To contact ESTOFEX you can send an e-mail to .
You are welcome to put a link to the ESTOFEX web site on your site.
We do not mind if you load our forecast map directly on your
site, provided that you add a link to ESTOFEX and add a notice that the
map is created by the ESTOFEX team, like "Forecast provided by ESTOFEX".
In order to add the most recent forecast map to your site, use this link:
or the latest XML file:
Any other link may not work. We reserve the right to discontinue the permission to deep-link to the map in the future. Please be aware that the forecasts are published under a Creative Commons Attribution-Noncommercial-Share Alike 3.0 license, and that usage not complying with this license is illegal.
The European Storm Forecast Experiment is an initiative of a team of European meteorologists, and students in meteorology, and serves as a platform for exchange of knowledge about forecasting severe convective storms in Europe and elsewhere.
Our goals are:
to issue daily forecasts of severe convective weather in Europe, using an ingedients-based forecasting methodology
to enhance the understanding of severe convective storms in Europe, both among ourselves and others
to promote and demonstrate the use of the European Severe Weather Database (ESWD) as a basis for severe weather research
to verify the forecasts, based on ESWD and lightning data
ESTOFEX issues "Storm Forecasts" on a daily basis. These bulletins, accompanied by a map, address the threats posed by severe convective storms in Europe. The focus of the forecasts is the threat for hail, severe wind gusts, and tornadoes that these storms pose. Since May 1st 2009, the threat of excessive convective rainfall is included. Additionally, areas where lightning is expected are highlighted.
To communicate the magnitude of the threat of hail, severe wind gusts, tornadoes, and excessive precipitation, ESTOFEX uses threat levels. Threat levels are issued based on the estimated probability of severe weather resulting from convective storms. A storm is considered severe or extremely severe when it is accompanied by one or several phenomena that meet the criteria listed below.
severe convective weather phenomena:
hail with a diameter of at least 2.0 cm
wind gusts with a speed of at least 25 m/s (92 km/h or about 48.6 knots)
excessive rainfall of at least 60 mm (explained elsewhere in this FAQ)
extremely severe convective weather phenomena:
hail with a diameter of at least 5.0 cm
wind gusts with a speed of at least 33 m/s (about 119 km/h or 65 knots), or
a tornado of class F2 or stronger
The threat levels are expressed as a range of probability that severe weather occurs within a radius of 40 km from a location. For example, a probability of 10% in your location means that in one out of ten times such forecast is issued, severe weather would occur close to you. For the size of a typical threat area, for example the size of the northern half of Germany, this could correspond with 3-4 reports of severe weather (not counting very closely spaced reports). We obtained the probability values by conducting a verification of the threat level areas with the severe weather reports entered into the European Severe Weather Database (ESWD). For now, only regions of Europe with a high reporting rate were used. For every point on the map (see verification page) the occurrence of severe and extreme convective weather was counted, and compared to number of times a certain threat level was issued. This resulted in an average frequency that severe weather occurred for each threat level: the probability of severe weather to occur when it was forecast. With these results, we started to label the lines with the probabilities that represent the threat level boundaries.
This level is valid for the area that does not fall under level 1, 2, or 3, implying that the expected probability of severe convective weather appears insignificant. There is 0-5% probability that a severe convective storm will occur within 40 km radius of a location. By the definition of probability isolines, the probability is just below 5% in the direct vicinity of level 1 contours.
Threat level 1 corresponds to a probability of 5% to 15% that a severe convective storm will occur within 40 km radius of a location. This is the most common threat level and is used when a low threat of severe weather has been identified.
Threat level 2 corresponds to a probability greater than 15% that a severe convective storm will occur within 40 km radius of a location. This is a significant threat that expresses larger confidence in the occurrence of severe storms. While extremely severe weather phenomena are not part of the criterion, there is a slightly increased probability for these as well (3-5%).
