1. Introduction
The European changing climate and its variability have been extensively studied through observations and model outputs (Gudmundsson and Seneviratne 2016; Sousa et al. 2019; Brunner et al. 2020a; Coppola et al. 2021). Particularly, the Mediterranean basin is a hotspot for enhanced warming and precipitation decline, especially in summer (Zappa and Shepherd 2017; Tuel et al. 2021; Cos et al. 2022). In the Iberian Peninsula, studies have documented more frequent and intense climate extremes, including warm seasons, droughts, heatwaves, and heavy rainfall events (Turco and Llasat 2011; Calbó et al. 2012; Barrera-Escoda et al. 2014; Llasat et al. 2014).
Large-scale atmospheric circulation patterns (CPs), distant weather event connections, and local factors like soil moisture–atmosphere interactions (Fischer et al. 2007; Zampieri et al. 2009; Ríos-Cornejo et al. 2015; Miralles et al. 2019; Materia et al. 2022), land–sea contrasts (Lambert and Chiang 2007; Mora 2014; Brogli et al. 2019), and aerosol concentration (Nabat et al. 2014; Drugé et al. 2019; Wild et al. 2005) influence these extremes. Mean sea level pressure patterns, including the North Atlantic Oscillation (NAO) and atmospheric blocking, are well-documented mechanisms related to extreme temperature and rainfall episodes at various time scales (Folland et al. 2009; Jury et al. 2019). For instance, NAO modulates rainfall and temperature variability in Mediterranean areas exposed to westerlies during winter (Krichak and Alpert 2005; López-Moreno et al. 2011; Vicente-Serrano et al. 2011) and shows high correlations with precipitation and surface temperature in the eastern Mediterranean (positive and negative, respectively) during summer (Folland et al. 2009; Bladé et al. 2012).
However, regions like eastern Iberia, normally leeward from westerlies, exhibit lower correlations with the principal modes of variability of the NAO, Scandinavian Blocking, or Atlantic Cyclogenesis. Here, the western Mediterranean oscillation (WeMO)—defined by the difference in sea level pressure between the anticyclone over the Azores and the depression over Liguria (Italy)—was found to be the index most significantly correlated with annual, monthly, and daily precipitation on the littoral fringe of the region (Martin-Vide and Lopez-Bustins 2006; Gonzalez-Hidalgo et al. 2009). WeMO’s negative phase is associated with extreme torrential episodes in northeastern Iberia, especially in autumn months, whereas its most positive phase usually takes place in January, when the synoptic and sea temperature conditions of this time of the year inhibit torrential events (Lopez-Bustins et al. 2020). Precipitation variability is more influenced by synoptic and regional circulation and cyclogenesis in this Mediterranean region, rather than only the dominant NAO large-scale pattern in winter.
At a regional scale, frameworks based on CPs provide insights into regional climate drivers, supporting diagnostics, prediction, and climate change projections. These frameworks have been employed globally to understand the link between atmospheric circulation and regional-to-local climate at different spatiotemporal scales (Huth 2000; Muñoz et al. 2016; Moron et al. 2015, 2016; Stryhal and Huth 2019; Doss-Gollin et al. 2018; Faranda et al. 2020; Gadouali et al. 2020; Delgado-Torres et al. 2022; Pineda et al. 2023).
CPs have been used to characterize weather and climate surface conditions (Schuenemann and Cassano 2009; Espinoza et al. 2013; Tencer et al. 2016; Gibson et al. 2016), forecast localized extreme conditions at the subseasonal scale (Mastrantonas et al. 2022b,a; Krouma et al. 2022), attribution of observed trends and future changes (Nilsen et al. 2014; Fleig et al. 2015; Cahynová and Huth 2016; Espinoza et al. 2021; Olmo et al. 2023), and process-based model evaluation (Huth 2000; Bettolli and Penalba 2014; Muñoz et al. 2017; Fernández-Granja et al. 2021; Olmo et al. 2022).
In Europe, CP classifications from reanalysis and global climate models show changes in synoptic patterns (e.g., Herrera-Lormendez et al. 2022; Fernández-Granja et al. 2023), working with the predefined Lamb airflow weather types, such as increased frequencies of CPs with weak pressure gradients, mainly at the expense of decreasing frequencies of westerlies. In the Euro-Mediterranean region, different studies (e.g., Gadouali et al. 2020; Mastrantonas et al. 2021) have employed Empirical Orthogonal Functions (EOFs) and the k-means nonhierarchical clustering method to identify distinct structures associated with extreme precipitation, especially during winter and autumn.
Even so, few studies have assessed the temporal variability of atmospheric CPs and their influence on temperature and rainfall variability over the Iberian Peninsula. Some climate extremes like heavy rainfall events and heatwaves have been analyzed through CPs, clustering the days of interest (Gil-Guirado et al. 2022; Ventura et al. 2023). In northeastern Spain, Pérez-Zanón et al. (2018) showed that certain CPs with a low pressure system advecting warm and wet air from the Mediterranean Sea are conducive to severe rainfall in the region. These configurations indicate the presence of a warm and humid air mass at the surface, which is potentially unstable under the presence of a cutoff low at medium- and high-tropospheric levels.
A large body of literature (e.g., Muñoz et al. 2016, 2017; Moron et al. 2015; Doss-Gollin et al. 2018; Gadouali et al. 2020; Pineda et al. 2023) shows that a weather-typing approach provides a natural and unified framework to analyze cross time-scale interference of climate drivers, leading to prediction methodologies and climate model diagnostics. Climate variability across time scales can be described by the frequency and sequence of CPs, with forcings shifting the systems’ residence time (Muñoz et al. 2017). This approach helps diagnose a wide range of variables physically consistently, highlighting the causes of biases rather than focusing on the biases themselves.
Analyses focusing on a particular time scale tend to provide an incomplete picture of regional climate drivers (Thomson et al. 2018). However, up to the authors’ knowledge, a comprehensive CP study across multiple time scales—from daily to long-term trends—has not been explored in the Iberian Peninsula as in other regions (Muñoz et al. 2016, 2017; Moron et al. 2016; Espinoza et al. 2021). For instance, trends in weather types across western Europe have indicated major changes, especially in winter and summer, such as increasing temperatures (Riediger and Gratzki 2014; Cahynová and Huth 2016). Thus, given that many local or regional events are driven by large-scale circulation, a better comprehension of how much they affect the local climate can provide useful information for climatic variability and change characterization.
All of this reinforces that CP classifications are useful for climate diagnostics and prognostics and adaptable for specific climate zones and user needs. Increasing the number of clusters can reveal less frequent patterns relevant for describing precipitation and temperature variability in areas with low correlation to the main large-scale drivers, like the Mediterranean Iberian coast. Thus, evaluating sensitivity to different domain sizes and clusters helps understand the strengths and weaknesses of atmospheric circulation classification frameworks.
In this context, this study aims to develop a CP evaluation framework to assess temporal variability across multiple time scales and identify atmospheric circulation’s influence on the Iberian Peninsula’s regional climate. We construct a hierarchical classification of daily CPs and describe the main synoptic processes and their relationship with temperature and precipitation. We evaluate sensitivity to different methodological procedures, including spatial domains, number of clusters, and their variations across multiple time scales using observational and reanalysis data.
The paper is structured as follows: section 2 shows the reanalysis and observational data used here and presents the methodology, section 3 describes the main outcomes of this evaluation and, and finally, section 4 presents a discussion and some final remarks of this study.
2. Data and methodology
a. Reanalysis and observational data
For constructing the CP classifications, daily mean sea level pressure (SLP) data are employed, during the period 1950–2022, from the ECMWF ERA5 reanalysis (Hersbach et al. 2020). The selection of this variable is due to SLP fields being typically good indicators of humidity advection at the surface level in the Mediterranean. To obtain a wider picture of the synoptic environment in each CP, anomaly fields of midlevel geopotential height (500 hPa) and low-level winds (zonal and meridional wind components at 850 hPa) are also considered. ERA5 has hourly outputs provided at a 25-km spatial resolution, which are aggregated at a daily time scale for this analysis.
