The damaging tornado in Luxembourg on 9 August 2019: towards better operational forecasts

WeatherVolume 76, Issue 8 p. 264-271 Research ArticleOpen Access The damaging tornado in Luxembourg on 9 August 2019: towards better operational forecasts Luca Mathias, Corresponding Author Luca Mathias luca.mathias@airport.etat.lu MeteoLux, Air Navigation Administration, Findel, Luxembourg *Correspondence to: L. Mathias luca.mathias@airport.etat.luSearch for more papers by this authorPatrick Ludwig, Patrick Ludwig Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology, Karlsruhe, GermanySearch for more papers by this authorJoaquim G. Pinto, Joaquim G. Pinto Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology, Karlsruhe, GermanySearch for more papers by this author Luca Mathias, Corresponding Author Luca Mathias luca.mathias@airport.etat.lu MeteoLux, Air Navigation Administration, Findel, Luxembourg *Correspondence to: L. Mathias luca.mathias@airport.etat.luSearch for more papers by this authorPatrick Ludwig, Patrick Ludwig Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology, Karlsruhe, GermanySearch for more papers by this authorJoaquim G. Pinto, Joaquim G. Pinto Institute of Meteorology and Climate Research, Karlsruhe Institute of Technology, Karlsruhe, GermanySearch for more papers by this author First published: 05 May 2021 https://doi.org/10.1002/wea.3979AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Graphical Abstract On 9 August 2019, a devastating tornado hit southwestern Luxembourg and caused widespread damage and many injuries. The atmospheric environment, in which the tornado occurred, was characterised by a moderate latent instability, very strong vertical wind shear and high storm-relative helicity. The radar images show typical characteristics of a tornadic supercell, that is, a hook echo and a well-defined velocity couplet. To assess the risk of tornadoes associated with supercells in the future, the Luxembourgish weather service (MeteoLux) elaborated an ingredients-based forecast concept. On 9 August 2019, a devastating tornado hit southwestern Luxembourg and caused widespread damage and many injuries, being one of the most severe convective weather events affecting Luxembourg in decades. We provide a thorough examination of the environmental conditions, which favoured the tornadogenesis, and an analysis of the parent supercell. The predictability is briefly investigated using operational numerical weather prediction model output. The atmospheric environment was characterised by a moderate latent instability, very strong vertical wind shear and high storm-relative helicity. The radar analysis shows that the tornadic supercell had typical characteristics, that is, a hook echo and a well-defined velocity couplet. It turns out that the updraft helicity as a forecast parameter in convection-resolving models provides useful guidance for assessing the risk of supercells and tornadoes during the operational workflow. Following this event, MeteoLux (the national weather service of Luxembourg) initiated a project to elaborate a concept for assessing and communicating the tornado risk associated with supercells. Introduction On the late afternoon of 9 August 2019, a supercell thunderstorm crossed the border region of northeastern France and the Grand Duchy of Luxembourg, producing a damaging tornado along its path. Severe tornadic wind damage was reported in Rodange, Lamadelaine, Pétange and Bascharage (Figure 1). For instance, roughly 400 trees and a total of 310 houses were damaged in Bascharage, 50 of which lost their roofs (Gemeng Käerjeng, 2019). At least 80 people had to be sheltered in hotels or other accommodation. Two seriously injured persons and 17 minor casualties are attributed to the tornado. The vortex lasted for about 15min and travelled a distance of 18 to 20km (Mathias, 2020; cf. Figure 1). The tornado was rated as F2+ based on the scale currently in development by a steering group lead by the European Severe Storms Laboratory (Groenemeijer et al., 2018), which corresponds to estimated maximum wind speeds of approximately 241kmh−1 (150mph). The estimated translation speed (speed of advance) of the tornado ranged between 17 and 19ms−1 and its maximum path width exceeded 500m (Mathias, 2020). In the aftermath of this extreme weather event, the total insured losses were estimated to be at least €100 million. Figure 1Open in figure viewerPowerPoint Topographic map of the investigation area (shaded orange in the inset on the upper-left-hand