A seeing disk observed through a slit should---in principle---reside at a fixed coordinate along the slit. One common departure from this assumption is in the case of high-bandwidth spectrographs (usually low resolution) observed at low airmass in good seeing, in which case atmospheric dispersion causes the seeing disks to reside at different positions along the parallactic angle. That's usually not a problem for high resolution spectrographs, in which the seeing disk within an echelle order appears at about the same place.
Empirically we see a pin-cushion-like distortion to the position of the seeing disk compared to our simplistic slit coordinate reference frame, s(x,y). My hunch is that the origin to this pin-cushion is more likely intra-spectrograph optical camera effects than it is atmospheric dispersion, but the manifestation of the effects are the same, so it's helpful to think of it like dispersion.
No matter, the most expedient way to deal with this pincushion is to add a slope and curvature parameter to the trace center-position, as a function of s. These may become tricky to train for the faintest brown dwarfs, but we also want to know the trace positions to high accuracy in these cases to improve weighting, so it's worth getting this right.
A seeing disk observed through a slit should---in principle---reside at a fixed coordinate along the slit. One common departure from this assumption is in the case of high-bandwidth spectrographs (usually low resolution) observed at low airmass in good seeing, in which case atmospheric dispersion causes the seeing disks to reside at different positions along the parallactic angle. That's usually not a problem for high resolution spectrographs, in which the seeing disk within an echelle order appears at about the same place.
Empirically we see a pin-cushion-like distortion to the position of the seeing disk compared to our simplistic slit coordinate reference frame,
s(x,y). My hunch is that the origin to this pin-cushion is more likely intra-spectrograph optical camera effects than it is atmospheric dispersion, but the manifestation of the effects are the same, so it's helpful to think of it like dispersion.No matter, the most expedient way to deal with this pincushion is to add a slope and curvature parameter to the trace center-position, as a function of s. These may become tricky to train for the faintest brown dwarfs, but we also want to know the trace positions to high accuracy in these cases to improve weighting, so it's worth getting this right.