1 Introduction

Heavy rainfall can cause natural hazards such as landslides or floods damaging infrastructures (e.g., rail networks), affecting hydroelectric power production (e.g., power lines), or causing a decline in agricultural industry (e.g., crop loss) (Kuriqi et al. 2019; Lee et al. 2017; Lorenzo et al. 2013; Rahmani and Harrington 2019). At the same time, extreme winds can also cause important and immediate hazards to society. These events can impact afforestation (e.g. fallen trees), and security of large buildings and infrastructures (e.g. damage to roads), and are also often responsible for the worst wind turbine damages in wind farm facilities (Forzieri et al. 2018; Piasecki and Żmudzka 2022; Pryor and Barthelmie 2021; Quine 2000). Such hazards can cause casualties and damages, with direct and indirect economic repercussions (Li et al. 2021; Mansour 2019), making changes in extreme events a very active research area.

In the last decades, there has been a remarkable number of extreme climate events related to precipitation and wind over Europe, which have posed a threat to human settlements across the territory (Armada Brás et al. 2023; Küfeoğlu et al. 2014; Macdonald 2012; Ulbrich et al. 2003). The occurrence of these events can be particularly damaging when they occur individually (Berz 2005; Della-Marta et al. 2009; Madsen et al. 2014; Wang et al. 2011). However, it must also take into account that these phenomena can occur simultaneously, causing frequently most severe and, in some cases, irreversible impacts on human and natural systems (Zscheischler et al. 2018, 2020). Thus, in recent years there has been a growing interest in studying this type of phenomena. Concurrent precipitation and wind events have been recently explored over Europe (Bloomfield et al. 2023; De Luca et al. 2020; Hénin et al. 2021; Martius et al. 2016; Owen et al. 2021) finding evidence that they tend to systematically occur. Nevertheless, this occurrence varies between regions with high percentages of co-occurring extremes detected during winter along the western coast of the Iberian Peninsula, northwestern central Europe, the western coast of Norway, and the eastern coast of Greece. Lowest co-occurrences were observed in other locations such as eastern Norway and Sweden.

Over western Europe, several studies of strong storms have securely established that both wind and precipitation damage might co-occur during the same weather system (Fink et al. 2009; Liberato 2014; Otto et al. 2018), being the Iberian Peninsula (IP) one of the most affected regions as the first point of arrival of Atlantic disturbances (Liberato and Trigo 2014; Pereira et al. 2018). Nevertheless, studies focused on future changes in the co-occurrence of extreme precipitation and wind speed events at a regional scale over the IP are still scarce. The present study aims to advance the understanding of concurrent extreme wind and precipitation events over the IP (Fig. 1), analyzing the projected changes in these phenomena throughout the 21st century (2021 to 2100) based on CMIP6 model simulations. Future conditions are evaluated considering a moderate (SSP2 − 4.5) and a high (SSP5 − 8.5) radiative forcing pathways.

Fig. 1
figure 1

Map of the study area

Full size image

2 Data and Methods

2.1 ERA5

Daily precipitation and near-surface (10 m) wind speed data with 0.50° spatial resolution from ERA5 were used to compute historical concurrent events over the IP. ERA5 is based on the latest global reanalysis data set provided by the European Centre for Medium-Range Weather Forecasts (ECMWF) (Hersbach et al. 2020). To evaluate the historical performance of CMIP6 models, data from 1995 to 2014 were considered. These historical data were considered up to 2014 to match the historical period corresponding to the CMIP6 simulations, ending in this year.

2.2 CMIP6 Models

Daily precipitation and near-surface (10 m) wind speed data derived from six CMIP6 global climate models (Table 1) were considered to analyze changes in concurrent events across the IP. Data from historical (1995 − 2014) and future (2021 − 2100) simulations under two different emissions scenarios (SSP2 − 4.5 and SSP5 − 8.5) were used. The choice of these models was conditioned by those that performed best in the study area (Brands 2022) and which also provide data available for both scenarios.

