A total of 13 commercial airplanes were struck by lightning in October (10 in 1 day) and December (3 on 3 separate days) 2011 in the main Finnish Helsinki–Vantaa airport. The number of lightning-struck airplanes is extremely large, considering the time of year and the small number of flashes by the storms. This paper indicates the characteristics of these cases regarding the synoptic situation as well as their forecasting. There were remarkable differences in the operational models; the high-resolution nonhydrostatic model was superior in predicting the convective nature of the event compared to the coarser-resolution hydrostatic model. The interview of the pilots of the struck airplanes shows that the pilots did not receive detailed information to avoid the situation; also, the lightning strike affected the pilots, even causing temporary loss of sight and hearing. Luckily, no fatalities or severe damage to the airplanes occurred. The most interesting case is 19 October 2011; during this single day, a total of 10 airplanes were struck. The analysis suggests that a major cause for the large number of struck airplanes is that the planes took off directly into the convective core of the storm and the planes initialized the flashes themselves. However, the time of the year, the near position of the storm area relative to the takeoff path, and the necessity to use only a certain takeoff path because of the direction of wind makes the convective scenario difficult to predict and avoid. The pilots have expressed interest in receiving training for these cold-season thunderstorms.

Detail of Fig. 1. See p. 848 for more information.

Detail of Fig. 1. See p. 848 for more information.

One day in October 2011, 10 commercial planes took off into a convective storm in Finland, triggering lightning that temporarily blinded some of the pilots—demonstrating the need to improve warnings for this unseasonable weather.

It is estimated that every commercial airplane is struck on average once per year by lightning (Uman and Rakov 2003), and several studies have been focused on the “triggering effect” of an airplane to lightning (Clifford and Kasemir 1982; Mazur 1989; Moreau et al. 1992). According to Rakov and Uman (2005), an airplane hit is typically a single event, and only rarely are several planes hit within the same storm, for example, in Los Angeles on 24 February 1987 when at least six airplanes were hit within only a couple of hours. We will show a similar case with 10 hits during a single evening.

Despite the high peak current of a lightning flash, it is a very rare case that an airplane is severely damaged by the flash (Cherington and Mathys 1995); the lightning protection system of the airplane prevents the lightning current from entering the critical parts—say, fuel tanks—of the plane. However, minor damages, such as small holes, are reported (Plumer and Robb 1982; Uman and Rakov 2003).

Maybe the most famous and important, regarding the development of safety regulations, airplane accident by lightning occurred on 8 December 1963 in Maryland (Civil Aeronautics Board 1965); PanAm flight 214 exploded and crashed when gasoline fumes were ignited by a lightning flash, killing all 81 people onboard. After the incidence, the Federal Aviation Administration (FAA) regulated all airplanes flying in the U.S. airspace to have special lightning rods installed.

Other reported lightning-caused incidents include the fuel tank explosion of the Iranian commercial aircraft over Spain in 1976 (National Transportation Safety Board 1976) with the death of 17 persons; the Lineas Aéreas Nacionales Sociedad Anonima (LANSA) flight 508 in Peru in 1971, which caused the death of about 90 persons; an accident in Germany in 1988, when an aircraft lost its wing after being struck by lightning (21 casualties); and the glider accident of 1999 in Bedfordshire, United Kingdom. In the latter case, the glider was apparently literally blown apart by a high-peak current positive ground flash (AAIB 1999).

On 19 October 2011, several commercial aircrafts were struck by lightning during their approach or departure in the surroundings of the terminal control area (TMA) of Helsinki–Vantaa airport in southern Finland. Ten aircrafts, of three different aircraft types, reported a lightning strike and several departing aircrafts had to return to the airport because of a minor technical problem or because of a momentary blindness or deafness. Some of the pilots reported the whole windshield to be illuminated by electricity (St. Elmo's fire). A photograph showing one of the hits is depicted in Fig. 1. In December 2011, there were three more incidences.

Fig. 1.

Lightning strike to an airplane near Helsinki– Vantaa airport at 1736:32 UT C 19 Oct 2011. Pilot's comment: “It looked like a bucketload of sparkles had been thrown to my cockpit window.” (Courtesy: Pavel Shatylovich.)

Fig. 1.

Lightning strike to an airplane near Helsinki– Vantaa airport at 1736:32 UT C 19 Oct 2011. Pilot's comment: “It looked like a bucketload of sparkles had been thrown to my cockpit window.” (Courtesy: Pavel Shatylovich.)

