Meeting with Steve Warren and Andy Adamson at Imperial to discuss UKIDSS 
observing strategy                                           14 May 2003
------------------------------------------------------------------------

(These notes might seem a bit wandering, but the meeting was like that).

The first question was to find out why I wanted to attend the meeting. The
answer was to find out what was happening on the observing strategy front
and to see if I could contribute to the task.

AA described the Query Tool (QT). This selects the MSBs that will be
observed. The QT looks in the data base and takes information from the
observer (water vapour, seeing, ...). The current ranking uses
airmass/lowest airmass (close to culmination). This tweak was need to
complete some of the observations. There is also a science ranking (in front
of decimal point for survey and behind decimal point for within a survey).
At the moment, seeing is not scaled with airmass.

In November, the consortia were asked to provide criteria needed for each
survey.

DWE asked for a clarification: all observation selection will be automatic.

The idea is to have a monthly tweaking of priorities in the QT if there are
any problems.

Parts within a survey can have different seeing requirements (eg LAS).

Seeing (AA): median is 0.4-0.6" (seasonal). Very steep to 0.3. Tail out to
1.2 with half the power to 0.75 (not too sure about tail description).
Perhaps give up bad seeing and call it weather.

SJW suggested classifying standard seeing as <0.7", poor as 0.7-1.2 and >1.2
as bad. The consortia would then specify alternate programmes for poor
seeing. Standards could be done for bad seeing?

AA said that his simulations used a rate of completion (run rate) in the
selection the observations. If a survey was falling below its run rate then
it would be given a boost. Also said that seeing was a better criterion than
air mass.

In SJW's simulations, the density of observations required (in RA) was used
as a criterion to boost certain observations.

UDS can't finish if only observe within 1 hour of meridian.

AA's simulation used hour-by-hour seeing changes.

Lunch: DWE raised the issue of the Moon. AA thought that it probably wasn't
important, but no-one has looked into this yet. This would mainly affect the
J(?) band. There might be implications on the availability of guide stars.
He said that he would check up on this.

AA showed some plots of sky brightness. It was high at dusk, but not at
dawn. This was to do with heating and atmospheric chemistry etc. SJW
wondered about restricting the time of night to what filters can be used eg.
perhaps no J for first 3 hours of a night.

SJW asked what sequence of filters should be done. Should there be a
restriction? Calibrating strategy: AA thought JHK each time standards were
done - not much extra overhead. SJW wanted more flexibility in the
calibrating strategy. (The calibrations also provide seeing measurements.)

AA said that the QT works in the short term ie. is reactive, chooses the
next observation when it is required. SJW wanted an observing plan for the
night.

AA Photometricity: thin cirrus 20% of time. Thick cirrus useless for
astronomy. This information comes from the autoguider. Another
classification is locally clear - patchy cirrus. This is done using a
hand-held detector (Russian night-sight). 

AA's view: MSBs ~ half an hour. Doesn't want them to be > 1 hour since
timescale of change of observing conditions is about half an hour.
(see note at end)

Less than 50% of time is photometric in all directions (+15% locally clear).
Bottom line for MSBs and survey definition are two cases: clear and thin
cirrus.

Some surveys will have different priorities (this is the bit behind the
decimal point) within the survey eg. within DXS to match a HST field. (The
UDS will not be moved. HST field is within DXS, but not UDS.)

AA: The MSBs that need to be observed are defined by the Survey Definition
Tool (SDT). This is done every 2 months or so - defines the run rate.
Then the QT chooses an MSB from this list each time one is needed.

The SDT has two parts (1) visualization bit and (2) MSB definition bit. One
of these is done, but AA couldn't remember which one. (2) does the tiling of
defined areas (taking into account the availability of guide stars). The
area definition is rectangular and can be in equatorial, Galactic or Sloan
(eta,lambda) coordinates. DWE raised the point about circular regions since
a number of surveys have them. AA said that there was also a low-level
definition where the pointings are specified "by hand". (I'm not sure if
this means that they are going to look at circular definitions or leave it
with the low-level method).

