Successful commissioning of AAOmega - Australian · PDF file · 2014-04-22fact if one overlooks some frantic activity associated ... and WHT is coalescing into a coherent picture

AN

GLO

-AU

STR

ALIA

N O

BS

ER

VATOR

Yc

on

ten

ts

3 IRIS2 and the global star formation rate over most of history (Rob Sharp et al.)6 Probing the peak of the age of quasars using 2dF and SDSS (Kuenley Chiu et al.)11 End of an era: the 6dF Galaxy Survey takes its final spectrum (Heath Jones and the 6dFGS Team)14 The Radial Velocity experiment (The RAVE Team)18 Melbourne Observatory & the genesis of astrophysics in Aust. (Jenny Andropoulos & Wayne Orchiston)21 AAOmega commissioning update (Rob Sharp & Will Saunders for the AAOmega team)24 A new field configuration algorithm for AAOmega (Brent Miszalski et al.)

N E W S L E T T E R

N U M B E R 1 0 9 F E B R U A R Y 2 0 0 6

AAOmega installed in the Coude West room at the AAT.

An r~22 radio selected emission line galaxy from science verificationdata using nod and shuffle for 2 hours on source exposure time.

Successful commissioning of AAOmega

ANGLO-AUSTRALIAN OBSERVATORY NEWSLETTER

FEBRUARY 2006page 2

DIR

EC

TO

R’S

ME

SS

AG

E

DIRECTOR’S MESSAGE

AAOmega’s final commissioning run and Science Verification observations were completed last night. The articleby Rob Sharp and Will Saunders on page 21 of this newsletter reports on the commissioning process and providesthe latest information on the performance of the instrument. The bottom line is that AAOmega appears to beperforming very close to specification, and that the remaining issues are expected to be resolved prior to the startof the first allocated science observations at the end of February. The nod & shuffle and cross-beam switchingmodes have both been fully commissioned, and initial tests of mini-shuffling, allowing nod & shuffle observationswith all fibres, indicate that this important mode will also work as intended. This is an excellent outcome and atribute to the extraordinary efforts that the AAOmega team have put into designing, building and commissioningthe instrument.

AAOmega Science Verification data were taken over 8 nights for 11 programs selected by the service time committee,showcasing the range of AAOmega’s capabilities using a wide variety of instrument setups. These observationswill be reduced and made publicly available as soon as possible. The first allocated science observations withAAOmega will begin on 22 February. The SPIRAL integral field unit for AAOmega will be commissioned later in thesemester, with the first allocated science observations beginning on 27 June.

Semester 06A is the first in which time was allocated by the unified Anglo-Australian Time Allocation Committee(AATAC; see page 17 of the last newsletter). The Chair of AATAC, Martin Asplund, reports outcomes from theAATAC meeting on page 29 here. There was a healthy over-subscription for AAT time, and a strong response(11 proposals) to the call for large observing programs. In this first round, AATAC has approved components of onelarge program outright (the Anglo-Australian Planet Search) and made allocations of time for pilot observations tofour other large programs planning to use AAOmega.

Depending on the outcomes of these pilot studies, AATAC may choose to select one or more of these programsfor long-term support, or it may select another large program proposal from among the new proposals submittedfor Semester 06B. These large programs should produce much of the flagship science from the AAT over the nextfew years, just as the 2dF galaxy and quasar surveys did in the recent past, as reflected, for example, by the newsarticle in Nature (2006, 439, 251) comparing the impact of various telescopes.

The AAO is about to undergo a significant and necessary review by the Australian Government. The review willfocus on the future of the AAO, both in terms of positioning in the run-up to the foreshadowed withdrawal of the UKfrom the AAT Agreement in 2010, and the form the organization will take thereafter, when the AAO becomes awholly Australian entity.

The review is being conducted under the auspices of the Minister for Education, Science and Training, who hasappointed a review panel consisting of Dr Ian Chessell (chair; former Chief Defence Scientist), Professor GarthIllingworth (UCSC/Lick) and Professor John Storey (UNSW). The panel will consider the recommendations of theAustralian Astronomy Decadal Plan 2006–2015 and take as input submissions from all interested parties. Theannouncement of the review and the call for submissions will appear on the Department of Education, Science andTraining website (http://www.dest.gov.au/). In order to provide focus to submissions, the panel will release anissues paper, discussing the background to the review and highlighting the critical concerns, in early February.Submissions may be made to the review panel via the secretariat at DEST ([email protected].) before the deadline,24 March. The review panel will meet and visit the AAO for four days 10–13 April, during which time they will talkto the major stakeholders in the AAO, including the AAT Board, representatives of the user community and AAOstaff. It is expected that the panel will report back and make recommendations to the Minister by June 2006.

