FlareNet : A Deep Learning Framework for Solar Phenomena Prediction

Solar activity can interfere with the normal operation of GPS satellites, the power grid, and space operations, but inadequate predictive models mean we have little warning for the arrival of newly disruptive solar activity. Petabytes of data collected from satellite instruments aboard the Solar Dynamics Observatory (SDO) provide a high-cadence, high-resolution, and many-channeled dataset of solar phenomena. Several challenging deep learning problems may be derived from the data, including space weather forecasting (i.e., solar flares, solar energetic particles, and coronal mass ejections). This work introduces a software framework, FlareNet, for experimentation within these problems. FlareNet includes components for the downloading and management of SDO data, visualization, and rapid experimentation. The system architecture is built to enable collaboration between heliophysicists and machine learning researchers on the topics of image regression, image classification, and image segmentation. We specifically highlight the problem of solar flare prediction and offer insights from preliminary experiments.

[1]  D. S. Bloomfield,et al.  A COMPARISON OF FLARE FORECASTING METHODS. I. RESULTS FROM THE “ALL-CLEAR” WORKSHOP , 2016, 1608.06319.

[2]  C. J. Wolfson,et al.  The Atmospheric Imaging Assembly (AIA) on the Solar Dynamics Observatory (SDO) , 2011 .

[3]  R. Stein,et al.  Solar Surface Magneto-Convection , 2012 .

[4]  Didier Sornette,et al.  Self-Organized Criticality and Earthquakes , 1989 .

[5]  W. Pesnell,et al.  The Solar Dynamics Observatory (SDO) , 2012 .

[6]  E. Lu,et al.  Avalanches and the Distribution of Solar Flares , 1991 .

[7]  M. Aschwanden Irradiance observations of the 1–8 Å solar soft X-ray flux from goes , 1994 .

[8]  C. J. Wolfson,et al.  Design and Ground Calibration of the Helioseismic and Magnetic Imager (HMI) Instrument on the Solar Dynamics Observatory (SDO) , 2012 .

[9]  J P Eastwood,et al.  The Economic Impact of Space Weather: Where Do We Stand? , 2017, Risk analysis : an official publication of the Society for Risk Analysis.

[10]  P. McIntosh The classification of sunspot groups , 1990 .

[11]  Tang,et al.  Self-Organized Criticality: An Explanation of 1/f Noise , 2011 .

[12]  K. Shibata,et al.  Solar Flares: Magnetohydrodynamic Processes , 2011 .

[13]  Kris Hauser,et al.  Incorporating side-channel information into convolutional neural networks for robotic tasks , 2017, 2017 IEEE International Conference on Robotics and Automation (ICRA).

[14]  D. Hathaway The Solar Cycle , 2010, Living reviews in solar physics.