Radiation Belt Data Science using the Solar Dynamics Observatory

The SDO’s (Solar Dynamics Observatory) AIA (Atmospheric Imaging Assembly) has been observing the Sun since 2010, imaging the star in several wavelengths. In those images, a de-spiking algorithm was introduced into the processing pipeline of the images to remove bright points, called “spikes,” that were caused by energetic radiation at the geosynchronous location of the spacecraft. The number of spikes that were observed and removed are reported in the metadata of the images. Later studies by Dr. Spyros Kasapis under the mentorship of Dr. Barbara J. Thompson explored the composition of these spikes to explain their origin. Kasapis et al. found that the GOES-14 satellite and SDO’s orbits aligned twice a day, at the same time. Since GOES-14 and SDO were observing the same space, they started correlating the data of the radiation belt’s composition from GOES-14 with the nspikes observed by SDO, finding a good correlation with electrons in the 40keV energy band. They then trained a machine learning model to be able to predict the flux of 40keV electrons based on the number of spikes detected by SDO, reaching the mathematical relation: GOESe 40keV = 1.1604 nSpikes. Building on the Kasapis et al. work, we created a system to allow a user to pull SDO metadata of a given date range from the Joint SDO Operations Center (JSOC), containing the nspikes, exposure time of the image and other relevant information, and then filtering the data to remove any images that were captured during calibration routines or when the sun is eclipsed by the earth and is therefore not visible then applying the Kasapis et al. mathematical relation to obtain a prediction of the 40keV electron flux as well as obtaining Geocentric Equatorial Inertial (GEI) coordinate data for SDO at 1-minute cadence and then interpolating it to refine its precision. The GEI coordinates are then used to derive the geomagnetic parameters L-shell, magnetic local time and magnetic latitude. This component of the system is getSpikes.py. The system is also continuously pulling near-real time data from JSOC, generating the same data as getSpikes.py but within 5 to 6 minutes of current conditions. All of this data is then stored into an SQL database that is accessed by the user interface to examine the spikes data and obtain up-to-date predictions of electron flux.