Volatiles and Fluids in Subduction Zones

Mantle wedge hydration

In December 2004 the great Sumatra earthquake and tsunami affected the entire Indian ocean. Roughly three month later a second large earthquake hit Sumatra. Historically, it was always Sumatra and never Java that was affected by seismic events of magnitude Mw>8. Serpentinization of the mantle wedge - the main focus of research in A5 - could explain the seismogenic behaviour along the Sunda arc.

In many subduction zones the downdip limit of thrust earthquakes approximately coincides with the intersection of the thrust with the forearc mantle. This limit may be explained by aseismic hydrous minerals present in the mantle wedge. During subduction the downgoing plate experiences higher pressures and temperatures. Consequently, water trapped in the crust and mantle is gradually released by dehydration and infiltrates the overlying forearc mantle forming serpentinite and brucite, changing the strong, dry rocks into weak, hydrous rocks (Bostock et al., 2002; Hyndman and Peacock, 2003). These frail rocks do not support seismogenic slip and may therefore limit the size of the rupture zone and hence earthquake magnitude.

Bouguer gravity anomalies over the Sunda arc correlate well with the occurrence of large megathrust earthquakes; negative anomalies mark segments characterized by larger earthquakes while positive anomalies indicate lower seismic potential. This feature, known as Trench Parallel Gravity Anomalies (TPGA), has been observed world-wide (Song and Simons, 2003). We (Grevemeyer and Tiwari, 2006) used thermal models and structural constraints derived from seismic and gravity data to explain seismogenic behaviour in the Sunda subduction zone. With respect to Java, oblique subduction of young oceanic crust shifts the seismogenic coupling zone roughly 40 km trenchward offshore of northern Sumatra and increases the width of the locked megathrust. A prominent positive gravity anomaly offshore of Java is caused by a shallow mantle wedge underlying the forearc basin. A serpentinized mantle wedge would limit the width of the coupling zone off Java to only 30-40 km, compared to >120 km offshore of Sumatra. Thus, upper plate structure along the Sunda arc may explain seismogenic behaviour and distribution of earthquake hazards.

A number of studies suggest that the forearc mantle might generally be serpentinized and therefore too weak to support seismogenic stick-slip behaviour (e.g., Bostock et al., 2002; Hyndman and Peacock, 2003). For southern and northern Sumatra, this phenomenon would explain why the downdip limit of the seismogenic zone coincides with the intersection of the forearc Moho with the downgoing plate. Furthermore, if serpentinized, the shallow mantle wedge off Java would limit the size of the seismogenic megathrust and explain the relationship between seismic moment release and TPGA. Comparing structural evidence from Sumatra and Java clearly suggests that a negative Bouguer anomaly occurs over "normal" trenches and marine forearc's, while a positive Bouguer anomaly indicates high density material occurring at shallow depth. We suggest that features controlling seismogenic behaviour along the Sunda trench may explain similar features observed elsewhere. In terms of this interpretation, a shallow forearc mantle causes a positive TPGA and therefore controls prominent lateral differences in TPGA and may support the idea that forearc serpentinization is a global phenomenon. However, if a forearc mantle would not be hydrated, a positive TPGA could be misleading, resulting into a destructive megathrust earthquake in an area believed to be safe.

Results are published by Grevemeyer and Tiwari (Overriding plate controls spatial distribution of megathrust earthquake in the Sunda-Andaman subduction zone, Earth Planet. Sci. Lett., 251, 199-208, 2006).