UKIDSS first data release (DR1)

Steve Warren¹, Nigel Hambly² and Simon Dye³

Figure 1: Proportion of DR1 data with a given seeing.

The first large release of UKIDSS survey-quality data, DR1, took place on 27 July 2006. This followed on from the Early Data Release, that took place on 10 February 2006 (reported in the Spring 2006 Newsletter), which was a small prototype dataset. DR1 contains all the data from the 05A and 05B WFCAM blocks that passed quality control. The vital statistics of DR1 are summarised in Table 1. As before the data release comprises two databases. The DR1 database consists of fields where full coverage of the filter set exists for the particular survey, and is a subset of the DR1+ database, which includes also data for fields with partial filter coverage i.e. all data that passed quality control. The summary provided in the table is for the smaller DR1 database. But overall in DR1+ there is coverage in one filter or another of over 1000 sq degs. DR1 marks the point where UKIDSS overtook 2MASS in terms of source photons collected - quantified by the product: telescope collecting area x f.o.v. x summed integration time. At this point UKIDSS has completed some 7% of the 7-year plan.

Full details of DR1, including maps of the areas covered, are contained in a submitted paper, Warren et al. (2006, astro-ph/0610191). The image quality is good, with median seeing 0.82"+/-0.14". The distribution of the seeing, over all the frames released in DR1, is illustrated in Fig. 1. In fact some small improvement to this quantity may be expected for future releases, since the 05A data suffered from imperfect alignment of the optics, and compensation of instrument flexure was not implemented until after the end of 05B.

The next release, DR2, is penciled in for the end of February 2007.

Table 1: Depth and coverage in fields with the full filter complement
in UKIDSS DR1

Survey	Area	Filters	K-band depth
	(sq. degs)		(5sigma Vega)
Large Area Survey	190	YJHK	18.2
Galactic Clusters Survey	52	ZYJHK	18.2
Galactic Plane Survey	77	JHK(+H2)	18.1
Deep ExtraGalactic Survey	3.1	JK	20.7
Ultra Deep Survey	0.8	JK	21.6

Accessing WFCAM data from the WFCAM Science Archive

Mike Read

As part of its involvement in the Vista Data Flow System (VDFS) project, the Wide-Field Astronomy Unit (WFAU) in Edinburgh has developed the WFCAM Science Archive (WSA).

The WSA holds the calibrated image and catalogue data products generated by the Wide Field Camera (WFCAM) on UKIRT and processed by the Cambridge Astronomical Survey Unit (CASU). (Raw WFCAM data is archived by CASU and ESO.) The primary contents of the archive originate from the UKIRT InfraRed Deep Sky Surveys (UKIDSS): Large Area Survey (LAS), Galactic Plane Survey (GPS), Galactic Clusters Survey (GCS), Deep Extragalactic Survey (DXS) and the Ultra Deep Survey (UDS). Released in July 2006, the first UKIDSS data release (DR1) is formed from several Terabytes of data and contains over 100 million sources. In addition the WSA also archives non-survey data taken for open-time programmes (e.g. PATT, service data, Japanese and University of Hawaii data, etc.).

On transfer to WFAU, the catalogues and image meta-data are passed through a number of curation tasks prior to being loaded into a Relational DataBase Management System (RDBMS). The curation process also produces a number of enhanced data products: for example, object catalogues merged across several passbands, deep stacks for the DXS and UDS, and cross-matching with other data products (e.g. SDSS and 2MASS).

Figure 1: Screen shots from the WSA. Top row from left to right: Batch pixel finding charts (MultiGetImage); menu driven query builder; browsable archive listing. Middle row: SQL query form; SQL query results; results displayed in TOPCAT. Bottom row: schema browser; image cut-outs, library jpegs of archived files. Click on the above image for a detailed view.

Users can access the data held in the WSA in a number of ways via a set of web forms. The services provided currently include: browsing and downloading archived FITS images and catalogues, creating image cut-outs, performing simple cone-searches on catalogues, cross-matching archived sources with a supplied list of coordinates and using the power and flexibility of the Structured Query Language (SQL) to mine the data. A simple illustration of the last technique is shown in the central three panels of the screen shot in Figure 1. Here a query selecting the type (star/galaxy) and J-H and H-K colours of objects in a given region of the LAS is submitted. The results are written to a FITS catalogue which are then launched into the Starlink package Topcat (Tool for OPerations on Catalogues And Tables) and used to produce a colour-colour plot. This particular example took under a minute from query to plot. A cookbook is provided to help astronomers unfamiliar with SQL.

Future plans for WSA include implementing Virtual Observatory access via Astrogrid and possibly services providing user-defined image mosaicing and stacking.

Users must be registered with the WSA in order to access UKIDSS data or their non-survey (PATT, etc.) data whilst still under proprietary periods. Details on registration and further information on the WSA are given on the website. Feedback and questions should be addressed to wsa-support at roe-dot-ac-dot-uk.