Threat level 3 is issued when there is a significant threat (probability greater than 15%) of extremely severe convective storms to occur within 40 km radius of a location. Level 3 is rarely issued and implies that a major severe weather outbreak is expected. Examples include the derecho (a thunderstorm system producing widespread extreme wind gusts) that struck a large part of Germany on July 10th 2002. Major tornado outbreaks also require a level 3. An example is the outbreak that occurred on June 25th 1967 across France, Belgium and the Netherlands, and the more recent case of tornadoes and large hail across Poland on August 15th 2008.
Apart from the threat areas, the maps show two yellow lines: the areas where lightning is forecast, separating very low | medium | high probability of lightning to occur within 40 km radius of a location on the map.
For more information about severe storms see the research and education page.
Excessive precipitation is statistically the type of severe weather with the most impact. The resulting flash floods, landslides and river floods can cause many casualties and extensive damage. However, since ESTOFEX does not currently include any hydrologists and because local surface information is not available, we only forecast excessive precipitation. For verification purposes, we are forced to use a criterion that progressively excludes higher rain sums over longer reporting periods from being considered severe, in an attempt to exclude the majority of non-convective events.
The threshold is chosen so that in most types of terrain flash flooding is indeed likely, and this threshold will not be reached by every thunderstorm. Instead, we focus on forecasting special situations in which precipitation accumulates by thunderstorms which are slow moving in a moist environment, backbuilding or constantly regenerating, or forced by upslope flow in mountainous terrain. Our aim is to forecast events occurring under convective conditions, not the entirely stratiform events (as we also do with wind gusts). Because of this, it is possible that lightning reports will have to be considered alongside reports during verification.
It was chosen not do this, because of limitations of our available time, and our wish to focus on deep organized convection. In addition this would require knowledge of local details in non-convective climatology at the scale of entire Europe (please consult your regional weather service). We are well aware that this selective approach creates problems in marginally convective/non-convective situations. In case of a storm depression with severe widespread winds, we issue threat levels when the forecast convective organization appears to be capable of enhancing the already severe "background" weather threat, for example by bow echoes and supercells. In such cases it is particularly recommended to read the forecast text, which conveys the total threat better than the choice of threat levels.
When considering numerical model output and observational data, such as soundings, we use an "ingredients-based" methodology:
Several physical parameters need to come together in order to produce a storm and severe weather. Low-level moisture and steep lapse rates combine into latent instability. Instability needs a trigger in order to be released and generate a storm: deep lifting. This can be generated by various mechanisms at different scales that also have an influence on the coverage of storms. Once a storm forms, its development is strongly influenced by the interaction with the flow in its environment. Depending on the direction and magnitude of the winds at various altitudes (among other factors) the storm may grow into several types like multicell clusters, squall lines, and supercell storms. In the forecast, we use several parameters that have been proven to have a predictive value of storm types. Upon having gained a thorough picture of the factors that influence the organization of the storms on a particular day, we determine the likelihood of severe weather based on the characteristics of a particular storm environment as well as the expected storm type and coverage, and translate that conceptual picture into threat levels.
We have been running a verification of the lightning forecasts since 2006, see the Verification section for detailed information. Also, first verification efforts of the severe-weather forecasts have been realized in collaboration with the University of Oklahoma (see the literature page). While there is a lack of severe weather reports reaching us, a verification of the threat level areas has been started. Some results are presented in the new description of the threat levels. The probability of severe weather events indeed increases according to the forecast threat level. The probability of detection, based on reported events, inside at least a level 1 is: gusts: 57%, significant gusts: 81%, hail: 60%, significant hail: 84%, tornadoes: 36%, significant tornadoes: 67%. This shows that extremely severe events are less likely to be missed, but note that around 80% of extremely severe events did not fall inside the higher threat levels (2 and 3) which are considered most appropriate. There is room for improvement. Results were presented at the European Conference on Severe Storms in October 2009. The quality of data is crucial for a more reliable verification. Please help us and other scientists: submit to the ESWD!
