On the predictability of the giant hail event in Catalonia

In the late afternoon of 30 August 2022, an extraordinary supercell storm formed over the eastern Pyrenees. The storm quickly started to move to the right of the mean wind as it entered the district of Girona in Catalonia, producing a swath of very large hail (≥ 5 cm) between Esponnellá at 16:50 UTC and Tamariu at 17:34 UTC, after which it moved over the sea.

Sandwich satellite image of the hailstorm over Northeast Catalonia on 30 August 2022 at 17:30 UTC, and surface station measurements of the wind, temperature (red), and dew point (green).

Multiple reports of hailstones larger than 10 cm in diameter were collected with the largest stones estimated to be 12 cm. Based on some videos, the hail fall was relatively dense for stones of that size. The impacts of the storm were high: Besides serious damage to roofs and cars, 67 injuries and even one fatality resulted in the town of La Bisbal d’Empordà. 28 people had to be taken to the hospital, including one serious head injury. Based on our study on hail impacts across Europe, this was the first direct hail fatality in Europe since 1997. Furthermore, the number of injuries ranks as the third highest recorded in the ESWD for large hail events.

ESWD reports submitted for the hailstorm. Photo source:https://mobile.twitter.com/Supercelulas

Given the societal and economic impact of this hailstorm and the fact that the giant hail has already been reported 18 times in Europe this year, we look at the predictability of this particular event from the perspective of the large-scale, pre-convective environment addressing these two questions:

  1. How likely was the convective initiation?
  2. How likely were the initiated storms to produce giant hail?

Limiting ourselves to the large-scale environment we do not address two important sources of data: high-resolution convection-allowing models and nowcasting data, such as products based on radar

How likely was the convective initiation?

With an abundance of low-level moisture along the coastline and steep mid-tropospheric lapse rates advected from the interior of Iberia, high convective available potential energy (CAPE) was present to support the development of severe thunderstorms. 

With this potential being present, the most important question was if a trigger strong enough to set this energy free would be available in this environment. The synoptic-scale lift was forecast only over extreme northeastern Spain and in the upper troposphere near 300 hPa, but not at lower levels. With the absence of fronts or other large-scale air-mass boundaries, the mesoscale lift had to come from an upslope flow of maritime air against the high terrain.

Locations of where large-scale upward (red contours) and downward (blue contours) vertical motion is to be expected based on Q-vectors at the 300 hPa pressure level.

Convective initiation across Spain was complicated by a substantial amount of convective inhibition (CIN), negative energy to be overcome before a storm can form, especially over the southern part of Iberia. Near the eastern coastline, CIN rapidly increased from the mountains towards the coastline, restricting the ability of the storms to tap into the moisture- and CAPE-rich air mass. The largest area of relatively low CIN (< 50 J/kg) existed over far northeast Iberia, where the supercell formed. 

Combining the lowest CIN and the presence of at least some synoptic-scale lift over far northeast Spain, with hindsight it is possible to pinpoint this area as one with the highest probability of storm formation. Severe hailstorms are often isolated cells, rather than storms which are embedded in a larger convective system. On the 30 of August, the absence of widespread mesoscale lift and the presence of some CIN in the environment probably helped to limit the number of storms that formed to the one storm that produced the giant hail.

While it is easy to retrospectively explain the isolated nature of the storms on this day, beforehand it was not possible to state with certainty what will be the exact track of the storms or whether there will be three or no storms at all.

Forecast of the ECMWF IFS model of CAPE (colors) and areas of convective inhibition (whitish shading) at 15:00 UTC.

How likely were the storms to produce giant hail?

The supercell moved into the environment that has been found to be very conducive to severe weather, featuring high CAPE and strong vertical wind shear. Considering the model-simulated Skew-T and the surface observations from the area (temperature of 29, dewpoint of 23°C, and 5 m/s SSE wind), very large hail production was supported by:

  1. High values of CAPE (MLCAPE ≈ 4000 J/kg) with large amounts of CAPE found in the temperature zone < -10°C
  2. Vertical wind shear supportive of supercells. Very large hail occurs almost exclusively with this type of convection. Furthermore, strong shear resulted in a strong inflow into the storm. Based on the simulated hodograph, observed surface wind and observed storm motion, the surface inflow into the storm was almost 20 m/s. Strong inflow supports wide updrafts and wide updrafts lead to long hail embryo residence times in the favorable growth zone.
  3. Unidirectional vertical wind shear (i.e. straight hodograph). Straight hodographs have been found to be more conducive to large hail growth than curved hodographs. 

Both low-level shear and storm-relative helicity were quite weak in this case. Unlike for tornadoes, high values of these parameters are not necessary for very large or even giant hail.

ICON-EU forecast of CAPE, 0-6 km bulk shear at 15 UTC with the model Skew-T and hodograph for the location represented by the red cross. The red star within the hodograph marks the observed surface wind and the blue star the observed storm motion. The black arrow is the surface storm-relative wind (i.e. inflow at that level).
ECMWF and GFS forecast soundings for 15 UTC.
Note the underestimation of surface dewpoint.

Supercell also profited from being the only storm around with no disruption to its inflow and updraft. Its deviant motion to the right was also more pronounced than anticipated by the Bunker’s ID method, suggesting the presence of a strong mesocyclone.

Was it possible to make a confident forecast of the storm producing hail reaching 10 cm? Such a forecast would be useful, as 10+ cm hailstones have a higher probability of causing both damage and injuries compared to 5+ cm hailstones. However, hail diameter doesn’t linearly increase with increasing CAPE and shear. For example, increasing CAPE may even limit the large hail production beyond some point. There are likely other factors that influence the trajectory of hail embryos through the updraft and their residence time in a zone of abundant super-cooled water droplets. Some of these are covered in a lecture by Matthew Kumjian. Testing these factors against a large sample of very large or giant hail cases will perhaps bring us even closer to confident forecasts of such devastating hailstorms.