side). The analysed tornado track is denoted by the filled polygon with a dashed centre line. The automated weather station providing the data shown in Figure 5 is located in Rodange. The inset photograph of the tornado originates from a video, which was taken by an unknown author in Pétange (Luxemburger Wort, 2019). Recent studies have shown that tornadoes can be observed almost everywhere in Europe (e.g. Groenemeijer and Kühne, 2014; Antonescu et al., 2016; Antonescu et al., 2017). On average, 200 to 400 tornadoes are reported over the European land surface each year. Between 1950 and 2013, tornadoes caused 316 fatalities in Europe (Antonescu et al., 2017). Examples of strong tornadoes near Luxembourg are the tornado in the Belgian town Léglise on 20 September 1982 (Caniaux, 1984) and the tornado in the German city of Trier on 7 October 1988 (Trierischer Volksfreund, 2008). The aforementioned high impacts in southwestern Luxembourg motivate a detailed investigation of this hazardous weather event. Hence, we investigate the synoptic and mesoscale environment in which the tornado-producing thunderstorm formed using operational numerical weather prediction (NWP) model data. Moreover, measurements within the tornadic wind circulation from an automated weather station as well as radar and lightning observations are used to describe the evolution of the tornadic storm. The performance of two operational convection-resolving NWP models covering Luxembourg are also briefly discussed and suggestions for operational forecasters are given. The last section of this paper provides a summary and conclusion with a short description of the tornado warning process at MeteoLux, which was developed after this event. Synoptic-scale overview At 1200 utc on 9 August 2019, an upper-air ridge extended from Algeria over the Alps towards Denmark and an upper-level low was located near Ireland. This synoptic-scale pattern resulted in a strong southwesterly flow in the mid to upper troposphere over western Europe (Figure 2(a)). A cyclonically curved jet streak with maximum wind speeds of 65 to 70ms−1 was located over the Bay of Biscay and Brittany (France). Low vorticity values indicate no significant forcing for large-scale upward motion over northeastern France (Figure 2(a)). The mid- to upper-tropospheric low was associated with a deep surface low centred near the southwest coast of Ireland (Figure 2(b)). The corresponding cold front extended from the North Sea to the Iberian Peninsula. The air mass within the warm sector, which originated from the subtropical North Atlantic basin, was characterised by high moisture content, particularly near the prefrontal surface pressure trough over northern France, reaching total column water vapour values of up to 50mm (Figure 2(b)). As the cold front was orientated nearly parallel to the upper-level flow (cf. Figures 2(a) and (b)), it moved relatively slowly eastward and crossed Luxembourg between 1800 and 1900 utc. Figure 2Open in figure viewerPowerPoint ECMWF analysis of the synoptic-scale conditions on 9 August 2019 at 1200 utc over Western Europe. (a) 500hPa geopotential height (black lines; gpm), 300hPa wind speed (shaded; ms−1) and areas of 300hPa relative vorticity exceeding 0.00015s−1 are denoted by the dashed orange lines. (b) Mean sea level pressure (white lines; hPa), precipitable water (shaded; mm) and areas of 700hPa relative humidity exceeding 80% are denoted by the dashed magenta lines. The analysed location of the surface frontal boundaries by the German Weather Service (DWD) is superposed (blue line: cold front, red line: warm front, dark purple line: occluded front). When comparing this large-scale pattern to the one observed during the tornado outbreak in Western Europe in June 1967 (Dessens and Snow, 1989; Antonescu et al., 2020), similarities are identified regarding the location of the vertically stacked low pressure system and the presence of a prefrontal pressure trough superposed by a strong southwesterly flow in the middle troposphere. Contrarily to the recent case, the synoptic regime during the severe tornadic storm in northern France on 3 August 2008 was strongly forced given the close proximity of a short-wave trough and a mid-level jet streak (Tuschy, 2009; Wesolek and Mahieu, 2011). Mesoscale storm environment The low-level conditions were characterised by a prefrontal mesoscale low pressure area over northern France between 1200 and 1500 utc on 9 August 2019 (Figure 3). A subtropical air mass with high 925hPa wet-bulb potential