Introduction

Despite global efforts to protect vulnerable coastlines from erosion, the challenge of coastal erosion persists. Coastal regions are among the most economically, ecologically, and socially valuable environments on Earth, supporting biodiversity, tourism, fisheries, and human settlements. Monitoring and managing dynamic coastal environments also supports global sustainability goals, as recent study emphasize the role of environmental data in guiding climate and urban resilience efforts1. Nevertheless, these dynamic landscapes are increasingly threatened by coastal erosion, which affects 24% of global sandy beaches, with retreat rates exceeding 0.5 m per year2,3,4,5, putting communities and ecosystems at risk6. Rising sea levels, extreme weather events, and human interventions such as dam constructions have further exacerbated shoreline retreat7, making long-term monitoring and mitigation strategies essential. Ghana is no exception, with several communities (Fuveme, Anyanui, Dzita, etc.) facing heightened erosion8 driven by sea-level rise and human activities (sand mining, coastal engineering)9 leading to coastline retreat. Ghana’s coastline is roughly 550 km long and approximately half is at risk of erosion10,11. Based on geomorphologic traits, the coastline is divided into three main zones—eastern, central, and western—each exhibiting varying levels of vulnerability to erosion12. The western coast, which covers about 95 km of the total shoreline, consists of flat and wide beaches backed by coastal lagoons. It extends from the Ghana-Ivory Coast border to the mouth of the Ankobra River. The central coast is enclosed with headlands, sand bars, and sand spits that surround coastal lagoons. It is defined by sandy shorelines that are usually sandwiched between rocky headlands and capes. This coastline extends 296 km from Prampram to the southernmost point of Ghana. The eastern coast, which is about 149 km stretches from the Laloi lagoon west of Prampram to Aflao (the border with Togo) in the east. Unlike the western and central coasts, the eastern coast is the most vulnerable, subject to various natural and anthropogenic disturbances, which have resulted in notable shoreline changes over the years as highlighted by 13 and 14. This region has a sandy coastline with strong waves, where winds affect the direction of the waves, creating a high-energy environment in the surf zone12.

One particularly vulnerable area on the eastern coast is the Keta Municipality15. This area has gained attention as a noteworthy case study due to its history of severe erosion impacts. The region has experienced persistent erosion for over a century, with historical rates reaching approximately four (4) meters per year before the construction of the Keta Sea Defence Project16. Coastal erosion in this region has destroyed homes, forcing numerous families to seek temporary accommodation, such as tents on the beach, due to the loss of their houses17. In November 2021, a severe tidal wave displaced about 3,000 houses in the Keta municipality, with Abutiakope, Kedzikope, and Keta Central being the most affected18. In spite of this, the Keta Municipality’s coastal urbanization has been growing due to commercial and economic activities19. Nevertheless, this growth puts additional pressure on the region’s already scarce land area, increasing its susceptibility to the negative consequences of future erosion20. To address these issues, the Keta Sea Defense Project was started in 2000 21. However, the region still suffers from severe erosion, particularly in areas close to defense.

Erosion rates along Ghana’s eastern coast have been evaluated in previous research12,13,22,23, which have shed light on the coastal dynamics of the area. But these investigations were frequently limited by the drawbacks of conventional techniques, such as manual digitization of historical maps24 and field surveys12, which are labor-intensive and prone to human error. Because of this, earlier studies frequently lacked the capacity to thoroughly track changes in the shoreline at the necessary scale and frequency. To get over these restrictions, remote sensing and GIS developments have become essential instruments for precise, efficient, and extensive coastal monitoring. These technologies enable the extraction of shoreline positions from satellite imagery over long time periods, allowing for quantitative analysis of erosion and accretion trends. Automated methods, like CoastSat, address the challenges of conventional shoreline monitoring by providing consistent, objective, and scalable shoreline detection. CoastSat has become a popular alternative to conventional and other neural network-based techniques for coastline surveillance due to its easy-to-use interface and smooth integration with the Google Earth Engine cloud platform. The positional accuracy of this sophisticated technique has been evaluated In a number of locations throughout the globe, including southwest France (RMSE of 12.8 m) and northeastern Australia (RMSE of 8.2 m)25,26, . Furthermore, the use of subpixel extraction techniques such as CoastSat has grown to resolve coastline position changes at a finer scale.

