Ambient Noise Refections: Interferometry

Here we show preliminary results attempting to find and use the reflection portion of the ambient noise correlation functions. The noise distribution preferentually generates horizontally traveling energy, so refections and refractions tend to be very weak.
The images beloware very detailed (big) and can be a bit slow in loading.

A stack of all virtual sources. All 5200 virtual source are summed together in the above image. It shows there are three well-resolved phases. The horizonatal axis is distance in km. The vertical axis is time in sec. Both positive and negative time are shown. The arrival nearest the horizontal axis is the P-wave (fastest arrival). The slowest arrival is the fundamental-mode Rayleigh wave. The multiple strands are likely caused by stacking over many sources. In between the P and Rayleigh wave is likely a higher-mode Rayleigh wave and possibly the S-wave. P-wave detail. The image show the detail of the P-wave taken from the figure to the left. Note the curvature of the arrival, which means the wave is "diving" and not just traveling along the surface. This means the correlation is reconstructing some aspects of the vertical paths. The P-wave velocity varies from 2.0 Km/sec to 2.86 km/sec. The multiples are strands are likely caused by stacking over multiple sources.

To make a stacked section, we first try isolate a region that has wide-offset and avoids the refracted P-wave and surface waves. Thus we mute the date in a region shown below. The logical in choosing wide-offset is to select the portion of the reflections that are more likely to generated by ambient noise correlations, because they have more of a horizontal component to their raypath.

Mute Window. The white lines mark the zone where data are stacked. The mutes avoid the P-waves above and the surface-waves below. Stacked section. The result of stacking the virtual mid-point gathers within the mutes show to the left. There is some apparent energy in the stack, but we have not verified that it is real.

We have also looked into forming an image using the data in a cone under the surface wave. We can for example make a zero-offset section by taking the auocorrelations of the traces. In the example below, we have stacked the autocorrelations in 200m-radius bins along a N-S line approximately in the center of the array. The results were deconvolved with a predictive deconvolution operator to remove some of the ringing. There is an apparent reflector on the left side about 4 sec deep.

Here we show preliminary results attempting to find and use the reflection portion of the ambient noise correlation functions. The noise distribution preferentually generates horizontally traveling energy, so refections and refractions tend to be very weak. The images beloware very detailed (big) and can be a bit slow in loading.

A stack of all virtual sources. All 5200 virtual source are summed together in the above image. It shows there are three well-resolved phases. The horizonatal axis is distance in km. The vertical axis is time in sec. Both positive and negative time are shown. The arrival nearest the horizontal axis is the P-wave (fastest arrival). The slowest arrival is the fundamental-mode Rayleigh wave. The multiple strands are likely caused by stacking over many sources. In between the P and Rayleigh wave is likely a higher-mode Rayleigh wave and possibly the S-wave.	P-wave detail. The image show the detail of the P-wave taken from the figure to the left. Note the curvature of the arrival, which means the wave is "diving" and not just traveling along the surface. This means the correlation is reconstructing some aspects of the vertical paths. The P-wave velocity varies from 2.0 Km/sec to 2.86 km/sec. The multiples are strands are likely caused by stacking over multiple sources.

To make a stacked section, we first try isolate a region that has wide-offset and avoids the refracted P-wave and surface waves. Thus we mute the date in a region shown below. The logical in choosing wide-offset is to select the portion of the reflections that are more likely to generated by ambient noise correlations, because they have more of a horizontal component to their raypath.

Mute Window. The white lines mark the zone where data are stacked. The mutes avoid the P-waves above and the surface-waves below.	Stacked section. The result of stacking the virtual mid-point gathers within the mutes show to the left. There is some apparent energy in the stack, but we have not verified that it is real.