When and where was the last focus? Was it reliable (e.g., did the DR detect plenty of stars, were the E-W or N-S tilts large?). Was it done too close to the moon?
Is the ellipticity bad? If so, it could be because the focus is poor or you are seeing astigmatism.
Is the DVS open or closed?
Is the telescope close to a bright moon? If so the guiding may not be good on faint guide stars.
Is there any connection with change of position in the sky? You can check UKIRT_LOG.log but positions are also in the file headers and HA (degrees), declination and airmass are in the night log.
Are there large differences between the actual and demand positions for the hexapod (U, V, X, Y, Z)? These can be looked at in the channel archiver (e.g., mcu:hex:x0:dem:val and :rbk) or in the file headers themselves (e.g., M2_X, M2_XD). Differences shouldn’t be more than 5 hundredths although they can build up if the guiding has not been switched off for a while (e.g., UDS field).
Are there galaxies in the field that could have a detrimental effect on measured seeing?
Is the DR having a problem and/or have unguided frames been included in the group? Check the DR logs and the file headers include “GUIDING” which tells you if guiding was on, off or paused.
Weather problems? Is the telescope pointing into the wind? Bear in mind that if the wind is from the east the UKIRT readings will be incorrect for both direction and speed. You should confirm direction and speed with the CFHT numbers.
Is the guide rate frequency low? (again available in the headers: AGFREQ). Remember the lower the frequency the better the image quality. The ideal rate is 60-Hz.