The positional accuracy of this advanced technique has been assessed in various parts of the world including northeastern Australia (with a mean RMSE of 8.2 m) and southwest France (RMSE of 12.8 m)25,26. Moreover, to resolve changes in shoreline position at a finer scale, subpixel extraction methods like CoastSat have been increasingly adopted. According to studies by25,27,28 these techniques greatly improve shoreline detection accuracy by enabling the identification of changes smaller than a single pixel.

These advancements make automated shoreline monitoring methods well-suited for dynamic coastlines, such as the eastern shore of Ghana, where accurate, regular observation is necessary due to swift and subtle coastal changes. To address this need, this study aims to automatically extract historical shoreline positions from Landsat satellite imagery using CoastSat and quantify shoreline change rates (erosion and accretion) over a 40-year period (1986–2024) using DSAS. The adoption of this automated technique is a direct response to the need for more efficient, scalable, and accurate shoreline monitoring. In contrast to traditional approaches that are laborious and frequently prone to errors, the capacity of CoastSat to process shorelines in batches guarantees increased efficiency and precise, long-term tracking of shoreline movement. By offering timely, data-driven insights that guide adaptive coastal management methods along Ghana’s eastern coast, this capacity is especially useful for creating early warning systems and facilitating proactive responses to concerns of coastal erosion.

Study area

The coastline of Ghana is located along the Gulf of Guinea in West Africa. The coastline extends across four administrative regions of Ghana: Volta Region, Greater Accra Region, Central Region, and Western Region (as shown in Fig. 1). The coastline makes up about 7% of Ghana’s total land area and is home to approximately 25% of the country’s population22,23. It is economically significant, with over 80% of the nation’s businesses and industries concentrated along the coast, making it a central hub for commercial, industrial, and residential activities29,30.

Fig. 1
figure 1

Study Area Map of Ghana’s Coastline Highlighting the Eastern Coast.

Full size image

This study focuses on the eastern coast of Ghana located approximately between latitudes 5°45′N and 6°10′N and longitudes 0°45′E and 1°10′E. The eastern coastline of Ghana extends across the Volta Region and Greater Accra Region, featuring significant physical features such as the Volta River Estuary, which influences sediment transport, and the Keta Lagoon, a major brackish water body with important ecological and hydrodynamic interactions. The area experiences a Tropical Savanna climate (Aw), with an average annual temperature of 28 °C, ranging from 23 °C in August and 32 °C in March. Due to annual movement in the inter-tropical convergence zone and related changes in wind directions, rainfall distributions show a bimodal pattern, with a main rainy season from April to July and a secondary season from September to November. Annual precipitation averages 740–1000 mm, with June being the wettest month and January the driest31. Humidity varies between 60% in the dry season and 75% in the wet season. The coastal region is also influenced by waves generated by the local monsoon and storms originating from the Atlantic Ocean, with average tidal wave heights of 1.4 m, reaching peak values of 1.9 m and 2 m. These processes contribute to longshore drift and sediment transport, key factors influencing shoreline changes in the study area. The primary occupation of the people living in this area is artisanal fishing, which greatly enhances their standard of living and the local food supply32,33. Another important economic activity in the region is farming and agriculture, especially small-scale irrigation for vegetable growing like the production of shallots34,35.

Geologically, the area is made up of Holocene and Pleistocene deposits in the Quaternary period, which are mainly formed of unconsolidated sands, clays, and gravels36, therefore it tends to erode faster than consolidated coastal areas37. The elevation of this region is approximately 4.48 m (14.7 feet) above sea level, with geomorphological features such as sandy beaches, and lagoons contributing to its diverse landscape and influencing shoreline dynamics.

Map created using ArcGIS Desktop version 10.8 (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview).

