Characteristics of high-resolution marine reflection profiling sources

Abstract

Midwater signatures of eight common types of marine seismic sources have been recorded and characteristics of importance for high-resolution reflection profiling have been analyzed. Analyzed characteristics include wavelet shape, peak frequency, bandwidth, repeatability and directivity. Digitization rates required to describe the signatures before any processing and after whitening deconvolution were determined. It was intended that most of the broad range of available source types be represented. Toward that end, both resonant and impulsive types were studied; the impulsive types being both electric-discharge and pneumatic. They included a conventional sonar transducer, a chirped sonar transducer, a boomer, a plasmagun, a multi tip sparker, a watergun, an airgun and three sizes of sleevegun. It was observed that the peak frequency of each conventional sonar is similar to its nominal value and that of the chirped sonar is about the midpoint of its nominal sweep bandwidth. The dominant frequencies of the electric-discharge impulsive sources are about an order of magnitude higher than those of the pneumatic impulsive sources. Among impulsive sources, the boomer has the highest peak frequency and the watergun has the broadest bandwidth. The sleevegun and airgun were found to be comparable, both having low frequencies, poor repeatability and weak directivity. The conventional sonar and the boomer are the most repeatable sources. The boomer shows the strongest directivity with the chip sonar and the watergun being the most nondirectional. Sample rates required to describe the unprocessed signatures vary from 2 to 88 times the peak frequency and from 1 to 4 times the bandwidth 40dB below peak power. The required rate depends largely on source type but can also be affected by energy level. Bandwidth seems to be a more stable indicator than peak frequency. Rates required to describe the results of applying whitening deconvolution to signatures were about 1.5–3 times higher than rates calculated for the unprocessed signatures. This leads to the conclusion that the digitization rate required in a specific situation can depend as much on the intended processing scheme as it does on source type.