DuneFront deliverable D4.1 - Physical boundary conditions over European coasts

DuneFront Deliverable 4.1 December 2024 Physical Boundary Conditions DuneFront – D4.1 Deliverable information Title Physical Boundary Conditions Deliverable number D4.1 WP number 4 Author(s) Bruno Castelle (UBx), Maxime Dahirel (UGent) Lead beneficiary UBx Contributors Bruno Castelle, Maxime Dahirel Type Report Dissemination level Public How to cite Castelle, B., Dahirel, M. (2024). Physical boundary conditions, Version 1.0, DuneFront Project Deliverable 4.1, Université de Bordeaux Copyright license* © Authors and DuneFront consortium, 2024-2027. This report is openly licensed via CC-BY. For the separate license on the data layers associated with the report, see link in Section 6: Data availability. Versioning and contribution history Version Date Authors (Institution) Notes Version 0.1 28/11/2024 UBx, UGent Version to be checked and approved by DuneFront consortium Version 1.0 19/12/2024 UBx, UGent Final version approved by all Beneficiaries Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union. Neither the European Union nor the granting authority can be held responsible for them. Cover page The DuneFront Project is working to improve coastal protection across Europe by using Nature-based Solutions (NbS), such as Dune-Dike hybrids (DD-hybrids), to defend coastlines from extreme weather and rising sea levels. Deliverable 4.1, titled “Physical Boundary Conditions”, focuses on collecting and mapping key physical boundary conditions that affect the effectiveness of these solutions. The aim is to create a consistent, high-quality dataset that helps understand how DD-hybrids are, and will be, affected by waves, tides, weather patterns, and climate change. In order to provide useful information to the species distribution models developed in WP4.2, the study domain extends beyond the European border, covers coastlines from the Mediterranean, Black Sea, Baltic Sea, and European Atlantic, Channel, and North Sea regions, but excludes Macaronesian territories of the European Union (Canary Islands, Azores, Madeira). The dataset includes various physical boundary conditions categorized into three main groups: coastline boundaries (e.g., coastline type, orientation), hydrodynamic boundaries (e.g., wave height, tide range, storm surge), and weather conditions (e.g., temperature, precipitation, wind speed). These variables were sourced from open datasets, including satellite-derived products, numerical simulations, and global models, with detailed descriptions of each source and variable resolution provided. For future projections, the study incorporates both model-informed climate projections (CHELSA database) and linear extrapolation from recent past trends. The projections cover two periods (2041-2070 and 2071-2100) under different climate scenarios and Shared Socio-economic Pathways (SSPs). The dataset also considers interpolation techniques and addresses limitations such as the coarse resolution of wave data and potential inaccuracies in satellite-derived shoreline trends. The study area covers 113,761 km of coastline and includes 252.5-m spaced (average) 390,240 transects. The analysis reveals significant spatial variability, with only 6.6% of the coastline consisting of sandy beaches. The proportion of sandy coasts varies widely by country, with nations like Norway contributing a large portion of the coastline but having a very low percentage of sandy beaches. In contrast, countries such as Belgium, the Netherlands, and Portugal, which have coastlines predominantly made up of sandy beaches, are considerably above the global average. The report also highlights the variability of hydrodynamic conditions, such as wave energy, tides, and storm surges, which are crucial for understanding coastal erosion and dune dynamics. The Atlantic coast experiences high-energy wave conditions with large wave height and long wave period, whereas the Mediterranean and other enclosed seas see much lower wave height and period. The analysis of tidal conditions shows significant spatial differences, with the Severn Estuary and Bay of Mont-Saint-Michel having the highest astronomical tides in Europe. Storm surge is notably higher along the North Sea and other regions exposed to strong winds and having wide, shallow continental shelves. The Severn Estuary, for example, experiences the highest combined tidal and storm surge levels at 11.64 m above mean sea level. Weather conditions across the European coastlines are also assessed, with data from two datasets (CHELSA and ERA5) showing a strong agreement on average air temperatures, precipitation, and cloud cover. The northern latitudes tend to have higher cloud cover and lower temperatures, while the southern Mediterranean coasts are warmer and drier. Precipitation variability is evident, with west-facing coastlines, such as those in Scotland and Norway, experiencing higher rainfall, while regions like the Mediterranean see lower precipitation levels. These patterns reflect the climatic diversity across Europe’s coastlines. The report also highlights the limitations of the data mostly due to the coarse resolution of the models used to further interpolate boundary conditions along the coast, as local weather patterns (like sea breezes) are not fully captured. The document also provides projections for future boundary conditions, using climate models to predict temperature and precipitation changes under different emission scenarios. The analysis shows that Baltic coastal regions especially are likely to experience the strongest warming, while areas such as the UK and Ireland may see less warming. Precipitation changes also vary geographically, with increased rainfall predicted for the northwestern coasts and declines in Mediterranean regions. The study emphasizes the importance of using projected climate data rather than extrapolating recent trends, due to potential mismatches between recent trends and future projections. Despite some limitations discussed in this report, the data provides a reliable overview of the physical boundary conditions affecting the implementation and impact of DD-hybrids in Europe and beyond. Table of Contents 1. Introduction ............................................................................................................................................................................. 