Impacts of Storm “Zyprian” on Middle and Upper Atmosphere Observed from Central European Stations

Abstract

Mesoscale convective systems are effective sources of atmospheric disturbances that can reach ionospheric heights and significantly alter atmospheric and ionospheric conditions. Convective systems can affect the Earth’s atmosphere on a continental scale and up to F-layer heights. Extratropical cyclone “Zyprian” occurred at the beginning of July, 2021 and dominated weather over the whole of Europe. An extensive cold front associated with “Zyprian” moved from the western part to the eastern part of Europe, followed by ground-level convergence and the formation of organized convective thunderstorm systems. Torrential rains in the Czech Republic have caused a great deal of damage and casualties. Storm-related signatures were developed in ground microbarograph measurements of infrasound and gravity waves. Within the stratosphere, a shift of the polar jet stream and increase in specific humidity related to the storm system were observed. At the ionospheric heights, irregular stratification and radio wave reflection plane undulation were observed. An increase in wave-like activity was detected based on ionograms and narrowband very-low-frequency (VLF) data. On directograms and SKYmaps (both products of digisonde measurements), strong and rapid changes in the horizontal plasma motion were recorded. However, no prevailing plasma motion direction was identified within the F-layer. Increased variability within the ionosphere is attributed mainly to the “Zyprian” cyclone as it developed during low geomagnetic activity and stable solar forcing.

1. Introduction

The energy budget of the ionosphere fluctuates due to solar cycles, seasons, and daily changes. While external solar forces primarily drive this variability ([1,2,3,4] and others), wave coupling from below also contributes significantly, especially on consecutive days with stable solar forcing (see for instance [5,6,7,8]). Data from various solar cycles indicate that the Earth’s atmosphere’s sensitivity to solar forces evolves over time due to its complexity and long memory, complicating the identification of causal mechanisms. Different atmospheric regions react distinctly to various solar activity parameters. Studies have explored these connections, such as solar activity’s influence on the stratospheric ozone [9,10], stratospheric temperature [11], the Quasi-Biennial Oscillation [12,13], stratospheric, and others. The troposphere, the lowest atmospheric layer, is characterized by decreasing temperature with altitude and contains most atmospheric water, influencing weather patterns. The stratosphere plays a crucial role in absorbing and dissipating waves due to very strong zonal winds. There are also some other phenomena in the stratosphere, like sudden stratospheric warmings or QBO, which affect mesospheric and probably ionospheric dynamics as well. That is why it is very important to know the condition of the stratosphere. The ionosphere, shaped by solar radiation, displays variability and stratification into D, E, and F layers, with the highest ionization typically in the F or F2 layer. Ionospheric variations are described by long-term cycles and short-term fluctuations and influenced by solar and geomagnetic activity (see for instance [2,4,14]). European stations show correlated variations, and a 1000 km scale supports a link between the lower atmosphere and the F2 region [15]. Probabilistic models used in [16] identify three dependency groups based on tropospheric climate regimes: East Asian (cold), European (moderate), and North American (arid).

The Earth’s atmosphere is intricately connected through various processes across multiple scales. Vertical coupling involves dynamic, chemical, and electrical interactions, with energy transfer facilitated by atmospheric waves, such as gravity waves (GWs) and tides, which originate in the troposphere and propagate to higher layers, like the ionosphere [17]. Notably, auroral activity is a key source of GW at high latitudes [18] and references therein, while other atmospheric levels contribute at mid and low latitudes. These waves influence circulation and variability in the stratosphere and alter the thermal gradients in the mesosphere [19]. The impact of GW extends to the ionosphere, with effects shaped by wave properties and ionospheric conditions [20,21,22,23]. Sources of GW include orographic features, convection, wind shear, and jet streams [24]. Jets and fronts are significant wave generators, as observed in their increased activity near these features (e.g., [19,25,26,27,28,29,30,31,32]). The dispersive nature of GWs means that different wavelengths travel at varying speeds [33]. Convection-induced GWs cover a wide range of horizontal scales, and their energy can create disturbances observable in the mesosphere and lower thermosphere. A detailed review on GWs near jets and fronts from observations, theory, and modeling is provided by [34] and by [30]. GWs excited from mesoscale deep convection are of a wide range of horizontal scales of a few to tens (e.g., [35,36]) up to several hundreds of km (e.g., [37]). Wave activity is not uniformly distributed, peaking in winter at higher latitudes and around equinoxes at lower latitudes. Accurate studies of ionospheric variability benefit from considering lower atmospheric influences.

The observed morphology of the GW potential energy suggests a controlling factor that moves eastward; the synoptic scale system, which moves eastward with time and has seasonal variation in activity and can also be attributed to extratropical cyclones [38]. The high GW potential energy regions are located around the northern part of Eurasia, northeast part of America, and North Atlantic Ocean where the track density of extratropical cyclones is high [39,40]. High GW potential energy values above continents are explained by topographic effects [41].

Within a real meteorological geopotential data study, [28] identified a center of cyclones, a center of anticyclones, jet streams, a saddle point between a cyclone and an anticyclone, and an outlying area of vortices being effective meteorological sources that are able to influence the upper atmosphere. Observations of wave signatures in the upper atmosphere with a source in the meteorological mesoscale system were reported for instance in connection with a fast passage of frontal cyclone “Fabienne” [42] and several summer convective events “Bernd” [43,44]. Studies provided by [45,46,47,48,49] suggested that moist processes may enhance GW generation compared to dry convection only. The theoretical study by [45] shows that moist processes play an important role in the generation and amplification of GWs. Polar-orbiting satellite observations document that deep convective storms and mesoscale convective systems are frequent and efficient sources of GWs observed in the middle atmosphere.

