Global Long-wavelength Body-wave Catalog (GLoBoCat) for Lower Mantle P and S Wave Travel Times

Abstract

Long period seismic body wave travel times are a foundational source of information regarding the Earth's interior. Manual methods provide travel time catalogs with high internal consistency, but have not kept pace with the rapid growth in teleseismic waveforms. Here we use attention-based machine learning trained on prior manual datasets to develop a catalog spanning 2010-2023 which is consistent with previous catalogs and easy to update. The new catalog exceeds millions of measurements for P and S wave travel times and retains features of the previous datasets. Comparing our data to short period travel times, we find the best fit slope is different for early versus late arrivals, and that this difference is larger for P waves. The change in slope we observe is opposite to that proposed for wave propagation effects and warrants further investigation into the origin of this apparent dispersion in Earth's lower mantle.