The threat posed by climate change (CC) on the coast is taking on an increasingly significant role today and on the agenda of policy makers and coastal managers. Advanced models indicate that by the end of the century, significant changes in climate will occur. In addition to sea level rise (SLR), variations in maritime climate, precipitation patterns, and consequently, river flow and sediment supply, are expected (Fox-Kemper et al., 2021; Ranasinghe et al., 2021; Cooley et al., 2022). However, these changes in sediment dynamics are not solely driven by climate but are also strongly influenced by human activities. Interventions in river catchments – such as dams, water abstraction, and land use changes – as well as coastal infrastructures like harbors, breakwaters, and promenades, can significantly alter sediment supply and disrupt natural coastal processes (Yang et al., 2020). Some impacts of human interventions on coastal morphology have not yet fully manifested. By 2100, these changes driven by human activity will be exacerbated by fluctuating sediment availability and climate variability across all global change scenarios (Caretta et al., 2022). In this context, coastal management decisions should be guided by well-informed coastal evolution plans that account for the scientific advancements made regarding future climate and the modelling of processes, as well as for the legal framework of each country. Despite the research advances and the raising concern of the administration, the transfer of knowledge between the academic and non-academic worlds is not as effective as desirable (Magaña et al., 2020). As a result, the implementation of local, urgent, and poorly informed decisions that disregard the complexity of coastal systems and the effects of changing global climate conditions is further hindered by legal and administrative frameworks that are misaligned with the insights of the scientific community.

In Andalusia (Spain), the service of the regional government’s department responsible for authorizing the use and occupation of the coastal zone has promoted the development of a methodology to study the spatial and temporal variability of the coast in a CC scenario. The methodology has been implemented along the approximately 1,000 km of coast in the region. This initiative, which was launched in 2019, was carried out as part of two projects that were awarded to the University of Granada (UGR) and the Temporary Joint Venture (TJV) Estudio 7 – SandS. The UGR was responsible for developing the general methodology, coordinating the work, and implementing it in the provinces of Granada and Almería. Meanwhile, the TJV implemented it in the provinces of Huelva, Cádiz and Málaga.

This article presents the methodology developed within the framework of the so called ICCOAST project and its application in the coastal provinces of Granada and Almería.

Most of the research advances regarding the modelling of long-term coastal processes at study sites used reduced-complexity evolution models with simple geometries and stationary input conditions. Baquerizo and Losada (2008) highlighted that coastal morphology is the result of the random occurrence of storm and non-storm events and that those deterministic approaches cannot cope with the inherent randomness of the processes. Therefore, any predictive model for the evolution of the littoral zone over years or decades must be based upon statistical tools capable of dealing with the uncertainty of the prognosis. Some works have addressed this problem in the last decades. Vrijling and Meijer (1992) put the focus on this question and used Monte Carlo simulations to illustrate how to assess the probability distribution of the shoreline position for an idealized coast with assumed theoretical distributions of the forcing random variables. Reeve and Spivack (2004) introduced the use of statistical moments to forecast coastline evolution and its variability due to wave climate. In the direction of Vrijling and Meijer (1992); Payo et al. (2004) proposed a general methodology for the assessment of intrinsic uncertainty to be applied for the long-term prediction of the evolution of a certain morphological feature driven by climatologic agents, suitable to be applied, with forecasting purposes, to morphodynamic problems evolving in time and space. In their application to an ideal morphology, they used an analytical one-line solution obtained by Payo et al. (2003) forced with different equally likely wave climate time series realizations. Later approaches included the alongshore variability of sediment transport as well as the impact of a source of sediment due to river discharges (Losada et al., 2008). A probabilistic approach using a Monte Carlo wave climate simulation and a one-line model was performed by Wang and Reeve (2010) to assess long-term beach evolution in the vicinity of detached breakwaters. Ruggiero et al. (2007) also used a one-line model forced with several wave climate scenarios and different sediment supply boundary conditions to obtain probability density functions of shoreline position. Callaghan et al. (2008) analyzed extreme values of beach erosion with estimations made from simulated wave climate through a structural function based on an equilibrium profile concept. In the same line, Davidson et al. (2010) used synthetic times series to analyze the statistical behavior of the shoreline evolution with a beach profile model. Later, Ranasinghe et al. (2012) proposed a model to study long term coastal recession due to the combined action of storms and SLR in a beach profile. Stripling et al. (2017) used synthetic wave and river discharge scenarios, generated via Monte Carlo methods, to conduct a probabilistic analysis of shoreline evolution on a flat beach in the presence of a groyne or a sediment source. Meanwhile, Ding et al. (2018) developed a probabilistic framework based on Monte Carlo simulations that combines a stochastic wave climate generator with a shoreline evolution model to simulate the shoreline response along an idealized coast.

Most of the approaches address intrinsic uncertainty; however, model errors also introduce the so-called epistemic uncertainty (Vitousek et al., 2021). In relation with the later one, Vitousek et al. (2021) used an ensemble Kalman filter model to assess intrinsic uncertainty of coastal changes in a beach profile and to address epistemic uncertainty. Toimil et al. (2021) made long-term multiensemble coastal erosion projections emphasizing the visualization of the uncertainty cascade and the contribution of various uncertainty sources to the overall uncertainty. Thiéblemont et al. (2021) proposed an extra-probabilistic framework for obtaining shoreline projections that consider both intrinsic and epistemic uncertainties.

Some works faced the modelling of large-scale coastal stretches. Among them, the work by van Maanen et al. (2016) who used integrated models that gather different approaches to deal with a diverse coast of about 70 km at the east of England. Vitousek et al. (2017) presented CoSMoS-COAST model that includes longshore and cross-shore transport and studied with it the evolution of about 500 km of coast for a hindcast period. The model was fed with wave climate and SLR and uses historical shoreline information to improve estimates with a Kalman filter approach that parametrizes unresolved processes.

In the last years, efforts are focusing on making projections of CC induced impacts on the coast. One of the pioneering research projects is due to Zacharioudaki and Reeve (2011) who made a statistical analysis of shoreline variation of an ideal beach using a one-line model forced with different 30 years long wave climate projections. Their analysis focused on monthly and seasonal changes departing from a fixed position as the shoreline was set back to its initial shape after each single month or season. Panzeri et al. (2012) presented a GIS-based tool designed to simulate shoreline evolution and assess coastal vulnerability under changing wave climate and sea level rise scenarios. The tool enables probabilistic and scenario-based analyses through the integration of wave propagation models and shoreline response models, supporting comparative evaluations of long-term coastal impacts. Wainwright et al. (2015) analyzed long-term trends of coastline recession due to present day sediment budget and sea-level rise as well as the short-term with the method by Ranasinghe et al. (2012) and studied their combined effect in Narrabeen beach (Australia). More recently, Toimil et al. (2017) proposed a methodology to study recession due to waves, storm surge and SLR with a cross-section-based equilibrium model and applied it to several beaches along the Spanish north coast. Álvarez-Cuesta et al. (2021a) proposed a methodology that uses multimodel projections of climate to calculate ensemble shoreline changes over a 40 km of a coastal stretch using the shoreline evolution model presented in Álvarez-Cuesta et al. (2021b) that accounts for long-shore and cross-shore processes as well as for SLR and unresolved processes with a data assimilation technique as in Vitousek et al. (2017). Vitousek et al. (2021) obtained shoreline projections in Tairua beach (New Zealand) and reduced model and intrinsic uncertainties using an ensemble Kalman filter cross-shore model. d’Anna (2022) followed a similar approach for Truc Vert beach (France) and analyzed the relative impacts of uncertain input variables on shoreline changes. Toimil et al. (2023) focused on the estimation of projections of coastal flooding coupled with coastal erosion along a coastal Mediterranean stretch.

In addition to those studies, there are conceptual works that have highlighted the main challenges to be faced in this line of research and inspired some of the mentioned works. Among them, it is remarkable the work by Ranasinghe (2016) who made a review aimed at facilitating best practice to quantify physical impacts of CC on coasts at local scales (<10 km). In that work, he pointed out that, on sandy beaches, those impacts are driven by SLR as well as wave conditions and river flow and proposed a methodological framework to assess local scale impacts. More recently, Toimil et al. (2020) reviewed the challenges of studying the risks of global changes in coastal zones, highlighting, among others, the multi-impact assessment considering climate change as well as other socio-economic pathways. Splinter and Coco (2021) outlined the present condition of shoreline modelling for periods less than a century and offered insights into future challenges and opportunities. Among other, they highlighted the importance of dealing with non-stationarity and the limitation of the assumption of unlimited sediment supply.