In conclusion, the combination of very favorable large-scale conditions for hail with isolated convective initiation resulted in a perfect scenario for a damaging hailstorm. Such knowledge provided a good chance to correctly nowcast the event once the storm entered a supercell stage.

The derecho and hailstorms of 18 August 2022

On 18 August, severe storms occurred in a swath from Menorca through Corsica, northern Italy, Slovenia, Austria, and southern Czechia. The event featured giant hail up to 11 cm in diameter and extremely severe wind gusts up to 62.2 m/s. In total, 12 people died and 106 people were injured by wind and hail. All fatalities, and most of the injuries, were caused by a long-lived convective system that produced a swath of severe to extremely severe wind gusts. The windstorm event can easily be classified as a derecho, a particularly long-lived and severe convective windstorm as the official criteria for a derecho were clearly met: its damage path exceeded 1000 km in length and wind gusts > 32 m/s were measured in Corsica, Italy, and Austria along the path of the storm.

Temporal evolution of ESWD reports for the event between 00 and 24 UTC.

This article describes in more detail the evolution of the event and its relation to the environment. 

Temporal evolution of the derecho
The first severe wind gust report was observed at 06:15 UTC in western Corsica. The last severe wind gust report was reported at 17:45 UTC from Czechia. Severe weather reports indicate that the derecho lasted at least 11 hours and 30 minutes. However, it is almost certain that the severe wind gusts occurred already sooner than 06:15 UTC. The severe wind gusts probably occurred as soon as 05:30 UTC, when the bow echo began to form. This means that the severe wind production in the storm lasted likely for more than 12 hours.

17 August

21:00 – 00:00 UTC: Storms initiated over the Balearic Sea. One of the storms became a right-moving supercell and intensified.

18 August

00:00 – 03:00 UTC: The right-moving severe storm grew in size and impacted Menorca. Heavy rainfall and severe wind gusts were reported. Further storms formed to the north.

Airmass RGB showing a severe storm over Menorca 
and a pronounced trough over western Spain and the Balearic Sea. 
Courtesy: EUMETSAT.

03:00 – 05:00 UTC: The storms continued to grow upscale and became a squall line. In the same period, a supercell ahead of the main storm system impacted Lavagna and Sestri Levante in Liguria, Italy with wind-driven hail up to 5 cm in diameter. 22 people were injured in this event. Cars, roofs, and facades of houses were badly damaged by hail.

Wind-driven hail damage to the house facade in Sestri Levante, Italy. Photo by Gherardo Ghotti.

05:00 – 06:00 UTC: The squall line transitioned into a bow shape just ahead of reaching Corsica with an increase in lightning activity and rapid acceleration towards the island’s western coastline. 

06:00 – 07:00 UTC: The bow echo passed Corsica with extremely severe wind gusts up to 62.2 m/s in Marignana, a station located on a hill. Two other stations in Northwest Corsica measured wind gusts above 50 m/s. 4 people died and 10 were injured in this time frame. Severe winds capsized boats or blew the boats against the rocks and beaches. Damage to the roofs and power lines was reported.  The updrafts of the storm, as well as lightning activity,  weakened as the bow crossed the island. 

07:00 – 08:00 UTC: The bow echo passed the northwestern part of Corsica and continued over the sea. 1 person died in this part of Corsica and wind gusts reached up to 34.4 m/s. Updrafts increased in intensity again as the bow entered the warm sea, as indicated by an uptick in lightning activity. 

08:00 – 09:00 UTC: The bow echo impacted the western Italian coastline of southern Liguria and Tuscany with extremely severe wind gusts up to 38 m/s, causing widespread wind damage to trees and roofs. Some roofs were completely destroyed or blown away. 2 people died and 45 were injured. Lightning activity weakened and also the wind damage became more isolated as the storm passed the Apennines. 

Roof damage in Ortonovo, Liguria, Italy. 
Photo by Città Della Spezia.

09:00 – 10:00 UTC: Lightning activity of the storm increased and new updrafts developed over the lowlands behind the Apennines. 

10:00 – 11:00 UTC: The storm system moved over the Veneto and Emilia-Romagna regions of Italy. Severe wind gusts reached up to 38 m/s and 9 people were injured. 

11:00 – 12:00 UTC: The lightning activity decreased again as the storms reached the border between Italy and Slovenia. 2 injuries were reported in this period.

12:00 – 13:00 UTC: The system produced severe wind gusts over Slovenia and moved over southern Austria, where the storms intensified again. At the same time, a severe storm over Menorca produced very large hail up to 7 cm in diameter. 

13:00 – 14:00 UTC: The system produced wind gusts up to 38.6 m/s in south-central Austria, mainly in Styria and Carinthia. 2 people died and 13 were injured in Mettersdorf. High voltage power lines were brought down by the winds. Storms were now traveling north.

14:00 – 15:00 UTC: The convective system continued producing severe winds over Austria with a maximum gust of 34.4 m/s as it traveled north. 3 further fatalities occurred in Styria as trees fell on hikers trying to protect themselves from the storm. 

15:00 – 17:00 UTC: The system started decaying shortly after 16 UTC over northern Austria. Still, it managed to produce severe wind gusts over south-central Czechia. 

Further storm activity in the evening yielded very large to giant hail in Toscany and Marche regions, Italy. The largest hail fell in Macerata Feltria, measuring 11 cm across and injuring 1 person. Giant hail damaged cars, roofs, and solar panels.

Giant hail with size estimated at 11 cm across. Photo by Irene Gaggia.

The environment of the storms and its relation to their evolution.

