Abstract The data set that is analyzed in this study consists of strong motion and digital velocity records obtained during the last 6 yr for earthquakes along faults in Baja California North and southern California. From this set, local magnitudes for 60 earthquakes recorded at stations on hard rock and sediments are calculated following the technique proposed by Kanamori and Jennings (1978). The calculated near-source motion magnitudes ( M LSM ) are then compared with reported local magnitude values ( M L ) which are in the range from 3 to 6.6. From these comparisons, it is found that for earthquakes with magnitude M L between 3 and 5.5, almost all the M LSM values estimated from data registered on sediments exceed the M L values by amounts which at the shorter distances (less than about 10 km) may be as high as an order of magnitude, or even higher for a few cases. The strong motion records associated with these high M LSM values have highly coherent pulse-like S -wave signals of very short period and large accelerations. Similar estimations of local magnitude from more limited data obtained on hard rock were comparable or lower than the standard M L values. We conclude that the high M LSM values are the result of a combination of high-frequency source spectra and a strong shear wave energy amplification due to the presence of sediments. A sediment amplification factor of 3.2 is inferred from the data and is in good agreement with a theoretical factor of 3.4 estimated on the basis of the McMechan and Mooney (1980) velocity model for the Imperial Valley. It is also found from the analyzed data that near-source local magnitude decreases between 0 and 10 km, indicating that for smaller earthquakes the attenuation function, -log A 0 (richter, 1958), used in local magnitude determinations, needs to be modified to take into account the characteristics of the seismic energy radiated at short distances. Near-source strong motion data for the 1979 Imperial Valley ( M L = 6.6) and 1980 Victoria ( M L = 6.1) events do not show the strong discrepancy between M LSM and M L observed for the smaller magnitude events. Factors that might be responsible for this include nonlinear sediment attenuation for larger amplitude waves, as well as complex near-source spreading out of the waveform resulting from larger source size (near-source magnitude saturation). We give a numerical example, using small earthquakes as Green9s functions, to show that for a complex source function represented as a superposition of smaller earthquakes, M LSM and M L for the 1980 Victoria earthquake can be explained by the magnitude saturation phenomenon.
Read full abstract