Session 8 | Nowcasting and forecasting of severe weather and forecaster training

Session 8

Nowcasting and forecasting of severe weather and forecaster training

Orals MO7

| Mon, 17 Nov, 17:45–18:15 (CET)|Room Hertz Zaal

Orals WE2

| Wed, 19 Nov, 11:30–13:00 (CET)|Room Hertz Zaal

Orals TH1

| Thu, 20 Nov, 09:00–10:45 (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, P52–73

Posters TH4

| Attendance Thu, 20 Nov, 14:30–16:00 (CET) | Display Wed, 19 Nov, 09:00–Thu, 20 Nov, 18:30|Poster area, P52–73

Orals MO7: Mon, 17 Nov, 17:45–18:15 | Room Hertz Zaal

17:45–18:00

ECSS2025-151

Status and perspectives of SINFONY – the seamless combination of Nowcasting and NWP at DWD

Ulrich Blahak and the Team SINFONY & Friends

DWD's new Seamless Integrated Forecasting system (SINFONY) is entering a new phase.

Up to now the focus was initial development until "operational fitness", to improve very-short-range forecasting of intense convective events from observation time up to 12h, with radar at the heart of it. Having DWD's warning process (forecasters, automated systems) as well as German flood forecasting authorities in mind, seamless ensemble products in observation space (radar reflectivity composites, precipitation fields and convective cell objects) were developed, as well as informations on the probability of hazards like heavy precipitation, hail, wind gusts and lightning.

In the last eight years we developed
1) Radar Nowcasting ensembles for areal precipitation, reflectivity (STEPS-DWD) and convective cell objects including hail and life cycle information (KONRAD3D-EPS) with good forecast quality up to 1-2 hours.
2) A new regional NWP ICON-ensemble model (ICON-RUC-EPS) with LETKF assimilation of 3D radar volumes, cell objects, Meteosat VIS and IR channels and hourly new forecasts on the km-scale, whose forecast quality exceeds Nowcasting after about 1h. Advanced model physics (2-moment microphysics with prognostic hail) lead to an improved forecast of convective clouds and to more consistent model equivalents for radar data and visible/infrared satellite data based on detailed forward operators (EMVORADO and RTTOV-MFASIS).
3) An optimal combinations ("blending") of Nowcasting and NWP ensemble forecasts in observation space, which constitute the seamless forecasts of the SINFONY. Gridded combined precipitation and reflectivity ensembles (INTENSE) are targeted towards hydrologic warnings. Combined Nowcasting- and NWP cell object ensembles (KONRAD3D-SINFONY) help evolve DWD’s warning process for convective hazards towards flexible “warn-on-objects". Reflectivity composites and 3D cell objects are computed with the same software as for observations and Nowcasting.
4) Common Nowcasting and NWP verification systems for precipitation, reflectivity and cell objects help to continuously improve the SINFONY components.

Now we will expand in two directions.

First, the existing prototypes will be installed and maintained operationally on a long-term basis. We already got the ICON-RUC-EPS into operations in July 2024.

On the other hand we will further improve the systems in a new project phase 2025-2028 towards seamless forecasts for other parameters/phenomena and other user groups:
- Temperature, wind, cloudiness, fog, solar radiation, visibility, ceiling for
- aeronautical forecasts, renewables, customer data portals,
- with good year-round performance,
- and seamless forecasts beyond 12h.

For this, we try to improve the ICON-RUC forecasts for these parameters, we blend ICON-RUC into ICON-D2 and we integrate satellite data into our Nowcasting and combined products.

How to cite: Blahak, U. and the Team SINFONY & Friends: Status and perspectives of SINFONY – the seamless combination of Nowcasting and NWP at DWD, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-151, https://doi.org/10.5194/ecss2025-151, 2025.

18:00–18:15

ECSS2025-105

Combining Nowcasting and NWP for Improved Thunderstorm Prediction with KONRAD3D-SINFONY

Nora Linn Strotjohann, Lukas Josipovic, and Ulrich Blahak

KONRAD3D-SINFONY (K3DS) aims to provide the most accurate prediction of convective events by combining cells detected in radar observations with predictions from the ICON Rapid Update Cycle (ICON-RUC) model. To achieve this, we move model cells spatially and temporally and modify them to match observed cells better and we add nowcasting cells to the ensemble. For increasing lead times, the impact of observation and nowcasting diminishes and after about two hours the prediction becomes purely model based.

K3DS depends on many underlying systems such as the radar data, the cell detection algorithm, nowcasting, the model, and simulated radar sweeps. To improve K3DS systematically, we set up a framework to recalculate and evaluate days with strong convection in summer 2024. This allows us to measure the performance of K3DS and test which modifications yield more realistic predictions. Our tests can, hence, reveal disprepancies between predicted and observed convective cells and can provide hints how these underlying systems can be improved.

A current focus of our work is the visualization and interpretation of the K3DS results. As part of this effort, we transform the K3DS cell ensemble into a probabilistic forecast, which we aim to optimize. In the future, this probabilistic prediction can serve as the basis for issuing automated thunderstorm potential warnings with lead times of up to six hours. The automation of these warnings is currently under development within the framework of the DWD RainBoW project.

How to cite: Strotjohann, N. L., Josipovic, L., and Blahak, U.: Combining Nowcasting and NWP for Improved Thunderstorm Prediction with KONRAD3D-SINFONY, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-105, https://doi.org/10.5194/ecss2025-105, 2025.

Orals WE2: Wed, 19 Nov, 11:30–13:00 | Room Hertz Zaal

Chuck Doswell Memorial Session

11:30–12:00

ECSS2025-19

keynote presentation

The Reserch Labs Component of the National Weather Service Flash Flood Forecasters Training Course (~1978-1997): A Memorial to Chuck Doswell

Harold Brooks and Charles Doswell

In the mid-1970s, 3 flash floods that resulted in more than 400 deaths in the US led to the creation of a Flash Flood Forecasting Course. Two days of the two week course were led by researchers from the predcessor of NOAA's Office of Atmospheric Research, initially developed by Bob Maddox, Charlie Chappell, and Ray Hoxit. Soon, other scientists were added to the teaching and development group, notably Chuck Doswell. While the rest of the two week course focused on rules of thumb and the mechaniscs of issuing warnings in the system, the lab portion focused on scientific understanding and the dundamnetal idea that forecasting is hard.

In the early 1990s, Doswell modernized much of the course, dragging Harold Brooks into the teaching duties, until the in-person course was eliminated several years later and replaced by a distance learning module .The course led to the 1995 papter, "Flash Flood Forecating: An Ingredients-Based Approach." In January 2025, Chuck passed away. Since then, the 35 mm slides that covered the two days have been recovered. I'll use those slides to provide a flavor for the course and how it affected forecasting and research. It is anticipated that audience participation will be part of the presentation.

How to cite: Brooks, H. and Doswell, C.: The Reserch Labs Component of the National Weather Service Flash Flood Forecasters Training Course (~1978-1997): A Memorial to Chuck Doswell, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-19, https://doi.org/10.5194/ecss2025-19, 2025.

12:00–12:15

ECSS2025-227

The Summer 2024 Met Office Convection Nowcasting Testbed

Matthew Clark, Dan Suri, Brian Golding, Katie Norman, and Rosie Nation

The testbed concept, where researchers, developers, and users work together intensively for several hours at a time on a series of forecasting and verification exercises, is a proven concept for accelerating the development of new tools and forecasting techniques. At the Met Office, testbeds are one of the recommended platforms by which to evaluate decision aids and models, forming a critical part of Science to Services processes, and they have been held annually since 2020.

In summer 2024, a testbed was held at the Met Office to showcase two new observations-based nowcasting tools: the PLUVIA mesoanalysis, which blends observations from a variety of sources and model data to produce gridded fields of near-surface variables, and the PLUVIA cell tracker, which uses volumetric radar data to identify and track individual convective cells. In addition to increasing Operational Meteorologist familiarity with these new tools, the testbed aimed to assess the extent to which these tools provided added value for the nowcasting of convection and associated hazards in the UK.

The testbed included two specific forecasting tasks. Firstly, identification of mesoscale regions within the UK deemed to be at heightened risk of a given convective hazard (usually, 1-hour rainfall totals exceeding a threshold), as identified by a Lead Operational Meteorologist at the start of each testbed day. Teams were asked to define and then refine their delineated risk areas at intervals specified by the Lead Operational Meteorologist. This task was run as a forecast denial experiment, in which access to the abovementioned nowcasting tools was restricted for some groups, and the results compared for groups with and without access to these tools. Secondly, teams were asked to select a location under threat from an existing convective cell, and to predict the 1-hour rainfall accumulation at their selected location, using all available information including that provided by the nowcasting tools.

In this presentation, we will give an overview of the experiments and outline the key results, including details of how the new nowcasting tools were found to add benefit. We will also offer some lessons learned following our experiences of running a testbed of this nature.

How to cite: Clark, M., Suri, D., Golding, B., Norman, K., and Nation, R.: The Summer 2024 Met Office Convection Nowcasting Testbed, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-227, https://doi.org/10.5194/ecss2025-227, 2025.

Discussion Post a comment

12:15–12:30

ECSS2025-31

AR-CHaMo probabilistic forecasts of convective hazards with ECMWF ensemble

Ivan Tsonevsky, Pieter Groenemeijer, Francesco Battaglioli, Tomàš Púčik, Andreea Barascu, and Mateusz Taszarek

Convective hazards pose a significant threat to society and are among the most challenging phenomena for forecasting. The increasing spatial and temporal resolution of (European Centre for Medium-range Weather Forecasts) ECMWF Forecasting system makes it increasingly more suitable for applications targeting severe convection. The European Severe Storms Laboratory (ESSL) has developed additive logistic regression models (AR-CHaMo) for hail and severe convective wind gusts using the ERA5 re-analysis, observations of lightning and severe convection from around the world. ECMWF and ESSL collaborate in a project to apply AR-CHaMo to the ECMWF ensemble (ENS) to provide global predictions of convective hazards in the medium range. Probabilistic products for large hail and severe wind gusts from ENS are undergoing thorough testing, evaluation and skill assessment. Latest results suggest that AR-CHaMo forecasts on ECMWF ENS provide skilful predictions of large hail and severe wind gust risk many days in advance. ESSL and ECMWF continue to refine and develop these probabilistic products by testing more predictors, extending training datasets and including more convective hazards such as tornadoes. These products are already available via the ESSL’s Weather Displayer and are undergoing testing during the regular annual Testbeds. Evaluation of AR-CHaMo forecasts has also been performed internally at ECMWF on a number of severe convection events. The use of the ECMWF’s Integrated Forecasting System (IFS) model levels is capable of providing more accurate predictors and hence better probabilistic AR-CHaMo forecast.

How to cite: Tsonevsky, I., Groenemeijer, P., Battaglioli, F., Púčik, T., Barascu, A., and Taszarek, M.: AR-CHaMo probabilistic forecasts of convective hazards with ECMWF ensemble, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-31, https://doi.org/10.5194/ecss2025-31, 2025.

12:30–12:45

ECSS2025-3

Europe's Elevated Mixed Layer: New Insights into the Spanish Plume

David M. Schultz, Daniel J. Kirshbaum, and Martin V. Young

The canonical Spanish plume is a synoptic pattern associated with deep moist convective storms over western continental Europe and the UK. A large-amplitude trough or cut-off low in the jet stream extending to low latitudes produces a long fetch of southerly or southwesterly deep tropospheric flow. The near-surface air passes across the heated elevated terrain of the Iberian Peninsula and travels into western Europe. Thus, the preconvective environment is characterized by an elevated mixed layer of hot dry air with steep lapse rates (i.e., the Spanish plume airstream) overtop a warm surface layer and capping inversion, resembling the loaded-gun convective sounding.