The first storms over the Balearic Sea developed in the forward flank of the trough, in the exit region of a cyclonically curved jet stream. These storms were already forming in an environment of 0-6 km bulk shear > 20 m/s and plentiful CAPE. The transformation into the bow echo occurred in an environment featuring MLCAPE > 3000 J/kg, which is extraordinarily high, and 0-3 km bulk shear > 25 m/s, conditions very favorable for intense convective windstorms.

The very high CAPE values developed due to the presence of abundant low-level moisture and steep mid-tropospheric lapse rates that had been advected from Sahara. Such pronounced overlap of CAPE and shear was present west of Corsica and between Corsica and the coastline of Tuscany. Further north and northeast, CAPE was lower, but bulk shear remained favorable for well-organised squall lines (> 15 m/s in 0-3 km layer) as far as central Austria. 

ECMWF forecast of MLCAPE and 0-6 km bulk shear combined with a forecast Skew-T and hodograph for a gridpoint west of Corsica. 

Besides the extremely severe wind gusts, the environment was very conducive for (very) large hail. ESSL’s experimental model-based forecasts AR-CHaMo predicted a large hail probability (within 40 km of a point) of over 60% in an area between Corsica, central and northeastern Italy in the 24-hour period between Thursday and Friday 06:00 UTC.

24-hour probability of large hail and lightning based on AR-CHaMo and the combination of 3 models.

While the presence of high CAPE and very strong shear helped the intense bow echo to develop, the crucial point was the upscale growth from the isolated storm to the squall line between Menorca and Corsica. We speculate that the development of the cold pool within the storm helped with the upscale growth. While the maritime boundary layer remained very moist, the presence of drier air and steep lapse rates above 900 hPa could have created strong downdrafts. Another factor may have been the mesoscale lift ahead of the progressing cold front that removed the CIN from the environment west of Corsica. The storm then tracked along the warm front towards Italy, where the further lift was provided by a warm air advection regime and a convergence zone. This lift also helped to remove the CIN from the environment. South of the warm front, CIN values were again very high, preventing convective initiation at this stage. Note that along the coast of Tuscany, vertical wind profiles would support intense tornadic supercells, but the organization of convection into a large-scale bow-echo lowered the threat of tornadoes compared to the severe wind gusts.

Sandwich imagery of the storm about to impact Italy at 07:45 UTC, combined with surface 
observations at 07:00 UTC. Positions of the three ICON-EU derived 
Skew-Ts are shown using the letters A, B and C. Note the temperature gradient 
along the warm front and the increase in CIN south of the front.

Looking at the surface observations, a cold pool was present during all stages of the derecho. The temperature of the cold pool of the storm remained constant throughout its evolution, between 20 and 21 degrees. The temperature difference between the cold pool and the surrounding environment increased as the system moved from the maritime areas further inland. Over western Corsica, the temperature decreased by 6 – 7 degrees when the gust front passed. Near the coastline of Tuscany, the temperature dropped only by 4 degrees. The temperature difference increased towards Austria and Slovenia with a drop of 9 to 13 degrees. This case shows that a strong cold pool is not necessary for a powerful convective windstorm.

The derecho went through several cycles of strengthening and weakening. By far the most impressive phase of strengthening occurred just west of Corsica as the squall line developed into a bow echo. The squall line was originally traveling towards the east. As the bow began to develop, the storm accelerated forward and its direction of movement turned more towards the north. The central part apex of the bow echo accelerated to an incredible speed of 40 – 50 m/s, compared to a mean wind of 17 m/s, which suggests the presence of a very strong rear inflow jet. The storm moved in the direction of the 0-3 km shear vector, i.e. in the direction of the strongest lift along the cold pool. 

Bow-echo evolution over Corsica as shown by radar between 06:00 and 06:30 UTC. 
The location of the most severe wind gust is shown by the star.
Source:  kachelmannwetter.com, meteologix.com

During the course of its path, the storm weakened twice as it passed the high terrain of Corsica and the Apennines. Downslope winds and perhaps lower CAPE disrupted the generation of new updrafts, which can be seen in the satellite imagery and lightning detection data.

A sequence of sandwich satellite images showing a lack of overshooting top development 
as the system crossed Corsica. 

The updrafts also weakened as the system moved through northeastern Italy. Over Slovenia, numerous severe wind gusts were observed well away from the heavy precipitation cores, suggesting a persistence of the outflow. Further intensification occurred as the outflow encountered the main Alpine ridge, over which new strong cells erupted. This was followed by an intensification of severe wind gusts, reaching over 32 m/s at several stations. At this time, the environment was still characterized by 0-3 km bulk shear exceeding 20 m/s.

Sandwich imagery from 13:30 UTC shows intensifying updrafts over the Alps. 
Yellow squares represent severe wind gust reports based on the ESWD.

The system finally died after moving over the area of weaker 0-3 km vertical wind shear, higher LFC, and marginal CAPE over northern Austria and southern Czechia.   

Windstorms and tornado outbreak of February 2022

This short blog post provides an overview of severe weather associated with the recent windstorms and discusses why the windstorm of 16 – 17 February ended up as a prolific tornado producer.

Multiple severe windstorms affected Europe in the period of 16 – 21 February. 2814 reports of severe wind gusts were submitted to the ESWD in this period, the majority of them in the belt from the British Isles through northwestern France, BENELUX, Germany into Czechia, northern Austria, and Poland. The evolution of severe weather reports for the windstorm of 16 and 17 February is shown below

Temporal evolution of severe weather reports between 16 February 12 UTC and 17 February 12 UTC. Yellow squares represent severe wind gusts and red triangles tornadoes.

Each of the windstorms also involved a strongly-forced convective line that formed along the cold fronts ahead of the fast-moving short-wave troughs. In many areas, the passage of the convective line was accompanied by an increase in wind gusts severity. The environment featured marginal CAPE with a low Equilibrium level (suggesting low-topped storms) and strong vertical wind shear in the lower troposphere with 20 – 30 m/s of 0-1 km bulk shear. Very strong flow just above the ground explains the severity of wind gusts observed along the track of the storms.

