Performance assessment of real-time earthquake localization algorithms for Earthquake Early Warning Systems in the Ibero-Maghrebian Region

Abstract. We evaluate the performance of two real-time earthquake location algorithms, RTLoc and MParLoc, implemented in the Earthquake Early Warning System (EEWS) QuakeUp. Specifically, we assess whether the recently developed MParLoc algorithm—which incorporates P-wave arrival times, P-peak amplitude ratios, and back-azimuth estimates—improves earthquake localization accuracy in the Ibero-Maghrebian Region (IMR). Analyzing M_w ≥ 4 earthquakes recorded in the IMR between 2011 and 2026, we find that MParLoc generally outperforms RTLoc. It yields hypocentral locations significantly closer to reference solutions, particularly during the earliest stages after earthquake detection, when only a few stations and short P-wave windows are available. Although both methods converge as more P-wave picks accumulate, MParLoc’s early-stage precision is critical. Hypocentral accuracy strongly influences the distance term in the ground motion prediction equations used to generate real-time shakemaps, which define the spatial distribution of expected ground shaking and delineate Potential Damage Zones (PDZs). The results indicate that implementing MParLoc in PDZ-based EEWS like QuakeUp would significantly improve warning performance in the IMR, especially during the initial stages of an earthquake, by reducing the time to first alert, increasing lead-time for emergency safety actions, and providing valuable scientific and operational information for emergency response.