GCS Implementation Strategy: draft for discussion ------------------------------------------------- 1.0 INTRO For the purposes of splitting this complicated question into manageable chunks with a logical structure, I suggest we consider the following: o GCS target survey definitions - areas covered, field centres, their labels, overlaps o Priorities - relative priority of targets (2yr & post-2yr), field priority, priority relative to other UKIDSS sub-surveys o Rules - Fundamental limits, filter combination & order, calibration considerations, dither/jitter/coadd sequences o General points - MSB preparation, QC considerations, experience from other surveys 2.0 GCS TARGET SURVEY DEFINITIONS 2.1 Uniformly overlapping (5%) fields - But: is it necessary to avoid bright stars (our targets have many bright stars visible to the naked eye!) If we want to probe the IMF down to low limits, blending by bright stars will severely affect the photometry at the lowest levels; hence suggest to "mask" out bright stars, or at least avoid having too many of them in our fields. How bright is bright (need commissioning data)? If it's too complicated, then we may have to ignore (GPS are taking this attitude?) but the other problem with BS is spikes and ghosting producing spurious object detections, so it might be advantageous to try to minimise those. - Also: uniform field distribution will probably need to be adjusted by small amounts to attain sufficiently bright guide stars on the guide CCD. 2.2 Field labelling - fields labelled according to increasing distance from cluster/association "centre" - results in numeric IDs based on proximity to "centre" in order of scientific priority (see later), spiralling outwards from the centre (is this clear? Does it make sense?!). 2.3 Targets - targets as defined in Table 1 of the 2 year plan (revised version of original GCS target list) - circular areas except for Taurus-Auriga 2.4 Overlaps - Hyades area excludes those fields included in the Pleiades survey - Taurus-Auriga area excludes those areas included in the Hyades, Pleiades and Perseus OB2 surveys - Coma-Ber overlaps with the LAS; Alpha Per overlaps with the GPS: how do we handle these? LAS head (RFJ) suggests that GCS should get to Coma-Ber first... 3.0 PRIORITIES 3.1 Priority relative to other UKIDSS subsurveys - Are the combined survey targets sufficiently well spread in RA that this will not be a problem? If not, the problem may get worse as the surveys approach completion and different fields conflict for the same observing time. 3.2 First 2yr target priorities - 1:Pleiades, 2:Alpha Per, 3:Praesepe, 4:IC 4655 (JHK) - 5:Tau-Aur, 6:Orion, 7:Sco (JHK; central areas) - 8:Per OB2 (JHK) - 9:Hyades, 10:Coma Ber (K only) - 11:Tau-Aur, 12:Sco (K only; remaining areas) 3.3 post-2yr target priorities - same order as 3.2 (switching K only with JHK as appropriate) 3.4 Priority within a given GCS target - proximity to cluster/association centre (ie. in order of ID label) What about severe crowding on ground-based images? Very centre might not in all cases be the best, highest priority target as you won't be able to see anything other than the brightest stars. Scattered light bear the cluster centres may mean that photometry is less accurate and depth is compromised... however, relatively few frames will be affected so perhaps it is simplest on balance to use proximity as the primie priority. - areas automatically contiguous and expanding 3.5 Other priority considerations - restrict maximum usable airmass such that the observations making up individual target surveys are evenly spread rather than all occuring at once (and therefore leaving holes in the GCS, and potentially combined UKIDSS, schedule). On the other hand, perhaps best to try to cover any single cluster/association as extensively as possible at any given time, and not to jump from target to target: radial coverage will be of interest... - any other "relaxing" of the prioritisation outlined in 3.1 - 3.4 ? 4.0 RULES NB: Site stats can be obtained from [1] UKIRT Newsletter No. 10 (issue 10) page 7 etc, [2] ESPAS Site Summary Series: Mauna Kea (http://www.eso.org/gen-fac/pubs/astclim/espas) [1] quotes median seeing at 0.6 arcsec with < 0.7 arcsec for 70% of the time. [2] quotes 0.5 arcsec at 0.5um. Note that [2] quotes Mauna Kea as being 45% photometric on average! 