Search

Seismic observations of underground nuclear explosions provide crucial data on source yield and depth that cannot easily be estimated from other geophysical methods. However, it is difficult to obtain reliable yield estimates for test sites for which we do not have direct seismic calibration experiments. To obtain source information from uncalibrated sites and paths, local and regional seismic records of six, proximal, declared underground nuclear explosions in North Korea are used to compute spectral ratios of narrow‐band waveform envelopes of body‐wave coda that remove path and site effects to reveal precise, relative source moment. The yields of these explosions are obtained from the observed source ratios by simultaneously fitting the classical source model of Mueller and Murphy (1971), https://doi.org/10.1785/bssa0610061675 to all event pairs. The source model provides an impressive fit to the observations considering that the P phase coda derived source spectral ratios did not require a prior knowledge of the source or regionally calibrated corrections to be applied to the data. However, the observed corner frequencies from S wave coda spectral ratios are lower than the source model predictions, but are well fit by models calculated using the corner frequency consistent with the Fisk conjecture. The results presented here provide novel constraints on the spectral distributions of the energy radiated by the sources of the DPRK test series, and allows for an independent evaluation of existing estimated source model parameters. Plain Language Summary: It is difficult to determine the size and depth of an underground nuclear explosion only using observations of the ground shaking caused by its detonation. Optimally the test site is first calibrated using a "ground truth" explosion with a known depth and size. However, for many sites of interest, such as the location of the six declared North Korean nuclear tests occurring between 2006 and 2017, this information is not available. In order to characterize these uncalibrated underground explosions, we use ground shaking measured by sensors located within ∼1,000 km of the test site to determine "coda spectral ratios." These coda spectral ratios use the measurements of two explosions which were recorded at the same seismic sensor, and both occurred at roughly the same location, to remove the changes in the wave caused by the rock, as the wave traveled through the Earth's crust from the site of the explosion to the location of the sensor. The resulting measurement is an estimate of the amount of energy created by the explosion for each frequency of vibrational wave that radiates out from the explosions. The size of these nuclear tests is then determined by fitting physical models of explosions to the derived data. Key Points: The P wave coda spectral ratios match the predicted source spectral ratios calculated using the Mueller and Murphy (1971), https://doi.org/10.1785/bssa0610061675 source model (MM71)The observed S wave corner frequencies are consistently lower than those predicted by MM71, and provide support for the Fisk conjectureThe coda spectral ratios obtained in this study are used to estimate new independent yield estimates of the North Korean tests [ABSTRACT FROM AUTHOR]