Session 5 | Radar and non-satellite remote sensing studies of storms

Session 5

Radar and non-satellite remote sensing studies of storms

Orals TH6

| Thu, 20 Nov, 16:45–18:15 (CET)|Room Hertz Zaal

Posters TU4

| Attendance Tue, 18 Nov, 14:30–16:00 (CET) | Display Mon, 17 Nov, 09:00–Tue, 18 Nov, 18:30|Poster area, P34–3

Posters TH4

| Attendance Thu, 20 Nov, 14:30–16:00 (CET) | Display Wed, 19 Nov, 09:00–Thu, 20 Nov, 18:30|Poster area, P34–3

Orals: Thu, 20 Nov, 16:45–18:15 | Room Hertz Zaal

16:45–17:00

ECSS2025-269

Three years of monitoring severe hailswaths across Canada using radar

Julian C. Brimelow, Mark Gartner, and Sudesh Boodoo

One of the primary goals of the Northern Hail Project (NHP) is to generate the first detailed hail climatology for Canada. To do this, we are adopting several independent approaches. One of these is the use of the Maximum Expected Size of Hail (MESH) from radar data. Most of the 33 new S-band dual-polarization radars have been operational since early 2022. Using these radar data allowed us to identify hailstorms at high-spatial temporal resolution wherever we have radar data. In this research, we focused on manually identifying (and digitizing) severe hailswaths using the MESH data from 2022 through 2024. A severe MESH hailswath is one that has a continuous 40 km long or greater track of 10 mm pixels, and must include at least 2 adjacent pixels of 30 mm or greater. A total of almost 2,000 severe hailswaths have been identified by radar to date. Although the regional year-to-year variability is significant, our analysis has identified the Canadian Prairies and far western Ontario as hot spots for long-lived, severe hailstorms. Some of the severe hailswaths in the dataset are impressive, extending over 500 km and lasting up to 6 hours. The widest hailswath in our MESH dataset is approximately 50 km across. Even though most of the MESH hailswaths in our database have occurred near or just to the north of the Canadian/U.S border, some hailswaths have occurred at the edge of our available radar network range, with the most northern MESH hailswath terminating at a latitude of 58.0 degrees north in Saskatchewan. Moving forward, we will continue to monitor severe hailswaths using a semi-automated algorithm that draws on machine vision and machine learning techniques and will be trained on the existing dataset. In 2022, we used the MESH product produced by the National Oceanic and Atmospheric Administration-Multi-Radar Multi-Sensor (NOAA-MRMS) and switched to the Environment and Climate Change Canada (ECCC) MESH product in 2023 and 2024. Although the products from the two groups are generally in good agreement, we noted that there are notable differences in the MESH values at times. Reasons for these discrepancies are being investigated using ground reference data collected by the NHP.

How to cite: Brimelow, J. C., Gartner, M., and Boodoo, S.: Three years of monitoring severe hailswaths across Canada using radar, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-269, https://doi.org/10.5194/ecss2025-269, 2025.

17:00–17:15

ECSS2025-208

Object-based hail-size detection and nowcasting in Switzerland using random forests on polarimetric radar data

Martin Aregger, Olivia Martius, Urs Germann, and Alessandro Hering

Hail poses a significant and growing economic threat to Switzerland. In recent years, both large and small hail have caused substantial damage, with large hail leading to record property losses reported by insurance companies. However, small hail is also not without peril. It can facilitate flooding by clogging drainage systems when it occurs in large quantities. Additionally, it may cause devastating damage to crops, especially in combination with strong winds. Consequently, to better understand the hail hazard, a comprehensive detection of hail of all sizes is needed.

Radar-based hail detection in the complex Swiss orography faces significant challenges regarding retrieval quality and visibility due to beam-shielding and ground clutter effects. The recent upgrade of the national weather radar network (Rad4Alp) to polarimetric C-Band Doppler radars has brought significant improvements in both data quality and availability, providing a rich dataset for analysis.

Here, we present a new machine learning-based hail size discrimination algorithm using polarimetric radar data trained on a unique dataset of high-density crowdsourced hail reports. The algorithm uses an object-based approach; Individual storms are detected, and a wide range of predictors is extracted from newly created polarimetric radar composites for each storm. The resulting dataset is used to train a random forest for hail size detection, which significantly outperforms the currently operational hail size discrimination algorithm MESHS (maximum expected severe hail size).

We further assess the importance of various radar signatures for hail size classification through feature importance analysis of the trained model. The most impactful predictors include both conventionally used quantities, such as echotops and maximum reflectivities, and lesser-known ones like detected ice hail column height and hail differential reflectivity (HDR). Finally, we demonstrate the potential of the generated object-based dataset for hail size nowcasting, indicating its broader utility beyond discrimination.

How to cite: Aregger, M., Martius, O., Germann, U., and Hering, A.: Object-based hail-size detection and nowcasting in Switzerland using random forests on polarimetric radar data , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-208, https://doi.org/10.5194/ecss2025-208, 2025.

Discussion Post a comment

17:15–17:30

ECSS2025-217

A data-driven approach to predicting Severe Weather utilizing dynamic definition via dual-polarization products

Matej Murín, Matej Choma, and Jakub Bartel

As extreme weather events grow more frequent and severe, timely and accurate severe weather warnings are becoming more critical than ever before. With the complex dynamics of weather phenomena, it is primarily thanks to weather radars that operational meteorologists are able to monitor these situations and give out warnings for areas with high possibility of occurrence. With the growing availability of computational resources and the rise of data-driven methods, applying machine learning to severe weather nowcasting has become both feasible and increasingly effective. However, the classification of weather to be severe is often domain-specific and subjective, leading to a difficult objective definition that would stay true in any given geographical location. It is for this reason that we at Meteopress have trained a deep learning model that is able to work with dynamic definitions. When severe weather is defined as a combination of dual-polarization radar variables, such as copolar correlation coefficient or specific differential phase, paired with reflectivity, the model is able to adapt to these definitions and produce domain-specific outputs. Moreover, it enables the user to define what kind of weather they are interested in. It utilizes recent and current weather radar measurements to produce outputs 90 minutes into the future. The model’s output can be calibrated through targeted evaluation to meet specific performance criteria, such as maintaining a minimum expected recall or precision.