7 1.1 Work Package 4 overview .......................................................................................................................................... 7 1.1.1 General objectives ............................................................................................................................................... 7 1.1.2 Work Package sub-tasks, milestones and deliverables .............................................................. 7 1.2 Aims and objectives of D4.1 ..................................................................................................................................... 7 2. Dataset collection and processing .......................................................................................................................... 9 2.1 Dataset overview ............................................................................................................................................................. 9 2.1.1 Study domain .......................................................................................................................................................... 9 2.1.2 Datasets and physical variables ............................................................................................................... 10 2.2 Future projections ......................................................................................................................................................... 14 2.2.1 Climate model-informed projections (CHELSA) ............................................................................ 14 2.2.2 Linear extrapolation from current conditions ..................................................................................15 2.3 Interpolation and limitations ..................................................................................................................................15 3. Physical boundary conditions maps under current conditions .........................................................18 3.1 Coastline layers ...............................................................................................................................................................18 3.2 Hydrodynamic layers ................................................................................................................................................. 19 3.3 Weather layers ................................................................................................................................................................. 21 4. Future projection of boundary conditions ...................................................................................................... 24 4.1 Weather layers: Expected future climate under different climate scenarios (CHELSA data) ................................................................................................................................................................................................ 24 4.2 Weather layers: Comparison between CHELSA predictions and extrapolations based on recent past trends ......................................................................................................................................................... 26 5. Conclusions .......................................................................................................................................................................... 28 6. Data availability .................................................................................................................................................................. 28 7. References ............................................................................................................................................................................. 29 List of abbreviations Abbreviation Explanation WP Work Package DD Dune-Dike DD-Hybrid Dune Dike-Hybrid NbS Nature based Solution SSP Shared Socio-economic Pathway GCM Global Circulation Model CMIP6 6th Coupled Model Intercomparison Project ECMWF European Centre for Medium-Range Weather Forecasts ERA5 ECMWF Reanalysis version 5 C3S Copernicus Climate Change Services CDS Climate Data Store GIS Geographic Information System ISO International Standards Organization CHELSA Climatologies at High resolution for the Earth's Land Surface Areas NA Not Available 1. Introduction 1.1 Work Package 4 overview 1.1.1 General objectives The primary objectives of Work Package (WP) 4 are to quantify and map large-scale physical, biological, and socio-economic boundary conditions that are expected to influence the effectiveness of Dune-Dike-hybrid (DD-hybrid) Nature-based Solutions (NbS). WP4 focuses on gathering and utilizing spatial information to link up morphological changes and environmental conditions (WP6), to support DuneFront’s numerical modeling initiatives (WP7, WP11, WP13), physical experimental campaigns (WP12), and data-driven assessments for upscaling efforts (WP14). This work package also aligns with Task 7.2 in another WP, which aims to collect extreme storm boundary conditions at the demonstrator level, including multiple climate change scenarios, to inform physical modeling efforts. 