To explain the discrepancies observed between the explosively induced collapse and some natural collapses, we propose a source process model for the explosively induced collapse. The model is pictured in Figure 8 along with a model for spontaneous collapses for comparison. The basic difference between these two models is the inclusion of phase a) in the explosively induced collapse model. This phase represents the rapid material failure during an explosively induced collapse because of the sudden removal of the supporting pillars. This failure process, or crack forming process, seems to be an effective seismic source.
Figure 9 shows the second time derivative of M33 which corresponds to the force representation of the source. It is similar to the force time function of a spall accompanying an underground explosion. We modeled the observed data with a simple spall source.
Figure 10 shows that the fit is close especially for the early parts of the long period signal.
Figure 11 is the isotropic spectrum of the source moment rate tensor. Unlike typical explosion source spectrum, we see a peaked spectrum for the induced collapse. In order to validate our observations, we analyzed the data from another explosively induced collapse at the same mine. The source configurations of both collapses were similar.
Figure 12 shows the moment tensor results. Although the source also starts as expansive, the time histories are more complex. Cross-correlation between the regional P waves from the two collapses indicates that this collapse might be a two event collapse. This example demonstrates the complexity of the problem.