Subduction Parameters Controlling the Occurrence of Shallow and Deep Slow‐Slip Events (SSEs)

Slow‐slip events (SSEs) are transient aseismic fault‐slip phenomena that release tectonic stresses in a variety of tectonic environments, including subduction zones. In subduction margins, SSEs commonly occur along the plate interface at shallow (<20 km) and deep (30–60 km) depths. Here, we explore nonlinear relationships between subduction features (geometric, kinematic, and structural parameters) and the occurrence of SSEs, using two machine learning algorithms based on decision trees. Our results indicate that SSEs preferentially occur in subduction interfaces with low dip angles (<20°), slow convergence rates (<6 cm/yr), and/or along‐interface heterogeneity in physical and structural properties, often associated with fluid‐rich conditions. Shallow SSEs are further associated with slab roughness, erosive margins, and relatively high lithostatic pressures compared to other regions at similar shallow depths. In contrast, deep SSEs mostly occur in young slabs showing pronounced negative shear‐wave velocity anomalies and relatively low lithostatic pressures at depths greater than 20 km. These patterns identify the most influential controls on shallow and deep SSEs, revealing a depth‐dependent interplay among geometric, kinematic, and rheological factors. Moreover, we find that most subduction zones could host SSEs at specific depths, with ∼70% of the evaluated subduction margins showing medium to relatively high susceptibility to host SSEs. Our global model identifies subduction margins prone to SSEs that may frequently remain undetected. This study provides a framework for future monitoring efforts and helps assess the implications of SSEs for seismic hazard.