AbstractWe investigated the effect of the radiation pattern on station and network magnitude estimates (mb) by using centroid-moment-tensor (CMT) solutions published by the seismological group at Harvard University. The magnitude data used in this study were P amplitudes provided by the Prototype International Data Center (PIDC). When relating the station magnitude residuals to the calculated relative radiation factors, we found a small but clearly observable dependency. By compensating for this effect after inverting for station corrections and new event magnitudes, we found that the variance of the station magnitude residuals was reduced by about 2.6%. In parallel, we estimated a set of mean station corrections with and without corrections for the radiation pattern. Similarly, we recomputed the average network magnitudes (mb), which in some cases differed by up to 0.15 magnitude units from the uncorrected average network mb values. In conclusion, we found that the effect of the source radiation pattern on the amplitudes used for mb estimation on a global scale is relatively small compared to effects from other factors. Only about 24% of the predicted radiation pattern signal can be found in the observed data. However, when best possible mb values are needed (e.g., to investigate magnitude relations or station corrections for specific ray paths), compensating for the radiation pattern will help to reduce the scatter. |