temperatures ranging from 20 to 23°C (see figure 3(a) in Mathias, 2020) covered this area. To examine the dynamic and thermodynamic conditions of the pre-convective environment in the region where the tornadic thunderstorm developed and intensified (cf. dashed green outlined box in Figure 3), area-averaged vertical profiles at 1300 utc were computed based on the 1200 utc forecast runs of two different operational NWP models (Figure 4): Applications de la Recherche à l'Opérationnel à Mésoéchelle (AROME; Seity et al., 2011) operated by Météo-France and Consortium for Small-scale Modeling (COSMO-D2; Baldauf et al., 2018) operated by the German Weather Service (DWD). The profiles show a mean specific humidity of about 13gkg−1 in the lowest 500m, whereas the moisture is better mixed in the AROME profile. Furthermore, a conditionally unstable lapse rate between 850hPa and 600hPa is evident in both profiles (Figure 4). However, major differences of in the lapse rate are found near the surface, which impact the convective available potential energy (CAPE) and convective inhibition (CIN). The near-surface air is warmer in AROME than in COSMO-D2 due to less cloudiness (not shown), resulting in higher (lower) CAPE (CIN) values (cf. Figures 4(a) and (b)). Since AROME depicts the near-surface conditions more accurately than COSMO-D2 in the defined area (cf. near-surface temperature in Figure 4(a) with Figure 5(b) at 1300 utc), the tornadic storm likely encountered CAPE values between 600 to 900Jkg−1 and CIN values up to −40Jkg−1, as suggested by AROME (Figure 4(a)). We also considered the vertical sounding from Paris-Trappes (France) at 1200 utc, which was situated closer to the frontal boundary. However, this observed sounding is not representative for the low-level atmospheric conditions over northeastern France close to Luxembourg, that is, there is no veering of the wind below 700hPa, and the near-surface temperature was lower. Finally, low-level convergence zones in the surrounding area of the mesoscale surface low provided a strong lifting mechanism to overcome the initially strong CIN, thus completing the list of necessary ingredients for the development of deep moist convection (Johns and Doswell III, 1992). Figure 3Open in figure viewerPowerPoint Forecast of the pre-convective environment for 1300 utc on 9 August 2019 by the 1200 utc run of AROME. Mean sea level pressure (black lines; hPa), most-unstable CAPE (shaded; Jkg−1) and 0–6km shear vector (blue wind barbs; kts). The mesoscale low is denoted by the black ‘L’. The dashed green outlined box indicates the area considered for the vertical profiles shown in Figure 4. Figure 4Open in figure viewerPowerPoint Skew-T log-p diagram of an area-averaged vertical profile (area is denoted by the dashed green box in Figure 3) at 1300 utc on 9 August 2019 forecast by the 1200 utc run of (a) AROME and (b) COSMO-D2. The red (green) curve represents the temperature (dew point) and the dashed black curve represents the ascent trajectory of the most unstable parcel. The most-unstable CAPE (CIN) is indicated by the area shaded in transparent red (blue). The black arrow in the hodograph represents the storm motion of a right-moving supercell computed after Bunkers et al. (2000). The wind barbs are displayed in knots using standard notation. The lifting condensation level (LCL) marks the expected cloud base of a thunderstorm. Figure 5Open in figure viewerPowerPoint In situ measurements with a temporal resolution of 10min from the automated weather station located in Rodange (cf. Figure 1) between 1200 utc and 1800 utc on 9 August 2019. (a) Maximum wind gusts (purple line; kmh−1) and mean wind speed (orange line; kmh−1) during the preceding 10min (1kmh−1 = 0.278ms−1), and corresponding mean wind direction (yellow dots; °). (b) Instantaneous temperature (red line; °C) and dew point temperature (green line; °C) measured 2m above the ground. (c) Instantaneous surface pressure (black line; hPa). (d) Instantaneous relative humidity (blue line; %) measured 2m above the ground. The vertical shear of wind speed and direction is a crucial ingredient for well organised deep moist convection, with stronger shear tending to favour the occurrence of severe weather (e.g. Rasmussen and Blanchard, 1998; Púčik et al., 2015; Taszarek et al., 2017). In general, wind profiles at 1300 utc are similar for AROME and COSMO-D2 (cf. Figures 4(a) and (b)). Firstly, very strong 0–6km bulk shear with values slightly exceeding 25ms−1 overlapped with the prevailing latent