Seismograms of 320 earthquakes (1,486 observations) from short-period seismometers occurring from January 1969 to April 1971 and 91 earthquakes (257 observations) during 1971 have been used to establish a relationship between total signal duration and the local Richter magnitude for the CIT and BHSN telemetered seismic networks in southern California. The data have been fitted using regression analysis to relationships of the form
where τ is the total duration in seconds and Δ is the epicentral distance in kilometers. These relations explain up to 88 per cent (CIT) and 94 per cent (BHSN) of the variation in the data and yield magnitudes having standard deviations as low as 0.15 (CIT) and 0.14 (BHSN) magnitude units. It has been found that the local magnitude based on signal duration is relatively insensitive to variations in azimuth and source effects.
In view of the limited distribution and low magnifiation of the Wood-Anderson torsion seismometer, and the previously recognized problems of “saturation” and instrument response associated with the amplitude technique, it is concluded that the method of duration applied to vertical short-period seismograph records will greatly improve the assignment of local magnitude to earthquakes in the southern California region.