Severe convective storms, in particular supercells, are occasionally responsible for a large number of property losses and damage in Spain. This paper aims to study the synoptic configurations and pre-convective environments in a dataset of 262 supercells during 2011–2020 in Spain. The events are grouped into supercells with hail (diameter larger than 5 cm) and without hail and the results are compared. ERA5 reanalysis is used to study the synoptic configurations and proximity atmospheric profiles related to the supercell events at the initial time. In addition, temperature, convective available potential energy, convective inhibition, lifting condensation level, level of free convection, height of freezing level, wind shear and storm-relative helicity are obtained for each event. Results show that supercells are more frequent on the Mediterranean coast during the warm season. Some of the variables analyzed present statistically significant differences between hail and non-hail events. In particular, supercells with hail are characterized by higher median values of most-unstable convective available potential energy than supercells without hail.
Thunderstorms and their associated phenomena (lightning, hail, wind or flash floods) have a great influence on human activities due to their destructive consequences (Martín et al., 2020; Taszarek et al., 2020a; Rodriguez and Bech, 2021). Europe is regularly threatened by severe thunderstorms (Dahl, 2006), causing considerable economic loss, having a social impact, and endangering aviation safety (Nisi et al., 2016; Mohr et al., 2017; Antonescu et al., 2017; Kunz et al., 2020; Chernokulsky et al., 2020; Gatzen et al., 2020). Thus, improving the knowledge on the genesis and life cycle of thunderstorms is a constant endeavor in the meteorological community.
Thunderstorm cells can be formed either in a discrete and isolated form or in large and organized systems, (e.g., squall lines). Three different thunderstorm types are defined by the US National Weather Service (NWS, 2019) based on their structure, organization and size: ordinary cell, multicell and supercell. Browning (1962) defines supercells as thunderstorms occurring in a significantly vertically sheared environment, which contains a deep and persistent mesocyclone, representing the most organized, severe and long-lasting form of isolated deep convection phenomena. These systems are linked to hail reports – including hail diameters larger than 5 cm – and EF2 tornadoes or higher (Duda and Gallus, 2010; Quirantes et al., 2014; Blair et al., 2017). Supercells are common phenomena in spring and summer (Brooks et al., 2019), and can be detected through Doppler radar data to confirm the associated mesocyclone (Blair et al., 2011; Kahraman et al., 2017). Moreover, ground-based or satellite lightning detection systems can be useful as supporting information (Bedka et al., 2018; Galanaki et al., 2018). However, local phenomena associated with these systems, such as hail, require observational reports to be confirmed. Since many of these events occur in unpopulated areas, the observed weather reports have a spatial bias toward the most populated areas (Groenemeijer et al., 2017; Edwards et al., 2018). In recent years, thunderstorm and severe weather reports have increased due to the accessibility of the general population to new technologies, especially thanks to smart phones and social networks. This has enabled an improvement in databases related to thunderstorms and their effects, with increasing availability of information on these events (Elmore et al., 2014; Krennert et al., 2018; Taszarek et al., 2020b). Nevertheless, a rigorous quality and validity control, through validated observational data (radar, satellite, etc.), should be applied for these severe-weather reports to be scientifically valid (Dotzek et al., 2009).
Domain and orography of the study area (m). White star indicates Maestrazgo area.
Mainly due to orography, lower convective available potential energy (CAPE)
and wind shear (WS), supercells in Europe tend to be less severe – return
periods of hail
Since severe supercells have caused substantial property damage and economic losses in recent years in Spain, with around 300 casualties due to thunderstorms from 1987 to 2020 (Consorcio de Compensación de Seguros, 2020), this study aims to provide a better understanding of the supercell synoptic and mesoscale environments. To this end, a set of events are selected from the supercell database of Martín et al. (2020), which are then categorized into two distinct groups, i.e., with hail diameter larger than 5 cm (SP-HAIL) and without hail (SP-NONHAIL). Manzato (2012) recorded hailstorms using hailpads to perform a hail climatology analysis in northeast Italy. Merino et al. (2013) also used hailpads data to study the synoptic and mesoscale configurations for hailstorms in southwestern Europe.
