Automated analysis of radio and auroral data sets to understand substorm-driven energy deposition into the polar atmosphere

Space Weather describes environmental conditions in space that can have an impact on Earth, and severe Space Weather is listed as one of the highest priority natural hazards in the UK Government’s National Risk Register. Conditions in Earth’s magnetosphere are constantly changing over the substorm cycle: a build-up and explosive release of magnetic flux and energy from the interaction between the solar wind and the planetary field. At substorm onset, enormous amounts of energy are released, and much of this energy makes its way toward the polar regions, dissipating into the upper atmosphere.

We now have access to long time series of Auroral Kilometric Radiation (AKR) – measurements of Earth’s radio emissions which respond to substorms and serve as an excellent remote proxy of this cycle of energy build-up and deposition. Preliminary work in Southampton’s Space Environment Physics group has involved the development of simple algorithms to pick out the main features in the AKR spectrum. This proposed project will require data intensive methods to analyse more than a decade of satellite and ground-based auroral camera data, employing machine learning algorithms to automate the search for specific features which are indicative of dramatic Space Weather events impacting Earth’s upper atmosphere.

This project will involve handling large datasets from spacecraft in orbit around Earth which have been monitoring the radio emissions quasi-continuously since the 1990s. These datasets will be systematically probed for changes in power, polarization and frequency which are all characteristic signatures of this energy cycling process. In addition, we have a huge catalogue of ground-based auroral camera data over more than a decade. The student will apply automated feature identification methods to this data stream to compare and contrast with the radio emission response to Space Weather.

The student will build domain knowledge of radio emissions from examination of classic case studies which show the response to substorm onset. This will then be used to build machine learning training libraries. Important challenges include exploring how radio signatures alter as they propagate through the ionosphere, and monitoring by multiple spacecraft in different positions will be critical here (with research results directly relevant to the satellite communications sector). The main focus and deliverable at the end of the project will be sophisticated automated algorithms to select relevant features from large catalogues of space-based radio and ground-based auroral data.

The SPITFIRE DTP programme provides comprehensive personal and professional development training alongside extensive opportunities for students to expand their multi-disciplinary outlook through interactions with a wide network of academic, research and industrial/policy partners. The student will be registered at the University of Southampton and hosted at the Department of Physics and Astronomy. Specific training will include:

At the start of their PhD, the student will attend a national summer school in solar-terrestrial physics. An advanced summer school is also available in the second year. Further opportunities for knowledge development will be provided through the postgraduate lecture series and research seminar series in Physics & Astronomy. The student will have access to the training programme for DISCNet, a Southampton-led Data-Intensive Science Initiative which covers machine learning and big data. Development of programming skills will be a key element of the training provided, as computer programming will be integral to the data analysis undertaken