The Atmospheric Infrared Sounder (AIRS) on NASA’s Aqua satellite and the Infrared Atmospheric Sounding Interferometer (IASI) onboard the European Metop satellites can detect GWs in the upper stratosphere [50,51,52]. The Day/Night Band (DNB) of the Visible Infrared Imaging Radiometer Suite (VIIRS) instrument aboard the NOAA/NASA S-NPP, NOAA-20, and NOAA-21 satellites detects the GWs in airglow layers (ca 85–100 km) on Moonless nights, above darker regions (not affected by light pollution) [53]. Other space-borne instruments, such as Mesospheric Airglow/Aerosol Tomography and Spectroscopy (MATS), provide limb-oriented measurements, aimed at 3D GW structures [54] at the mesopause levels.

Convective storms and related phenomena are a well-known source of infrasound signals. Pulse-shaped infrasound signals recorded during convective storms were analyzed in a number of studies and were associated with lightning [55,56,57,58]. The spectrum of infrasound emitted by a lightning discharge is flat with several peaks between 0.1 to 10 Hz. Other sources in convective storms are microbursts and tornadoes [59]. Sources that generate infrasound in atmospheric convection were analyzed in a model study by [60]; they particularly include a source associated with phase transitions of moisture and, in tornadic storms, a source associated with adiabatic vortex fluctuations and turbulence. Observations of infrasound generated by large convective systems were reported in the ionosphere [61,62,63]. Due to atmospheric filtering, only low-frequency signals of the order of 10⁻³ Hz (dominant periods of 3–4 min) were able to reach ionospheric heights.

Atmospheric waves are manifested as a limited-duration oscillation of atmospheric parameters (temperature, pressure, etc.). The basic governing dispersion equation for acoustic-gravity waves (can be found for instance in [2]).

$${\omega }^4-{\omega }^2{\omega }_a^2-{\mathit{k}}_x^2{c}^2\left({\omega }^2-{\omega }_g^2\right)-c^2{\omega }^2{\mathit{k}}_z^2=0$$

(1)

tight together property of the medium in the parameter c sound speed, ω_a acoustic cut-off frequency, ω_g buoyancy frequency and wave property, ω frequency and k wave vector. Only waves satisfying the dispersion relation can propagate within the medium. Wave propagation is dispersive and anisotropic. Two families of waves can propagate: acoustic waves based on frequencies greater than the acoustic cut-off frequency and gravity waves based on frequencies lower than buoyancy frequency.

Particular parts of acoustic-gravity waves are infrasound waves based on frequency far from the acoustic cut-off frequency. Infrasound occupies the frequency range from approximately 0.0033 to 20 Hz. Infrasound emitted by a source at the ground and propagating in the atmosphere is refracted back at the altitude where the effective sound speed ratio is

$$C_{ratio}>1,$$

(2)

$$C_{ratio}={\displaystyle \frac{c_{eff,z}}{c_{eff,ground}}}$$

(3)

the effective sound speed, $c_{eff},$ is obtained as follows:

$$c_{eff}=\sqrt{{\gamma }_{g}RT}+\mathit{n}·\mathit{u}$$

(4)

${\gamma }_g$ is the specific heat ratio, $R$ is the gas constant, $T$ is the temperature in K, $\mathit{n}$ is the unit vector in the direction of infrasound propagation, $\mathit{u}$ is the wind speed vector, $c_{eff,z}$ is the effective sound speed at altitude z, and $c_{eff,ground}$ is the effective sound speed at the ground.

Thus, wind and temperature fields form waveguides through which infrasound can propagate to distances up to thousands of km from the source [64]. A part of infrasound energy is not trapped by the waveguides, and it escapes to the thermosphere/ionosphere. The launch angle of those rays is up to 20–30° from the vertical [65,66]. While infrasound attenuation is low in the lower and middle atmosphere, infrasound propagation in the thermosphere is limited by strong signal absorption [67]. Signals of frequencies higher than 1 Hz are absorbed below ~110 km; as a result, only low-frequency infrasound can propagate to the ionospheric heights [68].

In the atmosphere, the gravity wave domain covers frequencies up to approximately 0.002 Hz or oscillations based on periods larger than approximately 7 min in the troposphere and about 15 min in the ionosphere. The frequency or period range depends on the atmospheric region. The propagation of GWs is restricted by the property of the medium and wave frequency into the following range:

$${\varphi }_{min}=si{n}^{-1}\left({\displaystyle \frac{\omega }{{\omega }_g}}\right)$$

(5)

$${\varphi }_{max}=\pi -si{n}^{-1}\left({\displaystyle \frac{\omega }{{\omega }_g}}\right)$$

(6)

Thus only a part of the GW generated in the troposphere can propagate upward and eventually reach the height of the ionosphere.

This study aims in documentation of the coincidental observation of the gravity and infrasound waves generated by the mesoscale convective systems associated with the extratropical cyclone “Zyprian”. It puts together different types of observations and different manifestations of infrasound and GWs within different and distant atmospheric regions. Despite a large set of data, there are certain height gaps where the observation is very limited if even possible.

2. Method

2.1. Processing of Observation Data

2.1.1. Meteorological Data

In meteorology, we use weather radars, satellites, and lightning-detection systems to observe and interpret storms on a wide-area basis in order to see the evolution of large systems that affect the state of the troposphere. Point measurements at meteorological stations are used to detect changes in various meteorological elements (air temperature and pressure, wind direction and speed, precipitation), e.g., during the passage of storms.

Meteorological satellites detect radiance reflected and emitted by clouds, surface of the Earth, or cloud-free atmosphere (in various atmospheric absorption bands). Besides direct use in forecasting and nowcasting, satellite data serve as one of the key inputs for numerical weather-prediction models and are broadly used for climate monitoring, atmospheric composition monitoring, and many other applications. In this study, we focus mainly on satellite observations of convective storms [69,70,71,72] and their characteristics as a possible source of GWs. The geostationary satellite MSG (Meteosat Second Generation) is used to document the evolution of convective storms, while polar-orbiting NOAA-19 is used to show detailed storm-top structures shortly before sunset on 8 July 2021.

