Quantifying Sensitivities of PSHA for France to Earthquake Catalog Uncertainties, Truncation of Ground-Motion Variability, and Magnitude Limits

Abstract The results of this study clearly identify four key parameters controlling the estimation of probabilistic seismic hazard assessment (PSHA) in France in the framework of the Cornell–McGuire method. Results in terms of peak ground accel-eration demonstrate the equally high impact, at all return periods, of the choice of truncation of the predicted ground-motion distribution (at 2r) and of the choice between two different magnitude-intensity correlations. The choice of minimum magnitude (3.5/4.5) on hazard estimates can have an important impact at small return periods (1000 years), whereas the maximum magnitude (6.5/7.0), on the other hand, is not a key parameter even at large return periods (10,000 years). This hier-archy of impacts is maintained at lower frequencies down to 5 Hz. Below 5 Hz, the choice of the maximum magnitude has a much greater impact, whereas the impact due to the choice of the minimum magnitude disappears. Moreover, variability due to catalog uncertainties is also quantified; these uncertainties that underly all hazard results can engender as high a variability as the controlling parameters. Parameter impacts, calculated at the centers of each source zone, show a linear trend with the seismicity models of the zone, demonstrating the lack of contributions coming from neighboring zones. Indeed, the region of influence that contributes to the PSHA es-timate at a given site decreases with increasing return periods. The resulting overall variability in hazard estimates due to input uncertainties is quantified through a logic tree, obtained coefficients of variation vary between 10 % and 20%. Until better physical models are obtained, the uncertainty on hazard estimates may be reduced by working on an appropriate magnitude-intensity correlation.