hide
Free keywords:
-
Abstract:
Seismicity usually exhibits a non-Poisson spatiotemporal distribution and could undergo
nonstationary processes. However, the Poisson assumption is still deeply rooted in current
probabilistic seismic hazard analysis models, especially when input catalogs must be
declustered to obtain a Poisson background rate. In addition, nonstationary behavior
and scarce earthquake records in regions of low seismicity can bias hazard estimates that
use stationary or spatially precise forecasts. In this work, we implement hazard formulations
using forecasts that trade-off spatial precision to account for overdispersion and nonstationarity
of seismicity in the form of uniform rate zones (URZs), which describe rate
variability using non-Poisson probabilistic distributions of earthquake numbers. The
impact of these forecasts in the hazard space is investigated by implementing a negative-
binomial formulation in the OpenQuake hazard software suite, which is adopted
by the 2022 Aotearoa New Zealand National Seismic Hazard Model. For a 10% exceedance
probability of peak ground acceleration (PGA) in 50 yr, forecasts that only reduce the spatial
precision, that is, stationary Poisson URZ models, cause up to a twofold increase in
hazard for low-seismicity regions compared to spatially precise forecasts. Furthermore,
the inclusion of non-Poisson temporal processes in URZ models increases the expected
PGA by up to three times in low-seismicity regions, whereas the effect on high-seismicity
is minimal (∼5%). The hazard estimates presented here highlight the relevance, as well as
the feasibility, of incorporating analytical formulations of seismicity that go beyond the
inadequate stationary Poisson description of seismicity.