1.1.2 Work Package sub-tasks, milestones and deliverables WP4 is subdivided into 3 sub-WPs listed below, each associated with a deliverable, with milestone (M4.1) involving the compilation of all boundary conditions for transfer to other WPs: WP4.1 Physical Boundary conditions: with D4.1 (December 2024, this report) consisting in a catalogue/list/database of physical coastal boundary conditions for subsequent numerical/physical investigations. W4.2 Distribution of important species: with D4.2 (March 2025) consisting in catalogue/list/database of the distribution of important species, providing essential data for data driven analysis and subsequent modelling. W4.3 Socio-economic/admin boundaries: with D4.3 (October 2024) consisting in a catalogue/list/database of socio-economic and administrative boundaries for subsequent data driven analysis and upscaling analysis (Lojek et al., 2024). 1.2 Aims and objectives of D4.1 The objective of WP4.1 is to deliver consistent, high-resolution physical data—covering hydrodynamics, weather, shoreline changes, and more—at a European scale along the coast of Europe under current conditions, with extrapolations for future scenarios. To ensure data consistency across the entire European coastline, we prioritized European and global-scale datasets over combining smaller datasets from various sources, which could lead to biased or incomplete insights into European-scale physical boundary conditions. Physical boundary conditions impacting the performance of DD-hybrid NbS can be grouped into three main categories: coastline, hydrodynamics, and weather. These datasets are derived from various sources, primarily satellite-based shoreline data, numerical hindcasts, and projections, and are subsequently interpolated along the entire European coast. Historical trends and projections across different variables provide a foundation for new insights into future physical boundary conditions. This report is structured as follows: Section 2 describes the study area—which extends well beyond the European border—the data sources and parameters for physical boundary conditions, as well as the methods used for interpolation and forecasting future conditions. Section 3 presents maps, close-up views, and initial statistical analyses of selected physical boundary conditions under current conditions, with a similar treatment of future scenarios in Section 4. Finally, conclusions are presented in Section 5, and a persistent DOI link to access all data products in Section 6. 2. Dataset collection and processing 2.1 Dataset overview 2.1.1 Study domain DuneFront focuses primarily on the implementation of DD-hybrid NbS along the European Union coast. However, the species distribution models developed in WP4.2—to identify and map current and future distributions of species of conservation concern or those functionally important for coastal dune development—require physical boundary conditions that extend beyond Europe’s borders. Consequently, the WP4.1 physical boundary conditions coverage has been expanded, as shown in Figure 1. The dataset covers the entire Mediterranean, Black Sea and Baltic Sea coastlines sensu lato, as well as the European Atlantic, Channel and North Sea coasts from Gibraltar up to the approximate latitude of the Arctic circle in Norway (matching the latitudes of the northernmost coast points in mainland EU territory, in the Baltic Sea). By contrast, Macaronesian territories of the European Union (Canary Islands, Azores, Madeira) were not included during this coverage expansion, as they are geographically and biologically distinct from the European mainland’s which is the target of WP4.2 (e.g. Hernández-Cordero et al. 2015). Overall, the dataset covers 113,761 km of coastline, defined by 390,240 transects spaced on average by 251.52 m, with a minimum and maximum spacing at the highest and lowest latitude of the domain of 200 m and 431.8 m, respectively. techniques were used depending on the dataset. These interpolation techniques and limitations are listed in Table 3 below. Additionally, we note that for a few variables, a limited (well below 1%) number of transects in our final dataset contain missing values, due to either blank pixels in raw data that our procedure propagated, or to mismatches between our transect and raw data (for instance, some transects located deep into fjords in our coastline dataset were mistakenly seen as inland, rather than coastal, in some of our data sources with coarser resolution). For these, we encourage readers to impute values at these locations, if needed, as they judge most appropriate for their use case. There is a wider missing values issue in wave (height, period and direction) trends layers however, again tracing back to the raw data source. Here, we recommend users do not impute missing values, but only use the transects with available data. Table 3. Interpolation techniques and limitations Boundary condition category Data source Interpolation Primary limitations Hydrodynamics ERA5 (Hersbach et al., 2023) Nearest (water grid cell) neighbor The data is interpolated from a global wave model with a 0.5° grid resolution, meaning that wave transformations