The continuation of the thunderstorm season up to December was caused by a very mild and humid weather pattern. Because of cool southwesterly airstream from the Baltic Sea, the air mass gained energy from the warm sea surface. Air masses were not particularly unstable, but the upper troposphere was cold. Therefore cumulonimbus (Cb) towers were comparatively thin (their tops at 6–7 km AGL). Because of the relatively warm sea surface temperatures and coastal convergence in southwesterly flow, conditions were favorable for forced lifting.

Wintertime thunderstorms, defined as lightning-producing storms occurring between October and April with a ground temperature of 0°C or below, in Finland have been described by Rinne (2009). The results show that thunderstorms are possible almost every month in the wintertime with minor maximums in November and February. Most of the winter the thunderstorms occur within the occlusion front in warm advection, with convection available potential energy (CAPE) values near zero. In almost all of the cases, the sea areas were open without ice cover.

A 30-yr climatology of thundersnow events in the contiguous United States was studied by Market et al. (2002). Their results show the typical characteristics, including the synoptic environment, wind directions, and surface and dewpoint temperatures, of the events. Comparing their results to those of Rinne (2009) mentioned above, it seems that a thundersnow event is often a very localized and short-lived phenomenon. However, because of the relatively high temperatures (e.g., in December +3°–8°C) in the cases analyzed in this paper, our cases here cannot be qualified as winter thunderstorms (or thundersnow) but rather as cold-season thunderstorms.

The variability of wintertime weather conditions is large in Finland. Basically, whenever there is a wide low pressure area west of Scandinavia and a mild southwesterly flow prevailing in Finland, the conditions may be favorable for thunderstorms. This is especially true if there is an upper trough or frontal system coming from the southwest. During mild winters such situations are quite common. However, during a cold anticyclonic weather type (the “real winter conditions”), there may be long periods without any potential for cold-season thunderstorms.

The average number of thunderstorm days during the cold season (October–April) is shown in Fig. 2. Most of these storm days occur in early October, when the sea is warm. Interestingly, a maximum is found over the mainland near Helsinki; one explanation for this maximum is that in that area stands one of the highest radio masts of Finland, the Kivenlahti mast (325 m). This suggests that the mast increases the number of thunderstorm days during the cold season. Apparently, the physical explanation is that when convective clouds, not yet producing lightning, move over the Kivenlahti mast, the mast triggers a flash; this effect has resemblance to the lightning ignition by an airplane in favorable conditions.

Fig. 2.

Average number of cold-season thunderstorm days in 2002–11. The unit is thunderstorm days per cold season. The x- and y-axis values are kilometers to the east and north, respectively, based on the Finnish Uniform Coordinate System.

Fig. 2.

Average number of cold-season thunderstorm days in 2002–11. The unit is thunderstorm days per cold season. The x- and y-axis values are kilometers to the east and north, respectively, based on the Finnish Uniform Coordinate System.

Schultz (1999) investigated lake-effect snowstorms with and without lightning in two locations, northern Utah and western New York. The results show that the most useful parameters for forecasting lightning during lake-effect snowstorms are low-tropospheric temperatures and lifted index. The cases with lightning have substantially higher temperatures and dewpoints in the lower troposphere and lower lifted indices than the cases without lightning. The CAPE, which is often used as an indicator of warm-season thunderstorm potential, was not a useful predictor of lightning during wintertime lake-induced snowstorms.

In Finland, which is situated in northern Europe between latitudes 60° and 70°N, the thunderstorm season is highly concentrated in the summertime (May–September). Thunderstorms occur also outside this period, but their effect to the annual sum of flashes is practically negligible; however, they do affect the number of thunderstorm days, because a single flash is enough for a thunderstorm day. Cold-season thunderstorms are especially interesting because of the following two reasons:

  • 1) Their forecasting is difficult because the warning algorithms are designed for summertime convection.

  • 2) People are less prepared for winter lightning because of the rarity of the phenomenon.

Regarding reason 2, cold-season thunderstorms may be even more dangerous than summer storms because of their infrequent and sudden nature. The same conclusion can be found in Gough et al. (2009; summarized in Hemink et al. 2010), who showed statistics and prediction methods of cold-season thunderstorm around the Schiphol international airport in Amsterdam, the Netherlands. Their study indicates that actually most of the lightning encounters by airplanes occur during the cold season (October– April), although the warm season (May–September) is clearly the most abundant period of thunderstorms and lightning. According to Gough et al. (2009), the situation is the same in the United States.