A review of 102 journal articles mentioning the Spanish plume paints an unevidenced, inconsistent, unclear, and inaccurate picture. Some articles correctly employ the original definition of the Spanish plume airstream as the elevated mixed layer; others incorrectly apply the term to the surface (sometimes humid) airstream. Confusion extends to the origin of the airstream, which has been variously described as the Iberian Peninsula, northern Africa, or both, often unevidenced. Some air in so-called Spanish plumes does not even cross Spain.

We examine a Spanish plume case from 1–2 July 2015. We calculate air-parcel trajectories to determine four airstreams responsible for the unstable thermodynamic profile over the UK: (1) a near-surface continental airstream transporting hot, moist boundary-layer air from France; (2) a lid (850-hPa) airstream descended from the mid-troposphere over the eastern North Atlantic and western Iberian Peninsula, initially warm and dry, but gradually cooling and moistening and rising while traveling northward, forming a layer of convective inhibition; (3) a subtropical airstream of hot, dry air with steep lapse rates traveling poleward from North Africa and the Mediterranean then ascending; and (4) a middle-latitude upper-level trough airstream traveling eastward from the North Atlantic Ocean then ascending.

These results challenge the canonical Spanish plume synoptic pattern in three ways. First, most air reaching the UK does not travel over the Iberian Peninsula, particularly the near-surface air from France. Second, the steep lapse rates were pre-existing from the subtropics rather than created when passing over the Iberian Peninsula. Third, the lid results from subsidence rather than surface heating over the Iberian Peninsula. Thus, the synoptic-scale pattern appears to have a larger control over the thermodynamics of the Spanish plume airstream than heating from the Iberian Plateau, suggesting that the canonical conceptual model for the Spanish plume requires revision.

How to cite: Schultz, D. M., Kirshbaum, D. J., and Young, M. V.: Europe's Elevated Mixed Layer: New Insights into the Spanish Plume, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-3, https://doi.org/10.5194/ecss2025-3, 2025.

12:45–13:00

ECSS2025-299

The ESSL: Europe’s Centre of Competence on severe convective storms approaches its 20 year anniversary

Pieter Groenemeijer and Alois M. Holzer and the ESSL employees and Executive Board

When, before its founding in 2006, the European Severe Storms Laboratory (ESSL) was conceived by its initiator, Dr. Nikolai Dotzek, he had the vision of it becoming a leading “Center of Competence” on convective storms in Europe. Since its inception, the Laboratory has gradually expanded and now employs 15 staff members, including part-time personnel.

ESSL’s main activities include the development and maintenance of the European Severe Weather Database, conducting comprehensive research on severe weather phenomena, including the assessment of new forecasting tools, organizing the European Conferences on Severe Storms, and providing training for weather forecasters. These efforts are made possible through the support of institutional and individual members, funding from international, national and regional agencies, and the valuable contributions of volunteers who report severe weather events to ESSL.

Over time, ESSL has also built strong, long-term collaborations with organizations such as ECMWF, EUMETSAT, and various national weather services. Since 2024, the two legal entities in Germany and Austria responsible for ESSL’s operations, together with its employees, have co-owned a commercial venture that facilitates the use of ESSL’s hazard models and the popular online Weather Data Displayer by members and others.

Some of ESSL’s other current developments include the organization of the field campaign on Thunderstorm Intensification from Mountains to plains (TIM) and leveraging several new avenues in research and education enabled by the classification of ever more European meteorological data as “Open data”.
In our presentation, we will reflect on ESSL’s achievements over the years, highlight key milestones and events, anticipate the celebration of its 20th anniversary in 2026, and share a forward-looking perspective on the Laboratory’s future development.

How to cite: Groenemeijer, P. and Holzer, A. M. and the ESSL employees and Executive Board: The ESSL: Europe’s Centre of Competence on severe convective storms approaches its 20 year anniversary, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-299, https://doi.org/10.5194/ecss2025-299, 2025.

Discussion Post a comment

Orals TH1: Thu, 20 Nov, 09:00–10:45 | Room Hertz Zaal

09:00–09:30

ECSS2025-225

keynote presentation

Recent work at Météo-France’s Nowcasting department on warnings of exceptional rainfall and thunderstorm objects

Thibaut Montmerle, Renaud Tzanos, Enzo Pottez, Gabriel Arnould, Dorian Jaubert, and Jean-Marc Moisselin

This keynote will focus on recent research and developments at Météo-France's Nowcasting department. Firstly, the APIC warning system for exceptional rainfall at municipal level will be presented. This system was historically built on the comparison between observed rainfall rates of different durations and local climatological values. After aggregation at municipal level, warnings are displayed and sent automatically to over 13,000 institutional users. A 3-hour rainfall forecast, based on PIAF forecast ensembles that mix QPE and QPF extrapolation from the AROME-NWC model, has recently been added to these warnings. The way in which preferred values are deduced from these ensembles and how this information is communicated to subscribers will be described.

At the same time, much effort has been put into improving and completing the OPIC radar-based storm objects, which are detected every 5 minutes in the French radar mosaic, tracked on consecutive images and whose severity level is characterized on the basis of its attributes. Two different approaches were first tested to predict object contours up to an hour in advance : (1) by advection of points sampled from the contour and (2) using a convolutional neural network learned over 3 years of OPIC observation. An interesting complementarity was found, since the probability of occurrence of storm objects deduced from the second method includes the contours forecasted by the first. Machine learning algorithms were also calibrated on 7 months of object attributes collected during their tracking, such as morphological parameters, co-located observations of different types, environmental variables deduced from NWP models. The results and validation of the resulting one-hour forecasts of general severity levels and hazards such as precipitation and hail will be discussed.

How to cite: Montmerle, T., Tzanos, R., Pottez, E., Arnould, G., Jaubert, D., and Moisselin, J.-M.: Recent work at Météo-France’s Nowcasting department on warnings of exceptional rainfall and thunderstorm objects, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-225, https://doi.org/10.5194/ecss2025-225, 2025.

Discussion Post a comment

09:30–09:45

ECSS2025-5

An object-based method to study the life cycle of mesoscale convective systems and their environment from cloud-resolving AROME-France simulations

Gabriel Arnould, Thibaut Montmerle, Lucie Rottner, and Jean-Marc Moisselin

The aggregation of storms cells into a single entity generally results in mesoscale convective systems (MCSs) that extend over hundreds of kilometres and frequently produce severe hazards over large areas. The ambient conditions associated with the life cycle of MCSs in the central United States have been studied for more than twenty years, with the most impactful ingredients including potential instability, low-level moist-air advection and deep layer wind shear. However, the way in which the mesoscale environment influences the strengthening or dissipation of MCSs has received little attention in western Europe, although it could provide guidelines for MCS nowcasting. This lack motivates our study based on more than 150 simulated MCSs detected and tracked in mainland France from the outputs of the AROME-France convection-allowing numerical prediction model. MCS objects are detected using a convolutional neural network with simulated reflectivity and infrared brightness temperature images as input. Statistically, the life cycle of simulated MCSs is consistent with that of observed MCSs tracked in radar and satellite images from the same period, showing a classical three-stage pattern with development up to around 30 % of total lifetime, maturity between 30 and 60 % of lifetime and weakening thereafter. Next, we use two complementary methods to quantify the changes that the surrounding 3D environment undergoes during the life cycle of the simulated MCSs. In the first method, ambient variables are defined by averaging the AROME fields in a certain area around the objects. The second method introduces an original ring-shaped composite map approach in which the fields around the objects are projected onto a standard polar grid, enabling the surrounding environment of MCSs of different shapes or sizes to be examined statistically. The main results indicate a monotonic evolution of ambient variables over the life cycle, with the MUCAPE showing the most pronounced and significant decrease linked to drying and cooling at low level, ahead of the MCSs. Changes in upper-level wind are noted but their impact on the MCSs is less clear. At the end, our study highlights some of the ingredients responsible for MCS maintenance in western Europe, providing guidelines for the development of an object-based MCS nowcasting tool.

How to cite: Arnould, G., Montmerle, T., Rottner, L., and Moisselin, J.-M.: An object-based method to study the life cycle of mesoscale convective systems and their environment from cloud-resolving AROME-France simulations, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-5, https://doi.org/10.5194/ecss2025-5, 2025.

Discussion Post a comment

09:45–10:00

ECSS2025-102

Physical Scales Matter: The Role of Receptive Fields and Advection in Satellite-Based Thunderstorm Nowcasting with Convolutional Neural Networks

Christoph Metzl, Kianusch Vahid Yousefnia, Richard Müller, Virginia Poli, Miria Celano, and Tobias Bölle

Machine learning has a significant impact on severe weather nowcasting, with purely data-driven
models quickly moving into the focus of current research over traditional physically motivated
methods. Yet, recent work on radar-based precipitation nowcasting suggests that blending physical
insight—like advection—with deep learning can improve forecast skill. In this talk, we will present
our attempt to test how general this idea really is by applying it to satellite-based thunderstorm
nowcasting for the first time.
The central question we address is when and why advection should improve ML-based forecasts.
Using a simple scale argument, we show that advection through optical flow algorithms helps
preserve storm patterns within a model’s receptive field, which is critical for nowcasts at longer lead
times. To test this, we trained convolutional neural networks to nowcast thunderstorms based on
satellite imagery and lightning observations, with and without incorporating advected inputs.
When considering average performance, the impact is modest. But once we break down the results
by lead time and wind speed, a clear pattern emerges: the benefit of including advection emerges
after about 2 hours of lead time and grows with increasing wind speed, in agreement with our
physical reasoning.
Our findings highlight the importance of considering physical scales when designing and evaluating
ML-based forecasting systems.

How to cite: Metzl, C., Vahid Yousefnia, K., Müller, R., Poli, V., Celano, M., and Bölle, T.: Physical Scales Matter: The Role of Receptive Fields and Advection in Satellite-Based Thunderstorm Nowcasting with Convolutional Neural Networks, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-102, https://doi.org/10.5194/ecss2025-102, 2025.

10:00–10:15

ECSS2025-302

Blending NWP forecasts for the prediction of convective hazards

(withdrawn)

Pieter Groenemeijer, Tomas Pucik, Clemens Wastl, Phillip Scheffknecht, Ivan Tsonevsky, Francesco Battaglioli, and Mateusz Taszarek

10:15–10:30

ECSS2025-109

Severe storm signatures in three-dimensional lightning data (LMA): isolated lightning on the overshooting cloud tops

Oriol Rodriguez, Oscar van der Velde, Ferran Fabró, Nicolau Pineda, Marta Balagué, and Joan Montanyà

Lightning Mapping Array (LMA) networks are used to detect very high frequency (VHF, 60–66 MHz) electromagnetic sources associated with lightning channels inside clouds, providing three-dimensional lightning data. Over the past decades, LMA has been used in scientific campaigns with the aim of improving our understanding of electrification processes and revealing 3D lightning signatures related to severe weather. The use of real-time LMA data in weather surveillance and nowcasting provides valuable additional information to weather forecasters, playing an essential role in early warning alerts alongside other remote sensing systems such as traditional lightning detection networks, radar, and satellite.

Since 2023, an operational LMA network has been running in real time in Catalonia (northeastern Iberian Peninsula) as a collaboration between the Meteorological Service of Catalonia (SMC) and the Technical University of Catalonia (UPC). The XCALMA (extended Catalonia LMA) network consists of 30 LMA stations, spanning from the coast to the Pyrenees to cover the entire 32,000 km² country, making it the largest LMA network in Europe.

Several severe weather signatures can be detected using LMA data. One of them is the presence of isolated VHF detections in the overshooting top (OT), observed throughout the entire OT lifecycle. These detections, also known as sparkles, have been shown in some studies to precede surface severe weather reports—even occurring earlier than the commonly used lightning jump signature.

