How well do Probabilistic Seismic Hazard Analysis (PSHA) maps forecast ground shaking due to earthquakes? This question is central to ensuring the safety, security, and economic well-being of citizens. PSHA maps are an important product for users including seismologists, engineers, insurers, and policymakers. PSHA, which has been used worldwide for almost 50 years, uses estimates of the probability of future earthquakes and the resulting shaking to predict the shaking expected with a certain probability over a given time (Cornell, 1968; Field, 2014). Extensive research is ongoing into how well hazard maps perform relative to these expectations and how the maps can be improved. This dissertation explores aspects of PSHA and its components - from the challenges of modelling earthquake histories with temporal clusters, to the task of comparing model predictions with observations of shaking intensity using performance metrics. The former proposes an alternative to the traditional earthquake cycle model in which a fault's past influences its future likelihood of experiencing an earthquake. The latter takes advantage of hindcasting - using past data to evaluate models which forecast the future. Ideally, we would compare hazard maps to data collected after the maps were made, however, due to the long recurrence times of large earthquakes relative to post-map observation periods, this is not typically feasible. Part of this research involved collecting, compiling, and consistently interpreting 162 years of seismic intensity data in the California Historical Intensity Mapping Project (CHIMP). Using CHIMP data and historical seismic intensity compilations from Italy, France, and Nepal, I compared the maximum observed shaking in an area to that predicted by PSHA models. Assuming the datasets to be correct, it appears that PSHA models overpredict shaking, even correcting for the time period involved. Assuming the PSHA models are correct, a shaking deficit exists between the model and observations. Possible reasons for this apparent discrepancy between the model and observations are threefold: 1) the observations could be biased low; 2) the observation period has been less seismically active than typical - either by random chance or temporal variability due to stress shadow effects; 3) the model overpredicts, due to limitations of either the earthquake rupture forecast or the ground motion models (GMMs).

Creator

• Salditch, Leah Marschall

DOI

• https://doi.org/10.21985/n2-my3w-v278

Subject

• Geophysics

Language

• en

Alternate Identifier

• http://dissertations.umi.com/northwestern:15803
• etdadmin_upload_852540

Keyword

• Seismology
• Seismic hazards

Date created

• 2021-01-01

Resource type

• Dissertation

Rights statement

• In Copyright