Materials and methods

The study utilized satellite imagery acquired from Landsat missions (Landsat 5, 7, 8, and 9), covering the period from 1986 to 2024 (shown in Table 1). The satellite data retrieved from the USGS earth explorer platform at Level 2 (L2) processing using the Google Earth Engine which has a spatial resolution of 30 m for multispectral bands and 15 m for the panchromatic band (available for Landsat 7 and 8), with a temporal resolution of 16 days. Landsat was chosen due to its long historical archive (dating back to 1972), making it ideal for long-term shoreline change analysis. Unlike Sentinel-2, which began operations in 2015, Landsat provides a multi-decade dataset, allowing for a more comprehensive assessment of coastal evolution.

Based on the geomorphological features and management practices of the region, the study area was divided into three (3) zones (shown in Fig. 2). Zone A represents a protected beach area, characterized by coastal defenses such as sea walls and groynes. Zone B, on the other hand, is primarily made up of unprotected coastline segments where shoreline dynamics are dominated by natural processes. Zone C is located close to an estuary, where sedimentation is greatly influenced by interactions between riverine and marine processes.

Fig. 2
figure 2

Location of the various zones along the coastal stretch of Ghana. Map created using ArcGIS Desktop version 10.8 (https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview).

Full size image

Shoreline extraction

The CoastSat Toolkit was used to extract shorelines automatically25, using open-access satellite imagery from the Google Earth Engine platform. CoastSat is an open-source Python toolbox created to digitize a time series of coastline positions along sandy beaches. For the analysis of long-term shoreline changes along Ghana’s eastern coast, the toolkit is ideal since it enables a quick and reliable coastline extraction method over lengthy periods of time.

  1. 1.

    Image Acquisition and Cloud Masking: The CoastSat Toolkit uses a variety of pre-processing techniques to ensure the accuracy and reliability of shorelines extracted from satellite photos. The method sets out with cloud masking, using a pre-computed per-pixel cloud mask from data sources like the European Space Agency (ESA) for Sentinel-2 and the United States Geological Survey (USGS) for Landsat. This makes it possible to calculate the proportion of cloud cover in a picture. To guarantee that only cloud-free data is processed further, images that beyond a user-defined cloud cover threshold are eliminated. A stricter threshold (e.g., 10%) might unduly restrict the dataset and reduce temporal coverage38;39, so the cloud cover threshold in this study was set at 20% to allow for the inclusion of a sufficient number of satellite images while ensuring that cloud cover does not significantly obscure the region of interest.

  2. 2.

    Pre-Processing for Image Enhancement: To enhance the spatial resolution of the images for optimal shoreline detection, CoastSat employs specific techniques based on the satellite sensor. For Landsat 7 and Landsat 8, the high-resolution panchromatic band (15 m) is used to improve the resolution of multispectral bands (30 m). This is achieved through a data fusion method that employs principal component analysis, enabling a seamless enhancement of the spatial resolution. Landsat 5 images, lacking a panchromatic band, are down-sampled to 15 m resolution using bilinear interpolation, which has been shown to improve shoreline detection. These preprocessing steps ensure high-quality input for the subsequent shoreline extraction process. The images are already orthorectified by the data providers, and their georeferencing accuracy is preserved in the output files containing the extracted shorelines.

  3. 3.

    Image Classification after Pre-Processing: The shoreline detection process in CoastSat relies on a robust sub-pixel resolution algorithm designed to accurately delineate the boundary between water and sand. The shoreline is defined as the instantaneous sand-water interface at the moment of image acquisition. The first step in this process involves image classification, where a pre-trained Multilayer Perceptron Neural Network classifier assigns each pixel in the image to one of four classes: sand, water, white-water, and other land features. This classifier uses a set of 20 features, including pixel intensities from five spectral bands (Red, Green, Blue, NIR, and SWIR1), commonly used spectral indices (e.g., water and vegetation indices), and the variance of these features in a 3 × 3 pixel window. Sand pixels, which have low variance, can be distinguished from other bright things, such rooftops, which usually have more variance, due to this variance. The classifier is a dependable tool for image classification in a variety of coastal situations because it was trained on a global dataset of manually labeled pixels and showed 99% accuracy during validation.

  4. 4.

    Water-Sand Differentiation using MNDWI: After classification, sand and water pixels are distinguished using the Modified Normalized Difference Water Index (MNDWI). The green and SWIR1 bands are used to calculate this index, and Otsu’s algorithm—which maximizes inter-class variance—is used to find the ideal threshold (Eq. 1) for differentiating sand from water.

