The Space Weather Prediction Effort at GSU: Many Projects in Parallel Within a Unified Architecture

The Georgia State University Astro-informatics Cluster is a strongly
integrated interdisciplinary research group between the departments
of Computer Science and Physics & Astronomy. Since its foundation in
2014 the cluster has been solely focused on space weather prediction.

Our current research program consists of over a dozen projects, each
carried out in parallel by faculty and their graduate and undergraduate
students. Each project forms a building block for a comprehensive and
unique space weather data mining and forecasting using cutting edge
machine learning and modeling approach. Several of these projects are
presented at this meeting in more detail here in posters by our graduate
students.

In my contribution I will present the overall design and architecture
of our prediction system, the place of each individual effort within it,
and describe future projects to complete the system.

A partial list of efforts underway or completed is:
- Thoroughly verified, complete, connected, and machine learning ready
data bases of flares, CME's, and SEP's (benchmarks), made publicly
available
- A Systematic search for flare precursors in magnetic and spectral
data, including use of GSU's South Pole Observatory data, polarity
inversion line morphology, and GONG H-alpha filament observations
- A computer science study of the machine learning method and data
requirements for optimal space weather prediction
- Mitigation of solar limb limitation of the quality of magnetic field
data for machine learning
- Integration of SoHO EIT and MDI data labels with those from HMI and
AIA on SDO
- Flare prediction from time series data rather than static snapshots
- Constructing synthetic time series data for improving forecasting
- Simulated real-time predictions

This is merely a sample of projects going on. None of this work would
be possible without the dedication and persistence of our standing army
of students.