Unveiling subduction-related seismicity: towards a new global database

Subduction zones host the world’s largest earthquakes. Recent studies suggest complex interactions between megathrust, upper plate and intraslab seismicity. Understanding the spatio-temporal relationships of seismicity within subduction zones is challenging, yet essential for accurate seismic hazard assessment. A prerequisite for conducting these analyses involves the availability of a reliable and readily updatable dataset that classifies subduction seismicity into the three categories above. In this work, we compile a comprehensive global database of subduction-related earthquakes from 1976 to 2023, using the ISC-GEM catalog (Storchak et al., 2013; 2015; Di Giacomo et al., 2018) for events with magnitude Mw ≥ 5.5. Building on Heuret et al. (2011), we define 505 trench-normal transects across all active subduction zones, spaced at 1-degree intervals along the trench, partially overlapping and extending 120 km on both sides of a vertical plane. The seismicity in each transect is initially categorized into shallow (≤ 70 km) and deep (> 70 km) events. We then focus on the megathrust region, identifying earthquakes exhibiting a thrust focal mechanism with the  azimuth and dip of the focal planes aligning with the strike and dip of the megathrust along the transect. Subsequent steps involve categorizing the remaining earthquakes in the transect as either upper or subducting plate events. The classification uses the Slab 2.0 model (Hayes et al. 2018) when available, determining whether each earthquake occurs above or below the slab top surface.  In regions lacking Slab 2.0 data, geometric criteria are applied, considering the distance of the hypocenter to the trench. For each transect, this workflow yields three distinct seismicity classes: megathrust, upper plate, and intraslab earthquakes. Subsequently, seismicity from individual transects is merged into 62 broader segments (Heuret et al., 2011), ensuring the uniqueness of earthquakes in each segment. This automated workflow ensures the application of objective classification criteria and facilitates efficient analyses with each update of the ISC-GEM catalog. We compare key seismic parameters (e.g., maximum magnitude, number of events, cumulated seismic moment, recurrence time) across the different categories and segments. Additionally, we evaluate the correlation with a wide range of geological, geophysical, and geodynamical factors. This process not only provides an overview of the global behavior of subduction-related seismicity but also allows for the statistical identification of the combination of factors influencing subduction seismicity.