It is important that all stakeholders in the AAO, and especially Australians, respond to this opportunity to shapethe long-term future of the Observatory.

Matthew Colless

ANGLO-AUSTRALIAN OBSERVATORYNEWSLETTERFEBRUARY 2006

page 3

SC

IEN

CE

HIG

HL

IGH

TS

IRIS2 AND THE GLOBAL STARFORMATION RATE OVER MOST OFHISTORYRob Sharp (AAO), Andrew Bunker(Exeter), Michelle Doherty (ESO), IanParry (Cambridge) & Gavin Dalton(Oxford)

Since the first appearance of the notorious “Madau-Lilly”diagram in the mid-1990s (Madau et al. 1996, Lilly etal. 1996; Pei & Fall 1995), which represented the firstattempts to trace the variation of the global star formationrate over the history of the Universe via comparison ofstar-formation rates at different redshifts, astronomershave puzzled over the true form of the phenomenon.With star-formation indicators plagued by innumerableissues (dust reddening, indicator calibrationuncertainties, dust reddening, initial mass function (IMF)uncertainties, dust reddening, underlying absorptionspectra, dust reddening... pretty much everything really)requiring the use of large correction factors, it is easyto infer quite the wrong history when contrastingobservations made with different techniques. Ideally onewishes to take one reliable and well understood (or atthe very least uniformly biased) tracer of star-formationand observe this tracer over an extended redshift range.

Unfortunately one of the most widely used tracers,H-alpha emission (which directly traces emission fromthe hot massive star population and hence the currentstar-formation rate) becomes inaccessible to opticalspectrographs at modest redshifts (z<0.5). To follow it,

one must head into the depths of the deep, butdepressingly not very dark, forest of OH air-glowemission lines in the near IR (Iwamuro et al. 2001).

Our collaboration has used the CIRPASS (Parry et al.2004; Doherty et al. 2004) IR fiber MOS system (bothat the AAT and the WHT), initially bringing the AAO-FOCAP system out of retirement, and lately the IRIS2slit mask system, to pursue H-alpha star-formation rates(and H-beta/[OIII] in higher redshift source, allowing limitsto dust reddening and metallicity effects to be tested)over the redshift range 0.8<z<1.4 (see Figures 1 & 2).

Target selection

Working at intermediate resolutions (R~2000–3000),between the OH sky lines, one can take advantage ofthe intrinsic darkness of the night sky at near-IRwavelengths (Maihara et al. 1993). We identify targetsfrom a range of sources (for example the CFRS, Deep2,GOODs and VVDS programs) with spectroscopicredshifts confirmed at optical wavelengths. Accurateredshift information is required to prioritize targets whichwould place any H-alpha emission at a wavelength whichfalls between the strong OH night sky lines. We havefound that simple photometric redshift estimates arenot sufficient for such work, with the typical uncertaintyof the redshift of a source amounting to a substantialfraction of the wavelength range accessible, even to largeformat 1 & 2K IR arrays, at the moderate resolutionrequired to resolve out the sky emission. One wouldthen have no proper information as to where H-alphawould fall within the OH forest and quantitativedetermination of the underlying H-alpha emission levelis next to impossible. Identifying emission lines amongstthe residual OH forest, at the low flux levels of interest,when the correct wavelength range is not knownaccurately is a rather futile task. With future OH

Figure 1: Fibering up a plug plate. The CIRPASS systeminitially used the AAO FOCAP plug plate system at theCassegrain focus of the AAT (shown here) and the WHT.While outdated by the standards of modern robotic fiberingsystems, such as AAOmega, the FOCAP plug-plate systemhas the advantage that our army of graduate student andpost-doc “plate pluggers” had lower running costs (andmarginally better conversation) than most robot systems.

Figure 2: A CIRPASS MOS spectrum for an H-alpha emitterin SSA22. H-alpha is clearly visible, and at the expectedwavelength.

1.195•104 1.200•104 1.205•104 1.210•104 1.215•104 1.220•10Wavelength (A)

−6.00•10−17

−4.00•10−17

−2.00•10−17

−2.34•10−24

2.00•10−17

4.00•10−17

6.00•10−17


FEBRUARY 2006page 4

SC

IEN

CE

HIG

HL

IGH

TS

suppression instruments, one may well be suppressingthe very signal one seeks.