  5. 5.

    Shoreline Delineation at Sub-Pixel Resolution: Pixels categorized as “white-water” or other land features are not included in the thresholding procedure in order to reduce errors brought on by wave-induced foam or other ambiguous elements. The Marching Squares technique, which creates an exact contour of the sand-water interface, is then used to outline the shoreline at sub-pixel resolution.

Table 1 Image acquisition details.
Full size table
$$\:MNDWI=\:\frac{SWIR\:1-G}{SWIR\:1+G}$$
(1)

CoastSat offers a reliable and accurate coastline extraction method by fusing sophisticated image classification and sub-pixel resolution detection with robust pre-processing algorithms. Reliable findings for monitoring and analyzing coastal dynamics are ensured by their capacity to manage issues including cloud cover, changing resolutions, and environmental noise. Its value for tracking shoreline changes in areas with complicated coastal dynamics has been further established by the successful demonstration of CoastSat’s applicability in a variety of worldwide coastal environments 38,40,41.

Shoreline change analysis

Following their extraction, the shorelines were loaded into a GIS program (ESRI ArcGIS 10.8) for further investigation (Fig. 3, Fig. 4, and Fig. 5). The shoreline change analysis was conducted using the USGS’s Digital Shoreline Analysis System (DSAS)42. DSAS, developed by the U.S. Geological Survey is a well-validated and widely used GIS-based tool for shoreline change analysis. It automates the generation of transects and calculations of shoreline change rates, making it more efficient than manual methods. DSAS uses shoreline vector data to determine shoreline change rates. This method analyzes temporal and spatial patterns in shoreline position by constructing a hypothetical baseline with transects that are perpendicular43. In this study, a baseline was constructed onshore, parallel to the general shoreline orientation, serving as the reference for transect generation and distance measurements. All spatial data were projected into the Universal Transverse Mercator (UTM) coordinate system to ensure accurate distance and area calculations.

Fig. 3
figure 3

Shoreline Change Metrics (LRR, NSM, EPR) for Zone B. (a) Linear Regression Rate (LRR) map; (b) Linear Regression Rate (LRR) versus Transect ID plot (c) Net Shoreline Movement (NSM) map; (d) Net Shoreline Movement (NSM) versus Transect ID plot; (e) End Point Rate (EPR) map; (f) End Point Rate (EPR) versus Transect ID plot.

Full size image
Fig. 4
figure 4

Shoreline Change Metrics (LRR, NSM, EPR) for Zone A. (a) Linear Regression Rate (LRR) map; (b) Linear Regression Rate (LRR) versus Transect ID plot; (c) Net Shoreline Movement (NSM) map; (d) Net Shoreline Movement (NSM) versus Transect ID plot; (e) End Point Rate (EPR) map; (f) End Point Rate (EPR) versus Transect ID plot.

Full size image
Fig. 5
figure 5

Shoreline Change Metrics (LRR, NSM, EPR) for Zone C. (a) Linear Regression Rate (LRR) map; (b) Linear Regression Rate (LRR) versus Transect ID plot (c) Net Shoreline Movement (NSM) map; (d) Net Shoreline Movement (NSM) versus Transect ID plot; (e) End Point Rate (EPR) map; (f) End Point Rate (EPR) versus Transect ID plot.

Full size image

Transects were generated perpendicular to the baseline at regular intervals of 50 m, with a maximum search distance of 500 m to ensure intersection with all shoreline positions. The transects represent cross-sections for calculating shoreline change metrics. A total of 211 transects were generated for Zone A, 349 for Zone B and 501 transects were generated for Zone C. Furthermore, the Shoreline change rates were computed using the following statistical tools in DSAS, which provide a suite of metrics to quantify shoreline movement.

  1. 1.

    The Net Shoreline Movement (NSM) metric quantifies the total distance of shoreline change between the oldest and most recent shorelines.

  2. 2.

    The End Point Rate (EPR) represents the annualized rate of shoreline change calculated by dividing the total distance of shoreline movement by the time interval between the oldest and most recent shorelines.

  3. 3.

