Abstract
<title>Abstract</title> <p>The frictional resistance of an oceanic-plate subduction interface cannot be measured directly at full scale. The interface is buried, hydrated, thermally evolving, geometrically curved, and composed of locked asperities, creeping patches, fluids, sediments, altered minerals, and episodic slow-slip or seismic rupture zones. This study tests whether a structural-state friction architecture, previously used for inaccessible large-scale friction problems, can be meaningfully projected onto subduction observations. Two independent observational projections are used. First, heat-flow and thermal-model studies constrain the long-term effective friction coefficient of subduction interfaces to low values, typically near mu' = 0.03-0.07, far below laboratory Byerlee friction of about 0.6. A reduced structuralstate effective-friction closure is fitted to these interval constraints and compared with a constant effective-friction baseline. Second, USGS earthquake catalog data are downloaded for 12 representative subduction regions from 1900 to the present, using M >= 7 events. The total seismic moment release is compared against convergence speed, thermal state, slab geometry, and structural-state projection candidates. The heat-flow audit shows that the structural-state closure falls inside all compiled effective-friction intervals and gives a slightly smaller interval loss than the best constant effective-friction model. The USGS audit is more discriminating: simple convergence speed performs poorly, while thermal state and a speed-thermal-dip projection perform better under leave-one-out testing. A naive structural loading index fails, showing that the formula cannot be treated as a universal speedonly friction law. The strongest supported conclusion is therefore bounded: subduction effective friction is compatible with a structural-state closure, but the observable projection must include thermal-fluid state and slab geometry. External universality remains unproven.</p>