How to cite: Murín, M., Choma, M., and Bartel, J.: A data-driven approach to predicting Severe Weather utilizing dynamic definition via dual-polarization products, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-217, https://doi.org/10.5194/ecss2025-217, 2025.

17:30–17:45

ECSS2025-137

Radar Updraft Proxies for Supercell Tornadogenesis Prediction

Michael French, Erik Creighton, and Darrel Kingfield

The most fundamental feature of a convective storm is its updraft. The updraft plays a crucial role in the life cycle of a storm and the production of its hazards, including tornadoes. Recent studies, mostly using numerical models, have proposed direct links between updraft properties and tornado formation. However, it is difficult, outside of specialized field campaigns, to directly sample the vertical velocities that define a storm’s updraft. A lack of observations complicates efforts to verify proposed connections from theory and modeling studies between characteristics of storm updrafts and tornadogenesis.

An alternative to direct observations of updrafts is to leverage features in radar data that result from vigorous updrafts. In this way, characteristics of a remote sensing updraft proxy may be used as an estimate of the characteristics of the storm updraft. For our work, we use a polarimetric radar signature, the Z_DR column, as a proxy for supercell updrafts. We aim to study the theorized connection between tornadogenesis and both updraft size and updraft vertical alignment. The former link results from a 2021 paper that found evidence that storms with larger Z_DR column areas were more likely to be tornadic than non-tornadic. The latter inquiry is based on studies that have found that supercell updrafts that are more vertically aligned (upright) are more likely to produce tornadoes.

In this study, we analyze a large sample of Z_DR column areas and vertical alignments in tornadic and non-tornadic supercells. For the updraft area part of the study, we examine a large number of WSR-88D volumes with a focus on whether Z_DR column areas ≥ 40 km², which we call “immense” updrafts, are only present in tornadic supercells. If so, the implication is that an immense updraft serves as a sufficient condition for imminent tornado formation. For the updraft alignment part of the study, we use the centroid of the Z_DR column at storm midlevels and the storm hook echo at low levels to estimate updraft tilt in supercells; we also use radial velocity data to estimate mesocyclone tilt. The main objective is to identify if there is a difference in updraft (or mesocyclone) tilt between tornadic and non-tornadic supercells. In addition, we determine if there are tilt “cutoff” values large enough to significantly lower the probability of tornado formation. We also discuss study limitations in using radar proxy data and the implications of study results on operational nowcasting of imminent tornadogenesis.

How to cite: French, M., Creighton, E., and Kingfield, D.: Radar Updraft Proxies for Supercell Tornadogenesis Prediction, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-137, https://doi.org/10.5194/ecss2025-137, 2025.

17:45–18:00

ECSS2025-284

How well can European radars detect supercells and tornadoes?

Bram van 't Veen, Pieter Groenemeijer, and Tomas Pucik

As more European weather services release their single-site volumetric radar data as open data, ESSL has integrated a large number of these radar sites into a unified viewer. This tool enables the monitoring of convective storm activity across a broad region of Europe, including countries such as France, the Netherlands, Germany, Denmark, Czechia, Austria (partially), Poland, Romania, Estonia, and Finland.

The web-based interface offers intuitive functionality, allowing users to zoom in and out, switch between radar sites, change scan elevation, and easily navigate through time. It also provides seamless switching between various radar parameters, including reflectivity (Z), radial velocity (V), correlation coefficient (CC), differential reflectivity (ZDR), and, for some radars, specific differential phase (KDP). A promising method for identifying (large) hail involves the use of an RGB composite image, where reflectivity, correlation coefficient, and differential reflectivity are represented by the red, green, and blue channels, respectively.

Using several case studies involving strong supercells, both tornadic and non-tornadic, we analyzed how effectively mesocyclones and tornadic rotation can be identified. The results highlight the influence of radar configurations, primarily C-band, across different countries. Key factors affecting detection include the use of dual- or triple-PRF (Pulse Repetition Frequency), which extends the Nyquist velocity range, as well as the extent of velocity data filtering. Other relevant factors include scan update intervals, gate length, beamwidth, and scanning strategies.

How to cite: van 't Veen, B., Groenemeijer, P., and Pucik, T.: How well can European radars detect supercells and tornadoes?, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-284, https://doi.org/10.5194/ecss2025-284, 2025.

Discussion Post a comment

18:00–18:15

ECSS2025-157

Framework for weather radar data processing.

Piotr Szuster and Joanna Kołodziej

This study introduces a comprehensive framework for generating 3D Cartesian products from weather radar network data to support operational forecasting and numerical weather prediction. Traditional radar data is typically represented in spherical-polar coordinates, which complicates composite product generation, geospatial referencing, and assimilation into weather models. To address this we propose a system that converts these data into a unified 3D Cartesian format, enabling more flexible product generation and visualization.

The framework is modular and includes components for data acquisition, transformation, visualization, and export. It supports various radar data formats, including proprietary ones, and implements multiple interpolation techniques and sampling strategies. The system was developed in C# as a standalone Windows Forms application for interactive analysis and rapid development.

Experimental evaluation demonstrated the system’s ability to produce 72 products, such as CAPPI, VIL, and Echo Tops, from both single and multi-radar datasets efficiently. The processing time scaled predictably with domain size and number of radars. Though current evaluations focused on runtime and functional validation, future work will emphasize optimizing interpolation methods and enabling autonomous operations for real-time forecasting.

The proposed framework represents a robust solution for converting complex radar datasets into actionable 3D products, improving forecasters’ situational awareness and enhancing input for numerical weather models.

How to cite: Szuster, P. and Kołodziej, J.: Framework for weather radar data processing., 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-157, https://doi.org/10.5194/ecss2025-157, 2025.