    The Linear Regression Rate (LRR) calculates the rate of change by fitting a least-squares regression line to all available shoreline positions.

The calculated change rates were linked to the transects and visualized in ArcGIS. Transects were symbolized based on erosion (negative change rates) and accretion (positive change rates) trends. The spatial distribution of shoreline change rates was analyzed to identify erosion hotspots and areas of significant accretion. In this study, the confidence interval for the DSAS analysis was set at 90%.

Results

The results of the study, which examines shoreline dynamics along Ghana’s eastern coast with a focus on the temporal and spatial trends in shoreline movement, are presented in this section.

Shoreline dynamics analysis for zone A

The coastline change investigation along Zone A demonstrates significant heterogeneity in erosion and accretion processes, with diverse patterns seen across the northern, middle, and southern parts. These results are represented in the maps and transect plots (Figs. 4a–f), which provide a detailed visual description of these changes.

In the northern segment, the LRR analysis indicates substantial erosion, with rates ranging from − 6.1 to −10.4 m/year, signifying persistent shoreline retreat. This observation is evident in Table 2, where the maximum erosion rate for LRR is recorded as −1.49 m/year (ID: 110), while the corresponding maximum cumulative shoreline loss under NSM is -103.8 m in certain areas such as Vodza and Kedzi. The EPR values reinforce this trend, highlighting a maximum erosion rate of -2.95 m/year (Table 2). As depicted in Figs. 3b and 4a, the red zones signify erosion while accretion is represented in blue. The plot in Fig. 4b predominantly features negative values, demonstrating continuous shoreline retreat across this segment. Such high erosion rates are likely a result of intense wave action and sediment loss, exacerbated by the negative influence of protective measures. The percentage of negative transects (erosional) for Zone A is 35.29%, suggesting that erosion dominates nearly a third of the transects.

The middle segment also exhibits notable erosion, albeit to a slightly lesser extent than the northern portion. Here, the LRR values range between − 4.1 m/year to −0.5 m/year, indicating moderate to severe shoreline retreat. The EPR results confirm this trend, while the NSM reveals cumulative shoreline losses of up to −60 m in some transects. From Fig. 4c, it is clear that the shoreline in this section is marked by a mix of erosion and localized minimal net shoreline change, as reflected by small patches of stability amidst the dominant red zones. The transect plot (Fig. 4d) further reinforces this observation, showing predominantly negative values interspersed with occasional minor peaks. These patterns suggest that the erosion in this segment, though widespread, is less severe compared to the northern segment.

In contrast, the southern segment displays a markedly different trend, with evidence of significant shoreline accretion. The LRR values indicate positive rates of up to + 12 m/year, reflecting substantial land gain. This accretion is further confirmed by the EPR results and NSM values, which highlight cumulative shoreline advances exceeding + 200 m. As shown in Fig. 4e, the blue-highlighted regions signify areas experiencing substantial accretion, likely due to sediment deposition processes and possible coastal protection measures. The corresponding transect plot (Fig. 4f) shows a dominance of positive values, which confirms the shoreline advancement in this segment.

Shoreline dynamics analysis for zone B

The northern part (around Dzelukope) is characterized by positive LRR values (light blue to dark blue) indicating accretion. The southern parts (near Woe and Anloga District) display negative LRR values (orange to red), signifying erosion. LRR values vary from − 1.6 m/year to 5.0 m/year (Fig. 3a and b). Table 2 reveals that the maximum accretion rate is 4.98 m/year (ID: 2), accompanied by cumulative shoreline advances under NSM reaching + 143.49 m (ID: 48). Accretional transects account for 62.46% of the total, indicating a dominant trend of shoreline advancement.