IRIS2 observing

The IRIS2 slit mask acquisition and observing systemis impressive. With minimal effort the Skycat-GAIA andIRIS2 interface software allows alignment of the mask,in translation and rotation, with a number of fiducial starsin each field. After some initial experimentation theprocess became rapid and reliable allowing confidencein the result, which greatly reduced the stress levels forour inexperienced IRIS2 observers (IRIS2 is somewhatdifferent to 2dF!), associated with the multi hourexposures needed to record the H-alpha emission. Infact if one overlooks some frantic activity associatedwith a 2am GRB override, smooth, confident observingwould appear to be a feature of the IRIS2 system.

Unfortunately the excellent ORAC-DR software, whichhandles more common IRIS2 observing modes, doesnot reduce slit-mask data. IDL scripts and common IRAFtasks were used to process the data (cosmic rayrejection, A–B position beam switching, rectification ofwavelength scales and long-slit sky subtraction). Initialinspection of the data at the telescope, using even theserather primitive tools, revealed numerous H-alpha, andseveral [OIII] emission lines (see Figure 4). The detailedanalysis of the data has been somewhat delayed by

AAOmega commissioning.

The future

The analysis of the observations from both the AATand WHT is coalescing into a coherent picture of thez~1 star formation rate, as measured via H-alpha(Doherty et al. in prep). Our results will be presentedshortly in the context of similar works (Glazebrook etal. 1999; Tresse et al. 2002).

IR MOS spectroscopy is coming of age. Large formatIR arrays give us the real-estate to record multiple,moderate resolution spectra at once while improveddata quality allows one to push the limits of sensitivityrequired to make useful comparison with alternate SFRindicators. The wealth of data available from opticalredshift surveys at z>0.8 allows one to tune sampleselection to take maximum advantage of the dark skybetween OH lines, taking much of the guesswork outof the measurement.

While we all wait with bated breath for theimplementation of OH suppression fibers (Bland-Hawthorn 2005), the immediate future promises muchthrough the use of fantastic facilities such as the soon-to-be-commissioned FMOS/Echidna system at Subaru.

With the huge multiplex of Echidna and the light graspof an 8 meter telescope, FMOS presents the tantalizing

Figure 3: To the left we see a raw IRIS2 slit mask exposure. 15 minutes on sky is required for the interline regions tobe limited, not by readnoise, but by sky/scattered light levels. Cosmic ray rejection was performed using a custom IDLroutine which owes its origins to the IRAF/STSDAS task CALNICA. Due to the catastrophic failure of the IRIS2 sciencegrade detector earlier in the year, the engineering grade detector is seen here. There are a number of cosmeticdefects in the detector. However, many of these are removed by beam switching yielding, for the most part, excellentdata. The four fiducial stars used to align the mask can be clearly identified. Individual objects are allocated 8arcsecond slits with objects selected to give the maximum target density on the mask. Slits are then grown outwardsto fill in the gaps where no new object slits could be placed, yielding a longer spectrum for a better long slit skysubtraction. Observations are performed with the target nodded between two positions on the slit 4 arcseconds, 10pixels, apart. To the right we see the result of a single A–B position 15 minute beam switch. First order sky subtractionof this sky limited frame is good. A full long slit subtraction is performed after wavelength rectification. Using thebeamswitch technique removes the need for independent dark exposures.


page 5

SC

IEN

CE

HIG

HL

IGH

TS

opportunity to build upon the work described here,building larger samples, and to further address questions,such as the reddening and metallicity effects, raised byH-beta/[OIII] observations (Shapley, Coil & Ma 2005).

References

Bland-Hawthorn J. 2005 AAO newsletter 108 4Doherty M., Bunker A., Sharp R. et al. 2004 MNRAS 354L 7Glazebrook K., Blake C., Economou F., Lilly S. & Colless M.

1999 MNRAS 306 843

Iwamuro F. et al. 2001 PASJ 53 355Lilly S.J, Le Fevre O., Hammer F. & Crampton D 1996 ApJ

460L 1Maihara T. et al. 1993 PASP 105 940Madau P., Ferguson H.C., Dickinson M.E., Giavalisco M.,

Steidel C.C. & Fruchter A. 1996 MNRAS 283 1388Parry I., et al. 2004 SPIE 5492 1135Pei Y.C. & Fall S.M. 1995 ApJ 454 69Shapley A., Coil A.L. & Ma C.-P. 2005 ApJ 635 1006Tresse L., Maddox S.J., Le Fevre O., Cuby J.-G. 2002

MNRAS 337 369

Figure 4: A number of H-alpha emitters are clearlyvisible in this IRIS2 J band slit mask image. Thissection of a beam-switched A-B frame, at a mid pointof the data reduction process, shows clear H-alphaemission from two sources in SSA22. The field alsocontains a number of H-beta/[OIII] emitters, not shownhere. A close up of the long slit sky subtractedspectrum of one of the sources, along with its finalsummed spectra, is also shown.