SJW said that some surveys will want to do all the filters close to each other
(within a lunation) - important for variables. This seemed to be a new one
for AA and he seemed hesitant about it, but made a note.

MSBs microstepping etc.: UDS and DXB need to think on this. Does anyone want
3x3 microstepping?

There are some overlap problems. What is shift (n) for microstep? It's half
(or third) of a pixel plus n which matches up with the autoguider CCD. This
n reduces the edges of a frame which in turn causes a problem for the
macrostep overlap.

SJW jitter: Need to ask groups how they want this done. This needs to be
defined. This has an implication on the SDT - perhaps no guide stars will
be available with the jitter.

AA: there is a settle time for the guide star of about 1s. This might be in
addition to the readout overhead for the 5sx2 mode. Does 10s become 12s?

Bright star: persistence and neighbourhood. The SDT will flag MSBs that have
bright stars in them.

AA said that a simulator should probably be done at JAC. This is because it
needs to be tied in with the QT quite closely (various databases etc.). It
looks like there is no need for me to do any form of simulations. With AA
and SJW having done tests and a probable simulator at JAC, I would be
duplicating work.

Survey manager: Old plan (2 years ago) - based at JAC
New plan 2 half posts: (1) JAC week-to-week checking and tweaking - at JAC
because of time it takes to send data to C/b. Also can react to weather
better.
(2) UK (CASU? or WFAU?) monitoring the pipeline output, reacting on ~2 month
timescale to tweak surveys.
Close communication between (1) and (2).

(SJW will be back on full lecturing next year)

Design of processable MSB: This was a Mike topic, but I suggested that JRL
and STH would be better bods to discuss this (especially considering
differences in what people think an MSB is - see note at end).

Quality control will be at the summit and at CASU (AA thought that it would
be better to come from the end of the pipeline rather than the archive.) How
will these two interact? How often? How automatic? Quality control will be
at the level of an MSB is OK or not OK. AA wanted the interaction to be by
email (effectively by hand) if the two quality controls are different. (But
if there were more than 20% different, AA would then think automatic might
be more reasonable). AA did not want any bad data going into the archive.

The Observing Tool will show the progress to date of the survey.

Planning commissioning: who does it? How to do a calibration will be part of
the commissioning. The commissioning run should be driven from the tests and
not from science. Tests: throughput, image quality, do the filters work,
alignment (this will take most of time), mesostepping (to get idea of
vignetting). Mark does not want any science pressure during commissioning.
If the telescope works, then we start PATT/UKIDSS straight away (for 2
months) with no intermediate programmes.

They have sorted out staffing at JAC for 2 months of commissioning issues,
but AA wants UKIDSS volunteers as well.

AA said that WFCAM will be on UKIRT for 60% of the time, of which 80% will
be UKIDSS and 20% PATT. PATT and UKIDSS will not be mixed. PATT will have
separate 1-2 week blocks.

AA wants to know what STH is doing with the calibrations. He has seen the
work packages. There is one night of service time available for this. AA
will contact STH.

AA wanted to know what fraction of time nothing would get done with WFCAM
(due to visible surveys having been completed). If WFCAM is on the telescope
evenly spread throughout the year then there will be holes and bumps (where
survey has difficulty completing). Therefore uneven scheduling of WFCAM on
telescope. Even so, AA still found holes. Was concerned, but his simulation
didn't weight by run rate (equivalent to SJW's weight by observation
density), and SJW didn't find holes.

DWE asked AA for some of the detailed seeing, photometricity, weather data
to be released. Will do so. Photometricity a bit more difficult since not
collected yet. Perhaps in a month.

How were SJW and AA's simulations done? SJW has not much more than 2-year
plan from last year. AA will write up a few notes.

It seemed to me that the main purpose of the meeting was for SJW to get a
set of questions together to ask the 5 surveys in order to definitively
define the surveys.

                           -----------------------

(After a quick chat with Simon, I'm getting the impression that JAC and CASU
have very different ideas as to what an MSB is. This might be terminology
and a JAC MSB (which contains standards) is a CASU "observing block". This
needs to be sorted out).