Moon ghosts in WFCAM data

Daniel Mortlock¹, Steve Warren¹, Dave Henry² and Andy Adamson³

The very large number of frames taken for the UKIDSS project has made it possible to assess the end-to-end performance of the telescope, the WFCAM instrument, the pipeline, and the WFCAM Science Archive (WSA), and to identify and diagnose problems with the data. While in most respects (depth, image quality, uniformity) the data have lived up to expectations, a number of unpredicted artefacts have been identified. These include cross-talk images of bright stars, channels with bias offsets, and images of dirt on the field lens -- a rogues' gallery is provided in Dye et al. (2006). Most serious are a number of large, bright ghost images, an example of which is shown in Fig. 1. This article describes how the cause of these bright ghosts has been deteremined, and how they can be eliminated.

Figure 1: Image showing the four WFCAM arrays as they are positioned on the sky. The diamond in the centre represents the WFCAM CCD (optical) autoguider. A bright moon ghost appears in array 2 (bottom left). The artifact in array 1 (bottom right) is caused by a support pad residual on the field lens, which has since been removed.

As can be seen from Fig. 1, ghosts such as this are sufficiently large and bright to render the affected exposure unusable. Anecdotally, ghosts were believed to occur when the Moon is near the field, but this was not always the case, leaving the cause, and the solution, unclear. The problem was deemed serious enough to undergo the painstaking task of visually inspecting all available UKIDSS exposures as part of the quality control for the EDR and DR1. The JPEG images provided by the WSA make the task straightforward, if still timeconsuming -- in all, over 10000 images were viewed. In all, some 6% of exposures are adversely affected by ghosts (although this includes negative images caused by a positive ghost in the frame used for sky subtraction). This is clearly an unacceptably large figure, corresponding to a loss of 60 nights observing over the lifetime of UKIDSS.

Figure 2: Plot of frames with Moon ghosts (filled black symbols) and unaffected frames (open symbols, coloured according to Moon phase, from red = full to blue = new). The axes are the (absolute) angular separation of the Moon from the field in RA and dec. Ghosts apparently occur in a restricted patch that approximately corresponds to the position of the arrays in an expanded mirage of the focal plane (also illustrated).

Having determined the scale of the problem, the next step was to confirm that the Moon was in fact the source of the ghosts and then to find a way to avoid this in the future. Under the assumption that the ghosts are caused by a reflection, we explored a variety of variables including angular distance between the field and the Moon, angular distance of the Moon from the line of the slit, and Moon phase, plotting the affected and unaffected frames. It became evident that ghosts almost always occur when the Moon is within 30 deg of the field, in which case a ghost has about a 30% chance of appearing. A clear picture emerged, illustrated in Fig. 2, when the data were plotted in terms of (absolute) angular separation of the Moon from the field in RA and dec. This assumes that the problem is symmetric in RA and Dec, and folds the sky into a quadrant. It is immediately clear that ghosts only occur in a patch of this quadrant. Effectively the figure is a mirage of the focal plane, expanded by a factor of ~60, and a ghost occurs when the Moon lands on an array. The filling factor of arrays on the focal plane is about 0.3, in agreement with the observed fractional occurence of ghosts when the Moon is nearer than 30 deg. This hypothesis was partially confirmed by a controlled experiment, undertaken on an engineering night, when a series of images at selected offsets from the Moon was taken, but only along the diagonal of Fig. 2.

Final confirmation of this picture came when the original optical analysis to check for ghosts was extended to larger angles than previously considered. Ray tracing successfully reproduced the phenomenon in detail, correctly accounting for the ghost shape and location, for a given angular displacement of the Moon. The ghost arises when the moon shines obliquely on the autoguider auxiliary lens, above the field lens, and the light path undergoes a double internal reflection in the field lens. Shadowing by the barrel of the auxiliary lens assembly itself explains the form, steadily reducing the size of the ghost at larger angles, until the ghost disappears at an angle of 31 deg.

As important as it is to understand the nature of this problem, the real aim of the investigation was to eliminate the ghosts. The simplest solution is to avoid observing within 30 deg of the Moon. Indeed this procedure was followed throughout the 06A WFCAM block. Nevertheless because many UKIDSS fields lie near the ecliptic, this has serious implications for scheduling, and adversely affected progress in some fields in 06A. Fortunately the optical analysis indicates that a simple conical baffle below the auxiliary lens will eliminate the problem. The baffle is being manufactured, and it is intended that it be installed and its effectiveness tested at the start of the 06B observing block.

References
Dye, S., et al., 2006, MNRAS, in press (astro-ph/0603608)

View from the Top

Thor Wold

We're back to cass now; WFCAM has come off for a spell, and we are hard-pressed to try to get back some of the cass time that was utterly wiped out earlier this year by the "weather of biblical proportions". At the time of the publication of the last Newsletter, we were unknowingly just entering a period of rather unprecedented bad weather...

