A Bayesian Approach to Estimating the Source of Historical Earthquakes From Intensity Data: Application to the Eastern Sunda Arc, Indonesia, 1681-1877

Date: 4/26/2019

Time: 09:15 AM

Room: Pike

Historical reports of earthquake effects from the period 1681 to 1877 from Java, Bali and Nusa Tenggara, Indonesia, are used to estimate the source mechanisms of some of the most significant earthquakes. After assigning intensity values based on original historical reports, the source is estimated by undertaking a grid search of source parameters (magnitude, location, depth, strike and dip). For each parameter combination, we use ground-motion models and ground motion to intensity conversion equations to forward model intensity at each intensity data point. Due to limitations in the spatial coverage of intensity data inherent in reports from the archipelago, which predominantly focus on impacts to Dutch colonial economic interests, a range of parameter combinations is possible for each event. Bayesian inference is therefore used to calculate the distribution of source parameters given the historical intensity data. A Bayesian approach allows us to include our prior beliefs about the source; most significantly by applying a Gutenberg-Richter distribution to the prior magnitude distribution, we account for that fact that smaller earthquakes occur more frequently than larger earthquakes. The approach also allows us to account for uncertainty in the choice of ground motion model. The results demonstrate that large intraslab earthquakes have been responsible for major earthquake disasters in Java, including an Mw ∼ 7:4 intraslab earthquake near Jakarta in 1699 and an Mw ∼ 7:8 event in 1867 in Central Java. The results also highlight the potential for large earthquakes to occur on the Flores thrust, with a cluster of large earthquakes rupturing the Flores thrust in 1815, 1818, and 1820. We do not find conclusive evidence for the occurrence of large earthquakes on the Java megathrust during the time period, although smaller events may have occurred.

Presenting Author: Jonathan D. Griffin

Authors