SUSHI II Long-term stress modelling: Implications for large earthquake forecasting Suleyman S. Nalbant, John McCloskey, Shane Murphy, Nuno Simao and Tony Lindsay Environmental Sciences Research Institute University of Ulster, Coleraine, Co. Derry, N. Ireland

Background There were two gigantic earthquakes on the Sunda megathrust in 2004 and 2005 with 3 months separation. Stress transfer from these earthquakes sparkled a worry that Mentawai section would rupture soon

Background There were historical evidences that supported this worry. A potential earthquake on the Mentawai section with a magnitude in the range of M~ would have disastrous consequences. A tsunami as a result of this potential earthquake would invade population centres along the west coast of Sumatra, specifically the city of Padang with 800,000 population.

Padang McCloskey et al EPSL 2008 Knowing slip distribution is important

Motivation Can we forecast a future earthquake with its possible rupture area and slip distribution on the Sunda megathrust? To have a reliable answer to this question –stress/strain accumulation over at least a seismic cycle should be known; Initial stress level before the last seismic cycle Stress perturbation due to the all earthquakes in the last seismic cycle Stress accumulation over inter-seismic periods.

Strain accumulation It is defined as the ratio of interseismic slip rate to plate convergence rate Interseismic slip rates are inverted from geodetic and palaeogeodetic measurements. strong coupling is a necessary but not sufficient condition for high slip in a single event Chlieh et al. (2008) JGR Natawidjaja et al. (2006) JGR

Rupture planes of all earthquakes M>7.0 occurred between 1797 and 2005 (22 earthquakes)

Slip Function for earlier historical earthquakes Seismic moment is kept the same

We are calculating stress changes assuming a ‘zero’ stress level in 1797.

The rupture limits may have been defined by the pre‐existing stress state shaped by previous earthquakes

Rupture extend and slip distribution of the 1797 EQ Natawidjaja et al (JGR)

Rupture extend and slip distribution of the 1833 EQ Natawidjaja et al (JGR) One thing we know for sure that slip distributions and rupture extend weren’t like this.

What do we know? Newcomb and McCann JGR 1987 Increasing Information

Class 4 February Eq Using Monte-Carlo method, we can produce many scenario earthquakes satisfying the observations

Location (lat/long) -0.06/98.21 Magnitude:7.7 Class 3 December Eq

Coral IDLatitudeLongitude Vertical Displacement(m) 22 Class and 1833 Eqs. Sumatra

Maximum Magnitude Minimum Magnitude Maximum Slip Fault Coupling

Year Month Day Long Lat Depth Mag. #Scenario Eq All events Class 1 Modern Eqs

Slip deficit scenarios; Which History is most likely?

A preliminary result The well-resolved red bulge under Siberut would produce an M8.7 earthquake if it all failed simultaneously.

What’s next? Genetic algorithm for improved inversion for corals Searching for further coral displacements to constrain other historical earthquakes – particularly 1861 and 1935 Use Bayesian techniques to assess the probability of each model given the data and uncertainty Look at the stress field before recent damaging earthquakes Communicate risk mitigation by engaging key stakeholders, so that science-based hazard maps can be presented in a readily-understood local context

From Loading and Earthquakes to Absolute Stress? Slip Deficit Initial Slip Deficit Seismic Slip Interseismic Loading + - = Interseismic Loading = Plate Velocity Coupling  Seismic Slip = Sum of Slip distributions