Supercell Thunderstorms on September 7, 2024, in Greece: Documentation and Predictability ^†

Abstract

On September 7, 2024, a deep convection event was observed in Northern and Central Greece, and based on radar data analysis, three supercells were identified. One of these, the most intense with maximum radar reflectivity of 68 dBZ, had a lifetime of almost 7 h and covered a distance of more than 200 km, producing damaging winds and large hail along its track. The goal of this study was to analyze this case using radar data and to evaluate the predictability of such a high-impact event using a numerical weather prediction model. The Weather Research and Forecasting (ARW-WRF) model was used to perform an array of simulations, and using multiple initialization times, the influence of lead time was examined. Furthermore, the dependence of the results on the choice of parameterization scheme used in the model is assessed below. The model performed satisfactorily in predicting intense storm activity, without reaching the extreme values observed by the radar.

1. Introduction

Supercell thunderstorms, although the rarest type of storm, are responsible for high-impact hazardous weather phenomena as most become severe events during their lifetime. Supercells (SCs), defined by the presence of a deep, persistent, rotating updraft (mesocyclone), are highly organized storms, and are the longest-lived form of deep convection [1,2]. Usually, they are characterized by destructive power, large hailstones, damaging straight-line winds, flash floods, and sometimes tornadoes, resulting in devastating damage to property and agriculture, as well as injury or loss of life.

Several previous studies have shown that the development of supercells is not unprecedented in Europe [3,4], and even Greece is vulnerable to the occurrence of supercells, mainly during the warm season of the year, while their associated severe weather phenomena produce disastrous consequences with the greatest socioeconomic impacts [5]. The catastrophic supercell that affected northern Greece on July 10, 2019, was one such event, which caused widespread damage to agriculture and properties, while seven fatalities and 120 injuries were reported in the prefecture of Chalkidiki [6,7].

Recent studies indicate that as global warming continues to rise, mainly due to human activity, extreme weather events will become more frequent and intense in future climates, with significant spatiotemporal shifts for supercell occurrences in the USA [8], while in Europe, a future climate simulation shows an average increase of supercell occurrence by 11% [9]. Therefore, it is imperative to have more accurate forecasts concerning the occurrence of supercells, and it is necessary to be able to predict the evolution, strength, longevity, and motion of such storms, as the proper reaction to such situations is of great importance.

The investigations of this study focused on the following two main aspects: Firstly, to provide supercell storms documentation based on radar data analysis, and secondly, to assess the predictability of the high-impact event, using a numerical weather prediction model.

2. Data and Methods

In the present study, the synoptic and mesoscale environments on September 7, 2024, are examined using surface and upper-level data. Emphasis is placed on supercell identification based on radar data from two C-band weather radars, which are situated in the Filyro and Liopraso sites (Figure 1). Both radars are equipped with the TITAN software [10]. The Hellenic Agricultural Insurance Organization (ELGA) operates these radars during the warm season of the year, within the framework of the Greek National Hail Suppression Program (GNHSP). The study area covers the regions of Northern and Central Greece. A radar volume scan completes approximately every 3.5 min, and the radar data are used to examine the development, evolution, and movement of the storms.

The numerical simulations were performed using the non-hydrostatic Weather Research and Forecasting modeling system with the Advanced Research dynamic solver (WRF-ARW, 3.9) [11,12]. Three model domains were used, covering (a) Europe and northern Africa (d01), (b) the Ionian Sea, the Balkans, and part of Asia Minor (d02), and (c) Northern and Central Greece (d03), with a grid size of 15, 5, and 1.67 km, respectively (Figure 1). Results from d03 were acquired every 10 min to follow the fast-evolving nature of a thunderstorm.

The effects of model microphysics were investigated by performing simulations using the WSM6 [13] and the Ferrier-ETA [14] schemes, while the model initialization time effect was investigated using two initialization times: 12 UTC, 06/09/2024, and 00 UTC, 07/09/2024 (GFS 3-hourly forecasts, 0.25° × 0.25°). In this way, four simulations were performed, namely WSM6_12, WSM6_00, ETA_12, and ETA_00. RRTMG represented longwave and shortwave radiation, and the NOAH Unified model soil physics in all runs.

