Our seismic explorations have led us into a detailed examination of the propagation of seismic waves in the earth. Lamb’s 1904 theoretical solution to the problem for a homogeneous isotropic earth did not, as he pointed out, agree with a typical seismogram.
A typical seismogram appears to be the result of a series of unrelated motions. Small changes in shot or reception point appreciably alter the character of the seismogram. Some form of scattering could account for this. Yet a series of observations with multiple seismometers showed patterns of radiation apparently coming from the source. This led to the hypothesis that there was efficient conversion of body waves into surface waves. A ground-motion study seemed to corroborate this in that it showed Rayleigh motion directly after the compressional and before the time of a shear arrival.
Seismic model experiments this hypothesis. In addition, it was shown that not only do compressional waves produce surface waves at a surface discontinuity, but the converse also holds. The process seems efficient.
We conclude that the seismogram is the result of extremely complex interactions of waves with the uneven earth’s surface and interior scattering in an inhomogeneous nonisotropic earth.
There is at present no simple way in which these difficulties may be circumvented. A relatively large number of seismograms in various regions have been obtained. Different techniques, for example multiple and vectorial seismograms, have been employed. The shot position was changed when feasible, up to 30 km.
Straight lines through the travel-time plot are idealized average representations. These travel-time relationships indicate average compressional velocities of 6 km/sec in the upper crust and 8 km/sec at the outer mantle. In addition, there is a system of extremely prominent second arrivals forming a cusped travel-time curve.
These basic data can be accounted for by two simple crustal models with two types of lower crustal transitions. On the one hand, the compressional velocity of 6 km/sec increases slowly with depth to about 7 km/sec at 30 km depth (under Maryland). At this depth there is an abrupt (less than 0.5 km) increase to 8 km/sec. On the other hand, the velocity of 6 km/sec remains about constant with depth to about 24 km depth; in a zone of about 6 km depth the velocity increases at an increasing rate to 8 km/sec at 30 km depth.
This transition is deeper (40–45 km) under Minnesota and Tennessee where the surface elevation is higher but is only 30–35 km under the Colorado Plateau. This transition might well mark the lower crustal boundary—the Mohorovičić discontinuity.
No strong second-arrival system was observed in the states of Washington or California. It is concluded that the lower crustal transition there is more broken or more confused.
These are the simplest crustal models. The data, precise as they are, allow the admission of certain complicating hypotheses owing to the very complex nature of the earth and its surface. The velocity may increase or decrease with depth (in certain restricted ways), but the average characteristics must adhere closely to the simple models.
The choice of a more complex model must await newer methods or interpretations of seismic data. We have found no evidence of systematic intermediate crustal layering.
