*** Originally written in german by Armin K. (Team Lower Rhine), translation by Patrick B. (Team Sauerland)
Few things in 2014, in terms of weather, were as remarkable as the severe thunderstorms that took place around this year’s Pentecost and the 9th of June in particular. Wind speeds up to 146 km/h (≈ 91 mp/h = Hurricane force, Sources: Unwetterzentrale Germany, Meteogroup) were measured during a significant Bow-Echo event, leading to (insurance) damages well in the excess of 600 million Euros (according to german news magazine „Der Spiegel“ ) aswell as 6 fatalities and more than 60 injuries.
The most affected areas were around Duesseldorf (capital city of Northrhine-Westphalia), Rhein-Kreis Neuss and big parts of the urban area Ruhr district.
The following analysis takes a closer look on this day’s incidents, based on a meteorological and physical point of view, to answer the question, why before mentioned damages could occur.
To do that, selfmade images of the NinJo-System (Deutscher Wetterdienst – DWD) will be used; these were provided by the Meteorological Institute of the University of Bonn.
Mesoscale Convective Systems (MCSs) are spacious convective clusters, that are characterized in their expansion and by being long-lived.
In contrast to mid-latitude countries, where they are most common in the summer, MCSs can occur throughout the whole year in the tropics.
To form an MCS, quite a big amount of (potential) instability is needed, which is described by the KO-Index and the „Convective Available Potential Energy“ (CAPE). Furthermore, an inertial lifting trigger is needed to set off convection. This could be a front / convergence or even a mountain range. Divergence of a jetstreak / QG-Forcing (DPVA, WA-Maximum) in the 500 / 300 hPa – layers can trigger convection aswell.
Another criterion, especially regarding the durability of the system, is vertical wind shear. Higher amounts of vertical wind shear can lead to a better organization of convective systems, although on its own, it is not triggering convection. It has to be overlapping larger amounts of CAPE.
As wind shear can be depicted as vectors, it is divided into „speed shear“ and „directional shear“.
To measure a combination of both forms of shear, the storm-relative helicity (SRH) was introduced. It is positive, therefore supporting convection, if the windspeed increases in greater heights and at the same time its direction turns to the right.
According to Maddox (1980), two forms of MCSs are occuring on the meso-alpha scale: circular and linear MCSs.
Latter are also known as squall-lines, that are most often seen just before or at cold fronts or other air-mass boundaries. A high amount of wind shear (baroclinicity) is present, that results in a linear organisation of convection; the vertical increase of windspeed in lower layers of the troposphere leads to a certain durability of the system (Rotunno, Weismann, Klemp, 1988).
A more detailed classification of this kind of MCS was done by Parker and Johnson. According to them, a linear MCS consists out of a part with convective precipitation and one with stratiform precipitation. Depending on how these regions are orientated, they can be classified as leading stratiform, parallel stratiform or trailing stratiform:
1. trailing stratiform: The cold-pool / leading-edge of the system shifts in a progressive way and, relatively seen, faster than the base current. The area of stratiform precipitation spreads in a retrograde way.
2. leading stratiform: the stratiform area of precipitation spreads faster than the leading-edge in shifting direction, thus can be found in front of the strongest precipitation.
3. parallel stratiform: the stratiform area of precipitation spreads parallel to the leading-edge.
Other sub-types of linear MCSs are defined by the spatial form of their convective area of precipitation (e.g. if a squall-line consists of multiple cells, that can be optically devided by their maximum reflectivities on radar, it is called „broken line“).
If strong vertical wind shear is present, some parts of the line may be faster than others, which can lead to a bow-shaped pattern noticeable on radar. These segments are then called „bow-echoes“. Multiple lined-up bow-echoes then form a „line echo wave pattern“ or LEWP. A very progressive and far spreading linear convective system is known as „Derecho“. These systems are very rare in central Europe, but if they occur, they lead to enormous damages on a greater scale, mostly due to downbursts.
Circular MCSs are most common in the tropics, where huge amounts of CAPE are overlapped by weak wind shear. The top of the system spreads in a radial way along the tropopause-border; due to the weak flow aloft, it remains its circular shape. Maddox developed a special case for circular MCSs, which is known as the mesoscale convective complex or MCC. Following features must be apparent on IR-sattelite images:
1. Cloud-top temperatures have to be less than -52°C in an area of 50000 km² and less than -32°C in an area of 100000 km².