Posters TU4: Tue, 18 Nov, 14:30–16:00 | Poster area

Display time: Mon, 17 Nov, 09:00–Tue, 18 Nov, 18:30

P34

ECSS2025-2

Validating the Swiss automatic mesocyclone detection algorithm over the Highveld of South Africa: A case study event

Christina Liesker, Liesl Dyson, and Monika Feldmann

Supercell thunderstorms, characterised by a rotating updraft (mesocyclone), are known to produce severe weather. They frequent the South African Highveld during spring and early Summer (October to December). On the 28^th of November 2013 at least seven supercells moved across the Gauteng province, with golf to tennis ball sized hailstones reported. This event was one of the two supercell event days in that month, that contributed to just over EUR 101 million in insurance claims. While supercells can be manually identified using Doppler radar, algorithms to automatically detect the mesocyclone on the Doppler velocity field were first developed in the 1980s and have since been improved and adopted globally. The use of such algorithm is novel to South Africa. This study aims to evaluate the performance of the mesocyclone detection algorithm developed by Switzerland (with the same operational parameters), in detecting supercell thunderstorms on the Irene single polarised S-band radar, during the 28^th of November 2013 case study. This case study was chosen for its complex thunderstorm dynamics within a highly sheared environment, including the development of severe multicells, supercells and bow echoes. The results of the mesocyclone detection algorithm for cyclonic (clockwise) rotation were compared to the manual supercell database. In addition, the reflectivity and Doppler velocity field were reanalysed to ensure events were not missed. All supercells within the manual database were identified by the algorithm and some additional events and time steps were accurately detected. However, at least 5 events were not associated with a supercell and events were often identified earlier and/or ended later than manually detected. Cyclonic rotation associated with the poleward side of a bow echo often occurred within this case study. This is a first step towards automatically identifying supercells with an overall goal of improving the nowcast and warning of such events. Future work will include adjusting the algorithm thresholds to improve detection, as it was originally set to account for the complex terrain and lower shear environment over Switzerland.

How to cite: Liesker, C., Dyson, L., and Feldmann, M.: Validating the Swiss automatic mesocyclone detection algorithm over the Highveld of South Africa: A case study event, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-2, https://doi.org/10.5194/ecss2025-2, 2025.

Discussion Post a comment

P35

ECSS2025-48

PrecipType: Dual-pol C-band radars for analyzing the precipitation type on the ground

Felix Erdmann, Sylvain Watelet, Maarten Reyniers, and Dieter R. Poelman

Belgium has complete coverage with dual polarization Doppler radars. The horizontal and vertical polarizations allow not only the estimation of precipitation amounts, but also gathering of additional information about the detected hydrometeors, e.g. their size and shape. Horizontal reflectivity in combination with radar variables using signals from both polarizations, such as differential reflectivity, correlation coefficient and specific differential phase, are further analyzed to estimate the hydrometeor class at the height of the radar beam. Hydrometeor classification algorithms (HCAs) suggest the most probable hydrometeor class for each pixel of the radar image. Then the ground transition applies the method of Steinert et al. (2021) [1] to estimate the precipitation type on the ground.

Currently, four different HCAs are being tested at the Royal Meteorological Institute of Belgium (RMI) and Belgian C-band radars: (i) an algorithm adapted from the Australian Bureau of Meteorology Research Center (BMRC) [2], (ii) the Dolan algorithm [3], (iii) the Besic algorithm [4] and (iv) the HCA based on Zrnic et al. (2001) [5] as implemented in the Python library Wradlib [6].

Our poster briefly introduces the HCAs and then compares their results for cases of winter weather and convective storms in 2025. Reports from the RMI weather app from the population are used to verify the results.

References:

[1] Steinert, J., P. Tracksdorf, and D. Heizenreder, 2021: Hymec: Surface Precipitation Type Estimation at the German Weather Service. Wea. Forecasting, 36, 1611–1627, https://doi.org/10.1175/WAF-D-20-0232.1.

[2] based on Keenan, T., 2003: Hydrometeor classification with a C-band polarimetric radar, Aust. Met. Mag. Hydrometeor 52 (2003) 23-31.

[3] Dolan, B., S. A. Rutledge, S. Lim, V. Chandrasekar, and M. Thurai, 2013: A Robust C-Band Hydrometeor Identification Algorithm and Application to a Long-Term Polarimetric Radar Dataset. J. Appl. Meteor. Climatol., 52, 2162–2186, https://doi.org/10.1175/JAMC-D-12-0275.1.

[4] Besic, N., Figueras i Ventura, J., Grazioli, J., Gabella, M., Germann, U., and Berne, A., 2016: Hydrometeor classification through statistical clustering of polarimetric radar measurements: a semi-supervised approach, Atmos. Meas. Tech., 9, 4425–4445, https://doi.org/10.5194/amt-9-4425-2016.

[5] Zrnić, D. S., A. Ryzhkov, J. Straka, Y. Liu, and J. Vivekanandan, 2001: Testing a Procedure for Automatic Classification of Hydrometeor Types. J. Atmos. Oceanic Technol., 18, 892–913, https://doi.org/10.1175/1520-0426(2001)018<0892:TAPFAC>2.0.CO;2.

[6] https://docs.wradlib.org/en/latest/notebooks/classify/2d_hmc.html, last access 2025/05/22

How to cite: Erdmann, F., Watelet, S., Reyniers, M., and Poelman, D. R.: PrecipType: Dual-pol C-band radars for analyzing the precipitation type on the ground, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-48, https://doi.org/10.5194/ecss2025-48, 2025.

Discussion Post a comment

P36

ECSS2025-60

Ionospheric respons to gravity waves induced by tropospheric deep convection in mid-latitudes

Kateřina Potužníková, Petra Koucká Knížová, Mošna Zbyšek, Kouba Daniel, Chum Jaroslav, Petr Zacharov, and Martin Setvák

Our case study investigates coincidental occurrence of intense tropospheric convection and upper atmospheric variability, focusing on gravity wave (GW) generation in the tropospheric height and its possible effects on the ionospheric layers E, sporadic E (Es), and F2. Our analysis presents: several extreme convective storm events that occurred over Central Europe under tropical air mass conditions, not directly linked to the passage of a cold front. Two primary classifications of convective systems are considered: (1) rapidly propagating supercells characterized by strong, rotating updrafts; and (2) long-lived linear mesoscale convective systems (MSCs) connected to derecho. The latter are distinguished by their broad horizontal extent, long lifetimes of several hours and storm trajectories exceeding 400 km. Based on satellites data, meteorological radar data, and ionospheric radar and Doppler sounding techniques, we investigate ability of such types of severe weather events acting as an effective sources of gravity waves capable of propagating into the middle and eventually reaching the upper atmosphere. We focus on effects of GW within the stratosphere and ionosphere in E, Es, and F2 regions. Results contribute and enlarge our investigation of connection between the intense activity of the lower atmosphere and the dynamics of the upper atmosphere in mid-latitudes.

