1. Introduction
Maritime transport is essential to the world economy as over 90% of the world’s trade is carried by sea, being the most cost-effective way to move bulk goods and materials worldwide [
1]. Transportation relies on ports to supply services to freight and ships, which cover a range of functions that involve port operability. Demand for a port is strongly correlated to climatic conditions, and it can also be affected by the vulnerability of its facilities to disruption from extreme weather events, affecting the port’s reliability and operability, which may lead to a shift to alternative ports [
2]. Directive 2009/18/EC of the European Parliament and the Council recognises the importance of reducing the risk of future marine casualties caused by heavy weather damage. Recent hazardous weather created favourable conditions for floods in coastal areas, testing the resilience of the present port infrastructures, where climate change can diminish their ability to cope with future challenging weather and environmental conditions.
Climate change is underway and will likely intensify over the following decades, affecting extended lifetime infrastructures, such as ports designed to withstand extreme weather events that occur at least every 50 years [
3]. In addition, by virtue of the port locations in marine environments (estuaries, lagoons, rias and coastal zones) they are exposed to a range of climate drivers that determine extreme sea levels, including mean sea levels, astronomical tides, storm surges, waves and river discharge. Climate change causes modifications in the frequency and intensity of extreme sea levels [
4,
5,
6,
7,
8], and the intensification of severe storm surges and changes in wave patterns [
9,
10,
11] that will modify the impacts of extreme weather events [
5,
12,
13,
14,
15,
16,
17] and hydrodynamic circulation in the vicinity of and inside ports [
18,
19,
20,
21,
22]. These impacts will affect maritime traffic, berthing [
23,
24,
25], infrastructure, equipment, port operations [
23] and consequently, will threaten economic growth [
26,
27]. For instance, some ports are located in low-lying areas, being affected by rising sea levels and storm surges, while others are situated in marine areas, subjected to extreme weather events such as storms and waves, which could intensify under climate change. Therefore, improving the knowledge about port structure resilience is crucial to understanding how climate change will affect a given port, through an accurate assessment based on analysis of the port’s particular climatic vulnerabilities.
Coastal flooding is linked to storm surge events that generate an abnormal sea level rise above the predicted astronomical tide. When storm surges and high tides coincide, it can cause extreme flooding in these regions [
28,
29]. Climate change can amplify the impacts of these events due to the predicted mean sea level rise in the upcoming years [
5,
30,
31,
32,
33]. By following this line of reasoning, the extreme sea level (ESL) has been used in several studies [
19,
28,
29,
30,
34,
35,
36,
37,
38] to assess coastal flooding and its impacts on structures and land use. The flood hazard is correlated to the intensity and periodicity of the drivers (relative sea level, storm surge, riverine flows, and wave-induced sea level), showing higher risk when the combination of several drivers reaches their highest values. The flood hazard assessment is commonly estimated by the probability that a flood of a particular intensity occurs over an extended period, and this relationship between the probability and intensity gives rise to the concept of a return period, which represents the probability of a flood intensity being exceeded in a given year.
The northwest Iberian Peninsula (NWIP) coast is crossed by one of the most important and busiest shipping lanes in the Atlantic Ocean. Due to the intensive maritime traffic, this area has vital sea lanes and traffic separation schemes [
39] in the exclusive economic zone of Portugal [
40] and Spain [
41]. This fact confers great responsibility on maritime security, assuring navigation in the respective coastal waters and aiding vessels in avoiding accidents like the well-known Prestige oil spill [
42]. This coast is characterised by highly variable coastal systems (estuaries, rias and lagoons), namely the Rias Altas and Baixas (Galicia, Spain), and the Aveiro lagoon and Tagus estuary (Portugal). These systems have the characteristics of natural harbours and led to the establishment of the ports of Vigo (Rias Baixas), Aveiro (Aveiro lagoon), Lisbon (Tagus estuary) and more recently, the A Coruña Outer Port (northern coast of Galicia) (
Figure 1).