Table 1 CMIP6 models used in the comparison. N indicates the number of grid elements (lon × lat), GR the average horizontal grid resolution near the study area (lon × lat, º), and V the number of vertical layers for the atmospheric model’s component. The r1i1p1f1 ensemble was used for all the ESMs
Full size table

To evaluate the CMIP6 model’s performance, data for each climate model were re-gridded into 0.50°×0.50° of spatial resolution based on a linear interpolation to allow for comparison with ERA5. Then, a mathematical comparison between these models and ERA5 was performed from 1995 to 2014. Previous studies (Laurent et al. 2021; Rickard et al. 2016) employed Root Mean Square Error (RMSE) and Bias errors to compute a score and measure the match between climate models and observed data, thereby ranking the climate model’s performance. Following the methodology established by these authors, RMSE, Bias, and the correlation coefficient (R) for daily precipitation and wind speed were considered to calculate the score of agreement (SR), which represents the accuracy and variance between CMIP6 models outputs and ERA5 data. Following Pereira et al. (2023), the RMSE, Bias, and R metrics were used to calculate the SR score for wind speed and precipitation using the Eq. (1):

$$\:SR=\left(RMSE+\:\left|Bias\right|\right)\:\text{x}\:\left(1-R\right)$$
(1)

RMSE, Bias and R were calculated as:

$$\:RMSE=\sqrt{\frac{\sum\:_{i=1}^{N}{\left({M}_{i}-{O}_{i}\right)}^{2}}{N}}\:\:$$
(2)
$$\:Bias=\frac{\sum\:_{i=1}^{N}\left({M}_{i}-{O}_{i}\right)}{N}\:$$
(3)
$$\:R=\frac{{\sum\:}_{i=1}^{N}\left({M}_{i}-{\stackrel{-}{M}}_{i}\right)\left({O}_{i}-{\stackrel{-}{O}}_{i}\right)}{\sqrt{{\sum\:}_{i=1}^{N}{\left({M}_{i}-{\stackrel{-}{M}}_{i}\right)}^{2}}\sqrt{{\sum\:}_{i=1}^{N}{\left({O}_{i}-{\stackrel{-}{O}}_{i}\right)}^{2}}}\:\:\:$$
(4)

Where, \(\:N\) is the number of match-ups, \(\:M\) stands for CMIP6 and \(\:O\) for ERA5 data.

The lower the SR, the better the agreement between CMIP6 and ERA5. This score indicates the accuracy and variance between datasets allowing to establishment of a model ranking.

The overlapping percentage (OP, Perkins et al. 2007) has also been calculated for both datasets. This is a very simple metric to quantify the similarity between datasets frequently used in the analysis of climate data and climate modeling (Costoya et al. 2020; Houndekindo and Ouarda 2024). This statistic is clear, easily interpreted, and directly comparable across variables, providing a quantitative measure of similarity. The OP was calculated following the Eq. (5):

$$\:OP\left(\%\right)=100\:\text{x}\sum\:_{1}^{n}minimum\:\left({Z}_{CMIP6},{Z}_{ERA5}\right)\:\:$$
(5)

Where, n represents the number of bins (1 mm and 1 m/s for precipitation and wind speed respectively), ZCMIP6 is the frequency of values in a given bin from CMIP6, and ZERA5 is the frequency of values in a given bin from ERA5 data. If a model simulates the ERA5 conditions perfectly, the OP will be equal to 100%.

2.3 Defining Concurrent Events

Compound precipitation and wind extremes were defined as the simultaneous extremes at the same grid point (Ridder et al. 2020, 2021). Precipitation and wind speed data above the 95th percentile for the reference period (1995–2014) were considered as extremes. Previous works have analyzed extreme events considering the 98th or 99th percentile (Martius et al. 2016). The 95th percentile was used since this threshold ensures a sufficient amount of data for statistical analysis (Li et al. 2022; Plavcová and Urban 2020; Zhang et al. 2021). A co-occurrence was recorded if precipitation and wind speed above this percentile occurred within the same day, at the same grid point.

The analysis was carried out for winter (DJF), spring (MAM), summer (JJA), and autumn (SON). Then, changes in the number of occurrences were evaluated for the two scenarios considering three periods (near-term (2021 − 2040), mid-term (2041 − 2060), and long-term (2081 − 2100) concerning the reference period (1995 − 2014). The trend significance (statistical significance level computed at 5%) was assessed through the nonparametric Mann-Kendall’s test (Kendall 1975; Mann 1945).