Our previous studies identified simultaneous (a few hour delay) effects in the ionospheric parametrs from the lower ionosphere up to peak of F2 layer. Our study indicates differences in the ionospheric response according to the GW source. In particular, we focus on sporadic E characteristics, its occurrence and blanketing ability. Es layer formation in midlatitude, during low geomagnetic conditions and stable solar forcing, is dependent on neutral atmosphere motion, hence is influenced by GW propagation or depositing their momentum. Sporadic E layers in well pronounced summer phenomenon, while the tropospheric convective events appeared to intensify according to recent climatic studies.

This research was supported by the Johannes Amos Comenius Programme (P JAC), project No. CZ.02.01.01/00/22_008/0004605, Natural and anthropogenic georisks, by the GA ČR GA25-14158L, and also by the ESA project HORIZON 2020, PITHIA-NRF 101007599.

How to cite: Potužníková, K., Koucká Knížová, P., Zbyšek, M., Daniel, K., Jaroslav, C., Zacharov, P., and Setvák, M.: Ionospheric respons to gravity waves induced by tropospheric deep convection in mid-latitudes, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-60, https://doi.org/10.5194/ecss2025-60, 2025.

P37

ECSS2025-76

Polarimetric radar observations of the Jersey tornado and hailstorm of 1 – 2 November 2023

Matthew Clark

At approximately midnight on 2 November 2023, a supercell thunderstorm struck Jersey in the Channel Islands. This thunderstorm formed just ahead of the cold front of an intense extratropical cyclone (Storm Ciarán) which subsequently produced widespread damaging winds over northern France and adjacent areas during the morning of 2 November.

The supercell thunderstorm produced a tornado of intensity T6/IF3 which tracked over eastern parts of Jersey, and hail of diameter ≥5 cm over a swath several kilometres wide in central and eastern parts of the island. The tornado damage track was continuous from the south coast of the island at St Clement, where the tornado made landfall, to the northeast coast at Fliquet where the tornado exited the island.

The tornadic storm passed within a few kilometres of the Channel Islands Doppler, polarimetric, C-band radar, located on the south-western tip of Jersey. In this presentation, radar observations of the storm will be explored, focussing on the signature of a tornado debris cloud in polarimetric fields and a collocated reflectivity ‘debris ball’. The observations meet the accepted operational definition of a Tornado Debris Signature (TDS), comprising a small region of exceptionally low correlation coefficient (<0.8) collocated with differential reflectivity ≤0 dB, reflectivity >35 dBZ and a velocity couplet in the radial wind field. To the best of our knowledge, this is the first documented polarimetric TDS in northwest Europe. In vertical section (constructed from available plan position indicator scans at different elevation angles, translated horizontally to account for the movement of the storm between scan times), the debris signature comprised a column of low correlation coefficient and high reflectivity that tilted north-northeast with height and extended to at least 2 km above ground level.

How to cite: Clark, M.: Polarimetric radar observations of the Jersey tornado and hailstorm of 1 – 2 November 2023, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-76, https://doi.org/10.5194/ecss2025-76, 2025.

Discussion Post a comment

P38

ECSS2025-94

Determining the location of feeder clouds in a convective storm system based on radar observations

Tsvetelina Dimitrova, Evgeni Livshits, Viktoria Pencheva, Stefan Georgiev, and Denislav Bonchev

The movement of convective storms is directly dependent on the direction and speed of their evolution. Radar, aircraft, and satellite studies have established that this is due to merging of inflow feeder clouds (FCs). FCs are responsible for the intensification of the thunderstorms. One of the key hail suppression concepts is the early rainout of feeder clouds. Therefore, accurately determining the location of FCs within the convective storm system is a priority for hail suppression operations.
Convective storm development can be visualized as a continuous process of formation, growth, dissipation, merging, or splitting. By placing the origin of a Lagrangian coordinate system at the center of a convective cell, the internal dynamics of the convective storm and the overall convective system can be tracked.
To determine the location of FCs, the following vectors are used: the mean wind vector (Vc) – wind at 600 hPa level, the storm motion vector (Vs), and the evolution vector (Ve) – which indicates the direction and speed of merging of the main convective cell and FCs, defined as vector difference between Vs and Vc.
High-resolution visible (HRV) satellite data were used to verify the identified FCs locations.
Convective storms occurring on June 3 and 12, 2024, were analyzed. The observed convection included multicellular storms, with a supercell developing on June 12. Radar data and evolution vectors, calculated using the Lagrangian method, were used to assess storm evolution and FCs positioning.
The study showed that:
- The calculated locations of FCs based on the evolution vectors closely match the visual feeder cloud structures observed in satellite imagery.
- The orientation of the feeder clouds can be estimated by using radar data.
- In both multicell and supercell storms, the orientation of most feeder cloud lines was consistent, suggesting their potential use as predictors for subsequent severe storm development.

How to cite: Dimitrova, T., Livshits, E., Pencheva, V., Georgiev, S., and Bonchev, D.: Determining the location of feeder clouds in a convective storm system based on radar observations, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-94, https://doi.org/10.5194/ecss2025-94, 2025.

P39

ECSS2025-98

Cyclic supercell Storm in Bulgaria observed on 13 June 2024: thermodynamic conditions, evolution and structure

Tsvetelina Dimitrova, Nadezhda Kadiyska, Stefan Georgiev, and Viktoria Pencheva

In 2024, several supercell thunderstorms were registered in Bulgaria between April and June. A long-lived isolated supercell developed on 13 June 2024. This supercell persisted for over 7 hours, traveling more than 350 km and producing hail larger than 2 cm in diameter, which caused significant damage.

A thermodynamic analysis was conducted for estimation of the severity of the thunderstorm development. The obtained high values of CAPE, strong 0–6 km wind shear, and a typical curved hodograph shape indicated the potential of development of supercells.

To investigate the supercell’s evolution, radar data (updated every 4 minutes) and high-resolution visible (HRV) satellite imagery (at 5-minute intervals) were analyzed. Key radar parameters included maximum reflectivity (Zmax), the Vertically Integrated Liquid above the freezing level (dVIL), and VIL Density (VILD), defined as the ratio of VIL to the height of 15 dBZ contour.

Four cycles of intensification were distinguished during the mature severe stage of the supercell development. Most of the typical radar signatures of powerful convective cells, such as V-shaped radio echo, comma shape, Bounded Weak Echo Regions (BWER), Three-Body Scatter Spikes (TBSS), and Side lobe, were registered.