In this study, we focus on the analysis of a selection of severe storms reported in Catalonia between 2023 and 2025 that exhibited this signature, including large-to-giant hail and tornadic events. We combine LMA data with C-band radar and satellite observations to evaluate the presence of lightning in the OT as a potential predictor of severe weather.

How to cite: Rodriguez, O., van der Velde, O., Fabró, F., Pineda, N., Balagué, M., and Montanyà, J.: Severe storm signatures in three-dimensional lightning data (LMA): isolated lightning on the overshooting cloud tops, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-109, https://doi.org/10.5194/ecss2025-109, 2025.

10:30–10:45

ECSS2025-173

Fast Approximation for Diagnosing Convective Cloud Top Heights

Vinko Šoljan and Jadran Jurković

To avoid hazards associated with deep convective clouds, airplanes must fly above or around their tops. Therefore, accurate diagnosis of deep convective cloud top height (CTH) is crucial in aviation meteorology.

We employ an established operational method that involves comparing infrared satellite brightness temperature (BT) with a calculated parcel curve temperature. The intersection of BT and the parcel curve corresponds to a theoretical cloud top pressure level, which is directly related to altitude (flight level) in the standard atmosphere.

Typically, the parcel curve is calculated iteratively from surface temperature and dewpoint, but this process is computationally intensive for large datasets. In the first phase of this study, inspired by previous work on non-iterative calculations of moist adiabats, we found that the best approximation for moist adiabats is 5th-degree polynomial, with variable coefficients which are all functions of the wet bulb potential temperature. These coefficients can also be approximated with 4th-degree polynomials. In this approximation, a total of 6 polynomials (comprising 30 coefficients) must be evaluated, rendering it computationally very efficient. This represents a novel approach, as previous non-iterative approximations of moist adiabats employed a total of 200 coefficients and a different methodology to model the changing shape of moist adiabats.

In the second phase of the study the developed approximation was implemented in an operational environment. For the method to perform effectively for elevated convection, the temperature and dewpoint of the most unstable layer should be used for calculating the moist adiabat. For this purpose, the layer with the maximum equivalent potential temperature is assumed to be the most unstable layer. Additionally, it is important to note that this method's validity is limited to convective clouds, as other cloud types lack the updraft required for temperatures to follow moist adiabats.

The aim of the third phase of this study is the validation of the convective CTH diagnostic method. This validation can be challenging due to the lack of ground truth CTH data. This presentation will demonstrate the performance of our method across various convective situations and convective cloud top ranges (e.g., 6000-15000 m) by comparing the calculated CTH with radar vertical cross-sections, which are taken as ground truth. We also compared it with other similar products, such as the NWC SAF CTTH and radar ECHO TOPS products. Based on the analysis of all considered cases, we can conclude that our new method demonstrates excellent performance.

How to cite: Šoljan, V. and Jurković, J.: Fast Approximation for Diagnosing Convective Cloud Top Heights, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-173, https://doi.org/10.5194/ecss2025-173, 2025.

Discussion Post a comment

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

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

P52

ECSS2025-179

Training forecasters on the use of new generation satellite products

Vesa Nietosvaara, Natasa Strelec Mahovic, and Ivan Smiljanic

The complete constellation of Meteosat Third Generation (MTG) consists of three spacecrafts: two imaging satellites and one sounding satellite. The first imaging satellite, MTG-I1, is now operational under the name Meteosat-12. The satellite carries two important instruments – Flexible Combined Imager (FCI), a successor of SEVIRI on MSG, and a Lightning Imager (LI), the first space-based instrument monitoring lightning occurrence over Europe, Africa and even parts of South America, North America, and Asia from geostationary orbit.

The first sounder satellite, launched in July 2025, with its Infrared Sounder (IRS) will provide hyperspectral soundings of the atmosphere from geostationary orbit. It will track the three-dimensional structure of atmospheric water vapour and temperature. The Infrared Sounder instrument will detect instability in the atmosphere before clouds have formed.

In addition to these geostationary satellites, EUMETSAT Polar System Second Generation (EPS-SG) satellites will bring new observational capabilities from the polar orbit. This mission will continue the dissemination of information to improve numerical weather prediction (NWP) and support Nowcasting applications, especially at high latitudes, such as in Scandinavia.

All these new capabilities bring a huge amount of new data and products, which will be a challenge to absorb in the operational services. In our talk we will describe how EUMETSAT, together with its training partners in Europe and Africa, addresses this challenge. A variety of training material used in several types of training courses, as well as examples of key application guidelines presented to forecasters, will be shown. Finally, highlights from the evaluations provided by course participants, will be used to demonstrate the level of satisfaction with the courses and the need for improvements.

How to cite: Nietosvaara, V., Strelec Mahovic, N., and Smiljanic, I.: Training forecasters on the use of new generation satellite products , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-179, https://doi.org/10.5194/ecss2025-179, 2025.

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

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

P41

ECSS2025-14

GNSS Storm Nowcasting Demonstrator for Bulgaria

Guergana Guerova, Jan Douša, Tsvetelina Dimitrova, Anastasiya Stoycheva, Pavel Václavovic, and Nikolay Penov

Global Navigation Satellite System (GNSS) is an established atmospheric monitoring technique delivering water vapour data in near-real time with a latency of 90 min for operational Numerical Weather Prediction in Europe within the GNSS water vapour service (E-GVAP). The advancement of GNSS processing made the quality of real-time GNSS tropospheric products comparable to near-real-time solutions. In addition, they can be provided with a temporal resolution of 5 min and latency of 10 min, suitable for severe weather nowcasting. This paper exploits the added value of sub-hourly real-time GNSS tropospheric products for the nowcasting of convective storms in Bulgaria. A convective Storm Demonstrator (Storm Demo) is build using real-time GNSS tropospheric products and Instability Indices to derive site-specific threshold values in support of public weather and hail suppression services. The Storm Demo targets the development of service featuring GNSS products for two regions with hail suppression operations in Bulgaria, where thunderstorms and hail events occur between May and September, with a peak in July. The Storm Demo real-time Precise Point Positioning processing is conducted with the G-Nut software with a temporal resolution of 15 min for 12 ground-based GNSS stations in Bulgaria. Real-time data evaluation is done using reprocessed products and the achieved precision is below 9 mm, which is within the nowcasting requirements of the World Meteorologic Organisation. For the period May–September 2021, the seasonal classification function for thunderstorm nowcasting is computed and evaluated. The probability of thunderstorm detection is 83%, with a false alarm ration of 38%. The added value of the high temporal resolution of the GNSS tropospheric gradients is investigated for a storm case on 24–30 August 2021. Real-time tropospheric products and classification functions are integrated and updated in real-time on a publicly accessible geoportal.

How to cite: Guerova, G., Douša, J., Dimitrova, T., Stoycheva, A., Václavovic, P., and Penov, N.: GNSS Storm Nowcasting Demonstrator for Bulgaria, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-14, https://doi.org/10.5194/ecss2025-14, 2025.

P42

ECSS2025-39

SWING: A Post-Processing Algorithm for Improved Nowcasting and Environmental Safeguard

Martina Lagasio, Elena Oberto, Lorenzo Campo, Francesco Silvestro, Maria Laura Poletti, Massimo Milelli, and Antonio Parodi

This work presents SWING (Score-Weighted Improved NowcastinG), a novel post-processing algorithm designed to improve the accuracy and reliability of very short-term rainfall forecasts (nowcasting). SWING enhances the spatial and temporal predictability of convective rainfall events by combining high-resolution numerical weather prediction (NWP) outputs from the WRF model with radar-based nowcasting from the PhaSt system.

The algorithm operates by merging three forecasts over a 6-hour time window, updated every three hours, and weighing them based on recent performance. This evaluation is performed through an object-based comparison of modelled and observed rainfall fields using merged radar and rain gauge data. Each forecast is assigned a Reliability Score (RS) derived from spatial overlap, rainfall intensity, and object morphology, ensuring the final blended forecast maximizes accuracy while minimizing false alarms.

SWING has been running continuously for over a year, integrating high-resolution forecasts from the WRF model—updated every three hours using 3DVAR radar reflectivity assimilation and lightning data nudging—with radar-based nowcasting from the PhaSt system through a blending technique. A seasonal-scale validation of SWING against the standalone deterministic model run has been continuously performed since the system became operational.

SWING is fully automated and capable of generating rainfall scenarios and impact-based warnings through the output of a hydrological model (Continuum). Its rapid update cycle (3-hourly) makes it particularly suitable for operational early warning contexts where expert manual intervention is not feasible.

To assess its versatility, SWING is being extended within a multi-model forecasting framework. Preliminary tests will be presented to assess the algorithm’s ability to ingest and process outputs from different models, opening the door to ensemble-based rainfall scenarios and more robust hazard forecasting.

This research is conducted within the PNRR RAISE initiative - Spoke 3, which develops innovative technologies for environmental safeguard in water, air, and soil domains over the Ligurian region.

How to cite: Lagasio, M., Oberto, E., Campo, L., Silvestro, F., Poletti, M. L., Milelli, M., and Parodi, A.: SWING: A Post-Processing Algorithm for Improved Nowcasting and Environmental Safeguard , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-39, https://doi.org/10.5194/ecss2025-39, 2025.

Discussion Post a comment

P43

ECSS2025-41

Impacts of Lapse Rate Adjustments on Convective Parameters in the PLUVIA Mesoanalysis System

Billie Mackenzie, Katie Norman, Matt Clark, Andrew McNaughton, Anna Booton, and Ed Pavelin

The PLUVIA Mesoanalysis is a high-resolution, hourly updating analysis of near-surface variables over the UK. The Mesoanalysis combines observations from Automatic Weather Stations (including citizen Automatic Weather Stations via the Weather Observations Website) with lagged ensemble Unified Model data (T+4 to T+9 hours) using Ensemble Optimal Interpolation. Analysed variables include mean sea-level pressure, temperature, dewpoint temperature, wind speed and direction, visibility, Convective Available Potential Energy (CAPE), Convective Inhibition (CIN) and vorticity. The high spatial density of ingested observations (~3,000 sites over the UK) allows the Mesoanalysis to resolve localised features such as convergence lines, regions of enhanced CAPE or reduced CIN, and cold pools, which may have a bearing on the initiation and/or evolution of convection. This allows Operational Meteorologists to make inferences about the likely evolution of convection on Nowcasting timescales. The Mesoanalysis also provides analysis minus model background fields, to highlight where conditions are deviating from model expectations. The Mesoanalysis has been successfully trialled as a prototype and is currently being prepared for operational release onto the Met Office’s new supercomputer.

A key part of this work has been correcting errors that arise from the use of a smoothed orography field in the Unified Model (UM). Due to the smoothing, the model elevation is sometimes substantially different from the true elevation which can lead to error in the analysed temperature and/or dewpoint temperature fields. To address this problem, we have implemented variable lapse rate corrections to the temperature and dewpoint temperature fields. This requires approximating the local lapse rates using a nearest grid point neighbour gradient fitting approach. The lapse rates obtained are then multiplied by the height difference between the smoothed UM model and higher resolution (2km gridded) orography to give a temperature and dewpoint temperature correction at each grid point. In this presentation, we will give details of the correction methodology and explore how the correction influences other analysed variables, including the visibility, CAPE, and CIN. We will explore a range of case studies including cases of UK convection that resulted in substantial flooding impacts.

How to cite: Mackenzie, B., Norman, K., Clark, M., McNaughton, A., Booton, A., and Pavelin, E.: Impacts of Lapse Rate Adjustments on Convective Parameters in the PLUVIA Mesoanalysis System, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-41, https://doi.org/10.5194/ecss2025-41, 2025.

P44

ECSS2025-56

A new low-cost Unmanned Aerial Vehicle (Meteo-Dron) for monitoring upper air weather data and severe weather phenomena

Cesar Azorin-Molina, Amir Pirooz, Nicholas Kay, Jose Gomez-Reyes, and Carlos Calvo-Sancho

In the framework of the ThinkInAzul programme, the WIND-COAST project is a joint collaboration between the CSIC, NIWA and UOA aimed at designing and developing a new inexpensive “Meteo-Dron” for monitoring weather data across the low and mid-levels of the troposphere (up to 5,000-7,000 m a.s.l.). The Meteo-Dron is based on a DJI Matrice 350 RTK drone, equipped with the LI-550 TriSonica Mini Wind & Weather Sensor as its size and weight make it perfect for Unmanned Aerial Vehicle (UAV). The Meteo-Dron reports wind speed, direction, air temperature, humidity, pressure, tilt, and compass data.

The prototype has already been tested and calibrated in the wind tunnel of UOA to correct motion errors and evaluate its performance in different conditions of wind and turbulence. Field campaigns already started in September 2024 in New Zealand and Spain, first by flying the Meteo-Dron near a 10-m weather station from NIWA. The Meteo-Dron has potential in the long-term to be the substitute of existing operational radiosonde systems such as sounding balloons, which are very expensive and have relatively high environmental impact. The use of Meteo-Dron will lead to better real-time monitoring and forecasting of extreme weather events, in a more sustainable and less costly way. For instance, this novel equipment could improve convection-permitting models by sampling water content with high accuracy; e.g., HARMONIE does not properly estimate this parameter and the potential energy available to develop deep convection. Therefore, its ability to monitor wind storms and capabilities to improve the nowcasting of severe weather could be very useful for different socioeconomic sectors. For instance, the Meteo-Dron can have a wide range of applications, as e.g. being used by the General Directorate for the Prevention of Forest Fires in Valencia, supporting both the extinguishing and emergency management tasks.

How to cite: Azorin-Molina, C., Pirooz, A., Kay, N., Gomez-Reyes, J., and Calvo-Sancho, C.: A new low-cost Unmanned Aerial Vehicle (Meteo-Dron) for monitoring upper air weather data and severe weather phenomena, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-56, https://doi.org/10.5194/ecss2025-56, 2025.

P45

ECSS2025-73

Synoptic and mesoscale conditions of deep moist convection during the cold season in Croatia

Domagoj Dolicki, Petra Mikus Jurkovic, and Maja Telisman Prtenjak

In this study, thunderstorm activity during the cold part of the year was analyzed based on Thunderstorm Intensity Index (TSII) data on a predefined grid with a mesh of 3 km × 3 km in Croatia. The study covered a five-year period from 2016 to 2020, focusing on the months from October to March. The goal of the research was to conduct a spatial and temporal analysis of thunderstorm activity and determine the synoptic and thermodynamic conditions under which it occurs. The analysis aims to provide an overview of the fundamental characteristics, thereby improving the understanding of deep moist convection in the cold part of the year, which poses a significant challenge in operational weather forecasting due to its lower frequency and more difficult intensity assessment.

On surface synoptic charts the occurrence of surface frontal disturbances is detected and using 500 hPa synoptic charts the upper-level weather regime is determined. Thermodynamic and kinematic parameters are calculated from radiosonde profiles from stations in San Pietro Capofiume, Udine, Brindisi, Pratica di Mare (Italy), Zagreb and Zadar (Croatia), using the thundeR free software package.

A total of 296 convective days were selected for analysis from the observed period. The results indicate that synoptic forcing plays a significantly greater role in the development of convection during the cold part of the year compared to the warm part, while the dominant upper-level flow regime is southwesterly at the leading side of the upper level trough. The obtained values of Convective Available Potential Energy (CAPE) in the cold part of the year are much lower than those in the warm part. Additionally, most thunderstorms develop under conditions of strong vertical wind shear, indicating that the atmospheric environment conducive to winter thunderstorms is predominantly a high shear – low CAPE environment.

How to cite: Dolicki, D., Mikus Jurkovic, P., and Telisman Prtenjak, M.: Synoptic and mesoscale conditions of deep moist convection during the cold season in Croatia, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-73, https://doi.org/10.5194/ecss2025-73, 2025.

P46

ECSS2025-81

Integrating KONRAD3D-Sinfony Ensemble Information into the Nowcasting Guidance System NowCastMIX

Michael Debertshäuser, Paul James, Gergely Bölöni, and Manuel Werner

Short-term warnings for severe thunderstorms are generated at the German Weather Service (DWD) through the NowCastMIX system, which provides automated warnings for the next 60 minutes. NowCastMIX processes meteorological fields from various sources, including NWP, radar, surface station reports, and lightning detections. Every five minutes, the system employs a hierarchy of fuzzy logic sets to calculate the potential for heavy rain, hail, and severe gusts based on the available data. Categorical thunderstorm warnings are then issued for detected cells, and regions requiring warnings are identified by condensing the information into clusters.
In previous work, the original KONRAD cell detection scheme within NowCastMIX was replaced by a newly developed and more sophisticated cell detection and tracking algorithm called KONRAD3D. This advancement enabled a more precise three-dimensional analysis of convective cells based on radar volume scans, leading to improved characterization of storm structures. Building on this foundation, the next step is to leverage the latest development: KONRAD3D-SINFONY. This extended version not only identifies convective cells but also generates ensemble-based information about future cell behavior.
Although NowCastMIX incorporates ensemble data from numerical weather prediction (NWP) models to characterize the near-storm environment, a limitation remains due to its primarily deterministic nature, which does not fully account for the inherent uncertainty in convective development. To address this, we investigate how the ensemble-based output of the newly developed KONRAD3D-SINFONY system can be utilized within the NowCastMIX framework. Our research focuses on methods to incorporate probabilistic cell detections into the existing warning system, enabling a more nuanced representation of forecast uncertainty.

How to cite: Debertshäuser, M., James, P., Bölöni, G., and Werner, M.: Integrating KONRAD3D-Sinfony Ensemble Information into the Nowcasting Guidance System NowCastMIX, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-81, https://doi.org/10.5194/ecss2025-81, 2025.

Discussion Post a comment

P47

ECSS2025-91

Detection and nowcast of severe thunderstorms over Slovenia

Matic Savli, Matevž Osolnik, Janko Merše, Barbara Gabrovšek, and Eva Bezek

A method for detection and nowcast of severe thunderstorm events in the area of Slovenia is going to be presented.

Detection of intense convective cells is based on meteorological radar measurements and lightning discharges.
The pysteps module is used to search for closed areas above a certain threshold of radar reflectivity, and
then it determines convective cells from the closed areas using several additional criteria related to the lightning activity.
A fine-tuning of detection is needed to provide optimal performance. This is achieved by first conducting a thorough analysis of the
convective seasons from May to September (2020-2023) followed by an optimization approach which tries to provide the best configuration.
Primarily, the performance of detection is verified by the intervention data of the administration for civil protection and disaster relief (URSZR), where the events of hail, severe
winds and floods of stormwater were jointly analysed.
Due to the specifics of the interventions' database, the optimization and verification are performed only in the predefined 13 areas
where a sufficient density of reports are expected.
Each event detected is tracked back in time, which provides the ability to follow the lifetime of the specific thunderstorm event.
Detection generally performs well, and the verification results are comparable to the results of other studies performed in Slovenia or abroad.

Events detected at the target time are nowcasted in to the future.
This is currently performed by a nowcast of radar reflectivity field alone, followed by a repeated detection and tracking performed at the forecasted times.
The pysteps module is used to perform nowcast, since it already provides a list of well accepted methods.

The whole framework is currently used by operational subjective weather forecast process at the Slovenian NMS.
Also it is already used operationally to support automated warning system for a specific user.
Due to the modular approach, the framework has potential to be extended for various external costumer needs.

How to cite: Savli, M., Osolnik, M., Merše, J., Gabrovšek, B., and Bezek, E.: Detection and nowcast of severe thunderstorms over Slovenia, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-91, https://doi.org/10.5194/ecss2025-91, 2025.

Discussion Post a comment

P48

ECSS2025-92

Improvement of lightning nowcasting model using convective cell-based radar motion vectors

MyoungJae Son, Hae-Lim Kim, and Mi-Kyung Suk

The Korea Meteorological Administration (KMA) has operated a lightning nowcasting model based on radar-derived motion vectors using MAPLE(McGill Algorithm for Precipitation nowcasting by Lagrangian Extrapolation) since 2015. This model provides 10-minute interval forecasts with lead times of up to 6 hours for use by both forecasters and the general public.
In this study, we present a newly developed lightning nowcasting model designed to extend lead times and enhance the timeliness and accuracy of lightning risk alerts. Unlike conventional methods that calculate motion vectors across the entire precipitation field, the proposed model automatically identifies convective cell areas with high lightning potential based on the ETOP30 threshold (reflectivity ≥ 30 dBZ). Within these selected regions, sequential Hybrid Surface Rainfall (HSR) radar fields are analyzed using the Variational Echo Tracking (VET) algorithm, which estimates high-resolution motion vectors (1 km, 10 min) by optimizing a cost function that minimizes differences in reflectivity across three consecutive radar images.
To mitigate limitations of convective cell-based motion vector fields, the MAPLE motion vector field at previous 10 minutes in real-time is used as a background field to correct initial estimation errors in the VET algorithm. This hybrid approach enables more accurate tracking of convective cell evolution and movement, while also reducing the delivery time of lightning nowcasting information by approximately 7 minutes compared to the previous model.
Validation using 16 lightning cases from 2023 to 2024, comparing the nowcasting fields to LINET lightning observations, showed that the new model achieved an average 1-hour forecast CSI of 0.55, POD of 0.57, and FAR of 0.08. A peak CSI of 0.68 was recorded during a band-type lightning event on July 7, 2024.
These results demonstrate the improved performance of the proposed model in operational lightning nowcasting and highlight its potential for enhancing real-time risk assessment and public weather services.

KEYWORD
Convective Cell, HSR, Radar Motion vector, Lightning Nowcasting, ETOP30, VET

Acknowledgements:
This research was supported by the ”Development of Integrated radar analysis and customized radar technology (KMA2021-03021)” of “Development of integrated application technology for Korea weather radar” project funded by the Weather Radar Center, Korea Meteorological Administration.

How to cite: Son, M., Kim, H.-L., and Suk, M.-K.: Improvement of lightning nowcasting model using convective cell-based radar motion vectors, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-92, https://doi.org/10.5194/ecss2025-92, 2025.

P49

ECSS2025-103

Using Surface Moisture Flux Convergence for Convective Nowcasting

Pere Cladera, Sergio Gallego, and Francesc Figuerola

For several years, the Meteorological Service of Catalonia has operated a dense network of automatic weather stations, which the Forecast and Surveillance Team uses to monitor hazardous meteorological conditions in real time. In convective scenarios, wind direction data have proven useful for identifying low-level convergence zones, which are often indicative of potential convective development. However, a more objective parameter was needed to assess whether these convergence areas provide sufficient forcing to initiate or sustain convection.

This need is addressed through the use of Moisture Flux Convergence (MFC), a variable that combines specific humidity advection with wind field convergence to quantify the potential for moisture-driven vertical ascent. MFC is calculated by interpolating wind components and specific humidity from the station network onto a regularly spaced grid.

The resulting surface MFC fields are employed as a nowcasting tool (0–2 hours), capable of identifying areas of moist air convergence and mesoscale boundaries between surface air masses. These zones often coincide with key ingredients for convective initiation or maintenance. Initial applications show that surface MFC performs particularly well in warm-rain scenarios, characterized by high moisture content and low cloud bases. It also shows good performance in deep convection events with similarly low cloud bases. Its effectiveness is more limited in storms with elevated cloud bases and in regions with complex terrain, where wind interpolation becomes less reliable.