2. The before mentioned aspect has to be apparent for at least 6h.
3. Eccentricity during maximum horizontal spread has to be 0,7 or more.
Severe weather phenomena such as heavy precipitation, (large) hail and storm to gale force winds are most common during MCSs/MCCs. Latter are primarily occuring in linear MCSs, which can result in damages to large areas. A well known example would be the derecho, that hit Berlin in 2002, where windgusts up to 150 km/h (93 mp/h) were measured.
Circular MCSs – and MCCs in the first place – are very rare in Europe, as there are strong wind currents present in higher layers of the atmosphere nearly all the time (criterion 3 is not fulfilled). Relatively seen, they are most likely in the end of summer.
Figure 1: Weather Situation, Source: NinJo, DWD
An upper low in the west of the UK and a ridge that spreaded from northern Africa to the Baltic Sea induced a strong southerly to south-westerly flow in higher layers of the atmosphere over central Europe on the 9th of June 2014 (see Figure 1). In contrast to that, the surface-based flow was quite diffuse, only the upper low over the Atlantic was present in lower levels aswell.
Figure 2: Weather Situation, Source: NinJo, DWD
Fig. 2 shows the course of the polar jet, which reflects the position of before mentioned upper low. Maximum windspeeds reach up to 120 kts in the north-west of the Iberian Peninsula. After having a meridional orientation over the UK in front of the trough, the jetstream does an anticyclonal turn to the east over the North Sea and southern Scandinavia. All in all, this was a typical weather situation in front of a trough, which often is the base for severe thunderstorm outbreaks over central Europe.
An important factor for mesoscalic weather development on the 9th of June 2014, was a waving frontal zone over the western parts of France, which, together with the before mentioned jetstreak, marked a baroclinic Area. Fig. 3 (front analysis created with NinJo) shows the airmass boundary, aswell as a prefrontal convergence, that, like it’s often the case, formed in the warm sector.
In the near of that convergence zone, three convective clusters were able to form over Germany, France and Poland .
The „L“ in France marks a present, thermal induced low, that later moved north-eastwards.
Figure 3: Front Analysis 06 UTC, Source: NinJo, DWD
In the following, the existing conditions will be analyzed regarding the support of convection.
A parameter to characterize vertical layer instability is CAPE. In the following analysis, only Mean-Layer-Cape (ML-CAPE) by COSMO-EU was used. The advantage of that form of CAPE is, that temperature and dewpoints of the lowest 50 hPa are averaged for numerical definition of the parcel trajectory . This often leads to more realistic results compared to the classic definition of CAPE with surface baced values. A reason for this is, that CAPE reacts very sensitive to small variations of surface-based moisture, which often leads to exaggerated and non-representative CAPE-values. Furthermore, it has to be noticed that, when interpreting CAPE, the assumptions of the derivation have to be taken into account. In reality, air packages are affected by entrainment (first assumption to derive CAPE) and, atleast on a small scale, no hydrostatic conditions are present (second assumption to derive CAPE), which always leads to smaller actual CAPE-values than calculated.
Figure 4: ML-CAPE and Wind Shear, Source: NinJo, DWD
Fig. 4 furthermore shows CAPE being overlapped bei DLS (Deep Layer Shear) and LLS (Low Level Shear), which are, as already mentioned, an essential basis for organization of moisture convection. You can see, that especially in the evening (around 18 UTC), very high instability was present (i.e. for usual Central European conditions). According to the COSMO-EU analysis, ML-CAPE values reached 2000-2500 J/kg in large areas, peaking over 3000 J/kg. An interesting addition is, that even in the morning (06 UTC) 2000 CAPE and above were reached in parts of France and Belgium. Large parts of the area witnessed negative KO-Index values aswell (not shown), which was also a sign for the potential instability of the atmosphere over Central Europe.
The airmass was overlapped by DLS values of about 15 m/s over central Germany and nearly reached 30 m/s over the Normandie in the vicinity of the jet. If you now add the windprofiles in 850 and 500 hPa (12 UTC), you can see a right turning wind profile, also known as „veering“, which results in positive helicity.
Figure 5: Sounding Bergen, Source: NinJo, DWD
The vertical layers will now be analyzed in detail, using a Skew-T-Log-P diagram (Fig. 5).