FEBRUARY 2006page 6

SC

IEN

CE

HIG

HL

IGH

TS

PROBING THE PEAK OF THE AGEOF QUASARS USING 2DF ANDSDSSKuenley Chiu, Karl Glazebrook (JHU),Brian Boyle (ATNF), Scott Croom (AAO),Wei Zheng (Exeter), Terry Bridges(Queens), Rob Sharp (AAO), ZlatanTsvetanov (JHU)

Forty years ago, a new type of object began to beobserved by astronomers curious about unusual radio-emitting sources on the sky [1]. These objects –compact and bright in visible light images, yet alsopuzzlingly radio-loud unlike stars – were eventuallyunderstood to be highly luminous sources at previouslyunimagined cosmological distances. Given the namequasars (and synonymously QSOs and/or AGNs), theyopened a new frontier of extragalactic observations withtheir extreme brightness and redshifts. In the decadessince those first basic discoveries, quasars have playeda leading role in expanding our observable horizon.Indeed, with quasars steadily illuminating the pathtowards the high redshift domain, these once rare andexotic objects have become regarded as common toolsfor observers seeking information about the environment,composition, and state of the distant universe itself.Recent discoveries have brought us to the frontier ofz~6 using ground-based telescopes, where the epochof reionization has been tantalizingly suggested by the

few brightest quasars that can be found [2].

Quasars are intimately linked to the populations of blackholes that occupied the early universe, and to the laterpopulations of galaxies we see today. Looking back fromthe present in time and distance, one of the most strikingfirst observations about quasars was their rising numberdensity with increasing redshift [3]. Beyond z~3, quasarnumbers were also found to decrease rapidly [4]. Morerecent work by collaborations such as the 2dF QuasarRedshift Survey [5] and the SDSS Spectroscopic QuasarSurvey [6] have produced the clearest picture of quasarevolution yet, identifying a peak in the numbers of quasarsaround approximately z~2.5 – the epoch of maximumquasar activity in the universe. Thus, what we envisionnow is that the age of black hole formation in the earlyuniverse led to the conditions making quasar activitypossible, and that quasars grew and then declined withtime, giving way to the age of galaxies that we ourselvesoccupy.

But what is the detailed chronology of this sequence ofevents? And in what particular order did the mass thatformed luminous structures collapse? The exact timesat which these various populations flourished holdconsequences for constraining the rate of structureformation in the early universe – currently a topic ofsignificant interest as these several areas converge withthe accumulation of theoretical models and observationalevidence.

Fig.1 At z~2–3, the spectra of quasars appear very similar to normal Galactic stars when sampled bybroadband photometric filters. Optical surveys select candidates for followup spectroscopy based on colors,and in such cases, cannot distinguish between high-value quasar targets, and the relatively more commonstars.


page 7

SC

IEN

CE

HIG

HL

IGH

TS

Our team sought to carry out a project that might shedlight on the question of when exactly quasarsexperienced their maximum period of activity. Fordespite the importance of the z~2–3 redshift range forquasar evolution, the peak has remained only roughlyconstrained due to a coincidence of quasar and stellarcolors and shapes. At these and other problematicredshift ranges, the colors of normal Galactic starsmuddy the task of distinguishing quasar candidates inimaging and following them up with spectroscopicresources. Large numbers of stars reduce the efficiencyof quasar surveys in these regimes, and any consistentlyselected sample is overwhelmed with possible targets(see Figures 1 and 2). Given finite spectroscopicresources, survey groups naturally concentrate ontargets with the highest success rate. The SDSS, forexample, observes only 10% of the possible targets inthe z~2.5 area because of the heavy contamination.As a result, in all of the catalogs of quasar populationsat z~2.5 to date, a significant drop in quasar numbersis reported. And while a decrease in numbers may beauthentic in part, its true magnitude and location inredshift space is hidden behind the difficulty of observingthese candidates and confirming their identities asquasars.

In a program generously allocated time on the 2dFinstrument at the AAT, we conducted this project toidentify a large sample of quasars around the z~2.5

peak of their evolution, using SDSS imaging coupledwith a dedicated spectroscopic discovery survey. Wewould deliberately focus on these contaminated regionsof quasar color space in order to recover hiddenpopulations of quasars and improve the estimates ofthe parent population of quasars around the peak ofactivity.

