** Next:** Time-distance functions compared
** Up:** Spectral factorization of seismic
** Previous:** On the assumption of

Figure shows a cube of raw velocity data from the
MDI instrument.
The data have been transformed to Cartesian coordinates by
projecting high-resolution data from an area approximately
18 square onto a tangent plane.
The object in the center of the time-slice (top of cube) is a
sun-spot (irrelevant to this study).
The sample spacing is 1 minute on the time-axis and approximately
825 km on the two spatial axes.
**tallcube
**

Figure 2 Cube of MDI data.

Time-variable features of Figure fall into two
distinct spectral windows.
The low temporal frequency events (<1.5 mHz) are related to
solar convection, while the higher frequency events
are related to acoustic wave propagation.
We were interested in studying acoustic wave phenomena; so as a
preprocessing step, we removed the lower frequency spectral
window by applying a low-cut filter to the data.

Figure shows the impulse response derived by
multi-dimensional spectral factorization. Due to the lack of coherent
reflectors within the solar interior, no reflection events are
visible. The strong vertical velocity gradient causes the first
arrival to be a diving wave. Later arrivals are the multiple events
of increasing order (PP, PPP etc.) illustrated in
Figure .

**kolcube
**

Figure 3 Three-dimensional minimum phase
impulse response.

**multiples
**

Figure 4 Schematic showing some of the
raypaths visible in helioseismic time-distance plots.

** Next:** Time-distance functions compared
** Up:** Spectral factorization of seismic
** Previous:** On the assumption of
Stanford Exploration Project

5/27/2001