The motivation of this paper is to show the statistics of cold-season thunderstorms in southern Finland in 2011, which caused 13 hits to commercial airplanes near the Helsinki–Vantaa airport; 10 of the hits occurred during a single thunderstorm. The hit percentage is large, considering the overall low number of occurred strokes, about 130 per day within 50 km from the airport. Although no serious damages occurred, the incidents raise two fundamental questions:

  • 1) What was the primary cause for the large number of the lightning-strike incidences in 2011?

  • 2) How could have the incidences been prevented?

We examine these questions with the available meteorological observations, and with the information obtained by interviewing the pilots of the lightning-struck airplanes.

MATERIALS AND METHODS.

Lightning location system.

The Lightning Location System (LLS) of the Finnish Meteorological Institute (FMI) is part of the Nordic Lightning Information System (NORDLIS; Mäkelä et al. 2010); each participating country shares the raw sensor data from all of the NORDLIS sensors and processes the lightning data independently. The system detects primarily ground-to-ground strokes in the low-frequency (LF) domain. However, some of the located events are classified as cloud flashes, according to the peak-to-zero time of the lightning waveform (Schulz et al. 2005; Mäkelä et al. 2010).

The estimated flash detection efficiency of NORDLIS in southern Finland is above 90%, and the median location accuracy is about 500 m (Mäkelä et al. 2010). The relatively good performance and large coverage of NORDLIS is essentially due to the Nordic cooperation; without it, for example, the FMI LLS would have much smaller coverage and poorer efficiency.

Weather radar.

The main weather radar serving Helsinki Vantaa airport is the Vantaa radar, located 8 km from the end of runway. It is a C-band dual-polarization radar; technical details and the measurement program are described in Saltikoff and Nevvonen (2011). Location of the radar is indicated in Fig. 3. The aviation forecaster has several radar products available, and a subset of them has been used in this study: constant level reflectivity image [constant altitude plan position indicator (CAPPI]) at 500-m altitude, maximum height of +20-dBZ threshold (TOPS; labeled as “risk of thunder” for the aviation users), vertical profile of average wind and reflectivity in 30-km cylinder around the radar [volume velocity processing (VVP)], and a range– height indicator (RHI) north and south of the radar. The TOPS product indicates the intensity of the updraft, therefore also showing the higher possibilities for the production of lightning. The TOPS product has been used for years with success in the nowcasting of thunderstorms in Finland. Hydrometeor classification based on dual-polarization parameters is shown on vertical RHI and conical plain position indicator (PPI) surfaces.

Fig. 3.

(left) Finland and the surrounding areas and a (right) zoom-in image showing the position of the Helsinki– Vantaa airport, Vantaa weather radar, and a 25-km-radius range circle around it, the Jokioinen atmospheric sounding station, and the located lightning (crosses) on 19 Oct 2011.

Fig. 3.

(left) Finland and the surrounding areas and a (right) zoom-in image showing the position of the Helsinki– Vantaa airport, Vantaa weather radar, and a 25-km-radius range circle around it, the Jokioinen atmospheric sounding station, and the located lightning (crosses) on 19 Oct 2011.

Mesoscale analysis system.

FMI operates the Local Analysis and Prediction System (LAPS; http://laps.fsl.noaa.gov/; Albers et al. 1996; Toth et al. 2011) for production of 3D analysis fields of different weather parameters. Within LAPS observations are fitted to the coarser first-guess background field from the global numerical weather prediction model of the European Centre for Medium-Range Weather Forecasts (ECMWF) by using mainly a multiscale successive correction method, and high-resolution topographical datasets are taken into account while creating the final high-resolution analysis fields.

Pilot interview background

A few days after the incidences on 19 October, FMI was in contact with the airlines and pilots of the lightning-struck airplanes. Discussions revealed that the case was indeed extraordinary and that it should be investigated further. Also, many of the pilots informed the authors of their interest to receive more information regarding the synoptic situation in order to be better prepared if something similar happens in the future. Therefore, it was decided that the FMI would collect information and feedback via a questionnaire from the pilots. A total of 6 replies out of 10 were received. The questions and answers are shown in the “Results” section.

