The 2006 North Korean nuclear test - analysis of the explosive component using moment tensor inversion based on a 3D earth model

Full text: Due to its low yield, the mb 3.9 North Korean nuclear test on 9 October 2006 could not be verified undoubtedly using solely seismic methods. The low signal-to-noise ratio (SNR) limits conventional discrimination methods such as P/S amplitude-ratios, and thus we perform a moment tensor inversion (MTI) to estimate the explosive (isotropic) component of the event. Long-period recordings are mainly controlled by the seismic source mechanism, while at shorter periods variations of the crustal structure have a stronger influence on the waveforms. On the other hand, long periods have a lower SNR than shorter periods. To obtain a good SNR for source-sensitive periods, we use an intermediate pass-band (T = 10-50 s) and include a regional 3D velocity structure deduced from the Crust 2.0 model to calculate synthetic waveforms. We study the influence of the crustal model on the inversion result using regional recordings with source-receiver distances up to 3000 km and compare the results with a tectonic mb 4.0 event that occurred in a distance of 34 km on 16 April 2002. We observe a higher isotropic part (Miso) for the nuclear test compared to the tectonic event. However, the absolute value of Miso cannot be determined because a trade-off with the inversion damping exists. To study the influence of seismic noise on Miso, we add artificial noise to the recordings of the tectonic event. Compared with the MTI of the original data we also find an increased Miso, but with other dependencies on the damping compared to the nuclear explosion data. We show trade-off curves of damping versus model length (as a function of MW) that can be a helpful tool for a timely verification of nuclear tests. (author)