Drawing on the above, this paper compiles the knowledge of peer review literature on flood modelling in selected ports of the NWIP coast, based on the different geomorphological environments that characterise the NWIP coast, which present them with distinct challenges related to climate change impacts. The paper is structured as follows:
Section 2 provides a description of the studied coastal environments and inherent ports;
Section 3 characterises the climate drivers;
Section 4 provides the port flood assessment procedures;
Section 5 provides the flood assessment; and in
Section 6 the conclusion and implications are listed by identifying the most vulnerable ports to climate change according to their location. It is noteworthy that this review focuses on the most pessimistic RCP8.5 scenario of the IPCC for three climate periods, between 1979–2005, 2026–2045 and 2081–2099, corresponding to the historical, near future and far future periods. These periods were chosen considering previous works on coastal hazard assessments [
43,
44,
45,
46]. Additionally, a no-flood condition is compared with the flood extent under the climate change scenarios for 10-, 25- and 100-year return periods for storm surges, riverine input and wave regime (hereupon referred to as
Tr10,
Tr25 and
Tr100, respectively), identifying the port terminals susceptible to floods.
4. Port Flood Assessment Procedures under Future Conditions
Ocean circulation is a crucial climate regulator through complex and diverse mechanisms that interact to produce this circulation. Wind-driven combined with density-driven circulation defines ocean circulation from a global to a regional scale. This circulation is still poorly understood due to its volatility to climate drivers and their spatial variability, which are predicted to change during the following decades [
130,
131].
The flood assessment of port facilities under future conditions requires the simulation of the flow and short wave propagation at a local scale. As the mean sea level rises and storms become more intense, the prediction of these drivers becomes essential to address the climate change impacts on ports, which are susceptible to being flooded due to their location. This assessment is only viable through numerical modelling due to the large spatial and temporal scales involved. Several numerical models are suitable for modelling the hydrodynamics in deep, coastal and shallow waters. The model selection should consider the knowledge of the model’s limitations, the processes under study and the available data.
Considering these criteria, assessing future flood impacts is not straightforward, it requires data collection regarding meteo-ocean drivers and the statistical analysis best suited to identify the most accurate and reliable drivers to be used as forcing conditions in the numerical models. The meteo-ocean drivers are available in numerous databases, with the principal resource for future climate data being the global climate models (GCMs) from the different phases of the coupled model intercomparison project (CMIP). The most extensive databases are related to phase 5 of the CMIP (
https://www.wcrp-climate.org/wgcm-cmip/wgcm-cmip5 (accessed on 1 December 2022)), while the newest phase 6 is being published (
https://www.wcrp-climate.org/wgcm-cmip/wgcm-cmip6 (accessed on 1 December 2022)).
However, the CMIP6 GCMs have a coarse resolution, whereas CMIP5 GCMs have been used and applied to several regions by using a downscale of the GCMs and obtaining higher resolution regional climate models (RCMs). These RCMs can be found in other online databases where future wind and wave fields are available (Copernicus, the World Meteorological Organization’s Catalogue for Climate Data, the Commonwealth Scientific and Industrial Research Organisation (CSIRO), among others). These fields can be used in third-generation spectral wind–wave models (e.g., Delft3D though the wave module [
132], WAVEWATCH III [
133], SWAN [
134], among others) in order to downscale the global results at refined resolutions. The results from wave models can be used to determine the currents in the hydrodynamic model by using an offline or a two-way wave–current interaction, which is limited to the features of the hydrodynamic model and the type of data. A summary of the procedure followed in this study to assess the flood risk under the future winter climate is shown in
Figure 6.
4.1. Wave GCM Statistical Analysis
As mentioned before, it is essential to identify the best GCMs that characterise the wave climate of the study area. For this purpose, historical and future
Hs and
Tp data can be obtained from the simulations available in the online databases as a starting point. The skill of the CMIP5 GCMs available from the CSIRO database [
46] was evaluated in previous studies [
50,
51,
113] for the Atlantic Arc and the NWIP coast. This performance metric of GCMs has been applied [
135,
136,
137,
138,
139,
140] to explore the frequency and severity of climate extremes. Other indicators, such as root mean square error (RMSE) and bias were also used to assess the performance of various climate parameters [
141,
142,
143,
144,
145,
146] and the best GCM to characterise the winter wave climate in the study area [
34]. The authors applied the RMSE and the bias for the winter months in the NWIP coast (December, January and February) [
147] for the return periods of
Tr10,
Tr25 and
Tr100 and found that the BCC-CSM1.1 GCM represents the best climate model to reproduce the NWIP coast extreme wave climate.
