The robustness of seismic moment and magnitudes estimated using spectral analysis

calculate seismic moment. This is an important topic for operators and regulators who require good magnitude estimates when monitoring induced seismicity. It is therefore imperative that these parties know and understand what errors exist in given magnitude values, something that is poorly reported. This study concentrates on spectral analysis techniques and compares inline image computed in the time and frequency domains. Using recordings of inline image events at Cotton Valley, east Texas, the maximum discrepancy between inline image estimated using the different methods is 0.6 units, a significant variation. By adjusting parameters in the inline image calculation we find that the radiation pattern correction term can have the most significant effect on inline image, generally up to 0.8 units. Following this investigation we make a series of recommendations for estimating microseismic inline image using spectral methods. Noise should be estimated and removed from recordings and an attenuation correction should be applied. The spectral level can be measured by spectral fitting or taken from the low frequency level. Significant factors in obtaining reliable microseismic inline image estimates include using at least four receivers recording at ⩾1000 Hz and making radiation pattern corrections based on focal mechanism solutions, not average values.