Document Type


Date of Degree Completion

Spring 2019

Degree Name

Master of Science (MS)


Geological Sciences

Committee Chair

Breanyn MacInnes

Second Committee Member

Lisa Ely

Third Committee Member

Walter Szeliga


Reliable tsunami early warning forecasts rely on accurate initial modeling conditions and interpretations of subduction zone behavior in a multi-century perspective. GPS and seismologic data were introduced this past century to study rupture dynamics in detail, however limited information is known about ruptures that pre-date the 20th century. I propose a methodology that uses statistics to better understand these pre-20th century ruptures. This methodology applies the historical and geologic tsunami record as a means to select a suite of tsunami simulations from earthquake source solutions. I chose south-central Chile (46°S to 30°S) to test this new methodology; it has an extensive earthquake historical record at 47 different coastal sites, some of which date to the 16th century. Between 1570 and 1960, this region experienced at least 17 tsunamigenic earthquakes. In addition to evaluating possible source solutions for these earthquakes, my methodology also allows the test of whether subducted fracture zones, like the Mocha fracture zone (MFZ) in south-central Chile, controls rupture propagation (as previously hypothesized). For this research, I used GeoClaw, a numerical tsunami modeling code, to simulate 423 forward-modeled Mw 8.7 - 9.5 earthquake scenarios with stochastic, variable slip distributions. I used Akaike’s Information Criterion (AIC) to identify significant earthquake parameters (Mw and slip location) of 17 events by statistically selecting source models that had similar simulated wave heights to known observations in the historic and geologic record. For example, I concluded from AIC that the 1960 event was a Mw 9.3 rupture with high slip concentration (~ 30 m) at ~ 39-40ºS, and the 1730 event was a Mw 9.3 rupture with shallow maximum slip at ~ 36ºS; both solutions support the MFZ hypothesis. The AIC results generally agree with previously estimated magnitudes within the literature and were validated by using root mean square error RMSE values. I produced high resolution maps at three coastal sites with well-known tsunami observations for further refinement of potential rupture scenarios. Defining historical rupture characteristics gives insight regarding temporal and spatial variabilities of locking zones. This information may be useful for predicting future near-field tsunami wave heights for particularly vulnerable coastal regions.