500 hPa geopotential height, temperature, and wind (upper panel) and MLCAPE and 0-1 km bulk shear (lower panel) for 17 February, 18 February, and 21 February windstorms.

Due to the strong low-level shear, enhanced tornado threat accompanied the passage of all the windstorms. Between the night and morning hours of 17 February, a tornado outbreak occurred in a swath from northeastern Germany to south-central Poland, Malopolskie region. By 28 February, 22 tornadoes causing 2 fatalities and 5 injuries were reported in the ESWD, while further site surveys are ongoing by Skywarn Polska, and it is likely that the number of tornadoes will grow. 12 out of 22 tornadoes were strong and rated as IF2.

While the 17 February convective windstorm ended as a prolific tornado producer, the other two situations did not result in a single tornado despite a very strongly sheared environment. Overall, three windstorms occurred in this period: the 16 – 17 February windstorm (Dudley), the 18 February windstorm (Eunice), and the 20 – 21 February windstorm (Franklin).  The 16 – 17 February and the 20 – 21 February windstorms featured well-organized convective systems with persistent bowing segments and inflow notches. The 18 February convective line was broken with no accompanying bowing segments.

The first difference among the three situations was in the orientation of the convective line, the direction of the mean wind (green arrow), and the direction of the movement of convective elements, such as bow-echoes or inflow-notches (red arrow). In all three situations, the convective elements within the line moved to the right of the mean wind. The largest angle between the mean wind and the orientation of the convective line was noted for the 17 February windstorm over Poland, suggesting the strongest lift on the leading edge of the convective line (perpendicular being the most “favorable” configuration). The individual convective elements also deviated most from the mean wind in this case.

Relationship of convective system orientation and motion of individual elements (such as bow-echoes) to the wind profile characteristics for different cases. Upper panel: Black lines denotes the orientation of the convective system, green arrows represent the mean wind, and red arrows the direction of motion of individual convective elements. Lower panel: Hodographs with 0-1 km SRH highlighted for observed storm motion of convective elements.

Hodographs based on the ICON-DE allow us to reconstruct the storm-relative helicity available to the convective elements in different situations. Considering the observed motion of the convective elements (marked in the hodograph by blue stars), the wind profile on 17 February over Poland had by far the most SRH. The inflow into the storm would be very strong (> 20 m/s) and contain almost purely streamwise vorticity (meaning aligned with the flow, making it “helical”). Strong inflow combined with large amounts of streamwise vorticity helped to create a favorable environment for the development of strong rotation in parts of the advancing convective system.

The reason for a higher amount of helicity was a stronger southerly component of the flow compared to the other situations. This is evident both from the hodographs and the surface station measurements. As the convective line crossed Central Europe between 16 and 17 February, 10 m wind over northern Germany had a smaller southerly component than over central Poland.

EUCom radar composite combined with surface station data (temperature and dew point shown in red and green color respectively) and ESWD reports (red triangles correspond to tornadoes and yellow squares to severe wind gusts).

The amount of helicity in the near-surface inflow seems to be the main difference between the tornado-producing convective windstorm over Poland on 17 February and the other windstorms. 17 February also had the most abundant low-level moisture content (and the highest CAPE values) as well as the line broken into many different convective elements (bow-echoes and inflow notches). Compared to the Kyrill tornado outbreak on 18 January 2007, cold pools were weaker on 17 February 2022. 

Hailstorms of 2021

2021 was a very big year for severe hailstorms across Europe. As of 31 October, we have received 5195 reports of large (≥ 2 cm) hail, 871 reports of very large hail (≥ 5 cm), and 29 reports of giant hail (≥ 10 cm) in the European Severe Weather Database. These numbers are far greater than in the previous years. The number of very large hail reports collected in 2021 is more than twice more than the second most active year, which was 2019. By 31 July, the sum of large hail reports exceeded the sum of reports in the past years collected over the whole 12 months.

2021 does not stand out as much when looking at the number of days with large and very large hail. The inflation in the number of hail reports was caused by very large amounts of reports submitted for some individual days, such as 24 June or 8 July. Any trends in hail reports can not be used to infer the trend in the frequency of severe hailstorms in Europe due to a large number of non-meteorological factors, such as the increases in reporting rates across different countries.

A number of large and very large hail reports and days between 2006 and 2021 in the period of 1 January – 31 October.

The 2021 severe hail season was particularly active in the pre-Alpine region, both on the northern (Switzerland, Austria, Germany) and southern flank (Italy), as well as across eastern Poland and southeastern Czechia. The largest hail was reported on 24 June in Poland with an estimated hail size of at least 13.5 cm. The largest number of injuries, 12,  was reported from the 26 September hailstorm that hit Tuscany, Italy. 84.7% of the submitted hail reports included estimate or measurement of hail size. 

Spatial and size distribution of 2021 hail reports across Europe till October 2021.  Size ratio expresses the percentage of reports that include hail size.

Below you can find a selection and description of some of the most severe hail days and hailstorms in 2021. This list is non-exhaustive, so a severe hailstorm that affected your region may not be included. Red font indicates a particularly severe event. In case you feel that we have missed a particularly severe event in this list or it is not included in the ESWD, please let us know and/or submit your reports at eswd.eu!

ESSL would like to express gratitude to all voluntary observers and networks for submitting severe weather reports for their respective countries.