18 UTC radio soundings of Bergen were used, as the Essen soundings were not available the whole day. The Temp shows a moist boundary layer, in which the spread of temperature and dewpoint is minimal. Above, in the middle of the troposphere, a dry inversion with dry adiabatic lapse rates was present. Temperature and dewpoint values are then again converging in higher altitudes, which leads to a so-called inverted-V profile. The dry layer linked to that is called elevated mix-layere or EML.
EML in Central Europe often forms in south-westerly flow (like in this case). Air is heated over the spanish Highlands which forms a very turbulent and well mixed boundary layer with dry to over-adiabatic lapse rates. This airmass is advected with south-westerly flow towards Central Europe. As it is always situated in front of the trough, it is lifted at the warm conveyor belt. These processes are nearly adiabatic, which leads to that the layer remaining its structure but getting elevated.
If at the same time greater amounts of surface-based moisture is present, a moist boundary layer with dry air above forms, which leads to high amounts of potential instability. The above mentioned inversion kind of caps the surface-based moisture, which results in, due to evapotransporation and moistureadvection, steadily increasing moisture and instability. This is happening until an external trigger breaks the inversion and the CAPE is set free. This weather situation is also known as „loaded-gun“ and is one of the potentially most dangerous weather situations in Europe.
The actual lability-energy for Bergen reached, depending on which package you looked at, 2000 – 3000 J/kg.
The bright red area shows the so called surface-based or SB-Cape and describes the energy, that is available to a rising air package until it reaches the CCL. It reaches higher values than CAPE-LFC (dark red area) but the crucial criterion to release the potential energy is reaching the triggering temperature, which, in this case, was too high (34° C). ML-CAPE couldn’t be sketched in, but reached values of about 2000 J/kg. Furthermore, you can see the before mentioned veritable wind shear: Especially the low levels showed a strong right turning aswell as an increase of windspeed, which were the results of the Ekman-Spiral and surface-based warm air advection. The resulting values of helicity reached 200 J/kg in this sounding, which in combination with the high amounts of instability led to the suspicion of organized convection. The surroundings were more than supporting to form spacious convective clusters.
As the currents in high altitudes ran almost parallel/straight to each other with no markable short-wave troughs, you couldn’t expect stronger lifting impulses. The needed convection-triggering drive had to be delivered in lower levels, e.g. due to convergences or orographic lifting.
As already shown in the 6 UTC frontal analysis, an MCS was active over France in the wake of a shallow thermal low. After weaking of the MCS, a left over outflow boundary of it induced forming of single cells over western Germany, which then formed a new MCS, that won’t be further discussed.
Figure 6: Radar- and Satelliteimage, 14 UTC, Source: NinJo, DWD
Figure 7: Radar- and Satelliteimage 15 UTC, Source: NinJo, DWD
Concentration now layed on the developments over France, where new convection formed on a convergence.
A remarkable development was a supercell in the near of Paris, which can be found in Fig 7 (orange circle). Reflectivities were peaking the scale, french stormchasers measured up to 65 dBZ and hailstones with sizes up to 8 cm (over 3 inch) in diameter were found.
This cell and the stratiform area of precipitation will now be analyzed in detail, as it formed an unusually big and damaging MCS.
East of the french convection, an area with low cloudiness is present. It therefore was radiated permanently. The reason for that was, amongst other things, caused by the dissipating MCS, whose sinking air led to dissolving cloudiness. Furthermore, early convection over western Germany led to a surface-based increase of moisture. Due to radiation, lapse-rates in in the mid-tropospheric altitudes got steeper , which led to fast building potential instability.
If you now analyze the preceding airmass in the satellite and radar images of 14 UTC (Fig. 6), you can see a small area of cloudiness right in front of the cells (marked in red). These clouds are high-based Cumuli and Altocumuli castellani, which are an indicator for instability in the mid-tropospheric altitudes. They could form above the dry inversion and thus were elevetad. As they were inside the EML, little moisture was available, so deep convection was inhibited at first.
A crucial factor for further development was, that the east-moving cells in France were getting outflow-dominant and the resulting coldpool spreaded eastwards rapidly. If you now take a look on the Sat-Loop, you can notice the Ac cas getting overtaken by the convection’s cloudtops.
Between 15 and 16 UTC an explosive increase of convective activity of the (at first) aging system took place. Following factors were responsible for that:
1. Interaction of coldpools with orography (Ardennes) and environmental shear, which led to strong vertical movement at the leading edge.
