Why Does Psha Overpredict Historically Observed Shaking Data?
Description:
Probabilistic seismic hazard assessments (PSHA) based on current practice consistently appear to overpredict historically observed shaking data. Comparing the forecasts and datasets, independently prepared by different groups using similar methods, PSHA results (mean PGA converted to MMI or other intensity measures using published conversion equations) from California, Japan, Italy, Nepal, and France display a similar trend, even correcting for the lengths of the observation times. This overprediction is unexpected because recent numerical simulations show that if key PSHA parameters such as recurrence rate, ground motion prediction equations, and associated uncertainties are appropriate, the observed fractional exceedance is equally likely to be above or below that predicted. Furthermore, the standard deviation of fractional exceedances should decrease for larger ratios of observation time to return period length (i.e., observations should be closer to the prediction as observation time increases). However, in most regions studied to date, the observations of historical shaking data are below the predictions of PSHA, and no consistent improvement appears with longer observation times. The consistent overpredictions across a variety of tectonic settings suggest possible systematic biases. Our studies suggest that the discrepancy is possibly due to the ground motion intensity conversion equations. Additionally, the median PGA hazard fractile may be more appropriate than the mean for performance evaluations, as it tends to decrease the overprediction. Understanding the causes of this discrepancy between forecasted PGA and observed ground shaking is important for improving seismic hazard assessments.
Session: Advances in Probabilistic Seismic Hazard Analysis and Applications
Type: Oral
Date: 4/19/2023
Presentation Time: 02:00 PM (local time)
Presenting Author: Leah Salditch
Student Presenter: No
Invited Presentation:
Authors
Leah Salditch Presenting Author Corresponding Author leah.salditch@gmail.com U.S. Geological Survey |
Molly Gallahue molly@earth.northwestern.edu Northwestern University |
James Neely jneely@uchicago.edu University of Chicago |
Seth Stein s-stein@northwestern.edu Northwestern University |
Norman Abrahamson abrahamson@berkeley.edu University of California, Berkeley |
Susan Hough hough@usgs.gov U.S. Geological Survey |
|
|
|
Why Does Psha Overpredict Historically Observed Shaking Data?
Category
Advances in Probabilistic Seismic Hazard Analysis and Applications