4.2. Wave Dynamical Downscaling
The coarse resolution of the GCMs is not suitable to be used at a regional scale, which makes necessary dynamical downscaling to propagate the wave to a finer resolution at the port entrance. The third-generation spectral wave models can be used to perform the downscaling of the selected GCM. The downscaling can be conducted by implementing the model in a nested approach, where the domain resolution increases towards the study area. An example of a nested domain comprising three sets of domains with different resolutions is shown in
Figure 7.
The larger domain (L1) extends from 34.54° N and 14.54° W to 45.54° N and 5.54° W, followed by the second domain (L2) from 37.54° N and 12.54° W to 44.54° N and 6.88° W, and the third set of domains (Punta Langosteira (L3), Vigo (L4), Aveiro (L5) and Tagus (L6)) which correspond to the local ports domains. The resolution applied to each domain was the ratio of 1/3 from the previous domain, creating a resolution of 1/3°, 1/9° and 1/27° for domains L1, L2 and L3-L6, respectively. The numerical bathymetry was built based on the General Bathymetric Chart of the Oceans (GEBCO) (
www.gebco.net (accessed on 1 November 2022)), the European Marine Observation and Data Network (EMODnet) (emodnet.ec.europa.eu/en/bathymetry (accessed on 1 November 2022)) and the International Hydrographic Organization Data Centre for Digital Bathymetry (IHO DCDB) (
www.ncei.noaa.gov/maps/iho_dcdb (accessed on 1 November 2022)) database. Regarding the coastal and port regions, the use of high-resolution bathymetry is required, which was obtained from the port’s administration.
The wave field (
Section 4.1) is imposed at the open boundaries as a time- and space-varying data with the highest temporal resolution available for high accuracy results, which is 6 h for the BCC-CSM1.1 GCM. A wave spectrum of 25 frequencies (0.0418–1 Hz) and 36 directional bands can be applied using the SWAN numerical model [
34,
51,
113]. The model must also be forced with the wind GCM or RCM (if available) corresponding to the wave GCM/RCM used. For this region, only the BCC-CSM1.1 GCM r1i1p1 is available, which can be obtained through the CEDA archive (
https://archive.ceda.ac.uk/ (accessed on 1 November 2022)), the Coordinated Regional Climate Downscaling Experiment (CORDEX) (
https://cordex.org/ (accessed on 1 November 2022)) or the European Centre for Medium-Range Weather Forecasts (ECMWF) (
https://www.ecmwf.int/ (accessed on 1 November 2022)). This GCM is characterised by a horizontal resolution of 1.125° × 1.125° with a 3-hourly temporal resolution.
The wave model’s physical parameters included a stage 1 IEC model [
34,
50] and triads, bottom friction, depth-induced breaking and quadruplets [
51,
113,
148]. The simulations were run for the climate periods between 1979–2005, 2026–2045 and 2081–2099, corresponding to the historical, near future and far future periods.
4.3. Wave Parameters Characterisation
The
Hs,
Tp and
Dm (mean wave direction) parameters are often used to study extreme waves, including runup, erosion, and wave-driven longshore transport, which can change coastal processes during these extreme events [
144,
149]. Also, coastal flooding highly depends on
Hs and
Tp combined with high tides and storm surges [
150]. The wave parameters are obtained by the model used to downscale the GCM wave fields, which can be analysed to calculate the return periods for the wave parameters. The hourly values obtained over 1979–2005 were used to determine the maximum
Hs and
Tp and fitted to statistical distributions to estimate
Tr10,
Tr25, and
Tr100. The
Dm can be estimated by the wave arrival predominance. Similar to [
34], the mean wave direction was obtained for Lisbon, Vigo and A Coruña Outer Port (
Figure 8).
The historical (1979–2005), near future (2026–2045) and far future (2081–2100) return periods of
Dm,
Hs and
Tp can be observed in
Table 5.