Herein, these systems are analyzed through their synoptic and mesoscale environments using ERA5 reanalysis (Hersbach et al., 2020). ERA5 is a state-of-the-art, high-resolution reanalysis, which has yielded successful results in studies related to severe local thunderstorm environments over North America (Coffer et al., 2020; Li et al., 2020; Taszarek et al., 2020a, b) and Europe (Taszarek et al., 2020a, b; Calvo-Sancho and Martín, 2021), to tornadic environments in the Iberian Peninsula (Rodríguez and Bech, 2021), and to microbursts (Bolgiani et al., 2020). On the other hand, severe storms have been forecast using numerical weather prediction models and alternative methods based on them, as seen in Gascón et al. (2015) survey. Their study selected the Showalter Index, dew point temperature at 850 hPa, storm relative helicity between 0 and 3 km, wind speed at 500 hPa and wet bulb zero to develop a logistic equation that generated the probability of severe thunderstorm development, obtaining a robust instrument to forecast severe thunderstorms in the Ebro Valley region.
This work is organized as follows. The database and methodology are described in Sect. 2. Section 3 presents the discussion of the main results related to synoptic and convective variables associated with SP-HAIL and SP-NONHAIL events. Finally, the main conclusions are summarized in Sect. 4.
The supercell sample used is selected from the Spanish Supercell Database
(Martín et al., 2020) for the period 2011–2020. This dataset is formed
by confirmed (i.e., Doppler radar images, hail greater than 5 cm reports,
tornadoes greater than EF2 or images of the event) and medium–high
confidence (detected in non-Doppler radar images but without direct
observation; see Fig. 4 in Martín et al., 2020) supercell events
through reports from volunteers and collaborators. This dataset comprises a total of 1758 supercells, of which 262 correspond to confirmed
supercells and 1495 to medium–high confidence supercells. It is worth noting
that even though the database covers all of Spain's national territory, there are no
events reported in the Canary Islands. The database defines the supercell
spatial life cycle through an ellipse in a Geographical Information System
and collects additional information associated with the events, e.g., hail
diameter, tornado intensity. In the current study, only the confirmed
supercells are selected. These are then categorized as SP-HAIL and
SP-NONHAIL according to the observation of hailstones with a diameter larger
than 5 cm or the lack of these reports. It should be noted that in this
study the Spanish Supercell Database was cross-matched with the European
Severe Weather Database (ESWD) and Notification System for Singular
Atmospheric Observations (SINOBAS). The validation results show that more than
80 % of the SP-HAIL events from the Spanish Supercell Database are
included in the ESWD and SINOBAS datasets. Finally, the initial formation
time (
The ERA5 reanalysis (Hersbach et al., 2020) is selected for the study of the
synoptic characteristics and the convective variables involved in the
supercell development. This is the fifth generation reanalysis created by the
European Centre for Medium-Range Weather Forecasts (ECMWF). It is provided
with a horizontal grid resolution of 0.25
Following the methodology of Calvo-Sancho and Martín (2021) and Gensini
et al. (2021), supercell soundings for SP-HAIL and SP-NONHAIL events are
built for
Synoptic pattern composites of both types of events are created to describe and compare the common large-scale features. The ERA5 atmospheric fields used to compute the composites are 500 and 300 hPa geopotential height, mean sea level pressure, dew point, wind direction and wind speed at 10 meter above sea level, 700–400 hPa integrated mean of omega vertical velocity, and 0–6 km WS. These atmospheric variables have been used in studies related to spatial patterns of hailstorms (Merino et al., 2013; Melcón et al., 2017), supercells (Gropp and Davenport, 2018) and other types of thunderstorms (Mora et al., 2015).