The NSM results (shown in Fig. 3c and d) quantify the total shoreline change over the study period, providing a cumulative perspective. The northern section near Dzelukope displays positive NSM values (blue shades), suggesting shoreline advancement over time. The southern section, particularly around Woe, exhibits negative NSM values (red shades), with values ranging from − 116.4 m to 143.5 m, highlighting areas with shoreline retreat and advance. The total movement corroborates the patterns seen in LRR and EPR results. The EPR results (shown in Fig. 3e and f) reflects the net shoreline movement between the start and end years of the analysis, and the pattern generally aligns with the LRR results for Zone B. The northern coastline near Dzelukope shows positive EPR values (light to dark blue), similar to LRR results. The southern portions of Zone B, especially near Woe, show negative EPR values (red shades). EPR values range from − 3.1 m/year to 4.1 m/year, indicating that the rates of erosion are slightly higher in magnitude compared to accretion. The consistency between LRR and EPR validates the observed patterns of erosion and accretion. The higher erosion rates in the southern parts may be exacerbated by natural processes, lack of protective structures or anthropogenic activities, while the accreting areas likely benefit from localized sediment deposition.

Shoreline dynamics analysis for zone C

In the western segment of Zone C, the shoreline exhibits significant erosion, with the NSM (Fig. 5c) indicating cumulative shoreline losses of up to −272.6 m. The LRR results (Fig. 5a) corroborate this observation, showing erosion rates ranging from − 28.7 m/year to -11.8 m/year, while the EPR analysis (Fig. 5e) confirms these findings with values between − 30.5 m/year to −15.7 m/year. This severe erosion, particularly evident in the areas around Fuveme and the western boundary of Anyanui, is highlighted by dominant red zones on the maps and consistent negative values on the transect plots (Fig. 5b, d and f). The erosion in this section is likely exacerbated by wave energy, sediment loss, and the absence of coastal protection structures.

In the central and eastern segments of Zone C, the shoreline displays a mixed trend of localized accretion and moderate erosion. The NSM results show cumulative shoreline movements ranging from − 30 m to + 70 m, indicating that certain areas experience minimal net shoreline change. The LRR (Fig. 5a) and EPR values (Fig. 5c) reveal more moderate erosion rates, typically ranging between − 2 m/year to -3.5 m/year. This segment, which includes parts of Dzita and Anloga, demonstrates a more variable shoreline change pattern, as seen in the transect plots, where negative values dominate with occasional positive peaks. Table 2 highlights the maximum accretion rate as 9.61 m/year (ID: 87) and cumulative shoreline advances under NSM reaching + 335.47 m, primarily in the eastern portion. Accretional transects dominate at 59.48%, suggesting localized sediment redistribution and stabilization processes.

Table 2 Summarizes the shoreline change metrics (NSM, EPR, and LRR) for three coastal zones (A, B, and C) along Ghana’s Eastern coastline. These metrics are critical for assessing the magnitude and direction of shoreline movement, both in terms of erosion and accretion.

Table 2 Summary of Shoreline Change Metrics.
Full size table

Discussion

The deployment of CoastSat offered a distinct methodological benefit, especially in terms of its capacity to reliably assess historical data and automate coastline extraction. CoastSat handled the large spatial and temporal dataset needed for this investigation with efficiency and repeatability, in contrast to traditional methods that are labor-intensive and susceptible to human bias. Its ability to analyze publicly available satellite imagery further highlights its scalability and suitability for comparable research in other areas with sparse field data.

Shoreline changes along Ghana’s eastern coast varied significantly across Zones A, B, and C due to human and natural influences. These findings highlight key geomorphic and sediment transport processes, offering insights for sustainable coastal management.

Zone A demonstrated mixed patterns of erosion and accretion, with erosion dominating the northeastern segments and accretion prevalent in the south. The significant erosion rates (−10.5 m/year) observed in the Kedzi community (near the Keta Lagoon) can be attributed to wave dynamics and sediment transport disruptions caused by upstream structures as highlighted by various researchers 44,24,45. In contrast, the accretion observed in the southern areas (+ 10.8 m/year) around the Emancipation Beach in Keta reflects the effectiveness of the Keta Sea Defense Project in trapping sediments and stabilizing the shoreline locally. Analysis by44 of the period preceding the KSDP (1986–2001) indicates that erosion was the predominant process along the shoreline, affecting approximately 70% of the area recording erosion rates ranging from 0.1 to 15.4 m/year. In contrast, the period from 2001 to 2011, following the implementation of the KSDP, exhibits a significant shift, with accretion becoming more dominant, accounting for approximately 80% of the coastline while erosion rates between Keta and Blekusu remained high, with some sections experiencing rates of up to 9 m/year. However, this intervention also highlights the potential for downdrift erosion, as sediment supply to adjacent areas is interrupted. These erosion and accretion rates are consistent with the values presented in Table 2.