Overall, surface MFC offers a valuable complement to traditional observational tools, enhancing spatial and temporal resolution for short-term convective monitoring and operational decision-making processes.

How to cite: Cladera, P., Gallego, S., and Figuerola, F.: Using Surface Moisture Flux Convergence for Convective Nowcasting, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-103, https://doi.org/10.5194/ecss2025-103, 2025.

P50

ECSS2025-112

A Multimodel Approach for Forecasting of convective weather in Support of BULATSA Air Traffic management

Ilian Manfov and Rosen Penchev

A Multimodel Approach for Forecasting of convective weather in Support of BULATSA Air Traffic management

Authors: Ilian Manafov PhD BULATSA – Bulgarian Air Traffic Services Authority

Rosen Penchev BULATSA – Bulgarian Air Traffic Services Authority

Abstract

Background:

The impact of adverse weather condition on European ATM Network have increased significantly in recent years, which led to a significant increase in delays in summer 2024. In an attempt to mitigate the impact of weather, Bulgarian Air Navigation Service Provider BULATSA has introduced a special weather procedure that expected from weather forecasters to provide more accurate and spatially detailed predictions of severe weather areas within the Bulgarian airspace.

Methods:

This study introduces a multimodel forecasting approach aimed at improving predictions of hazardous convective phenomena. Operational weather forecasters at BULATSA utilize a variety of weather models. The proposed method combines special chosen NWP parameters from four numerical weather prediction models including ICON-EU, ECMWF deterministic model and two non-hydrostatical regional models: BULATSA-WRF and version of ALADIN integrated by Bulgarian National Institute of Meteorology and Hydrology.

A new forecast product, CB-Conditions, is developed based on the intersection of forecasted parameters from two of the models. It identifies areas with a high likelihood of hazardous CB activity. The overall probability of occurrence is derived using a weighted average of probabilities from all four models.

Validation and Tools:

To assess the accuracy and determine optimal model weights, at least 20 cases of intense convection will be analyzed. The CB forecasts are drawn as a polygons and verified against archived data from BULATSA Weather Radar network and LINET network.

Results and Conclusion:

Preliminary results indicate promising performance of the CB-Conditions product. The multimodel approach demonstrates potential to significantly enhance the accuracy and operational value of convective weather forecasts, contributing to safer and more efficient air traffic management.

How to cite: Manfov, I. and Penchev, R.: A Multimodel Approach for Forecasting of convective weather in Support of BULATSA Air Traffic management , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-112, https://doi.org/10.5194/ecss2025-112, 2025.

Discussion Post a comment

P51

ECSS2025-116

Increasing NWP thunderstorm predictability using ensemble data and machine learning

Kianusch Vahid Yousefnia, Christoph Metzl, and Tobias Bölle

Forecasting thunderstorms several hours in advance remains challenging due to the increasing forecast uncertainty in numerical weather prediction (NWP) models with lead time. Ensemble systems address this limitation by providing multiple scenarios consistent with forecast uncertainty and are well known to enhance the skill of deterministic systems. This study introduces a simple yet novel analytic expression that quantifies the improvement in Brier Skill Score (BSS) achieved by averaging over binary classification predictions from multiple ensemble members. This is particularly relevant to severe weather forecasting, where the task often involves estimating the probability of events such as lightning, heavy rainfall, or hail. The derivation of the formula relies only on the assumption that ensemble members are indistinguishable. We validate this expression using SALAMA 1D, a recent machine learning (ML) model designed to predict thunderstorm occurrence from convection-permitting ICON-D2-EPS ensemble forecasts over Central Europe. Our formula accurately captures the impact of ensemble averaging on the ML model's performance, which, in this case, results in extending the model’s 5-hour deterministic skill out to 11-hour lead times. While we use an ML model to exemplify our formula, its validity extends also to traditional (non-ML) approaches for severe weather identification in NWP data. Furthermore, we show that ML models like SALAMA 1D, which are trained using observations as ground truth labels, can identify patterns in thunderstorm occurrence that remain predictable for longer lead times compared to raw NWP output. Our findings offer insight on the use of ensemble forecasts of thunderstorm occurrence and support the growing use of ML techniques in severe weather forecasting.

How to cite: Vahid Yousefnia, K., Metzl, C., and Bölle, T.: Increasing NWP thunderstorm predictability using ensemble data and machine learning, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-116, https://doi.org/10.5194/ecss2025-116, 2025.

Discussion Post a comment

P52

ECSS2025-117

Probabilistic Localized Radar-Based Nowcasting of Flood-Inducing Rainfall Events

Daniel Eduardo Villarreal-Jaime, Patrick Willems, Lesley De Cruz, and Ricardo Reinoso-Rondinel

Accurate short-term rainfall forecasts, also known as nowcasts, are essential for building effective flood early warning systems, especially for convective events in urban areas with rapid hydrologic response. In addition, developing and using probabilistic ensemble forecasts can provide decision makers and stakeholders in highly populated areas with a clearer understanding of forecast uncertainty, supporting better flood risk management and response planning.

Traditional rainfall nowcasting techniques, such as extrapolation with Lagrangian persistence, are not able to predict the growth and decay of precipitation. To address these limitations, deterministic methods like RadVIL, which uses mass balance equations of the Vertically Integrated Liquid (VIL), and Spectral Prognosis (SPROG), which performs a spectral decomposition of rainfall fields and an autoregressive (AR) model, have been developed and improved over time. Methods, like SPROG-Localized (SPROG-LOC) and Autoregressive Nowcasting using the VIL (ANVIL), improve the internal evolution of rainfall fields by adding localization in the AR model and the accuracy for intense rainfall by using an autoregressive integrated (ARI) model on the VIL, respectively. Additionally, the probabilistic version of SPROG, called the Short-Term Ensemble Prediction System (STEPS), adds stochastic noise to include uncertainties in the precipitation providing an ensemble nowcast.

Building on these methods, we propose Short-Term Autoregressive Nowcasting (STAN), a novel integrated approach designed to leverage the strengths of existing techniques. STAN enables the seamless combination and reproduction of previous methods, incorporating improvements such as adaptive localization, which extends the lifetime of small convective cells. This approach aims to improve nowcasting performance, particularly in cases with large, non-uniformly distributed precipitation areas and isolated convective features.

While we are currently optimizing and evaluating the STAN configuration to achieve the best possible performance, we will present results with nowcasting lead times up to 2 hours of our deterministic and probabilistic approaches for different precipitation events that caused flood in Belgium. Preliminary results show that STAN in the deterministic version, which is using the VIL, an AR model and adaptive localization, shows a better Fraction Skill Score (FSS) with high precipitation thresholds (> 5 mm/hr) and Root Mean Squared Error (RMSE) than SPROG. In the probabilistic version, when stochastic noise is included, STAN shows better performance in FSS, RMSE, Equitable Threat Score (ETS) and False Alarm Ratio (FAR) than STEPS for high thresholds.

How to cite: Villarreal-Jaime, D. E., Willems, P., De Cruz, L., and Reinoso-Rondinel, R.: Probabilistic Localized Radar-Based Nowcasting of Flood-Inducing Rainfall Events, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-117, https://doi.org/10.5194/ecss2025-117, 2025.

Discussion Post a comment

P53

ECSS2025-122

The Devastating Convective Wind Event of 24 July 2023 in La Chaux-de-Fonds, Switzerland : Causes, Probable Mesoscale and Storm-Scale Mechanisms at Play and Nowcasting Implications

Lionel Peyraud, Aude Untersee, Stephan Vogt, Marco Stoll, Barbara Galliker, and Isabelle Bey

On July 24^th 2023, a low-topped high-precipitation (HP) supercell thunderstorm struck the city of La Chaux-de-Fonds in northwest Switzerland and caused over 135 million Swiss Francs of material damages, tore down approximately 22,000 trees, damaged 3,000 buildings throughout the city, resulted in one fatality, 45 injuries and inflicted deep psychological trauma on its inhabitants. A MeteoSwiss automated ground station at the local airport in the direct path of the storm recorded a validated maximum 1-sec convective wind gusts of 217 km/h. After an extensive post-event analysis utilizing multiple data sources (including but not limited to numerical weather model data, satellite/radar imagery/algorithms, a damage survey, high resolution aerial photography, video footage and ground station data), it was determined that both a microburst and tornado were responsible for the devastating wind damages rated IF2 on the International Fujita Scale. It is hypothesized that this hybrid outcome was a result of a phasing of specific meteorological parameters and phenomena at various spatial and temporal scales. The data seem to show that the remnants of a rear-inflow jet (RIJ) present from an earlier bow-echo phase of the storm coupled with a descending reflectivity core (DRC) associated with the supercell’s rear-flank downdraft (RFD) collapsed to the ground as it was approaching the city. This phasing at the storm-scale generated a powerful wet microburst within the RFD which seems to have helped initiate a tornadic circulation just to its north as the overall HP mesocyclonic circulation literally engulfed the city and surroundings, traversing it in a few minutes. The rapid evolution of this event stresses the importance of disposing of and developing new nowcasting tools and techniques aimed at increasing warning lead times for these type of convective wind events, even if by only a few minutes. This unique case within the Jura topography also highlights the complex interactions that can take place as dynamic convective storms, including supercells, impringe on this mountain chain and helps explain perhaps why the Jura region has been known to have spawned several strong and even violent tornadoes in the past. This case will hopefully motivate further observational and perhaps high-resolution convection-allowing numerical modeling studies aimed at better understanding what makes this region particularly prone to occasional tornadoes and help determine what role the complex Jura topography may actually play.

How to cite: Peyraud, L., Untersee, A., Vogt, S., Stoll, M., Galliker, B., and Bey, I.: The Devastating Convective Wind Event of 24 July 2023 in La Chaux-de-Fonds, Switzerland : Causes, Probable Mesoscale and Storm-Scale Mechanisms at Play and Nowcasting Implications, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-122, https://doi.org/10.5194/ecss2025-122, 2025.

Discussion Post a comment

P54

ECSS2025-124

Deep Learning vs. Traditional Satellite-Based Thunderstorm Nowcasting: Outline of a Model Benchmark Study

Philipp Straub, Christoph Metzl, Richard Müller, Virginia Poli, Miria Celano, and Tobias Bölle

The ability to produce reliable short-range forecasts for thunderstorms is crucial for issuing timely emergency warnings to the general population, protecting vital infrastructure and alerting first responders in advance. Regularly affected by thunderstorms and associated hazards are remote areas, mountainous regions and air traffic, necessitating broadly available nowcasting solutions. Geostationary satellites provide an ideal data source for thunderstorm nowcasting in these cases.

Traditional approaches employ deterministic algorithms to predict the occurrence and evolution of thunderstorms essentially by solving the optical flow problem. Recently, machine learning (ML) has emerged as an alternative and already shown promising results. While U-Net-based architectures have proven to be a very robust baseline, ML-based methods currently represent an extremely active area of research and optimal approaches have yet to crystallise. Particularly for satellite-based thunderstorm nowcasting, the superiority of either traditional methods or a specific ML architecture has yet to be established.

It is our goal to narrow this knowledge gap by performing an extensive benchmark. In this talk we present a preliminary study featuring a set of nowcasting tools continuously improved over the past 20 years at the German Aerospace Centre (DLR) as well as a newly developed ML-model. The presented evaluation pipeline is developed jointly with the German Meteorological Service (DWD) and the regional Italian meteorological service Arpae Emilia-Romagna as part of the Italia-Deutschland Science-4-Service network. It serves as a baseline for a broader follow-up study including satellite-based thunderstorm nowcasting operationally used at DWD and Arpae. All models are evaluated over a period of multiple years for a fixed region in central Europe, using only satellite imagery from the SEVIRI instrument onboard MSG as input, while lightning recorded by the LINET network serves as ground truth. The accuracy of each model is measured for a set of lead times up to 180 minutes according to established metrics in a homogeneous setting.

