We demonstrate a method that derives a more accurate measurement of ultrasonic attenuation by using sweep-type signals than by using impulse-type signals. We obtained spectral amplitude of the sweep signal in frequency-time domain using the continuous wavelet transform (CWT) and estimated attenuation in the time-frequency domain using the spectral-ratio method. The advantage of this method is independent on the effect of windowing. We used partially frozen brine as a solid-liquid coexistence system to investigate attenuation phenomena. Ultrasonic wave-transmission measurements on an ice-brine coexisting system were conducted to examine the influence of unfrozen brine in the pore microstructure on ultrasonic waves. We observed the variations of a 100–1000 kHz sweep signal transmitted through a liquid system to a solid-liquid coexistence system, changing its temperature from 20°C to –15°C. We quantitatively estimated attenuation in a frequency range of 350–600 kHz by considering different distances between the source and receiver transducers. Finally we demonstrated the possibility of sweep signal to estimate attenuation.
Attenuation is a fundamental property of rocks as it controls the decay in amplitude of seismic waves and the accompanying change in frequency. Attenuation is strongly dependent on lithology, pore fluid properties, and transport properties. There are many existing methods of estimating attenuation: the spectral ratio method (Gladwin and Stacey, 1974), the amplitude-decay method (Badri and Mooney, 1987), the rise-time method (Gladwin and Stacey, 1974), the centroid frequency shift method (Quan and Harris, 1997), wavelet modeling (Jannsen et al., 1985), the pulse-broadening method (Hathley, 1986), and the inversion method (Amundsen and Rune, 1994). For attenuation analysis, it is often necessary to extract the first few cycles of the direct arrival waveforms in order to isolate them from unnecessary events. However, the effect of windowing degrades the attenuation estimates, and the estimates are dependent on window length (Sams and Goldberg, 1990).
Sweep-type signal is the most extensively used land seismic exploration technique. In conventional sweep-type data processing, a received trace is cross-correlated with source sweep to convert the extended sweep signal into an impulse signal. Sun and Milkereit (2006) utilized uncorrelated sweep data in order to measure velocity dispersion and attenuation. Sun and Milkereit (2006) developed a time-domain method, the so-called crosscorrelation with a moving window (CCMW) method that obtains the time-frequency relation by crosscorrelating the received sweep with a portion of the source sweep, tapered with a time window which centers at a known frequency. Sun and Milkereit (2006) also investigate different approaches to obtaining the time-frequency relation, such as the cross-spectrum method (Donald and Butt, 2004) and the time-frequency decomposition method. In the time-frequency decomposition method, the difference between time-frequency (time-amplitude spectrum) relations of the source and the received sweeps gives frequency-dependent travel times or amplitude ratios. Sun and Milkereit (2006) also indicate that the time-frequency decomposition method is not suitable for field data because this method handles ambient noises very poorly and the results are not stable. Nevertheless, the time-frequency decomposition method can be utilized to estimate attenuation with the spectral ratio method in the condition of low-noise.