For lightning detection, we used measurements from the Blitzortung.com network, which uses very-low-frequency receivers to determine the location of lightning based on the exact time of arrival of the signal [73]. Thanks to the wide network of volunteers forming this measurement network, the sufficient quality location of lightning discharges in Central Europe is possible. The lightning activity map is used to track the movement of severe thunderstorms and their lightning activity.

Although weather stations measure weather only at discrete points and are influenced by the surrounding terrain, they allow continuous measurements of important weather elements. Changes in the time series of temperature, humidity and air pressure, wind direction, and wind speed can accurately detect the passage of various meteorological phenomena, such as the passage of a front or a convective storm.

2.1.2. Infrasound Stations and Data Processing

To observe infrasound at the ground, arrays of microbarometers are employed. Microbarometers are highly sensitive sensors able to record high-frequency pressure fluctuations in the infrasound range down to amplitudes of the order of mPa. Pressure recordings contain a mixture of infrasound signals and background noise. To distinguish between the signal and noise, microbarographs are deployed in arrays. It is assumed that infrasound is recorded as a coherent signal at the respective array elements, while background noise is characterized by variations in phase and amplitude [74]. Various methods have been developed for extracting signals from noise; we use the Progressive Multi-Channel Correlation (PMCC) [75,76] in our study. The method is able to extract low-amplitude coherent signals from background noise [74]. The minimum number of array elements necessary for infrasound detections is three. Recommendations on the geometry of infrasound arrays, i.e., inter-element distances and their spatial arrangement, can be found, e.g., in [77,78].

The infrasound array PSZI is located in Northern Hungary at 47.92°N 19.89°E. It is equipped with four SeismoWave MB3d microbarometers. The array aperture of 250 m determines optimum array performance in the frequency band between 0.11 and 2.72 Hz [79]. The pressure recordings are stored with the sampling frequency of 100 Hz and resampled to 20 Hz when processed with PMCC. The Czech array PVCI is located at 50.53°N 14.57°E. Three sensors of the type Infrasound Gage ISGM03 are distributed in an equilateral triangle with inter-element distances of 200 m. The optimum operational range is therefore similar to that of PSZI, and it is from 0.14 to 3.4 Hz [79]. Data are stored with the sampling frequency of 25 Hz. Resampling of the waveform data is not applied when processing infrasound detections.

Infrasound detections at PSZI and PVCI are processed with DTK-GPMCC 6.3.0 software, the core of which is the Progressive Multi-Channel Correlation (PMCC) detection algorithm [75,76]. The PMCC configuration is individually optimized for each of the arrays based on recommendations by [78,79,80]. Infrasound detections are processed in the frequency range of 0.09–7.1 Hz at both stations. The band width goes beyond the optimum operational range, particularly at the high frequency limit. However, a good resolution can be expected for arrival parameters of the infrasound thunders thanks to a high signal-to-noise ratio of those pulse-shaped signals.

The West Bohemia Czech Infrasound (WBCI) large aperture array is used for the monitoring and analyses of GW activity in the troposphere. It consists of four sensitive absolute microbarometers (6000-16B-IS, Paroscientific, Inc., Redmond, WA, USA) located in Studenec (STC), Luby (LBC), Nový Kostel (NKC), and Vackov (VAC) at mutual distances of 4–10 km in the western part of the Czech Republic (see [81] for the exact coordinates) and is suitable for the analysis of GWs and long-period infrasound in the range 0.0033–0.4 Hz [82].

2.1.3. Stratospheric Data

The stratosphere is characterized mainly by the zonal mean flow, where meridional wind plays a marginal role. Moreover, it is very dry in comparison with the troposphere, so we can identify even small changes in the specific humidity. Unfortunately, direct measurements in the stratosphere are very rare and we have to use satellite observations for temperature or humidity or mostly reanalyze datasets, which provide all parameters in a regular grid network with regular time steps. That is why we can analyze stratospheric behavior during the storm in one-hour resolutions in specific grid points (in our case, we selected Prague, 50°N 15°E).

2.1.4. Ionospheric Data

The ionosphere is remotely monitored by ground-based instruments based on the radio wave transmission and observation of the vertically or obliquely reflected waves. The principle of the method lies in the property of plasma that reflects all the electromagnetic waves with frequencies below the plasma frequency (related to the local electron concentration). Measuring the time of flight of the reflected wave allows for the determination of the vertical electron density profile. The sounding frequencies used for the routine measurement are 1–20 MHz. A detailed description of the vertical ionospheric sounding can be found, for instance, in [2]. The same principle, the wave reflection, is used in continuous Doppler sounding measurements at fixed frequencies with a high sampling rate. The applied methods of GW analysis using continuous Doppler sounding are described in [83,84]. In the study, we use data from European observatories equipped by Digital Portable Sounders DPS 4D (a detailed description of the instrument can be found in [85] or https://ulcar.uml.edu/digisonde_dps.html, accessed on 2 September 2024), which form part of the Global Ionosphere Radio Observatory (GIRO). Furthermore, we analyze and interpret ionograms (height-frequency characteristics of the ionosphere) and manually scale the recorded traces of the reflection from each layer; using the inversion technique NHPC [86], we obtain full electron density profiles and derive the main characteristics from them. In detail, we analyze the behavior of the critical frequency foE and foF2 and their corresponding heights hmE and hmF2 derived from inverted profiles. The DPS 4D system measures plasma motion below the F2 layer maximum as a separate measurement after ionogram recording. Data are scaled and interpreted using a routine described by [87,88].