22 May 2021: Numerous severe hailstorms produced very large hail up to 7 cm across in Stavropolskyi Kray in Russia.

12 June 2021: Very large hail up to 9 cm across fell in the town Tábua, Portugal.

14 June 2021: 40 sheep were killed by large hail near Ankara, Turkey.

16 June 2021: Very large hail up to 6 cm across damaged numerous roofs and cars in Krasnodarskyi kraj, Russia. 

19 June 2021: Severe hailstorms affected the Central, Midi-Pyrénéés, Bourgogne, Picardie and Franche-Comtéregion regions in France. The largest hail fell in Vercel-Villedieu-le-Camp with a diameter up to 10 cm, causing significant damage to cars and roofs. This was the first giant hail report in Europe in 2021. Very large hail up to 7 cm also fell in Luxembourg

20 June 2021: Very large hail was reported in the Rhone-Alpes region, France, and Luzern canton in Switzerland. 

21 June 2021: Widespread area spanning a belt from southern France through Switzerland into northwestern Austria was affected by severe hailstorms. Southern suburban areas of Munich experienced hail up to 5 cm across. Severe hail damage was reported from the border area of Austria and Bavaria with the largest hail size of 6.5 cm.

22 June 2021: 7 cm hail was reported in southern Russia. Two long-tracked hailstorms affected southern Germany and northwestern Austria. Severe damage to cars from 6 cm large hail was reported from the town of Gmunden, Austria.

23 June 2021: Severe hailstorm with hail up to 7 cm across caused damage on the border of Austria and Czechia.

24 June 2021: 809 severe hail reports were submitted to the ESWD for this day, 615 of them from Poland. This day has broken the record for the maximum daily number of hail reports submitted to the ESWD, as well as the maximum daily number of hail reports submitted for a single country. Several hail-related injuries were reported to the ESWD from this day along with widespread hail damage. In some villages and towns, most roofs were damaged or destroyed by hail. 

Very large hail was reported from Austria, Czechia, Slovakia and Poland. Giant hail was reported from Poland and Austria. In Poland, giant hail was observed in multiple locations with the largest hailstone, measuring at least 13.5 cm across, reported from Tomaszów Mazowiecki, breaking the national record. The largest hail fell from a left-moving supercell. In Austria, giant hail was reported in two locations with the largest hailstone falling in Hollabrunn and an estimated diameter of 12 cm. Supercell that spawned the violent tornado between Breclav and Hodonin in Czechia also produced hail up to 9 cm on the border of Czechia and Austria. 

A photo of the largest hailstone of the 2021 hail season in Europe that fell in Tomaszów Mazowiecki, Poland. The lower-end estimate using a ruler in the hand of the person that took the photo was placed at 13.5 cm. Photo by Magdalena Cecotka.

25 June 2021: Severe hailstorms affected southeastern Austria, Croatia, Hungary, northern Slovakia and southern Poland. The largest 8.5 cm hailstone was reported from Czarny Potok in Poland.

28 June 2021: Very large hail was reported from Switzerland and southern Germany. 11 people were injured by hail in Switzerland and a giant, 10 cm, hail was observed in Wolhusen with significant damage to roofs, windows, and cars. 

29 June 2021: Multiple severe hailstorms were reported in the Alpine area and parts of central Europe. A Hailstone of 8 cm in diameter was photographed in Tarcento, Italy.

30 June 2021:Very large hail up to 7 cm fell in northern Slovakia. A long track hailstorm across southern Poland produced giant hail up to 11.5 cm that weighed up to 200 g.

1 July 2021: Very large hail up to 7 cm across reported in Bulgaria. 

8 July 2021: Very large hail was reported in Italy, Switzerland and Czechia. 169 large hail reports were submitted to the ESWD from Italy. The largest hail, 11 cm in diameter, fell in Rozanno, on the southwestern outskirts of Milan.

Very large to giant hailstones in Rozanno, Italy on 8 July 2021.
Image source: Tornado in Italia (author unknown)

9 July 2021: Very large to giant hail was reported from Croatia (up to 9 cm across), Hungary (up to 9 cm across) and Poland (up to 11 cm across). The largest hail across Poland was again produced by a left-moving supercell.

13 July 2021: Very large hail was reported from northern Italy (up to 7 cm in diameter) and southwestern Czechia (up to 6 cm in diameter).

14 July 2021: Very large hail up to 8 cm in diameter was reported from eastern Austria in the early morning hours.

15 July 2021: Very large hail was reported in Serbia and Romania (up to 7 cm in diameter).

25 July 2021: Very large hail, up to 8 cm in diameter, observed in Lombardia.

26 July 2021: Several long-tracked hailstorms formed over southern Germany and northern Italy. The largest hailstone reached 8 cm in diameter. One of the hailstorms crossed the highway near Parma, severely damaging hundreds of cars that were undrivable afterward. Minor injuries were also reported. Very large hail was also reported in Austria and Poland.

Examples of damage inflicted by wind-driven hail on 26 July 2021. Photos were taken by Rosella Rotelli and inmeteo.net. 

30 July 2021: Very large hail up to 7 cm across damaged roofs in Rochefort-Samson, France.

31 July 2021: Numerous severe hailstorms occurred in eastern Spain and northern Italy. Giant hail up to 10 cm across fell in Peñíscola, the Valenciana region of Spain. In Italy, the largest hail fell in Piemonte with a diameter up to 8 cm across multiple locations. 

1 August 2021: An early morning hailstorm produced giant hail up to 11 cm in diameter in Azzano Decino, Friuli-Venezia Guilia region of Italy. Serious damage to roofs and cars was reported. In the afternoon, very large hail was observed in Romania.

15 August 2021: Severe hailstorms formed over eastern France, southern Germany and eastern Slovakia. Maximum hail size reached 6 cm. 

23 August 2021: Very large hail up to 8 cm in diameter was reported from southern Russia.

18 September 2021: Damaging hailstorm affected Tivat, Montenegro in the morning hours. Large hail caused significant damage to windows, cars, and crops

26 September 2021: 12 people were injured by very large hail in Vaglia, Tuscany. Significant damage was done to cars, roofs, windows, and crops.

