Source Time Functions of the Dprk Announced Nuclear TESTS of 2006, 2009 and 2013
Thursday, 18 December 2014: 5:15 PM
The concept of source scaling is used to determine the source time functions of the three announced nuclear tests of 2006, 2009 and 2013 of the Democratic People’s Republic of Korea (DPRK). Two explosions at essentially the same source location, but at different times, have scaled source time functions. In an x-y plot of these functions, the ground motion axis (y) and the time axis (x) scale by the same factor, which is the cube-root of the energy ratio. The seismograms from these two events, measured at a given seismometer, share the same path effect, or Earth impulse response. The two seismograms and the scaling law provide three equations that can be solved for the three principal unknowns: the two source time functions and the Earth impulse response. This concept is applied to seismograms recorded at Mudanjiang (MDJ) for the three DPRK events, using two novel steps. First, the spectral ratio of the two source time functions is calculated in the time domain directly from the seismograms. Second, the scale factor is calculated directly from the data. The energy partition is independent of event size. So the ratio of the energies of the P-waves, for example, is equal to the ratio of the source energies. The ratio of the energies of the Pn phase was computed for the 2009 and 2013 events. The scaling factor is the cube-root of this ratio. The source time functions of the 2009 and 2013 events were recovered using the two seismograms and the scaling law, and the seismograms were deconvolved for these source time functions to recover the Earth impulse response. If everything is done correctly the recovered Earth impulse response should be the same for each event. I show that the recovered impulse responses for the 2009 and 2013 events are nearly identical. Presuming the DPRK test site to be in granite, relative reduced displacement potentials for the three events give yields of 0.11, 1.98 and 7.11 kt, with errors up to 46 %.