2.4 Study Area

The IP is located in the southwestern Europe. The northern and western regions are bordered by the Atlantic Ocean, whereas the southern and eastern regions, are influenced by the Mediterranean Sea (Fig. 1). The territory presents a complex orography with several mountain ranges separated by broad plains or depressions, that causes great contrast in the atmospheric conditions among regions (Royé et al. 2019). This orography, combined with Mediterranean and Atlantic low-level moisture, as well as mesoscale and synoptic scale thermal and dynamical forcing, endows this region with remarkable climatic diversity, ranging from oceanic and Mediterranean to continental climates (Diaz-Poso et al. 2023; Rodriguez and Lemus-Canovas 2023).

Temporal variability of precipitation over the IP presents strong spatial gradients and seasonal character (Parracho et al. 2016; Serrano-Notivoli et al. 2018). Large spatial contrast can be observed among wet mountainous regions to the north, and dry plain regions in the south. In autumn, winter and spring, precipitation is mostly due to synoptic perturbations moving eastward from the Atlantic Ocean, causing higher precipitation in the western regions. On the other hand, in summer, the precipitation is mostly associated with convective storms due to ground heating and high moisture content and can occur in the central and eastern parts of the IP (Serrano et al. 1999).

Regarding wind, the complex orography of the IP leads to the emergence of numerous regional winds, mainly resulting from air flows driven by heat flux and convective activity associated, for example, with storms (Lorente-Plazas et al. 2015; Ortega et al. 2023). Thus, regional winds are usually generated by small-scale processes, and capturing the orographic characteristics of these winds requires high-resolution products, generally with a spatial resolution finer than 10 km and a temporal resolution of hours. Wind also responds to changes in large-scale circulation, mainly controlled by internal decadal ocean-atmosphere oscillations, nevertheless, identifying the influence of the various scales of movement involved in the wind response is a complex task. However, areas with common features of wind speed have been reported over the IP in the literature. The windiest areas are, generally, the Galician coast, the Ebro valley and the Strait of Gibraltar (Lorente-Plazas et al. 2015). Concerning seasonal wind regime, spring tends to be the windiest season for most of the wind locations. However, the western coast can also present a maximum strength in summer, and the northern area in the late winter. Autumn is usually the least windy season for most places and winter can present high variability (Lorente-Plazas et al. 2015; Martin et al. 2011).

3 Results

3.1 Model Evaluation for Precipitation and Wind Speed

To analyze the performance of CMIP6 models, the score (SR) and overlapping percentage (OP) of each model in simulating precipitation and wind speed was calculated seasonally for the historical period (1995 − 2014, Figs. 2 and 3). The analysis for precipitation (Fig. 2) indicated a higher agreement between CMIP6 and ERA5 results during the summer, with the lowest SR values. For the rest of the seasons, similar SR values were obtained for each model, although with slightly higher SR values than for summer. Analyzing the OP for precipitation data, values were greater than 80% for all models and seasons indicating that, in general terms, model results can simulate the precipitation with an adequate correspondence in model performance.

Fig. 2
figure 2

Seasonal scores (SR) and overlapping percentage (OP) calculated for precipitation data for each climate model

Full size image
Fig. 3
figure 3

Seasonal scores (SR) and overlapping percentage (OP) calculated for wind speed data for each climate model

Full size image

Regarding wind speed (Fig. 3), if each season is considered individually, the CMCC-ESM2, EC-Earth3, and GFDL-CM4 models showed the lowest SR with similar values between them. The BBC-CSM2-MR model showed the highest SR for the four seasons indicating the weakest match between CMIP6 and ERA5. The OP analysis confirmed these results with values lower than 55% and 65% for the BBC-CSM2-MR and MRI-ESM2-0 models respectively. On the other hand, the three previously mentioned models with the lowest SR, also showed the highest values for the OP in all seasons (greater than 70%). The MPI-ESM-1-2-HR also showed high OP values, although these values fall below 70% during the winter season. These results highlight the performance of CMCC-ESM2, EC-Earth3, and GFDL-CM4 as the best models simulating wind speed. Following these results, a multi-model ensemble mean from these three models was considered for subsequent analysis.