The present work is part of the ICCOAST project which objective was the development of a methodology to make joint projections of erosion and flooding coastal impacts due to CC along the Andalusian coast. In this article and in a companion paper by Otiñar et al. (2025), we present its implementation to the Andalusian coastal provinces of Granada and Almería, located along the Mediterranean coast for the 21^st century (2025 – 2100), under the RCP 8.5 climate scenario. To the authors knowledge, the methodology addresses for the first time some of the challenges highlighted in theoretical works such as, the sediment supply by watercourses, the geological setting, and the non-stationarity of climate. The differences with previous works can be assessed in Supplementary Table S1 of the Supplementary Material (SM) that summarizes the main characteristics of this research and the above-mentioned methodologies. Among them, it is remarkable the way of dealing with the multimodel approach, in which future climate is dealt with a multivariate non-stationary multimodel ensemble characterization (Lira-Loarca et al., 2021) that differs from the approach by Toimil et al. (2021, 2023), Álvarez-Cuesta et al. (2021b) and Vitousek et al. (2021) who follow the methodological approach suggested by Ranasinghe (2016). We jointly consider maritime and hydrological agents for rivers and ephemeral river courses and their sediment supply as well as their effect on sea-levels and, therefore, the extension of flooding. They are treated with Guadalfortran model (Ávila, 2007; Losada et al., 2011) that solves Saint-Venant equations for non-stationary-1D flux. For ephemeral river courses, the catchments’ contribution is estimated with a model that uses the Soil Conservation Service curve number method (SCS, 1956; Kumar and Vijay, 2003) fed with precipitation projections. Regarding the morphodynamic processes, we use a version of CoastalME (Payo et al., 2017), a shoreline reduced complexity evolution model that has been improved in collaboration with the British Geological Survey development team. The implementation of the methodology is illustrated with results from Guainos Beach. In a companion paper by Otiñar et al. (2025), we present the procedure for the statistical analysis of the results, aimed at facilitating informed decision making, and also present the overall results obtained along the analyzed coast.

The paper is organized as follows. Data and materials section include the description of the study site (Section 2.1), the models employed for reproducing the physical processes (Section 2.2) and the data used for the analysis (Section 2.3). The methodology is shown in Section 3, and the results of its application are gathered in Section 4. The Section 5 closes the paper with the discussion and conclusions.

The methodology assumes that, for a given realization of wave climate over a specific time interval, morphological changes can be analyzed as the cumulative effect of a sequence of climate states during which conditions remain approximately stationary (Baquerizo and Losada, 2008; Payo et al., 2008). This assumption enables the decoupling of hydrodynamic and morphodynamic processes, thereby reducing the complexity of the modelling approach (De Vriend et al., 1993). By repeating this procedure multiple times, it is possible to generate different, equally probable realizations and to analyze dynamic random sets related to coastal evolution and flooding occurrence.

For each realization, the methodology begins with an initial configuration of a specific coastal area, defined by bathymetry, terrain characteristics—such as sediment properties and land use—and the presence of infrastructures. Future climate projections derived from multiple combinations of GCM–RCM models are also known. The methodological steps are detailed below and illustrated in Figure 2.

After summarizing the results of the physical environment characterization, we present the results for Subunit 1 of Tract ‘A01’ in Almería province, identified as ‘A01-001’. In the companion paper by Otiñar et al. (2025), we also present overall results obtained along the analyzed coast. Within this subunit is Guainos Beach, a semi-urban beach composed of dark gravel and sand, approximately 700 meters in length and 15 meters in width, with negligible tidal variation O(40 cm). The beach is bounded by a small rocky cliff to the west and a transverse breakwater to the east. It has three additional transverse breakwaters at its easternmost end. The Guainos River creek flows into the sea near the western end. Precipitation within the basin is scarce, with prolonged dry periods during the summer—typical of semiarid regions.

