validation and Tidal Correction

To validate the accuracy of the satellite remote sensing data extraction, this study employed Global Navigation Satellite System-Real Time Kinematics (GNSS-RTK) techniques. In August 2023, 25 profiles perpendicular to the coastline were established along the sandy coast of the study area. GNSS-RTK was utilized to obtain geographic and elevation data for these beach profiles, using the CGCS2000 coordinate system.

Tidal correction data were obtained from the FES2014 global tidal model [44], developed by Archiving, Validation and Interpretation of Satellite Oceanographic (AVISO+) data. This model offers tidal datasets derived from tidal hydrodynamic simulations, long-term altimetric data, and tide gauge assimilation. For tidal correction, instantaneous tidal levels with a 15 min time resolution were extracted from the FES2014 model. The details of the verification and tidal correction methods will be discussed in the subsequent sections.

2.2.4. Multivariate ENSO Index

To assess the specific impacts of climate variability on wave climate change, this study employs the Multivariate ENSO Index (MEI) [45] as an indicator of ENSO variability. The MEI is one of the most comprehensive indicators for describing ENSO events and can be used to identify periods of El Niño and La Niña. The monthly MEI datasets were sourced from the National Oceanic and Atmospheric Administration (NOAA) Physical Sciences Laboratory (https://psl.noaa.gov/enso/mei/, accessed on 15 June 2024). El Niño (La Niña) events are defined as periods when the running average of the MEI > 0.5 (<−0.5) for at least five consecutive months, with the peak MEI > 1 (<−1) [46]. In this study, three typical El Niño events (1997–1998, 2009–2010, and 2015–2016) and four typical La Niña events (1998–2000, 2007–2009, 2010–2012, and 2020–2023) were recorded over a 30-year period from 1994 to 2023 (Figure 3).

3. Methods

3.1. Shoreline Extraction

To generate the satellite-derived shorelines dataset, this study utilized the Python-based open-source toolkit CoastSat [47,48] to obtain high-frequency time series of shoreline positions along sandy coasts. CoastSat utilizes the Google Earth Engine (GEE) platform to collect, download, and process Landsat 5, 7, and 8 (Tier 1 Top-of-Atmosphere) and Sentinel-2 (Level-1C) images, enabling the extraction of sub-pixel resolution instantaneous waterlines with an accuracy of 10–15 m [48]. The CoastSat toolbox incorporates a supervised image classification algorithm based on a neural network model that classifies four distinct categories: “sand”, “whitewater”, “water”, and “other” (Figure 4b). To improve the accuracy of sand–water boundary detection, the sub-pixel resolution technique of the Modified Normalized Difference Water Index (MNDWI) [49] is applied within CoastSat to automatically extract instantaneous waterlines from selected images (Figure 4c). The MNDWI is an enhancement of the traditional Normalized Difference Water Index (NDWI) and has become a widely accepted method for differentiating water bodies from land features. Compared to NDWI, MNDWI is more effective at distinguishing water from high-reflectance surfaces such as sand, which makes it particularly useful in coastal studies. The calculation formula is as follows:

$$\mathrm{M}\mathrm{N}\mathrm{D}\mathrm{W}\mathrm{I}=\frac{SWIR1-G}{SWIR1+G}$$

(2)

where SWIR1 represents the pixel intensity of the shortwave infrared band (1.55–1.65 μm), and G corresponds to the green band (0.52–0.6 μm). The MNDWI value is calculated for each pixel, resulting in a grayscale image with values ranging from −1 to 1. Then, the Otsu method [50] is employed to determine an optimal threshold from the probability density functions of the four labeled classes (Figure 4d). Finally, contours corresponding to the MNDWI threshold were mapped onto the grayscale image, enabling the extraction of instantaneous waterlines.

In the study area, 8 rectangular Regions of Interest (ROIs) were defined along the sandy coast. For each ROI, CoastSat was employed to identify and download available satellite images from 1994 to 2023. Using a predefined cloud cover threshold of 50%, approximately 5000 usable images were acquired. To analyze and evaluate the dynamic evolution of the sandy shoreline, transects orthogonal to the coastal baseline were established at 100 m intervals, each extending 500 m. A total of 529 transects were generated and numbered sequentially from south to north. Specifically, Xiaohai includes transects No. 1–104, HZ spans transects No. 105–319, Yudaitan comprises transects No. 320–433, and Boao consists of transects No. 434–529 (Figure 5). After extracting the instantaneous waterlines from the satellite images, CoastSat automatically calculates the intersection points of each two-dimensional waterline with the orthogonal transects, producing time series data on cross-shore distance changes along each transect (Figure 6).