3.2 Historical Performance for Compound Events

To analyze the frequency of concurrent extremes over the historical period (1995 − 2014), maps of the number of occurrences for ERA5 and CMIP6 multi-model ensemble were calculated (Fig. 4). CMIP6 models depicted a similar pattern to ERA5 for all seasons indicating that the spatial distribution of compound events was correctly captured. In addition, a considerable regional variability of such events was observed for each season and both datasets.

Fig. 4
figure 4

Number of occurrences of compound events (days) from 1995 to 2014 for ERA5, CMIP6 ensemble, and differences between them

Full size image

Winter and autumn maps presented a similar pattern with more occurrence of compound events in western IP. The highest number of events (around 50 days) were detected over the northern areas of Portugal and Galicia.

In spring, a spatial distribution similar to that of winter was observed. A decrease in the number of events was detected throughout the territory with maximum values (around 40 days) in northern Portugal. In summer the occurrence of compound events decreased markedly, showing less than 25 events in the northwestern area and less than 10 events for the rest of the peninsula.

The differences between datasets indicated some variances (Fig. 4, last column). For winter, CMIP6 results showed more events than ERA5 over the northern plateau and all the eastern part. On the contrary, in the western and southern regions, ERA5 presented more concurrent events than CMIP6. For spring, overestimations from CMIP6 models were more visible across large parts of the territory except over the southern plateau and Galicia. Summer season presented the lowest differences (< 10 days) for the whole region with a slight overestimation in the northeast and a slight underestimation in the southwest. Finally, for autumn, CMIP6 hindcast presented higher values than those corresponding to ERA5 concentrated over the west-central region.

3.3 Future Projections

3.3.1 Projected Compound Events

To investigate future changes in compound events, trends from 2021 to 2100 were evaluated for each season under future SSP2 − 4.5 and SSP5 − 8.5 scenarios (Fig. 5). The analyses were carried out by calculating the future occurrence of compound events based on the 95th percentile of the reference period (1995–2014).

Fig. 5
figure 5

Computed trends (events per decade) in the occurrence of future compound events in the period 2021 − 2100 under SSP2 − 4.5 and SSP5 − 8.5. Dots indicate significant trends

Full size image

For winter, considering the SSP2 − 4.5 scenario, a general decreasing trend over most of the IP is predicted, with maximum significant values (close to -1 event per decade) over the northwest region. Upward trends are observed scattered in central areas, eastern coast, and north of Galicia. Under the SSP5 − 8.5 scenario, is foreseen a similar spatial distribution with negative trends over northern and southern regions. Nevertheless, clear positive trends are projected over central Portugal and the southern plateau.

For spring trends range from 0.5 to -0.5 events per decade all over the territory and for both scenarios. Negative trends for the moderate emission scenario are mainly predicted on the western side of the peninsula. Positive trends are only projected over Galicia and some southeastern areas. For the high emission scenario, the spatial pattern is slightly different with decreasing trends over the northern plateau, Cantabrian mountains, and the northeasternmost region. Positive trends are mainly detected throughout the southern areas and west of the Pyrenees.

For summer, the SSP2 − 4.5 scenario presents a small number of grid points with significant trends ranging from 0.5 to -0.5 events per decade. Negative values are mainly projected over the northern and southeastern areas, while positive trends are foreseen over the southwestern areas and the northeasternmost region. The area of significant trends increased for the SSP5 − 8.5 scenario, where large areas of negative trends were detected across the territory.

For autumn, negative trends for most of the peninsula with scarce positive trends dispersed in southern Portugal and northern Spain are predicted under the SSP2 − 4.5 scenario. The highest negative trend values (close to -1 event per decade) are observed over the central area. For the SSP5 − 8.5 scenario, more positive trends are observed over the eastern region, while negative trends are projected in southern Portugal, with values around to -1 event per decade.

The generalized trends in the occurrence of future compound events indicated a decrease in a large part of the IP during 2021 − 2100. Nevertheless, some areas clearly showed positive trends, highlighting the need to perform a more detailed analysis.