2.1.5. Narrowband VLF Data

The lower ionosphere (~60–100 km) represents the transition region between the neutral atmosphere and the ionosphere and a prominent buffer zone for energy transfer from the Sun to the Earth [1,89,90]. The in situ observation of this region is restricted by the fact that stable satellite orbits are not feasible at these altitudes, while conventional meteorological balloon measurements only reach altitudes of a few 10 km. The so-called narrowband VLF technique uses the powerful very-low-frequency (VLF, 3–30 kHz) signal of known frequency emitted by military transmitters for remote sensing of this region [91]. The transmitters emit a phase-modulated radio signal that is strongly influenced by the instantaneous state of the lower ionosphere as it propagates in the waveguide formed by the Earth’s surface and the lower ionosphere. Thus, the amplitude and phase of the received signal carries information on the actual state of the lower ionosphere. The reflection height of VLF waves is in the 60–75/75–90 km altitude range on the dayside/nightside, respectively.

Gravity-wave-related amplitude perturbations in narrowband VLF signals have been detected in connection with the solar terminator [92,93], tropical cyclones and other large-scale meteorological systems [94,95,96,97], and a solar eclipse [98], as well as with thunderstorms [99,100]. In the present study, the occurrence of GWs in the lower ionosphere in relation to the “Zyprian” storm is investigated based on narrowband VLF amplitude data. The analysis is limited to the night periods, when GW-related amplitude fluctuations are much more frequently observed with this measurement technique [101].

Narrowband VLF data have been recorded at the Tihany Geophysical Observatory (TGO, 46.9°N, 17.89°E) in Hungary, Central Europe, since 2007. The station collects radio signals from a total of 17 VLF transmitters, of which 8 channels detect European transmitters. For the present study, the German VLF transmitter (call sign: DHO, frequency: 23.4 kHz) was selected, which in our experience, is the most stable of the 8 European VLF transmitters, at least in the studied period. One further advantage of the DHO transmitter is that the midpoint of the DHO-TGO propagation path lies close to the continuous Doppler sounding system operated by the IAP in the western part of the Czech Republic.

After reading the VLF amplitude data sampled at 20 Hz, median filtering was applied to remove outliers associated with measurement errors, and 15 s time series were computed by averaging. The daily data have been rearranged into noon-to-noon blocks to facilitate a joint analysis of the night periods, which were defined to cover the period between 20:30 and 02:30 UT. This time interval was selected based on the visual inspection of the amplitude recordings as the longest time interval unaffected by the strong amplitude variations associated with the solar terminator. Finally, low-frequency variations were filtered by subtracting the 12 min running average curve, and the continuous wavelet transform has been applied to determine the wave-like activity in the lower ionosphere at different periods (from 0.5 up to 30 min).

2.2. Spectral Analyses

As the atmospheric waves are quasi-periodic and time-limited oscillations of atmospheric parameters (temperature, pressure, etc.), suitable tools for detection are spectral analyses. In order to capture wave-like oscillations within a particular dataset, we apply Fourier transform and Wavelets (for instance [102]).

3. Results

Cyclone “Zyprian” (informally named this way by the Institute for Meteorology of the Free University in Berlin) is an event from July 2021. The center of extratropical frontal cyclone “Zyprian” moved from the Bay of Biscay across the English Channel into the North Sea. “Zyprian” determined the dynamic circulation pattern in the troposphere over the entire European continent for several days. A strong cold front separated the warm and moist air flowing into central and eastern Europe in the southwesterly cyclonic flow from the cold air in the rear of the cyclone over western Europe. The front was associated with a meridional jet stream in the upper troposphere. The rugged orography of central Europe led to the formation of local cyclones in the lower troposphere along the frontal boundary. High values of both low-level and deep wind shear, a high temperature lapse rate, and sufficient moisture create a convective environment suitable for the formation of extremely intense storms over central Europe. Severe thunderstorms occurred over central and eastern Europe over three days from 7 to 9 July. Extremely severe and well-organized storms moved in a relatively narrow band along a front across Austria, the Czech Republic, and Poland in the afternoon and evening of 8 July. Measured lightning activity and records of observed hazardous phenomena (torrential rainfall, hail greater than 5 cm in diameter) confirm both the high intensity and the persistence of these convective systems.

The detailed description of the development of the synoptic situation from 6 to 9 July, 2024, is given in Appendix D. The surface synoptic maps show the passage of the cold front, its waving over central Europe on 7 and 8 July, and its subsequent dissipation near the surface on 9 July. The upper level charts of 8 and 9 July show that the cold front was strongly pronounced throughout the vertical profile of the troposphere. At the same time, the charts of both days show significant vertical wind shear, which supported massive convection leading to the development of thunderstorm clouds and the generation of gravity waves.

Cyclone “Zyprian” occurred during a rather stable solar and geomagnetic activity situation. According to the space weather information issued by NOAA (ftp://ftp.swpc.noaa.gov/pub/warehouse/2021 accessed on 29 August 2024), geomagnetic field activity was quite unsettled for the period 5–11 July. Some extended periods of southward Bz were followed by a few isolated unsettled periods on 5–6 July. Another period of an unsettled state was observed on 10 July, which very probably has been caused by weak influence from a positive polarity coronal hole that rapidly decayed as it crossed the central meridian of the Sun. The remainder of the period was quiet. As it is seen on all the solar wind parameters (Bz, magnetic field angle, density, speed, temperature), the studied period is rather stable. The Kp value is less than 3 and the minimum of the Dst index during the period was −18 nT (at 6 July at 1 UT), which means a geomagnetically quiet, undisturbed period. The solar activity was also quite calm; the most energetic flare during the period was a C7.0 class flare occurring on 9 July around 10:50 UT. Figure 1 shows the recurrent high-speed solar wind streams originating in coronal holes, with a spike on 1 July 2021, several days before the onset of the “Zyprian” storm.

