The Sun can be very active at times, resulting in very powerful ejections of highly energetic elementary particles that could have dramatic consequences for Earth: power grid failure, circuit burnout, disruption of radio communication and GPS, civil aviation etc. which can bring chaos in our day-to-day life. Early detection and prediction of such events in advance can help brace for the incoming impacts at Earth’s orbit. Remote observation of “space weather” allows for preventative action to be taken during the passage of the burst on Earth, e.g., by allowing vital electric systems to be shut down to protect them.

The last generation of radio telescopes (as well as space-borne probes) allow us to detect and quantify incoming solar bursts through their signature emitted as radio waves, as soon as they are emitted from the Sun. The remote signal measured from solar bursts can be turned into dynamic movies showing the evolving structure with time (around the Sun) and with frequency (as a sound spectrogram). This information allows researchers to evaluate potential  threats with regard to solar of burst size, energy and trajectory crossing Earth orbit.

Photo Credit: NASA

 The EXTRACT project is validating its extreme data mining workflow in the TASKA use case. This case aims to develop the different aspects and technical solutions for building a fast-response system that will identify the presence and morphology of the bursts using trained Machine Learning/Deep learning (ML/DL) networks deployed on cloud architectures.

• How can we efficiently detect and model the various solar radio emissions from observation with radiotelescopes?  

• How can we develop an automatic data processing and identification system that will enable quick response time to derive burst parameters and location from near real-time?

• How can ML/DL in a Cloud-based framework contribute to the deployment of such alert systems?