To characterize the convective environment, several thermodynamic and
kinematic variables are calculated using the thundeR R language package
(Taszarek et al., 2021) for each vertical profile. The selection of these
parameters (Table 1) is based on similar studies related to severe
convective storms in the United States and Europe (Rasmussen and Blanchard, 1998; Kaltenböck
et al., 2009; Westermayer et al., 2016; Rodríguez and Bech, 2018, 2021;
Taszarek et al., 2020a; Davenport, 2021). The 2 m temperature (T2M) and
dew point (DWPT) are selected. CAPE and convective inhibition (CIN) using
most-unstable (MU), mixed-layer (ML; averaged over 0–500 m above ground
level) and surface-based (SB) parcels are calculated using the virtual
temperature correction (Doswell and Rasmussen, 1994). The deep-layer bulk
WS over 0–6 km (WS06) and the effective bulk wind difference (EBWD;
limited to the layer in which CAPE
Description of the convective variables.
The application of the non-parametric Mann–Whitney test (Mann and Whitney,
1947) is used to establish statistical differences (at
Mann–Whitney test results of all variables analyzed
for SP-HAIL and SP-NONHAIL events at
The spatial and temporal distribution of supercells for both SP-HAIL and SP-NONHAIL formed in the Spanish mainland are first assessed. The main results relative to large-scale composites, and the thermodynamic and kinematic variables involved in supercell formation in the domain, are presented and discussed in the following two subsections.
The spatial distribution of the reported supercell episodes (Fig. 2a) shows that most of the events for both SP-HAIL and SP-NONHAIL took place in the eastern half of Spain. The Ebro Valley and the Mediterranean coastal area accumulate 79.9 % of the SP-NONHAIL and 88.3 % of SP-HAIL. This is consistent with lightning observations in Spain, as the eastern Iberian System area (white star in Fig. 1) has the highest density of lightning flash per year (Mora et al., 2015). This area favors convective initiation and supercell formation due to low-level convergence (northwesterly–southeasterly and south westerly–easterly winds), upper-level forcing for ascent, low–medium-level moist convection coming from the Mediterranean Sea and strong diurnal heating (Mora et al., 2015).
The temporal distribution of supercell events (Fig. 2b) matches the
warmest and stronger insolation months (July and August accumulate 53.3 %
of the SP-NONHAIL and 74.4 % of the SP-HAIL storms) in the study area,
since deep convection is a necessary condition for the formation of
supercells (Markowski and Richardson, 2011; Miglietta et al., 2017; Taszarek
et al., 2019). This is consistent with other studies on convective storms in
Europe that assess the higher thunderstorm frequency in summertime, when the
diurnal heating is stronger (Merino et al., 2013; Kotroni and Lagouvardos,
2016; Taszarek et al., 2018, 2019). The hourly distribution
of the supercells (Fig. 2c) shows a concentration of the events during the
late afternoon (summer local time is coordinated universal time plus 2 h, UTC
Synoptic pattern composites for the most relevant atmospheric variables in SP-HAIL and SP-NONHAIL events are presented in this subsection to describe and compare the large-scale characteristics.
The 500 hPa geopotential height (colored; dam), 300 hPa geopotential
height (blue contours; dam) and mean sea level pressure (black lines; hPa)
composites at
Non-substantial differences between SP-HAIL and SP-NONHAIL are found in the
mean sea level pressure (Fig. 3). However, the 500 hPa geopotential height
displays a SP-NONHAIL composite with a deeper trough and weaker geopotential
height gradient in comparison with SP-HAIL. A similar situation is shown by
the 300 hPa geopotential height. This atmospheric configuration promotes
weak WS in upper levels, which could be indicative of weaker convective
environments (Weisman and Klemp, 1982; Brooks et al., 2003; Taszarek et al.,
2017). Although there are differences between SP-HAIL and SP-NONHAIL, both
geopotential configurations promote upper-level positive vorticity advection
(not shown) and divergence over Spain, which favor a stronger upper-level
forcing (Markowski and Richardson, 2011). Values of 1.1 Pa
Mora et al. (2015) studied electrical thunderstorms in the northern plateau of Spain during 2000–2010, finding that 31 % of these thunderstorm episodes were linked to upper-level troughs. These episodes were characterized by strong baroclinic short waves and deep troughs at 500 hPa, which is a pattern very similar to the one shown in SP-HAIL and SP-NONHAIL composites in Fig. 3, respectively. Therefore, the results are in line with those of Mora et al. (2015), showing that supercell episodes in Spain are associated with troughs at upper and medium levels of the troposphere. Overall, the higher convective activity is located in the eastern part of Spain, corresponding to the right side of the troughs, with the thermal lows at the center of Spain.