ESSL Testbed 2021

10th edition of the ESSL Testbed is upon us and will be held in four weeks: 1418 June, 2125 June, 59 July and 1216 July. The second and third weeks are for invited experts, or participants who have already been to the Testbed before or other courses. Like last year, the Testbed will be held online. 44 participants have registered, 12 people for the regular and 10 for the expert weeks.

Compared to 2020, there are a few new aspects. First, we have improved our Weather Data Displayer, which has a new, darker layout, and now allows a smoother switching between different regions. Moreover, a transparency option for data layers has been added. There is also more NWP data in the displayer compared to last year, including ensemble output from ICON-D2 (DWD) and C-LAEF (ZAMG). The third convection allowing model is Harmonie (KNMI). A large part of the data processing is now done in the European Weather Cloud, with thanks to ECMWF.

Besides daily forecasts of convective storms, we are going to evaluate new tools developed for forecasting and nowcasting of convective storms. This year we will concentrate on:

  1. Extreme Forecast Index and Shift of Tails for CAPE and CAPE-shear from ECMWF
  2. C-LAEF ensemble prediction system from ZAMG
  3. ICON-D2 ensemble prediction system from DWD
  4. Modified Lightning Potential Index from DWD
  5. KONRAD3D cell-tracking nowcasting tool from DWD
  6. STEPS-DWD radar nowcasting tool from DWD
  7. NowcastSAT satellite nowcasting tool from DWD

We are looking forward to all the participants and interesting discussions on these new products and the hopefully interesting weather situations!

We welcome external persons to join the daily Testbed weather briefings from Tuesday through Friday during Testbed weeks starting at 11:00 CEST (0900 UTC) until approximately 12:30 CEST (1030 UTC). To join, visit this link: https://bluejeans.com/720241930

Afternoons with convection

*** UPDATE on recordings ***

We will post links to the recordings of the webinars here as they become available (this also depends whether our speakers agree to having their talk recorded)

Matthew Kumjian: Forecasting large hail

John Peters: New insights on supercell dynamics

Cameron Nixon: Using hodographs

*** UPDATE on how to connect to the webinars ***

To connect to the webinars, just follow the link:
https://bluejeans.com/2536484999 and Bluejeans will guide through you through the rest! Make sure to reserve 5 – 10 minutes before the start of the webinar to allow your system to set up the Bluejeans and connect to the webinar.

The most important rule is to make sure your microphone is muted during the presentation! This will be done by default upon your arrival to the webinar, but please make sure this will be the case also individually. Depending on the number of participants, discussion will be either done throught the chat or more interactively.

*** GENERAL INFORMATION concerning the webinars ***

Last year we brought you a series of 3 “refreshers” to get ready for convective season. While the convective season picks up over Europe, this year we are organizing a series of 3 webinars called “Afternoons with convection” with severe weather experts to help you expand your horizons in convection forecasting!

The webinars will take place on the Wednesdays 2, 9 and 30 June and will start at 13 UTC (15:00 CEST). Each webinar consists of a lecture by a storm researcher followed by a discussion and concludes with a discussion of the current weather situation by Tomas Púčik and Christoph Gatzen (ESSL / ESTOFEX). The webinars end at 15 UTC.

You do not need to register for the webinars! On Mondays, two days before each webinar, we will post a link on ESSL’s Twitter (@essl_ecss) and Facebook channels and the webpage (essl.org) with instructions how to connect. As last year, we will use the BlueJeans videoconferencing system.

We are happy that the following expert researchers have offered to give a talk on their recent work in the Afternoons with Convection series:

2 June 2021, Matthew Kumjian (The Pennsylvania State University)
Forecasting and nowcasting large hail.

Large hail is the most costly thunderstorm-related threat and each year multiple events with hail-related injuries occur across Europe. Matt will discuss new findings of his research group concerning the physics of large hail formation, such as how the pre-convective environment influences the growth of large hailstones and how to recognize truly damaging hailstorms.  

9 June 2021, John Peters (Naval Postgraduate School)
New insights on dynamics of supercells

Understanding why storms behave in a certain way is very important for all forecasters. John will talk about his novel research results concerning various aspects of storm dynamics, including how shear modulates the width and strength of convective updrafts, in what ways updrafts of supercells are special, and what role entrainment plays in the evolution of updrafts.

30 June 2021, Cameron Nixon (Central Michigan University)
 Using hodographs to their fullest potential.

Long? Short? Curved? Straight? Hodographs offer a holistic approach to study vertical wind shear in the context of convective storm dynamics. Cameron will talk about how hodographs can predict supercell behavior, hazards, and even visual appearance, and how learning patterns rather than bulk shear or SRH will improve our forecasts in the future.

New study on climatology of severe convective storms and their environments

A two-part study on the climatology of severe convective storms over Europe and the U.S. was recently published in the Journal of Climate. The study was led by Dr. Mateusz Taszarek of the National Severe Storms Laboratory (Norman, OK, USA) and Adam Mickiewicz University (Poznán, Poland) with contributions from ESSL among others. The study used lightning detection data, European Severe Weather Database reports, Storm Data reports and the ERA-5 reanalysis to answer the research questions.

The first part, “Severe Convective Storms across Europe and the United States. Part I: Climatology of Lightning, Large Hail, Severe Wind, and Tornadoes“, deals with the climatology of lightning and severe weather reports. In general, there is more lightning over the U.S. and thunderstorms tend to last longer and produce more strikes than over Europe. Over continental Europe, most of the thunderstorms occur during the day, but the opposite is true for the continental U.S., where more than 50% of lightning occurs at night. A higher frequency of thunderstorms at night is observed over the seas, particularly over the southeastern Mediterranean.

