Search

Earthquake hazard assessments and rupture forecasts are based on the potential length of seismic rupture and whether or not slip is arrested at fault segment boundaries. Such forecasts do not generally consider that one earthquake can trigger a second large event, near-instantaneously, at distances greater than a few kilometres. Here we present a geodetic and seismological analysis of a magnitude 7.1 intracontinental earthquake that occurred in Pakistan in 1997. We find that the earthquake, rather than a single event as hitherto assumed, was in fact an earthquake doublet: initial rupture on a shallow, blind reverse fault was followed just 19 s later by a second rupture on a separate reverse fault 50 km away. Slip on the second fault increased the total seismic moment by half, and doubled both the combined event duration and the area of maximum ground shaking. We infer that static Coulomb stresses at the initiation location of the second earthquake were probably reduced as a result of the first. Instead, we suggest that a dynamic triggering mechanism is likely, although the responsible seismic wave phase is unclear. Our results expose a flaw in earthquake rupture forecasts that disregard cascading, multiple-fault ruptures of this type.

Earthquake hazard assessments and rupture forecasts are based on the potential length of seismic rupture and whether or not slip is arrested at fault segment boundaries. Such forecasts do not generally consider that one earthquake can trigger a second large event, near-instantaneously, at distances greater than a few kilometres. Here we present a geodetic and seismological analysis of a magnitude 7.1 intracontinental earthquake that occurred in Pakistan in 1997. We find that the earthquake, rather than a single event as hitherto assumed, was in fact an earthquake doublet: initial rupture on a shallow, blind reverse fault was followed just 19 s later by a second rupture on a separate reverse fault 50 km away. Slip on the second fault increased the total seismic moment by half, and doubled both the combined event duration and the area of maximum ground shaking. We infer that static Coulomb stresses at the initiation location of the second earthquake were probably reduced as a result of the first. Instead, we suggest that a dynamic triggering mechanism is likely, although the responsible seismic wave phase is unclear. Our results expose a flaw in earthquake rupture forecasts that disregard cascading, multiple-fault ruptures of this type.