The 2 m dew point temperature (contours;
One of the main features favoring deep moist convection is the moisture at lower and medium levels (Taszarek et al., 2019). Figure 4c depicts statistically significant differences in the DWPT values between SP-HAIL and SP-NONHAIL, with these differences being clearly stronger on the Spanish Mediterranean coast. Inland, a notable difference in DWPT is seen for SP-HAIL over the Ebro Valley (Fig. 4b), along with a stronger easterly wind flow. This would be a consequence of the geopotential and thermal low configuration described earlier, which induces humid air advection from the Mediterranean Sea. According to the DWPT climatology (not shown), the DWPT in the Ebro Valley and the Mediterranean coast is higher in August (when SP-HAIL is predominant; Fig. 2b) than in July. The convective processes are then supported by the favorable environment that promotes deep convection in those zones and are pushed by the south-westerly flows. This process is consistent with the results of the supercell observations for the period 2011–2020 (Fig. 2a).
The 700–400 hPa omega vertical velocity (contours; Pa s
The omega vertical velocity composites show statistically significant
differences between SP-HAIL and SP-NONHAIL at
Vertical WS promotes thunderstorm organization and its longevity. However,
excessive WS can be unfavorable to weak updrafts in environments of low
instability and, furthermore, can be disadvantageous to convection
initiation by increasing entrainment (Markowski and Richardson, 2011).
Figure 5 shows similar strong WS06 values (
As described in the convective variables methodology (Sect. 2.3), results
of the T2M, DWPT, CAPE, CIN, MLLCL, MLLFC, FZH and WS variables from the
ERA5 database are presented in this subsection. These results are shown as
violin plots, where the probability density distributions of each variable
can be seen as well as the differences between SP-HAIL and SP-NONHAIL
events at
The 90th percentile (based on MUCAPE values) composite soundings at
Based on the synoptic compositing methodology, schematic SP-HAIL and
SP-NONHAIL composite soundings at
T2M and DWPT distributions and boxplots for SP-HAIL and SP-NONHAIL
at
The distribution for T2M and DWPT shows differences between both types of
events, which are statistically significant (Table 2) for both variables.
Different distributions can be seen in Fig. 7. The T2M for the SP-HAIL
distribution depicts a lower variability and larger median value (Table 3)
than the corresponding SP-NONHAIL. The T2M maximum (minimum) for SP-HAIL is
33.0
Median values for each parameter analyzed for SP-HAIL and
SP-NONHAIL events at
As in Fig. 7, but for
CAPE is a common and useful forecast tool, in combination with vertical WS, for predicting supercells and hail (Kaltenböck et al., 2009; Merino
et al., 2013). The distributions in the parcel measurements of MUCAPE,
SBCAPE and MLCAPE are shown in Fig. 8a. Results show statistically
significant differences between the distributions for SP-HAIL and SP-NONHAIL
(Table 2). It is noteworthy that CAPE distributions follow positive skew
distributions for SP-NONHAIL events, with the SP-HAIL median values being notably
larger than SP-NONHAIL. However, the 25th percentile and median values for
SP-HAIL SBCAPE, 758 and 1231 J kg
It is well known that due to limited vertical resolution, reanalyses do not
represent capping inversions very well (Nevius and Evans, 2018; Coffer et
al., 2020; Taszarek et al., 2021). Here, CIN distributions in the parcel
measures MUCIN, SBCIN and MLCIN are displayed (Fig. 8b). The SP-HAIL and
SP-NONHAIL differences for MLCIN distributions are statistically significant
(Table 2). However, SBCIN and MUCIN differences are not statistically
significant, presenting negative skew distributions for both events. The
SP-HAIL events show a very weak inhibition layer (Fig. 8b) with an MLCIN
median of
As in Fig. 7, but for MLLCL, MLLFC, FZH and FZH_W.
