The useful property of all these fillings of space is that the spacing between the grid points is basically constant. Two-dimensional ``pixels'' have about equal area, likewise three-dimensional voxels have about equal volume.

Although mesh points are exactly evenly spaced along the helix and approximately evenly spaced in the crossline dimension, unfortunately, the angles between neighboring points are continuously changing.

I hope I am wrong, but this appears to defeat the many useful tricks we play in cartesian space such as using autoregression for multidimensional prediction and using prediction-error filters to characterize multidimensional spectra. Thus we cannot interpolate by expressing physical stationarity as stationarity of filter coefficients.

4/20/1999