Numerical weather prediction (NWP)

FMI uses two operational limited-area NWP models. High-Resolution Limited-Area Model (HIRLAM; Undén et al. 2002) is a hydrostatic primitive equation model covering all of Europe with a 16.5-km grid size. In addition to traditional larger-scale models, FMI operates the nonhydrostatic mesoscale NWP model HIRLAM–Aire Limitée Adaptation Dynamique Développement International (ALADIN) Research on Mesoscale Operational NWP in Euro–Mediterranean Partnership (HARMONIE). This model can provide high-resolution precipitation data, both in space (2.5 km) and time (hourly or less), and it has a more detailed description of precipitation physics (Seity et al. 2011) than previous NWPs, empowering better simulations of heavy rainfall episodes (Niemelä 2009; Bengtsson and Niemelä 2008; Niemelä et al. 2007). Furthermore, HIRLAM is designed to be used in scales where all convective flow structures need to be parameterized (Kain and Fritch 1990; Kain 2004). In HARMONIE the deep convective flow structures are assumed to be resolved explicitly (kilometer scale), leaving the parameterization problem only for nonprecipitating shallow convection (Siebesma et al. 2007).

Other

Upper-air soundings from Jokioinen, Finland, were used to analyze low-tropospheric winds and shear, and to derive the vertical temperature differences and different stability indices in the studied cases. The sounding closest to the event in time was used. The Jokioinen observation and sounding station is located about 100 km northwest of the Helsinki–Vantaa airport (Fig. 3).

RESULTS.

In this section we first show the general synoptic scenario of all four cases of 2011: 19 October, and 4, 14, and 26 December. Then we concentrate on the case of 19 October, which is the most interesting regarding the influence it had on the Helsinki–Vantaa airport.

Synoptic situation of cold-season thunderstorms in 2011.

Table 1 shows information about four thunderstorm events that occurred late 2011. The first one (19 October 2011) was the most active, with lightning activity lasting for several hours during the late afternoon and evening along a southwest– northeast-oriented area passing the Helsinki–Vantaa airport (Fig. 3). The synoptic setting of this case is described more thoroughly in the section “Case 19 October 2011.”

Table 1.

Four convective events with lightning in southern Finland during late autumn and early winter in 2011. (a) Data from Jokioinen upper-air sounding station (source: the University of Wyoming), (b) surface observations at Helsinki–Vantaa airport (complemented with the most severe MET AR report), (c) detected lightning strokes, and (d) Vantaa radar data within a 100-km radius: maximum height of +20-dBZ isoline (“thunderstorm risk indicator”) and the largest reflectivity (dBZ).

Four convective events with lightning in southern Finland during late autumn and early winter in 2011. (a) Data from Jokioinen upper-air sounding station (source: the University of Wyoming), (b) surface observations at Helsinki–Vantaa airport (complemented with the most severe MET AR report), (c) detected lightning strokes, and (d) Vantaa radar data within a 100-km radius: maximum height of +20-dBZ isoline (“thunderstorm risk indicator”) and the largest reflectivity (dBZ).
Four convective events with lightning in southern Finland during late autumn and early winter in 2011. (a) Data from Jokioinen upper-air sounding station (source: the University of Wyoming), (b) surface observations at Helsinki–Vantaa airport (complemented with the most severe MET AR report), (c) detected lightning strokes, and (d) Vantaa radar data within a 100-km radius: maximum height of +20-dBZ isoline (“thunderstorm risk indicator”) and the largest reflectivity (dBZ).

The first three cases (19 October, and 4 and 14 December) were quite similar with a wide low pressure area west of Finland and a southwesterly airstream prevailing in southern Finland. The upperair lapse rates show surprisingly similar values, for example, the temperature difference between surface and 700 hPa being 17°–18°C. These values are equivalent to those found by Schultz (1999) during wintertime thunderstorm events in western New York. The lifted index (LI) and total totals index (TOTL) for the three cases are quite similar, indicating high probability of showers and thunder (especially the TOTL values); the CAPE values were very low, which is also in good agreement with the results by Schultz (1999). The temporal evolution of the TOTL index and the vertical wind profile for two of the cases according to LAPS are shown in Fig. 4. The left panel contains all the cases; however, it is complemented with a nonthundery case from 27 December 2011, when organized convection was formed but without lightning. The cases seem to be organized into two groups, the one showing potential risk for thunderstorms (TOTL values above 50) and the other showing slightly lower values.

Fig. 4.

(left) Stability index (total totals) at Vantaa airport. Values above 50 indicate risk of moderate–severe thunderstorm. (right) LAPS wind profiles at Vantaa airport. Dashed and solid lines show profiles for 1500 UT C 19 Oct 2011 and 0300 UT C 26 Dec 2011, respectively.

Fig. 4.

(left) Stability index (total totals) at Vantaa airport. Values above 50 indicate risk of moderate–severe thunderstorm. (right) LAPS wind profiles at Vantaa airport. Dashed and solid lines show profiles for 1500 UT C 19 Oct 2011 and 0300 UT C 26 Dec 2011, respectively.

