Severe hailstorms are becoming more frequent in Central Europe, showing increasing interannual variability. The Pre-Alpine and Alpine region is significantly affected due to its complex terrain, which initiates convection and can intensify many hail-favoring processes. In particular, large hail events are often very local phenomena and are becoming increasingly intense. Ground-based observations from weather radars are the most reliable for detecting hail; however, they are challenging in the Alpine region due to interference at mountain ranges.

Passive Microwave satellite observations offer a valuable alternative for detecting hail: a hail probability can be directly derived from Passive Microwave channels with a high spatial coverage. However, this data is only available at certain times during satellite overpasses, thus capturing only snapshots of a few of these events. Visible, near-infrared, and infrared data from MSG give the highest spatial and temporal coverage. Though not directly sensitive to hail, its high spatiotemporal resolution can identify early stages of severe storm development leading to large hail formation by peculiar characteristics in their spatiotemporal evolution.

Whereas various ML approaches already classify spatial cloud patterns from satellite measurements, the temporal component of cloud development remains less explored. This work aims at classifying the evolution of typical cloud patterns leading to severe storms over the Alpine region. In particular, learning about formation of hail storms and the conditions in which they develop.

We initially adopted a supervised deep-learning framework for classifying spatiotemporal cloud evolution patterns as hail and non-hail, obtaining better performance than a simple logistic regression. We trained the network using MSG timeseries displaying cloud evolution patterns and labeled the presence of hail using a hail probability product based on passive microwave radiometry. Then, we exploited the same architecture in a self-supervised approach, using the labeled dataset as fine-tuning to capture hail development characteristics better. We characterize such classes of cloud developments in a spatiotemporal embedding, exploiting its characteristics and investigate the physical properties of the classes with ancillary datasets.

How to cite: Bigalke, P., Acquistapace, C., Corradini, D., and Laviola, S.: Learning about (hail-) storm development exploiting supervised and self-supervised deep learning and MSG imagery, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-29, https://doi.org/10.5194/ecss2025-29, 2025.

We present a few case studies demonstrating the application of a neural network (NN) model for detecting overshooting tops (OTs) using high-resolution visible (HRV) channel from the SEVIRI instrument on MSG satellites. The model was trained on manually labeled data using OT shadows (database created by Ján Kaňák) and the product of this model provides per-pixel probabilities of OT presence.

To assess the model's relevance for severe weather forecasting, we compared the detected OTs with reports from the European Severe Weather Database (ESWD). The comparison showed varying levels of agreement - several OTs corresponded well with hail or another type of events, while in other cases, strong convection was detected without reported impacts, and vice versa. This may be due to multiple factors on both sides - limitations in our model’s predictions as well as in the event database (e.g., missed reports).

This variability highlights both the usefulness and the limitations of OT detection as a proxy for severe weather. The model performs well in identifying deep convective features but should be interpreted alongside other data sources for operational use.

Our results suggest that ML-based OT detection from HRV imagery can contribute to nowcasting applications, especially when integrated with additional observational and model data.

How to cite: Doležalová, A., Seidl, J., and Šťástka, J.: Application of Machine Learning to Severe Weather Prediction from Storm Top Indicators, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-88, https://doi.org/10.5194/ecss2025-88, 2025.

Climate change is intensifying and increasing the frequency of storms in the Alpine region. However, accurately capturing these events remains a challenge for current weather models due to the complexity of atmospheric processes over mountainous terrain. Realistic rainfall predictions require precise cloud representation, especially for deep convective systems that cause extreme precipitation.

We present a framework for classifying cloud structures observed by the Meteosat Second Generation (MSG) geostationary satellite. Leveraging its good spatio-temporal resolution and extensive historical data, our approach exploits brightness temperature in the 10.8 µm infrared channel alone and in combination with the 6.2 µm water vapor channel. This synergy enables a better representation of diurnal cycles and cloud top heights. Our classification framework employs a self-supervised deep learning (DL) model to generate a feature space where cloud structures group together based on their semantic similarity.

We characterize the identified cloud classes using a range of physical parameters, including cloud properties, precipitation amounts, lightning activity, and morphological indices. Additionally, their diurnal and seasonal variability are analyzed to determine whether some cloud types are most likely to occur. Once the classes are physically described, cloud development are tracked in the feature space associated with extreme convective rainfall and hailstorms in the Alps, as recorded in the European Severe Storms Laboratory (ESSL) database. We study the transition of convective systems to extreme precipitation across space and assess the associated environmental conditions.