Figure 6 shows the changes in projected compound events under the SSP2 − 4.5 scenario. Winter projections corresponding to the near-term future indicate a general decrease over the peninsula except over the southwestern area with an increase of around 10%. A similar spatial pattern is observed for the mid-term future with a projected increase in the occurrence of compound events in the southwest and an enlargement of the affected areas. In this regard, this increase is higher (around 20%) than for the previous period. For the long-term future, results show an important increase all over the western and southern areas with occurrences greater than 30%.

Fig. 6
figure 6

Seasonal changes (%) in projected compound events under the SSP2 − 4.5 scenario

Full size image

For spring, a different spatial pattern is observed. For the near-term future, the northwestern and southeastern corners of the IP are projected to experience more compound events (around 20 − 30% and 90 − 100%, respectively). This increase in the southeast is also projected for the mid-term future, while a decrease is predicted over Galicia. In addition, in the Pyrenees region, an increase of 100% in compound events is projected. For the long-term future, an increase is again foreseen over Galicia, although with lower values than in the near-term future. Similar to the mid-term results, the northeastern areas are also expected to undergo more compound events.

Projected changes for summer considering the near-term future show an increase (up to 100%) in the compound events in the north. The mid-term and long-term futures depicted a similar pattern to the spring season but with lower magnitudes.

Finally, for the autumn season, positive differences over large eastern areas (the Mediterranean coast and some regions in the south) are projected for the near-term future. For the mid-term future, compound events are expected to increase over northeastern areas and some southern regions, while in the long-term future, a general decrease is anticipated throughout the territory except in the easternmost region.

Considering the SSP5 − 8.5 scenario (Fig. 7), in general, it is observed that spatial patterns are different from the ones corresponding to the SSP2 − 4.5 scenario.

Fig. 7
figure 7

Seasonal changes (%) in projected compound events under the SSP5 − 8.5 scenario

Full size image

For winter, a general decrease in the occurrence of compound events is projected for the near and mid-term futures, with small scattered areas with increments of 10%. Nevertheless, at the end of the century (2081 − 2100), the northwestern region will experience an important increase (up to 30%).

Projected changes, for spring, show a general decrease in the number of events as the century progresses. However, more compound events in the north (up to 100%) are projected in the near-term future. For this season, an increase between 20 and 60% in the occurrence of compound events over the Galicia region is projected throughout the century.

Summer results in the near-term future indicate some increases (up to 100%) over the northeastern, southeastern, and southwestern coasts. These increases tend to dissipate as the century progresses, and a general decrease is observed for the whole peninsula by 2081 − 2100.

For autumn, positive changes (10 − 20%) in the occurrence of compound events are predicted over the northeastern region in the near and mid-term future. These increases are projected to reduce in the long-term future.

3.3.2 Total Area Affected by Compound Events

The extent of total exposure to future compound events was also evaluated by calculating the percentage of grids showing an increasing or decreasing trend in Figs. 6 and 7, related to the total number of grid points covering the whole IP change (Fig. 8).

Fig. 8
figure 8

Percentage of grids showing increases (red bars) and decreases (blue bars) in the number of compound events under the SSP2 − 4.5 (solid bars) and SSP5 − 8.5 (striped bar) scenarios

Full size image

For winter, an increase in the overall percentage of areas showing more compound events is detected under both scenarios. Nevertheless, a higher percentage of the area affected by the moderate emission scenario is observed. For this scenario, the area experiencing compound events could increase from about 10% in the near-term future (2021 − 2040) to nearly 40% in the long-term future (2081 − 2100). For the high emission scenario, an area of less than 20% in the occurrence of compound events is obtained for the end of the century.

For spring, results revealed a decrease in the percentage of areas with positive trends in the occurrence of compound events for both scenarios. This season shows a different pattern from winter with a larger area experiencing compound events for the SSP5 − 8.5 throughout the century. In addition, it should be noted that at the beginning of the century, nearly 20 − 30% of the IP represents an increase in compound events, which is more than double that observed in winter for the same period for both scenarios.

Regarding summer, a similar pattern of a lower percentage of areas showing an increase in compound events throughout the century is also observed for both scenarios. Results suggest a higher percentage of areas (around 20 − 30%) that may experience more compound events at the beginning of the century, concerning the winter season.