4.4. Hydrodynamic Model
The Delft3D suite was used to simulate the Aveiro lagoon and Tagus estuarine hydrodynamics [
34,
62,
68,
87,
151], as well as the hydrodynamics of the Ria de Vigo and the coastal region around the A Coruña Outer Port [
60,
62]. The Delft3D-FLOW module [
152] was selected to simulate the tide, freshwater discharge and wind-driven flows, as well as the wave-driven currents with accurate results. The model implementations developed used local domains [
34,
68,
87,
151] and regional domains [
60,
62], focusing on hydrodynamic simulation. The numerical bathymetry used a combination of detailed topographic data from LIDAR surveys provided by the Portuguese General Direction of the Territory and bathymetry provided by the different port administrations under analysis. For detailed model implementation, the reader is referred to [
34,
62,
87]. The various model calibration was performed for different periods, showing high agreement between the modelled and observational data with high skill and low RMSE, where the best results were found in the vicinity of the interface between the coastal systems and the ocean, where the ports are located.
4.5. Flood Assessment Scenarios Design
A key challenge is to choose an appropriate method to predict floods under climate change scenarios. Several authors have obtained the return period of climate drivers and assessed the flood impact on coastal regions [
28,
29,
34]. A similar approach to [
34] was applied using the Delft3D FLOW and WAVE modules to propagate the wave parameters across the region (
Section 4.3) and simultaneously simulate the hydrodynamic conditions for each climate and return period. The hydrodynamic model uses the ESL, taking into account the mean sea level, storm surge height and astronomic tide, the riverine discharges and the wave parameters. The hydrodynamic implementation simulated the historical, near future and far future periods for the
Tr10,
Tr25 and
Tr100. A reference scenario representing the present extreme climate level was also determined (
Table 6).
These scenarios were imposed at the open ocean boundary of each model with the time series of the ESL, with nine days as the initial spin-up, and with a tide considering the reference value [
34]. After, the following two days considered the storm surge, tide and extreme river discharge, with the matching between their peaks representing the ESL for each climate period and return period (
Figure 9).
5. Flood Assessment
Flood extent maps and flooded areas of the port’s terminals are shown similarly to [
34]. The flood assessment considers the reference level as a starting point where no flood occurs in the terminals. The reference scenario considers the tide with the historical MSL and average riverine discharges. A comparison between the reference and the
Tr10,
Tr25, and
Tr100 of the ESL, storm surges, waves and riverine discharge scenarios is performed, indicating the port terminals susceptible to floods and the corresponding flooded area (
Table 7).
Previous works have shown that climate change is expected to increase Ria de Aveiro’s flood risk [
28,
30,
34,
70] due to increased oceanic and fluvial drivers, posing higher impacts at the end of the century [
19]. Most of the Aveiro port terminals lie near the lagoon entrance in a low-lying area exposed to oceanic drivers, such as storm surges. Ribeiro et al. [
34] observed that under the historical period (
Figure 10a), the flooded areas are located in the central lagoon region for a
Tr10 and will increase slightly with higher return periods. These flooded areas are expanded for higher return periods during the near future (
Figure 10b), but the flood risk will increase significantly (>60%) under the
Tr100 climate drivers at the end of the century (
Figure 10c). The authors also assessed which climate driver contributed most to the flood risk, observing that waves have a residual impact due to the lagoon geometry and the natural protection from the sea waves. The river discharges affect mostly the upstream terminals, whereas the MLS rise is an important driver of the flood risk due to its contribution to the total water level, which can be intensified under storm surge events, the most prevalent climate driver, which contributes to large flood extensions [
34].
Similar to the Aveiro port, the Lisbon port benefits from natural shelter, whereby its entrance orientation protects it from the most NNW energetic waves during winter. However, the low-lying margins of the estuary compromise the port’s resilience to rising sea levels caused by the climate drivers MSL, ESL and riverine discharges. As a result of these drivers, most of the Lisbon port terminals are threatened by floods (
Figure 11), with emphasis on the terminals located in the inner estuary for the historical period (
Figure 11a), as previously observed by [
83]. These terminals show significant sections of their area flooded under the
Tr10 scenario due to low-lying margins. The flood extent intensifies with higher return periods, being more notorious under the
Tr100, where most of the terminals in the inner estuary are exposed to intense floods. The flood extent for the historical period shows that Lisbon port’s resilience to floods is high for the terminals located at the entrance channel (10–16), while the terminals located inside the estuary are highly vulnerable to floods under the
Tr100 scenario.