The characterization of the study area resulted in the identification of nine tracts—three in Granada and six in Almería—and a total of 67 subunits within them. Supplementary Table S3 lists these tracts and subunits, including a representative element for each subunit, as well as the characteristics of the computational meshes used in CoastalME. Figure 3 shows the delimitation of the tracts over physical maps of the provinces of Granada (A) and Almería (B), along with the location of wave, river discharge, and precipitation datasets, as well as other elements such as the coastline, river courses, basins, and reservoirs. A total of seven rivers and 30 minor watercourses were considered. The information is zoomed into panel (C), which focuses on subunit A01–001 and shows the initial position of the mean water level line.

We have presented a methodology developed within the ICCOAST project for making joint projections of erosion and flooding due to the combined action of maritime and riverine forcings under the RCP 8.5 climate pathway. To the best of the authors’ knowledge, this is the first work that explicitly considers sediment supply from watercourses, geological setting, the presence of obstacles, and climate non-stationarity. The methodology is applicable to a variety of coastal typologies, including deltaic systems, natural pocket beaches, headland bay beaches, and other coastal landforms developed in the shelter of marine structures. Moreover, as discussed in detail in a companion paper by Otiñar et al. (2025), it allows for the assessment of the intrinsic uncertainty of a collection of random sets that can be tailored to the management requirements of the analysis.

As mentioned in Section 3.2, future climate is addressed through a multivariate, non-stationary, multi-model ensemble characterization (Lira-Loarca et al., 2021), which applies a Bayesian approach to marginal distributions to account for uncertainty regarding the true state of nature. The non-stationary framework, together with the inclusion of riverine forcings, enables the analysis of the spatiotemporal variability of flooding driven by both maritime climate and episodic river discharge events typical of arid and semi-arid Mediterranean coastal stretches. Additionally, this non-stationary treatment of climate forcings allows results to be distinguished across different time scales—such as monthly or touristic seasons—associated with annual climate variability. This seasonal analysis reveals significant differences between peak and off-peak seasons, enabling authorities to make informed decisions regarding authorizations for the use and occupation of the coastal zone. The choice of one year as the reference time unit in the non-stationary climate characterization and simulation is due to the limited length of available climate projections (20 years for wave climate), which prevents analysis at longer climate variability scales (Cobos et al., 2022c; Lira-Loarca et al., 2021), such as the 11-year Schwabe cycle or the 22-year cycle (Usoskin et al., 2004), both linked to solar activity and associated with major climate indices.

As also noted in Section 3.2, due to the unavailability of precipitation projections from the models used for wave and wind, the random variables representing river discharge and precipitation were assumed to be independent of the other variables. This assumption is justified by the semi-arid to arid nature of the river basins in the study area, where fluvial discharges are rare events typically decoupled from wind-wave storms (Losada et al., 2011). However, when this is not the case, a joint analysis is recommended. Given the large spatial scope of the ICCOAST methodology (over 1,000 km of the Andalusian coast), several simplifications have been made. First, the methodology decouples hydrodynamic and morphodynamic processes by keeping the seabed constant during each sea state. This assumption, based on the differing temporal scales of hydrodynamic and morphodynamic responses, is valid for short- to medium-term analyses, but does not capture the non-linear behavior of coastal systems over the long term (De Vriend et al., 1993). Additionally, a simplified morphological model is employed in which cross-shore processes are not considered. This approach assumes equilibrium beach profiles and treats cross-shore sediment transport as a perturbation superimposed on long-term coastal change (Payo et al., 2017). On the Mediterranean Andalusian coast—where the wave climate features nearly opposing dominant directions (Losada et al., 2011), typically approaching obliquely—alongshore sediment transport governs the sediment budget (see, e.g., Fredsoe and Deigaard (1992)). However, cross-shore transport may play a critical role in short-term changes, particularly during storm events and post-storm recovery (Bergillos et al., 2017), and may interact with alongshore transport. Despite the simplifications made, the methodology offers the advantage of being modular, allowing any of its components to be modified or enhanced to incorporate additional processes or to tailor the analysis to specific coastal stretches.