In Figure 2, we can observe that the Kp index, roughly during the period of the “Zyprian” storm, in most cases, did not exceed the value 3. This means that during the storm period, the geomagnetic situation can be described as calm. The ap index similarly indicates calm geomagnetic conditions. The following Figure 3 clearly shows that all the components of the geomagnetic field observed in the Budkov observatory remained quiet during the studied event.

3.1. Effects Observed in Troposphere

3.1.1. Tropospheric Observation

On 8 July, the convective environment for the formation of strong storms in Central Europe was characterized by very high values of deep and low-level wind shear, e.g., in Prague-Libuš at 12:00 UTC on 8 July, the shear in the 6 km layer was 29.6 m/s (in the 1 km layer 7 m/s), and in Munich, it was even 31.4 m/s. Mean CAPE values (Prague 430 J/kg, Munich 848 J/kg) and the high shear values were suitable for the development of organized convection with the possible occurrence of supercells or even tornadoes or other severe phenomena.

Along the axis of the low-pressure ridge, storms moved from the southwest across central Europe from northern Italy in a northeasterly direction to Poland, where they began to turn northward on 9 July. Numerous thunderstorms originating in the foothills of the Alps subsequently merged into two distinct mesoscale convective systems behind the Alps. Large hail was produced throughout the storms, with hail of 11 cm being recorded in Italy in the Piemonte region on 8 July and in Poland on 9 July and hail of 7 cm being reported in the Czech Republic on 8 July. In Italy, three tornadoes were observed, the strongest of F2 intensity in Veneto, another in Lombardy (F1) and Piedmont (F0), and one tornado was observed in Germany, in the Baden-Württemberg region (probably F0). The excessive lightning activity of the storms is presented in Figure 4, which shows the progression of the storms on 8 July towards the northeast and the northward turning of the storms on 9 July. These convective storms were therefore strong and well organized and moved in a relatively narrow band across central Europe, the strength of which is evidenced by a number of observations of dangerous meteorological phenomena.

The static image from polar-orbiting NOAA-19 satellite (Figure 5 and SUP SAT05) is a blended image of the visible and color-enhanced IR 10.75 µm band [103]. It contains information from two types of radiance in one image, from the reflected solar bands (showing cloud-top morphology) and from the emitted IR band (showing brightness temperature of the cloud), making this product very effective for the operational monitoring of convective storms, as well as for research purposes (case studies). In this image product, the coldest parts of storms are overshooting tops—parts of the storms where strong convective updraft penetrate the cloud-top equilibrium level, reaching up to 2 or 3 km into the lower stratosphere. These overshooting tops (together with their source updrafts) are probably the most efficient tropospheric source of the observed vertically propagating GWs [104,105].

The surface measurement record from the Institute of Atmospheric Physics (IAP) weather station in Figure 6a shows a typical cold front transition associated with thunderstorm activity. The front began to pass the station just before 16 UTC on 8 July, when the temperature began to fall rapidly, the pressure drop stopped, and the first showers appeared. Between 16 UTC and 03 UTC, the pressure value fluctuated considerably—up to 4 hPa in one hour—and the wind became gusty. This indicated the presence of severe storms. The sudden outflow of cold air together with precipitation (downbursts) and its spread over the ground caused sharp increases in pressure and wind speed. The frontal passage also caused a change in wind speed from easterly to westerly.

The surface measurement record at Sopron, Hungary in Figure 6b indicates the passage of a weakened cold front shortly after 20 UTC on 8 July, which dissipated at the surface after crossing the Alps. After the passage of the front, there was a sharp increase in air pressure and a change in the direction of wind speed from southeast to north. The precipitation recorded on 9 July confirmed the occurrence of a local thunderstorm in the afternoon. It formed in the area of the surface trough and passed over the Sopron station at 13 UTC. In the surface record, both a temporary increase in atmospheric pressure and a temporary decrease in air temperature reflected the outflow of cold air from the thunderstorm, the so-called gust front. The presence of early-afternoon thunderstorms around Sopron is also visible in the lightning distribution record (Figure 4).

3.1.2. Ground Infrasound Observations on 7–9 July 2021

Infrasound station PVCI registered two types of signals; pulse-shaped signals of median frequencies around 3.2 Hz dominated the observations (Figure 7a). Transiently a continuous signal of lower frequencies occurred (Figure 7b).

Pulse-shaped signals were first observed on 7 July, 2021, at 22:26 UTC and registered until 04:07 UTC on 8 July. The signals initially arrived from the back-azimuths of 127–144°, and the arrivals were gradually shifting to 99–116° with time (see Figure 8a). The back-azimuths followed the motion of the convective storms from the south to the north above the eastern regions of the Czech Republic (animation available in Supplementary Materials). The signals disappeared after 04:00 UTC, although the wind speed at PVCI was low until 08:00 UTC and detections were not masked by wind noise. The lightning activity ceased and decreased to zero after 04:00 UTC. The wind noise masked detections for most of the daytime on 8 July; the wind speed at the station was close to and repeatedly exceeded the critical value of 2 m/s between 08:14 and 19:27 UTC. Infrasound arrivals from the south-east occurred again at 19:33 UTC (Figure 8b). From 20:47 to 23:17 UTC, continuous signals prevailed with a few infrasound thunders recorded from 22:24 to 22:26 UTC. Minimum signal frequencies decreased down to 0.16 Hz, and the median of 1.6 Hz was lower compared to that of the lightning generated pulses. The signals arrived from a wide azimuth range between 125° and 165°. A gust front produced by the second wave of the storms was moving in this area, which could possibly be the source of the detected continuous signal. The back-azimuths point to the extensive convective storms in eastern directions from PVCI. From 00:30 to 04:15 UTC on 9 July, the observations were disturbed with gaps in the waveform data. Arrivals from the south-east were again registered at 02:44 UTC, at back-azimuths of 144°. The back-azimuths gradually shifted to the north-east, and the signals ceased after 08:40 UTC. Pulse-shaped signals prevailed from 02:44 UTC. The back-azimuths of 125–50° corresponded to the position of convective storms above eastern regions of the Czech Republic (animation available in Supplementary Materials).