During the evolution of the supercell:
- V-shaped radar echo was registered several times, followed by a significant deviation in the storm motion of up to 60–65 degrees from the mean wind flow.
- There is a pronounced relationship between the storm's cyclic behavior and variations in dVIL and VILD.
- The intensification of the cycles was associated with the new formation of feeder cells and their subsequent merging with the supercell.
- Maximum radar reflectivity exceeded 65 dBZ for more than 2 hours, with peaks reaching above 70 dBZ during the severe stage of the supercell.

How to cite: Dimitrova, T., Kadiyska, N., Georgiev, S., and Pencheva, V.: Cyclic supercell Storm in Bulgaria observed on 13 June 2024: thermodynamic conditions, evolution and structure, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-98, https://doi.org/10.5194/ecss2025-98, 2025.

P40

ECSS2025-110

Orographic and lake effects on the evolution of thunderstorms in the region of Lake Balaton, Hungary

Máté Kurcsics, István Geresdi, and Ákos Horváth

Lake Balaton is the largest lake in Central Europe. The Hungarian Meteorological Service operates a lake storm warning system by the lake to ensure the safety of sailors, fishermen and the thousands of tourists who visit the area. Thunderstorms are the most dangerous phenomena following windstorms caused by cold fronts due to wind gusts and lightning. Thunderstorms have a frequent occurrence in the region from May to August, occurring every three days on average.

Observations indicate that the 600 km² lake can significantly influence the evolution of these thunderstorms. It affects instability conditions and the moisture content, as well as the location of convergent and divergent areas. In calmer weather situations, when synoptic-scale processes allow local effects to dominate, significant lake circulation develops. Convection is therefore inhibited over the lake during the daytime, while it is promoted along the convergence lines over the shores. Additionally, circulation above the lake generates wind shear and vortices, which can intensify the evolution of thunderstorms.

To the northwest of Lake Balaton lies the Bakony Mountains, whose highest peak is around 700 m high. The orographic lifting mechanism can also initiate the formation of daytime thunderstorms. These thunderstorms can affect the weather over the lake by generating gust fronts, lightning and showers. The lake circulation and the most frequent prevailing flow can further enhance orographic lifting. In the convective season these additional effects often result in the first daily formation of thunderstorms over the Bakony Mountains. If the direction of the main flow is northwest, the sinking of air on the lee side may weaken or even dissipate cumulonimbus clouds.

This study used radar data, lightning and surface measurements to examine the development of thunderstorms in the region of the lake. Statistical analyses of the data and case studies are also provided, examining the various local effects. The WRF (Weather Research and Forecasting) numerical model was applied to simulate the impact of these local effects on thunderstorm evolution.

How to cite: Kurcsics, M., Geresdi, I., and Horváth, Á.: Orographic and lake effects on the evolution of thunderstorms in the region of Lake Balaton, Hungary, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-110, https://doi.org/10.5194/ecss2025-110, 2025.

P41

ECSS2025-114

Evaluating the effect of mass continuity, smoothness, and resolution constraints on thunderstorm wind fields using Dual-Doppler 3DVAR wind retrieval

Massimiliano Burlando, Priya Kumari, Hanna Beatriz Wollmeister Muñoz, Renzo Bechini, Djordje Romanic, and Alessandro Battaglia

Windstorms driven by thunderstorms are among the most hazardous weather events, capable of causing significant damages. In this study, Doppler radar observations are used to analyse the internal wind structure of storm. Wind kinematic within storm systems are retrieved using three-dimensional technique based on dual-Doppler variational approach, which integrates data from C-band and X-band radar system. Sensitivity experiments were conducted by varying resolution and the weights of the cost function terms, which control the extent to which the model enforces the mass continuity equation and smoothness in the domain. The technique was applied to a thunderstorm event that occurred in the Piedmont region of Italy. The retrieved wind profiles were validated against available LiDAR observations from surface up to 2000 m in height. Results show the noticeable changes in updraft and downdraft structure depending on the cost function weights. These smoothness constraints help reduce noise and make the wind field look more realistic. The study highlights the importance of mass continuity term in producing realistic wind fields and the potential of retrieving accurate wind information from Doppler radar data. Additionally, the findings of this research contribute to a better understanding of thunderstorm dynamics and offer valuable insights for enhancing nowcasting and risk mitigation strategies for localized windstorms.

How to cite: Burlando, M., Kumari, P., Wollmeister Muñoz, H. B., Bechini, R., Romanic, D., and Battaglia, A.: Evaluating the effect of mass continuity, smoothness, and resolution constraints on thunderstorm wind fields using Dual-Doppler 3DVAR wind retrieval, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-114, https://doi.org/10.5194/ecss2025-114, 2025.

P42

ECSS2025-170

Doppler wind lidar measurements during a thunderstorm at Cabauw, The Netherlands

Tiemo Mathijssen and Steven Knoop

The Royal Netherlands Meteorological Institute (KNMI) operates several ground-based Doppler lidars, which are laser-based remote sensing instruments to measure wind. We deploy long-range scanning Doppler lidars at our atmospheric research site Cabauw (Vaisala Windcube200S) since 2021 and at Amsterdam Schiphol Airport (Leonardo Skiron3D) since 2025, after a two-year deployment at Cabauw. Together with Rijkswaterstaat (RWS) we operate a network of short-range Doppler lidars (ZX lidars ZX300M) on the North Sea, on TenneT substations within offshore wind farms. Our Doppler lidars are based on infrared laser light and rely on aerosol scattering. As such, the wind measurements are typically bounded to the atmospheric boundary layer and are limited by (optically dense) clouds. The short-range Doppler lidars provide wind profiles up to a height of 300m.

The Doppler lidar instruments allow for unattended and continuous 24/7 operation and data is collected in near-real time. This enables the use of this data to forecasters for nowcasting, for data assimilation into numerical weather prediction models and for research purposes. These instruments can operate during severe storms, in which they can provide information of the wind field below the cloud base. Furthermore, Doppler lidars can provide boundary layer wind profiles prior to such storms, which could lead to better forecasting of those storms.

