Measuring Seeing at UKIRT

Tim Hawarden

Head of UKIRT Development, Joint Astronomy Centre, Hilo

NUMBERS IN SHORT SUPPLY

Eyeball estimates from the TV have been around for more than a decade, but are hard to calibrate in the presence of automatic gain control. Image cross-sections from the SCANCO autoguider were obtained, AAT-style, some years ago, but required operation with control settings which impaired our never-too-robust guiding performance, and the conflicting requirements of detectivity of faint objects and resolving images in any case mean that image scales are too small for any but the worst seeing to be properly resolved.
 

WHEN WE GET GOOD DATA, PROBLEMS GETTING ENOUGH OF IT...

It has long been recognised that nearly-ideal, simply-acquired, performance measures would be values of the FWHM (say) of well-focussed images taken with IRCAM using the x2, or preferably, the x5, magnifiers. However, these were provided as post-hoc add-ons, mainly intended for test purposes, and cause background level increases and consequently sensitivity losses. So they were
rarely used until word of regular undersampling with IRCAM in standard mode (x1 = 0.28" per pixel) began to spread; use of x5 is still a rarity, although the x2 magnifier is now used on 25-30% of IRCAM nights. This was encouraging, but it soon became apparent that observers are unlikely to bother with the x2 magnifier if the seeing is poor ( > 0.7" say) so that a uniform sampling policy would be required.

Unfortunately, installing and removing a magnifier, and taking measurements of the image quality with it, takes about half an hour, due to slewing, installing, focussing, observing, slewing, removing, refocussing ... etc. So it has clearly been impractical to require observers who are not using it already to install it several times a night to do image quality measurements in addition to their own
programme.

This has left us in a dilemma, and we have taken some deliberate steps to try to solve it. In particular, we feel a strong need to have a seeing measurement system in place and working, and enough data gathered to allow before and after estimates to be made of the effects of the mirror cooling system, which comes into operation late this year.
 

A PROTOCOL IN PLACE

After a little experimentation a standard procedure has been adopted. This protocol can be found on the public UKIRT web-page (http://www.jach.hawaii.edu/UKIRT/telescope/protocol.html). Observers are now asked to make hard-copy cross-sections of images from standard star measurements made soon after focus checks, around the middle of the first and second halves of the night (if this has the beneficial effect of gently persuading observers – who do like to know the seeing – to do more focus checks and observe more standards, which many do NOT like doing, it’s all to the good). The TSS’s then do the pencil-and-paper work and record the results in their nightly log, onto which I then pounce upon next morning and extract into the database ready for analysis and display.

In case the IRCAM observer is not observing at K at the right time, or at all (which is in any case rare!) there are now two  standardised system EXEC’s for use with the x2 magnifier and without. These secure 3 images at K. The observations are made using a subset of the UKIRT Faint Standards to ensure proper S/N on the one hand and a long enough exposure on the other (to avoid freezing the seeing and inadvertantly doing speckle imaging!). We find that together with averaging over the 3 images, this produces a robust result.

I would not ask observers to measure all 5 images from a normal standard observation unless they were short of entertainment, but hey! if you’re feeling generous...

Nevertheless, all measures are welcomed as long as they are reasonably carefully done according to the protocol. For overall statistics we use only two (possibly averaged) values per night, to avoid bias, but (for example) one or more series of good measures made approximatly hourly through several nights would be invaluable to start learning about trends and changes (I just don't have the gumption to lean that hard on observers to analyse so many images!).

In engineering time we have secured consecutive, carefully focussed, image sets with and without the x2 magnifier on 4 occasions. They trace a well-defined locus outlining the effects of undersampling (which can be dramatic: an image measured at 2.0 pixels FWHM with no magnifier is actually 1.05 pixels wide). We can use this relation to correct no magnifier images to equivalent x2 images – but the better x2 images are significantly undersampled too. We have extended the correction to the x2 data by assuming the same effects affect the better images in the same way too, something we will check and correct with the x5 magnifier now we know how serious undersampling effects can be.

In the meantime, the provisional correction curve allows us to confidently correct onto a fully sampled scale the larger images taken with no magnifier. The smaller images are done so with some reservations, and the x2 images similarly.

The figure displays a histogram of the results which will be kept updated on the UKIRT web site (http://www.jach.hawaii.edu/UKIRT/telescope/seeing.html).

At the time of writing the overall data set (95 measurements) has a median FWHM of around 0.52". This is in the K band and transforms to around 0.68" at V (0.55 microns).

We want to impress on our visiting observers our need for increasing and improving the database of seeing statistics at UKIRT. These numbers are the final validation of the things we do to make the telescope better: without them we are in reality firing blind. In addition, they are tools for the further investigation and understanding of the factors affecting the ultimate performance of the telescope and such understanding is the key to further development of this and many other telescopes.

In the meantime, please remember – GET THAT SEEING DATA ( please?).
 

Back to Contents