On Fri, 25 Jul 2008, Malcolm J. Currie wrote:
>>> What I *really* want to do is set this quality flag "badbol" for all
>>> data points _except_ for those defined by the ndf section badbol(16,20,)
>
> Had to leave before I could test this last night, but
>
> DIMENSION(3)
> COFRAME(PIXEL)
> .NOT. BOX( 16, 20, 0, 1, 1, 255 )
>
> worked for me. The first triple is the centre of the box; the second
> triple specifies the length of the box sides in pixels. Thus the "1, 1"
> picks out the single bolometer or the single spatial pixel given by the
> first-two arguments.
>
> You might need to change two of the BOX arguments. The 255 and 0
> specifies 255 z-axis (spectral) pixels centred at pixel co-ordinate 0.
> Now my cube only had 151 pixels along the spectral axis and ARD didn't
> complain. It appears therefore you have some latitude in the third and
> sixth arguments to ensure you capture the whole spectrum.
>
>
> Do we need to extend the ARD keywords for cubes? At present POLYGON,
> COLUMN, and ROW only apply in two dimensions. Should we add a new
> keyword BAND to specify a band in Z (in our case a spectrum) equivalent
> to COLUMN for X and ROW for Y?
Either that or we need a method to do something like:
DIMENSION(2)
COFRAME(PIXEL,NAXES=2)
.NOT. PIXEL(16,20)
with a way to pick which of the 2 of 3 axes match. Not so crazy as:
DIMENSION(2)
COFRAME(SKY)
.NOT. PIXEL(00:00:00,00:00:00)
does work (with some sensible RA/Dec naturally).
Peter.
|