Why are the Standard Probabilistic Methods of Estimating Seismic Hazard and Risks Too Often Wrong.

According to the probabilistic seismic hazard analysis (PSHA) approach, the deter- ministically evaluated or historically defined largest credible earthquakes (often referred to as Maximum Credible Earthquakes, MCEs) are “an unconvincing possi- bility” and are treated as “likely impossibilities” within individual seismic zones. However, globally over the last decade such events keep occurring where PSHA pre- dicted seismic hazard to be low. Systematic comparison of the observed ground shaking with the expected one reported by the Global Seismic Hazard Assessment Program (GSHAP) maps discloses gross underestimation worldwide. Several inconsistencies with available observation are found also for national scale PSHA maps (including Italy), developed using updated data sets. As a result, the expected numbers of fatalities in recent disastrous earthquakes have been underestimated by these maps by approx- imately two to three orders of magnitude. The total death toll in 2000e2011 (which exceeds 700,000 people, including tsunami victims) calls for a critical reappraisal of GSHAP results, as well as of the underlying methods. In this chapter, we discuss the limits in the formulation and use of PSHA, addressing some theoretical and practical issues of seismic hazard assessment, which range from the overly simplified assumption that one could reduce the tensor problem of seismic- wave generation and propagation into a scalar problem (as implied by ground motion prediction equations), to the insufficient size and quality of earthquake catalogs for a reliable probability modeling at the local scale. Specific case studies are discussed, which may help to better understand the practical relevance of the mentioned issues. The aim is to present a critical overview of different approaches, analyses, and ob- servations in order to provide the readers with some general considerations and constructive ideas toward improved seismic hazard and effective risk assessment. Specifically, we show that seismic hazard analysis based on credible scenarios for real earthquakes, defined as neo-deterministic seismic hazard analysis, provides a robust alternative approach for seismic hazard and risk assessment. Therefore, it should be extensively tested as a suitable method for formulating scientifically sound and realistic public policy and building code practices.