The fourth case, 26 December, was somewhat different: the lapse rates and stability indices show less instability than in the three previous cases. In this case, the thunderstorm was probably associated with the passage of a trough, preceded by warm advection and very strong vertical wind shear in the lower troposphere, which can also be seen in the analyzed wind profile from LAPS (Fig. 4). Based on the 0000 UTC 26 December sounding from Jokioinen, the surface and 850-hPa wind directions and speeds were 210° at 9 m s−1 and 240°at 34 m s−1, respectively. This situation resulted in stormy wind gusts at the surface later in the day, causing a lot of forest damage and electricity cuts. The common feature in all four cases is that a moderate or brisk southwesterly wind prevailed (Table 1, winds at Helsinki–Vantaa airport), advecting heat and moisture from the relatively warm Gulf of Finland toward the inland, thus promoting convective development.

Case 19 October 2011.

The 1800 UTC synoptic weather map and weather radar CAPPI image from southern Finland on 19 October are shown in Fig. 5. A wide low pressure area was located over Scandinavia, moving slowly northeast. A quite-strong southwesterly airstream prevailed in Finland, surface winds being around 10 m s−1 at the southwestern coastal areas and the 850/700-hPa winds being almost 20 m s−1 based on the Jokioinen sounding (Table 1) and LAPS analysis (Fig. 4). The incoming air mass had a long fetch over the relatively warm water (10°–13°C) of the Baltic Sea, resulting in a couple of degrees higher surface air temperatures in Finland's southwestern coast than farther inland. At upper levels, a tongue of cold air had pushed in over southern Scandinavia and western Finland. Based on the Jokioinen sounding, the temperature at the 500-hPa level was about −31°C, resulting in a vertical temperature difference of 38°C between the surface and 500-hPa level (Table 1). The relative humidity was high (80%–100%) up to the 650-hPa level and above that level the air was drier. Circumstances were quite favorable for convective development, which was indicated, for example, by the high TOTL index value, 59.2, based on the Jokioinen sounding and LAPS analysis. During the day, well-organized southwest–northeast-oriented convective lines formed over the western Gulf of Finland and at the coast west of the city of Helsinki. The individual convective cells moved to the northeast by the midtropospheric flow, while the area of heaviest convection moved very slowly east during the evening. (An animation is available as supplemental material online at http://dx.doi.org/10.1175/BAMS-D-12-00039.2.)

Fig. 5.

Synoptic situation in northern Europe at 1800 UT C 19 Oct 2011 (analysis by FMI). (top left) Weather radar inset from southern Finland (rectangle) at 1800 UTC.

Fig. 5.

Synoptic situation in northern Europe at 1800 UT C 19 Oct 2011 (analysis by FMI). (top left) Weather radar inset from southern Finland (rectangle) at 1800 UTC.

In the top panel of Fig. 6, we see the isolines of radar reflectivity averaged in a 30-km cylinder around the Vantaa radar; the bottom panel indicates the 15-min lightning rate within 25 km from the airport. The 20-dBZ isoline corresponds to a product used for thunderstorm warning, and it rises before a thunderstorm is reported in the aviation routine weather reports (METARs). Similar behavior was observed in the other cases, too. The lightning data show three peaks, at about 1415, 1700, and 1900 UTC. It is highly possible that some of the reported lightning strikes to the airplanes (black stars in the bottom panel) were actually triggered by the plane itself, like the flash (two strokes) at 1530–1545 UTC.

Fig. 6.

(top) Average reflectivity in 30-km cylinder around Vantaa radar 19 Oct and isolines at intervals of 5 dBZ. Height of +20 dBZ (red) is used as thunder indicator, when it reaches 6 km in summer. Interval +15–20 dBZ is shaded. (middle) Prevailing weather at the airport: SHR A—rain showers, TSR A—thunderstorm, TSGS —with graupel. (bottom) Lightning flash rate (strokes per 15 min). Stars indicate times when planes were hit.

Fig. 6.

(top) Average reflectivity in 30-km cylinder around Vantaa radar 19 Oct and isolines at intervals of 5 dBZ. Height of +20 dBZ (red) is used as thunder indicator, when it reaches 6 km in summer. Interval +15–20 dBZ is shaded. (middle) Prevailing weather at the airport: SHR A—rain showers, TSR A—thunderstorm, TSGS —with graupel. (bottom) Lightning flash rate (strokes per 15 min). Stars indicate times when planes were hit.