3.2. Tidal Correction and Outlier Correction

Given that satellite images are captured at various times and tidal levels, tidal correction is necessary to eliminate the effects of tidal fluctuations and adjust the time series of instantaneous waterline positions to a common tidal reference datum [51]. The beach foreshore is modeled as a linear inclined plane intersecting the sea surface at an angle β, allowing the satellite-derived waterlines to be projected from their instantaneous tidal elevations to a static reference elevation (Z_ref). The reference elevation in this study is based on the multi-year mean high water level recorded at the Boao tidal station, corresponding to the mean high water springs for the Qionghai–Wanning coast. This projection results in horizontal shoreline displacement (ΔX) along the transects as the instantaneous waterline is adjusted to the mean high water spring tide line, defined as follows:

$$\text{∆}X={\displaystyle \frac{Z_{ref}-Z_{tide}}{tan\beta }}$$

(3)

where Z_tide (m) represents the tidal level corresponding to the time of satellite images, and tanβ is the foreshore slope. The estimated foreshore slope is obtained using the CoastSat extension package, CoastSat.slope [52], which calculates the slope for each defined transect based on the satellite-derived shorelines and their corresponding tidal levels. Once the slope for each transect is determined, the average slope across all transects within the ROI is computed to obtain the static slope tanβ.

Satellite-derived shorelines often contain outliers from various sources [51]. In this study, outlier correction is applied after tidal correction in two main steps. Firstly, the maximum cross-shore distance change observed between consecutive satellite-derived shorelines is capped at 40 m to eliminate large, easily identifiable outliers. Secondly, the Otsu threshold range for delineating instantaneous waterlines is set within the range of [−0.5, 0], and any time series values falling outside this range are removed. This approach helps mitigate errors that may arise during boundary segmentation [47].

3.3. Satellite Data Validation

The quality of the CoastSat-derived shorelines is evaluated by comparing them with in situ measured beach profiles from August 2023 within the study area. Matching transects are selected in CoastSat, allowing for a time tolerance of 15 days between satellite image acquisition and in situ profile measurements. A scatter plot and best-fit linear regression illustrate the relationship between shoreline positions inferred from in situ profiles and those extracted from satellite imagery (Figure 7). The high correlation coefficient (R² = 0.98) and low root mean square error (RMSE = 3.93 m) demonstrate that the satellite-derived shorelines closely align with the actual shoreline positions.

4. Results

4.1. Shoreline Changes

Due to the irregular frequency of satellite-derived shorelines, a high density of extracted shorelines over short periods may introduce potential biases. To address this, cross-shore distances are averaged on a monthly basis to standardize the resolution of the shoreline time series along each transect to a consistent monthly interval. It is noticed that the resolution of satellite may limit the accuracy of captured dynamic shoreline since our research is strongly dependent on the data captured from satellite. However, the time span of this paper is 30 years, which provides a possible solution of using monthly averaged shorelines to compensate the error caused by the limited resolution of satellite data.

The Net Shoreline Movement (NSM) is calculated for each transect as the distance between the cross-shore distances from the oldest and most recent years, providing a measure to characterize long-term shoreline changes. The Theil-Sen estimator [53] is used to determine the rate of shoreline change. Positive values of NSM and the shoreline change rate indicate shoreline advancement seaward (accretion), while negative values denote shoreline retreat landward (erosion).

4.1.1. Long-Term Analysis of Shoreline Changes

The long-term shoreline evolution from 1994 to 2023 for the sandy coast is summarized in

Table 3. Calculated summary of long-term shoreline change rates from 1994 to 2023.