To link the occurrence of each synoptic structure or CP with different surface patterns, daily total precipitation (PR) and maximum and minimum temperatures (TX and TN, respectively) are employed. As the observational reference, we use the dataset provided by the Spanish Agencia Estatal de Meteorología (AEMET) in 1951–2020 with a spatial resolution of 5 km (AEMET-5), which is a gridded dataset of station data for daily PR, TX, and TN.
b. Clustering procedure
When it comes to synoptic classifications, many sensible aspects and methodological choices arise, including the use of raw or filtered data. In this regard, Moron et al. (2016) introduce an interesting discussion: while removing the mean annual cycle has the advantage of focusing on a specific time scale, emphasizing the impact of one or a few climatic phenomena, a potential side effect is the aliasing between time scales. This hinders the analysis of scale interactions and the modulation of the annual cycle (in its phase and amplitude). Hence, considering that we want to study the CPs impacting the Iberian Peninsula and their relationship with surface variables across multiple time scales, the SLP clustering is performed without any time filtering. This approach provides a large-scale overview of the atmospheric variability by considering, a priori, all time scales from daily to decadal (Moron et al. 2015, 2016; Muñoz et al. 2017; Espinoza et al. 2021; Olmo et al. 2022).
The SLP data are reduced through EOFs, a popular data reduction technique (Wilks 2019). In this work, 90% of the variance of the entire dataset is retained to optimize the clustering procedure by reducing data dimensionality via the EOFs and decreasing its noise. The generated SLP subspace is used as input data for a Ward hierarchical clustering, which minimizes the total within-cluster variance classifying the data utilizing a dendrogram (Ward 1963). The method consists of finding, at each step, the pair of clusters that leads to a minimum increase in the total within-cluster variance after merging. This increase follows a weighted squared distance between cluster centers. Here, the Euclidean distance is used to estimate the clusters and centroids. This statistical method has been used in previous studies over different domains with successful results (Espinoza et al. 2013; Castellano and DeGaetano 2016; Olmo et al. 2020).
c. Sensitivity analysis
Within the synoptic climatology community, it is well documented that the construction of classifications of CP needs an initial analysis of some climate data configuration choices and processes (Huth 2000; Muñoz et al. 2017; Hansen and Belušić 2021). The main interest of this work is to find a classification that successfully represents the variety of atmospheric states and manages to find differentiated surface structures associated with each CP over the Iberian Peninsula. Therefore, the clustering procedure is performed considering its sensitivity to 1) the spatial domain and 2) the number of clusters.
Three different domains of potential interest are explored (Fig. 1): D1, centered on the Iberian Peninsula (between 15°W–10°E and 32°–48°N); D2, covering the western Mediterranean and part of the North Atlantic Ocean (between 25°W–25°E and 30°–50°N); and D3, expanded over Europe, northern Africa, and a large portion of the North Atlantic Ocean (between 35°W–35°E and 20°–70°N).
Spatial domains employed for the classification of CPs (colored boxes D1, D2, and D3).
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
We provide robustness to this sensibility analysis by estimating the optimal number of clusters through a 1-yr-out cross validation of the PseudoF statistic. Once the EOFs are computed, 1 year of the period is left out and the remaining days are clustered with Ward’s method. Then, the PseudoF is estimated for 2–15 clusters, and this procedure is replicated as many times as years we have in the period of study (73 years). Finally, the mean and standard deviation of the PseudoF are computed for each number of clusters (Fig. 2). Note that other ways to validate the PseudoF metric have been analyzed—including a bootstrapping procedure and cross validations with different periods—and lead to consistent results (not shown).
PseudoF statistic changing the number of clusters for the classifications using different domains (rows). The black lines show the mean of all pseudoF estimations based on the 1-yr-out cross validation described in the main text, while the shaded color shows one standard deviation above and below the mean.
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
For D1, optimal PseudoF values are found for k = 3 and k = 5 (maximums of the curve). Results for D2 suggest the choice of k = 4 and k = 8, whereas for D3, no clear maximum is found (Fig. 2). In this case, the recommended option is to follow the “elbow” method, which consists of picking the elbow of the curve as a cutoff point, a point where introducing another cluster does not give a much better separation of the clusters. Thus, k = 4 and k = 5 arise as possible number of clusters for D3, since the curve is then decreasing faster. This selection allows us to intercompare the different classifications and evaluate if they are physically consistent, e.g., leading to similar physical mechanisms.
d. Link between circulation patterns and surface variables
Different aspects of the CPs and their influence on local rainfall and temperature across multiple time scales are explored. Once the classification of CPs is obtained, each day of the period of study is classified in one of the CPs, so climatological composites (averages) of the days within each CP can be constructed for different variables of interest. PR, TX, and TN anomalies are first estimated by removing their daily annual cycle, to filter seasonality and analyze the specific effect of the CPs occurrence in the surface variable. Then, the composites of anomalies are generated for each CP, replicating similar analysis performed in other studies (Gibson et al. 2016; Olmo et al. 2022, 2023). This is done for year-round (as the CP classification) and seasonal composites, to illustrate the influence of the CPs on an annual and seasonal basis.
The significance of the anomalies is tested based on Welch’s test, which compares the means between two groups not assuming that they have equal variances. The alternative hypothesis is that the true difference in means is not equal to zero (the difference between the climatological mean of a given variable and the conditional mean to the occurrence of a CP, as expressed by the corresponding anomalies values), with a confidence level of 95% (Wilks 2019).
Additionally, a cross-time-scale analysis is performed by studying the intraseasonal, seasonal, and interannual frequency of the CPs, following the recommendations by Muñoz et al. (2017). Particularly, their daily persistence and transition probabilities are calculated and tested for significance. This is done by means of a bootstrap technique. This procedure consists of randomly resampling 1000 times the vector that associates each day with a specific CP while preserving the probability of each CP and then using the 95th percentile of the estimated transition probabilities as a threshold.
The temporal evolution of CPs at a daily scale is illustrated through Klee diagrams, which consist of a matrix plot showing the daily evolution of the CPs during the period of study, and from which multiple time scales can be studied, including the daily transition probabilities (Muñoz et al. 2016, 2017). Furthermore, the monthly CP frequencies are calculated for different decades of the study period, so decadal variability can be observed.
The association between the synoptic patterns and the local conditions at the subseasonal scale is studied through a Kendall rank-based correlation analysis (95% confidence) (Espinoza et al. 2021). To this end, we work with detrended interannual time series of the CP frequencies and the surface variables on a monthly basis. For brevity, this is performed for the regional time series of three climatically homogeneous regions over Spain, taken from Martinez-Artigas et al. (2021).
The interannual seasonal frequencies of CPs are also analyzed and, to determine the presence of monotonic upward or downward trends in the CPs frequency series, the nonparametric Mann–Kendall test was implemented (Wilks 2019), at the 95% level of confidence. As part of this process, statistical relationships are computed between the synoptic patterns and the large-scale mechanisms involved in the NAO teleconnection. To this end, the NAO index is considered, which is based on the SLP difference between the subtropical (Azores) high and the subpolar low.
The positive phase of NAO reflects below-normal pressures over the northern–northeastern Atlantic—when the Azores High is stronger than usual—whereas its negative phase reflects an opposite pattern over these regions. The NAO monthly time series is provided by the U.S. National Oceanographic and Atmospheric Administration (NOAA). This index is aggregated seasonally and correlated with the interannual seasonal frequency of CPs using the linearly detrended time series and the rank-based τ Kendall’s correlation coefficient, with a confidence level of 95%. In addition, although NAO may not be the most suitable index to study the teleconnection during summertime, the summer NAO (SNAO) index was checked. SNAO is defined as the first EOF of the summertime extratropical North Atlantic SLP, which is characterized by a more northerly location and smaller spatial scale than its winter counterpart (Folland et al. 2009). The link with the CPs during the summer season leads to congruent results to the ones exposed in the present study (not shown). Thus, for this assessment, the same NAO index throughout the year is employed.
3. Results
a. Sensitivity to the domain choice
The seasonal cycles of the CPs according to different domains and choices of k are illustrated in Fig. 3. For the sake of conciseness, the patterns will be labeled as CPi-Dj.k, where i denotes the number of CP, j denotes the choice of domain, and k is the choice of the number of clusters. As an example, the classification into three groups based on the domain D1 is labeled as CPi-D1.3.
Seasonal cycles of the CPs as depicted by the ERA5 reanalysis in the different classifications (domains from D1 to D3, with two choices of k each). The x axis displays the 365 days of the year, while the y axis indicates the relative mean daily frequency of each CP.