instability over northeastern France and Luxembourg (Figures 3 and 4). However, most of the 0–6km bulk shear was concentrated in the layer below 700hPa, whereas the 0–1km bulk shear values were around 11ms−1 (Figure 4). In addition, strong veering of the wind with height was apparent below 700hPa (Figure 4), especially along the northern flank of the mesoscale surface low where easterly winds prevailed. Thus, large values of 0–3km storm-relative helicity 1 (200 to 300m2s−2) were present for both AROME and COSMO-D2 in the development area of the storm, though the curvature of the hodograph in the lowest 3km was more pronounced in AROME (cf. Figures 4(a) and (b)). Overall, the prevailing regime of moderate latent instability (CAPE between 500 and 1000Jkg−1), high bulk shear (>20ms−1 between 0 and 6km) and storm-relative helicity (>200m2s−2 between 0 and 3km) generally allows the development of supercell thunderstorms with a deep and persistently rotating updraft (mesocyclone), which may be capable of producing a tornado (e.g. Davies-Jones et al., 1990; Thompson et al., 2003). Due to enhanced values of 0–1km storm-relative helicity (100–150m2s−2), the tornado potential was indeed considerable over parts of northern France, eastern Belgium and Luxembourg (Figure 4). Values above 100m2s−2 have been associated with significant tornadoes in Europe (Taszarek et al., 2017; Taszarek et al., 2020). However, it is important to underline the potentially large variability of low-level helicity on a temporal and/or spatial scale (Markowski et al., 1998), meaning that the area-averaged 0–1km storm-relative helicity value at 1300 utc shown in Figure 4 is possibly not representative for the in situ environment near the tornadic storm. Moreover, the high relative humidity within the boundary layer corresponded to a relatively low cloud base height of 1000 to 1500m above mean sea level (cf. lifting condensation level (LCL) in Figure 4) and to a less negatively buoyant cold pool of the storm due to decreased evaporational cooling. These moist and strongly sheared conditions at lower levels tend to favour tornadogenesis under a sufficiently strong mesocyclone as shown by numerous studies (e.g. Markowski and Richardson, 2014; Coffer and Parker, 2017; Yokota et al., 2018). The aforementioned mesoscale tropospheric conditions compare reasonably well with the environment during the western European tornado outbreak in June 1967 (Dessens and Snow, 1989; Antonescu et al., 2020). However, in contrast to the violent tornado event in northern France in August 2008 (Tuschy, 2009; Wesolek and Mahieu, 2011), the helicity and shear values were significantly lower in the current case (this study). In situ measurements Figure 5 shows a sample of parameters measured at the surface within the tornadic environment, which was provided by an automated weather station located in Rodange in southwestern Luxembourg and operated by the private weather service Kachelmann Group. The tornado hit this weather station between 1530 utc and 1545 utc. Prior to the passage of the tornadic storm (1430 utc to 1530 utc), the temperature steadily decreased with the onset of precipitation while the dew point temperature remained between 20 and 21°C, resulting in an increase of the relative humidity to values of 80 to 90% (cf. Figures 5(b) and (d)). The pressure also decreased slightly due to the approach of the mesoscale low and weak winds blew from easterly directions (Figures 5(a) and (c)). When the tornadic storm hit the station, the temperature decreased by approximately 2 degC while the dew point temperature remained almost constant, hinting at a weak cold pool (Figure 5(b)). The winds veered rapidly to northwesterly directions and a peak gust of 35.5ms−1 (128kmh−1) was measured at 1540 utc (Figure 5(a)). Following Lee et al. (2004) and Yao et al. (2019), we assume that the relatively sharp pressure-drop observed at 1540 utc was partially caused by the short-term pressure perturbation of the tornadic wind circulation (Figure 5(b)). Storm cell analysis At 1200 utc on 9 August 2019, a major super-cellular thunderstorm developed to the southeast of Paris in northern France and moved east-northeastwards, producing a substantial amount of lightning flashes between the south of Luxembourg and the northern part of the German state Saarland by 1600 utc (Figure 6). This supercell produced hail and severe non-tornadic wind gusts along its path. Between 1200 and 1400 utc, a large-scale precipitation field with embedded and mostly weak convective