Regarding SLR, the model does not explicitly incorporate the so-called Bruun effect (Bruun, 1954; 1962), unlike other studies such as Álvarez-Cuesta et al. (2021a) and Toimil et al. (2021, 2023), which apply the original Bruun formulation. This formulation has been questioned due to challenges in estimating its parameters and concerns about its applicability beyond its originally intended context. Although alternative formulations have been proposed to address these limitations and extend their applicability (e.g., Dean and Houston, 2016; Dean and Maurmeyer, 2018; Edelman, 1969; Rosati et al., 2013), they are often based on site-specific field data and are not directly transferable to beaches with different characteristics. The approach presented here is more closely aligned with that of Ranasinghe et al. (2012), which uses synthetic climate and sea level data to force a morphological model.

In this study, a single time series corresponding to the ensemble mean of 20 GCMs is used to represent SLR. This contrasts with the approach of Álvarez-Cuesta et al. (2021b) and Toimil et al. (2021, 2023), which employ multiple SLR time series from different percentiles, treating the resulting outcomes as equally probable to reduce uncertainty. An alternative approach—not pursued here due to computational constraints—would be to present results under multiple SLR scenarios, in the style of a scenario-based analysis.

Compound flooding is addressed by integrating maritime and fluvial dynamics using two coupled numerical models. The sea level is set as the downstream boundary condition in the Guadalfortran model that simulates the river’s hydrodynamics, to achieve this coupling. This sea level incorporates contributions from the mean sea level, as well as from astronomical and meteorological tides. Consequently, water levels throughout the river channel are influenced by river discharge and maritime forcing. However, this approach neglects the effect of river discharge on the water levels of adjacent beaches. The level prescribed to the Guadalfortran model at the river mouth is also used as input in the CSHORE model, which simulates maritime hydrodynamics within the CoastalME model.

The resulting inundation domains, one from the fluvial model and one from the coastal evolution model, are represented in raster format and combined spatially to define the total flood extent. This modular coupling strategy allows processes to be integrated in a practical way, while ensuring that computational costs remain manageable.

Several simplifications have been made within this framework. Firstly, the seabed topography within the river channel is assumed to remain constant throughout the simulation. Secondly, the sea level is kept fixed for each three-hour sea state, in line with the wave climate input. While this assumption is suitable for capturing general patterns, it may overlook short-term nonlinear interactions between river discharge and maritime forcing. These interactions can be associated with rapidly evolving storm conditions or the interplay between tides, surges and waves, which operate at finer temporal scales. These limitations are well documented in the literature on compound flooding, where a lack of finer temporal resolution can lead to the underestimation of the impact of critical peaks (e.g. Moftakhari et al., 2017; Bilskie and Hagen, 2018). For this reason, this approximation may only be reasonably valid for small rivers with low discharge and microtidal environments, where the influence of riverine flows on sea level is minimal and the effect of waves within the river mouth is negligible.

The analysis of the variation in mean shoreline position is illustrated with results from Guainos Beach, where a net loss of beach area is observed on average. Superimposed on this general trend is a clear seasonal variation, attributed to the non-stationary treatment of climate. This seasonal variation is more pronounced during the first 15 years; thereafter, the influence of wave climate conditions gradually diminishes, and the trend appears to be primarily driven by SLR. Although this behavior could be partly attributed to the use of constant wave climate conditions derived from the 2026–2045 period (due to data gaps in wave projections), the lack of significant differences in wave climate compared to the later available period (2081–2100) suggests a progressive increase in the relative importance of SLR. Results regarding the compound flooding shows a strong variability of probability of flooding across months particularly at the final stretch of the river and its mouth.

Although there are no specific studies on the historical evolution or future projections of the variables analyzed in this article for Guainos beach, the companion paper (Otiñar et al., 2025) presents results based on multiple random sets for this beach and the wider coastal areas of Granada and Málaga. The future projections in that analysis are consistent with previously documented trends along the Andalusian coast, particularly regarding the spatial distribution of the most critical sectors and the expected intensity of impacts. Specifically, the identification of areas prone to erosion and flooding, as well as the magnitude of projected shoreline retreat, aligns with historical patterns observed by Prieto Campos and Ojeda Zújar (2024) in the provinces of Granada and Almería, based on shoreline evolution from 2001 to 2019. The agreement between these historical trends and the future projections, despite the different time periods and methods involved, supports the robustness of the modelling approach and strengthens confidence in the projections presented for the 21st century (2025–2100).