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
From a statistical point of view, the most robust clustering is found for optimal values of the PseudoF statistic. In terms of temporal variability, the different classifications agree that the summer season is generally dominated by few CPs, while the winter and transition seasons need a larger number of structures explaining the span of all atmospheric states. Again, increasing the number of clusters allows for more seasonal and intraseasonal variability in the CP frequencies.
Based on the D1 classification, the clustering into three groups (k = 3) exhibits three patterns that are similarly distributed throughout the year, although CP2-D1.3 seems more frequent during autumn. CP3-D1.3 is the least present, exhibiting relative maximum frequencies during the transition seasons. This classification cannot identify seasonal variability in the synoptic configurations, probably due to a small number of clusters combined with the choice of a small domain of influence (as D1 only covers the Iberian Peninsula and its surrounding ocean).
When moving to more clusters (k = 5), the CPs show more seasonal and intraseasonal variability. In particular, summer is dominated by two main structures (CP1-D1.5 and CP5-D1.5), whereas CP2-D1.5 and CP3-D1.5 are the most frequent during winter, being almost null during the summer months. CP4-D1.5 presents similar frequencies throughout the year, except for the summer season.
Similarly, the D2 classification with k = 4 shows a predominance of one structure during the summer (CP2-D2.4), while in the transition and winter seasons, a larger number of CPs are present. CP1-D2.4 is more frequent during winter, whereas CP3-D2.4 and CP4-D2.4 exhibit two relative maxima in autumn and spring. Splitting the classification into eight groups (k = 8), two structures are present during summer (CP2-D2.8 and CP5-D2.8) and introduce larger variability in CP structures during the winter season, such as CP6-D2.8 and CP7-D2.8. In this classification, CP2-D2.8 and CP3-D2.8 are much less frequent, occurring in all seasons in summer but with low frequency, while CP8-D2.8 has maximums in the transition seasons.
Finally, the seasonality of CPs for D3.4 is similar to that of D2.4, being summer dominated by one synoptic structure (CP3-D3.4) and the winter season exhibiting more variability. Nonetheless, in the D3 classification, only one structure has maximum presence during the transition seasons, although slightly contributing to the total frequency (CP4-D3.4). When considering k = 5, CP3-D2.5 CP is present in all seasons but summer, with maximum values during the transition seasons but with a small contribution to the total frequency.
As regards the election of a clustering domain, the CPs resulting from the election of different geographic boxes do not show great disparities in the location of the main SLP centers of action (as will be shown later). In general, twinlike CPs are obtained in the case of synoptic structures describing the more frequent SLP patterns. The similarity is particularly evident when a similar number of clusters are retained. Those results, joined with the need of a description of Mediterranean-driven atmospheric flows—from the East—lead to the election of D2 as the reference geographic domain. However, results obtained with D1 and D3 domains are added as in the online supplemental material.
Therefore, the following assessments will be focused on one classification (D2.8), as an optimal PseudoF value is found for k = 8, which is a number of patterns similar to the one employed in the literature (Delgado-Torres et al. 2022). Note, however, that this decision is flexible and easy to update thanks to the hierarchical nature of the clustering method, and it can be adapted to the user’s needs to describe the synoptic environment with more or less detail within this workflow.
b. CP description and link with surface variables
The CPs for the D2.8 classification are presented in Fig. 4. For an integrated characterization of the atmospheric structure, the first row shows the SLP structures, the second row shows the midlevel geopotential height at 500 hPa (shaded colors) and low-level wind (vectors) anomalies at 850 hPa (estimated vs the mean annual cycle), whereas the following rows present the PR, TX, and TN anomalies, estimated by removing their seasonal cycle, thus looking specifically at the contribution of each CP to the surface anomalies. Even though the clustering is performed based on the D2 domain (Fig. 1), the atmospheric patterns are plotted over an extended domain (similar to D3) to have a wider picture of the synoptic environment. Moreover, the percentual frequencies of the CPs are presented in the left panel of Fig. 6 on an annual and seasonal basis.
Classification of CPs using the D2 domain with eight groups (k = 8): (first row) spatial patterns of SLP over an extended domain; (second row) Z500 anomalies (shaded) and low-level wind anomalies (850 hPa; in vectors); (third to fifth rows) surface anomalies of PR, TN, and TX over the Iberian Peninsula. Each column represents one CP. Only significant anomalies based on Welch’s test (at the 95% confidence level) are plotted.
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
Visual inspection of the SLP patterns indicates that the dominant structure during summer (CP5-D2.8)—being the most frequent pattern of the D2.8 classification, of about 37% of the days as shown in Fig. 5—exhibits the subtropical Azores high pressure system centered over the Atlantic Ocean; as it makes an incursion over Europe, it may enhance stable conditions in western Europe. Low pressure values are identified over Iceland and northern Africa. This atmospheric configuration is conducive to warm and dry air advection toward southern Europe by directing the low-level flow to higher latitudes. When looking at the year-round anomalies, they are close to zero since CP5 is the most frequent pattern, dominant in summer. However, when taking a closer look at the CP influence on a seasonal basis (see Figs. A1 and A2 in the supplemental material), the anomalies are higher, showing slightly positive temperature values in summer and negative anomalies for temperature and rainfall during winter (Fig. A1). This pattern is spatially correlated with the positive phase of the summer NAO during high summer, with positive geopotential height anomalies over the British Isles and, sometimes, potential instability in some Mediterranean regions (Bladé et al. 2012).
Heatmap of Kendall’s τ monthly correlations between each D2.8 CP (y axis) interannual frequency and the surface variables time series (PR, TN, and TX). Results are shown for specific regions within Spain (1–3) based on the regionalization by Martinez-Artigas et al. (2021). Significant correlations at the 95% for the linearly detrended time series of confidence are labeled in bold.
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
The other frequent CP during summertime is CP2-D2.8 (but also present in the transition seasons), in which the anticyclonic center in the Atlantic is located further north, and the Z500 and low-level wind anomalies are also intensified (Fig. 4, second row). This pattern is prone to producing cold and dry airmass advections from northeastern Europe over Iberia and, therefore, is related to dry anomalies and negative (positive) temperature anomalies in central and northeastern Spain (western and southwestern Spain).
During winter and early autumn, the anticyclonic center in CP6-D2.8 is reinforced. Compared to CP3-D2.8, the subpolar low near Iceland is much more intense in the SLP field, whereas the high pressure system this time is centered over southern Europe and the Mediterranean, but spanned over most of the domain of study, diminishing rainfall development as stable conditions prevail (Fig. 4). This configuration is associated with a northward shift of westerlies, increasing (decreasing) precipitation in northern (southern) Europe (Bladé et al. 2012; Hurrell and Deser 2015). This aligns with the extended center of positive geopotential height anomalies at 500 hPa in southern Europe and a reinforcement of zonal wind anomalies at 850 hPa in northern Europe (second row). The stability associated with such a configuration promotes radiative heating during the day and cooling at night, explaining the sign of TN and TX anomalies over Spain (negative and positive, respectively).
As mentioned in previous lines, CP6-D2.8 is probably related to the positive winter NAO. In contrast, its opposite pattern may be shown in CP4-D2.8, usually present in the transition and winter seasons (Fig. 5, left panel). In this case, westerlies are reinforced (weakened) in southern (northern) western Europe, as can be seen by the negative and positive u-wind component anomalies in the second row of Fig. 4. This synoptic configuration is associated with the negative phase of winter NAO, which implies cooling and wetting conditions over southern Europe except for the Iberian Mediterranean coast (Bladé et al. 2012).
It is worthwhile noting that the flow interaction with the topography is visible when looking at the link between the CPs and precipitation and temperature anomalies. CP4-D2.8 produces low-level westerly winds through the Iberian Peninsula that interact with the Spanish central plateau. Similar interactions between the flow and the topography can be seen in the other CPs. This increases rainfall on the windward side of the plateau due to orographic lifting and decreases rainfall on the leeward side due to topography-forced subsidence when the flow passes through the Iberian Peninsula. This effect is seen as a reduction in the intensity of the precipitation anomalies in the Spanish lowlands and a shift in TX anomalies when CP4-D2.8 is present (from negative to positive) due to subsidence-induced adiabatic heating (Fig. 4).
