> SURFIT does not give us these coefficients. FITSURFACE does but doesn't
> seem to be as good as SURFIT (and requires an additional call to
> MAKESURFACE adding 0.1 sec to the pipeline run time).
SURFIT originates from wide-angle galaxy photometry from plates and
worries about filtering out galaxies and stars, and bins. FITSURFACE
is pure surface fitting; if there are regions you want to exclude you
have to mask them beforehand.
> We need to fit a 2d plane to our SCUBA-2 data. This works brilliantly
> using KAPPA SURFIT but we need to know the gradient from the fit because
> this provides an estimate (possibly) of the extinction.
SURFIT could write Chebyshev coefficents to the output NDF's SURFACEFIT
extension like FITSURFACE.
> At the moment we are considering simply collapsing the surface into X and
> Y and then calculating the gradient by looking at the data values at the
> extrema. Bit of a kluge but should work since we know it was a linear fit.
Yes a kludge but practical. However, that also involves two more task
invocations.
I can't think of anything else. ESP background fitting is local around
each bright galaxy rather than a global surface fit.
Malcolm
|