cells formed in the adjacent northwest sector of the isolated supercell storm as reflected by the lightning density (Figure 6). Around 1440 utc, a second super-cellular storm formed slightly to the north (cf. red rectangle in Figure 6). At 1500 utc, a relatively broad rotational circulation within that storm became apparent in the radial velocity data at an altitude of about 2 to 3km (see figure 8 in Mathias, 2020). Subsequently, the 50dBZ echo top reached a height of approximately 7km by 1505 utc due to the rapid strengthening of the mesocyclonic updraft, coinciding with a significant increase of the lightning activity (Figure 6). Figure 6Open in figure viewerPowerPoint Lightning density (flashes per km2) measured by the European Cooperation for Lightning Detection (EUCLID; Schulz et al., 2016) network on 9 August 2019. The path of the tornadic supercell is denoted by the red rectangle. As the supercell reached the Franco-Belgian border at 1515 utc, the reflectivity scans revealed a V-shaped form of the precipitation field associated with the forward-flank downdraft of the storm (see figure 9(a) in Mathias, 2020). A well-defined mesocyclone is identified at lower levels in the velocity scans, corresponding to a bounded weak echo region (area of low radar reflectivity associated with an intense updraft; Lemon, 1977) from about 2.5 to 5km altitude in the reflectivity data (not shown). However, the lowest elevation scan did not show yet a clear rotational circulation. While the storm moved along the Franco-Belgian border producing a lot of lightning (Figure 6), a distinct hook echo showed up at the southern tip of the supercell in the low-level reflectivity scan at 1524 utc (Figure 7(a)). The radial velocity scan at the lowest elevation angle revealed a signature of rotation at an altitude of about 1km, although significant filtering is obvious in the centre of the circulation (Figure 7(b)). This may suggest that the lower part of the mesocyclone strengthened between 1515 and 1525 utc, tornadogenesis occurring during the following 10min. Figure 7Open in figure viewerPowerPoint Reflectivity (dBZ; upper row) and radial velocity (ms−1; lower row) measured at 1524 utc, 1534 utc and 1544 utc with an elevation angle of 0.5° by the meteorological radar located in Wideumont (49.9°N, 5.5°E; outside of the area shown) to the north-northwest of the tornadic storm and operated by the Royal Meteorological Institute of Belgium (RMIB). Negative velocities indicate a relative movement towards the radar and positive velocities indicate a relative movement away from the radar. The mesocyclonic circulation is indicated by the blue circle in (b), (d) and (f). The hook echo of the supercell storm reached the French town Longwy at about 1534 utc (Figure 7(c)), when the storm's tornadic phase was already ongoing. The mesocyclone was still evident at this time in the radial velocity data. The scan at the lowest elevation angle revealed an azimuthal shear couplet with inbound velocities of about 23ms−1 and outbound velocities of about 17ms−1, representing the base of the mesocyclone associated with the tornado at the surface. Since filtering was again very prominent in the centre of the vortex signature (Figure 7(d)), it is probable that the strongest part of the mesocyclonic circulation could not be analysed. Between 1540 and 1545 utc, the vortex reached the Luxembourg town of Bascharage. A clear velocity couplet was again visible in the imagery at an altitude of roughly 1km (Figure 7(f)). However, the supercell storm weakened at this time, as the lightning density decreased considerably (Figure 6), suggesting that the mesocyclonic updraft was collapsing and the tornado rapidly dissipated soon afterwards. The potential causes for the dissipation were the weakening of the low-level mesocyclone by 1550 utc and the occlusion of the rear-flank downdraft, which is indicated by the narrowing of the weak-echo slot associated with the hook echo (Figure 7(e)). The remnants of the supercell then passed slightly north of the capital Luxembourg City between 1555 and 1610 utc, producing only sporadic lightning flashes (Figure 6). Overall, the supercell storm travelled a distance of approximately 100km while producing the tornado at the end of its life cycle. Predictability Since tornadoes cannot be intrinsically forecast using operational NWP models, the following discussion about predictability is based on the forecast of instantaneous updraft helicity (Kain et al., 2008) and simulated radar reflectivity at 