The hydrostatic model HIRLAM and the nonhydrostatic mesoscale model HARMONIE show a large difference in their analysis (Fig. 7). HIRLAM indicates only some cloud water in the lower levels and is not able to predict well the vertical structure of the storm, while HARMONIE seems to capture it largely in the same way as the weather radar sees it (Fig. 7).

Fig. 7.

The most intensive cell around 1830 UT C. (from the left) Cloud condensate in HIRL AM, clouds and hydrometeors in HARMONIE, radar reflectivity, and hydrometeor classification in radar (gray values indicate echoes classified as nonmeteorological). For each panel, width is 60 km and height is 10 km. The vertical axis is linear in radar images, but it is different for the two models because of the number of model levels in the boundary layer (13 and 20 model levels in the lowest 1,000 m, respectively).

Fig. 7.

The most intensive cell around 1830 UT C. (from the left) Cloud condensate in HIRL AM, clouds and hydrometeors in HARMONIE, radar reflectivity, and hydrometeor classification in radar (gray values indicate echoes classified as nonmeteorological). For each panel, width is 60 km and height is 10 km. The vertical axis is linear in radar images, but it is different for the two models because of the number of model levels in the boundary layer (13 and 20 model levels in the lowest 1,000 m, respectively).

Interestingly, a Finnish storm chaser, Pavel Shatylovich, took a photograph of one of the airplanes being hit (Fig. 1). The time of the photograph matches to a flash located at 1736:32 UTC. This single-stroke negative polarity flash has an estimated peak current of 7.4 kA (i.e., relatively low peak current). The airplane appears in the photograph as a bright spot in the middle of the lightning channel.

Table 2 summarizes the replies received from the pilots of the lightning-struck airplanes on 19 October. We have filtered the replies to some extent to save space.

Table 2.

Pilot questionnaire. (TCAS stands for traffic collision avoidance system.)

Pilot questionnaire. (TCAS stands for traffic collision avoidance system.)
Pilot questionnaire. (TCAS stands for traffic collision avoidance system.)

The main findings of Table 2 are the following:

  • Most of the pilots had experienced lightning hit before.

  • During the hit, practically all of the hit planes were approximately at the cloud-base height inside the cloud. One pilot reported the plane to be about 2 km away from the cloud.

  • Typically, a bright light was seen and a loud bang was heard. One pilot reported the whole windshield to be illuminated by St. Elmo's fire.

  • Prior to the hit, interference was observed in the radio (apparently due to the electricity of the cloud).

  • The forecast was not specific enough about the risk of thunderstorms of this magnitude.

  • More rapid response and warnings from the traffic control and airlines are encouraged.

  • Real-time lightning location data available at the cockpit would be useful.

DISCUSSION.

We return to the questions raised earlier. First, what was the primary cause for the large number of the lightning-strike incidences in 2011? The answer is a combination of four ingredients:

  • 1) The exceptionally warm early winter. This makes the occurrence of thunderstorms possible, but their forecasting (especially the magnitude) highly difficult.

  • 2) The direction of the wind at the Helsinki–Vantaa airport during a convective weather type. The direction of wind dictates the runway to be used. Especially in the 19 October case, many of the planes took off right into the core of the storm. In the worst case, three consecutive departing planes got hit by lightning.

  • 3) The lack of information. The convective situation of this magnitude in the cold season at high latitudes surprised both air traffic control and the pilots.

  • 4) The lack of air traffic control procedures. It seems there are no exact procedures on how to react in similar situations. The options are to delay the flights until the convective situation is over or to stay on the schedule and take the risk.

Second, how could have the incidences been prevented? One of the pilots stated ironically that this would have been easy by staying at home. Actually, the statement is quite accurate: it may well be that similar situations will occur anyway, because a rare cold-season thunderstorm is difficult to predict, and the delaying of flights is not economically profitable to the airlines. However, regarding small and isolated cold-season thunderstorms, which are the majority, for example, in the Amsterdam Schiphol Airport (Gough et al. 2009; Hemink et al. 2010), certain avoidance procedures can be successfully used. The problem with smaller cells is that usually the first flash is actually the one triggered by an airplane; this means that the thunderstorm area cannot be monitored prior to the hit with, for example, a lightning location system.