A comparison of MLLCL, MLLFC, FZH and FZH_W percentiles and distributions between SP-HAIL and SP-NONHAIL is presented in Fig. 9. The MLLCL has been an important discriminator between tornadic and non-tornadic supercells in the United States (Rasmussen and Blanchard, 1998; Thompson et al., 2003). Taszarek et al. (2020b) compare the MLLCL in severe thunderstorms with hail greater than 5 cm in the United States and Europe obtaining great similarities between both continents. However, in Europe, the MLLCL tends to have much less variability in supercell events, and thus the skill of this indicator is limited (Kahraman et al., 2017; Taszarek et al., 2020b). The results shown here for Spain are in line with the previous conclusion, since there are no statistically significant differences (Table 2) for MLLCL between SP-HAIL and SP-NONHAIL events. Rodriguez and Bech (2018) also observed this low MLLCL variability between tornadic and non-tornadic thunderstorms in the Iberian Peninsula. Nevertheless, Púčik et al. (2015) obtained a MLLCL median value of 1000 m for severe hail thunderstorms in Central Europe, that matches those herein described for SP-HAIL events (Table 3). On the other hand, the MLLFC does not show significant differences between SP-HAIL and SP-NONHAIL (Table 2), with similar median values and variability. The MLLFC and MLLCL median values (Table 3) are also similar to those obtained by Taszarek et al. (2020b) for Europe. Another important factor for SP-HAIL and SP-NONHAIL events is the freezing level, with differences between both types of supercells being statistically significant for FZH and FZH_W (Table 2). Both FZH and FZH_W distributions for SP-NONHAIL events present higher variability than for SP-HAIL events (Fig. 9), with FZH and FZH_W median values for SP-HAIL being higher than for SP-NONHAIL (Table 3).
As in Fig. 7, but for WS06 and EBWD.
Several studies (Rasmussen and Blanchard, 1998; Púčik et al., 2015; Taszarek et al., 2019) suggest that the severity of the convective storms depend on the relationship between CAPE and WS. Furthermore, deep moist convection tends to develop more organized systems as the WS intensifies (Markowski and Richardson, 2011). WS between 0 and 6 km and EBWD distributions for SP-HAIL and SP-NONHAIL events are displayed in Fig. 10. As expected, the WS06 and EBWD values are higher for SP-HAIL than for SP-NONHAIL with statistically significant differences in EBWD (Table 2). The WS06 and EBWD median values for SP-HAIL (Table 3) match the results of Taszarek et al. (2020b) for severe convective storms with large hail in Europe; nevertheless, the values for the United States are higher than those presented here. Despite this, supercell events in the domain of study can be explained by the presence of several mountain ranges in the study area (Fig. 1) where the WS is enhanced by the interaction between the wind field and orography, with a similar mechanism to the one observed in the Alps (Kunz et al., 2018; Taszarek et al., 2020b). However, ERA5 is limited in reproducing this enhancement due to its horizontal resolution.
Storm-relative helicity (SRH) is a frequent parameter used for forecasting
supercells and tornadoes since it quantifies the streamwise vorticity that
can support the development of rotating updrafts in right- and left-moving
supercells (in this survey only the right-moving measure is used;
Davies-Jones et al., 1990; Bunkers et al., 2002). Higher SRH values are
usually related to the development of the mesocyclones and large hail
formation (Rasmussen and Blanchard, 1998; Thompson et al., 2003). However,
Hannesen et al. (1998), Kaltenböck (2004) and Gascón et al. (2015)
suggested that complex orography such as in the current study domain can
increase directional shear at low levels due to flow disturbance. Resulting
large SRH03 values would produce favorable but not necessary conditions for
severe thunderstorm formation. The SRH01 and SRH03 distributions (Fig. S2, Supplement) show no statistically significant differences (Table 2)
between SP-HAIL and SP-NONHAIL. However, the SRH01 and SRH03 median values
are slightly higher for SP-HAIL than for SP-NONHAIL (Table 3), showing
similar variability. There are remarkable SRH03 90th percentile values for
both groups of events, at 225 m
Here, the environments of SP-HAIL and SP-NONHAIL events are characterized and compared in Spain from 2011 to 2020. Different atmospheric variables are retrieved from the ERA5 reanalysis to obtain the synoptic patterns and composite soundings at the formation time of the supercells. Thermodynamic and kinematic parameters related to convective environments are also calculated and compared between SP-HAIL and SP-NONHAIL events.