In the case of CP3-D2.8, which is present during winter and spring, the SLP pattern has a strong cyclonic system in high latitudes of the North Atlantic Ocean as in CP4-D2.8, but the high pressure system entering the continent is broken in the Mediterranean Sea, where a low SLP system is located. The midlevel circulation shows a negative center of anomalies in the North Atlantic Ocean as in CP4-D.8 but much less intense and constrained to higher latitudes, whereas a dipolar structure is found over subtropical latitudes (Fig. 4, second row). Consistent with the position of the anticyclone in the North Atlantic, an advection from the ocean is observed through northern Spain and France, which may be associated with the positive rainfall anomalies in the northeastern tip of the Iberian Peninsula.
CP7-D2.8 has a similar SLP structure to CP6-D2.8, but in this case, the high pressure system is located more to the west and south, on the Atlantic coast of the Iberian Peninsula and Northern Africa. CP7-D2.8 presents an intense Azores High centered in the Atlantic Ocean and low pressure values over northern Europe and the central Mediterranean. This can be associated with positive NAO flavors during transition months (Rousi et al. 2020). In this case, northern Iberia is still affected by westerlies, which causes an oceanic humid flow toward that region, whereas the rest of the peninsula is under drier and warmer than normal conditions. A dipolar structure is present at 500 hPa, with negative anomalies in central and eastern Europe and positive anomalies in the Atlantic coast of the peninsula.
CP8-D2.8 is a transitional CP—with the largest frequencies in spring and winter (Fig. 6)—related to a western Mediterranean cyclogenesis (strong negative Z500 anomalies are revealed over western Europe), which is depicted by the positive precipitation anomalies over many areas of the domain, but especially on the Mediterranean coast.
Frequency of CPs on an annual and seasonal basis. Percentages are estimated compared to (left) the total number of days by season. (right) Transition matrices for the D2.8 classification. Rows and columns indicate the number of CP occurring at the origin and end of the transition. Probabilities are estimated compared to the number of days of the starting CP in each transition, so each row in the panel sums 100%. Significant transitions according to a bootstrap procedure are highlighted with bold values.
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
CP1-D2.8—present in winter and autumn—has a strong high pressure system centered in eastern Europe. The Z500 anomalies suggest that this pattern could be related with a powerful blocking pattern leading to a retrograde circulation over the Mediterranean basin. The southern portion of this system has easterly winds reaching the Mediterranean coast of Spain, enhancing positive precipitation anomalies. This pattern has positive temperature anomalies, mostly found during autumn (see Fig. A2 in the supplemental material).
As presented in the supplemental material (see Figs. A3 and A4), the classifications of CPs based on the domains D1 and D3 depict physically consistent SLP structures to the ones described before, with the high pressure system over the Atlantic Ocean more or less intense and extended toward Europe and with lower pressure values over higher latitudes. The surface anomalies associated with these structures are generally congruent to the ones found for the D2 classification, indicating a physical agreement between classifications (not shown).
Complementing the previous assessment, the link between CPs and surface variables on a subseasonal scale was studied through a correlation analysis based on the interannual monthly series (Fig. 5), as described in section 2. For the sake of conciseness, a regionalization of Spain is taken from the work by Martinez-Artigas et al. (2021) and regional average time series of PR, TN, and TX are constructed. In Fig. 5, only one variable per region is shown for brevity, although the complete set of correlation heatmaps is presented in Fig. A5 as supplemental material.
Over region 1—covering most of the country except for the Mediterranean and northern Spanish coasts—CPs are significantly correlated with TX on a monthly basis. Consistent with the results from the surface anomalies (see Fig. 4, Figs. A1 and A2 in the supplemental material), CP6 shows high positive correlations in late winter and the transition seasons, particularly in February and March, indicating that TX is higher during these months when CP6 occurs, in line with the anticyclonic structure predominant over southern Europe. This signal is also present in CP1 but with a lower magnitude. On the contrary, CP8 and CP4 show general negative correlations throughout the year, even though only significant in spring and in autumn but only for CP8. In the case of region 2—covering the east coast of Spain and part of the inland territory—TN shows significant negative correlations with CP2 all year long, whereas an opposite link is found with CP4 but only significant from October to March. Over region 3, more patterns are significantly correlated with rainfall. As this is the rainiest area of the domain, like in CP7 and CP8, positive correlations are found with these structures but mostly significant in late spring and winter. CP3 also has a positive link with rainfall during winter months, with a maximum in January, while CP4 (CP1) shows significant positive (negative) correlations from October to March but, in the case of CP4 also in early summer (June).
It is worthwhile mentioning that a similar exercise was done using, instead of the CP monthly frequencies, a daily projection index to each CP measured by the Euclidean distance between each day and the cluster centroids (as employed in the clustering algorithm). This was explored to consider the spatial structure of each day and the similarities not only to the assigned CP but also to the other patterns. The primary outcomes of this correlation analysis align with the ones displayed here, indicating the subseasonal influence of the CPs on rainfall and temperature in the peninsula but through a slightly different approach (not shown).
In this way, the D2.8 year-round classification can detect an influence of the synoptic patterns in the regional rainfall and temperature monthly variability, making it a good indicator that could potentially help in subseasonal assessments of forecasts and climate simulations.
c. Temporal variability across time scales
Figure 6 (right) shows the transition probability matrix—between consecutive days—for the D2.8 classification, whereas the diagonal (from left to right) indicates the self-transition or persistence of each pattern. The probability is estimated compared to the frequency of the starting CP in the transition, so each row in the panel sums 100%. The diagonal presents significant probabilities (bold values); that is, the persistence probability for any CP (at a daily scale) is significantly higher than chance (at the one-sided 95% level of confidence, following a bootstrap procedure), consistent with previous studies in other parts of the world (e.g., Muñoz et al. 2016, 2017; Doss-Gollin et al. 2018; Gadouali et al. 2020, for southeastern South America, northeastern North America, and the Euro-Mediterranean region, respectively).
The persistence probability is maximum for CP5 (a probability of around 70%), in agreement with the discussion related to the seasonal cycle (see Fig. 3), indicating that this pattern dominates during the summer season. The strongest (but less frequent and nonsignificant) nonself transition is found from CP2 to CP5 (of 26.5%), since these two patterns are the predominant structures during summertime. CP2 has the third highest persistence in the CP classification, 55.4% of transitions, after CP6 with 56.9%.
Another combination of patterns that presents relatively high transition probabilities is CP8 to CP5 (of 26%), as CP8 has its maximum frequency during spring. Other smaller probability transitions, yet significant, are the CP3-CP4 and CP7-CP3 transitions—present during winter—whereas CP1-CP6, CP4-CP8, and CP6-CP7 also depict significant transitions up to 12%.
Notwithstanding, the transition probabilities for year-round statistics may not reflect the exact behavior of each season of the year. For this reason, the Klee diagrams are constructed and displayed in Fig. 7. The monthly frequencies of CPs separated by decades are also presented, allowing us to assess decadal variability and detect long-term changes in the CP frequencies.
(top) Klee diagrams (Muñoz et al. 2016) for the D2.8 classification. The x axis presents the years (1950–2022) and the y axis presents each day of the year. (bottom) The monthly frequencies of the CPs separated by decades (blue colors).
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
The analysis of the daily sequence of CPs throughout the year indicates that pattern transitions are typically faster in autumn and spring, occurring mostly daily, followed by the winter season. On the contrary, during the summer, the persistence of CP5 is noticeable, and it usually occurs before or after CP2, while transitioning to other structures like CP1 or CP6 usually takes longer. In agreement with the description above, CP2 persistence is much smaller than the one in CP5 but of about 7 days during summertime.
The Klee diagrams show us that the onset of the summer season often occurs with the CP8 and CP5 transitions mentioned before. In addition, CP1-CP6 and CP6-CP7 transitions typically occur during the winter season, but depending on the year, they can happen in early spring as well.
Furthermore, these diagrams also show that CP6 tends to be more frequent in the second half of the study period (during winter months), with the largest frequencies observed in the early and late 1990s during January and February and in the 2010s especially in November and December. This can also be observed in the decadal analysis (bottom panel of Fig. 7). Based on this assessment, it can be seen that CP frequency has less decadal variability during summer months in most of the patterns, except for the two structures that are predominant in that season (CP5 and CP2). In particular, CP5 has become more frequent in recent decades, with its maximum frequency evidencing a shift to late summer. On the other hand, CP7 and CP8 show similar behavior throughout the study period, with maximum frequencies in winter (December) and the transition seasons (April and November), respectively.