4.1 Fundamental limits - seeing: no GCS observations to be made if seeing is worse than 0.7 arcsec [some debate about this one. Hide has suggested 1.5 arcsec - this seems a bit large... may be necessary to alter this depending on crowding in the target field] - airmass: no GCS observations to be made if airmass > ? (stringent limit to get best possible data and to space GCS observations evenly?) [again, some debate: one suggestion was airmass always less than 2, another < 1.5] - non-photometric conditions: no GCS JHK observations to be done in such conditions, but K-only observations can be made provided seeing is better than 0.6 [Two possible uses of non-photometric conditions have been suggested: i) increase depth of sub-optimal exposures, and ii) disregard conditions, simply revisiting JHK frames taken in non-photometric conditions with snap-shots to calibrate. My feeling is that neither is worth pursuing: i) it is hard to get significant gains in depth in reasonable times, and ii) is not efficient simply because the GCS exposures are likely to be short anyway... one might as well retake the full exposure science frame in photometric conditions since the overhead on taking a snap-shot will be similar.] - sky brightness limits for JHK (see also coadds below) 4.2 Filter sequence - take JHK together [Hide has suggested doing one filter at a time while Paul Hewitt (for the LAS) points out that quasi-simultaneous has clearly been advantageous for eg. in SDSS. My feeling is there are additional efficiency advantages to taking JHK together. However, it may be necessary to limit filter changes to once every 30mins, say, if the overhead on moving the filters is large (eg. suppose it takes 20s to change the filter...?!) in which case it might be more sensible to change filter only after taking one or two "tiles"] - order: JHK can be taken in any order [once again, there will be efficiency gains in that filter changes will be fewer if we don't insist on the same order for each field.] - repeat K observations (for position measures at one epoch): must not be taken if epoch difference between original and repeat is less than ? years. [in the original UKIDSS GCS proposal, required epoch difference is stated as 5 years, but this won't work in practice if the total UKIDSS survey is to be 5 years, and in view of the time required to do all of the observations for a second time. It is probably sensible to prioritise repeat observations by the cluster motions - ie. give the highest expected pms the shorter time baseline, so the Hyades would get done first, then the Pleiades, then Praesepe, then Alpha Per... In any case, minimum epoch difference should be set at 3 yr] 4.3 Dither/jitter/coadds [Does jitter (UKIDSS Newsletter #7) mean microstep ?!] - (n + 1/2) pixel combined dither/microstep ? - 2x2 microstep ? - how to reach uniform faintness at all times: variable amount of coaddition ? depends on observing conditions, incl. Moon, photometric, airmass, etc... Will be hard to really get uniform, but perhaps we should go for a rule-of-thumb exposure time guideline and adjust according to conditions at observing time? - only microstep if seeing results in sampling worse than Nyquist ? [All of these are common to all wide UKIDSS surveys; I've chucked this last one in to see what people think. I don't think we can decide on any of the above in isolation within the GCS so will wait for a UKIDSS-wide debate] 4.4 Frequency of standard observations - how frequent do we think we will need ? Standards at at least 3 different airmasses per photometric night, ideally twice (at least) at each airmass, so that means roughly one standard field each 1.5 hours or so. 2MASS observed standard fields once every hour. - should we set up secondary standard fields for each target at large (how large?) distance from UKIRT FS equatorial standard fields? It has been suggested that we establish a cluster target field as a local secondary standard field to be revisited; this will be especially advantageous if there is a good distribution of colour in the secondary standard field stars. 4.5 Other calibration considerations - any need for offset skies ? As long as we can create good sky frames to subtract from the science images by using other science images taken close in time... - any other calibration considerations... ? 5.0 GENERAL 5.1 MSB preparation [It is my understanding that a "Survey Definition Tool" will be used to generate "minimum schedulable blocks" (MSBs). The MSBs then reside in a DB at the telescope, and the scheduler queries the DB to find out which MSBs to attempt in a given observing slot] - Can the Survey Definition Tool handle all of the complex criteria for the GCS ? Do arbitrary and complex "rules" (as above) propagate through to the scheduler via MSBs? Martin Folger is looking into circular areas being implemented in the definition tool, but is obviously keen to avoid spec creep and goalpost movement in the work he is doing... - If not, can we edit MSBs or otherwise generate them externally to the Tool? 5.2 QC considerations - What provision will there be for automatic QC and rescheduling? - How do we reschedule MSBs when the end user finds QC problems in frames 5.3 Experience from previous surveys - Are there any lessons to be learned from previous surveys, eg. 2MASS for IR-specific things [5.2 & 5.3 will be common to all UKIDSS surveys so we have not get too hung up on them within the GCS]