Our work aims to guide future research and development efforts, ultimately paving the way for improved satellite-based thunderstorm nowcasting.

How to cite: Straub, P., Metzl, C., Müller, R., Poli, V., Celano, M., and Bölle, T.: Deep Learning vs. Traditional Satellite-Based Thunderstorm Nowcasting: Outline of a Model Benchmark Study, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-124, https://doi.org/10.5194/ecss2025-124, 2025.

P55

ECSS2025-126

Nowcasting of Thunderstorm Hazards with Deep Learning: Performance Report of the First Convective Season in Operations

Ulrich Hamann, Luca Nisi, Irina Mahlstein, Matteo Buzzi, Michele Cattaneo, Néstor Tarin Burriel, Przemyslaw Juda, Nathalie Rombeek, George Pacey, Ophélia Miralles, and Jussi Leinonen

Thunderstorms pose serious risks to life and property through hazards such as lightning, heavy rainfall, hail, and strong winds. These events develop rapidly and affect localized areas, making timely and accurate short-term forecasts essential for early warnings to the public, emergency services, and infrastructure operators. On the sub-hourly scale, nowcasting - statistical forecasting based on the most recent observations - offers high spatial and temporal precision. In particular, deep learning models can effectively perform nowcasting by learning to approximate physical processes that are implicitly embedded in diverse observational datasets. These models can generate accurate, multi-hazard predictions in seconds, making them ideal for operational early warning systems.

COALITION-4 is a deep learning nowcasting model utilizing an advanced encoder-forecaster model to nowcast thunderstorm-related hazards such as accumulated precipitation, lightning occurrence, and hail probability up to one hour in advance. It leverages recurrent convolutional layers and integrates a range of predictor datasets, including radar observations from the Swiss dual-polarization radar network and Météorage lightning data. Input data undergo extensive preprocessing and data augmentation to improve generalization. The model is trained using GPU-accelerated optimization with an adaptive learning rate. The inference is performed for the entire domain of Switzerland.

Recently, COALITION-4 has been operationally deployed including continuous monitoring of product completeness and quality as well as fall back options to degraded modes when input data are incomplete. In the convective season 2025, the system is evaluated and compared to the current operational thunderstorm nowcasting system by forecasters at the MeteoSwiss forecasting centers who in turn inform different key users about thunderstorm hazards: cantonal authorities, civil protection agencies and fire brigades use the information to optimize response strategies during and after thunderstorms. In aviation, accurate short-term lightning forecasts can improve safety and efficiency by guiding temporary suspensions of airport ground operations. We will present both qualitative and quantitative comparisons with current operational thunderstorm nowcasting system of MeteoSwiss, focusing on warning lead times, spatial accuracy, and consistency of forecasts across multiple hazards under diverse convective conditions. This assessment and forecaster feedback informs the next steps in model development, including enhancing forecast quality, usability, interpretability and integration into operational workflows.

How to cite: Hamann, U., Nisi, L., Mahlstein, I., Buzzi, M., Cattaneo, M., Tarin Burriel, N., Juda, P., Rombeek, N., Pacey, G., Miralles, O., and Leinonen, J.: Nowcasting of Thunderstorm Hazards with Deep Learning: Performance Report of the First Convective Season in Operations, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-126, https://doi.org/10.5194/ecss2025-126, 2025.

Discussion Post a comment

P56

ECSS2025-133

Integrated use of MTG tools and radar for nowcasting storms: a case study from Istanbul Airport

Melek Erdal and Gizem Hodoglu

The precise nowcasting and forecasting of convective storms affects airport operations, as they are important to the safe use of aircraft, scheduling and affecting ground operations. The present study integrated the analysis of convective activity over Istanbul Airport from 20-21 September 2024 using single satellite imagery, ground based radar data and recently available products from the Meteosat Third Generation (MTG) for real time operability for forecast decision making.

The analysis commenced with continuous monitoring utilized through combined use of EUMETView satellite imagery and ground based weather radar observations of the same meteorological event. The RGB composites of air mass derived from the satellite data, infrared brightness temperature fields at different levels and visible reflectance channels allowed for the identification of large-scale atmospheric structure, upper-level forcing and cloud-top cooling trends. Meanwhile, the radar reflectivity provide a high level of resolution to track precipitation cores, echo tops and vertical development of the storm.

As the convective activity intensified, the FCI provided data that was able to allow clearer probability of cloud microphysical and dynamical properties, while the LI was used to monitor the persistence and detection of intra-cloud and cloud-to-ground lighting flashes. Following this, the distribution of density flash groupings and the extent of each flash was analyzed in conjunction with previous analysis in order to infer convective intensity and cell organization. To combine the multi-platform datasets, the ESSL Weather Displayer was used to simultaneously visualize satellite, radar, and model-derived fields.

This study shows that the integration of MTG data into operational meteorology provides substantial benefits for aviation operation, particularly at major hubs like Istanbul Airport where weather-related disruptions can have significant impacts.

How to cite: Erdal, M. and Hodoglu, G.: Integrated use of MTG tools and radar for nowcasting storms: a case study from Istanbul Airport , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-133, https://doi.org/10.5194/ecss2025-133, 2025.

P57

ECSS2025-142

Development and evaluation of a new convective index

Margarida Belo-Pereira

Deep convective clouds, such as towering cumulus and Cumulonimbus (Cb), can endanger lives and property, being also a significant hazard to aviation. This study presents the convective index (IndexCON) used operationally at the Portuguese Meteorological Watch Office. IndexCON forecasts (for steps ranging from 12 to 35 hours) were evaluated against lightning and precipitation observations. Reports of hail and tornadoes were also used to define a convective event. This evaluation was conducted over two years (January 2022 to December 2023) across mainland Portugal and its surrounding regions.

IndexCON integrates several prognostic variables from the European Centre for Medium-Range Weather Forecasts (ECMWF), including stability indices, cloud water content, relative humidity, and vertical velocity, using a fuzzy-logic approach. The index performs well during the warm season (May–October), achieving a probability of detection (POD) of 70%, a false alarm ratio (FAR) of 30%, and a probability of false detection (POFD) below 5%, resulting in a critical success index (CSI) above 0.55. However, IndexCON tends to overestimate convective activity during the cold season (November–April), leading to lower performance scores. This study demonstrates that optimising the membership functions can partially mitigate this overestimation.

In addition, the study evaluates other convective predictors, such as convective available potential energy (CAPE), Total-Totals (TT), K index (KI), lifted index (LIs), and Jefferson Index. Finally, the performance of IndexCON is illustrated in detail for two thunderstorm episodes occurring in different seasons, using satellite products, synoptic and radiosonde observations, lightning, and precipitation data. For these events, the Cb cloud tops derived from IndexCON were also compared with the NWC SAF cloud top product.

How to cite: Belo-Pereira, M.: Development and evaluation of a new convective index, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-142, https://doi.org/10.5194/ecss2025-142, 2025.

P58

ECSS2025-146

Thunderstorm nowcasting at IMGW

Przemysław Baran, Anna Jurczyk, Agnieszka Kurcz, Krystian Specht, and Jan Szturc

A storm detection model has been developed and implemented at the Institute of Meteorology and Water Management (IMGW), along with a forecast of the movement of storm cells with a leading time up to 60 minutes (TSP - Thunderstorm Prediction). The model is based on data from the 1-minute reports of the PERUN (lightning detection system): density of intercloud lightning, density of cloud-to-ground lightning, maximum lightning jump, number of lightning jumps, within 10 minutes. Another source of data are radar data from the POLRAD network and from neighboring countries: VIL (Vertically Integrated Liquid), EHT (Echo Top Height), CMAX (Column Maximum), CAPPI (Constant Altitude Plan Position Indicator), and the 0ºC isotherm altitude from the NWP COSMO model. In addition, Meteosat satellite data processed with NWC-SAF software: CTTH (cloud top temperature and height) and RDT-CW (rapidly developing thunderstorm – convection warning) were used. The heart of the system is a trained model based on the SVM (Support Vector Machines) method. Its calibration was carried out on observation data from synoptic stations located throughout Poland. This model determines the intensity class of the storm and the probability of its occurrence. The intensity forecast is based on the RDT-CW satellite product: the current and forecast storm intensity class and the forecast intensity class obtained with the SVM model. The probability predictions included RDT-CW, CTTH and parameters determined from the lightning dynamics analysis. The forecast of the movement of storm cells in the model is based on the displacement vectors obtained from the SCENE precipitation forecast model. A dedicated algorithm was developed for the graphical presentation of the TSP model results, which takes into account the uncertainty of the displacement vectors of the SCENE model, which affects the size of the area of potential storm occurrence in the assumed time horizon of 60 minutes. The results of the model have been used in services dedicated to aviation (Polish Air Traffic Control Agency) and are also presented to the public.

How to cite: Baran, P., Jurczyk, A., Kurcz, A., Specht, K., and Szturc, J.: Thunderstorm nowcasting at IMGW , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-146, https://doi.org/10.5194/ecss2025-146, 2025.

P59

ECSS2025-158

Towards a machine-learning enhanced nowcasting tool for storm severity analysis and prediction

Gergely Bölöni, Paul M. James, Michael Debertshäuser, and Susanne Theis

At the German Weather Service (DWD), an important tool for issuing severe weather warnings is the NowCastMIX system (James et al., 2018), which provides warnings for the next hour based on radar, lightning, NWP and other data. In its analysis step, NowCastMIX detects thunderstorm cells and assigns them a severity level based on remote sensing and numerical model data at their locations. In the prediction step, the thunderstorm cells are clustered and then spatially extrapolated according to the estimated motion vectors of the cells to provide a simple temporal evolution for the next hour. The extrapolated cell clusters retain the severity assigned to them in the analysis step.

One of the focal points of the work presented is the analysis step, in particular the assignment of severity to the detected storm cells. This is currently based on fuzzy logic, which allows estimation of severity based on a set of thresholds for measured or predicted meteorological variables. This approach works well, but systematic changes to the input systems (e.g. radar upgrades, improvements to radar products, major model upgrades) always require re-tuning of the fuzzy functions, making the adaptation of NowCastMIX relatively complex and costly. As an alternative, we present our attempts to replace fuzzy logic with machine learning. Although this approach is very attractive due to its simplicity and flexibility, it also has its limitations, which we will present in detail.

The other focus of the presentation is the prediction step. As mentioned before, this is based on linear advection in the current version of NowCastMIX. To improve this, we present an approach that incorporates the convective life cycle of cell clusters by predicting the temporal evolution of storm severity using machine learning.

How to cite: Bölöni, G., James, P. M., Debertshäuser, M., and Theis, S.: Towards a machine-learning enhanced nowcasting tool for storm severity analysis and prediction, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-158, https://doi.org/10.5194/ecss2025-158, 2025.