On an interannual basis, the evolution of the CPs frequency and their seasonal trends are shown in Fig. 8. In the D2.8 classification, CP5 is the most prevalent during the study period—in all seasons but winter—followed by CP2. However, as described in previous analyses, these structures are predominant in summertime, when these patterns appear almost exclusively, while the other structures rarely occur and show a reduced frequency in recent years. This is also reflected in the sign of the long-term trends, as CP5 depicts significant upward changes (95% level, using a Mann–Kendall test). The other structures show declines in their frequency, particularly CP6, although this pattern has a minimum frequency in the summer.
(top) Interannual relative frequency of CPs of the (colors) for the D2.8 classification separated by seasons of the year, expressed as a percentage. (bottom) The heatmaps of Kendall’s τ coefficient representing the seasonal trend for each case. Significant values at the 95% confidence level based on a Mann–Kendall test are marked with an asterisk.
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
On the contrary, in terms of long-term variability, CP6 has become more frequent in recent years during the winter season, in agreement with the analysis of the Klee diagrams of Fig. 7. This pattern exhibits the largest trend of all CPs, with significant upward changes in the expense of decreasing seasonal trends in most of the other synoptic structures. This season has the largest synoptic variability, as all patterns are present and show large fluctuations throughout the study period. In CP4 and CP5, a significant decline in their winter frequency is detected.
The transition seasons also show strong CP variability, although smaller than during winter, as CP5 is responsible for up to 40% of the total frequency. During the transition seasons, CP1 presents a significant increase on its frequency. This is probably at the expense of the decline on other CPs frequency, as CP5 during autumn and CP2 and CP4 during spring all exhibit significant downward changes.
In general terms, the seasonal-trend analysis indicates that the frequency of CPs changes during 1950–2022, with changes in all seasons but the highest trends presented mostly during winter.
As a final step, the association between the interannual evolution of the CP frequency and the NAO index is explored (Fig. 9). First, the NAO index shows changes toward a more present positive phase during winter (left). No clear changes are found in the rest of the seasons, although more negative values of the index seem to occur during summer and spring in recent years. When correlating those time series with the CP frequency, the strongest signals are found between NAO and CP4 with all significant and negative τ values for all seasons but particularly higher during winter. An opposite correlation of similar value is also detected with CP6 and of lower magnitude with CP7.
(top) NAO index interannual time series for each season of the year and the heatmap of Kendall’s τ correlations between the NAO time series and each D2.8 CP (y axis) interannual frequency for the different seasons of the year, separately (x axis). Significant correlations at the 95% for the linearly detrended time series of confidence are labeled in bold.
Citation: Journal of Climate 37, 21; 10.1175/JCLI-D-23-0735.1
Generally, winter is the season that presents the greatest association between the large scale and the CPs—as seen in the NAO analysis—with significant negative correlations also present for most of the patterns, but of lower magnitude than the ones previously mentioned. The other seasons of the year show more heterogeneous and nonsignificant associations, except for the negative correlations with CP4 (as discussed above) in spring and summer and the positive correlation between NAO and CP1 during spring.
4. Discussion and conclusions
Atmospheric circulation patterns (CPs) are closely linked to regional climate hazards in Southern Europe and the Mediterranean. Large-scale circulation variability considerably modulates the European climate, with phenomena like the North Atlantic Oscillation (NAO) affecting extreme weather events (Folland et al. 2009; Bladé et al. 2012; Jury et al. 2019; Mastrantonas et al. 2021). Most studies on the Iberian Peninsula focus on specific CPs, events, or seasons, but a comprehensive year-round classification of CPs, its variability across time scales, and the relationship with temperature and precipitation have not yet been explored.
This study aims to design a process-based framework to evaluate CPs’ temporal variability and their influence on the Iberian Peninsula’s regional climate across time scales. The motivation behind studying such classifications is that they can provide useful insights into regional climate drivers, aiding diagnostics, predictions, and climate change projections.
We analyze CPs using observational and reanalysis data, assessing sensitivity to different methodological procedures. The wide span of atmospheric states dominating the region is clustered through hierarchical clustering of daily mean sea level pressure (SLP) fields using all days in 1950–2022 from the ERA5 reanalysis.
Our sensitivity analysis shows minimal differences based on domain choice probably due to the homogeneous nature of SLP raw data. Increasing the number of clusters is beneficial for detailed regional descriptions, with an optimal classification of eight groups when using SLP over midlatitudes, covering the western Mediterranean to the North Atlantic Ocean next to the Iberian Peninsula, showing specific patterns associated with rainfall and temperature anomalies. The typical structures show an anticyclone over the Atlantic that often enters the continent—the subtropical Azores high, with varying intensity—while lower SLP is typically found over subpolar latitudes.
CPs display clear seasonality, with a leading and persistent dominant structure during summer (CP5), favoring dry and warm conditions, and CP2 being the second most frequent structure. In the transition and winter seasons, more CPs explain atmospheric variability. CP6 and CP7 are more frequent during winter, showing a strong contrast between midlatitudes and high latitudes with a strong anticyclone in Southern Europe and the North Atlantic Ocean, respectively, and an intense low pressure system in higher latitudes. Even though these patterns look similar in SLP, the systems’ differentiated position and spatial extent—as shown in the midlevel composites—lead to singular surface patterns, particularly in positive TN and PR anomalies in central and northern Spain, respectively. Moreover, CP4 and CP8 exhibit two relative maxima in autumn and spring, significantly promoting rainfall in large parts of the region.
The temporal variability of CPs is studied across time scales, showing that transitions between patterns are faster in autumn and spring, while summer is dominated by the two oceanic anticyclonic structures described before. Studying the transition probabilities from one CP to another can shed some light on forecast and prediction model biases due to the misrepresentation of synoptic conditions and their frequencies, as well as improve subseasonal and seasonal forecasts. Since the CPs are a synthetic version of the whole span of circulation fields, they may lead to higher predictive skill when used as predictors (Doss-Gollin et al. 2018).
CPs influence temperature and precipitation differently throughout the year, both on seasonal and intraseasonal scales, as shown by correlation values of varied intensity from month to month between the CP frequency and the surface variability. CP monthly frequency indicates decadal variability, particularly during winter and the transition seasons. Particularly, CP5 has shifted to higher frequencies in the late summer during recent decades.
The CPs in this work are seasonally modulated by large-scale structures, as depicted by the statistically significant link of CP4 and CP6 with NAO during winter. This result goes in line with the observed changes toward warmer and drier conditions over Southern Europe and the Mediterranean and previous literature on the role of NAO in the European climate (Bladé et al. 2012; Ríos-Cornejo et al. 2015; Cos et al. 2022).
Furthermore, the long-term changes of CPs are larger during winter, in agreement with Cahynová and Huth (2016), that detected significant trends in the frequency of winter circulation types reflecting the tendency toward NAO’s positive phase. Note, however, that changes in CP frequency are commonly not the main driver of regional climate change. The emergence of patterns, although dependent on the chaotic nature of the atmosphere, can also be modified in a global warming scenario (Faranda et al. 2020). Within-type variability—that is, modifications of the established CPs in their shape and intensity or processes at a smaller spatiotemporal scale—has been found to dominate the changes in different studies around the globe (Schuenemann and Cassano 2009; Herrera-Lormendez et al. 2022; Olmo et al. 2023).
Our findings show that combinations or sequences of daily CPs, rather than single patterns, impact temperature and rainfall across time scales, and those sequences tend to be different as in their frequency and persistence depending on the season, the type of the year (e.g., the phase of ENSO combined with other climate drivers changes the favored sequences in that year), the type of decade, etc. (Moron et al. 2016; Muñoz et al. 2016). Analyzing multiple time scales provides a better understanding of large-scale climate variability, crucial for assessing climate change and projections in hotspot regions like the Mediterranean basin (Cos et al. 2022).
Current uncertainties in climate projections are too large for regional decision-making or to generate a constrained GCM ensemble to apply downscaling techniques (McSweeney et al. 2015; Zhang et al. 2022; Ashfaq et al. 2022). Hence, determining the circulation characteristics and drivers that govern the regional climate becomes important. Once this knowledge is systematically generated, the potential to filter the model ensembles is opened up by selecting which models represent the relevant processes of the regional climate satisfactorily compared to observations (Palmer et al. 2023; Brunner et al. 2020b; Hegerl et al. 2021; Doblas-Reyes et al. 2013; Mahmood et al. 2022).
