Values of local magnitude ML, are calculated from 56 strong-motion accelerograms recorded in the Imperial Valley earthquake of 15 October 1979 according to procedures developed earlier (Kanamori and Jennings, 1978). These data, plus similar data from the San Fernando earthquake of 9 February 1971 and additional, less numerous data from several other California earthquakes, are used to investigate the use of different measures of distance in near-field determinations of ML: this investigation has relevance for similar uses of distances in determining seismic design criteria. In addition, the consistency of the values of ML found from the strong-motion data is examined from the viewpoint of assessing the need for any correction in the standard attenuation curve, −log10A0(Δ).
It was found that the most consistent values of ML result when distance is measured to the closest point on the surface trace of the fault if a site lies within a circle with diameter equal to the extent of faulting and centered on the center of faulting (center of energy release). Outside this circle, the distance measured to the center of the circle is recommended.
A consistent trend in the values of ML found from strong-motion records is seen in the data. The values start, at zero distance, at essentially the far-field value and then decrease to −1/4 unit at about 20 km. Then they rise to +1/4 unit at 50 to 60 km. A smooth revision to the standard attenuation curve is presented which removes this systematic trend.