along the coast (such as refraction, sheltering, etc.) are not accounted for. Due to the coarse resolution, the water levels at grid points may vary significantly, which could affect the consistency of the data. Overall, the wave conditions are expected to be broadly representative of coastal wave conditions. However, for embayed and/or sheltered coasts, which account for a large portion of the coastline in this dataset, the wave conditions—particularly wave height—are likely to be inaccurate due to the influence of the dominant wave incidence. Therefore, readers are encouraged to consult regional or local wave models for more accurate wave conditions specific to their site Wave indicator (Muis et al., 2022) Nearest neighbor Interpolation from an already interpolated coastal dataset, which may induce some slight smoothing TPXO Global (TPXO9_atlas_v 5) (Egbert and Erofeeva, 2022) Nearest (water grid cell) neighbor Tidal data may slightly differ locally around estuaries and bay due to tidal model resolution Weather ERA5 (Hersbach et al., 2023) Linear interpolation Given the coarse grid resolution, local weather effects (e.g. sea breeze, some catabatic winds) are not taken into account CHELSA version 2.1, Karger et al (2017, 2019, 2020,2021) Nearest grid cell The finer grid cell resolution is achieved using terrain-based downscaling of ERA data, accounting for wind effects (Karger et al., 2017). The apparent increased local precision in CHELSA is therefore fully dependent on the local performance of this downscaling procedure (see Karger et al (2017) for validation details) Another limitation of the present dataset concerns the satellite-derived products. As acknowledged by Luijendijk et al. (2018) and Castelle et al. (2024), the uncertainties in satellite-derived shoreline trends are approximately ±0.5 m (Luijendijk et al., 2018), with a potential bias towards accretion of +0.2 m/yr (Castelle et al., 2024) and, locally, flawed longterm trends. In line with the recommendations of Castelle et al. (2024), readers are encouraged to apply moving alongshore averaging to smooth out shoreline trend outliers, and to consider large uncertainties when planning to extrapolate past change to future shoreline position. In addition, although the new classification (Hulskamp et al., 2023) of sandy, muddy, and cliff coasts provides a better distinction between sandy and other environments (Castelle et al., 2024), some coastal sectors may still be misclassified. Quality checks conducted along specific sectors of the present dataset suggest that it is limited to a small range of coastal environments—e.g., some sandy intertidal zones are classified as muddy coasts. 3. Physical boundary conditions maps under current conditions In this section, we provide an overview of the dataset. For brevity, only a selected number of layers (out of 21 in total, representing current and past conditions) are mapped, along with some general statistics. For a more detailed examination, readers are encouraged to explore the layers provided in the various files (Table 2) using GIS software, for example. 3.1 Coastline layers The dataset covers 113,761 km of coastline, defined by 390,240 transects. Overall, the coastline studied here exhibits a wide range of coastal environments (see zoom in Figure 3) with significant spatial variability. Notably, only 25,770 km (6.6%) of the coastline is sandy. This low proportion is largely driven by countries like Norway, which contributes a substantial portion of the total coastline (≈18,000 km, even though it is not fully included in the dataset), but has an exceptionally low proportion of sandy beaches (≈2%). In contrast, countries such as Belgium, the Netherlands, Poland, and Portugal have coastlines where more than half consists of sandy beaches—well above the global average of approximately 31% (Luijendijk et al., 2018). Figure 3. Zoom on coastal type showing a large spatial variability in terms of coastal environments (sand, cliff, vegetated, mud, other) Amongst the 25,770 km of sandy coasts, according to the satellite-derived shoreline dataset approximately 13.5 % are eroding at a rate > 0.5 m/year, while a larger proportion (22.8%) is accreting, with a mean satellite-derived sandy shoreline trend of 0.28 m/yr. It is important, however, to remind that this satellite-derived shoreline product may be slightly biased towards erosion (Castelle et al., 2024) and that such trends are associated with substantial uncertainties. 3.2 Hydrodynamic layers Incident wave conditions, astronomical tides, and storm surges are critical factors in beach and coastal dune erosion. The three figures (Figures 4, 5 and 6) below provide insights into the spatial variability of these key variables on a European scale. Figure 4 highlights a clear distinction between the long-wave-dominated Atlantic coast of Europe, which is exposed to high-energy ocean waves generated in the North Atlantic (with average wave periods exceeding 8 seconds), and other regions, particularly along the more sheltered coastal sectors of the Mediterranean Sea, as well as the Adriatic, Black, and Baltic Seas, where the mean wave period is generally under 4 seconds. This difference is also reflected in the variability of time-averaged wave heights (not shown). Regarding wave direction (also not shown), waves consistently approach from the west-northwest along the Atlantic coast of Europe, while the angle of wave incidence shows much greater variability in the other regions. Figure 4. Map of averaged (1940-2023) mean wave period (wave_period in DuneFront_D4.1_means) in seconds. The two