Descriptive Statistics	Xiaohai	HZ	Yudaitan	Boao	Total
Transect-ID	1–104	105–319	320–433	434–529	1–529
Total number of transects	104	215	114	96	529
Length of shoreline (km)	10.4	21.5	11.4	9.6	52.9
Total number of transects where erosion was recorded	69	42	83	43	237
Total number of transects where accretion was recorded	35	173	31	53	292
% of total number of transects where erosion was recorded	66.35	19.53	72.81	44.79	44.8
% of total number of transects where accretion was recorded	33.65	80.47	27.19	55.21	55.2
Mean NSM (m)	−1.97	14.33	−15.61	−5.99	0.99
Maximum NSM (m)	75.87	47.03	39.38	99.11	99.11
Minimum NSM (m)	−20.7	−35.44	−80.06	−134.31	−134.31
Mean shoreline change (m/year)	0	0.55	−0.77	−0.14	0.03
Maximum erosion rate (m/year)	−0.46	−1.33	−2.37	−5.4	−5.4
Maximum accretion rate (m/year)	4.18	1.81	0.63	4.08	4.18
Mean erosion rate (m/year)	−0.23	−0.47	−1.18	−1.16	−0.77
Standard deviation of erosion rate (m/year)	0.12	0.39	0.7	1.33	0.84
Mean accretion rate (m/year)	0.45	0.8	0.32	0.68	0.68
Standard deviation of accretion rate (m/year)	0.77	0.46	0.21	0.98	0.63

. According to the results from the Theil-Sen estimator, the average rate of change in the sandy shoreline over the last 30 years has been 0.03 m/year, with an average NSM of 0.99 m. Among the 529 transects analyzed, 237 (44.8%) experienced erosion, with an average erosion rate of −0.77 m/year, while 292 (55.2%) exhibited accretion, with an average accretion rate of 0.68 m/year. Figure 8 illustrates the distribution of shoreline change rates along the 529 transects from 1994 to 2023, revealing uneven distribution across different subunits. Erosion hotspots are primarily concentrated around the Wanquan River mouth (northern Yudaitan and southern Boao) and the southern side of the artificial islands in central Boao. Accretion is most notable at the northern end of Xiaohai, the southern portion of HZ, and the northern end of Boao (Figure 9).

The long-term shoreline trends for four sections over the past 30 years reveal different distinct patterns. Xiaohai exhibits a dynamic equilibrium with an average NSM of −1.97 m, and most transects show shoreline change rates between −0.5 to 0.5 m/year. Significant accretion occurs at the northern end, where the maximum accretion rate reaches 4.18 m/year. HZ, the longest section, generally experiences accretion with an average rate of 0.55 m/year and an NSM of 14.33 m. Accretion is more pronounced in the southern portion, while the northern section remains in dynamic equilibrium. Only minor erosion is observed in the central portion of this section. Yudaitan shows the most intense erosion among the sections, with an average shoreline change rate of −0.77 m/year and an NSM of −15.61 m. The northern portion of the Yudaitan spit, particularly near the Wanquan River mouth, experiences severe erosion. Boao has an average shoreline change rate of −0.14 m/year and an NSM of −5.99 m. Shoreline change rates are highly uneven in this section. The southern side of the artificial islands experiences severe erosion at −5.40 m/year, while the northern portion shows notable accretion, with a maximum accretion rate of 4.08 m/year and a maximum NSM of 99.11 m.

4.1.2. Interannual Shoreline Variability

Figure 10 highlights the spatiotemporal evolution of cross-shore distance along 529 transects within the study area, revealing shoreline change patterns at seasonal to decadal scales. Distinct evolutionary patterns emerge across different periods and spatial distributions. Xiaohai shows overall dynamic equilibrium from 1994 to 2023, with minimal long-term shoreline shifts. HZ has experienced accretion over the past 30 years, with intensified accretion in the southern section, dynamic equilibrium in the northern section, and increasing erosion in the central portion. Yudaitan displays a significantly stronger erosion trend compared to the other sections. The northern area, especially near the Wanquan River mouth, experiences sustained and severe erosion, while the southern portion remains in balance between erosion and accretion. Boao generally maintains a dynamic equilibrium, except for the northern beach at the Wanquan River mouth, which follows the intense erosion pattern of northern Yudaitan. The northern portion of Boao exhibits complex evolution, with alternating periods of intense erosion and accretion.