Ultimately, this framework can enhance the evaluation of numerical weather prediction models by analyzing how simulated cloud evolution aligns with observed transitions in extreme events. Furthermore, it can be used to improve nowcasting and early warning systems for extreme precipitation by leveraging observation-based transition probabilities derived from past severe weather events.

How to cite: Corradini, D., Acquistapace, C., Bigalke, P., and Cattani, E.: Self-supervised cloud classification using satellite infrared imagery to characterize extreme precipitation events over the Alps , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-97, https://doi.org/10.5194/ecss2025-97, 2025.

The paper presents the results of validation of RDT SAF product based on MSG data with ENTLN data over Eastern Europe.  Ukrainian Total Lightning Network (UTLN), as a part of Earth Networks, installed in Ukraine is capable to detects the components of both intra-cloud (IC) and cloud-to-ground (CG) flashes with a high efficiency and very precise spatial detection (200 m) and covered area over Eastern Europe. Total lightning (IC and CG) provides significantly better identification of storm severity. The components of total lightning detection consist from lightning flashes and pulses, lightning cell (cluster of flashes), trajectory of lightning cell tracking, lightning flash rate (flashes/min) and a warning area ahead of the storm cell, based on combination of information from the cells, such as the moving speed and direction and size of the cell. The RDT (Rapid Development Thunderstorm) product is an object-oriented diagnostic for convective clouds or cells. RDT is based on MSG data and demonstrates tracks of clouds, identifies those that are convective (discrimination), and provides some descriptive attributes for their dynamics. The detection algorithm allows to define “cells” which represent the cloud systems. In the RDT algorithm, “cells” are defined on infrared images (channel IR10.8) by applying a threshold which is specific to each cloud system, depending on local brightness temperature pattern. We download RDT product using API of Eumetview(https://view.eumetsat.int/productviewer?v=default).

Validation process includes joint  analysis of 15 min RDT SAF, MSG RGB’s images  and ENTLN data with different levels of severity from 01.04.2023 to 31.05.2025. We use a specific software developed for visualization of RDT, NWP, RGB and lighting data to detect severe weather.  The analysis shows a good agreement between RDT and ENTLN data, especially from June to August, when the convective processes over Eastern Europe have maximum intensity and does not depend of geographical locations of severe weather objects. Some specific cases were analyzed in details.        

How to cite: Kryvobok, O., Kryvoshein, O., and Zabolotna, O.: Validation of RDT product based on MSG data with Ukrainian Total Lightning Network (UTLN) data, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-143, https://doi.org/10.5194/ecss2025-143, 2025.

Large hail events produced by severe convective storms (SCSs) have emerged as a critical concern for the insurance sector, driven by a significant rise in insured losses from severe hail events across densely populated regions of Europe in recent years. Further complicating the relationship between losses and hail events is shifts to the underlying probability of SCSs occurrence and severity due to anthropogenic climate change. Record-breaking hail events, such as those observed in France and Italy in recent years, underscore the evolving risk that hail poses currently in Europe and may further pose in the future under changing climatic conditions. 

As part of an overarching goal of improving probabilistic models for convective hazard occurrence (especially across data-sparse regions), in the present work, we seek to better understand the relationship between remotely-sensed convective signatures, like overshooting tops (OTs), with large hail occurrences in SCSs. Specifically, OTs retrieved from satellite data provide a uniquely consistent view, spanning multiple decades, of the most intense convective updrafts across Europe. By employing a recently developed 21-year dataset of convective storms and overshooting tops over Europe derived from MSG SEVIRI infrared imagery (developed by NASA), we will compare OT occurrence and intensity across Europe with ground-based severe hail reports from the European Severe Weather Database (ESWD). OT events are first assigned into clusters based on spatiotemporal constraints. The frequency of these clusters will then be compared to modeled, multi-decadal trends in (very) large hail from the Additive Regression Convective Hazard Models (AR-CHaMo) in order to gain a first-order understanding of the potential predictive skill of OTs for large hail. We will also provide preliminary results of using OTs as an input predictor of hail occurrence in the AR-CHaMo models of hail risk across Europe. Finally, near-storm environmental characteristics, derived from ERA5 reanalysis, will be used to compare attributes of the thermodynamic and kinematic vertical profiles that may, or may not, differentiate between OT clusters, including a cluster’s orientation, length, width, and the severity of hail (or the lack thereof) assigned to each cluster.

