Near-surface Q estimation: an approach using the up-going wave-field in vertical seismic profile data. Michelle Montano*, Don Lawton and Gary Margrave, CREWES, University of Calgary. Summary Understanding the wavelet evolution with depth is the key to estimate Q values in the subsurface. Knowing these values is important to enhance the vertical resolution of seismic data, improve seismic-well ties and can also be used for reservoir characterization. The direct down-going wavefield recorded in VSP data is the conventional input for Q estimation. However, estimation for the shallow layers may be problematic. In this study, we show that the up-going wavefield is an alternative to have more reliable estimations especially in the near-surface layers. Combining both estimations from the down-going and up-going wavefield provides the optimum understanding of Q variation with depth. Q values are estimated from synthetic VSP down-going and up-going wavefields by using the dominant frequency matching method. We also estimated Q from field VSP data by using the spectral-ratio method as well as the dominant frequency matching method. From the up-going wavefield, we obtained that QP values range from 20-28 from 66-250m depth. For the deeper layers, using down-going wavefield, the estimated QP values range from 51-61 from 250-500m depth. In comparison, we used the direct down-going shear wavefield for QS estimation and found values ranging from 21-34 from 200-420m depth. Introduction Estimating Q from the shallow down-going wavefield is a challenging task because the receivers are close to the source, S, (Figure 1a). This often causes an oversaturation in the amplitudes that may be problematic. Also, the wavefield has propagated for a short period of time and we may not see significant attenuation when we process the seismic data. However, shallow layers are expected to show low Q values because poorly consolidated rocks are usually present. One way to approach this problem is using the up-going wavefield to estimate Q in the shallow zone. By assuming that the source, S*, is at the reflecting interface, the receivers located in the shallow zone will be far from it (Figure 1b) and more reliable estimations could be obtained. Figure 1: (a) Down-going waves and (b) up-going waves propagating to the borehole receivers. Q estimation theory Spectral-ratio method If we consider two wavelets at times t1 and t2, in which t1 < t2, their amplitude spectra will be the following: | ( 1 , )| = | ()| − 1 . (1) | ( 2 , )| = | ()| − 2 . (2) Then, the log spectral-ratio, lsr, is the ratio of equations 1 and 2 (Margrave, 2013), (, Δ, ) = | ( 2 ,)| | ( 1 ,)| =− Δ , (3) where Δ = 2 − 1 . Equation 3 shows that lsr has a linear relationship with frequency. The interval Q between t1 and t2 can be computed by a least squares fit of a first order polynomial. Note that noise and also notches can be a problem for the spectral division. Dominant frequency matching method Quan and Harris (1997) first introduced a frequency-shift method to estimate Q in which they defined the centroid frequency of the input signal () and the output signal () as = ∫ () ∞ 0 ∫ () ∞ 0 , (7) and, R = ∫ () ∞ 0 ∫ () ∞ 0 . (8) S R (a) S S* R (b)