Seismic spectral models for chemical and nuclear explosions are used in many applications including network modeling and yield estimation. Here we compare the models presented in Denny and Johnson (1991) and Mueller and Murphy (1971) with each other and with new results from the Source Physics Experiments (SPE). We demonstrate analytically the two models are in substantial agreement for large and normally buried explosions, consistent with much of the historic data collected during American and Soviet nuclear testing. However, for small and/or deeply buried explosions, the spectral predictions of the two models can differ significantly. For example, the predicted yield of a 1 km deep, Mw 2 nuclear explosion differs by more than a factor of 5; and, for the same moment and depth chemical explosion, the difference is greater than a factor of 10. We compare the models with initial data from the SPE, which include small and overburied chemical explosions. The corner frequency of the one‐ton SPE explosion (SPE‐2) is slightly higher than the Mueller and Murphy (1971) model and approximately double the Denny and Johnson (1991) model prediction. The absolute moment of the one‐tenth ton SPE explosion (SPE‐1) is near the Denny and Johnson (1991) prediction and an order of magnitude smaller than the Mueller and Murphy (1971) prediction. The low‐frequency moment ratio for SPE‐2/SPE‐1 is more consistent with the Denny and Johnson (1991) model. The results presented here show the need for an improved explosion source model that can accommodate a wider range of yields and emplacement conditions.
Online Material: Moment magnitude, corner frequency, and yield for all geologic media.