How to cite: Coffer, B., Battaglioli, F., Groenemeijer, P., Bedka, K., Itterly, K., and Cooney, J.: Towards using overshooting tops in improving probabilistic risk models of (very) large hail across Europe, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-266, https://doi.org/10.5194/ecss2025-266, 2025.

The planetary boundary layer (PBL) contains the majority of moisture and links solar surface heating to increased convection  through vertical redistribution of heat and moisture. Such changes can influence the initiation and life cycle of convective clouds. Remote sensing of these parameters could be beneficial to short-range forecasting, nowcasting and process studies.

In a previous study we analysed the sensitivity of various channels to changes in boundary layer height (BLH) and moisture (BLM). Clear-sky, day-time observations in the near-infrared (NIR) at 0.9µm and thermal infrared (TIR) at 12 µm are primarily sensitive to total column water vapour (TCWV) and in the TIR to the skin temperature and emissivity. But they also exhibit distinct sensitivities to changes to BLH and BLM, respectively. The Flexible Combined Imager (FCI) onboard Meteosat-12 carries both these channels.

In an effort to exploit these sensitivities we build a look-up-table which ties BLH and BLM to the split window difference (SWD, brightness temperature difference between 11 and 12 µm) and the water vapour optical depth (calculated from the ratio of 0.865 and 0.914 µm). With an optimal-estimation inversion scheme, BLH and BLM may be retrieved at 10 minute intervals for clear-sky, day-time pixels.

Currently our effort lies on integrating these complementary NIR and TIR measurements to improve retrievals of TCWV and to derive additional information on the vertical moisture structure in the lower levels of the atmosphere.

How to cite: El Kassar, J. R., Carbajal Henken, C., Preusker, R., and Fischer, J.: Clear-sky Planetary Boundary Layer Characterisation Using Near- and Thermal-infrared Observations from Satellite-based Imagers, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-277, https://doi.org/10.5194/ecss2025-277, 2025.

The Flexible Combined Imager (FCI) on the MTG-I satellite includes a near-infrared channel centered at a wavelength of 0.91 µm, where radiation is partially absorbed by atmospheric water vapor. During daytime, in cloud-free regions, the signal detected from the Earth is weakened due to this absorption: first along the path from the sun to the Earth's surface, and then again from the surface to the satellite. By comparing this channel with a nearby one at 0.85 µm, it is possible to accurately estimate the total water vapor content. This method was notably advanced by Hans-Peter Roesli, who pioneered the use of the ratio between these two channels for such analysis.

At ESSL, we display the logarithm of this channel ratio, corrected for sun angle and satellite viewing angle, using an intuitive color map that ranges from bright yellow through green to blue and violet, representing increasing levels of water vapor. This visualization technique allows for the identification of mesoscale features such as sea-breeze fronts and drylines, which mark sharp gradients in low-level moisture that may otherwise go undetected. Furthermore, it enables the tracking of mid-tropospheric structures, such as pockets or filaments of dry or moist air.

These observed mesoscale features can provide valuable insights for forecasters, highlighting processes that may influence the development of convective storms. They also offer a way to evaluate the accuracy of Numerical Weather Prediction (NWP) models. However, a key limitation lies in the difficulty of determining the precise altitude at which moisture variations occur.

A promising future approach involves combining the differential total column water vapor data with information from hyperspectral sounders, such as IASI and especially the Infrared Sounder (IRS) on the MTG platform. Each method compensates for the other's limitations: while the differential water vapor signal does not specify the altitude of the moisture, the IRS can provide vertical resolution, although it is less accurate near the surface. This uncertainty near the ground can be reduced using total column water vapor estimates derived from transmittance measurements.

How to cite: Groenemeijer, P., Pucik, T., and Holzer, A. M.: Visualizing tropospheric humidity using differential water vapor transmittance between the 0.91 and 0.85 µm FCI channels, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-294, https://doi.org/10.5194/ecss2025-294, 2025.