Infrasound station PSZI did not detect the “Zyprian” event on 7 and 8, July which is likely due to unfavorable upwind conditions to the west. On 9 July, “Zyprian” moved further to the east. The array recorded the passage of the storms from 16:19 to 23:48 UTC. The back-azimuths gradually shifted from 236° to 12° and followed the path of the storms (Figure 9; animation available in Supplementary Materials). The recordings contained signals from lightning discharges, as well as continuous signals in the wide frequency range of 0.1–7 Hz.

Infrasound detections were compared to the lightning activity downloaded from the World Wide Lightning Location Network database. In the animations and in Appendix B, the lightning bolts’ origin time is shifted by the travel time calculated from its location to the infrasound array. This was performed in order to account for the time delay when analyzing the temporal and spatial correlation between the observations of the two technologies. For that direct wave arrival, an 300 m/s average speed was assumed. The lightning database was also filtered so that it only contained flashes in the azimuth range defined by the minimum and maximum values present in the infrasound bulletin and expanded by 10°. The association process is more detailed in, e.g., [106].

Further, the GW activity in the troposphere was analyzed using data from the WBCI large-aperture array. Figure 10 shows pressure recorded by the four sensors of the WBCI array (a), the periodogram (b), and the color-coded azimuth of propagation (c) from 7 to 9 July 2021. The azimuths are only displayed for large fluctuations that exceed 0.1 hPa in a given frequency band and if the estimated uncertainty in azimuth determination is less than 10°. Large amplitudes of GWs in the period range of about 20 to 60 min are observed on 8 July and on the beginning of 9 July, when the direction of propagation also changes from eastward to northward (blue color).

3.2. Effects Observed in the Stratosphere

The Earth Explorer Atmospheric Dynamics Mission Aeolus yields data from global observations of wind profiles from space using the active Doppler Wind Lidar (DWL) method [107]. The DWL measures 100 wind profiles per hour using both Rayleigh and Mie scattering methods [108]. The global wind profiles (along a single line-of-sight) are measured up to an altitude of 30 km to an accuracy of 1 m/s in the planetary boundary layer (up to an altitude of 2 km). The graphs in Figure 11 display the wind profiles measured by the Aeolus ESA satellite using the ALADIN instrument [109]. During the time period of three consecutive days (7–9 July 2021) based on Rayleigh scattering observations of wind profiles (in the middle panels) and based on the Mie MDS wind velocity profiles (on the right panels), an important feature is seen on the right panels. Red areas represent the subtropical jet, with blue being the polar jet. The gray part of the graph is without the lidar reflection, and the colored areas of the reflection indicate the storm. The upper panel (7 July) shows changes in the area in front of the Alps, the middle panel (8 July) over the territory of the Czech Republic, and on 9 July (lower panel), it can be seen that the polar jet became visible to the lidar as turbulent scattered reflection in the Czech Republic after the cyclone passage. The position of the polar jet is significantly shifted after the cyclone passed.

Plots in Figure 12 show the stratospheric specific humidity. Under normal conditions, the stratosphere is a dry region of the atmosphere, while the troposphere is a very strong dynamical barrier for most of the tracers including water vapor. There has to be exceptional input from the troposphere (i.e., a storm with the strong convective motion) to cross the tropopause barrier. The time evolution demonstrated on the plots shows the humidity increase above the tropopause in coincidence with the passage of cyclone “Zyprian”. We also analyzed other parameters, like temperature or winds, but the effect is visible mainly for the humidity.

3.3. Ionospheric Observation

First, we discuss the observation by CDS equipment. Figure 13 shows the recorded signal at two frequencies f = 4.65 MHz (Figure 13a–e) and f = 3.59 MHz (Figure 13f–j) for five consecutive days (6–10 July 2021), covering days around the event of the “Zyprian” passage. Only the sounding path from Panska Ves (50.528°N, 14.567°E) to Prague (50.041°N, 14.477°E) was selected for clarity. On the day before the studied event, a rather nice clear echo is seen on both frequencies. In the night hours, wave-like oscillations are visible on 6 July until the morning 8 July. Then, the character of the echo changes substantially into the spread echo in coincidence with the frontal passage. The wide-range noise around 17 UT on 8 July is caused by lightning nearby the receiving site. It clearly indicates the moment of the transition of the frontal border with a squall line across the measurement site.

The spread echo remains on the spectrograms until the next day. Then, the observed echo intensity decreases. The observed spread echo indicates that equi-density planes in the ionospheric plasma are not well defined. The reflected electromagnetic waves come from a wider range of heights and from different directions. Such a situation is often associated with the presence of atmospheric waves and small-scale irregularities. From the CDS spectrograms, it is well seen that the ionosphere is disturbed at the beginning of the transition of cyclone “Zyprian”.

The DPS 4D instrument provides two basic experiments—measurement of the ionogram and plasma drift. The ionograms provide the dependence of reflection height based on the sounding frequency. It provides information about main parameters—critical frequencies (maximum plasma frequencies of the layers) foE, foF1, and foF2 and virtual heights of the layers (h′E, h′F1, h′F2), height of layer maxima (hmE, hmF1, hmF2), and also qualitative information about special stratification (spread, cusp, etc.). The NHPC inversion routine is used to further derive the true-height profiles of the electron concentration [104].