The interannual signal of cross-shore shoreline distance from 1994 to 2023 is characterized by alternating erosion and accretion. It reveals a slow but steady seaward migration trend at a rate of 0.06 m/year over the 30-year period (Figure 11). Additionally, this study investigates the interannual shoreline evolution in response to ENSO variability (Figure 11) and identifies distinct patterns. El Niño-dominated years typically exhibit seaward shoreline migration, with the shoreline advancing during these periods. Conversely, during La Niña phases, the shoreline consistently retreats landward, indicating more pronounced erosion trends. Notably, the shoreline demonstrates a stronger response to La Niña events compared to El Niño events.

4.1.3. Seasonal Shoreline Variability

Cross-shore distances along the majority of transects in the study area exhibit significant seasonal oscillations, which are represented as high-frequency vertical stripes in Figure 10. The amplitude of these oscillations varies irregularly across different sections. Specifically, seasonal shoreline changes are characterized by seaward accretion during the summer and landward erosion in the winter. Figure 12 presents the monthly average cross-shore shoreline distance for transect No. 60 in Xiaohai, illustrating this seasonal variation, with values peaking in summer and reaching a minimum in winter. The seasonal erosion–accretion cycle is most pronounced in Xiaohai and southern HZ, while northern Boao displays less pronounced seasonal changes.

4.1.4. Anthropic Interventions

The study area has undergone varying degrees of anthropogenic intervention over the past few decades. Artificial engineering structures are typical examples, primarily distributed along both sides of the Xiaohai Lagoon inlet and the Boao coastline (Table 1). With the development of tourism, numerous offshore artificial islands have been constructed around the eastern coast of Hainan Island over the past decade, including the artificial island named Coral Island in northern Boao (Figure 1c), which began construction in 2011 but was later halted due to ecological concerns. In the shadow zone of Coral Artificial Island. The shoreline of the beaches located on the northern side of the island, represented by transect No. 518, experienced a notable landward retreat in cross-shore distance following the construction of the artificial island, indicating erosion (Figure 13a). By contrast, as shown in Figure 13b, after the construction of Coral Island began in 2011, the shoreline within the artificial island’s shadow zone significantly advanced seaward due to accretion. However, after 2018, the shoreline was artificially reshaped landward due to dredging activities. These dynamics have resulted in a pronounced erosion–accretion anomaly on the northern side of Boao (Figure 8, Figure 9 and Figure 10).

Breakwaters are also typical anthropogenic structures along the study coast. A breakwater was constructed south of the Xiaohai Lagoon inlet in 2013 (Figure 1e). Figure 14 clearly illustrates the impact of the Xiaohai Breakwater on the southern shoreline. Following the completion of the breakwater in 2013, Transect No. 103 exhibited significant accretion, with the cross-shore distance advancing by approximately 75–100 m compared to the pre-construction period.

4.2. Wave Climate Characteristics

4.2.1. Characteristics of Significant Wave Height, Period, and Energy Flux

The wave parameters, including significant wave height (Hs), mean wave period (Tp), and wave energy flux (P), exhibit notable seasonal variations in the study area throughout the 30-year period. Figure 15 presents the monthly averages of Hs, Tp, and P at the E1, E2, and E3 sites, demonstrating consistent distribution patterns. Situated in a low-latitude, tropical monsoon region, seasonal wave variations are predominantly influenced by monsoons. During the winter half-year (October to March), northerly winds prevail, resulting in waves with Hs ranging from 1.22 to 2.26 m, Tp between 6.36 and 7.42 s, and P from 4.65 to 18.5 kW/m. By contrast, the summer half-year (April to September) is dominated by southerly winds, leading to smaller Hs (0.89 to 1.13 m), shorter Tp (5.18 to 6.22 s), and lower P (2.12 to 3.84 kW/m). The highest values of these parameters are recorded in December, significantly exceeding those of other winter months, while the lowest values occur in June and July, with minimal variation between May and September.

Figure 16 illustrates that the multi-year average values of Hs, Tp, and P are highest at E3 (Hs = 1.41 m, Tp = 6.39 s, P = 7.24 kW/m), followed by E2 (Hs = 1.35 m, Tp = 6.31 s, P = 6.50 kW/m), and lowest at E1 (Hs = 1.28 m, Tp = 6.25 s, P = 5.76 kW/m). The average distribution patterns of these wave parameters demonstrate a gradual increase from north to south.