The results yield several conclusions; the most important are listed below:
There are notable differences in the spatial and monthly distributions of supercells in Spain. The eastern half of Spain accumulates 79.9 % of the
SP-NONHAIL and 88.3 % of the SP-HAIL events. July and August accumulate
53.3 % of the SP-NONHAIL and 74.4 % of the SP-HAIL supercells. Most
events are initiated between 12:00 and 19:00 UTC with a peak at 15:00 UTC. The synoptic pattern composites show a deeper trough for SP-NONHAIL in
comparison with SP-HAIL composites at 500 hPa, with the largest height
gradients corresponding to SP-HAIL. Strong upper-level forcing is promoted
by vorticity advection and upper-level divergence. Surface humidity is
influenced by the 10 m winds, being higher for SP-HAIL. The conjunction
of these factors with wind convergences enables initiation of convection. The T2M and DWPT values are related to supercell monthly distributions
with higher values corresponding to the warm season and minimum values to
the cool season. Both variables are statistically different between SP-HAIL
and SP-NONHAIL, with values being larger for the first group. Environments of SP-HAIL events are characterized by approximately 2 times larger MUCAPE median values than for SP-NONHAIL events. Moreover,
higher CAPE, FZH and EBWD values and lower MLCIN results are found for
SP-HAIL compared with SP-NONHAIL. The differences for these parameters
between both events are statistically significant at Based on the ERA5 characterization results for SP-HAIL events in Spain, 75 % of the supercells present T2M
Finally, convective environments have been shown to be important factors to
supercell formation and development. Thus, although ERA5 resolution improves
previous reanalyses, more research is needed with high-resolution models,
facilitating the study of the interactions between large-scale and convection
processes in the genesis and development of hail supercell events. In
addition, further research on the influence of orography in the genesis
of supercells would be interesting.
ERA5 reanalysis is available from the Copernicus Climate Change Service Climate Data Store (
The supplement related to this article is available online at:
CCS, JDF, YM, MLM, PB and JJGA designed the study. CCS and JDF performed the analysis and wrote the first manuscript. PB, JJGA, MS, and MLM supervised and reviewed the manuscript. DSM and JIF provided computational and software support. All authors discussed the results and edited the manuscript.
The contact author has declared that none of the authors has any competing interests.
Publisher’s note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Carlos Calvo-Sancho and Javier Díaz-Fernández acknowledge the grant support from the Spanish Ministerio de Ciencia, Innovación y Universidades (FPI programs PRE2020-092343 and BES-2017-080025, respectively). Yago Martín acknowledges the grant support from the European Union (Marie Skłodowska-Curie Programs 101019424). We thank the three anonymous reviewers for their time and their detailed and constructive comments that helped to improve the presentation of our results.
This research has been supported by the Ministerio de Ciencia e Innovación (grant nos. PID2019‐105306RB‐I00, CGL2016-78702-C2-1-R, and CGL2016-78702-C2-2-R), the ECMWF Special Projects (grant nos. SPESMART and SPESVALE), the H2020 Marie Skłodowska-Curie Actions (grant no. 101019424), and the Ministerio de Ciencia e Innovación (grant nos. PRE2020-092343 and BES-2017-080025).
This paper was edited by Johannes Dahl and reviewed by three anonymous referees.