Overall, we highlight here the usefulness of CP classifications for assessing climate in the Euro-Mediterranean region, supporting evaluation and attribution frameworks for observational and model outputs in climate change contexts (Stryhal and Huth 2019; Faranda et al. 2020; Olmo et al. 2020). Moreover, the hierarchical nature of the clustering method allows for detailed analysis of regional circulation structures while preserving the original atmospheric classification (Espinoza et al. 2013).
This study lays the groundwork for a process-based evaluation framework to assess climate simulations, identifying strengths and limitations in representing the Iberian climate, similar to the approach proposed in previous studies (e.g., Muñoz et al. 2017; Olmo et al. 2022, 2023). The same set of weather types can diagnose a range of variables, shedding light on model biases and increasing confidence in future projections in a global warming scenario.
Acknowledgments.
This work was done within the CLIMCAT project: “Plan for comprehensive climate change information for Catalonia.” ÁGM was partially supported by the Grant RYC2021-034691-I, funded by MCIN/AEI/10.13039/501100011033 and the European Union NextGenerationEU/PRTR. The authors thank Margarida Samso Cabre, Saskia Loosveldt Tomas, and Pierre Antoine Bretonniere for the technical support (BSC).
Data availability statement.
The ECMWF ERA5 reanalysis can be found online at https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5. The AEMET-5 km dataset is available online at https://www.aemet.es/en/serviciosclimaticos/cambio_climat/datos_diarios. The NAO index is constructed by NOAA and available online at https://www.ncei.noaa.gov/access/monitoring/nao/.
REFERENCES
Ashfaq, M., D. Rastogi, J. Kitson, M. A. Abid, and S.-C. Kao, 2022: Evaluation of CMIP6 GCMs over the CONUS for downscaling studies. J. Geophys. Res. Atmos., 127, e2022JD036659, https://doi.org/10.1029/2022JD036659.
Barrera-Escoda, A., M. Gonçalves, D. Guerreiro, J. Cunillera, and J. M. Baldasano, 2014: Projections of temperature and precipitation extremes in the North Western Mediterranean basin by dynamical downscaling of climate scenarios at high resolution (1971–2050). Climatic Change, 122, 567–582, https://doi.org/10.1007/s10584-013-1027-6.
Bettolli, M. L., and O. C. Penalba, 2014: Synoptic sea level pressure patterns–daily rainfall relationship over the Argentine Pampas in a multi-model simulation. Meteor. Appl., 21, 376–383, https://doi.org/10.1002/met.1349.
Bladé, I., B. Liebmann, D. Fortuny, and G. J. van Oldenborgh, 2012: Observed and simulated impacts of the summer NAO in Europe: Implications for projected drying in the Mediterranean region. Climate Dyn., 39, 709–727, https://doi.org/10.1007/s00382-011-1195-x.
Brogli, R., N. Kröner, S. L. Sørland, D. Lüthi, and C. Schär, 2019: The role of Hadley circulation and lapse-rate changes for the future European summer climate. J. Climate, 32, 385–404, https://doi.org/10.1175/JCLI-D-18-0431.1.
Brunner, L., A. G. Pendergrass, F. Lehner, A. L. Merrifield, R. Lorenz, and R. Knutti, 2020a: Reduced global warming from CMIP6 projections when weighting models by performance and independence. Earth Syst. Dyn., 11, 995–1012, https://doi.org/10.5194/esd-11-995-2020.
Brunner, L., and Coauthors, 2020b: Comparing methods to constrain future European climate projections using a consistent framework. J. Climate, 33, 8671–8692, https://doi.org/10.1175/JCLI-D-19-0953.1.
Cahynová, M., and R. Huth, 2016: Atmospheric circulation influence on climatic trends in Europe: An analysis of circulation type classifications from the COST733 catalogue. Int. J. Climatol., 36, 2743–2760, https://doi.org/10.1002/joc.4003.
Calbó, J., A. Sanchez-Lorenzo, A. Barreda-Escoda, and J. Cunillera, 2012: Climate change projections for Catalonia (NE Iberian Peninsula). Part II: Integrating several methodologies. J. Mediterr. Meteor. Climatol., 9, 13–24, https://doi.org/10.3369/tethys.2012.9.02.
Caliński, T., and J. Harabasz, 1974: A dendrite method for cluster analysis. Commun. Stat., 3 (1), 1–27, https://doi.org/10.1080/03610927408827101.
Castellano, C. M., and A. T. DeGaetano, 2016: A multi-step approach for downscaling daily precipitation extremes from historical analogues. Int. J. Climatol., 36, 1797–1807, https://doi.org/10.1002/joc.4460.
Coppola, E., and Coauthors, 2021: Climate hazard indices projections based on CORDEX-CORE, CMIP5 and CMIP6 ensemble. Climate Dyn., 57, 1293–1383, https://doi.org/10.1007/s00382-021-05640-z.
Cos, J., F. Doblas-Reyes, M. Jury, R. Marcos, P.-A. Bretonnière, and M. Samsó, 2022: The Mediterranean climate change hotspot in the CMIP5 and CMIP6 projections. Earth Syst. Dyn., 13, 321–340, https://doi.org/10.5194/esd-13-321-2022.
Delgado-Torres, C., D. Verfaillie, E. Mohino, and M. G. Donat, 2022: Representation and annual to decadal predictability of Euro-Atlantic weather regimes in the CMIP6 version of the EC-Earth coupled climate model. J. Geophys. Res. Atmos., 127, e2022JD036673, https://doi.org/10.1029/2022JD036673.
Doblas-Reyes, F. J., and Coauthors, 2013: Initialized near-term regional climate change prediction. Nat. Commun., 4, 1715, https://doi.org/10.1038/ncomms2704.
Doss-Gollin, J., Á. G. Muñoz, S. J. Mason, and M. Pastén, 2018: Heavy rainfall in Paraguay during the 2015/16 austral summer: Causes and subseasonal-to-seasonal predictive skill. J. Climate, 31, 6669–6685, https://doi.org/10.1175/JCLI-D-17-0805.1.
Drugé, T., P. Nabat, M. Mallet, and S. Somot, 2019: Model simulation of ammonium and nitrate aerosols distribution in the Euro-Mediterranean region and their radiative and climatic effects over 1979–2016. Atmos. Chem. Phys., 19, 3707–3731, https://doi.org/10.5194/acp-19-3707-2019.
Espinoza, J. C., J. Ronchail, M. Lengaigne, N. Quispe, Y. Silva, M. Bettolli, G. Avalos, and A. Llacza, 2013: Revisiting wintertime cold air intrusions at the east of the Andes: Propagating features from subtropical Argentina to Peruvian Amazon and relationship with large-scale circulation patterns. Climate Dyn., 41, 1983–2002, https://doi.org/10.1007/s00382-012-1639-y.
Espinoza, J.-C., P. A. Arias, V. Moron, C. Junquas, H. Segura, J. P. Sierra-Pérez, S. Wongchuig, and T. Condom, 2021: Recent changes in the atmospheric circulation patterns during the dry-to-wet transition season in south tropical South America (1979–2020): Impacts on precipitation and fire season. J. Climate, 34, 9025–9042, https://doi.org/10.1175/JCLI-D-21-0303.1.
Faranda, D., M. Vrac, P. Yiou, A. Jézéquel, and S. Thao, 2020: Changes in future synoptic circulation patterns: Consequences for extreme event attribution. Geophys. Res. Lett., 47, e2020GL088002, https://doi.org/10.1029/2020GL088002.
Fernandez-Granja, J. A., A. Casanueva, J. Bedia, and J. Fernandez, 2021: Improved atmospheric circulation over Europe by the new generation of CMIP6 earth system models. Climate Dyn., 56, 3527–3540, https://doi.org/10.1007/s00382-021-05652-9.
Fernández-Granja, J. A., S. Brands, J. Bedia, A. Casanueva, and J. Fernández, 2023: Exploring the limits of the Jenkinson–Collison weather types classification scheme: A global assessment based on various reanalyses. Climate Dyn., 61, 1829–1845, https://doi.org/10.1007/s00382-022-06658-7.
Fischer, E. M., S. I. Seneviratne, P. L. Vidale, D. Lüthi, and C. Schär, 2007: Soil moisture–atmosphere interactions during the 2003 European summer heat wave. J. Climate, 20, 5081–5099, https://doi.org/10.1175/JCLI4288.1.
