Earthquake Catalog Completeness

Complete earthquake catalogs are often required in studies of seismicity. This assessment is not simple because seismicity varies in both time and space. Several issues and problems related to estimating the completeness of earthquake catalogs will be discussed. We investigate two specific methods of analysis for completeness: (1) a method based on the assumption that the Gutenberg-Richter, or b-value, curve is linear and (2) a method that relies on the day-to-night modulation of the noise threshold. In the first method departures from a linear fit to the b-value at small magnitudes are believed to be due to missed events because of an incomplete catalog. In the second method it is assumed that the number of recorded events will exhibit a day-to-night modulation as the seismic detection threshold approaches the noise background. Both methods have implicit assumptions and limitations associated with their proper use and reliability. In particular, at low magnitudes the effects of swarms and quarry blasts must be taken into account. The completeness levels of seismic catalogs from Hokkaido, Japan and Parkfield, California will be investigated using these methods of analysis. A comparison of results shows that there may be a significant difference in the level of completeness obtained from these methods.