following figures show the variability of coastlines in terms of potential water level, with the highest astronomical tide (Figure 5) and the 100-year return period storm surge (Figure 6). The highest tides are located in the Severn Estuary (England/Wales), peaking at 7.87 m according to TPXO, and the Bay of Mont-Saint-Michel in France, peaking at 7.85 m according to TPXO. Tides are very low in the Mediterranean Sea and in the Adriatic, Black, and Baltic Seas, while they are much more variable along the coasts of the Atlantic and North Sea. The storm surges (Figure 6), on the other hand, are typically most significant in areas that are exposed to potentially the strongest winds and are facing a wide, shallow continental shelf, which increases storm surge amplitude. This is why the largest storm surges are typically observed along the coasts of the North Sea, with an estimated 100-year return storm surge peaking at 5.84 m in the Elbe and Jade-Weser estuaries in Germany. Overall, the estimated current maximum potential water level elevation (astronomical tide + 100-year return storm surge) at the study area is 11.64 m above mean sea level in the Severn Estuary (England/Wales). Figure 5. Map of highest astronomical tide level (highest_astro_tide in DuneFront_D4.1_means) in meters ² Figure 6. Map of 100-year return Storm Surge (ss_100yr in DuneFront_D4.1_means) in meters 3.3 Weather layers In contrast to the hydrodynamic layers, some of the weather boundary conditions were available from two datasets (Table 1). Figure 7 compares the average temperatures from CHELSA and ERA, showing a strong overall agreement between the two datasets (R2 = 0.97, RMSE = 0.81°), which supports the reliability of using either dataset. The deviation from the x = y line (with CHELSA giving generally higher values) reflects the different temporal coverage of the two datasets and global warming. Overall, all weather datasets reveal a clear latitudinal gradient. For instance, higher cloud coverage (Figure 8) and lower temperatures (Figure 9) are observed at the northernmost latitudes. Minimum cloud coverage (0.23) is recorded along the southern coast of Cyprus, while the cloudiest coastal region (0.84) is located in western Scotland. The coldest coastal area (274.45 K) is found in Russia near the White Sea, whereas the warmest coastal region (294.82 K) lies along the eastern Nile Delta in Egypt. Greater variability is observed for precipitation (Figure 10). While some latitudinal gradients persist, west-facing coastlines also tend to experience higher precipitation rates. Figure 7. Smoothed density plot of CHELSA average air temperature (CHELSA_tas) and ERA5 average temperature (tmp) across the entire coastline Figure 8. Map of cloud cover (cloud_cover in DuneFront_D4.1_means) Figure 9. Map of average air temperature (CHELSA_tas in DuneFront_D4.1_CHELSA_means) Figure 10. Map of average total annual precipitation (CHELSA_pr in DuneFront_D4.1_CHELSA_means) 4. Future projection of boundary conditions We provide a brief overview of a few of the “future conditions” layers, along with general statistics and comparisons. For brevity, here we only focus on weather layers. As in Section 3, readers are encouraged to use e.g. GIS software to explore the layers in more detail. 4.1 Weather layers: Expected future climate under different climate scenarios (CHELSA data) The correlations between transect-level CHELSA present (DuneFront_D4.1_CHELSA_means) and projected (DuneFront_D4.1_CHELSA_projections) temperature and precipitation values are extremely strong, irrespective of climate scenario and time horizon (Pearson’s r values always > 0.96 and very often > 0.99). As a result, qualitative spatial patterns in future weather conditions are almost identical to those for present conditions shown Figures 8-9, and we do not redescribe them here. Instead, we show in Figures 11 and 12 the differences between present-day (1989-2018) and near-future (2041-2070) temperatures and precipitations, using the intermediate SSP3-RCP7 (“ssp370”) as an example. Figure 11 shows that the Baltic coasts are predicted to experience the strongest warming under that scenario, with expected annual temperatures over 2 Kelvin (2°C) higher than the “present”. By contrast, Atlantic coasts, especially in the UK and Ireland, are expected to see the lowest levels of warming. However, we must note that in this intermediate emission and near-future scenario, there is still no transect that is predicted to experience less than a 0.7 Kelvin (0.7°C) temperature increase compared to the “present”. Figure 12 shows that spatial heterogeneities in precipitation visible in Figure 10 will increase in the near future. Indeed, the coastal regions experiencing the most precipitation in the present day (Scottish and Norwegian coasts especially) are predicted to receive even higher amounts in the near future, with increases of over 0.15 m and up to nearly 0.5 m in some cases. By contrast, many transects in the Mediterranean region, which are already among the driest of our dataset, will see further declines in precipitation rates. Figure 11. Map of the difference between future annual temperature (2041-2070, SSP3-RCP7) and “present day” values (i.e. difference between CHELSA_tas_2041_2070_ssp370 in DuneFront_D4.1_CHELSA_projections and CHELSA_tas in DuneFront_D4.1_CHELSA_means) Figure 12. Map of the difference between future annual precipitation (2041-2070, SSP3-RCP7) and “present day” values (i.e. difference between CHELSA_pr_2041_2070_ssp370 in DuneFront_D4.1_CHELSA_projections and CHELSA_pr in DuneFront_D4.1_CHELSA_means)