Discussion Post a comment

P60

ECSS2025-163

SALAMA 1D: Identification of thunderstorm occurrence from convection-permitting forecasts of vertical profiles using deep learning

Kianusch Vahid Yousefnia, Christoph Metzl, and Tobias Bölle

Thunderstorms pose significant risks to society and the economy due to hazards such as heavy precipitation, hail, and strong winds, necessitating accurate forecasting to mitigate their impacts. Convection-permitting numerical weather prediction (NWP) models can explicitly resolve convective processes, but predicting thunderstorms from their output remains challenging since there is no obvious variable that directly indicates thunderstorm occurrence. Many approaches rely on combining multiple single-level variables, such as convective available potential energy (CAPE), which are derived from state variables like temperature, pressure, and specific humidity, and act as surrogates for thunderstorms. In this study, we present a deep neural network model that bypasses surrogate variables and instead directly processes the vertical profiles of state variables provided by convection-permitting forecasts. Our model, SALAMA 1D, analyzes ten different NWP output fields, such as wind velocity, temperature, and ice particle mixing ratios, across the vertical dimension, to produce the corresponding probability of thunderstorm occurrence. The model’s architecture is motivated by physics-based considerations and symmetry principles, combining sparse and dense layers to produce well-calibrated, pointwise probabilities of thunderstorm occurrence, while remaining lightweight. We trained our model on two summers of forecast data from ICON-D2-EPS, a convection-permitting ensemble weather model operationally run by the German Meteorological Service (DWD), using the lightning detection network LINET as the ground truth for thunderstorm occurrences. Our results demonstrate that, up to lead times of (at least) 11 hours, SALAMA 1D outperforms a comparable machine learning model that relies solely on derived variables. Additionally, a sensitivity analysis using saliency maps indicates that the patterns learnt by our model are to a considerable extent physically interpretable. This work advances NWP-based thunderstorm forecasting by demonstrating the potential of deep learning to extract valuable predictive information from high-dimensional NWP data while preserving model interpretability.

How to cite: Vahid Yousefnia, K., Metzl, C., and Bölle, T.: SALAMA 1D: Identification of thunderstorm occurrence from convection-permitting forecasts of vertical profiles using deep learning, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-163, https://doi.org/10.5194/ecss2025-163, 2025.

Discussion Post a comment

P62

ECSS2025-196

The Role of CIN Depth in Regulating Deep Convection Initiation

Adam Houston, Stephen Shield, and Kylee Matousek

Convective inhibition (CIN) reduces the kinetic energy of rising air but, when controlling for the magnitude of CIN, the depth of the CIN layer could impact the likelihood of deep convection initiation (DCI). Considering thermodynamics alone, CIN depth should not impact vertical velocity at the LFC, but it would impact the transit time of air passing through the CIN layer. This could impact entrainment within this layer. Moreover, when ascent below the LFC is driven by non-thermodynamic forcing imparting kinetic energy to air passing through the CIN layer, ascent through the CIN layer is not controlled solely by the integrated buoyancy but also by the evolution of this forcing. During transit of air through the CIN layer it is expected that non-thermodynamic forcing will evolve significantly. Thus, DCI is likely to depend on transit time through the layer and, by extension, the CIN layer depth.

Results will be presented from analysis of environments near more than 60,000 observed DCI points. Vertical profiles of the atmospheric state are approximated using RAP/RUC analyses and are compared to vertical profiles at the same time but away from initiation points (referred to as Null points). Results show that CIN depth is among the most important distinguishing parameters and is more important than CIN in differentiating DCI environments from Null environments.

Idealized numerical experiments were also conducted to explain the importance of CIN depth while controlling for CIN magnitude. A generic non-thermodynamic impulsive initiation mechanism is imposed below the LCL. Experiments reveal a systematic decrease in the likelihood of DCI as CIN layer depth is increased for a given CIN. Specifically, for deeper CIN depth, a longer traverse of parcels through the inversion along with natural relaxation of dynamic forcing means that air reaching the LFC encounters upward forcing that is insufficient to carry it vertically even as thermal instability is released. Unexpectedly, clouds at the LFC in environments with deeper CIN layers are actually larger. Ordinarily, larger clouds would be expected to be more likely to yield DCI but, in these simulations, are not well correlated with the likelihood of DCI.

How to cite: Houston, A., Shield, S., and Matousek, K.: The Role of CIN Depth in Regulating Deep Convection Initiation, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-196, https://doi.org/10.5194/ecss2025-196, 2025.

P63

ECSS2025-211

Nowcasting convective hazards in complex topography using machine learning

George Pacey, Ulrich Hamann, Ophélia Miralles, and Olivia Romppainen-Martius

Machine learning (ML) is currently revolutionising weather prediction at the short- to medium-range timescale. Nowcasting using ML has attracted less attention in comparison, especially for convective hazards such as lightning, hail and extreme precipitation. Conventional nowcasting approaches for convective hazards are typically based on Lagrangian extrapolation by advecting radar or satellite observational fields. Limitations of this approach include difficulties representing the intensification and decay of convective systems as well as identifying convective initiation during the forecast window. ML presents a potential avenue to make significant improvements to existing approaches by leveraging the large amount of historical data available from different sources (e.g. dual-pole radar, Meteosat satellites, lightning networks).

At MeteoSwiss, a probabilistic deep learning nowcast model for lightning, hail and precipitation is currently being operationalised for use in Switzerland. The model outperforms Lagrangian benchmarks when validating on one convective season. Furthermore, a seamless (nowcasting through to short- and medium-range) ML-based prediction system is envisaged for the coming years.

Here, we extend the current framework and explore where further gains may be possible by investigating advanced network architectures, novel input features and diversifying the training data. We focus on the Swiss radar domain, which presents unique challenges due to complex alpine topography. Nowcasts are generated at a 1 km and 5 minute spatial and temporal resolution, respectively, up to a lead time of one hour. Our work contributes towards the continued improvement of ML-based nowcast models, providing vital guidance and damage mitigation for sectors including emergency services, aviation and the public.

How to cite: Pacey, G., Hamann, U., Miralles, O., and Romppainen-Martius, O.: Nowcasting convective hazards in complex topography using machine learning, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-211, https://doi.org/10.5194/ecss2025-211, 2025.

P64

ECSS2025-226

Automated Severe Thunderstorm Oulooks from thundeR Package (ASTORP)

Mateusz Taszarek and Patryk Matczak

Construction of thunderstorm environmental datasets consisting of severe weather reports from 4 continents (Europe, Australia, North America, South America), global lightning detection data, SPC and ESTOFEX convective outlooks, and over 700 convective parameters derived from hybrid-sigma levels of global ERA5 reanalysis dataset allowed development of machine learning models aimed at predicting probability for the occurrence of non-severe, severe and significant severe convective storms. In the first step of model development, database was organized to choose the best environmental proxies for the identification of: (1) lightning occurrence given any environment, (2) severe hail occurrence given lightning, (3) severe tornado occurrence given lighting, (4) severe wind occurrence given lightning, (5) significant hail occurrence given severe hail, (6) significant tornado occurrence given severe tornado, and (7) significant wind occurrence given severe wind. In order to avoid biases towards certain geographical areas, the process of selecting best predictors for hail, tornadoes and wind have been performed separately for each continent and then best predictors contributing to final models were selected with the assumption that they should work on each evaluated domain. Among those parameters, a best combination (leading to highest skill) of 5-7 ingredients were selected for each model. In the second phase, models were used to produce historical convective outlooks consistent with SPC and ESTOFEX risk levels methodology for the period 2015-2023 and using ERA5 reanalysis. Based on those outlooks, a calibration process was performed to better fit modeled convective hazard risk probabilities. Since march 2025, a final calibrated models are used with operational GEFS for both Europe and the United States, producing convective outlooks 4x daily with a forecast up to 9 days. In this work we will present a methodology of constructing ASTORP models and show sample forecasts generated with those models and operational GEFS during spring and summer of 2025 for Europe and the United States.

How to cite: Taszarek, M. and Matczak, P.: Automated Severe Thunderstorm Oulooks from thundeR Package (ASTORP) , 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-226, https://doi.org/10.5194/ecss2025-226, 2025.

P65

ECSS2025-246

Localized heavy rainfall prediction using selective cloud-radar data assimilation based on automated cumulonimbus tracking

Ryohei Kato, Shingo Shimizu, Tadayasu Ohigashi, Takeshi Maesaka, Ken-ichi Shimose, and Koyuru Iwanami

Rapidly developing cumulonimbus clouds can cause localized heavy rainfall, leading to significant damage such as sudden rises in river water levels and even loss of human life. Therefore, there is a strong demand for earlier and highly accurate prediction methods. In this study, we developed a selective data assimilation method utilizing ground-based scanning type Ka-band radar (cloud radar), which can detect cloud droplets smaller than raindrops, to predict localized heavy rainfall from the cloud development stage before precipitation starts.

Previous research (Kato et al., 2022, WAF) successfully predicted localized heavy rainfall approximately 20 minutes ahead by assimilating high-frequency (every 1 minute) 3D special observational data obtained from simultaneous sector scanning by three cloud radars into a cloud-resolving numerical model (CReSS) with a horizontal grid spacing of 700 m. However, the assimilation method used (nudging-based humidification of cloud regions) assimilated not only developing clouds but also weakening clouds, resulting in unnecessary humidification and false precipitation forecasts.

To overcome this issue, we propose a new method that selectively assimilates only developing clouds. Specifically, we applied an automatic cumulonimbus cloud tracking algorithm (AITCC) to the cloud radar data, automatically extracting parameters such as cloud area and maximum reflectivity for each cell. By analyzing the temporal changes of these parameters, we could automatically distinguish between rapidly developing clouds and other clouds.

We verified the effectiveness of the proposed method through the special observational case and found that false precipitation forecasts observed without selective assimilation were suppressed. Consequently, the method successfully predicted true localized heavy rainfall accurately. In the future, we plan to further validate this method using multiple cases to achieve rapid and highly accurate predictions of localized heavy rainfall starting from the pre-precipitation stage.

How to cite: Kato, R., Shimizu, S., Ohigashi, T., Maesaka, T., Shimose, K., and Iwanami, K.: Localized heavy rainfall prediction using selective cloud-radar data assimilation based on automated cumulonimbus tracking, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-246, https://doi.org/10.5194/ecss2025-246, 2025.

P66

ECSS2025-248

Development of cell tracking method using multiple thresholds for obtaining accurate storm motion

Ken-ichi Shimose, Shingo Shimizu, and Ryohei Kato

One representative indicator of the probability of a supercell is the storm relative environmental helicity (SREH), which is calculated from the storm motion and the horizontal vorticity produced by the storm's environmental winds. Generally, storm motion and horizontal vorticity are calculated from soundings, objective analysis, and numerical forecasts; SREH is a very sensitive indicator to storm motion, so calculating SREH from more accurate storm motion will increase the possibility of determining whether a developing cumulonimbus cloud will become a supercell. In this study, more accurate SREH is calculated for each cumulonimbus cloud by using cell tracking technology to calculate accurate storm motions.

The cell tracking method used in this study is based on an automatic cumulonimbus cloud tracking algorithm (AITCC; Shimizu and Uyeda, 2012). Generally, when tracking a cell, a precipitation intensity or reflectivity intensity area exceeding a single threshold value is tracked as a cumulonimbus cloud. However, when a single threshold value is used, it is difficult to continuously track cells that merge or split or are detected at the very edge of the threshold value. Therefore, in this study, multiple threshold values (specifically, from 43 to 53 dBZ at 1 dBZ intervals) are used for cell tracking to develop a method to robustly track cells even when cell mergers or splits occur. Using the developed method, SREH for each cumulonimbus cloud can be calculated more accurately. In the future, we plan to use these methods to develop nowcasting methods for more accurate prediction of the probability of supercell occurrence for the next hour.

How to cite: Shimose, K., Shimizu, S., and Kato, R.: Development of cell tracking method using multiple thresholds for obtaining accurate storm motion, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-248, https://doi.org/10.5194/ecss2025-248, 2025.