In Zone B, the shoreline exhibited significant variability in erosion and accretion trends (Table 2) despite the absence of coastal protection structures along the while stretch. Localized accretion was observed in some areas, with rates of up to + 4.01 m/year, likely influenced by natural sediment deposition processes or geomorphic characteristics. However, erosion dominated much of the zone, with rates reaching − 3.05 m/year in more dynamic sections. The average NSM value of 20.42 m (Table 2) reflects relatively moderate cumulative changes compared to Zones A and C. Notably, the erosion and accretion rates in Zone B are less critical than those observed in Zones A and C, where extreme shoreline retreat and advance present more significant challenges for coastal management. This highlights Zone B as an area where natural processes remain relatively balanced, though continued monitoring is essential to identify potential future risks.

Zone C displayed the most pronounced spatial variability, with significant erosion dominating the western portions and accretion observed in the eastern sections near the Volta River estuary. Erosion rates reached − 30.49 m/year, while accretion peaked at + 8.79 m/year, as indicated by EPR metrics (Table 2). The high erosion rates in the west can be attributed to both natural (wave action, sediment transport) and anthropogenic factors (costal engineering, sand mining). A key factor contributing to the dominance of erosion in Zone C is the construction of the Akosombo and Kpong dams in the early 1960s. There are sediment shortages throughout the downdrift coast, especially close to the Dzita beach, as a result of these dams’ substantial reduction in the Volta River’s sediment discharge46. According to47, erosion has accelerated as a result of this sediment deficit. Additionally, the Ada Sea Defense structures at the mouth of the Volta River play a significant role in sediment dynamics. While these structures protect localized areas, they restrict sediment transport downdrift, exacerbating erosion in adjacent areas such as Fuvemeh and Anyanui beach, which has been reported to erode at a faster rate48.

In addition to coastal defense structures and upstream dams, other anthropogenic activities such as sand mining and gravel extraction significantly impact shoreline dynamics along Ghana’s eastern coastline. Unregulated sand and gravel mining for construction depletes beach and river sediment supplies, exacerbating shoreline retreat. Field observations from this study confirm the presence of gravel mining activities at Dzita (Fig. 6), an area that has experienced significant erosion.

Fig. 6
figure 6

Small-scale graving mining observed at Dzita46.

Full size image

The observed shoreline changes can also be linked to coastal hydrodynamics, particularly wave energy distribution, tidal currents, and sediment transport dynamics. Along Ghana’s eastern coast, longshore drift is a dominant factor influencing sediment movement, with a net transport direction from west to east49. This aligns with the erosion observed in updrift areas and accretion in downdrift regions (Zone A), particularly near estuaries like that in Zone C50,51 .

Furthermore, the presence of wave refraction near estuarine inlets, such as the Volta River Estuary, alters sediment deposition patterns52,53, contributing to localized shoreline variability54. Seasonal variations in wave energy and storm events also intensify erosion in exposed coastal sections. Studies indicate that wave climate across West Africa influences beach responses and shoreline changes at inter-annual timescales54,55.

The results are consistent with past studies showing how Ghana’s coastline dynamics are influenced by both natural processes and coastal defenses. For example, the erosion rates in Zone A corroborate previous studies by 24 13, which highlighted the vulnerability of the northeastern coastline to sediment depletion. Similarly, the dynamics observed in Zone C supports findings by56 which highlight alarming erosion rates, with reports showing shoreline retreat of nearly ten meters per year near the estuary. Zone B, on the other hand, is unprotected but seems to be stable for the time being, which may account for the lack of study interest in this area. Nonetheless, ongoing observation is necessary to spot any new indications of instability, especially in regions that are prone to erosion.