(a) Annual mean number of hours with lightning, (b) fraction of nocturnal lightning, (c) mean number of hours with lightning per thunderstorm day, and (d) mean number of flashes per thunderstorm hour over the United States (1989–2018) and Europe (2006–18). Data are presented in 0.25° boxes with a 0.75° × 0.75° spatial smoother. Nocturnal lightning is defined when a sun angle for a specific grid and date is below 0°.
Copyright of American Meteorological Society, CC-BY License

Over the U.S., the fraction of lightning hours associated with severe weather reports is higher than over Europe with the exception of winter, when a seasonal maximum is observed over both areas.  One can also clearly see stronger reporting inhomogeneity across European countries in contrast to the more homogeneous Storm Data that cover the U.S.. Compared to Europe, extreme events are considerably more frequent over the United States, with a maximum activity over the Great Plains. However, the threat over Europe should not be underestimated, as severe weather outbreaks with damaging winds, very large hail, and significant tornadoes occasionally occur over densely populated areas

Fraction of hours with lightning associated with tornadoes, large hail, and severe wind over the seasons in the United States (1989–2018) and Europe (2006–18). Data are presented in 0.25° boxes with a 0.75° × 0.75° spatial smoother.
Copyright of American Meteorological Society, CC-BY License.

The second part, “Severe Convective Storms across Europe and the United States. Part II: ERA5 Environments Associated with Lightning, Large Hail, Severe Wind, and Tornadoes“, deals with the environments of severe convective storms and their hazards derived from the ERA-5 reanalysis. The U.S. experiences more extreme severe convective storm environments than Europe when considering how often high CAPE, strong vertical wind shear and high Storm Relative Helicity (SRH) occur. This explains why tornadoes are rarer over Europe than over the U.S. On the other hand, 0-3km CAPE is higher and low-level lapse rates are steeper over Europe.

(a) Scaled mean number of hours per year with lightning and (b) conditional probability as for significant tornados (F2+) given specific ML WMAXSHEAR (product of square root of 2* MLCAPE and 0-6 km bulk shear) and 0–1-km storm-relative helicity parameter space.
Copyright of American Meteorological Society, CC-BY License.

While environments conducive to severe convection (characterized by the simultaneous occurrence of CAPE and shear) occur more frequently over the U.S., European severe storm environments more frequently result in storms. 30 – 40% of the thunderstorms over the midwest of the U.S. form in conditions conducive to severe weather, in contrast to 10 – 25% of thunderstorms over central and western Europe. Over Europe, the highest fraction of thunderstorms forming in severe environments is found over the Balearic Sea and the northern Adriatic Sea.

(a) Annual mean number of situations (hours) with severe environments and lightning detection (at least 2 flashes), (b) probability of convective initiation in severe environment (fraction of severe environments associated with lightning), and (c) probability that a developing thunderstorm will be be associated with severe environment (fraction of lightning events associated with severe environments).
Copyright of American Meteorological Society, CC-BY License.

Additionally, an animation of the annual cycle of lightning activity, MLCAPE and 0-6 km vertical wind shear over both areas can be seen below. Over Europe, one can see a shift of thunderstorm activity from land to sea as the year progresses from spring and summer to autumn. In both areas, the seasonal increase in CAPE is accompanied by a decrease in shear and vice versa. The U.S. Midwest sees a pronounced combined occurrence of high CAPE and strong shear in the spring, while it occurs across western Mediterranean in autumn.

Taszarek, M., Allen, J. T., Groenemeijer, P., Edwards, R., Brooks, H. E., Chmielewski, V., & Enno, S. (2020). Severe Convective Storms across Europe and the United States. Part I: Climatology of Lightning, Large Hail, Severe Wind, and Tornadoes, Journal of Climate, 33(23), 10239-10261. Retrieved Dec 9, 2020, from https://journals.ametsoc.org/view/journals/clim/33/23/jcliD200345.xml

Taszarek, M., Allen, J. T., Púčik, T., Hoogewind, K. A., & Brooks, H. E. (2020). Severe Convective Storms across Europe and the United States. Part II: ERA5 Environments Associated with Lightning, Large Hail, Severe Wind, and Tornadoes, Journal of Climate, 33(23), 10263-10286. Retrieved Dec 9, 2020, from https://journals.ametsoc.org/view/journals/clim/33/23/jcliD200346.xml

Severe weather outbreak in Slovakia and Poland on 4 – 5 October 2020

Severe weather outbreaks in autumn are quite typical over the Mediterranean but rarer over continental parts of Europe. On 4 and 5 October, an unusual synoptic-scale situation led to an outbreak of severe thunderstorms over eastern Slovakia and Poland.

During these two days, a total of 36 large hail reports (with 5 reports of hail exceeding 5 cm in diameter), 71 reports of severe wind gusts (with several instances of F1 damage), 4 reports of damaging lightning and an F1 tornado report were entered into the European Severe Weather Database (eswd.eu). The map below gives an overview of the impacts. While severe convective wind gusts are not uncommon with deep low pressure systems over this part of Europe, hail reaching 7 cm or more in diameter is exceptional for this time of year.

Severe weather reports on 4 and 5 October 2020. Data source: eswd.eu

The outbreak was caused by severe thunderstorms developing in a very warm and moist airmass that had been advected northward from the southern Mediterranean Sea (Fig 2). Mixing ratios of 10 – 12 g/kg allowed 500 – 1500 J/kg of MLCAPE to build over the area. Combined with strong vertical wind shear (exceeding 20 m/s in the 0-3 km layer), thunderstorms quickly turned severe as they organized into supercells and bow-echoes.

IR satellite imagery combined with an ECMWF forecast of specific humidity at 500 m AGL (g/kg, filled contours) and lapse rates between 850 and 500 hPa (K/km, red to purple contours). Forecast sounding and hodograph are representative of the location indicated by red cross.

More information concerning the F1 tornado over southern Slovakia can be found here. ESSL would like to thank Skywarn Polska and the Slovak Hydro-Meteorological Institute for providing severe weather reports!

