Equations for determining amplitude magnitude (MX) and duration magnitude (MF) that employ all calibrated instruments in the USGS short-period telemetered seismic network in northern California (NCSN) were developed and tested against a set of 1276 earthquakes from 1986 to 1990 that were analyzed on the Caltech-USGS processing system (CUSP). The expressions for decay of amplitude and record duration in these equations are functions of distance alone. Sensitivity corrections for both MX and MF are simply the logarithms of the ratios of the magnification of the reference instrument to that of the instrument actually used. Component corrections were chosen so as to minimize the dependence of instrument site residuals on instrument component. MF site residuals were found to be closely linked to MX site residuals in a manner that suggests both depend primarily on site amplification. Both MX and MF site residuals vary systematically with bedrock lithology: older well-consolidated rocks produce negative residuals (smaller amplitudes and shorter durations) and younger unconsolidated rocks produce positive residuals (larger amplitudes and longer durations).
Average station magnitude residuals are virtually independent of distance from the epicenter to at least 800 km; and MX-MF, averaged over 0.5 unit magnitude intervals, is less than 0.05 from M0.5 to M5.5.
Comparison of MX and MBK (ML, UC Berkeley) for 293 events in both the CUSP data set and the Berkeley catalog shows that calculated MX s are marginally larger than the corresponding MBK s. MX-MBK averages about + 0.04.
The characteristics of the standard Wood-Anderson seismograph employed to calculate MX are: free period 0.8 sec, damping constant 0.8, and static magnification 2080.