Here we present an example of Doppler lidar measurements during a severe convective weather event of 9th of July 2024, a day at which KNMI issued an orange weather warning (“code oranje”) for the whole of the Netherlands. At the time both long-range scanning Doppler lidars were operational at Cabauw. The Skiron3D was performing Velocity Azimuth Display (VAD) scans providing vertical profiles of horizontal wind speed and direction with a range resolution of 25m and 100m respectively with an update frequency of approximately 3 minutes. The Windcube200s was performing continuous vertical velocity measurements with a temporal resolution of 1 s and a vertical resolution of 75m.

On radar imagery and lightning detection system is observed that the showers and lightning arrives at 17:10 UTC. Signs of approaching severe weather are already visible in the Doppler lidar data starting at 15:00 UTC, indicated by a growing layer with a differing wind direction. Strong updrafts are observed shortly before the arrival of the showers. Heavy rainfall obscured the lidar measurements only for a very short period, such that valuable information can be derived from the lidar measurements before, during and after the storm.

How to cite: Mathijssen, T. and Knoop, S.: Doppler wind lidar measurements during a thunderstorm at Cabauw, The Netherlands, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-170, https://doi.org/10.5194/ecss2025-170, 2025.

P43

ECSS2025-213

Use of polarimetric radar data for identification of processes in convective storms

David Rýva

Since 2015, the Czech weather radar network (CZRAD) has consisted of two dual-pol C-band weather radars. This allows us to use polarimetric radar data for many purposes, including better correction for attenuation, which can provide better precipitation estimation data, improved removal of ground clutter and WLAN interference, and enhanced detection of hailstorms. The use of polarimetric data should also allow us to identify updrafts in convective storm cells based on the detection of so-called ZDR-columns and, more recently, KDP-columns. These are column-like features with positive ZDR and KDP values that can reach altitudes several kilometers above the freezing level. The main reasons for this are the presence of liquid water droplets in the warmer updraft air and their ability to remain in a supercooled state for some time in rapidly rising air parcels. The main goals of this study are to verify the potential of polarimetric data for better nowcasting of storm core behavior based on the detection of ZDR-columns and KDP-columns, and to present the advantages and weaknesses of automated algorithms for identifying warm vs. cold rain processes in convective storms. Such knowledge could significantly help forecasters with severe weather warnings, especially in cases of severe hailstorms or torrential rains. In the present analysis, we focused on case studies of several selected situations that occurred during storm seasons from 2018 to 2024. We studied the ability of various polarimetric data to help predict the severe behavior of convective storms and their accompanying phenomena.

How to cite: Rýva, D.: Use of polarimetric radar data for identification of processes in convective storms, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-213, https://doi.org/10.5194/ecss2025-213, 2025.

P44

ECSS2025-214

Field test of mobile solid state X-band radar in central Europe

Filip Najman and Miloslav Staněk

During the summer and autumn of 2025, we conducted an extensive field testing campaign of a newly developed mobile solid-state X-band weather radar system by Meteopress, with the primary aim of evaluating its capabilities in a variety of Central European meteorological and geographic environments. The radar is designed for portability and flexibility, incorporating a 1.2-meter parabolic antenna and a low-power solid-state transmitter. The entire system is mounted on the bed of a standard pickup truck, enabling rapid deployment, minimal logistical requirements, autonomous operation, and adaptable scanning strategies. These features make it particularly well-suited for short-term field missions, mobile nowcasting support, and targeted atmospheric research campaigns.
The campaign focused on collecting high-resolution radar data during a wide range of weather scenarios, with a special emphasis on convective storms. Over the course of several months, the radar was deployed to numerous locations, primarily across Czechia, where we had a licence for using the radat in the whole country. Locations include both flat lowland areas and regions with more complex terrain such as hills and valleys. Deployments were guided by real-time weather forecasts and were timed to intercept developing thunderstorms. In addition to convective events, the radar also observed episodes of stratiform precipitation, providing opportunities to assess its sensitivity and performance in lower-intensity rainfall situations.

How to cite: Najman, F. and Staněk, M.: Field test of mobile solid state X-band radar in central Europe, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-214, https://doi.org/10.5194/ecss2025-214, 2025.

Discussion Post a comment

P45

ECSS2025-215

Monitoring of convective storms in Central Europe using C-band mobile solid-state weather radar

Miloslav Staněk and Filip Najman

In 2023, Meteopress company developed a mobile C-band weather radar, MASEC (Mobile Automatic Self-Erecting Containerised C-band radar). This solid-state weather radar stands out for its rapid deployment - fully operational in under 10 minutes - and its ability to be dismantled and relocated just as quickly using a standard truck, offering high mobility. Another key advantage is its low power consumption (around 750 W), allowing deployment in areas without access to the electrical grid. It runs on batteries, which can be recharged using photovoltaic panels. In its current setup, fully charged batteries enable up to four days of continuous operation, with the potential to double that by adjusting the scanning strategy. The radar’s standard operational range is approximately 300 km.

In summer 2023, MASEC was tested at the Dlouhé stráně in the Jeseníky Mountains, Czechia, where all aspects of its use in operational meteorology were evaluated. MASEC captured several significant weather phenomena, including supercells and the development of squall lines. Further tests followed in autumn on Schöckl Mountain near Salzburg, and in 2024, the radar was moved to Sazená Municipal Airport near Prague. There, various settings were tested, focusing on advanced scanning strategies, tuning data resolution, implementing new methods for signal processing and new computation of derived products from the radar data. These derived products included estimating horizontal wind shear, detecting mesocyclones, and identifying microbursts from high-resolution data. For monitoring convective storms, a specialized shear scanning strategy was adopted. It featured a Doppler radial velocity range of 125 km with staggered pulses and polarimetric products and reflectivity up to 150 km at 0.5° resolution. Operating throughout the summer season, MASEC successfully detected multiple intense supercells with large hail and bow echoes with embedded mesovortices.

The poster presents selected cases of severe supercells and squall lines analyzed using this specialized shear scanning strategy. It highlights insights from both polarimetric data and derived products from radar data, such as horizontal wind shear estimates, mesocyclone detection, and microburst identification. A comparison between the specialized and standard scanning strategies used in 2023 is also included, along with an exploration of how modified scanning can be used to monitor thermal turbulence near Prague.

How to cite: Staněk, M. and Najman, F.: Monitoring of convective storms in Central Europe using C-band mobile solid-state weather radar, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-215, https://doi.org/10.5194/ecss2025-215, 2025.

