The SOHO/EIT solar flare catalog

In this work, we attempt to produce a database of solar flares detected by the Extreme ultraviolet Imaging Telescope (EIT) aboard the Solar and Heliospheric Observatory (SOHO) in accordance to the Geostationary Operational Environmental Satellite (GOES) flare catalog. The flare detection effort for the pre-SDO (Solar Dynamics Observatory) era available is only the GOES catalog with spatial and temporal specifications along with an associated active region (AR) number. The GOES system records flares through hard X-ray detectors and provides flare location on the Sun using it's SXI (Solar X-ray Imager) instrument (Bornmann et al., 1996). However, many flare locations and/or AR numbers are unavailable in the GOES catalog. For machine learning purposes, it is crucial to establish the correct locations of solar flares in the GOES catalog with other space-borne instruments (Aydin et al., 2019). Hence we choose the duration of EIT operations, i.e., from 1996 to 2010, for analysis of an extensive data set from the instrument.

Considering the GOES flare catalog as the primary source, the recorded peak time of the flare is used as a reference to select the closest Extreme UltraViolet (EUV) image of the Sun from EIT plus a consecutive one in order to detect brightenings. The follow-up image to the peak time could be before or after the peak depending on the temporally closest available. Although the cadence of EIT was revised from two images per day to 12 minutes in mid-1998 until 2010 (ESA, 2017), this still causes significant data gaps in capturing much of the flaring phenomenon. Therefore, we monitor possible errors in flare detection by parameters such as temporal overlap, observational wavelength, maximum cadence limit of 24 minutes, and considering only full field of view (FOV) images. Over 12 years of EIT data has been downloaded via the Virtual Solar Observatory (VSO).

Nonetheless, this flare data from the SOHO/EIT period proves essential to the study of the solar atmosphere as the instrument has covered most of Solar Cycle 23. The flare catalog will be part of a machine learning benchmark dataset for solar flare prediction.