Magnitude Bias in Global Bulletins

Jack Murphy and Brian W. Barker, Maxwell Technologies

While the scatter is large, there is a clear and significant offset between the m_b values of the Prototype International Data Center (PIDC) and those of the US Geological Survey (USGS) National Earthquake Information Center (NEIC), with the IDC values being lower by about 0.3 magnitude units. In fact, the offset is even larger than it appears here in that the NEIC began incorporating PIDC data into their published m_b estimates as the Group of Scientific Experts Technical Test-3 (GSETT-3) experiment progressed (a practice which has since been discontinued by USGS) and this resulted in a systematic decrease in the NEIC m_b values by about 0.1 magnitude units during the period of 1995 to 1996. Thus, the actual offset between the PIDC (and now IDC) and USGS NEIC historical m_b values is about 0.4 magnitude units.

PIDC and USGS NEIC mb values for 7,026 events from 1995 and 1996.

Although there is currently no consensus on all the causes of this offset, the following contributions have been identified.

(1) The PIDC/IDC employs the Veith/Clawson correction factors for the effects of epicentral distance and focal depth on m_b, while NEIC employs the original Gutenberg/Richter corrections. The difference is further complicated by the fact that the IDC default focal depth is zero, while the NEIC default focal depth is 5, 10 or 33 km, depending on event

location. It has been demonstrated that the combination of effects due to these two differences contributes about 0.1 magnitude units to the observed offset.

(2) During the time period represented, single station data reported to the USGS by Australian (and perhaps other) stations corrected maximum trace deflections to ground motion using the nominal instrument gain at 1 Hz rather than the gain at the measured period of the peak. The difference led to overestimation of the single station m_b values by more than 0.2 magnitude units on average at a number of stations. Although the Australians have subsequently modified their data reduction procedures and the USGS has demonstrated that this practice was not widespread among the stations reporting to NEIC, Australian station data contributed nearly 20% of the non-IDC station data used by the USGS to estimate the m_b values in the figure. As a result, the difference contributed to the m_b offset shown.

(3) The primary and array stations that contribute to the IDC were carefully selected to be low noise stations. Extensive statistical analyses have now indicated that they are also low signal stations. Application of station corrections derived with respect to large network (IMS) m_b averages produces an average increase in the IDC m_b values by about 0.2 magnitude units.

(4) It can be seen from the figure that the observed offset between the two magnitude measures seems to increase somewhat above about m_b = 5. It is believed that this shift is primarily due to the fact that the IDC P-wave maxima are automatically determined from the first 6 seconds of the initial P-wave signal, while the USGS values are defined as the maxima in the whole P-wave group. This can lead to significant differences for large, slowly emergent earthquakes for which the maximum P-wave amplitude may occur later than 6 seconds after P-wave offset.

Additional Information

For additional information, see Murphy, J. R. and B. W. Barker (1996), "Preliminary Evaluation of Seismic Magnitude Determination at the International Data Center (IDC)" Presented at the 1996 AGU Fall Meeting, 15 December 1996.