According to Figure 17, the time series for the annual average of these three wave parameters from 1994 to 2023 show a generally slow upward trend. The average increasing rates for Hs, Tp, and P across the three sites are 0.0004 m/year, 0.0006 s/year, and 0.0017 kW/year, respectively. Additionally, the response of wave parameters to ENSO events is analyzed. During El Niño periods, Hs, Tp, and P typically show a downward trend, while these parameters significantly elevated during La Niña periods. Notably, peak values in the time series of wave parameters often coincide with La Niña periods. For instance, during the 2011 La Niña event, the annual average values of Hs and P reached their highest levels in the past 30 years.

4.2.2. Wave Direction Characteristics

The mean wave direction (Hd) in the study area is significantly influenced by prevailing wind patterns and undergoes notable changes with the seasonal shifts of the monsoon. Given the consistent Hd characteristics across the three sites, E2 is selected for detailed analysis of monthly variations (Figure 18). From September to February, the northeast monsoon dominates, leading to waves primarily from the ENE direction, with the highest frequency of ENE storm waves occurring during winter (December to February). March and July–August serve as transition periods between the northeast and southeast monsoons, resulting in shifts in Hd. In March, ENE waves remain dominant, although S-SE waves also begin to appear. During the July–August period, the wave direction gradually transitions counterclockwise from S-SSE to E-ENE. From April to June, the wave direction becomes more dispersed, primarily due to the influences of the southeast and southwest monsoons, leading to waves predominantly from the S and SSE directions.

The long-term Hd in the study area predominantly aligns with the ESE (106.55°). The time series of annual average Hd values over the past 30 years reveals a general clockwise shift at a rate of 0.0669°/year, reflecting a gradual shift from ESE towards a more easterly direction (Figure 19). Hd also responds significantly to ENSO variability, with a pronounced clockwise shift during El Niño periods and a counterclockwise shift towards a more southerly direction during La Niña events (Figure 19).

4.2.3. Extreme Wave Climate Characteristics

In this study, extreme wave forcing is defined as instances where wave energy flux exceeds the 95th percentile (P₉₅), representing the top 5% of wave energy flux events. The annual trend of P₉₅ from 1994 to 2023 shows a gradual increase at a rate of 0.1562 kW/yr (Figure 20). Analysis of wave data from site E2 reveals that the average wave direction during these extreme events (Hd₉₅) typically ranges between E and NE, with ENE as the dominant direction (Figure 21). Over the 30-year period, the average Hd₉₅ shows a clockwise shift at a rate of 0.1843°/yr. The annual time series reveals two significant peaks around 2003 and 2013, with most years’ Hd₉₅ values remaining close to the average (Figure 22). The response of extreme wave conditions to ENSO variability shows a general decrease in P₉₅ during El Niño periods and an increase during La Niña periods (Figure 20). However, no consistent pattern is observed in the response of Hd₉₅ to ENSO events (Figure 22).

5. Discussion

5.1. Impact of ENSO on Wave Climate

ENSO, the dominant mode of interannual climate variability in the Pacific Ocean, influences nearshore wave climates through its teleconnections with coastal ocean–atmosphere processes [54]. Current research on ENSO’s impact on regional wave climates primarily focuses on coastal regions around the Pacific Rim [13,15,16,55]. However, the influence of ENSO on wave climate variability in the South China Sea, particularly in the sea area of east Hainan Island, remains systematically unexplored.

In this study, we used the wave climate data from site E2 as a proxy to represent conditions in the study area. Pearson correlation analysis reveals significant negative correlations between the annual mean MEI and both Hs (r = −0.666, p < 0.01) and P (r = −0.619, p < 0.01) (

Table 4. Pearson correlation analysis between mean annual MEI and wave climate data.