2. Elevated convection getting surface based due to 1., leading to sudden entrainment of surface-based moisture.
3. Convection moving into radiated regions.
Figure 8: Radar- and Satelliteimage, 17 UTC, Source: NinJo, DWD
Figure 9: HRV-Satelliteimage, 17:30 UTC, Source: NinJo, DWD
The rapid strengthening and the increasing size of the increasingly organizing system can be seen on both radar and satellite images (see Fig. 8). Within 30 minutes, the reflectivity of the now called MCS increased to more than 50 dBZ in a broad strip. Lightning activity increased aswell, 1000 strikes were recorded during 15 minutes.
The NinJo-based HRV image (Fig. 8) also shows the forming of an over-shooting top in the area of the leading edge, where the maximum vertical movement would occur.
The textbook plate-shaped cloud-tops spreaded radial, which could be compared to the eruption of a volcano or an explosion. Within an hour, the area of the cloudtops nearly doubled. Measurings stated that the eccentricity was 250 km / 300 km = 0, 83, which is the value that would fulfill one criterion of gaining MCC status.
If you take a closer look on the cloud-top temperatures, that are colored in IR-satellite imagery for values below -32°C, you can see, that the area of 75000 km² is not quite that of a MCC.
The MCS now crossed the German border. The most active region was located between Aachen and Heinsberg.
Even before 18 UTC, you could see that the system began to organize in a linear way (see Fig. 8) ; it was clear very soon, that downbursts would be the system’s main hazard.
Figure 10: Radaranalysis and Theta-e-Measurements 18 UTC, Source: NinJo, DWD
Added to the NinJo-based radar-analysis of 18 UTC (Fig. 10) are so called pseudofronts. These can be compared to „normal“ fronts, which is shown by additionally shown surface-based measurements for equivalent-potential temperatures. These dropped about 10 K after the MCS went through (60° C -> ≈ 50° C ).
Therefore, the MCS can be classified as a kind of mesoscalic low, whose southern part is faster than the northern one. As stated in the introduction, this kind of radar echo is called bow-echo. This bow-echo started accelerating, which was caused by a phenomenon described by Weismann, the rear-inflow-jet (RIJ). This mesoscalic process often occurs when CAPE values of 2000 J/kg or more are present and an enormous cold pool was formed. The system, and in particular its updraft, tilts backwards, because the cold-pool circulation over-compensates the shear-circulation. Essential are strong horizontal driving gradients in the rear sections of the MCS, that create vortexes around a horizontal axis (ξ -vorticity). This creates cyclonal rotation in mid-tropospheric altitudes and anticyclonic vorticity in lower levels of the rear section of the cluster. Between those vortexes, pressure compensation induces a „belt“ of strong winds, that is known as the rear-inflow-jet.
Figure 11: Vertical Scan 19 UTC, with additional information, Source: NinJo, DWD
These processes are illustrated in the 19 UTC vertical scan (Fig.11, cut in west-east direction through the system), at that time, the MCS happened to be in its major-phase. You can see the area of highest reflectivities being pushed „forward“ (to the east) in the lowest altitudes by the rear inflow jet (RIJ, dark red) and the coldpool and at the same time being tilted towards the west in greater heights. You can also notice, that the RIJ was feeding the system with dry air of the mid-troposphere (lower reflectivities push forward along the RIJ). This led to (precipitation) evaporation which consumed a lot of latent heat and increased the cold pool’s strength even more.
Figure 12: Radaranalysis and Theta-e-Measurements 20 UTC, Source: NinJo, DWD
Even the Azimut-radar images at around 20 UTC (Fig.12) show the before mentioned structures, the course of the vertical descending RIJ is roughly marked by the black arrow. Unusually strong wind gusts (near surface) were present with the arrival of the RIJ (the airport of Duesseldorf measured windspeeds up to 144 km/h (89 mp/h) for example ). According to the following damage analysis, some parts of the Ruhr Area witnessed windspeeds up to 150 km/h (93 mp/h), although Bemoulli- and Venturi-effects between houses could have played a greater role.
Weather phenomena were not as severe at the southern end of the bow-echo, due to anticyclonal rotation at the line end vortex, although severe wind gusts and heavy rainfall were still occuring in that area.