Autumn results show a contrasting pattern between both scenarios. Under the high emission scenario, the percentage of areas with increasing trends in the occurrence of compound events shows similar values along the near and mid-term future periods (around 10%), with a slight decrease by the late century (around 3%). Nevertheless, the moderate emission scenario depicted the same pattern as for the previous seasons, with a decrease in the percentage of areas where an increase in compound events is expected. Moreover, for this season, the maximum percentage (around 40%) of the area with an increase in projected compound events is predicted in the near-term period.

4 Discussion

4.1 Historical Performance

In the current context of climate change, there are new challenges that require us to make an effort to adapt to changes that are occurring and are expected to occur. Weather and climate phenomena affect sectors as diverse as energy, agriculture, and health, as well as the ecosystems that surround us. If we consider extreme phenomena, the problem becomes more acute and it is therefore necessary to understand what is happening at a regional scale.

Compound wind and precipitation extremes can exert significant societal and economic impacts, for example, increasing the time for soil moisture, overloading emergency services and generating financial implications for the insurance industry. The IP is often affected by these compound events. Our analysis from the spatial distribution of seasonal compound events over the historical period (1995 − 2014), showed the ability of CMIP6 models to capture the spatial distribution of historical compound events (Fig. 4). A high occurrence of such events in the past decades was detected along the western IP during winter. This pattern is in agreement with a recent work by Hénin et al. (2021), who found that these types of concurrent events are more likely to occur in the northwestern area in wintertime. Nevertheless, the results of our study also indicated that these compound events occur throughout seasons. Thus, although these phenomena are less frequent, it is important to analyze seasonal variations as they can help to understand possible future changes in these events.

Past studies focused on historical compound wind speed and precipitation extreme events over Europe, also found a high percentage of concurrent events on the western coast in winter (De Luca et al. 2020; Martius et al. 2016; Owen et al. 2021; Ridder et al. 2020). This area is affected by extratropical cyclones, which are associated with extreme precipitation and wind speed in Europe during winter (Martius et al. 2016; Pfahl 2014; Raveh-Rubin and Wernli 2015). In addition, these compound events also occur along with frontal structures and storm clustering happening over western Europe (Priestley et al. 2018; Schemm et al. 2017). The western IP coast, especially the northwestern area, is the region most exposed to North Atlantic disturbances (Eiras-Barca et al. 2018; Liberato and Trigo 2014; Pereira et al. 2018). Some previous results also suggest that high co-occurrences have both extremes caused by the same weather system while low co-occurrences are observed in regions where the extremes come from different weather systems (De Luca et al. 2020; Martius et al. 2016; Owen et al. 2021).

The occurrence of compound events in autumn and winter, mainly in the western part of the IP, could also be analyzed in terms of atmospheric dynamics over the North Atlantic. Thus, circulation patterns have been extensively used to enhance the understanding of atmospheric dynamics variability and weather extremes (Breton et al. 2022; Grams et al. 2017). Previous studies found that the North Atlantic Oscillation (NAO) has a key role in winter precipitation and wind regimes over the IP explaining a great percentage of climatic variability over the past decades, mainly over the western IP (Barton et al. 2022; Chidean et al. 2018; Turki et al. 2023). Several studies have focused on how large-scale circulation influences wind variability over the IP analyzing the North Atlantic Oscillation (NAO) (Garcia-Bustamante et al. 2012; Jerez et al. 2013), although there are other modes of variability that can also affect the Western Mediterranean climate. Thus, for example, the Eastern Atlantic (EA) pattern is correlated with precipitation regime for all seasons mainly over the northwestern IP (Lorenzo et al. 2008). In summer months, the most frequent weather type over the IP is the thermal low, which causes warm and humid air flows inland (Ortega et al. 2023). Some studies have also identified significant correlations between the Atlantic Multidecadal Oscillation (AMO) and the climate of the Mediterranean region involving precipitation, especially in summer and the intermediate seasons (Sutton and Dong 2012; Zampieri et al. 2017).