		Hs	Tp	P	Hd	P95	Hd95
MEI	Pearson correlation	−0.666 **	−0.151	−0.619 **	0.501 **	−0.365 *	0.019
	Sig.(2-tailed)	<0.001	0.426	<0.001	0.005	0.047	0.923

** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

). Persistent El Niño periods are generally associated with decreases in Hs, shorter Tp, and lower P values. Conversely, La Niña events typically coincide with higher P intervals (Figure 17). The annual mean Hd exhibits a significant positive correlation with MEI (r = 0.501, p < 0.01), with Hd shifting clockwise toward the east during El Niño periods and shifting closer to the SE direction during prolonged La Niña conditions (Figure 19). Changes in P and Hd during extended La Niña events tend to be more pronounced than those observed during El Niño events (Figure 17 and Figure 19). Additionally, the annual mean P₉₅ value shows a negative correlation with MEI (r = −0.3653, p < 0.05), indicating that extreme wave events occur more frequently and with greater intensity during La Niña years.

The evolution of sandy shorelines, although not directly related to ENSO, exhibits significant and consistent responses within the study area during prolonged warm and cold phases of ENSO over the 30-year analysis period (Figure 11). During El Niño years, sandy beaches generally show a trend in accretion, whereas La Niña events correspond to net erosion of the beaches. In addition to ENSO, other atmospheric teleconnection patterns, such as the North Pacific Oscillation (NPO), may also significantly influence long-term wave climate variability in the study area [56]. Additionally, climate variability through teleconnections can lead to increased sea level anomalies and storm surge events, which impact sandy coastal erosion or accretion. When changes in wave climate coincide with storm surges and sea level anomalies, the resulting impact on coastal erosion can be more severe [57].

5.2. Relationship Between Sandy Shoreline Evolution and Wave Climate Changes

The eastern Hainan Island coast, characterized as a wave-dominated, microtidal environment, experiences patterns of erosion and accretion that are primarily governed by seasonal variability in the wave climate. This variability drives seasonal sediment transport, which, in turn, regulates shoreline evolution. A distinct seasonal pattern emerges in both wave parameters and cross-shore shoreline distances: from July to December, P increases, Hd shifts counterclockwise, and the shoreline retreats landward. Conversely, from January to June, P decreases, Hd shifts from ENE to S, and the shoreline advances seaward (Figure 12 and Figure 13). During summer, short-period, low-energy waves dominate, facilitating onshore sediment transport and leading to beach accretion. By contrast, winter is characterized by high-energy waves that promote offshore sediment transport, resulting in beach erosion. Seasonal variations in Hd result in corresponding reversals in longshore currents and sediment transport directions, significantly influencing the shoreline’s seasonal evolution. The overall orientation of the sandy coast transitions from NNE-SSW in the north to NNW-SSE in the south. Strong ENE waves in winter enhance southward longshore sediment transport, favoring accumulation in the southern beach, while S-directed waves in summer cause sediment redistribution to the northeast.

The interannual variability of the wave climate is characterized by anomalies in various wave parameters (Figure 17). Throughout the study period, significant upward trends are observed in Hs, Tp, P, and P₉₅, with a spatial pattern of increasing values from north to south (Figure 16). Notably, the rise in P₉₅ is more pronounced than that of P, aligning with the global trend in extreme events rising at a faster rate than average conditions [58]. Over the 30-year study period, Hd shows a slow clockwise shift, making it the parameter most closely associated with shoreline evolution. A significant positive correlation is found between the annual average Hd and the cross-shore shoreline distance (r = 0.565, p < 0.01) (

Table 5. Pearson correlation analysis between mean annual cross-shore shoreline distance and wave climate data.

		Hs	Tp	P	Hd	P95	Hd95
Cross-shore distance	Pearson correlation	−0.407 *	−0.026	−0.384 *	0.565 **	0.151	0.180
	Sig.(2-tailed)	0.026	0.893	0.036	0.001	0.427	0.341

** Correlation is significant at the 0.01 level (2-tailed). * Correlation is significant at the 0.05 level (2-tailed).

), indicating that shoreline changes are particularly sensitive to variations in incident wave direction. When Hd shifts counterclockwise towards a more easterly direction, the sandy shoreline in the study area tends to become unstable, typically responding with net erosion. Conversely, a clockwise shift of Hd generally correlates with a trend toward shoreline accretion (Figure 23).