Fleig, A. K., L. M. Tallaksen, P. James, H. Hisdal, and K. Stahl, 2015: Attribution of European precipitation and temperature trends to changes in synoptic circulation. Hydrol. Earth Syst. Sci., 19, 3093–3107, https://doi.org/10.5194/hess-19-3093-2015.
Folland, C. K., J. Knight, H. W. Linderholm, D. Fereday, S. Ineson, and J. W. Hurrell, 2009: The summer North Atlantic Oscillation: Past, present, and future. J. Climate, 22, 1082–1103, https://doi.org/10.1175/2008JCLI2459.1.
Gadouali, F., N. Semane, Á. G. Muñoz, and M. Messouli, 2020: On the link between the Madden-Julian Oscillation, Euro-Mediterranean weather regimes, and Morocco winter rainfall. J. Geophys. Res. Atmos., 125, e2020JD032387, https://doi.org/10.1029/2020JD032387.
Gibson, P. B., S. E. Perkins-Kirkpatrick, and J. A. Renwick, 2016: Projected changes in synoptic weather patterns over New Zealand examined through self-organizing maps. Int. J. Climatol., 36, 3934–3948, https://doi.org/10.1002/joc.4604.
Gil-Guirado, S., A. Pérez-Morales, D. Pino, J. Carlos-Peña, and F. López-Martínez, 2022: Flood impact on the Spanish Mediterranean coast since 1960 based on the prevailing synoptic patterns. Sci. Total Environ., 807, 150777, https://doi.org/10.1016/j.scitotenv.2021.150777.
Gonzalez-Hidalgo, J. C., J.-A. Lopez-Bustins, P. Štepánek, J. Martin-Vide, and M. de Luis, 2009: Monthly precipitation trends on the Mediterranean fringe of the Iberian Peninsula during the second-half of the twentieth century (1951–2000). Int. J. Climatol., 29, 1415–1429, https://doi.org/10.1002/joc.1780.
Gudmundsson, L., and S. I. Seneviratne, 2016: Anthropogenic climate change affects meteorological drought risk in Europe. Environ. Res. Lett., 11, 044005, https://doi.org/10.1088/1748-9326/11/4/044005.
Hansen, F., and D. Belušić, 2021: Tailoring circulation type classification outcomes. Int. J. Climatol., 41, 6145–6161, https://doi.org/10.1002/joc.7171.
Hegerl, G. C., and Coauthors, 2021: Toward consistent observational constraints in climate predictions and projections. Front. Climate, 3, 678109, https://doi.org/10.3389/fclim.2021.678109.
Herrera-Lormendez, P., N. Mastrantonas, H. Douville, A. Hoy, and J. Matschullat, 2022: Synoptic circulation changes over central Europe from 1900 to 2100: Reanalyses and Coupled Model Intercomparison Project phase 6. Int. J. Climatol., 42, 4062–4077, https://doi.org/10.1002/joc.7481.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Hurrell, J. W., and C. Deser, 2015: Northern Hemisphere climate variability during winter: Looking back on the work of Felix Exner. Meteor. Z., 24, 113–118, https://doi.org/10.1127/metz/2015/0578.
Huth, R., 2000: A circulation classification scheme applicable in GCM studies. Theor. Appl. Climatol., 67 (1–2), 1–18, https://doi.org/10.1007/s007040070012.
Jury, M. W., S. Herrera, J. M. Gutiérrez, and D. Barriopedro, 2019: Blocking representation in the ERA-Interim driven EURO-CORDEX RCMs. Climate Dyn., 52, 3291–3306, https://doi.org/10.1007/s00382-018-4335-8.
Krichak, S. O., and P. Alpert, 2005: Signatures of the NAO in the atmospheric circulation during wet winter months over the Mediterranean region. Theor. Appl. Climatol., 82, 27–39, https://doi.org/10.1007/s00704-004-0119-7.
Krouma, M., P. Yiou, C. Déandreis, and S. Thao, 2022: Assessment of stochastic weather forecast of precipitation near European cities, based on analogs of circulation. Geosci. Model Dev., 15, 4941–4958, https://doi.org/10.5194/gmd-15-4941-2022.
Lambert, F. H., and J. C. H. Chiang, 2007: Control of land-ocean temperature contrast by ocean heat uptake. Geophys. Res. Lett., 34, L13704, https://doi.org/10.1029/2007GL029755.
Llasat, M. C., R. Marcos, M. Llasat-Botija, J. Gilabert, M. Turco, and P. Quintana-Seguí, 2014: Flash flood evolution in north-western Mediterranean. Atmos. Res., 149, 230–243, https://doi.org/10.1016/j.atmosres.2014.05.024.
Lopez-Bustins, J. A., L. Arbiol-Roca, J. Martin-Vide, A. Barrera-Escoda, and M. Prohom, 2020: Intra-annual variability of the Western Mediterranean Oscillation (WeMO) and occurrence of extreme torrential precipitation in Catalonia (NE Iberia). Nat. Hazards Earth Syst. Sci., 20, 2483–2501, https://doi.org/10.5194/nhess-20-2483-2020.
López-Moreno, J. I., S. M. Vicente-Serrano, E. Morán-Tejeda, J. Lorenzo-Lacruz, A. Kenawy, and M. Beniston, 2011: Effects of the North Atlantic Oscillation (NAO) on combined temperature and precipitation winter modes in the Mediterranean mountains: Observed relationships and projections for the 21st century. Global Planet. Change, 77, 62–76, https://doi.org/10.1016/j.gloplacha.2011.03.003.
Mahmood, R., M. G. Donat, P. Ortega, F. J. Doblas-Reyes, C. Delgado-Torres, M. Samsó, and P.-A. Bretonnière, 2022: Constraining low-frequency variability in climate projections to predict climate on decadal to multi-decadal timescales – A poor man’s initialized prediction system. Earth Syst. Dyn., 13, 1437–1450, https://doi.org/10.5194/esd-13-1437-2022.
Martinez-Artigas, J., M. Lemus-Canovas, and J. A. Lopez-Bustins, 2021: Precipitation in peninsular Spain: Influence of teleconnection indices and spatial regionalisation. Int. J. Climatol., 41, E1320–E1335, https://doi.org/10.1002/joc.6770.
Martin-Vide, J., and J.-A. Lopez-Bustins, 2006: The Western Mediterranean Oscillation and rainfall in the Iberian Peninsula. Int. J. Climatol., 26, 1455–1475, https://doi.org/10.1002/joc.1388.
Mastrantonas, N., P. Herrera-Lormendez, L. Magnusson, F. Pappenberger, and J. Matschullat, 2021: Extreme precipitation events in the Mediterranean: Spatiotemporal characteristics and connection to large-scale atmospheric flow patterns. Int. J. Climatol., 41, 2710–2728, https://doi.org/10.1002/joc.6985.
Mastrantonas, N., L. Furnari, L. Magnusson, A. Senatore, G. Mendicino, F. Pappenberger, and J. Matschullat, 2022a: Forecasting extreme precipitation in the central Mediterranean: Changes in predictors’ strength with prediction lead time. Meteor. Appl., 29, e2101, https://doi.org/10.1002/met.2101.
Mastrantonas, N., L. Magnusson, F. Pappenberger, and J. Matschullat, 2022b: What do large-scale patterns teach us about extreme precipitation over the Mediterranean at medium- and extended-range forecasts? Quart. J. Roy. Meteor. Soc., 148, 875–890, https://doi.org/10.1002/qj.4236.
Materia, S., and Coauthors, 2022: Summer temperature response to extreme soil water conditions in the Mediterranean transitional climate regime. Climate Dyn., 58, 1943–1963, https://doi.org/10.1007/s00382-021-05815-8.
McSweeney, C. F., R. G. Jones, R. W. Lee, and D. P. Rowell, 2015: Selecting CMIP5 GCMs for downscaling over multiple regions. Climate Dyn., 44, 3237–3260, https://doi.org/10.1007/s00382-014-2418-8.
Miralles, D. G., P. Gentine, S. I. Seneviratne, and A. J. Teuling, 2019: Land–atmospheric feedbacks during droughts and heatwaves: State of the science and current challenges. Ann. N. Y. Acad. Sci., 1436, 19–35, https://doi.org/10.1111/nyas.13912.
Mora, C., 2014: Sea breeze and local weather conditions in a coastal Mediterranean mountain setting. Weather, 69, 184–190, https://doi.org/10.1002/wea.2258.