850hPa. The two considered forecast runs by AROME, initialised at 0000 utc and 0600 utc on 9 August 2019, feature strong convective cells over northeastern France, eastern Belgium and Luxembourg at 1500 utc (Figures 8(a) and (b)). The 0000 utc run suggested a cluster of four storms with supercellular characteristics (updraft helicity larger than 50m2s−2) moving northeastward over northern France (Figure 8(a)). The strongest supercell in this cluster showed updraft helicity values exceeding 350m2s−2 (Figure 8(a)). A more isolated, but weaker supercell is evident over far eastern Luxembourg (Figure 8(a)), with a strong updraft helicity of up to 200m2s−2 one hour before at 1400 utc near the Belgian town of Virton to the west of Luxembourg (cf. Figure 1). The 0600 utc AROME run also simulated widespread storms with embedded supercells over northern France (Figure 8(b)). By contrast, the COSMO-D2 model failed to initiate storms with a significant amount of updraft helicity (Figures 8(c) and (d)). While a few stronger convective cells are forecast over eastern Belgium at 1500 utc in the COSMO-D2 0600 utc run (Figure 8(d)), they do not show supercellular characteristics. The lack of storms over northeastern France and Luxembourg in both COSMO-D2 runs is likely related to high CIN (cf. Figure 4(b)) and/or weak low-level convergence. Hence, AROME provided in this particular case a better short-range guidance regarding the increased potential of supercells, despite being inconsistent with the spatial and temporal evolution of the storm cells. This suggests that a forecaster should consider multiple deterministic convection-resolving NWP models covering the forecast area or convection-resolving ensemble forecasts of a single model for the evaluation of such dangerous situations. Moreover, the updraft helicity is found to be a very useful forecast parameter to assess the risk of rotating updrafts in thunderstorms, which is in line with numerous studies (e.g. Kain et al., 2008; Antonescu et al., 2020). In particular, footprints of high hourly maximum updraft helicity can help to detect the potential of long-lived supercells. Clark et al. (2012) also pointed out that footprints of updraft helicity should always be considered when forecasting the potential of mesocyclonic tornadogenesis in combination with an analysis of the low-level environment. Figure 8Open in figure viewerPowerPoint Forecast of 850hPa simulated reflectivity (shaded in grey tones; dBZ) and updraft helicity integrated between 800 and 400hPa (shaded in colours; m2s−2) for 1500 utc on 9 August 2019 by the 0000 utc and 0600 utc runs of (a), (b) AROME and (c), (d) COSMO-D2. Summary and conclusions The synoptic-scale and mesoscale meteorological atmospheric ingredients for the impactful tornado on 9 August 2019 in southwestern Luxembourg were investigated. Furthermore, the evolution of the tornadic storm cell was thoroughly analysed, and a brief discussion of the predictability is presented in this study. The atmospheric setting was very conducive to the development of discrete and long-lived supercell thunderstorms over northeastern France, eastern Belgium, Luxembourg and western Germany, where moderate latent instability and strong vertical wind shear coincided ahead of a cold front. Moreover, a well-defined prefrontal mesoscale surface low was associated with high values of the storm-relative helicity in the lowest 3km at its northern flank. Together with high values of absolute and relative humidity in the boundary layer overlapping with 0–1km storm-relative helicity above 100m2s−2 in some areas, favourable lower-tropospheric conditions for mesocyclonic tornadogenesis existed. In situ measurements from a weather station in the tornado environment indicated contamination of the relative humidity in the boundary layer by preceding precipitation, which might have played a key role in conditioning the environment for tornadogenesis to some extent. The radar analysis of the right-moving tornadic supercell revealed a long-lived mesocyclone, which strengthened while the storm was moving towards Luxembourg. The intensification phase of the mesocyclonic updraft was accompanied by a strong increase of the lightning activity. The storm cell also exhibited a well-defined hook echo in the high-resolution radar imagery after the mesocyclone had strengthened. The formation of the hook echo signature preceded the tornadogenesis by about 10min. Reasons for the tornadolysis are proposed, namely the occlusion of the rear-flank downdraft and the