How do you predict thunderstorms?

share with:

The physical processes that cause thunderstorms are complex: in addition to atmospheric instability, moisture conditions that are influenced by the nature of the soil, the type of vegetation, or the relief profile must also be taken into account. Digital models used in weather forecasting are progressing, and spatial resolutions have been greatly improved.
However, forecasting is still often difficult, even though "Nowcasting" makes it possible to effectively anticipate a phenomenon within an hour.
It is mainly this method that we will describe in this article.

1) In order to predict thunderstorms, observations have to be made

Observation is the fundamental basis and the quality of the observation is essential in order not to introduce errors in the predictive model.
Predicting a thunderstorm for tomorrow, for example, means identifying the conditions conducive to convection: Unless there is little doubt that all the conditions are present, we will speak more commonly of a storm risk forecast, especially beyond a few hours and over a large area (for example the scale of a département in France).
In the case of Nowcasting, we will mainly use observation data of storm phenomenon, using satellite, radar or lightning detection networks.

The information from lightning detection networks is the only information available in true real time since lightning is detected "as it happens", generally only ten seconds after it actually occurs, unlike satellite and radar images which are sequenced (generally every 5 to 15 minutes).
It is for this reason in particular, but also because of their excellent detection efficiency1 that the use of "LLS" (Lightning Locating Systems) is generally the best alternative for "Nowcasting storms".
This method currently provides the best results in terms of anticipation for a time period of less than one hour. . Beyond this time frame, it is combined with weather forecasting, as observation of the phenomenon no longer makes it possible to estimate its movement as effectively.

2) How do you anticipate the movement of thunderstorms?

In order to understand how one can anticipate a phenomenon by simply observing it, it is important to realise that there are many types of thunderstorms. In terms of schematics, we can differentiate them by placing them into 2 very different categories:

Frontal storms

These are large-scale situations that can affect several hundred kilometres, and are most frequently encountered in our latitudes.
They often follow a south-north direction and their fairly linear configuration makes it possible to anticipate their movement.

Example : the presence of several storm fronts in Europe on 20/08/2019

storm fronts in Europe

Convective storms

These are isolated situations, such as orographic storms, which often form very quickly as a result of topography.
Anticipating them is more tricky than in the case of frontal storms, but they often affect particular areas such as the Alps or the Pyrenees Mountains.

Example : isolated thunderstorm in the Verdon (France) on 27/05/2019

Convective storms

Contrary to popular belief, few thunderstorms are sudden and unpredictable. One of our recent accidentology studies proves this and reports about 5% of cases over most of Europe2.

Specific example

Exemple orage

3) Can we also anticipate their severity?

By definition, any thunderstorm is dangerous and it is advisable to systematically get to a safe place. Nevertheless, some thunderstorm episodes can be particularly severe, and so it is essential to be able to predict them. One of the anticipation methods that has been developed very recently, consists in identifying a sudden increase in the number of lightning strikes, in particular intra-cloud activity.

This phenomenon, known as “lightning jump” is a proven precursor to many phenomena associated with storms, such as hail or strong gusts of wind below the thunderstorms3. Numerous studies exist on the subject, and 2 of our experts, Stéphane Pedeboy and Paul Barnéoud, have been working for several years on this very topic in order to improve the methods of storm prevention that are currently being used.

Stéphane Schmitt
File written by Stéphane S. He is our expert and trainer in storm risks.
He is (hyper)active in many organizations to reinforce awareness actions regarding storm risk. He contributes in the normative field and through numerous studies published during international conferences.

>> Would you like to learn even more about the topic? <<
Email the author of this blog at:

1Read for example this publication from 2020, which mentions that a satellite detects about half of the lightning located by high-precision lightning detection networks
2Read the publication here
3Read for example this publication
4Read for example this publication