The flood hazard increases for higher return periods with the increase of the ESL, whereas for future periods (
Figure 11b,c), the climate drivers, such as the MSL, promote large floods along the estuary margins [
86]. The terminals located in the entrance channel of the estuary face marginal inundations under the
Tr10 scenario, which increases significantly under the
Tr25 and
Tr100 scenarios. Lisbon port’s risk to floods is high with the far future climate drivers under
Tr10,
Tr25 and
Tr100 scenarios.
These floods are observed at present under dangerous meteorological conditions, and the increase in MSL is expected to enhance its extent and impact. Thus, the risk will increase significantly in some terminals under climate change scenarios, where large port sections may be flooded.
The Vigo port has high resilience to the climate drivers in the historical period, as only some marginal areas of the port terminals are inundated (
Figure 12a). The ports resilience is also observed for the near future climate drivers (
Figure 12b), where the flood extent impact slightly increases compared to the historical period. These floods occur at the margins of the terminals, where the slipways are located, which will not restrain the activities. The flooded area is considerably expanded for the far future climate drivers (
Figure 12c), representing some challenges regarding long-term climate change readiness for the uses and activities management and operability that lies at the estuary margins. During this period, the flooding of some port areas is quite significant, compromising the port activities in these sectors, particularly the central section of the port (4, 5 and 6). These flood patterns suggest that the climatic drivers considered are quite relevant for determining future flooding at this port, which could result in damage and interruptions to terminals in the distant future [
19,
153].
In contrast with the previous case studies analysed, the A Coruña Outer Port is located in a marine system influenced only by oceanic drivers. In addition, the port is relatively recent, and therefore climate change impacts were considered during its design and recent adaptations. Based on this, it is expected that climate drivers do not contribute to the flooding of the port terminals, as shown in
Figure 13. Note that terminal 1, an expansion terminal, is not in use and is intended to be flooded during high tides at the date of this assessment. Thus, the port shows high resilience to flooding, even under
Tr100 of the ESL. This resilience is also observed for future periods (
Figure 13b,c), where no flood is observed in the port terminals.
6. Conclusions and Implications
Climate change is expected to profoundly affect the facilities of the ports exposed to sea level rise and extreme weather events. The impacts can vary according to the geographic location and the surrounding pressure, from anthropogenic sources to morphological features, which can be identified as the significant interventions to the construction, operation and maintenance of the facilities that can result from the vulnerability to the various stresses along their life span. This issue is fundamental considering the exposition of some ports to climate drivers, whereas natural harbours shelter some ports, others rely on artificial structures that protect against high waves and strong currents.
The A Coruña Outer Port, Vigo, Aveiro and Lisbon ports were selected considering their locations, coastal, rias, lagoon and estuary environments, which may face different challenges under climate change. The flood risk of the ports was assessed under the historical and future climates. This assessment was conducted under extreme weather events during the winter season by considering climate change scenarios for 10-, 25- and 100-year return periods relating to storm surges, riverine input and wave regime. This analysis showed that the flood risk varies significantly according to the location of the ports, where some environments are more exposed to particular climate drivers than others, threatening the port’s performance.
Following the identification of the significant threats to each port, the flood risk due to the climate drivers was assessed, which can be summarised as:
Aveiro port, located in an inland lagoon system with a single connection to the ocean, serves as a natural shelter from hazardous weather and waves. This environment is characterised by low and flat topography making it susceptible to floods when sea levels or river discharges are high. The combination of these drivers increases the flood hazard drastically, which was observed in the Aveiro port under the Tr100 scenarios with several terminals subject to significant inundation. The flood extent increases significantly at most of the terminals under future drivers, highlighting the flood risk resulting from the mean sea level rise and storm surges, which can be enhanced by coincident river discharges;
Lisbon port is located in the Tagus estuary. The hydrodynamic drivers are directly influenced by the geomorphology of the estuary, with a deep inlet channel and a broad and shallow inner domain. These characteristics constrain ocean waves’ propagation and limit the influence of river discharges in the estuary. However, the estuary geomorphology, hydrodynamic conditions, and territory occupation promote large floods observed under the Tr100 scenarios for all climate periods. The flood risk in these scenarios is higher in the inner domain, where several terminals may be completely flooded because of the end of century climate drivers. This highlights the need to increase the port’s resilience to future ESL;
Vigo port is situated in the Ria de Vigo. The ria is characterised by a wide mouth and a funnel shape, and it is sheltered by the Cíes Islands, which grants protection from extreme waves. The port terminals face partial inundations under far future scenarios, with increased flood risk in the Tr100 scenarios for some terminals. The flood patterns show that Vigo port is susceptible to flooding from the combination of sea level rise and storm surges under climate change drivers;
A Coruña Outer Port is different from the other ports, being the most recent and located in a coastal environment, which is exposed to Atlantic storms. This port relies on an artificial shelter from hazardous weather provided by breakwaters. The port’s high resilience to climate change protects the port from floods driven by the ESL and storm surges.