Panels in Figure 14 show critical frequencies of F layer foF2 [MHz], the critical frequency of E-layer foE [MHz], peak heights of the F2 region hmF2 [km], and peak heights of E-layer hmE [km] with median values (red lines and dots). A comparison with monthly median values of foF2 shows that the days 6–10 July 2021, match the median values well. A significant decrease in minimal values of foF2 shortly after the “Zyprian” frontal passage occurred. During the observed days, a slight decrease in variance of foF2 compared to preceding days can be identified. The critical frequency foF2 stayed at lower values for the following three days compared to median values, as well as the days prior to the “Zyprian” passage. On both stations, lower concentration is recorded on the entire profiles for the first three days, including the “Zyprian” day.

All the ionograms were manually scaled and interpreted. A well-pronounced spread F echo was observed on DPS 4D ionograms after the “Zyprian” passage. This is in agreement with the CDS observation showing spread echo on both sounding frequencies. Spread F echo is a typical feature observed in connection with wave-like propagation from a lower atmospheric region and/or auroral zone.

Figure 15 shows examples of recorded ionograms. Before the arrival of “Zyprian”, the recorded ionograms were unusually noisy. The echo from the E layer is attenuated. Reflections from F1 and F2 are observable within the noisy ionograms (Figure 15a). After the “Zyprian” passage (Figure 15b,c), the noise based on frequencies below 3.5 MHz remains. Sporadic E starts to form (Figure 15b) and remains present (Figure 15c,d) during the night. Split on the F2 trace (better seen on extraordinary trace) is observable (Figure 15b,c). Figure 15d shows an ionogram with spread F echo. The spread F echo is recorded on ionograms 19.15–3.30 UT indicating a disturbed ionosphere. The ionosphere remains disturbed during the night and then returns to the regular stratification. During the day before the cyclone passage, the recorded ionograms indicate a well-stratified ionosphere without irregularities or undulation caused by propagating waves.

The situation observed by Sopron DPS 4D slightly differs. Sporadic E stratification is observed in almost all ionograms. Figure 16a shows a rather well-pronounced echo from all the layers. On the trace F2, there is a cusp present close to the peak. Similarly to Průhonice, the noise occurs on ionograms (Figure 16b–d), but it is lower. Figure 16c,d shows the spread F echo. This characteristic feature on the ionograms remains until 21:00–02:45 UT. Strong sporadic E is recorded practically throughout the night. For some of the observed sporadic E traces, the off-vertical echoes are very strong.

After the manual scaling of ionograms, the reflection pairs of the frequency and reflection height are further recomputed using the inversion technique NHPC to obtain the real-height profiles. Their temporal evolution is presented in the form of a profilogram in Figure 17. At both stations Průhonice and Sopron, lower concentration is recorded on the entire profiles for the first three days, including the “Zyprian” day. It seems that the lowest electron concentration is observed during the event day; however, due to blanketing by the sporadic E, it cannot be accurately analyzed. After the frontal passage, the electron density increases at both stations. Solar radiation is rather stable during the analyzed days, and the values of solar flux F10.7 cm are characterized by a mean value of 72.1 sfu, median value of 69.1 sfu, and EUV index Bremen composite MG II index mean value of 1.5464 and median value of 1.5442.

Directograms represent a way to obtain information about plasma flow from the ionogram measurement. The plot in Figure 18 demonstrates a vertical plasma motion at a height close to hmF2. The echo below the F2 layer peak recorded during regular ionogram sounding is used. The increase in the horizontal plasma flow is clearly visible in the middle part of the measurements of both stations. There is an apparent increase in the motion immediately after the cyclone passage above Průhonice, while the increase above Sopron is less pronounced but is also observable the following day. As it has been shown and illustrated by the satellite chart in the tropospheric part, the ionosphere above Sopron was influenced by a system somewhat less compared to Průhonice, and hence, it might be the reason why the change of plasma flow is less pronounced. Changes in the color indicate plasma flow shears. In this particular case, the flow changes from eastward to westward and back.

Figure 19 shows a sequence of SKYmaps that represents the follow-up drift measurement after the ionogram sounding. Plasma flow is monitored in a narrow interval of frequencies below foF2. Figure 19 demonstrates how the plasma flow rapidly changes within a short time. The measurement represents plasma flow at approximately the same height below the F2 peak. In Figure 19a (18:49 UT), the plasma flows in the north direction with the horizontal velocity v_h = 262 ± 141 m.s⁻¹. Recorded reflection points used for the evaluation of horizontal velocity form a clear bipolar pattern. Figure 19b shows a situation 60 min later (19:49 UT) when the observed plasma flow is substantially smaller, v_h = 21 ± 2 m.s⁻¹, and the distribution of the reflection points does not allow for a clear determination of the prevailing direction of the plasma flow. Figure 19c (21:04 UT) shows a similar situation as Figure 19a when the bipolar pattern of the reflection points is recorded. The prevailing plasma flow is in the north-east direction with a velocity v_h = 154 ± 116 m.s⁻¹. The last plot Figure 19d (21.34 UT) shows the decrease in the horizontal velocity v_h = 82 ± 57 m.s⁻¹ without an apparent bipolar pattern of the reflection points. It should be emphasized that geomagnetic activity is low, and hence, the distortion of the plasma flow due to auroral region activity is unlikely to be observed.

Figure 20 shows the observed VLF amplitudes and the corresponding nighttime wavelet spectrograms between 7 and 10 July 2021. One can clearly notice in this figure the recurrent nature of the VLF amplitudes from day to day, the pronounced difference between the daytime and nighttime signal levels, and the major amplitude changes associated with the passage of the solar terminator. During the daytime, except for early morning, signal losses between 7 and 8 UT and a few minor solar flares (e.g., on 9 July at about 11 UT), the amplitudes are more stable than during the night, when short periodic fluctuations can be observed. These fluctuations, for which wave periods fall in the typical period range of GWs (see the wavelet spectrograms), occur on all three nights but are most pronounced on 9 July, following the most intense period of the “Zyprian” storm.

