Coda duration magnitudes in central California: An empirical approach

Abstract

A new empirical coda magnitude M D is presented for the central California seismic network (CALNET), in which station corrections are introduced to reduce the influence of hypocenter and station distributions. The new magnitude scale is obtained by relating the lapse-time τ^ to local magnitudes M L using data from 55 earthquakes with 1.1 ≦ M L ≤ 5.6 from June 1977 through 1981. Magnitude estimates are significantly improved by introducing corrections for the seismograph instrument attenuation setting α ij and a site correction δ j . The magnitude of the i th event 〈 M D 〉, is the median value of individual M D ij = -1.03 + 2.10log 10 τˆ ij + 0.00268 τˆ ij + α ij + δ j , where τ^ ij is the lapse-time measured at the j th station. Local site effects influence the duration of the coda and cause stations to either overestimate or underestimate 〈 M D 〉, typically up to 0.5 magnitude units. The distribution of the site corrections is spatially correlated, changing coherently across major tectonic and geologic features such as the San Andreas fault. It appears that the site corrections are influenced by the physical attenuation properties near the station, seemingly contradicting the concept that the coda is related to back-scattering and attenuation in a volume with dimensions on the order of the hypocenter to receiver distance.