SPECIAL REPORT: ORAC - a new data acquisition and reduction system for UKIRT

Andy Adamson¹ & Gillian Wright²

¹UKIRT/JAC, Hawaii
²Astronomy Technology Centre, Edinburgh, U.K.

Introduction

ORAC is an acronym which stands for Observatory Reduction and Acquisition Control. The ORAC software was integrated with all UKIRT instruments in May, and is now in use for all UKIRT observing. It replaces all the software at UKIRT that interacts with users, providing a modern observing interface.

The goal of the ORAC project was simply to improve the publication rate of UKIRT. It will do this by making it simpler to prepare and carry out observations, and by giving observers excellent feedback on their observing as it goes along, thus reducing time wasted, and by producing near-publication quality reduced data at the telescope. There are many other ways in which ORAC was designed to provide a good base for future operations, however. The above elements are all required if we are to consider scheduling the telescope flexibly: (i) excellent remote preparation tools such that entire observing programmes can be prepared in the UK or elsewhere, to a level that someone else knows exactly what to do and (ii) sufficiently accurate, revealing and non-interactive data reduction pipelining that a non-expert can monitor data quality on behalf of an absent P.I. In addition to the benefits for the user, the ORAC project was designed to provide benefits to UKIRT operations "under the hood", by making software support and future development easier - it does this by using modular designs to achieve flexibility and extensibility. For example future new instruments will only need to provide the software that is specific to that instrument meeting the interfaces defined by ORAC and everything will work together. Additional software for flexible scheduling could be added in a straight forward way.

For many years, UKIRT observers have prepared their observations and executed them using "EXECs" and "CONFIGs" generated using the UKIRT_PREP system. This system, based on ASCII files generated using tools running on VT terminals, has stood UKIRT in good stead for many years, but is decidedly long in the tooth and is not appropriate to operations in the 21st Century. ORAC replaces both of these aspects of data acquisition at UKIRT and also provides a data reduction system which is data-driven and into which all future UKIRT facility instruments will slot. The new preparation system provides far more help than was possible with UKIRT_PREP - for example it has a powerful position editor with which mosaics can be set up and stored, and displayed on digital sky survey images. ORAC does not supersede the telescope control system; however, it does interface to the TCS and is capable of executing slews, offsets etc. Since UKIRT plans to upgrade the telescope control system to use the "PTCS" in the near future, ORAC was implemented with an interface to this new system. Although the original UKIRT telescope control system is currently still in use, it has been modified to "look like" the PTCS in some respects. Of course ORAC had to be compatible with existing UKIRT instruments, and some aspects of the old software were still good ideas, so in some ways ORAC will have a familiar "feel".

Naming conventions

The various components of ORAC with which observers now interact are:

ORAC-OT: Observing Tool (preparation of "Observations" and submission to database)

ORAC-OM: Observation Manager (selection of individual "Observations" at the telescope) and OM Sequence Console (display and control of detailed breakdown of observing sequences).

ORAC-DR: Data Reduction system

An "Observation" in the ORAC sense can embody a fairly lengthy sequence of events: for example, a cycle of 40 quads with CGS4 or an extended 5x5 UFTI jitter pattern would both be described as "Observations".

People

The ORAC was a collaboration between the UK-ATC and JAC. All of the following have contributed significantly to the new system: Alan Bridger, Gillian Wright, Frossie Economou, Andy Adamson, Min Tan, Malcolm Currie, Alan Pickup, Maren Purves, Russell Kackley and Nick Rees.

FIGURE 1: The ORAC-OT (left) and ORAC-DR (centre and right).
Click on the figure for an expanded view.

Data acquisition: what the user sees

ORAC-OT

The ORAC Observing Tool is based on the Gemini OT, with extensions to allow preparation of programmes for the UKIRT instruments. The centre spread in the Newsletter (Fig.1 above) shows an OT running on kauwa, the Linux PC at the summit. Your prepared observations are stored to a database using this facility.

ORAC-OM

This runs at the telescope (see the screen shot in Figure 2). Here you load your observing programme from the database, and select individual observations for execution - this is shown on the left of the figure.

FIGURE 2: The ORAC-OM (left) and ORAC-OM Sequencer (right).
Click on the figure for an expanded view).

ORAC-OM Sequence Console

Submitting an observation for execution from the OM pops up a version of the Sequence Console (if one is not present already for the instrument requested). This console (the much larger window at right in Figure 2) shows you the executable form of the currently-selected observation, unwrapped into its component parts, and gives you control of the observing sequence. For example, in this screen you can run a sequence from a given highlight position, abort a sequence, request that a sequence stops at a logical breakpoint (no more stopping quads by waiting for the last exposure!) and restart the sequence if you need to increase signal-to-noise. The sequence console is also responsible for showing you the status of the instrument - in the example shown in Figure 2 CGS4 is being used; the status information is at the bottom right of the console window.

Data reduction: ORAC-DR

This part of ORAC has been in use (and under continual development) since the arrival of UFTI in 1998. The pipeline is designed to require minimal interaction with the user - indeed once a copy of the pipeline is set into execution, the only interaction possible, apart from adjusting display parameters and doing slices etc in GAIA, is to terminate the pipeline. This is deliberate; the ORAC philosophy is to implement well-defined observing sequences and to produce standard reduction recipes which accompany them. The data reduction process then requires very little in the way of human interaction. This is not to say that the results of the pipeline cannot be easily inspected in an interactive way; the STARLINK GAIA display tool is used for most image display purposes, and this provides an increasing array of inspection tools (slices, spectral extraction, image analysis, seeing measurements etc.).

Because of the above philosophy, ORAC-DR takes its cue from the data themselves as to how it should reduce a given sequence of frames. Each OT observation contains a DRRECIPE instruction, which does nothing other than to make sure that each data frame has in its FITS header the name of a pre-defined DR recipe to be used on that frame. For example, a set of jittered UFTI frames might be flagged with a DR Recipe name of JITTER_SELF_FLAT. When those frames arrive on the raw data disk, ORAC-DR knows what to do with them without being told. Indeed the only argument (as opposed to command-line switch) which the oracdr command takes is a recipe name, so that you can override the recipe specified in the FITS header. This gives the flexibility required, for example, to use a non-flat-fielding reduction recipe if one exists.

Initial Results

ORAC has been used by all PATT programmes since 1-Aug-2000, and had been used by JAC staff members for engineering and their own programmes since the initial installation in May. Initial perceptions are that ORAC offers a considerable efficiency gain (of the order of 20%; we are working on calibrating this). A further gain is to be had when carrying out combined imaging/spectroscopy programmes: switching from CGS4 to UFTI is now extremely quick, limited only by the speed with which one can rotate the dichroic and set up on the target. Obviously the better prepared your program is, the more likely you are to see efficiency improvements.

Notes for Observers

Remember that the ORAC philosophy is to match observing sequences to suitable data reduction recipes. If you require a non-standard observing sequence (or believe that UKIRT should implement your sequence as standard!), contact the JAC ahead of time to enable us to implement data reduction recipes appropriate to the data your sequence will produce. A good example is multi-position nodding along the CGS4 slit, which while trivial to set up in the observing tool, is not catered for by the current set of DR recipes.

Significant Web Sites

You can get more detailed information on the ORAC project, its internal workings and details of the software, from the following site:

http://www.roe.ac.uk/atc/projects/orac/

and its JAC Mirror:

/UKIRT/software/orac/