Using genetic algorithm to constrain earthquake slip distributions from point surface displacement

• Main Author: Lindsay, Anthony
• Published: Ulster University 2016
• Subjects: 551.22
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.697547

Examining fault activity over several earthquake cycles is necessary for long term modeling of the faults strain budget and stress state. While this requires knowledge of coseismic slip distributions for successive earthquakes along the fault, these exist only for the most recent events. However, overlying the Sunda Trench, sparsely distributed coral microatolls are sensitive to tectonically induced changes in relative sea levels and provide a century spanning paleo geodetic and paleoseismic record. This thesis presents a new technique called the Genetic Algorithm Slip Estimator (GASE) to constrain slip distributions from observed surface deformations of corals. A suite of models consistent with the observations is identified and from them an ensemble estimate of the causative slip is computed. The technique is systematically tested using synthetic data. Applying the technique to observed coral displacements for the 2005 Nias-Simeulue earthquake and 2007 Mentawai sequence, modeled solutions reproduce features of slip (i.e. along strike extent, magnitude of slip and the presence of asperities), present in previously published inversions such as the magnitude and location of slip asperities. From the displacement data available for the 1797 and 1833 Mentawai earthquakes, slip estimates are presented that reproduce observed displacements. The areas of highest modeled slip in the paleoearthquakes are non-overlapping and our solutions appear to tile the plate interface, complementing one another. Using the slip estimated produced by the GASE technique the total stress field of the Sumatran section of the Sunda megathrust is modeled. Combining the coseismic stresses produced by the GASE models of the 1797 and 1833 paleo earthquake and by 25 other M >7.0 events that have occurred along the megathrust, along with the cumulative interseismic stressing since 1796, the evolving stress field of the Sunda megathrust is investigated. The use of GASE slip distributions for the 1797 and 1833 paleoearthquakes resolve some inconsistencies between the prestress state of the fault and coseismic slip reported by Nalbant et al. [2013]. However, further work is required to explore the sensitivity of the stress field to variation in slip. While this thesis describes the development and testing of the GASE technique for use with the coral displacement data along the Sunda Megathrust, it may be possible to adapt the algorithm to examine paleoseismic records elsewhere. Application of the technique to alternative sources of paleoseismic data may yield insight into the details of pre-instrumental earthquakes in other regions. In tum, more robust modeling of the past seismicity in these areas may allow better constraints on the future risks posed by earthquakes to at risk communities.