According to the analyzed data, we believe that the extent of similar situations can be highly reduced by training, better nowcasting tools, and with the aid of common procedures at the airports. Regarding the nowcasting tools based on weather radar, the height of +20-dBZ isoline (see the section “Weather radar”) was the best at indicating the active thunderstorms and separating thundery and nonthundery cases. Hydrometeor classification showed graupel associated with the thunderstorms, typically forming narrow and tall vertical pillars. As these pillars did not always extend to the lowest weather radar measurement, and as their diameter is typically only a few radar pixels, finding them in the standard images is a challenge. The lightning– ignition process is highly increased when an airplane flies into the convective core. When the first flash of the storm has occurred, a real-time lightning location system may give important information for pilots, forecasters, and air traffic control.

The operational models showed remarkable differences. The high-resolution HARMONIE model was superior at predicting the convective nature of the event compared to the coarser-resolution HIRLAM. HIRLAM was able to represent only the low-level cloud cover because of to its lower resolution and parameterized deep convection. On the contrary, high-resolution HARMONIE can simulate more realistic deep convective structures with high and narrow cloud/precipitation band. In this case, the mesoscale model with explicit treatment of deep convection and five-species prognostic microphysics parameterization clearly outperforms HIRLAM.

Stability indices such as the K index (KI), LI, and TOTL can provide useful information, from either upper-air sounding or analysis models, such as the LAPS system.

Finally, we note that it is possible that these kinds of cold-season weather phenomena will be more probable in the future if similar weather types occur more often because of climate change.

CONCLUSIONS.

Cold-season thunderstorms are rare events, which increase their threat to aviation safety; both pilots and forecasters can be surprised when they occur. Furthermore, especially at high latitudes, these thunderstorms occur often in the dark, which adds to their physiological effect on the pilots.

Some of the forecasting tools used in the warm season can also be used in the cold season. New and improved tools such as nonhydrostatic models and dual polarization radars have additional value in these cases. Three-dimensional radar data are useful in thunderstorm analysis even in the cold season. Besides forecasters, pilots also need training. The Lightning Location System is an excellent tool for pinpointing in real time where, when, and to what rate lightning is occurring.

Based on the analyzed cases, the most useful indicators of cold-season lightning in southern Finland are the following: the vertical temperature difference between the surface and midtroposphere (700/500 hPa); low-tropospheric wind shear; southwesterly flow (impact of warm sea water); and some convective indices, such as the “total totals” index and the lifted index. CAPE was not a useful predictor here.

Forecasters can have more realistic information on the characteristics of cold-season convective episodes from high-resolution mesoscale NWP than from coarser-resolution models (e.g., global models).

