Beyond Button Pushing: Probabilistic loss assessment in California
According to the Working Group on California Earthquake Probabilities, the likelihood of magnitude 6.7 and larger earthquakes occurring somewhere in California in the next 30 years is near certainty. Given the high level of earthquake risk in California, it is crucial to develop policies for risk mitigation, emergency preparedness, risk transfer and insurance, and buildings and infrastructure design that are informed by the best available science on earthquakes. Consequently, there is a need for improved understanding of the seismic risk in the State, including a better characterization of elements that comprise the risk: hazard, exposure, and vulnerability.
There are too few past earthquakes to use history as the only judge of underlying earthquake risk. Earthquake risk models allow us to bridge that gap by combining with history the best available science and engineering knowledge about earthquakes and their effects on society. In so doing, risk models can provide realistic estimates of the likelihood of future earthquakes and the potential societal impacts, such as damage to structures, economic losses, casualties, and business disruption. However, proprietary earthquake risk models are protected intellectual property, containing countless assumptions and methods that are disclosed only as necessary, as such, these models are black boxes. Model users can push buttons (vary the input) to produce results, but the effect of each button remains mostly hidden. The lack of transparency of these models means the uncertainties in the assumptions and model results are not known to users and stakeholders, which diminishes their credibility and hinders their adoption in earthquake risk management.
The Alfred E. Alquist Seismic Safety Commission (SSC) engaged the GEM Foundation to quantify and discuss the impact of various assumptions on earthquake model results for California, and also to investigate the treatment of uncertainty within the model. The results of this study will help identify the key factors that influence the risk results in California and improve understanding of the seismic risk in the state. Such improved knowledge can be leveraged by a wide range of stakeholders within the community to: enhance earthquake risk mitigation strategies; develop appropriate emergency preparedness and response plans; inform risk transfer and insurance mechanisms; and better manage risk-sensitive investment portfolios.
Objectives and Expected Outputs
The goal of the project was to show how important the quantification of uncertainty is in estimating and understanding California�s earthquake risk using OpenQuake.
Several objectives were achieved during the course of this project, including:
� Implementation within OpenQuake of the latest seismic hazard model for California based on the recently published Uniform California Earthquake Rupture Forecast version 3 (UCERF3), produced by the U.S. Geological Survey and the Working Group on California Earthquake Probabilities.
� Calculation of the average annual loss estimates for all 8,057 census tracts in California, using the seismic hazard model based on UCERF3.
� Establishing the range (distribution) of scientifically viable results for the chosen risk metrics by accounting for the various uncertainties in the hazard model.
� Identification of the components of the hazard model contributing most to the overall uncertainty in the risk metrics for the different exposure portfolios.
� Implementation of a model simplification (�logic-tree trimming�) software tool to reduce the number of computer runs and greatly speed up the time required for running the seismic risk model for California.
United States of America
United States of America
2014 - 2016