Search for Reflected Cosmic Rays in ARIANNA Data

The next generation of radio-based neutrino observatories is currently under development, aiming to detect neutrinos with energies more than 10¹⁷ eV. With their increased scale and sensitivity, it is crucial to focus efforts on removing rare, yet non-negligible, background events. In this presentation, I demonstrate rejection techniques for the so-called reflected cosmic rays (RCR). Using station data from the Antarctic Ross Ice-Shelf Antenna Neutrino Array (ARIANNA) experiment, I apply cross-correlation and machine learning techniques to identify RCR events. The trained convolutional neural network demonstrates high accuracy during training and passes several post-training checks. This rejection tool can serve as a noise filter for archival data, or potentially increase trigger rates operating as a real-time trigger.