← Back to Sessions

Does the Logic Tree Hide the Forest? Quantifying Uncertainties in Predicted Risk for Individual Model Settings in the Induced-Seismicity Hazard and Risk Analysis of the Groningen Gas Field

Description: 

The public Seismic Hazard and Risk Analysis (pSHRA) for production-induced seismicity in Groningen comprises a seismic source model (SSM) that provides a (spatio-temporal) seismicity rate and associated magnitude distribution, a ground motion model (GMM) to compute hazard, and a fragility and consequence model (FCM) that calculates Local Personal Risk for each building. Epistemic uncertainties of six key model parameters for the pSHRA are taken into account via a logic tree: 2 SSM (magnitude-frequency model, maximum magnitude), 2 GMM and 2 FCM model parameters. Due to the unique context of production-induced seismicity and its associated models, this logic tree may not cover all relevant epistemic uncertainties. Here, we aim to identify other model settings that may be a significant source of uncertainty.
To do so, a sensitivity analysis was performed for 19 model settings, which are a selection of deterministic input data, model parameters, model chain components, and model choices/assumptions. One model setting was varied per sensitivity test with respect to a base case. The metric used to quantify the sensitivity to each particular model setting was the mean risk over the entire building database. One or more variations per setting were tested. The risk sensitivity of a model setting is deemed ‘significant’ if a significance threshold of 1.4x the mean risk of the base case is surpassed. This threshold represents aleatoric variability, based on the effects of alternative choices in the earthquake completeness magnitude. Most sensitivity tests increase risk relative to the base case. The model settings with a significant sensitivity are: The inclusion of period-2-period correlation in the site response of the GMM, alternative magnitude-frequency models, adapted fragility parameters of unreinforced masonry in the FCM, and the use of the Coulomb stress measure in the SSM calibration. These results help focus ongoing research in pSHRA uncertainty reduction, and provide insight into possible candidates for inclusion in a logic tree.

Session: Advances in Probabilistic Seismic Hazard Analysis and Applications

Type: Oral

Date: 4/19/2023

Presentation Time: 03:00 PM (local time)

Presenting Author: Frans M. Aben

Student Presenter: No

Invited Presentation: 

Authors