A series of low-pressure systems set themselves up 200 miles west of Kaua'i, held in place by an unusual blocking pattern. If this had set up 100 or 200 miles further west, nobody would have noticed anything. As soon as one would weaken and start to move off to the north, it was replaced by yet another. One low-pressure system managed to sit in this spot for six weeks straight. Meteorologists here said this was a once-in-a-career happening; it made for the wettest March in Hawai'i in 55 years. They do hope that the data they have accumulated from this event will better prepare future forecasters for when it occurs again.

The flow over this island was such that we found ourselves in the tail of a continuous stream of high clouds and embedded thunderstorms giving the summit bouts of snow, ice, fog and utterly opaque cloud. The flow was from the "wrong" direction, meaning that the leeward and windward sides of the island reversed - this meant Hilo and Hamakua did not see the epic rainfall that the smaller, northern islands did; the mountains managed to block the weather from getting to the eastern side of this island. Winds were predominantly from the southeast or south.

Some of the Grim Statistics: in March, a dam burst on Kaua'i, killing 6 people. We had 22 days of flash-flood warnings, and only five days of tradewinds. There was a tornado on Lana'i, and large hail here on Hawai'i Island (both are extremely rare events). While Honoka'a and Waimea on this island got about 2" of rain in March, Lihue Airport on Kaua'i got 36.13" - ten times greater than the average rainfall for the month as determined from the past 30 years worth of data! The rains started February 14th and finally went away April 13th. Meanwhile, atop Mauna Kea...

                     total jac weather     jac weather web    ukirt weather
      date               web page hits  page hits per hour  loss percentage
      -------------  -----------------  ------------------  ---------------
      2006 april                 17727                24.6            30.2%
  --> 2006 march             --> 26753            --> 36.0        --> 87.3%
      2006 february              16220                24.1            48.3%
      2006 january               16769                22.5            36.5%
      2005 december              20380                27.4            30.8%
      2005 november              14524                20.2            28.2%
      2005 october               16670                22.4            24.9%

Personally, I did not have a full night of observing between Feb 14 and April 13. I lost four whole shifts where I never left HP even for a single night. I lost both of my two shifts in March in a row and again never left HP. Except for the occasional foray to the telescope to see how bad things were, we were imprisoned at HP.

I can state unequivocally that I have never experienced anything like this in 20 years on Mauna Kea! From what the meteorologists are saying, it is unlikely to happen again for 20-30 years, if not 55 (like this one). We hope...

Road improvements

A word of caution; when driving up Saddle Road to the Mauna Kea Access Road turnoff, be prepared to encounter construction. The overhaul of 18 miles of Saddle Road, from around the 19 mile mark to around the 35 mile mark at Mauna Kea State Park commenced towards the end of July. $59 million has been budgeted for the project, and scheduled completion is in early June 2008. We will keep you informed on progress, of course.

Currently, activity is concentrated on the intersection of Saddle Road and the Access Road, as they are starting from there and heading back to the 19 mile mark, which is about where the forest begins. During the realignment and construction back towards the 19 mile mark, there are likely to be significant periods of intense construction, temporary detours, etc. For the moment, if you are coming to visit us do expect to encounter heavy equipment and construction activity as you near the turnoff to Hale Pohaku. Later, while they will be straightening the roadbed, there will also be significant stretches where they will be following the existing roadbed, necessitating tearing it up entirely.

Installation of the new traffic signals at the intersection of Nowelu and Komohana near JAC has been underway now for several weeks, but we have yet to see any of the poles, as they have been involved with digging through the rock to install the under-road wiring. Traffic flow through this project has been a bit of a mess and highly varied as to which lanes are open on any given day. This ought to be up and working soon. Again, we hope.

Hale Pohaku

The work on replacing the outside siding of Building B at Hale Pohaku is finally making noticeable progress. They have managed to make the turn around the back side of the building and have finally begun finishing the last wall (the Mauna Kea side). In the past six months, they have spent the time replacing the decking on the walkway to the dorms, the entry and the south deck of the Commons Building, rather than just finishing off the Building B project. Periodically, they were trying to keep people out of the rooms on that side (just in case?), causing a bit of a crunch in the other two dorm buildings. Regular readers of this space will know that this project has been in a continuing status since its first mention in the fall 2004 edition of this Newsletter - meaning this project has passed its third anniversary. Well past time it was over with.

Internet connections have now been installed in all of the rooms. Otherwise things are much the same. I now know folks that book themselves into Building D with its miniature beds simply because they know the bedding will stay put and not come off (that's me - Ed!), since the sheets still don't quite fit the 'new' queen-size beds in B and C dorms. The hot/cold water surges are making a comeback, so residents of HP are again learning how to do the HP Shuffle to avoid being scalded/frozen. Put your dancing shoes on...

Aloha!