We appreciate your help! We are in great need of reports of severe weather to verify the forecasts. We cannot improve our forecasts if we don't know where severe weather has happened. So, submitting your severe weather report to the ESWD database will certainly help. Besides reports, we welcome more observational data, such as real-time radar, lightning and analysis tools. And it is now possible to support ESTOFEX by a donation.
ESTOFEX is looking for persons who:
are willing to contribute regularly in discussions
have experience in forecasting and/or research of thunderstorms and severe weather
had a preferably academic education in meteorology or physics
have qualities e.g. in programming, numerical modeling or statistics to advance our goals
If you feel suited to join us, please introduce yourself by sending an email to the above address. We especially welcome applications from people from southern Europe, Scandinavia and Eastern Europe.
No. We are not paid. In order to recover the costs of the web server and travel costs to Estofex meetings, we would welcome sponsoring of the experiment.
This is a misunderstanding. ESTOFEX does not issue warnings. At least not in the sense that the general public or professional sectors are advised to take action. In most countries, national meteorological institutes issue such warnings, while commercial companies may offer similar products. Please check the forecasts and warning of those organizations.
Their use is to raise awareness and to provide real-time education about severe weather forecasting. Severe storms occur almost every day somewhere in Europe, much more frequently than at a national level. Because of this, ESTOFEX can gain experience faster, and our forecasts are free of the pressure involved with the responsibility for warnings and communication to clients.
The forecasts produced by ESTOFEX are valid for an entire day or a large part of a day and cover large areas. Although certain non-meteorological users may find the information useful, our primary targets are meteorologically educated people. Therefore, the forecast texts are technical in nature. Similar to the Storm Prediction Center in the United States we can provide guidance to other meteorologists. Our forecasts may be one of the data sources for a meteorologist on duty to consult. It must be noted, however, that we are an 'experiment' and therefore cannot take any responsibility with respect to the quality or availability of the products.
ESTOFEX forecasters have given presentations and workshops on severe storm forecasting on several occasions, for example at the weather services of Czechia (CHMI), Germany (DWD), Austria (AustroControl), Finland (FMI), and at media companies (RTL and ORF television). If you are interested in a contribution by one or more of our forecasters, please contact us at , or contact the forecaster personally.
The forecast area has been chosen to be a contiguous geographic area covering a central part of Europe: the area from which we expected and received the most feedback initially. The choice was made without regard to political or tectonical reasons of what is "Europe". Its extent is a balance between the wish to be complete and cover the entire European continent including all its extremities, and the fact that doing so greatly increases the time and effort needed to produce a forecast. For example, including all parts of European Russia would increase the surface area by a factor 1.4, which is currently too big an expansion for one forecaster on duty to cope with. At this moment we will not expand in that direction. However, the ESTOFEX forecast area is not cast in stone and we may decide to include those areas when we are able to. For similar reasons, parts of southwestern Europe (Madeira, the Canary Islands) and northwestern Europe, like Iceland and the Far Oer, have been excluded, even though severe storms have been reported in those locations also.
Bologna Limited Area Model
cold air advection
convective available potential energy
cylonic vorticity advection
differential cyclonic vorticity advection
DLS or DL shear
Deep-layer shear: 0-6 km difference vector length
Global numerical model from European Centre for Medium-Range Weather Forecasts
elevated mixed layer
Global model of the Deutscher Wetterdienst (DWD)
density-weighted integral of CAPE
HIgh Resolution Local Area Model run at the Agencia Estadal de Meteorología
HIgh Resolution Local Area Model run at Dutch Royal Met. Inst. (KNMI)
lifted condensation level
level of free convection
LLS or LL shear
Low-level shear: 0-1 km difference vector length
local model of the Deutscher Wetterdienst (DWD)
mesoscale convective system
Mesoscale and Microscale Modeling System Version 5, NCAR
most unstable CAPE
most unstable lifted index
SRH or SREH
storm-relative (environmental) helicity
equivalent potential temperature
Unified model of U.K. Met. Office
upward vertical motion