Moron, V., A. W. Robertson, J.-H. Qian, and M. Ghil, 2015: Weather types across the Maritime Continent: From the diurnal cycle to interannual variations. Front. Environ. Sci., 2, 65, https://doi.org/10.3389/fenvs.2014.00065.
Moron, V., I. Gouirand, and M. Taylor, 2016: Weather types across the Caribbean basin and their relationship with rainfall and sea surface temperature. Climate Dyn., 47, 601–621, https://doi.org/10.1007/s00382-015-2858-9.
Muñoz, Á. G., L. Goddard, S. J. Mason, and A. W. Robertson, 2016: Cross-time scale interactions and rainfall extreme events in southeastern South America for the austral summer. Part II: Predictive skill. J. Climate, 29, 5915–5934, https://doi.org/10.1175/JCLI-D-15-0699.1.
Muñoz, Á. G., X. Yang, G. A. Vecchi, A. W. Robertson, and W. F. Cooke, 2017: A weather-type-based cross-time-scale diagnostic framework for coupled circulation models. J. Climate, 30, 8951–8972, https://doi.org/10.1175/JCLI-D-17-0115.1.
Nabat, P., S. Somot, M. Mallet, A. Sanchez-Lorenzo, and M. Wild, 2014: Contribution of anthropogenic sulfate aerosols to the changing Euro-Mediterranean climate since 1980. Geophys. Res. Lett., 41, 5605–5611, https://doi.org/10.1002/2014GL060798.
Nilsen, A., I. ane Fleig, L. Tallaksen, and H. Hisdal, 2014: Recent trends in monthly temperature and precipitation patterns in Europe. IAHS Publ., 363, 132–137.
Olmo, M., M. L. Bettolli, and M. Rusticucci, 2020: Atmospheric circulation influence on temperature and precipitation individual and compound daily extreme events: Spatial variability and trends over southern South America. Wea. Climate Extremes, 29, 100267, https://doi.org/10.1016/j.wace.2020.100267.
Olmo, M. E., J.-C. Espinoza, M. L. Bettolli, J. P. Sierra, C. Junquas, P. A. Arias, V. Moron, and R. Balmaceda-Huarte, 2022: Circulation patterns and associated rainfall over south tropical South America: GCMs evaluation during the dry-to-wet transition season. J. Geophys. Res. Atmos., 127, e2022JD036468, https://doi.org/10.1029/2022JD036468.
Olmo, M. E., M. L. Bettolli, and R. Balmaceda-Huarte, 2023: Model uncertainty in synoptic circulation patterns and precipitation changes in southern South America using CMIP5 and CMIP6 models. Climatic Change, 176, 171, https://doi.org/10.1007/s10584-023-03647-5.
Palmer, T. E., C. F. McSweeney, B. B. B. Booth, M. D. K. Priestley, P. Davini, L. Brunner, L. Borchert, and M. B. Menary, 2023: Performance-based sub-selection of CMIP6 models for impact assessments in Europe. Earth Syst. Dyn., 14, 457–483, https://doi.org/10.5194/esd-14-457-2023.
Pérez-Zanón, N., M. C. Casas-Castillo, J. C. Peña, M. Aran, R. Rodríguez-Solà, A. Redaño, and G. Solé, 2018: Analysis of synoptic patterns in relationship with severe rainfall events in the Ebre Observatory (Catalonia). Acta Geophys., 66, 405–414, https://doi.org/10.1007/s11600-018-0126-1.
Pineda, L. E., J. A. Changoluisa, and Á. G. Muñoz, 2023: Early onset of heavy rainfall on the northern coast of Ecuador in the aftermath of El Niño 2015/2016. Front. Earth Sci., 11, 1027609, https://doi.org/10.3389/feart.2023.1027609.
Riediger, U., and A. Gratzki, 2014: Future weather types and their influence on mean and extreme climate indices for precipitation and temperature in Central Europe. Meteor. Z., 23, 231–252, https://doi.org/10.1127/0941-2948/2014/0519.
Ríos-Cornejo, D., Á. Penas, R. Álvarez-Esteban, and S. del Río, 2015: Links between teleconnection patterns and precipitation in Spain. Atmos. Res., 156, 14–28, https://doi.org/10.1016/j.atmosres.2014.12.012.
Rousi, E., H. W. Rust, U. Ulbrich, and C. Anagnostopoulou, 2020: Implications of winter NAO flavors on present and future European climate. Climate, 8, 13, https://doi.org/10.3390/cli8010013.
Schuenemann, K. C., and J. J. Cassano, 2009: Changes in synoptic weather patterns and Greenland precipitation in the 20th and 21st centuries: 1. Evaluation of late 20th century simulations from IPCC models. J. Geophys. Res., 114, D20113, https://doi.org/10.1029/2009JD011705.
Sousa, P. M., D. Barriopedro, A. M. Ramos, R. García-Herrera, F. Espírito-Santo, and R. M. Trigo, 2019: Saharan air intrusions as a relevant mechanism for Iberian heatwaves: The record breaking events of August 2018 and June 2019. Wea. Climate Extremes, 26, 100224, https://doi.org/10.1016/j.wace.2019.100224.
Stryhal, J., and R. Huth, 2019: Classifications of winter atmospheric circulation patterns: Validation of CMIP5 GCMs over Europe and the North Atlantic. Climate Dyn., 52, 3575–3598, https://doi.org/10.1007/s00382-018-4344-7.
Tencer, B., M. L. Bettolli, and M. Rusticucci, 2016: Compound temperature and precipitation extreme events in southern South America: Associated atmospheric circulation and simulations by a multi-RCM ensemble. Climate Res., 68, 183–199, https://doi.org/10.3354/cr01396.
Thomson, M. C., Á. G. Muñoz, R. Cousin, and J. Shumake-Guillemot, 2018: Climate drivers of vector-borne diseases in Africa and their relevance to control programmes. Infect. Dis. Poverty, 7, 81, https://doi.org/10.1186/s40249-018-0460-1.
Tuel, A., S. Kang, and E. A. B. Eltahir, 2021: Understanding climate change over the southwestern Mediterranean using high-resolution simulations. Climate Dyn., 56, 985–1001, https://doi.org/10.1007/s00382-020-05516-8.
Turco, M., and M. C. Llasat, 2011: Trends in indices of daily precipitation extremes in Catalonia (NE Spain), 1951–2003. Nat. Hazards Earth Syst. Sci., 11, 3213–3226, https://doi.org/10.5194/nhess-11-3213-2011.
Ventura, S., J. R. Miró, J. C. Peña, and G. Villalba, 2023: Analysis of synoptic weather patterns of heatwave events. Climate Dyn., 61, 4679–4702, https://doi.org/10.1007/s00382-023-06828-1.
Vicente-Serrano, S. M., and Coauthors, 2011: The NAO impact on droughts in the Mediterranean region. Hydrological, Socioeconomic and Ecological Impacts of the North Atlantic Oscillation in the Mediterranean Region, S. Vicente-Serrano and R. Trigo, Eds., Springer, 23–40, https://doi.org/10.1007/978-94-007-1372-7_3.
Ward, J. H., Jr., 1963: Hierarchical grouping to optimize an objective function. J. Amer. Stat. Assoc., 58, 236–244, https://doi.org/10.1080/01621459.1963.10500845.
Wild, M., and Coauthors, 2005: From dimming to brightening: Decadal changes in solar radiation at Earth’s surface. Science, 308, 847–850, https://doi.org/10.1126/science.1103215.
Wilks, D. S., 2019: Statistical Methods in the Atmospheric Sciences. Elsevier, 840 pp.
Zampieri, M., F. D’Andrea, R. Vautard, P. Ciais, N. de Noblet-Ducoudré, and P. Yiou, 2009: Hot European summers and the role of soil moisture in the propagation of Mediterranean drought. J. Climate, 22, 4747–4758, https://doi.org/10.1175/2009JCLI2568.1.
Zappa, G., and T. G. Shepherd, 2017: Storylines of atmospheric circulation change for European regional climate impact assessment. J. Climate, 30, 6561–6577, https://doi.org/10.1175/JCLI-D-16-0807.1.
Zhang, M.-Z., Z. Xu, Y. Han, and W. Guo, 2022: Evaluation of CMIP6 models toward dynamical downscaling over 14 CORDEX domains. Climate Dyn, 62, 4475–4489, https://doi.org/10.1007/s00382-022-06355-5.