3. Results

3.1. Synoptic, Thermodynamic, and Shear Conditions

On September 7, 2024, at upper levels, a weak ridge was situated over Italy, producing NW flow over Greece at 00:00 UTC, while a low over the Black Sea was expected to deepen and move southwestwards later in the day. Additionally, some short-wave troughs were embedded in this circulation, associated with air masses of −10 °C, and gradually strong vertical motions developed over the area of interest. At middle levels, a thermal ridge located west of Greece established warm (18 °C) and moist air masses over the study area. In the upper troposphere, throughout the day, a jet streak over the Balkan region gradually entered the northern part of the country, strengthening the horizontal gradient, thus contributing to moderate vertical wind shear, while absolute vorticity advection intensified over Greece.

Furthermore, thermodynamic and kinematic variables of the 06:00 Thessaloniki sounding are calculated using SHARPpy version 1.4.0 [15]. The thermodynamic environment featured moderate instability, showing MUCAPE values of 1521 JKg⁻¹ forecasted to increase up to 1892 JKg⁻¹; thus, there was significant potential for thunderstorms with strong updrafts. Strong downdrafts were also anticipated as the DCAPE was 1199 JKg⁻¹. The fact that the FCST CIN was expected to be almost zero (−6 JKg⁻¹) intensified the potential for severe convection. The high values of the instability indices (TT: 51, KI: 36, LI: −6, SWEAT: 203) reflected the moist and unstable atmosphere, and the increased value (60 K) of the Soaring Index suggested a greater potential for thermal lift.

Supercells form in strongly sheared environments with favorable vertical wind shear, which causes the development of dynamic processes in the storm. On this day, the examination of the vertical profile of the atmosphere indicated a moderate wind shear with low-level wind shear (0–3 km) of 12 kts, deep-layer shear (0–6 km) of 30 kts, effective bulk shear (EBWD) of 31 kts, and BRN shear of 5 m² s⁻². These values, as well as the BRN (218), were best related to the development of multicell storms. Opposite to this, the Supercell Composite Index (SCI) was 1.4, which favored right-moving supercells. The SRH (0–3 km) of 59 m² s⁻² was within the limits (−40 to 175 m² s⁻²). Bunkers [16] found that either left- or right-moving supercells can occur. Furthermore, previous case studies concerning supercell developments in Greece found that such values can support supercells [17]. Additionally, the shape of the shear profile, as depicted in the hodograph, was essentially a straight-line hodograph, but with a clockwise turn at lower levels, which strongly influenced convective storms’ evolution and organization.

3.2. Description of the Event: Radar Analysis of Supercells Evolution

On September 7, 2024, around midday, the first convective echoes were detected from the Filyro radar in Northern Greece. One of those, at 13:15 UTC, presented supercell characteristics (SC1) for about 20 min east of Kilkis town. This supercell did not have remarkable features, although it reached 65 dBZ with tops up to 11.5 km. Its lifetime was only 1.7 h, and it traveled 28 km towards SSE with a mean speed of 16.5 km/h.

Later, after diurnal heating, which has a strong influence on the initiation of supercells [5], enhanced by dynamic forcing, two new cells popped up almost simultaneously (14:30) over mountainous terrain and initially moved southeastwards. Soon after, the propagation direction of the southern cell, which originated over Pindos Mountain, changed and gradually shifted more to the south as the storm evolved into a powerful supercell (SC2). At 15:32, SC2 presented quintessential supercellular signatures (Figure 2a), as the well-formed “Hook Echo” and the “Weak Echo Region” (WER). Beyond the 4.75 km level, the “Bounded Weak Echo Region” (BWER) was persistent, indicating that SC2 was associated with a significant mesocyclone. The characteristic signature of the V-notch was also well distinguished, while the anvil length was expanded to 160 km. SC2 maintained impressive supercellular features for about 4 h during its almost 7 h lifetime, covering the distance of more than 200 km through the Thessaly plain (Figure 2b). Along the supercell track, damaged swaths with hailstones up to 5 cm in diameter, gale-force winds, and flash floods were reported by many municipalities of the Thessaly region. According to ELGA, the total monetary compensation for hail and wind damage was significant.

