Basics of a Solar Flare Forecast

In a previous post I described the concept of space weather. Whilst monitoring current conditions around Earth, space weather forecasters will produce forecasts of the likelihood of solar eruptions occurring over the next few days. Solar flare forecasts are just one part of these daily guidance documents. To create a forecast, a few simple guidelines are generally followed, which I will outline in this post.

1. Setting the scene.

For any forecast, be it Earth weather or space weather, the forecaster needs to begin by describing what is happening now. They start by examining solar imagery, such as the MDI magnetograms used in Sunspotter. They will identify any features in these magnetograms which are already having, or are likely to have, an impact on space weather conditions in the coming days. See Figure 1 for an example of identifying active regions using magnetograms. In this particular example, regions have been numbered according to the NOAA Space Weather Prediction Centre sunspot numbering scheme – when a new region emerges onto the solar disk it is given a number in order. The Solar Monitor Active Region Tracker method outlined in a previous post is another way to identify regions of interest, which automatically produced the data sets you see in Sunspotter.

SDO/HMI magnetogram from 2015 August 22, source:

Figure 1: SDO/HMI magnetogram from 2015 August 22. NOAA active region 12403 might be of particular interest to a forecaster, with a ‘beta-gamma-delta’ classification. Figure source:

The forecaster will examine the history of these identified regions of interest– if something significant has happened recently this might get described in the guidance document, perhaps with images to help the viewer understand what happened and why.

2. What’s likely to happen (the forecast)?

Once the current situation has been described, it is time to move on to what is likely to happen over the period of the forecast. This can be separated into two parts:

2.1 The next 24 hours.

The forecaster will describe in the guidance document how any identified regions are likely to move or develop in the immediate future. For example is a sunspot region growing? Is it complex? Is it likely to produce flares? This is where your classifications in Sunspotter will help define how complex a region is! Once forecasters have decided on a classification, they will calculate the probability of a flare occurring in this region over the next 24 hours.

There are many ways to do this, including Bayesian methods (e.g., Wheatland et al), machine learning (e.g, Qahwaji et al), discriminant analysis (e.g., Barnes et al), and many more. A relatively simple statistical method is often used in operational flare forecasting, which starts with a large database of flare information from previous solar cycles. This database shows how many particular classes of flares each classification of active region produced in that time period. From this an average flare rate can be calculated, and using Poisson statistics a percentage probability of flare occurrence for the next 24 hours will be obtained for each region. See Bloomfield et al, 2012 for a more in-depth description of this method. A little human intervention is also involved here – if a forecaster thinks the value is too small or too big they can change it based on their experience! The percentages for each region can then be added to obtain the probability of a flare occurring across the entire solar disk over the next 24 hours.

2.2 The rest of the forecast period.

The forecaster will then take a briefer look at the whole sun over the next few days. If the next 24 hours look fairly quiet, but something interesting is expected to return to the solar disk in a couple of days time, then the probability of a flare occurring might be increased later on in the forecast period. Similarly if a particularly complex region is due to leave the disk in a few days, the probability might be decreased for that day. An example of a typical flare forecast is shown in Figure 2.

Example of a typical flare forecast for the next four days. Percentage probabilities of M- and X- class flares are included, as well as a description of current conditions.

Figure 2: Example of a solar flare forecast for the next four days. Percentage probabilities are calculated for two ‘levels’ that are related to M- and X- class flares. These are the two largest classes of solar flare, the ones most likely to produce radio blackouts. The forecast also mentions whether any M- or X- class flares occurred over the previous day, and a general description of current conditions. Figure source: Met Office (Crown Copyright)


3. Potential Earth impacts.

Once the forecaster has described what we expect to happen on the Sun, next it’s time to explain how this may affect life on (or in the vicinity of) Earth. For example can we expect radio blackouts? Do astronauts onboard the International Space Station need to postpone any space walks? This will be summarised as part of more general guidance documents, which include forecasts of other space weather phenomena such as coronal mass ejections.


These are the typical steps taken to create a flare forecast. However every day is different, and forecasters can diverge from this method if necessary! Check out current space weather conditions on, e.g., the SWPC and Met Office webpages.


Thanks to Senior Operational Meteorologist, Mark Sidaway, for his guidance when creating this blog post. All statements in this post are my own and not those of the Met Office.


Tags: , ,

About Dr Sophie A. Murray

Space Weather Research Scientist

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: