KEM9: Cumulative propagation effect of Groningen risk model component uncertainties on hazard and risk predictions
report
The Groningen Model Chain’s sensitivity on a number of model settings was tested. This was done by changing one model setting (model choice / assumption / parameter / input data) per sensitivity test with respect to a base case. The averaged risk of the test case is then compared with the base case. Most model setting variations result in a higher mean risk relative to the base case. We identified a small number of model settings that yield risk variations exceeding a ‘significance’ threshold of a factor 1.4 times the average risk of the base case. This threshold represents aleatory variability (irreducible uncertainty), based on the effects of alternative choices in the earthquake completeness magnitude Mmin. The model settings exceeding this threshold are: the inclusion of period-2-period correlation in the site response of the ground motion model (GMM), alternative magnitude-frequency (MF) models, adapted fragility parameters of unreinforced masonry (URM) in the fragility model, and the use of the Coulomb stress in the calibration of the seismic source model. Most of these ‘alternative model settings’ were previously identified as having substantial impact on the risk results. This study confirms these findings.
TNO Identifier
961795
Publisher
TNO
Collation
32 p.
Place of publication
Utrecht, The Netherlands