The other cell that originated NE of Kozani in the Western Macedonia region presented supercellular characteristics (SC3) at 15:11, about 15 min later than SC2, but both indicated maximum reflectivity of 68 dBZ close to 15:45 and had echo tops penetrating the tropopause, reaching about 15 km. The speed of both SC2 and SC3 was higher than 30 km/h, double that of SC1, and this was due to the intensification of the wind at middle and upper levels. SC3 from 15:39 to 15:50 indicated an impressive hail core, which is well reflected by the “Hail Mass Aloft” radar parameter that exceeds 1715 ktons (Figure 2c). Radar reflectivities greater than 55 dBZ are attributed to hail [18], and that was confirmed by the large walnut-sized hail reported from the greater area of Velventos, as well as from the Elassona and Tirnavos municipalities. After 16:35, SC3 revived and shifted towards the south as it approached the Tirnavos area (Figure 2d). It is notable to mention that in the frame of the GNHSP, both supercells were seeded during some period of their lifetime.

SC2 and SC3 had large dimensions (30–40 km diameter) and presented a discrete convective mode, although only SC2 was isolated, and this contributed to its longevity, as there were no interactions between neighboring storms. The persistent rotating updraft, which enhances vertical motion, and the updraft–downdraft configuration, which prevents precipitation from falling through the updraft, enable longer-lived supercells [1].

3.3. Numerical Simulations

All simulations predict storm activity in the area of the observed storms with varying intensity and duration. The common characteristic of all runs was the early onset of the storm activity by about two hours compared to the observed ones.

Regarding maximum reflectivity, ETA_00 resulted in the smallest values in both the intensity and area covered. In the ETA_12 run, the storms lasted longer and were more intense. WSM6_00 and WSM6_12 (Figure 3a) runs had the distinct feature that the predicted reflectivity appeared to be attributed to storms that happened in two parallel zones. Storm relative helicity (SRH), a parameter whose high values are closely related to intense thunderstorms (supercells), showed values greater than 350 m² s⁻² in the WSM6_12 run (Figure 3b). In the WSM6_00 run, the predicted SRH values were lower (but still close to 350 m² s⁻²) and had a rapid decay after their peak. ETA_12 and ETA_00 had a similar behavior with a swift decline of the SRH values after reaching peak values of more than 300 m² s⁻².

4. Conclusions

Supercells produce the most devastating convective weather and have the greatest impact on life and property. On September 7, 2024, some embedded short-wave troughs in the northwesterly flow brought unstable conditions over Greece, and three supercells developed in an environment with moderate CAPE but substantial wind shear. All of them maintained high reflectivity values of more than 65 dBZ and recorded distinct mesocyclone features for some periods of their lifetime, producing widespread damage along their track, mainly in the Thessaly region, where SC2 and SC3 persisted for several hours.

Although the data obtained from radars has improved the capability to detect signatures that characterize severe storms, it will be more useful to have the opportunity for early and accurate forecasts of supercells, and this was the goal of this study. Testing the possibility of forecasting supercell storms using the WRF-ARW model will help in making decisions concerning the severity of such storms and possible impacts.

The results showed that all simulations provided indications of the intense convective activity, although the forecasted potential shifted about two hours earlier. The two runs with an initialization time of 12 UTC gave better agreement concerning the intensity and duration of the storms, while the WSM6_12 provided better indications of the areas where supercells were most likely to occur. The predicted SRH, a parameter used to assess potential for storm rotation, indicated extreme values in good spatial agreement with the actual locations of the storms. More studies are necessary to optimize the performance of the WRF-ARW model in such severe weather conditions.