collapse of the mesocyclonic updraft. It was also shown that convection-resolving NWP models like AROME can forecast the formation of supercell storms within a certain region, and thus hint at the associated risk of mesocyclonic tornadogenesis. Furthermore, footprints of updraft helicity from operational model output may be very important for forecasters to identify supercells simulated by the model. The rapid passage of the tornado over Rodange, Lamadelaine, Pétange and Bascharage was one of the most damaging convective weather events in the Grand-Duchy of Luxembourg of recent decades. This hazardous event also highlights the necessity for further efforts regarding operational tornado forecasting and warnings in Europe, as pointed out by Rauhala and Schultz (2009) and Antonescu et al. (2017, 2018). Hence, in the aftermath of this event, MeteoLux elaborated an ingredients-based forecast concept for assessing the tornado risk associated with supercells during operational workflow within the short-term. This concept is mainly based on the output of NWP models and considers the updraft helicity as a key variable in addition to the ingredients-based forecast method. If a significant risk is forecast in the short-term (similar to the ‘tornado watch’ in the United States), the High Commission for National Protection will be informed via a web-based emergency management tool. MeteoLux will also continue to adapt the forecast concept in the future based on experience and new scientific insights on severe convective storms and tornadoes. Acknowledgements The authors want to thank ECMWF, DWD and Météo-France for the provision of NWP model data. The authors also thank Laurent Delobbe from RMIB for providing radar data. The authors would like to thank Stephan Thern from Siemens AG for providing EUCLID lightning data. The authors also thank the Kachelmann Group for archiving weather station data and making it freely available on its website (https://kachelmannwetter.com). Joaquim G. Pinto was supported by the AXA Research Fund (https://axa-research.org/en/project/joaquim-pinto). Patrick Ludwig was partially funded by the Helmholtz Climate Initiative REKLIM (regional climate change; https://www.reklim.de/en). Finally, the authors thank two anonymous reviewers for their constructive comments that helped to improve the manuscript. Endnote 1 Storm-relative helicity is a measure for the potential of updraft rotation in a thunderstorm, which is related to the amount of streamwise vorticity in the inflow of a storm (Davies-Jones, 1984). References Antonescu B, Schultz DM, Lomas F et al. 2016. Tornadoes in Europe: synthesis of the observational datasets. Mon. Weather Rev. 144: 2445– 2480. Antonescu B, Schultz DM, Holzer A et al. 2017. Tornadoes in Europe: an underestimated threat. Bull. Am. Meteorol. Soc. 98: 713– 728. Antonescu B, Fairman JG, Schultz DM. 2018. What is the worst that could happen? Reexamining the 24–25 June 1967 Tornado Outbreak over Western Europe. Weather Clim. Soc. 10: 323– 340. Antonescu B, Púčik T, Schultz DM. 2020. Hindcasting the First Tornado Forecast in Europe: 25 June 1967. Weather Forecast. 35: 417– 436. Baldauf M, Gebhardt C, Theis S et al. 2018. Beschreibung des operationellen Kürzestfristvorhersagemodells COSMO-D2 und COSMO-D2-EPS und seiner Ausgabe in die Datenbanken des DWD. https://www.dwd.de/DE/leistungen/modellvorhersagedaten/cosmo_d2_eps__documentation.pdf?__blob=publicationFile&v=2 (accessed 9 March 2021). Bunkers MJ, Klimowski BA, Zeitler JW et al. 2000. Predicting supercell motion using a New Hodograph technique. Weather Forecast. 15: 61– 79. Caniaux G. 1984. Trombes et chutes de grêle du 20 septembre 1982 sur les Ardennes. Direction de la météorologie nationale, notes techniques no. 8, Paris. http://pluiesextremes.meteo.fr/france-metropole/IMG/sipex_pdf/1982_09_20_trombre_grele_ardennes.pdf (accessed 9 March 2021). Clark AJ, Kain JS, Marsh PT et al. 2012. Forecasting tornado pathlengths using a three-dimensional object identification algorithm applied to convection-allowing forecasts. Weather Forecast. 27: 1090– 1113. Coffer BE, Parker MD. 2017. Simulated supercells in nontornadic and tornadic VORTEX2 environments. Mon. Weather Rev. 145: 149– 180. Davies-Jones RP. 1984. Streamwise vorticity: the origin of updraft rotation in supercell storms. J. Atmos. Sci. 41: 2991– 3006. Davies-Jones RP, Burgess DW, Foster M. 1990. Test of helicity as a tornado forecast parameter. Preprints, 16th Conference on Severe Local Storms, Kananaskis Park, AB, Canada. American Meteorological Society, pp. 588– 592. Dessens J, Snow JT. 1989. Tornadoes in France. Weather Forecast. 