From a holistic standpoint, a port’s flood risk under the future climate will increase significantly due to winter storms if climate change drivers accelerate as predicted. Port infrastructure needs to be strengthened to protect them from sea level rises, which brings local problems in terms of application and planning. There can be no single general solution at a port or national scale because the infrastructure and risk vary in this space. The combination of MSL rise and storm surges are the main causes of floods, but the effects of river discharge and waves cannot be excluded, as they can contribute significantly to the ESL in lagoons and coastal environments, respectively. Waves mainly affect the ports located outside of natural harbours, which can cause flood overtopping in these locations. In addition, a sea level rise is projected to double the frequency of coastal flooding at most locations [
21,
22], enhancing overtopping by energetic waves during hazardous weather.
Coastal flooding is the dynamic interaction of various oceanographic drivers [
154,
155,
156], which may exhibit high interannual variability, such as the North Atlantic Oscillation (NAO) [
157,
158,
159,
160]. Future model developments should consider other processes responsible for port flooding (e.g., harbour oscillations and NAO) and validate the flood extent using satellite-derived data. However, this task can be challenging due to the lack of good satellite coverage acquisition during storm conditions. Moreover, the flood assessments analysed have large time gaps and do not include the energetic winter storms that occurred after 2005 [
161,
162,
163,
164,
165,
166,
167]. Flood hazard studies should also consider the non-linear interaction between the climate drivers and their contribution to water level variability. Previous works [
168,
169,
170,
171] suggest that climate driver interactions cannot be neglected, which can occur when assuming the linear addition of sea level rise (among other climate drivers). These non-linear interactions can represent a significant part of the water level changes. Arns et al. (2015) [
172] assessed the impact of sea level rise on storm surge water levels and observed a slight increase in return water levels when considering the interactions between mean sea level, tide and storm surge (12 cm in a 50-year return) compared to those who neglect these interactions.
In the present form, the flood assessment may be underestimated and future studies should consider the most recent data. Other anthropogenic factors such as dredging, the expansion or construction of new ports, flood barriers, and other structures could modify the flood extent. Thus, future research should consider present and projected anthropogenic activities and uses when assessing the flood extent.
The definition of adaptation or mitigation countermeasures against flooding is not straightforward due to the high uncertainty of projections related to climate change’s multiple spatial and time scales. Climate projections rely on probabilities of specific processes occurring, and ports tend to focus on the observed general conditions and act accordingly by adapting or building new structures to reduce the impacts. A port’s planning horizons are also around 5 to 10 years [
173] and are subject to the volatility of changing business circumstances. These timescales are much shorter than existing infrastructures and typical climate change timescales, which are on the scale of thirty to hundreds of years, making it challenging to implement efficient long-term measures. For this reason, a port’s resilience and adaptation planning requires a risk assessment to design the adaptation planning, which could include the construction of seawalls, dikes or marine ecosystems, and relocation or elevation to compensate for a sea level rise [
19,
174].
Drawing on the above, planning responses must be applied to the resilience of the infrastructures and operational measures, like relocating the terminals and zones according to the commercial value of the services and goods. However, continuous observation and flood assessment must be performed uninterrupted due to the development of new climate models, which will result in more reliable and accurate climate change predictions and serve as a support toolkit to assist in the definition of adaptation, management, and resilience measures for future challenges.