Shoreline change patterns observed along the eastern coast of Ghana reflect broader global trends in coastal erosion and accretion. The shoreline change patterns observed along Ghana’s eastern coastline are consistent with global trends in coastal erosion and accretion. Similar to Ghana, regions such as the Nile Delta and Mississippi Delta have experienced significant shoreline retreat due to sediment retention from damming, sea level rise, and anthropogenic influences57. The Nile Delta has suffered extensive coastal retreat following the construction of the Aswan High Dam, which has drastically reduced sediment deposition58,59. The dam has trapped approximately 98% of the Nile’s sediment load, resulting in severe coastal erosion rates reaching up to 50 m per year in some locations.

From a methodological standpoint, the findings of this study align with previous research that employs remote sensing and GIS-based shoreline monitoring techniques. Studies such as60,61 have demonstrated the effectiveness of satellite imagery (Landsat and Sentinel-2) with DSAS in quantifying shoreline dynamics61. applied MNDWI-based classification for shoreline delineation in Hersek Lagoon, while60 successfully tracked shoreline modifications along Turkey’s Konyaaltı coastline using multi-source satellite imagery. The present study builds upon these approaches by applying CoastSat and DSAS to assess shoreline change along Ghana’s eastern coast over a 40-year period, reinforcing the utility of automated remote sensing methods for long-term coastal monitoring.

Beyond methodological advancements, this study also offers a broader spatial and temporal perspective on shoreline changes in Ghana. While previous research has shed light on localized shoreline dynamics along Ghana’s coast13,14,19,62, this study builds on their findings by examining a much larger area, spanning Zones A, B, and C (roughly 50 km). The addition of 1,275 transects spanning a variety of human and geomorphic circumstances made it feasible to comprehend the geographical heterogeneity in accretion and erosion patterns in more detail. This larger scope provides a macro-scale perspective that is essential for regional coastal management initiatives, in contrast to research that concentrates on particular hotspots. However, while this study provides significant insights into long-term shoreline evolution, certain limitations, such as the lack of local tide gauge data and seasonal variations, must be considered when interpreting the results.

Conclusion

By combining an open-source Python-based framework for automated shoreline extraction with the Digital Shoreline Analysis System (DSAS) to measure shoreline change rates, this study presents an innovative and efficient shoreline monitoring methodology. The use of CoastSat transforms the procedure by making it possible to observe shoreline dynamics on a wide scale over an extended period. This methodology handles significant advantages over traditional techniques, including automated shoreline extraction, which reduces manual effort, and the ability to handle high temporal and spatial variability—challenges that have hindered previous shoreline studies.

The results highlighted erosion dynamics in Zone A, particularly in the northeastern sections, with accretion observed near protective structures in the south. Zone B, though unprotected, exhibited moderate cumulative changes, suggesting relative stability compared to the other zones. In contrast, Zone C experienced the most severe erosion, exacerbated by sediment retention from the Akosombo and Kpong dams and sediment trapping by the Ada Sea Defense structures. Localized accretion near the Volta estuary confirmed the influence of sediment deposition processes, although this exacerbates downdrift sediment starvation.

High-resolution datasets and long-term monitoring, facilitated by CoastSat, address gaps in coastal management. This study reinforces the need for seasonal monitoring and highlights how hard engineering structures can intensify erosion.

Given the local variations in shoreline dynamics revealed by the study, it is recommended that coastal managers consider the use of UAVs (Unmanned Aerial Vehicles) as a next tool for high-resolution, real-time monitoring. To provide timely and more precise shoreline assessments, UAVs can complement existing remote sensing methods by addressing gaps in data resolution. This technology enhances decision-making and supports the development of adaptive coastal management strategies by providing high-resolution, localized data that can track coastal changes more precisely. The UAV-based approach demonstrated for Dzita beach46 can be replicated on other beaches along the Gulf of Guinea to support regional sediment management and shoreline monitoring.

Additionally, sustainable coastal management strategies should incorporate adaptive planning, nature-based solutions such as beach nourishment, and restrictions on high-risk coastal development. In areas experiencing severe erosion, a combination of soft engineering solutions and strategic land-use planning can help mitigate long-term shoreline retreat and protect coastal infrastructure.

Future research could integrate hydrodynamic modelling approaches such as Delft3D or XBeach to simulate wave dynamics, sediment transport, and shoreline response to natural and anthropogenic influences. This would complement remote sensing-based shoreline analysis and improve predictive capabilities for coastal management.