Effects of magnitude, depth, and time on Cellular Seismology Forecasts

Abstract

This study finds that, in most cases analyzed to date, past seismicity tends to delineate zones where future earthquakes are likely to occur. Network seismicity catalogs for the New Madrid Seismic Zone (NMSZ), Australia (AUS), California (CA), and Alaska (AK) are analyzed using modified versions of the Cellular Seismology (CS) method of Kafka (2002, 2007). The percentage of later occurring earthquakes located near earlier occurring earthquakes typically exceeds the expected percentage for randomly distributed later occurring earthquakes, and the specific percentage is influenced by several variables, including magnitude, depth, time, and tectonic setting. At 33% map area coverage, hit percents are typically 85-95% in the NMSZ, 50-60% in AUS, 75-85% in CA, and 75-85% in AK. Statistical significance testing is performed on trials analyzing the same variables so that the overall regions can be compared, although some tests are inconclusive due to the small number of earthquake sample sizes. These results offer useful insights into understanding the capabilities and limits of CS studies, which can provide guidance for improving the seismicity-based components of seismic hazard assessments.