Shoreline evolution is driven by gradients in longshore sediment flux, with oblique wave incidence inducing longshore sediment transport. The resulting longshore currents, generated by wave breaking, serve as the primary mechanism for sediment transport. The stability of longshore sediment transport largely depends on the prevailing wave direction. However, shifts in Hd caused by climate variability result in interannual anomalies in longshore sediment transport, thereby influencing the long-term evolution of sandy shorelines. Additionally, anomalies in storm wave direction can significantly alter longshore currents, affecting the redistribution of beach sediments along the coast. For example, around 2003, a notable shift in Hd₉₅ occurred, with a counterclockwise shift of 12.43° (Figure 22). During this period, extreme wave conditions predominantly from the east contributed to considerable net erosion of the coast (Figure 11).

A negative correlation is observed between the annual mean cross-shore distance and P (r = −0.384, p < 0.05) (Table 5). During years with positive anomalies in P, the coast generally experiences erosion, while negative P anomalies are typically associated with positive shoreline positions, indicating accretion. The correlation between P₉₅ and the annual mean cross-shore distance is relatively weak (Table 5), as beach responses to interannual extreme events and recovery are influenced by underlying seasonal patterns. Sandy beaches may endure severe erosion during consecutive storm events but tend to recover under normal wave conditions.

The response of sandy shorelines to wave climate is also influenced by coastal geomorphological features. The beach in the northern portion of HZ is positioned within the sheltered area of the Zhengmenling Headland (Figure 1b). Due to the protective effect of the headland, this region experiences reduced wave energy impact, resulting in smaller and more stable nearshore waves. Consequently, the long-term shoreline evolution in this area remains relatively stable, with a less-pronounced seasonal pattern of shoreline change (Figure 9 and Figure 10). By contrast, the central portion of HZ is situated in an open area, directly exposed to stronger wave energy, leading to a more prominent seasonal erosion–accretion pattern on the beach and a gradual trend in erosion under increasingly intense wave climate conditions. Deep-water wave climate data are used in this study to characterize nearshore wave conditions. However, as deep-water waves approach the shore, they tend to weaken and redistribute, a process largely influenced by nearshore geological features. The wave energy of incoming waves may undergo significant changes before reaching the beach, affecting their spatial distribution patterns. Nevertheless, trends observed in the deep-water wave climate are considered representative of nearshore trends [41,43,59].

5.3. Other Factors Influencing Shoreline Dynamics

In addition to long-term changes in wave climate driven by climatic variability, other factors such as anthropogenic interventions and river sediment supply may significantly influence the dynamics of sandy shorelines. In certain localized areas, these factors can dominate beach evolution, potentially exerting a greater effect on shoreline change than wave climate. Consequently, the shoreline’s long-term response to wave climate variations may be obscured by these influences.

The coastal engineering projects in this study primarily consist of artificial islands and breakwaters (Table 1). The construction of artificial islands typically alters hydrodynamic conditions and sediment transport patterns in the surrounding local sea areas, leading to either beach accretion or erosion [38,60]. In the shadow zone of Coral Island, the sheltering effect of the artificial island reduces hydrodynamic forces, causing sediment deposition and significant beach accretion (Figure 13b). Conversely, north of the shadow zone, where local hydrodynamics are more vigorous, the combined effects of incident wave action and longshore currents contribute to beach erosion and substantial shoreline retreat (Figure 13a). As another kind of typical man-made coastal engineering in the study area, Xiaohai Breakwater disrupted the longshore sediment transport, reducing local hydrodynamic activity in the wave shadow zone to the south, leading to gradual sediment deposition and a significant seaward advance of the shoreline (Figure 8, Figure 9 and Figure 10). Similarly, an artificial dike built in 1972 connected the spit north of the Xiaohai Lagoon inlet with Neizhi Island [61] (Figure 1e), blocking the strong northward longshore sediment transport and resulting in the southern portion of the Hele-Zhengmehai the most prominent accretion zone in the area (Figure 8, Figure 9 and Figure 10).