P67

ECSS2025-257

Meteosat-12 at DWD: first experiences with novel FCI and LI products

Gerrit Holl, Alexander Halbig, Jochen Richters, and Christian Herold

Since November 2024, forecasters at Deutscher Wetterdienst (DWD) have started evaluation novel products from the Meteosat-12 Flexible Combined Imager (FCI) and Lightning Imager (LI) instruments. The strongly improved spatial and temporal resolution and the new spectral abilities present a major improvement for forecasters. During winter 2024/2025, evaluation focussed on the novel FCI RGBs Day Cloud Phase and Day Cloud Type, on a Total Moisture Imagery visualisation based on the ratio between the 0.91 µm and 0.86 µm channels, and on Geo Colour imagery, a combination of True Colour during the day and a night-time visualisation originally developed by the Cooperative Institute for Research in the Atmosphere (CIRA). Since May 2025, evaluation has focussed on the Day Cloud Phase RGB, the infrared Sandwich visualisation, the Fire Temperature RGB, and the experimental Flash Geometry product from the Lightning Imager. Forecasters report that in the Day Cloud Phase RGB and IR Sandwich can increase the storm forecasting lead time by up to ten minutes. Although there are still problems with the Flash Geometry product, it may detect lightning before the ground-based LINET system does. The presentation ends with an outlook to further products from LI as well as the Infrared Sounder (IRS) on MTG-S1 and rapid scanning opportunities with MTG-I2.

How to cite: Holl, G., Halbig, A., Richters, J., and Herold, C.: Meteosat-12 at DWD: first experiences with novel FCI and LI products, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-257, https://doi.org/10.5194/ecss2025-257, 2025.

P68

ECSS2025-264

Challenges of developing AR-CHaMo for severe convective wind gust forecasting

Tomas Pucik, Francesco Battaglioli, Pieter Groenemeijer, Andreea Bărăscu, and Mateusz Taszarek

ESSL is developing additive logistic regression models (AR-CHaMo) for forecasting large hail, tornadoes, and severe convective wind gusts. Large hail and severe convective wind gust forecasting models applied to the ECMWF ensemble were tested during the 2024 summer severe convective season. The wind model was developed using a subset of severe wind reports from the United States and Europe from all seasons. We found that the severe wind gust model was less skillful than the large hail model. The main issues were too low probabilities in high CAPE and low to moderate vertical wind shear, and too high probabilities in situations characterized by weak elevated instability and strong mean flow in the lower troposphere.

Using the above-mentioned issues, we outline the main challenges associated with the severe wind development. The first challenge is dealing with the various quality of severe wind reports related to the observed wind intensity in the European and the United States severe weather databases. The second challenge is a large spread of situations that can result in severe convective wind gusts: from disorganised storms producing downbursts, through longer-lived convective systems in high CAPE and moderate to strong shear, to strongly forced narrow convective rainbands forming in marginal CAPE and very strong mean tropospheric flow. We propose a way to deal with these challenges and how to improve the performance of the severe convective wind gust forecasting model. At the same time, we invite a discussion on the topic from the present severe weather research community.

How to cite: Pucik, T., Battaglioli, F., Groenemeijer, P., Bărăscu, A., and Taszarek, M.: Challenges of developing AR-CHaMo for severe convective wind gust forecasting, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-264, https://doi.org/10.5194/ecss2025-264, 2025.

Discussion Post a comment

P69

ECSS2025-267

Tornadic Event in Portugal in March 2024: Synoptic environment and Forecasting

Pedro M. Sousa and Paulo Pinto

On 28 March 2024, a significant convective weather event impacted Portugal, producing multiple reports of severe wind, including conformed tornadoes, one of them ocurring in the vicinity of Lisbon. This study provides a concise analysis of the synoptic and mesoscale conditions that led to the development of these phenomena. The environment was characterized by an unstable air mass, supported by favorable thermodynamic and dynamic parameters, such as elevated CAPE and storm-relative helicity, supporting the potential for tornadogenesis.

Deterministic and ensemble numerical weather prediction (NWP) products from ECMWF and AROME models were used to assess precipitation patterns and wind gust forecasts. While ensemble outputs suggested potential for localized high-impact weather, there was substantial uncertainty regarding the intensity and spatial distribution of precipitation and maximum wind gusts. Mesoscale forecast products further supported the likelihood of severe
weather, as predicted by forecasters in advance.

We also present insights into the operational forecasting process at the time, such as weather bulletins and experimental nowcasting guidance for convective storms. Observations confirmed the occurrence of tornadoes in the Tagus estuary (close to Lisbon) and in Benaciate (Algarve), along with intense convective wind gusts in other locations. These observations aligned with the earlier model indications and operational forecasts, validating the usefulness of high-resolution ensemble forecasting and real-time convective diagnostics.

In summary, and from an operational weather forecasting perspective, this case study illustrates the combined value of synoptic analysis, ensemble prediction systems, and convective nowcasting tools for early detection and short-term forecasting of tornadic events in Portugal. It also emphasizes the critical importance of integrating deterministic forecasts with probabilistic guidance and observational confirmation, particularly in the context of high-
impact but low-frequency events such as tornadoes.

How to cite: Sousa, P. M. and Pinto, P.: Tornadic Event in Portugal in March 2024: Synoptic environment and Forecasting, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-267, https://doi.org/10.5194/ecss2025-267, 2025.

P70

ECSS2025-278

Cell Mergers, Boundary Interactions, and Convective Systems in Cases of Significant Tornadoes and Hail

(withdrawn)

Cameron Nixon, John Allen, Matthew Wilson, Matthew Bunkers, and Mateusz Taszarek

P71

ECSS2025-281

Precipitation classification functions for Northwest Bulgaria: GNSS IWV and Instability Indices

Martin Slavchev, Guergana Guerova, and Tsvetelina Dimitrova

Use of Global Navigation Satellite Systems (GNSS) tropospheric products to derive water vapour is a well established technique for atmospheric monitoring in Europe (GNSS meteorology). GNSS meteorology establishment across Europe was achieved within COST Action ES1206 ”Advanced Global Navigation Satellite Systems tropospheric products for monitoring severe weather events and climate” (GNSS4SWEC). GNSS4SWEC facilitated application of GNSS tropospheric products for severe weather forecasting and nowcasting. Precipitation monitoring during the convective storm season May-September is conducted routinely by the Bulgarian Hail Suppression Agency (HSA). From 2020, in Northwest Bulgaria 4 GNSS stations are processed in near-real time mode and vertically Integrated Water Vapour (IWV) is computed in an operational manner. As a part of a storm nowcasting demonstrator GNSS IWV and Instability Indices (InI) thresholds are implemented for Sofia and Central Bulgaria. In this work site-specific classification functions are computed for Northwest Bulgaria. A GNSS derived monthly IWV threshold separates well precipitation (P) and no precipitation (nP) groups in July, August and September. Probability of detection is between 77-100 % for July and 78-85 % for August. For July the false alarm ratio scores are high in the range 30-66 %, which limits the use of IWV. Classification functions based on InI and IWV have the best performance with an increase of probability of detection score by 16 % in July, 23 % in August, and 20 % in September and decrease of false alarm ratio score by 23 % in July, 22 % in August and 8 % in September.

How to cite: Slavchev, M., Guerova, G., and Dimitrova, T.: Precipitation classification functions for Northwest Bulgaria: GNSS IWV and Instability Indices, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-281, https://doi.org/10.5194/ecss2025-281, 2025.

P72

ECSS2025-308

Advancements in Automated Convective Cell Detection and Nowcasting at Deutscher Wetterdienst (DWD)

Manuel Werner, Lukas Josipovic, Robert Feger, Christian Berndt, and Cornelia Strube

At DWD, a semi-automated process is currently used to warn the public about thunderstorm-related hazards such as large hail, strong wind gusts, and heavy rainfall. Human forecasters are supported by algorithms and meteorological products that integrate information from various data sources.

A key component of this infrastructure is KONRAD3D, a tool designed for the detection, tracking, and nowcasting of convective cells, particularly thunderstorms. KONRAD3D uses three-dimensional, quality-controlled radar reflectivity data from DWD’s radar network as its primary data source. It generates warning indicators for hail, heavy rainfall, and wind gust threats.

In addition, the algorithm incorporates various supplementary data sources, including lightning data (LINET), DWD’s dual-polarization hydrometeor classification (to assess hail potential), grid-based, range-gauge-adjusted quantitative precipitation estimates (QPE) for evaluating heavy rainfall, and DWD’s mesocyclone detection for supporting wind gust warnings.

Furthermore, numerical weather prediction (NWP) data are used to derive the most unstable vertical trajectories, from which meteorological quantities such as CAPE, CIN, vertical wind shear, and helicities are calculated. Cell-based vertically integrated ice (VII) and VII density are also now provided. Finally, cloud top height information derived from satellite data has been integrated.

An evaluation of KONRAD3D’s cell detections and nowcasts against lightning observations has been conducted. However, assessing false alarms is non-trivial, as cell detections without lightning are permissible—early detection is a key objective. Therefore, we analyze cell tracks rather than individual cells in order to identify very short-lived, non-lightning tracks as potential false alarms.

This contribution presents the basic functionality and recent enhancements of KONRAD3D, outlines its integration into DWD’s warning infrastructure, and summarizes the statistical results that assess its performance.

How to cite: Werner, M., Josipovic, L., Feger, R., Berndt, C., and Strube, C.: Advancements in Automated Convective Cell Detection and Nowcasting at Deutscher Wetterdienst (DWD), 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-308, https://doi.org/10.5194/ecss2025-308, 2025.

P73

ECSS2025-315

Nowcasting tropical rainfall events using Commercial Microwave Links

Bas Walraven, Ruben Imhoff, Aart Overeem, Miriam Coenders, Rolf Hut, Luuk van der Valk, and Remko Uijlenhoet

To mitigate the impact of severe storms, accurate and timely high-resolution precipitation forecasts are crucial. In the tropics, however, many low- and middle-income countries typically lack the near surface rainfall sensors to provide such a nowcast. Weather radars are largely unavailable, and rain gauge networks are often sparse or poorly maintained, and not available in (near) real-time. Satellite precipitation products do provide valuable precipitation information in these regions, but often come with the drawback of having a spatial or temporal resolution too low to nowcast convective storms at the kilometer scale.

A viable and ‘opportunistic’ source of near-surface high-resolution space-time rainfall estimates is based on the rain-induced signal attenuation experienced by commercial microwave links (CMLs) in cellular communication networks. This idea exploits the fact that the EM waves travelling between two antennas on different cell phone towers are scattered and absorbed by raindrops causing the strength of the signal to be partially attenuated. Using the open source algorithm RAINLINK we convert this specific attenuation to path-averaged rainfall intensities, essentially creating a network of virtual rain gauges which we then interpolate to create 2D rainfall maps, every 15 minutes.

In this work we investigate the potential to use these CML derived rainfall maps as input into a conventional nowcasting algorithm, pySTEPS. The analysis is based on a CML network from Sri Lanka. The data set spans 15 months across 2019 and 2020. For each of the four monsoon seasons represented in the data set we define extreme events of different duration, ranging from 1 to 24 hours. These events are used as input to create probabilistic nowcasts in pySTEPS for lead times up to three hours. The nowcasts are evaluated spatially against the QPE at multiple catchments, and using 21 hourly rain gauges as an independent point reference source.

Based on our findings we point out the opportunities and limitations of using CML data for nowcasting tropical storms. Finally, we highlight where CMLs can complement other remotely sensed rainfall estimates, for example from geostationary satellites, to provide more accurate nowcasts and as such potentially have impact in an operational setting too.

How to cite: Walraven, B., Imhoff, R., Overeem, A., Coenders, M., Hut, R., van der Valk, L., and Uijlenhoet, R.: Nowcasting tropical rainfall events using Commercial Microwave Links, 12th European Conference on Severe Storms, Utrecht, The Netherlands, 17–21 Nov 2025, ECSS2025-315, https://doi.org/10.5194/ecss2025-315, 2025.

Discussion Post a comment