REFERENCES

REFERENCES
AAIB
,
1999
:
Schleicher ASK 21 two seat glider
.
AAIB Bull. 12/99, Reference EW/C99/04/02
,
15
Albers
,
S.
,
J.
McGinley
,
D.
Birkenheuer
, and
J.
Smart
,
1996
:
The Local Analysis and Prediction System (LAPS): Analyses of clouds, precipitation, and temperature
.
Wea. Forecasting
,
11
,
273
287
.
Bengtsson
,
L.
, and
S.
Niemelä
,
2008
:
Nested AROME simulations of convective precipitation
.
HIRLAM Newsletter
,
No. 54, HIRLAM, De Bilt
,
Netherlands
,
37
47
.
Cherington
,
M.
, and
K.
Mathys
,
1995
:
Deaths and injuries as a result of lightning strikes to aircraft
.
Aviat. Space Environ. Med
.,
66
,
687
689
.
Civil Aeronautics Board
,
1965
:
Pan American World Airways, Inc., Boeing 707-121, N709PA, near Elkton, Maryland, December 8, 1963. Aircraft Accident Report File 1-0015
,
22
pp
.
Clifford
,
D. W.
, and
H. W.
Kasemir
,
1982
:
Triggered lightning
.
IEEE Trans. Electromagn. Compat.
,
24
,
112
122
.
Gough
,
W. R.
,
J.
Hemink
,
S.
Niemeijer
, and
T. H.
Fahey
,
2009
:
The prediction and occurrence of aircraft lightning encounters at Amsterdam–Schiphol Airport
.
Proc. 47th AIAA Aerospace Sciences Meeting Including the New Horizons Forum and Aerospace Exposition
,
Orlando, Florida
,
AIAA, ASE-5
.
Hemink
,
H. J.
,
S.
Niemeijer
,
W. R.
Gough
, and
T. H.
Fahey
,
2010
:
Prediction and occurrence of aircraft lightning encounters around Amsterdam Airport Schiphol
.
KNMI Research Triennial Scientific Rep. 2007–2009, 96–100. [Available online at www.knmi.nl/research/biennial/07-09_H16.pdf.]
Kain
,
J. S.
,
2004
:
The Kain–Fritsch convective parameterization: An update
.
J. Appl. Meteor.
,
43
,
170
181
.
Kain
,
J. S.
, and
J. M.
Fritsch
,
1990
:
A one-dimensional entraining/detraining plume model and its application in convective parameterization
.
J. Atmos. Sci.
,
47
,
2784
2802
.
Mäkelä
,
A.
,
T. J.
Tuomi
, and
J.
Haapalainen
,
2010
:
A decade of high-latitude lightning location: Effects of the evolving location network in Finland
.
J. Geophys. Res.
,
115
,
D21124
,
doi:10.1029/2009JD012183
.
Market
,
P. S.
,
C. E.
Halcomb
, and
R. L.
Ebert
,
2002
:
A climatology of thundersnow events over the contiguous United States
.
Wea. Forecasting
,
17
,
1290
1295
.
Mazur
,
V.
,
1989
:
A physical model of lightning initiation on aircraft in thunderstorms
.
J. Geophys. Res.
,
94
(
D3
),
3326
3340
.
Moreau
,
J.-P.
,
J.-C.
Alliot
, and
V.
Mazur
,
1992
:
Aircraft lightning initiation and interception from in situ electric measurements and fast video observations
.
J. Geophys. Res.
,
97
(
D14
),
15 903
15 912
.
National Transportation Safety Board
,
1976
:
Special investigation report: Wing failure of Boeing 747-131, near Madrid, Spain, May 9, 1976
.
Bureau of Technology
,
38
pp
.
Niemelä
,
S.
,
2009
:
Verification of high-resolution precipitation forecasts by using the SAL method
.
HIRLAM Newsletter
,
No. 55B, HIRLAM, De Bilt
,
Netherlands
,
41
47
.
Niemelä
,
S.
,
U.
Andrae
,
B.
Sass
, and
B.
Anderssen
,
2007
:
Convective precipitation in AROME
.
HIRLAM Newsletter
,
No. 52, HIRLAM, De Bilt
,
Netherlands
,
33
38
.
Plumer
,
J. A.
, and
J. D.
Robb
,
1982
:
The direct effects of lightning on aircraft
.
IEEE Trans. Electromagn. Compat.
,
24
,
158
172
.
Rakov
,
V. A.
, and
M. A.
Uman
,
2005
:
Lightning: Physics and Effects
.
Cambridge University Press
,
687
pp
.
Rinne
,
J.
,
2009
:
Talviukkoset Suomessa 2003–2007 (Winter thunderstorm in Finland 2003–2007)
.
M.S. thesis, Dept. of Physics, University of Helsinki
,
72
pp
.
Saltikoff
,
E.
, and
L.
Nevvonen
,
2011
:
First experiences of the operational use of a dual-polarisation weather radar in Finland
.
Meteor. Z.
,
20
,
323
333
.
Schultz
,
D. M.
,
1999
:
Lake-effect snowstorms in northern Utah and western New York with and without lightning
.
Wea. Forecasting
,
14
,
1023
1031
.
Schulz
,
W.
,
K.
Cummins
,
G.
Diendorfer
, and
M.
Dorninger
,
2005
:
Cloud-to-ground lightning in Austria: A 10-year study using data from a lightning location system
.
J. Geophys. Res.
,
110
,
D09101
,
doi:10.1029/2004JD005332
.
Seity
,
Y.
,
P.
Brousseau
,
S.
Malardel
,
G.
Hello
,
P.
Bénard
,
F.
Bouttier
,
C.
Lac
, and
V.
Masson
,
2011
:
The AROME-France convective-scale operational model
.
Mon. Wea. Rev.
,
139, 3
,
976
991
.
Siebesma
,
A. P.
,
P. M. M.
Soares
, and
J.
Teixeira
,
2007
:
A combined eddy-diffusivity mass-flux approach for the convective boundary layer
.
J. Atmos. Sci.
,
64
,
1230
1248
.
Toth
,
Z.
,
S. C.
Albers
, and
Y.
Xie
,
2011
:
Analysis of finescale weather phenomena
.
Bull. Amer. Meteor. Soc.
,
93
,
ES35
ES38
.
Uman
,
M. A.
, and
V. A.
Rakov
,
2003
:
The interaction of lightning with airborne vehicles
.
Prog. Aerosp. Sci.
,
39
,
61
81
.
Undén
,
P.
,
and Coauthors
,
2002
:
HIRLAM-5 scientific documentation
.
HIRLAM-5 project, SMHI
,
144

Footnotes

A supplement to this article is available online (10.1175/BAM S-D-12-00039.2)

Supplemental Material