The areas most affected by riverine sediment supply in the study area are the Yudaitan spit and the northern beach at the Wanquan River mouth. These areas exhibit a pronounced erosion trend (Figure 8, Figure 9 and Figure 10), primarily due to the reduction in fluvial sediment supply. Over the past three decades, dam construction and engineering projects upstream in the Wanquan River basin have led to decreased runoff and weakened hydrodynamic forces. The interception of bed load and the relative reduction in suspended load have diminished sediment transport to the sea, resulting in finer sediment deposition. This substantial decrease in coastal sediment supply has caused accelerated erosion in these regions [36,62].

Notably, coral reefs are distributed along the Boao coast, where they effectively dissipate wave energy and provide significant protection to the sandy shorelines [35,63]. However, over the past 30 years, increased anthropogenic intervention, ecological degradation, and the potential impacts of global warming have led to severe coral reef degradation along the eastern coast of Hainan. Consequently, the wave energy impacting the Boao shoreline has progressively increased, leading to intensified erosion [64].

Overall, the long-term changes in the sandy shorelines result from a combination of climate change, natural processes, and intensified anthropogenic activities. For wave-dominated sandy coasts, regional-scale changes in wave climate are expected to increasingly influence shoreline evolution. With the escalation of anthropogenic activities, coastal development, and the potential effects of rising sea levels, the erosion of sandy shorelines may be further exacerbated. There are still gaps in our understanding of the relative impacts of natural processes, human interventions, and climate change, highlighting the need for further research based on localized observational and modeling data. As widely recommended, relevant authorities should incorporate the effects of climate change into the policy frameworks for shoreline management and coastal infrastructure development.

6. Conclusions

This study investigates shoreline evolution across various time scales and the impact of long-term wave climate changes along the sandy coast of eastern Hainan Island from 1994 to 2023, utilizing publicly available medium-resolution remote sensing imagery and ERA5 reanalysis wave datasets. It also examines the response mechanisms of shoreline dynamics to climate-driven wave climate changes.

Seasonal and interannual variations, as well as long-term trends in the shoreline, were investigated by analyzing 529 transects across four sections: Xiaohai, HZ, Yudai-tan, and Boao. A long-term analysis from 1994 to 2023 revealed that the sandy shoreline experienced an average rate of change of 0.03 m/year. Notably, 44.8% of the transects experienced erosion, while 55.2% recorded long-term accretion. Yudaitan emerged as the primary erosion hotspot, while HZ is a significant accretion zone. By contrast, Boao displayed considerable internal variability between erosion and accretion. The interannual shoreline variability showed alternating patterns of erosion and accretion, with an average interannual accretion rate of 0.06 m/year. Seasonal analysis indicated that most portions of the coast underwent distinct seasonal fluctuations, characterized by accretion during the summer months and erosion in the winter.

Over the past three decades, the average and extreme wave climate mechanisms in the sea area of east Hainan Island have become increasingly active, exhibiting pronounced seasonal patterns in wave parameters. Our results indicate that the ongoing changes in wave climate within the study area are driven by the ENSO climate pattern. The ENSO index shows a positive correlation with significant wave height, wave energy flux, and extreme wave forcing while demonstrating a significant negative correlation with changes in mean wave direction.

The multi-scale behavior of the sandy shoreline exhibits a discernible response to the persistent changes in wave climate triggered by ENSO viability. During El Niño events, beaches typically experience overall accretion, whereas La Niña periods lead to net beach erosion. Along the sandy coast of east Hainan, there is a significant correlation between wave energy flux, mean wave direction, and changes in cross-shore shoreline distance, with mean wave direction exerting the greatest influence on interannual variations in cross-shore distance. Long-term variations in wave climate drive the dynamics of shoreline erosion and accretion by affecting the gradients of cross-shore and longshore sediment transport.

The long-term changes in the sandy shorelines of eastern Hainan Island arise from a combination of climate change, natural processes, and intensified anthropogenic activities. The response of local long-term shoreline changes to wave climate variability is challenging to discern, as it is significantly influenced by human engineering interventions, such as artificial islands and breakwaters, as well as riverine sediment supply.

The findings of this study reveal that the relationship between sandy shoreline dynamics and ENSO is not negligible. It is essential to take this relationship into account when assessing sandy beaches’ vulnerability. Additionally, the findings of this study are directly relevant to coastal risk management policies and should be incorporated into the assessment of climate change impacts on coastlines to develop localized adaptation plans for the coastal area.