In recent years channel-wave seismology has shown itself to be a useful aid to coal mining. Small geologic faults which interrupt a coal seam can seriously disrupt mining by modern longwall methods. Such features have been accurately predicted at distances of several hundred meters by underground seismic surveys.
The useful range of the method in a particular seam depends critically on the rate at which channel waves in the frequency range of interest (50 to 500 Hz) are attenuated during passage through the coal; when no faulting is located by a given survey, it is always important to know at what range a fault would have been detected had it been present. To this end, the attenuation characteristics must be studied.
Frequently, coal seams are highly anisotropic wave-guides. Perpendicular to the bedding plane one or two “cleat” or cleavage planes exist, along which the coal is extensively cracked and jointed. Recent work by Crampin (1978, 1981) and by Hudson (1981) has shown that such systems of parallel cracks in crustal rocks lead to anisotropic propagation of body waves, with both velocity and attenuation varying considerably as a function of angle of travel through the cracked medium. The polarization of seismic energy propagating through such media can no longer be defined as simple shear or compression. It is reasonable to expect that guided channel waves in coal seams will exhibit similar properties.
A number of transmission surveys have been carried out to investigate both attenuation and anisotropy. The principal requirement for such surveys is that in-seam shot and geophone arrays be designed to provide a large range of both source-detector distance and angle of travel through the coal. Given this condition, anisotropy of group velocity can be detected by adapting Dziewonski's multiple filter analysis of single traces to transmission stacking over selected raypath angles (Dziewonski et al, 1969); attenuation similarly can be investigated by a least-squares regression of the logarithm of the amplitude spectrum of a particular mode onto the path length.
The methods and results of these investigations are reported here, the significance of anisotropy for the collection and processing of in-seam data is discussed, and consideration is given to the type and size of errors introduced into the position of fault predictions made on the basis of an isotropic model. The equivalence of Fermat's principle and Snell's law for anisotropic raypaths results in a computationally efficient modification to enable our previously reported dynamic trace gathering (DTG) stacking procedure to be used for such data.