Compound events can also occur due to severe convective systems. In fact, in the central and eastern part of the IP, episodes of intense precipitation and strong winds can be caused by convective storms, mostly during spring and autumn (Parracho et al. 2016; Serrano et al. 1999). These episodes have a much higher hourly precipitation intensity, and are accompanied by very strong wind gusts, but doesn´t last a few hours at the most. Nevertheless, CMIP6 data do not include sub-daily time scale observations and the spatial resolution is too coarse, resulting in the potential omission of local events due to the models resolution limitations. Such changes should be deeply explored when high-resolution projections become available to better understand the spatial and temporal evolution of compound events.

4.2 Future Changes

The generalized trends in the occurrence of future compound events indicated a decrease in a large part of the IP during 2021 − 2100 (Fig. 5). These trends could be analyzed in terms of the research on the future behavior of atmospheric dynamics over the area under study. Several studies have analyzed changes in weather regimes based on future projections for Europe. Results indicated a projected decrease in westerly weather patterns affecting the Iberian Peninsula for all seasons, leading to drier conditions in the future (Breton et al. 2022; Lorenzo et al. 2011). Additionally, our results could also be related to projections of the temporal behavior of the North Atlantic Oscillation (NAO) which has historically modulated precipitation on interannual timescales over southwestern Europe. Thus, an increase in the frequency and persistence of the positive phase of the NAO regime is expected in the future. This increase is associated with higher winter precipitation in northern Europe and correspondingly lower precipitation over the Mediterranean region (Fabiano et al. 2021; Ullmann et al. 2014).

In spite of the widespread trend towards negative tendencies, some areas clearly showed positive trends. Thus, to better analyze future changes in compound events, a near-term (2021–2040), mid-term (2041–2060), and long-term (2081–2100) periods were evaluated. Overall, projected changes for the SSP2 − 4.5 scenario (Fig. 6) indicated a significant regional variability during all seasons. For winter, the western areas are projected to experience more compound events affecting an increasing area throughout the century. These results could be related to future changes in extratropical cyclones. Although the response of these phenomena to climate change is uncertain, some previous studies have investigated this topic based on CMIP5 and CMIP6 models (Mizuta 2012; Priestley and Catto 2022; Zappa et al. 2013, 2015). General results indicated that the total number of cyclones is projected to decrease globally by the end of the 21st century, although extratropical cyclones are projected to increase in intensity and severity in the Northern Hemisphere during winter. A further increase in such events is expected during the winter months in central Europe, especially in those associated with strong precipitation. In addition, an increase in the number and intensity of cyclones associated with strong wind speeds is also projected over the United Kingdom and central Europe from December to February (Zappa et al., 2013).

In our study, the results for the spring, summer, and autumn (Fig. 6), showed a general decrease in the occurrence of compound events towards the end of the 21st century, with the most significant decline in autumn. Despite this general decrease, the occurrence of these phenomena over some northeastern regions in spring and summer is maintained throughout the century. Thus, although most of the past studies analyzed winter compound events, studying these types of events throughout the year is also important.

Winter projected changes for the SSP5 − 8.5 scenario (Fig. 7) indicated a slight increase of compound events over the northwestern region at the end of the 21st century. Nevertheless, this increase is lower compared to the SSP2 − 4.5 scenario, where large parts of the IP are projected to experience more compound events (Fig. 6). For spring, Galician territory may experience an increase in compound events throughout the century with higher magnitudes than for the SSP2 − 4.5, indicating amplified changes under the high emission scenario. Regarding summer and autumn, changes under moderate and high emission scenarios showed similar patterns with a general decrease in the occurrence of compound events throughout the century with small differences in magnitude.

Overall, these results indicated an increase in the areas of the IP affected by the occurrence of compound events during winter throughout the century, and under both scenarios. However, the affected area is larger for the SSP2 − 4.5 scenario. For the rest of the seasons, a decrease in the percentage of areas with the occurrence of compound events was detected throughout the century. Nevertheless, important differences were found at the beginning of the century, when these seasons showed a larger area experiencing compound events compared to the winter season.

