Solra Flare CNN Catalog

A convolutional neural network (CNN)–based detection framework is developed to construct a new catalog for solar flares. The method operates on high resolution (1s cadence) Geostationary Operational Environmental Satellites (GOES) soft X-ray data and identifies flare rise episodes. The method aims to complement and extend the GOES flare archive by alleviating limitations such as the reduced sensitivity to small or closely spaced events during periods of elevated background flux. Operating on high-resolution GOES SXR measurements, the framework is optimized to identify flare rise episodes through deep learning, rather than the complete rise–decay profile. By focusing on the rise episode, the approach partially relaxes the slow-driving (non-overlap) constraints inherent to many conventional flare detection algorithms, thereby allowing consecutive and overlapping events, particularly those initiating during the decay of a preceding flare, to be recognized. From 01 January 2018 to 22 August 2025, the algorithm detected 111,580 flare candidates, compared with 14,612 events in the corresponding GOES archive. For each candidate, the probability of being a true positive is quantified by a Bayesian inference.