Figure 13: Radaranalysis and Theta-e-Measurements 22 UTC, Source: NinJo, DWD
The outflow boundary of the MCS propageted southwards very fast, due to the southward pushing coldpool in the rear. The correlating pseudofront could be verified by surface-based measurements, around 21 UTC it was situated roughly between Adenau and Warburg. In the following hours, isolated convection with low (lightning) activity occured on that front (see radar image of 22 UTC, Fig. 13).
The mainpart of the MCS visible weakening tendencies after 22 UTC, although it was still obtaining unusually good organization. A reason for the weakening was the increasing retrograd tilting of the updraft, which led to a decreased vertical movement at the leading edge. A spacious area of stratiform precipitation developed in the rear of the system aswell.
Taking Maddox’s criteria for MCSs into account, one can see, that cloud-tops were expanding up to over 100000 km², but at the same time, eccentricity values dropped below 0,7.
The MCS moved towards the north-east during the night and it began to disorganize noticeably. A second pseudofront developed over Lower Saxony, where weak convektion was induced again (not shown). Taking a closer look on radar imagery between 0 and 2 UTC, one can see a weak cyclonal rotation of the stratiform area of precipitation. This was occuring due to increasing influences of the Coriolis force. Around 4 UTC, the heavily weakened system reached the polish border and later dissipated in a less dynamic and more stable environment.
In summary, this MCS was one of the most organized and damaging systems over Central Europe, especially in Germany, for years. It developed out of an old Cluster in France in an unusually unstable environment, where CAPE values reached up to 3000 J/kg. Notice again, that the necessary lifting trigger was not caused out of the high layers, e.g due to a jet exit. The mentioned jet was situated over western Europe, thus never had a direct influence on the event. QG-forcing can’t be named a triggering factor aswell, as the currents aloft were orientated in a straight line and no edging short-wave troughs with differential vorticity-advection were embedded. Of course, every current possesses vorticity that can be advected, but these were not the dominant drives in this case. The initialising factor was outflow-dominant convection over France, that established along the prefrontal convergence zone.
The outflow-boundary then interacted with the Ardenne’s orography, which led to vertical movement. The driving factors were of mesoscale nature.
After explosive strengthening during half an hour, high wind shear values led to a rapid straight-line clustering and the formation of a bow-echo. Within that rapidly accelerating segment, a huge cold pool caused a strong rear inflow jet, which, originating in mid-tropospheric altitudes, could push down towards surface-based layers and thus caused unusually strong hurricane force wind gusts.
The MCS’s / bow-echo’s climax occured over the western parts of Germany, between the southern Lower Rhine and the Ruhr Area, where the biggest damages since extratropical cyclone Kyrill (January 2007) were recorded, mostly due to fallen trees and destroyed roofs. It was an unlucky coincidence that this system moved over one of the most populated areas of Germany. A different track could have easily led to less damages to property and people. Reaching eastern Westphalia, the system began to weaken slowly, as the updrafts began to tilt backwards.
Regarding the analysis of the general weather situation, it turned out that NinJo was a very helpful tool. Especially the radar and satellite imagery together with its appealing additions were a big advantage.
The MCS’s structure, outflow-boundaries, aswell as environment variables (T, θe, v) were analyzed by surface-based measurements. The vertical cut radar imagery with its additions led to the verification of the RIJ, which was most likely the cause of the enormous wind speeds.
MCC criteria of Maddox could be verified by IR-satellite imagery aswell. Colorscheme was adjusted following the required maximum temperature (-32 ° C).
Essence of this research is, that this system was NOT an MCC. Although the MCS reached some criteria to form an MCC, the eccentricity was the essential part that was not reached (as it is often the case in mid-latitude countries). Eccentricity values during maximum expansion layed between 0,5 and 0,6.
According to Parker and Johnson’s classification of MCSs, this system was a trailing-stratiform system with an embedded bow-echo.
Rotunno, Richard, Joseph B. Klemp, and Morris L. Weisman. „A theory for strong, long-lived squall lines.“ Journal of the Atmospheric Sciences 45.3 (1988): 463-485.
Weisman, Morris L. „The role of convectively generated rear-inflow jets in the evolution of long-lived mesoconvective systems.“ Journal of the atmospheric sciences 49.19 (1992): 1826-1847.
Parker, Matthew D., and Richard H. Johnson. „Organizational modes of midlatitude mesoscale convective systems.“ Monthly weather review 128.10 (2000): 3413-3436.
Maddox, Robert A. „Mesoscale convective complexes.“ Bulletin of the American Meteorological Society 61.11 (1980): 1374-1387.