The original array recording made by Rob Williams of the USGS is plotted at the top. It shows a predominance of energy propagating northward across the array, from an Interstate highway a mile to the south. The tilt of the Rayleigh-wave arrivals across the array suggests a propagation velocity of 155 meters per second (or a slowness of 1/155 s/m). Given that this recording was made on Mud Island in the Mississippi River across from Memphis, such a slow velocity is not unexpected.
For the left channel the array recording is summed straight across all channels at the same time. If a wave were to hit all sensors of the array simultaneously from far to the side of it, or from far directly below, this flat-stack, zero-slowness beam forming procedure would emphasize such a wave.
For the right channel the beam is formed for the 155 m/s velocity from the south, and its 1/155 s/m tilt. Since this is close to the tilt of most of the ambient noise, the right channel dominates the recording. One could try a range of tilts to see which emphasized the channel the most, and thus extract the predominant propagation velocity from the array.
(164 kb) Beam-formed microtremor array recordTime increases toward the right in these plots: