Abstract
In many cases, the seismic response of bottom-simulating reflectors is characterised by low frequencies called low-frequency shadow. Generally, this phenomenon is interpreted as attenuation due to partial saturation with free gas. Actually, this frequency loss may have multiple causes, with a normal moveout stretch as a possible candidate. To analyse this phenomenon, we compute synthetic seismograms by assuming a lossy bottom-simulating layer, with varying quality factor and thickness, bounded by the upper hydrate-brine/gas-brine and lower gas-brine/brine interfaces. First, we estimate the shift of the centroid frequency of the power spectrum as a function of the travelled distance of the seismic pulse. Then, we perform one-dimensional numerical experiments to quantify the loss of frequency of the seismic event below the bottom-simulating reflector as a function of the quality factor of the bottom-simulating layer and its thickness (due to wave interference). Then, we compute shot gathers to obtain the stacked section, with and without the normal moveout stretch correction and with and without the presence of wave attenuation in the bottom-simulating layer. The results indicate that the low-frequency shadow due to the normal moveout stretch is stronger than that due to attenuation and may constitute a false indicator of the presence of gas. In fact, often, the low-frequency shadow overlies events with higher frequencies, in contradiction with the physics of wave propagation. This is particularly evident when the low-frequency shadow is so extensive that the presence of high frequencies below cannot be justified by the acquisition geometry.