Skip to main content
European Commission logo
English English
CORDIS - EU research results
CORDIS
CORDIS Web 30th anniversary CORDIS Web 30th anniversary

Flare Likelihood and Region Eruption Forecasting

Article Category

Article available in the following languages:

Solar flare forecasting service to be launched shortly

Humanity is becoming ever more vulnerable to adverse space weather conditions due to our increasing reliance on networked space-borne technologies. Losing one or more of the network nodes even for a short period of time will have major repercussions and cost billions of euro.

Digital Economy icon Digital Economy
Space icon Space

Adverse space weather results from solar flares and coronal mass ejections released from the turbulent and highly complex magnetic fields of active regions of the Sun. Understanding how active region magnetic fields evolve and behave will enable scientists to develop accurate and reliable space-weather monitoring and forecasting capabilities. The EU-funded FLARECAST initiative studied the drivers behind the triggering of solar flares to improve flare prediction through the application of physics, state-of-the-art mathematics, statistics, Big Data and machine learning. The initiative is an example of a research-to-operations project, using methodologies taken from textbooks and scientific articles to create arguably the most systematic flare prediction service worldwide. Advanced image-processing techniques were employed to determine the properties of solar active regions. These included area, magnetic flux, shear, magnetic complexity, helicity and proxies for magnetic energy from solar magnetogram and white light images in near-real time. The team correlated the results with solar flare activity and optimised prediction algorithms via statistical, unsupervised clustering and supervised learning methods. “This enabled researchers to validate image processing and flare prediction algorithms before launching a near real-time flare forecasting service,” says project coordinator Manolis Georgoulis. A range of disciplines utilised The consortium employed open-source Docker engine technology as a breadboard to facilitate the project’s infrastructure on a highly modular ensemble of Docker containers. “Big Data handling and machine learning showed that predicting solar flares is not and should not be just about heliophysics,” explains Georgoulis. “A combination of expertise from the mathematics, statistics, computer science and artificial intelligence communities is required to make a breakthrough in this area.” The project generated three databases with the FLARECAST native and external data amounting to a mind-boggling 240 terabytes. This indispensable collection of data will help to support many future research efforts. Georgoulis states: “Currently about 15 peer-reviewed papers highlight different aspects of the project, such as new and promising predictors, machine-learning algorithm performance, the relation or connection between flares and solar coronal mass ejections and other findings. More publications focused entirely on the project are planned for the near future.” Science supported around the world FLARECAST therefore forms the basis of a quantitative and autonomous active region monitoring and flare forecasting system, which will be of use to space weather researchers and forecasters, both in Europe and around the globe. They include scientists working in the field of solar physics and heliophysics who will use its results and databases to advance the understanding of the physics behind solar eruptions and future prediction efforts. The project will also help the machine learning and Big Data communities hone their skills and devise new methods, like the “hybrid” and “innovative” machine learning techniques developed during the project. The modularity and open-access nature of the FLARECAST infrastructure will enable other research teams to expand on and add more information (such as on flares, coronal mass ejections, and solar energetic particles) to an integrated space weather forecasting facility, thereby avoiding duplication of effort. “As a consortium, we have already seen preliminary expressions of interest from Europe and beyond, either for using the results of the service or for migrating the entire facility to their premises,” points out Georgoulis.

Keywords

FLARECAST, solar flares, space weather, machine learning, Big Data

Discover other articles in the same domain of application