Discussion Post a comment

P46

ECSS2025-256

Lightning Activity and its Temporal Progression Using GOES/GLM sensor During the Catastrophic May 2024 Floods of Rio Grande do Sul

Leonardo Calvetti, Clarissa Tavares, Wagner Loch, and Cesar Beneti

The May 2024 flood event was the most significant in the history of Rio Grande do Sul and among the largest recorded in Brazil, resulting in 182 fatalities, 29 missing persons, and affecting over 2 million individuals. Considered a major global climatic event of recent times, this tragedy mobilized the entire nation of Brazil, significantly influenced news broadcasting schedules, and remained a prominent topic on social media platforms for an extended period. Extreme weather events, such as floods and severe storms, underscore the importance of studying atmospheric phenomena, including lightning. Understanding the dynamics of lightning is crucial for forecasting and mitigating the impacts of these natural events, thereby contributing to public safety and the development of effective prevention strategies. The results indicated a direct association between lightning density and 87% of the precipitation events, and a "lightning jump" was observed minutes prior to the occurrence of the most significant precipitation intensities. The convection was sustained by a persistent low-level moisture flow from the Atlantic Ocean, which was anomalously warm. Consequently, the convection was not as deep as that generated by the South American low-level jet, but it persisted for many days. The lightning activity was also persistent and highly correlated with the rainfall.

How to cite: Calvetti, L., Tavares, C., Loch, W., and Beneti, C.: Lightning Activity and its Temporal Progression Using GOES/GLM sensor During the Catastrophic May 2024 Floods of Rio Grande do Sul, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-256, https://doi.org/10.5194/ecss2025-256, 2025.

P47

ECSS2025-275

Polarimetric Radar Signatures of ZDR and KDP in Tornadic Storms over Germany and their Use for Nowcasting

Erik Brune, Silke Trömel, and Lisa Schielicke

Germany has one of the highest densities of tornado reports in Europe with several damaging tornadoes observed every year. Although the German Meteorological Service (DWD) has a dedicated warning process in place, it relies on voluntary observers on the ground to confirm tornadoes. Thus, a robust and reliable radar-based detection algorithm for tornadic cells would represent a great improvement of the warning process. Previous studies by Loeffler and Kumjian (2018, Weather&Forecasting, 33(5), 1143-1157) and Loeffler et al. (2020, GRL,47(12), e2020GL088242) showed pathways to distinguish between tornadic and non-tornadic supercells based on signatures of differential reflectivity (Z_DR) and specific differential phase (K_DP) in polarimetric radar observations. Storm relative winds cause size sorting in precipitation, leading to a higher concentration of larger drops at the forward flank of a supercell and a localized maximum of Z_DR. Smaller drops are advected further into the core of the convective cell, causing enhanced values of K_DP. In most storms, ZDR and KDP signatures are spatially separated along the storm motion, but in tornadic supercells, and in some tornadic non-supercells, this separation tends to be more perpendicular to the motion vector. This study uses measurements of DWD's polarimetric C-Band radar network to investigate the separation signature in tornadic storms over Germany and its potential for nowcasting. This data is available since 2021 with a radial resolution of 250 m. The analysis includes 16 tornado cases observed in 2021 and 2022, including supercell and non-supercell tornadoes. An clustering algorithm and percentile-defined thresholds are exploited to identify and analyze the separation signature. Results confirm the existence of the signature in both cell types, showing more consistent separations in the supercell cases showing a good potential for nowcasting with lead times of 5 − 20 min. For non-supercell tornadoes, however, the value of the signatures is limited and can at best confirm the occurrence of a tornado. Results also show differing track characteristics of clusters with enhanced Z_DR and K_DP for both cell types. In supercells, the clusters tend to deviate less around the linear direction of storm motion, which may be linked to the degree of organization of the storm. An extended and revised version of the algorithm is assumed to significantly improve the warning process for tornadoes in Germany.

How to cite: Brune, E., Trömel, S., and Schielicke, L.: Polarimetric Radar Signatures of ZDR and KDP in Tornadic Storms over Germany and their Use for Nowcasting, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-275, https://doi.org/10.5194/ecss2025-275, 2025.

P48

ECSS2025-279

Tracking and Characterization of Convective Storms Through Their Lifecycle in the Monsoon Core Zone Using Polarimetric Radar Observations

Ashruba Ghorapade, Sachin Deshpande, Manisha Tupsoundare, Zhe Feng, Subrata Das, and Harshad Hanmante

Deep convective clouds range from isolated storm to organized mesoscale systems and play a vital role in Earth's energy and water cycles by transporting heat, moisture, and momentum from the boundary layer to the upper troposphere. During the Indian monsoon season, deep moist convection significantly influences the spatial and temporal distribution of rainfall. Observing convective storms/cells and their evolution with time is key for quantifying their behaviour and its dependence on the large-scale environment. One of the challenges in studying convective storm properties is observing the quick evolution of individual storms. Prior studies of deep convection in India have mostly used satellite data to describe the storm structure. However, studies on convective modes, their lifecycle, microphysics, and link to large-scale are lacking, which motivates this work.

Against this background, we identified and tracked storms using high-resolution C-band polarimetric (CPol) radar data (6 min, 0.5 km) obtained at IITM's Atmospheric Research Testbed (ART) facility at Silkheda in the monsoon core zone. The storms are identified based on the horizontal radar reflectivity texture. This Lagrangian tracking resulted in a storm lifecycle of over 63,000 tracks from June-September 2023, providing unique information on individual storm initiation and growth, location, area, and depth. Results show that about 80% of storms have a short life span of < 1 hr. and cover a small area of < 50 km², while 16% of storms survive up to 2 hrs. and acquire larger areas up to 100 km². During monsoon onset (June) and active (July) periods, storms are wider, more intense, and deeper compared to August and the withdrawal month (September). Diurnal cycle shows that between the noon and early evening hours, storms gradually deepen, intensify, and expand in area. Upon matching the radar tracked storms with large-scale data, it is seen that low-level humidity and CAPE have a major impact on the evolution of convective storm properties. Furthermore, using polarimetric variables, this work investigates the internal vertical microphysical structure of tracked convective cells. The high reflectivity values and greater specific differential phase up to 7 km at the cell initiation provide further evidence of the favorable moist low-level conditions. The radar derived statistics of convective storm lifecycles and interaction between storms described here have the potential to increase our understanding of factors controlling convective evolution and their representation in models.

How to cite: Ghorapade, A., Deshpande, S., Tupsoundare, M., Feng, Z., Das, S., and Hanmante, H.: Tracking and Characterization of Convective Storms Through Their Lifecycle in the Monsoon Core Zone Using Polarimetric Radar Observations, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-279, https://doi.org/10.5194/ecss2025-279, 2025.

P49

ECSS2025-287

Optimizing C-Band radar settings for mesocyclone and tornado detection

Bram van 't Veen and Pieter Groenemeijer

The detection of mesocyclones and tornadoes by radar heavily relies on the accurate measurement of Doppler wind speeds. The associated circulations are characterized by strong tangential gradients of the radial wind. With C-band radar, the type of radar most commonly used in Europe, such measurements are challenging. The main difficulty stems from the fact that the returned signal is not associated with one unique velocity v, but with a velocity v = v +/- N v_N whereby v_N is known as the Nyquist velocity. For C-band radars v_N is often between 5 and 10 m/s as a result of the other chosen settings. Dual- or triple-pulse repetition frequency (PRF) is a technique to mitigate this problem and effectively increase v_N substantially and to around or above 30 m/s, however usually introducing errors whereby velocities are offset from their real values by a multiple of 2N. In response, corrections and filters are applied, however any filtering must be tuned not to throw out genuine signal. Furthermore, wind speed in strong circulations may still exceed the extended Nyquist velocity of 30 m/s.

Multiple-PRF and other techniques and settings are used differently across Europe. Using ESSL’s radar viewer we have investigated how difficult it is to detect tornadoes and mesocyclones using those settings. In some countries no multiple PRF technique is used, rendering the detection of mesocyclones next to impossible. In others, excessive filtering of the extreme winds in mesocyclones occurs to varying degrees, hindering their confident detection. In another country, the effective beam width and gate length reduce the ability to detect tight circulations. In other countries, scans below 1.0° are missing, or a staggered scanning pattern of different elevations complicates the detection. The settings for detecting circulations appear to be best in Udine (FVG), Finland, and Germany. We present advice on how to optimize C-band radar settings for the detection of strong, tight circulations.

How to cite: van 't Veen, B. and Groenemeijer, P.: Optimizing C-Band radar settings for mesocyclone and tornado detection, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-287, https://doi.org/10.5194/ecss2025-287, 2025.

Discussion Post a comment

P50

ECSS2025-289

Detecting severe storms using an RGB composite combining polarimetric radar parameters

Bram van 't Veen, Pieter Groenemeijer, and Tomas Pucik

Besides reflectivity (Z) and radial velocity (V), polarimetric doppler radar provides much more information that is not often used by forecasters in real-time, at least in Europe. These data include among others the correlation coefficient (CC or RhoHV) and differential reflectivity (ZDR). By combining those data using the red, green and blue (RGB) color channels of an image these parameters can be combined in a visually pleasing way that contains a large amount of information. This idea was showcased by Trevor White (University of Alabama at Huntsville) who shared experimental images on social media. We think that this may be a way to make these polarimetric data more accessible to forecasters.

At the ESSL Testbed 2025 we tested the composite product with forecasters. The composite that we provided to them has reflectivity (Z) increasing from 30 to 60 dBZ mapped onto the red channel, correlation coefficient decreasing from 1.0 to 0.7 onto the green channel, and differential reflectivity (ZDR) increasing from 0 to 8 onto the blue channel. The alpha (opacity) channel is used to emphasize higher reflectivity and reduces to fully transparent as reflectivity decreases from 43 to -10 dBZ in a piecewise linear way.

The resulting RGB shows heavy rainfall as purple to magenta, hail as yellow or white, tornadic debris as green or yellow. Investigating the colors at various elevations can reveal details about the storm structure and microphysics, such as the vertical extent of hail in a storm. Although a storm can be studied using the individual parameters Z, ZDR, and CC, the RGB makes it easier. The method has some limitations, such as the fact that ground clutter can look similar to hail or tornadic debris, which can complicate the interpretation of the RGB.

We will provide several examples and summarize the feedback from forecasters.

How to cite: van 't Veen, B., Groenemeijer, P., and Pucik, T.: Detecting severe storms using an RGB composite combining polarimetric radar parameters, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-289, https://doi.org/10.5194/ecss2025-289, 2025.

Discussion Post a comment

P51

ECSS2025-305

Dual-Polarization Radar Insights into Regional Cloud Seeding Program over Central Saudi Arabia: Spring 2025 Observations

(withdrawn)

Ioannis Matsangouras, Stavros-Andreas Logothetis, and Ayman Mohmmed Albar

Posters TH4: Thu, 20 Nov, 14:30–16:00 | Poster area

Display time: Wed, 19 Nov, 09:00–Thu, 20 Nov, 18:30

ECSS2025-37

Classifying convective surface wind gusts from polarimetric radar data

Florian Ackermann, Monika Feldmann, Daniele Nerini, Martin Aregger, Josué Gehring, Simone Balmelli, and Olivia Martius

Thunderstorm cells generating severe straight-line winds pose significant risks to infrastructure in Switzerland, where current warning methods lack specific detection capabilities for these events. Five high-impact thunderstorm cases (2019–2023) were analyzed to identify indicative radar signatures, revealing that descending reflectivity cores, specific differential phase (KDP) cores, and midlevel radial convergence were the most consistent radar-based indicators of downbursts. A climatology of gust-producing cells was established by integrating 5-minute thunderstorm tracking data with weather station data, crowd-sourced reports, and European severe weather database records from May–October 2019–2023. This dataset was combined with composite radar metrics to train a Random Forest model for a binary classification into gust-positive and gust-negative cells. Input features included thunderstorm tracking attributes and the identified radar features, supported by the case studies. Lightning and convergent radial shear emerged as top predictors in analyses of feature importance. The model achieved a moderately skillful probability of detection and a very low false alarm ratio, while maintaining a good critical success index, highlighting that the model struggled to classify some gust-positive cells correctly, but rarely misclassified a gust-negative cell. This approach demonstrates the potential of improving nowcasting systems in Switzerland by applying machine learning on radar data to advance severe weather detection and warnings.

How to cite: Ackermann, F., Feldmann, M., Nerini, D., Aregger, M., Gehring, J., Balmelli, S., and Martius, O.: Classifying convective surface wind gusts from polarimetric radar data, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-37, https://doi.org/10.5194/ecss2025-37, 2025.