Projections of compound wind speed and precipitation extreme events at a regional scale are still scarce and do not allow a direct comparison with the results of this study. Some previous studies at a global scale (Ridder et al. 2022; Zhu et al. 2023) found a clear increase in compound events over large areas of central Europe at the end of the 21st century, particularly under high emission scenarios. The sign of change in Western Europe also indicated more frequent events in some areas of the IP under the SSP5 − 8.5 scenarios over the century. However, these projections of future changes were considered unclear, highlighting the importance of regional analysis.

Previous results suggested the uneven influence of climate change on the occurrence of compound events over the IP in the different seasons and periods analyzed. It is important to note that all the results of this study were obtained considering the occurrence of compound events on the same day and grid point, which is a conservative definition used in studies of different regions worldwide (Li et al. 2022; Ridder et al. 2020, 2021, 2022; Zhu et al. 2023). Previous publications have investigated occurrences expanding temporal and spatial shifts (Martius et al. 2016; Zhang et al. 2021; Zscheischler et al. 2021), finding that the average number of this type of compound events could increase. The present study demonstrated future increasing and decreasing trends throughout the IP with significant regional variability depending on the season analyzed. This variability reveals the importance of analysis at a local level to prevent and mitigate the consequences of future occurrences. These events can affect multiple sectors causing, for example, critical infrastructure damages, paralysis of public transportation, large-scale power outages, posing a major risk to the food supply chain, and amplifying threats to global food security (Bloomfield et al. 2023; Hillier et al. 2020; Li et al. 2022; Ward 2013; Zscheischler et al. 2018, 2020). Thus, it is important to explore projected changes of these compound events in the future, to advance in the knowledge of their occurrence to be able to integrate them into regional risk and adaptation management.

Compound wind and precipitation extremes occurring simultaneously over the same region may undergo large impacts on ecosystems and human society. The occurrence of these events over the IP could have important repercussions, for example, on energy supply from renewable sources. Thus, these events could significantly damage wind turbines in wind farms. In addition, wind turbines should restrict their operation during these events to minimize the cut-out events.

Wind energy has emerged as a leading contender in Europe due to the successful deployment of mainland wind turbines, an efficient and mature technology. In the IP, there are a significant number of wind farms, which continue to grow each year with new installations. As a result, wind energy is becoming the primary source of electricity generation in the region. Changes in compound wind and precipitation extremes may significantly influence electricity generation at wind farms over this territory. This could impact operating margins and investments within the sector, thereby influencing the overall economic landscape.

Important infrastructures such as rail networks can also be affected by concurrent events that create weather-related incidents (e.g., trees on line, landslides and delays to trains) leading to an associated cost. Spain has a highly developed railway infrastructure with a high-speed rail among the largest networks in Europe. Thus, compound events could adversely affect this infrastructure with respect to mobility and, even, environmental impact, given that rail transport has significant potential to achieve substantial reductions in greenhouse gas emission and energy consumption. High-speed rail operates on electricity, which benefits the environment by shifting passengers and freight to modes of transport with less environmental impacts.

Compound wind and precipitation extremes can have important societal and economic repercussions. Consequently, it is important to analyze an understand future changes in these events to evaluate and mitigate associated risks, as well as to improve predictability on sub-seasonal time-scales.

5 Conclusions

The co-occurrence of strong precipitation and wind events are hazards that can impact ecosystems and the lives of millions of people around the world causing severe property damage and fatalities. These events have shown high occurrences over the last decades along the western coast of the Iberian Peninsula, but studies focused on future changes in these phenomena at a regional scale are still scarce. This study presents the analysis of future changes in these compound events over the IP for the 21st century under two emissions scenarios (SSP2 − 4.5 and SSP5 − 8.5). The main conclusions of this study can be summarized as follows:

  • In a future climate, a general decreasing trend in the occurrence of concurrent extreme wind speed and precipitation events over most of the IP is predicted for all seasons, although some areas clearly show positive trends;

  • More compound events are expected during the winter for both scenarios in the western region, affecting a larger area in the moderate scenario;

  • The Galician territory may also experience an increase in the occurrence of these events in spring, with higher magnitudes under the high emission scenario;

  • The occurrence of these phenomena over some northeastern regions in summer is also projected across the century;

  • At the beginning of the century, spring, summer, and autumn seasons show a larger area experiencing compound events compared to the winter season.