To better visualize how the FMR boundaries intersect with the extraction
spheres, I have found it useful to create an \emph{extraction map}.
Each map is labeled by a
triple of parameters $(N,R,\Delta)$, which indicate, respectively, the
number of (cell-centered) points across one coordinate direction, the
extraction radius, and the half-width of the extraction shell.  The mesh
accurately portrays the grid refinement in each region, and the extraction
region is denoted by three circular lines; the middle line is the extraction
radius, and the two surrounding lines are the edges of the extraction shell.
Note that the extraction shells generically pass through mesh refinement 
boundaries.  One should also bear in mind that these two-dimensional
slices depict the extraction \emph{sphere} at its largest cross section
and that the refinement regions are cubical.

The overall quality of the waveforms extracted from the simulation can also
be verified by comparing results extracted at several distinct radii.  Because
the dominant $r$ dependence of $\Psi_4$ goes like $r^{-1}$, the easiest
way to compare multiple extractions is to shift the waveform in space and 
scale it by $r$ so that the different waveforms should lie nearly on top of 
each other; this agreement will not be perfect even in the
analytic solution since $\Psi_4$ also has terms that scale with higher
negative powers of $r$.
The wave extraction figures show both the raw 
$\Psi_{4,20}$ data (panel (a)) extracted at 
$3\lambda$, $4\lambda$, $5\lambda$, $6\lambda$, and $7\lambda$, 
as well as that same data shifted to $r=3\lambda$ and scaled up by $r$.
It is clear from panel (b) that the wave is faithfully propagated and extracted
at all radii considered.

References:
C. W. Misner, "Spherical harmonic decomposition on a cubic grid,"
Class. Quant. Grav. 21 (2004), S243.

D. R. Fiske, "Numerical studies of constraints and gravitational wave
extraction in general relativity," Ph.D. Dissertation, University of
Maryland, College Park (2004).