Our method was to begin with imaging data of the SloanDigital Sky Survey and follow up using the spectroscopiccapabilities of the 2dF instrument. Given the possibilitythat we would have hundreds of targets per squaredegree, the unique 2dF instrument provided the onlyspectroscopic resource capable of executing such aprogram. With the ability to rapidly configure (andreconfigure) 400 fibers over a wide field, and reach limitingmagnitudes close to the SDSS imaging limit within areasonable exposure time, 2dF was perfectly suited tocarrying out the program we envisioned.

On the surface, our task seemed simple – to observeas many targets as possible in the contaminated regionshown in Figure 2. But in practice this presented someinteresting technical challenges, and gave us excellentexperience in merging SDSS and 2dF practices. Ourmain task before observation was to design the efficientphotometric color selections that would create suitablesamples for input to 2dF. In the area where z~2.5quasars are found, the number of candidates rises

2.0 2.2 2.42.6

2.8

3.0

u-g

Fig.2 As quasar colors evolve with redshift (solid track) and collide with those of stars, forexample at z~2.5, spectroscopic surveys are overwhelmed by the number of candidates forfollowup. Surveys such as the SDSS must exclude such contaminated areas (central narrow boxwith labels 2.2 to 2.6).


FEBRUARY 2006page 8

dramatically as the stellar locus is encountered. Clearly,the extent of the selection into this contaminated areawould determine the number of objects produced forobservation. Figure 3 illustrates our candidate cuttingstrategy. Using an adjustable top which could betranslated up or down in SDSS g-r color, we were ableto increase or decrease the number of candidatesproduced. Calculations of candidate number versus thisadjustable selection criterion, as well asmagnitude, gave us the needed informationto design the appropriate color andmagnitude cuts to match the number oftargets to the 2dF fiber density (as seen inFigure 4). And with a “floating” magnitudelimit in each field, the 2dF configurationsoftware could be assured of a sufficientnumber of candidates even in naturally under-dense patches of sky.

One interesting technical aspect of 2dFbenefited our project and is worth a mention– as a side program, in addition to this z~3work, we pursued quasars at z~5.4–5.8located in another stellar-contaminatedregion of color space. The discovery of thisrange of quasars is hindered by a similarcollision of the quasar track with the stellarlocus in SDSS riz color space. Indeed, onlytwo quasars in this broad redshift range havebeen discovered to date [7,8] due to thisproblem. In contrast, 15+ z~6 quasars have

been found, which are believed to be even rarerobjects. The 2dF’s split spectrograph capabilityallowed us to observe both z~2.5 and z~5.6 targetssimultaneously, by placing a blue spectral gratingin spectrograph 1 and a red grating inspectrograph 2. Because z~2.5 targets areidentifiable with either grating, this allowed us toimplement both programs in a resource sharingscheme, so that an efficient allocation of fiberswould benefit both programs. The interesting fiberallocation issues posed by this split spectrographscheme, as well as the successful observation ofa z=5.03 quasar, were detailed in the September2001 issue of this newsletter.

Another observational enhancement we studiedand implemented was the “slow-nodding” procedureused to help reduce the effect of the many nightsky lines approaching the near infrared, whichotherwise affected our red grating identificationsseriously. As is all too familiar to astronomersworking in the optical where λ>8000, and in thenear-infrared, the increasing strength and numberof night sky lines severely hinders low resolution

spectroscopy at these wavelengths. Specifically, it isthe variability of the strength of the lines which causesinaccurate subtraction of their flux, and thus distortsthe spectrum of the underlying object. Slow-noddinginvolves observing a blank field (in each 2dF fiber)between 2 object observations for the same exposuretime. The roughly contemporaneous observation of sky

SC

IEN

CE

HIG

HL

IGH

TS

Fig. 4 The 2dF candidate number density (contour lines) was calculatedversus color selection and magnitude. By fixing the color selection acrossthe sample, and allowing the magnitude limit to float, under- and over-dense fields could be accommodated by the 2dF configure program.Horizontal line indicates our nominal survey limit of i=21.0

u-g

Fig. 3 Our selection strategy tuned the number of targetsobservable in each 2dF field by adjusting the candidate cut depthinto the stellar locus.


page 9

acquired in this way can be used to better subtract thesky flux affecting the adjacent object observations. Thisapproach is more successful than use of the 30 or sosky fibers allocated by default to each 2dF field, becausethe separate sky observation proceeds through theidentical prism, fiber, and overall optical path as eachobject observed, thus making the individual subtractionas accurate as possible. Figure 5illustrates the promising results ofthis technique.

With the observation strategyplanned, we travelled to the AAT fora dozen nights of observing spreadover two years. The central themeof the observing could be summedup in one word: stars. Plenty ofstars – we found more normal starsthan could ever be desired by anastronomer. But in addition, for eachnight of observing four to six 2dFfields, we also discoveredapproximately 30 new z~2.5quasars – a rate far exceeding theSDSS automated quasar survey perarea, precisely because ourprogram chose to focus on this peakrange of their numbers. In all, 70fields were exposed on the 2dFinstrument, each with a typical fiberallocation of 360 program objects,30 sky, 4 guide stars, and about 10broken fibers. A total of some 24500

object spectra were inspected, withmost of these being normal Galacticstars. 340 quasars were found in theredshift range z=1.95 to 3.28.5 additional quasars were discoveredin the range z=4.44 to z=5.28, selectedand observed by the experimental z~5.6high redshift observing program carriedout simultaneously. The quasars spana redshift range of 1.95 to 5.28, apparentmi magnitudes of 17.46 to 21.85, andabsolute Mi(z=2) magnitudes of –28.94to –24.34, as shown in Figure 6.

A number of interesting points haveemerged from the luminosity functionsand calculated density functions so far,though our analysis is ongoing. Wehave found that the binned luminosityfunction derived from the 340 quasarsdiscovered in this work shows aninteresting redshift evolution compared

to the analytic double power law fit by Richards et al.(which is very similar to that of Croom et al. (2004) atlow redshifts). The data suggest that the redshiftevolution of the quasar space density is both more rapidthan previously found, and reaches an absolute peakdensity higher (and later) than Richards et al. Oursparsely populated bins at redshifts greater than z~3

SC

IEN

CE

HIG

HL

IGH

TS

Fig. 5 The improvement of “slow-nodding” sky subtraction is illustrated. A2dF sky exposure is taken between two object exposures, allowing accuratesky to be measured for each object, through the identical fiber, prism, andcomplete optical path.

Fig. 6 The sample of 340 new z~2.5 quasars discovered in this work. Objectsare plotted in redshift versus absolute magnitude, using Mi(z=2) convention ofRichards et al. 2006. Our survey limit of i=21.0 is shown as the solid line,contrasted with the survey limits of the SDSS spectroscopic quasar survey, ati=19.1 (low-z) and i=20.2 (high-z).


FEBRUARY 2006page 10

SC

IEN

CE

HIG

HL

IGH

TS

Fig. 7 Bright quasar spacedensity versus redshift. Theresults of several groups areshown in a composite here,illustrating the general riseand fall of quasar numbersversus redshift, including thelatest work of Croom et al.(2004) and Richards et al.(2006) constraining the peakof activity around z~2.5. Ourdiscovered quasar sample(three red points with errors)suggests that the peakoccurs both higher and laterin redshift than found inprevious work, around z~2.8.

make it difficult to give definite statements about theendpoint of the trend and at what redshift the quasarnumbers begin their expected decline again. However,we are intrigued that the redshift peak of the quasaractivity is later than expected from previous models –i.e. at higher redshifts, around z~2.8 rather than z~2.4as predicted by Richards et al. and others (seeFigure 7).

As we had hoped, the excellent capabilities of 2dFcoupled with the SDSS allowed us to examine theimportant redshift range where quasars have their mostactive period of evolution. The natural difficulty of selectingtargets in this area resulted in an observationallyintensive, but valuable program that produced a sampleof 340 quasars around z~2.5. In addition to publishingthe catalog of discovered objects, a study of theluminosity function will be forthcoming – we have foundquite interesting preliminary results pointing to a peakin the quasar evolution later (at higher redshift) andstronger than found in previous work. Deeperobservations in this area may help to determine thereality of this peak, quantify its amplitude, and betterconstrain its redshift. Although 2dF has now beendecommissioned, it has regained a second life as the

central refitted component of the AAOmega instrument,a more sensitive multi-object and integral-fieldspectrograph for the AAT. Throughput, spectralresolution, and stability have improved, and theinstrument promises to continue producing thegroundbreaking large, faint, and high-redshift samplediscoveries that marked 2dF as a unique astronomicalfacility. The several collaborative groups involving SDSSand 2dF have already begun planning the productiveuse of AAOmega, and with new surveys such as UKIDSSproviding a rich source of targets, we will undoubtedlywitness many novel results from this instrument foryears to come.

