Windowless Q-factor tomography by the instantaneous frequency

The estimation of the Q factor of rocks by seismic surveys is a powerful tool for reservoir characterization, as it helps detecting possible fractures and saturating fluids. Seismic tomography allows building 3D macro-models for the Q factor, using methods as the spectral ratio and the frequency shift. Both these algorithms require windowing the seismic signal accurately in the time domain; however, this process can hardly follow the continuous variations of the wavelet length as a function of offset and propagation effects, and it is biased by the interpreter choice. In this paper, we highlight some drawback of signal windowing in the frequency-shift method, and introduce a tomographic approach to estimate the Q factor using the complex attributes of the seismic trace. We show that such approach is particularly needed when the dispersion is broadening the waveforms of signals with a long wave-path. Our method still requires an interpretative event picking, but no other parameters as the time window length and its possible smoothing options. We validate the new method with synthetic and real data examples, involving the joint tomographic inversion of direct and reflected signals. We show that a calibration of the frequency-shift method is needed to improve the estimation of the absolute Q factor, otherwise only relative contrasts are obtained.