A Challenging Tornado Forecast in Slovakia

Tornado forecasting can be very challenging, especially in low CAPE – high shear environments and when lower tropospheric shear is only locally enhanced. Such was the case of an F1 tornado that hit the village of Lekárovce in eastern Slovakia on 3 October 2018. This tornado is the main subject of a paper “A Challenging Tornado Forecast in Slovakia” that has been recently published in the journal Atmosphere. The study was led by Miroslav Šinger from Comenius University and Slovak Hydrometeorological Institute in Bratislava and co-authored by Tomáš Púčik from ESSL. 

Roof damaged by the tornado and a toppled concrete cross. Author: Martin Žec.

Authors of the study attempted to reconstruct the environment leading up to the tornado based on the observational datasets and compared it with the model data that was available to forecasters in the morning hours before the tornado. One of the main aims of the paper was to show whether observational data or the higher resolution run of the local model would improve the ability to identify conditions favourable for tornadogenesis.

Tornado occurred in the early afternoon hours underneath strong west-northwesterly mid to upper tropospheric flow at the flank of a deep low-pressure system. Enhanced lower tropospheric moisture combined with mid-tropospheric lapse rates of 6.5 K/km allowed for a build-up of marginal CAPE reaching 200 – 300 J/kg.

Model forecast of 3 October 2018 12:00 UTC (a) Average lapse rate in 900–600 hPa layer (black isolines) (°C·km−1) and dew-point temperature at 2 m (color scale) (°C); (b) surface-based convective available potential energy (CAPE; color scale) (J·kg−1) and deep-layer shear (wind barbs: full barb, 5 m·s−1; half barb, 2.5 m·s−1; flag, 25 m·s−1). Country names shown in italics; tornado location highlighted by blue star.

But while the deep-layer shear was very strong, models predicted a decrease in 0-1 km shear between morning and early afternoon hours with surface wind veering from East to West. However, easterly wind direction persisted over eastern Slovakia for much longer, yielding strong lower-tropospheric shear as the storms developed in the afternoon. Study shows that using both observational datasets and the higher resolution version of the local area model would alert the forecaster to the presence of strong lower tropospheric shear over the area of interest.

Hodographs representing the environment before tornadogenesis based on observational datasets (red), model forecast available to the forecasters (black), and the experimental higher resolution run (blue). Crossed circles represent the storm-motion based on Bunker`s method, orange represents the real vector derived from the radar data. Values of low-level wind shear (LLS) and storm relative helicity (SRH) in 0–1 and 0–3 km layers for individual hodographs are shown in top-right corner.

More details can be found here: Šinger, M.; Púčik, T. A Challenging Tornado Forecast in Slovakia. Atmosphere 202011, 821.

Tornado outbreak of 1967: How bad would it be today and could we forecast it?

The tornado outbreak of 2425 June 1967 remains the second deadliest tornado outbreak over Europe since 1950 after the tornado outbreak of 9 June 1984 over Russia. Over the course of two days, one F2 tornado, four F3 tornadoes, one F4 tornado, and one F5 tornado struck France, Belgium and Holland (Fig. 1), resulting in 15 fatalities and 234 injuries

Figure 1. Tornado tracks from 24 (red) and 25 (blue colour) June 1967.

In a study from 2018, a team led by Bogdan Antonescu looked at the details of the outbreak and at what would be the consequences if a similar tornado outbreak will occur 50 years later (i.e., 2017). This was done by transposing the seven tornado tracks from the June 1967 outbreak over the modern landscape. Due to urban growth, it is possible that tornadoes could cause even more impact than in 1967. Based on the statistics of fatality and injury rates associated with European extracted from the European Severe Weather Database, a similar tornado outbreak with the one that occurred in 1967, would result in 55–2580 injuries, and 17–172 fatalities. In the worst-case scenario, with tornado tracks moving over highly populated areas over the region,  up to 146 222 buildings could be  impacted with 2550–25 440 injuries and 170–1696 fatalities. This study clearly shows how impactful such a tornado outbreak could be to society.

A follow-up study that has just been published in Weather and Forecasting (link) hindcasts the tornado outbreak using an WRF-ARW simulation, with initial and boundary conditions provided by ERA-40 reanalysis and the highest-resolution domain with 800-m grid spacing. The model simulated an environment conducive for tornadic supercells with CAPE exceeding 2000 J/kg, 0–6 km bulk shear between 20–25 m/s and Storm Relative Helicity reaching 300 m²/s² in the 0–3-km layer. The model was also able to explicitly simulate a number of supercells over the region of interest (Fig. 2).  

Figure 2. Timestamps of radar reflectivity echo at 14, 15, 16 and 17 UTC on 24 June 1967.

One of the questions posed by the article is how would a forecast of such an event look like today? To do this, an ESTOFEX forecaster was presented with a set of forecast maps for both days (without knowing the dates) and asked to provide Day-1 outlooks. The forecaster issued Level 3 (the highest risk of severe weather) for both days, on  (Fig. 3). This paper demonstrates that, with our understanding of severe convective storms and state-of-the-art numerical modelling, a forecast of a tornado outbreak over Europe is possible. Given how much societal impact significant tornadoes can cause, addressing their threat should be part of any convective storm forecast.

Figure 3. ESTOFEX hindcast outlooks for a) 24 June 1967  and b) 25 June 1967.

Full versions of both studies can be freely accessed here:

Antonescu, B., J.G. Fairman Jr, and D. M. Schultz, 2018. What is the worst that could happen? Reexamining the 24–25 June 1967 tornado outbreak over Western Europe. Weather, Climate, and Society, 10, 323–340.

Antonescu, B., T. Púčik, and D. M.Schultz, 2019. Hindcasting the First Tornado Forecast in Europe: 25 June 1967. Weather and Forecasting, 35, 417–436.