References

[1] Schmidt, M. 1963, Nature, 197, 1040[2] Fan, X., et al. 2004, AJ, 128, 515[3] Boyle, B. J., Shanks, T., & Peterson, B. A. 1988, MNRAS,

235, 935[4] Schneider, D.P., Schmidt, M., and Gunn, J.E., 1994, AJ,

107, 1245[5] Croom, S.M., Smith, R.J., Boyle, B.J., Shanks, T., Miller,

L., Outram, P.J., & Loaring, N.S. 2004, MNRAS, 349,1397

[6] Richards, G.T., and SDSS collaboration, 2006, in press[7] Stern, D., et al. 2000, ApJ, 533, L75[8] Romani, R.W., et al. 2004, ApJ, 610, L9


page 11

END OF AN ERA: THE 6DF GALAXYSURVEY TAKES ITS FINALSPECTRUMHeath Jones and the 6dFGS Team

Observing for the 6dF Galaxy Survey (6dFGS) finishedon the 5th of January this year after nearly five years ofcollecting spectra on the UK Schmidt Telescope. Thisambitious programme seeks the redshifts of 150000southern galaxies and the peculiar motions for around10000 of the brightest. Being a public survey, the redshiftand spectral data for some 90000 sources have beenmade available in prior data releases in 2002, 2004 and2005. Accessing the data is possible through an onlinedatabase maintained by the Wide Field Astronomygroup at the Royal Observatory Edinburgh (http://www-wfau.roe.ac.uk/6dFGS/). The final instalment of data willbe released later in 2006.

The key science drivers for 6dFGS seek the total stellarmass of the local universe and its relationship toenvironment and bulk flow motions. To this end, the mainsurvey targets were selected from the Two Micron AllSky Survey (2MASS). Such near-infrared selection notonly furnishes a source list of the most evolved stellarpopulations, but also provides a direct link to theluminous mass of the system as contained in stars.This single feature is what sets 6dFGS apart from itsoptically selected counterparts such as the SDSS and2dFGRS.

Meeting the survey goals has been a challenge for the6dFGS, but it has been in the fortunate position ofcapitalising on lessons learned from forerunners suchas the FLAIR and 2dF surveys. Instrument design,survey strategy and software are just some of the areasin which the 6dFGS has benefited in this way.

The Final Data Release for 6dFGS will necessarily bemore comprehensive than all its predecessors. Beingthe last time we update the database, several monthsof exhaustive scrutiny of the dataset will need to takeplace. Manual examination of questionable redshifts,application of zero-velocity template shifts as well asheliocentric corrections, and manual cross-checking ofassociated field book-keeping are just some of the tasksthat will be undertaken during this time. For the onlinedatabase itself, there are plans to broaden its contentas well as provide additional links to supplementaryinformation about certain fields and the data theycontain.

Recent work by Tom Jarrett (IPAC/Caltech) hasexamined the issue of fibre cross talk, a difficult but(thankfully) rare occurrence in 6dFGS data. Astigmatism

towards the edges of the 6dF spectrograph means thatstrong emission-lines can spread and contaminateadjacent spectra if sufficiently bright (Figure 1). As aconsequence, neighbouring spectra gain additionalemission-line-like features that in some cases canconfuse redshift identification. Tom has software toolsto identify instances of this, which we will use to flagpotentially affected spectra in the database.

Figure 2 shows the map of 6dFGS field coverage as itstands at the end of the survey. It shows how the surveyhas essentially covered the entire southern sky. Theoriginal survey target was the completion of all 1595fields by the end of July 2005. However, a slightly higherthan expected fibre breakage rate meant that the averagenumber of useful fibres turned out closer to 100 thanthe 120 originally expected per field. Consequently, bymid-2005 the survey was 155 fields short of completecoverage, with many of these spring and summer fieldsaround 02 to 06 hrs in RA. Negotiations with the RAVEProject underway on the UK Schmidt Telescope led tothe acquisition of 49 nights after July 2005 by way ofswap and purchase so that 6dFGS could meet its goals.

Alas, Mother Nature was not kind to 6dFGS during thespring of 2005, and cloudy skies wiped many of the

(a)

(b)

Figure 1. (a) Close-up view of a portion of unreducedspectra showing [OIII] and Hβ features in a bright emission-line source at z=0.00247. Astigmatism at the field edgecauses the lines to de-focus slightly. (b) Same spectra,now reduced, showing how the spread of light has producedfalse emission-lines in the adjacent spectrum above theemission-line source.