4: 110– 132. Gemeng Käerjeng. 2019. Eise Magazin - Spezialausgabe Tornado. https://kaerjeng.lu/wp-content/uploads/2019/10/eise-magazin_SPECIAL_09-2019-BAT.pdf (accessed 9 March 2021). Groenemeijer P, Kühne T. 2014. A climatology of tornadoes in Europe: results from the European Severe Weather Database. Mon. Weather Rev. 142: 4775– 4790. Groenemeijer P, Holzer AM, Hubrig M et al. 2018. The International Fujita (IF) Scale - Tornado and Wind Damage Assessment Guide. Draft version 0.1, 48 pp. https://www.essl.org/media/publications/IF-scale_v0.10.pdf (accessed 9 March 2021). Johns RH, Doswell CA III. 1992. Severe local storms forecasting. Weather Forecast. 7: 588– 612. Kain JS, Weiss SJ, Bright DR et al. 2008. Some practical considerations regarding horizontal resolution in the first generation of operational convection-allowing NWP. Weather Forecast. 23: 931– 952. Lee JJ, Samaras TM, Young CR. 2004. Pressure measurements at the ground in an F-4 tornado. 22nd Conference on Severe Local Storms, Hyannis, MA, USA. https://ams.confex.com/ams/pdfpapers/81700.pdf (accessed 9 March 2021). Lemon LR. 1977. New severe thunderstorm radar identification techniques and warning criteria: a preliminary report. NOAA Technical Memorandum NWS NSSFC-1. ftp://ftp.library.noaa.gov/noaa_documents.lib/NWS/NWS_NSSFC/TM_NWS_NSSFC_1.pdf (accessed 9 March 2021). Luxemburger Wort. 2019. 100 Häuser in Petingen von Tornado abgedeckt. https://www.wort.lu/de/lokales/100-haeuser-in-petingen-von-tornado-abgedeckt-5d4da009da2cc1784e3496d6 (accessed 9 March 2021). Markowski PM, Richardson YP. 2014. The influence of environmental low-level shear and cold pools on tornadogenesis: insights from idealized simulations. J. Atmos. Sci. 71: 243– 275. Markowski PM, Straka JM, Rasmussen EN et al. 1998. Variability of storm-relative helicity during VORTEX. Mon. Weather Rev. 126: 2959– 2971. Mathias L. 2020. Tornado in south-western Luxembourg on 9 August 2019: meteorological context and damage assessment. https://www.meteolux.lu/fr/filedownload/418/tornado20190809_report_final.pdf/type/pdf (accessed 9 March 2021). Púčik T, Groenemeijer P, Rýva D et al. 2015. Proximity soundings of severe and nonsevere thunderstorms in central Europe. Mon. Weather Rev. 143: 4805– 4821. Rasmussen EN, Blanchard DO. 1998. A baseline climatology of sounding-derived supercell and tornado forecast parameters. Weather Forecast. 13: 1148– 1164. Rauhala J, Schultz DM. 2009. Severe thunderstorm and tornado warnings in Europe. Atmos. Res. 93: 369– 380. Schulz W, Diendorfer G, Pédeboy S et al. 2016. The European lightning location system EUCLID – Part 1: performance analysis and validation. Nat. Hazards Earth Syst. Sci. 16: 595– 605. Seity Y, Brousseau P, Malardel S et al. 2011. The AROME-France Convective-Scale operational model. Mon. Weather Rev. 139: 976– 991. Taszarek M, Brooks HE, Czernecki B. 2017. Sounding-derived parameters associated with convective hazards in Europe. Mon. Weather Rev. 145: 1511– 1528. Taszarek M, Allen JT, Púčik T et al. 2020. Severe convective storms across Europe and the United States. Part II: ERA5 environments associated with lightning, large hail, severe wind, and tornadoes. J. Clim. 33: 10263– 10286. Thompson RL, Edwards R, Hart JA et al. 2003. Close proximity soundings within supercell environments obtained from the rapid update cycle. Weather Forecast. 18: 1243– 1261. Trierischer Volksfreund. 2008. 20 Jahre danach - Wirbelsturm über Trier. https://tornadoliste.de/bilder/1988/081007tv.pdf (accessed 9 March 2021). Tuschy H. 2009. Examination of severe thunderstorm outbreaks in Central Europe. MSc thesis. Institute of Meteorology and Geophysics, University of Innsbruck, Austria. https://www.uibk.ac.at/acinn/theses/diploma-theses/tuschy_helge_2009_dipl.pdf (accessed 9 March 2021). Wesolek E, Mahieu P. 2011. The F4 tornado of August 3, 2008, in Northern France: case study of a tornadic storm in a low CAPE environment. Atmos. Res. 100: 649– 656. Yao D, Meng Z, Xue M. 2019. Genesis, maintenance and demise of a simulated tornado and the evolution of its preceding Descending Reflectivity Core (DRC). Atmosphere 10: 236 Yokota S, Niino H, Seko H www.essl.org/media/publica. 2018. Important factors for tornadogenesis as revealed by high-resolution ensemble forecasts of the Tsukuba Supercell Tornado of 6 May 2012 in Japan. Mon. Weather Rev. 146: 1109– 1132. Volume76, Issue8August 2021Pages 264-271 FiguresReferencesRelatedInformation