Figure 21 presents the observed VLF amplitudes and the corresponding wavelet spectrogram for the night of 8–9 July. On this night, which follows the most intense period of the Zyprican storm, fluctuations in the VLF amplitudes typically fall within the 5–15 min period range. Significant wave activity can also be observed around the 20–25 min period range at the beginning of the night and below 5 min after midnight.

4. Discussion

Extratropical cyclone “Zyprian” occurred at the beginning of July 2021 under quiet geomagnetic conditions with stable solar forcing. It dominated weather above Western Europe. The large frontal border was characterized by a high temperature drop with strong cyclogenesis. The frontal boundary moved rapidly across Europe. High wind gusts were recorded. Heavy rains caused large damages, especially in Germany. In the troposphere, a convective environment formed. It was characterized by high values of deep low-level windshear. Overshooting tops of clouds and strong updrafts were identified on the satellite observations. An excessive lightning activity was detected. Infrasound measurements recorded thunder-related signals from extensive convective storms. The azimuths corresponded to the position of convective storms above the analyzed region. The large-aperture infrasound array data revealed the occurrence of gravity waves with periods 20–60 min.

Overshooting tops of clouds and strong updrafts are recognized to be an efficient source of GWs. Above them, the stratosphere was dumped by humidity, which was seen well on the stratospheric reanalysis data. Further, in the stratosphere, a shift in the polar jet stream after the cyclone “Zyprian” passage was observed by the satellite Aeolus.

At the ionospheric heights, irregular stratification of the ionosphere was a well-pronounced feature. Such types of irregularities are usually observed in connection with the propagation of GWs. From the radio wave reflection echo, it was seen that reflection planes were distorted. Strong spread F echo was observed by both DPS 4D and CDS equipment during the “Zyprian” event. The distortion of isodensity planes was observed until the following day. Then, the ionosphere slowly returns to the normal stratification. A significant decrease in the critical frequency foF2 and electron density depletion in the entire profile was observed on the day of cyclone “Zyprian” passage above both ionospheric stations. The observed depletion of electron density was followed by a substantial increase on the following day. An increase in the hmF2 is seen on data from Průhonice station. Statistical analysis reveals effects on height of the E layer hmE on both stations. An increase in hmE is observed during the day of cyclone passage in Sopron data and the following day in Průhonice data. We deduce that propagating GWs affect the entire electron density profile. Analyses of plasma flow indicate an increase in horizontal velocities and fast changes in velocity values and in directions. Narrowband VLF data analysis identified an important wave activity in the 5–15 min period range after the cyclone passage and around the 20–25 min period range at the beginning of the night and below 5 min after midnight.

Our main findings can be summarized as follows:

• Extratropical cyclone “Zyprian” was formed under stable solar and low geomagnetic activity;
• Extratropical cyclone “Zyprian” dominated weather above (Central) Europe;
• A severe convective environment was formed with updrafts and overshooting tops of clouds;
• Excessive lightning activity, hails, wind gusts, and floods were observed;
• The stratosphere was dumped by humidity above clouds overshooting the tropopause;
• Lightning and motion of the convective storm were detected by infrasound arrays;
• The gravity wave structure with periods around 20–60 min propagating first eastward after the cyclone passage and later to northward were observed in the troposphere;
• The position of the polar jet is significantly shifted after the cyclone passage;
• The undulation of equidensity planes is manifested by specific type ionograms recorded by DPS 4D;
• Irregular stratification is recorded on ionograms (spread F, splits, cusps, etc.);
• Departures from the regular daily course of foE, foF2, hmE, and hmF2 are observed;
• Depletion of electron concentration in the entire profile during the day of cyclone passage followed by a substantial increase the day after;
• The horizontal component of the plasma drift velocity changes rapidly in both the direction and value;
• Directograms show a substantial increase in values of horizontal flow at the hmF2 height;
• No prevailing plasma motion in the horizontal plane can be identified;
• Gravity wave activity is observed in the lower ionosphere in two domains of about 5–15 min and 20–25 min period ranges.

5. Conclusions

Recently, the occurrence of severe weather systems is reported in connection with the global change. They bring severe weather and often cause damages in the infrastructures. As documented by many experimental and theoretical studies, such systems are effective generators of gravity waves that eventually propagate up to the upper atmosphere where they could have a significant impact. Our multiinstrumental observation of extratropical cyclone “Zyprian” describes observed changes in the atmospheric regions from troposphere up to the ionospheric F2 layer.

The observed variability in the properties of the lower and upper atmosphere is attributed to the influence of the extratropical cyclone as it developed and evolved during stable solar forcing and rather low geomagnetic activity. The observed variability we consider to be induced by the severe convective environment that developed in the troposphere due to coincidental time of occurrence across the atmosphere. Most of the observed changes in the ionosphere cannot be explained by corresponding changes in geomagnetic conditions and in the incoming solar radiation. Despite the large dataset, we are aware of the limitations of this study, due to height gaps (for instance in mesosphere) where data observation is limited or not possible.

It is important to carefully investigate the ionospheric effects of strong tropospheric phenomena in the future. Anthropogenic pollutants affect the lower atmosphere, where they induce the well-known global change phenomenon, leading, among other effects, to more frequent severe tropospheric events. However, they also affect the upper atmosphere (mesosphere, thermosphere, and ionosphere). Increased gravity wave activity triggered by strong tropospheric phenomena significantly affects the proper functioning of GNSS applications.