Earthquake Hazard When the Rate Is Non-Stationary: The Challenge of the U. S. Midcontinent

Abstract:

In July 2014, the U. S. Geological Survey released an update of the 2008 National Seismic Hazard Map for the coterminous U. S. The Map provides guidance for the seismic provisions of the building codes and portrays ground motions with a 2% chance of being exceeded in an exposure time of 50 years. Over most of the midcontinent the hazard model is derived by projecting the long-term historic, declustered earthquake rate forward in time. However, parts of the midcontinent have experienced increased seismicity levels since 2009 – locally by 2 orders of magnitude – which is incompatible with the underlying assumption of a constant-rate Poisson process. The 2014 Map acknowledged this problem, and for its intended purpose of underpinning seismic design used seismicity rates that are consistent with the entire historic record. Both the developers of the Map and its critics acknowledge that the remarkable rise of seismicity in Oklahoma and nearby states must be addressed if we are to fully capture the hazard in both space and time. The nature of the space/time distribution of the increased seismicity, as well as numerous published case studies strongly suggest that much of the increase is of anthropogenic origin. If so, the assumptions and procedures used to forecast natural earthquake rates from past rates may not be appropriate. Here we discuss key issues that must be resolved include: the geographic location of areas with elevated seismicity, either active now or potentially active in the future; local geologic conditions including faults and the state of stress; the spatial smoothing of catalog seismicity; the temporal evolution of the earthquake rate change; earthquake sequence statistics including clustering behavior; the magnitude-frequency distribution of the excess earthquakes, particularly to higher and yet unobserved magnitudes; possible source process differences between natural and induced earthquakes; and the appropriate ground motion prediction equations.