SC

IEN

CE

HIG

HL

IGH

TS



SC

IEN

CE

HIG

HL

IGH

TS

October and November 6dFGS nights out. Whatremains, as Fig 2 shows, is virtually full coverage exceptfor some areas around the LMC and the pole. The centralband is the Zone of Avoidance which 6dFGS did notattempt to cover. The final number of fields observedwas 1539, or 96.5% of the total. The actual fraction ofthe survey sky area covered is higher, since many ofthe unobserved fields overlap with observed fields.

The paper that accompanied the First Data Release(Jones et al. 2004) describes the main attributes of the6dF Galaxy Survey, including the instrument, targetselection and redshifting procedures. A comprehensiveintroduction to the online database is also given. TheSecond Data Release paper (Jones et al. 2005)summarises the main features of this second publicrelease and discusses limitations that current users ofthe data should be mindful of. A final data paper willcoincide with the Final Data Release and will characterisethe survey in its finished state, as well as provideluminosity functions and densities.

Figure 3 shows the K and bJ luminosity functions fromJones et al. (2006), from a larger set covering KHJbJrF.Luminosity function fits from a number of recent surveysare also shown for comparison. By this juncture in thesurvey, a little more than half the redshifts were in hand.The luminosity functions shown here improve on previous6dF measurements in a number of ways. First, thesample sizes are much larger than before – the K-bandsurvey alone totals some 60000 galaxies. Second, themagnitudes used in K have been updated with the latest2MASS total magnitudes, and no longer rely on ouroriginal total magnitudes, which were inferred from acombination surface brightness and isophotal magnitude(see Jones et al. 2004). Furthermore, the bJrF

magnitudes are the re-calibrated SuperCOSMOSmagnitudes from which plate-to-plate zero-point variationhas been removed (see Jones et al. 2005). Third, wehave made field flow corrections to our line-of-sightvelocities using software by J. P. Huchra, therebyremoving the peculiar velocity component of each galaxy.

(b)(a)

Figure 2. Equal area Aitoff projection of 6dFGS field coverage. Open circles denote all fields on the target list and filledcircles denote those comprising the survey.

Figure 3. Luminosity functions in K and bJ from the 6dFGS. Those from other recent surveys are also shown. Magnitudeoffsets for different passbands (where required) have been indicated in the key.


page 13

SC

IEN

CE

HIG

HL

IGH

TS

Finally, the Schechter fit has been convolved with themagnitude error distribution in each passband.

Of particular interest in Figure 3 is the upturn in galaxynumbers at the bright-end, an effect that has been notedby several authors previously. This excess of luminousobjects is due to the brightest cluster galaxies, whichare produced by the special merger and accretionprocesses that come into effect in the high-densityregime at the centre of cluster gravitational potentials.We examined the galaxies responsible for this upturnin some detail, to confirm that the effect was real andnot some measurement artifact. Figure 4 shows examplespectra from this luminous subset alongside 2MASSand SuperCOSMOS images. The imaging reveals thatthere are a number of close galaxy pairs in this sample.While they are not close enough to have inflatedmagnitudes due to image blending – 2MASS canindividually identify sources 5 arcsec apart – it supportsthe idea that many of these galaxies inhabit denseenvironments.

Figure 5 shows the luminosity density – a sort of volume-averaged spectral energy distribution (SED), obtainedby integrating over all five 6dFGS luminosity functions.The 6dFGS values agree with most estimates, finding aK-band value at the lower end of recent results for thisband. This is also the expectation of the SED for a 12Gyr-old stellar population with a 4 Gyr e-foldingexponentially decreasing star formation rate (dashedline, after Bell et al. 2003). Models with more rapidlydeclining star formation rates (say 2 Gyr) can produceSEDs more luminous in K, but at the expense ofoverestimating optical luminosities by factors of two ormore.

With the final 6dFGS photon captured, we celebratethe closing of another chapter in the illustrious historyof the UK Schmidt Telescope. We are grateful for thetalent and hard work of all Schmidt observers – pastand present – whose contributions to the 6dFGS havehelped in no small part to shape it into the excellentdataset it has become. Our sincere thanks are extendedto Donna Burton, Paul Cass, Kristin Fiegert, MalcolmHartley, Dionne James, Ken Russell, Fred Watson andthe late John Dawe. The legacy of the 6dFGS will bethe public availability of these data to general users,until such time as a more sensitive survey with equivalentfull-sky coverage pushes south of the equator.

More information on the 6dF Galaxy Survey can be foundat our web site: http://www.aao.gov.au/local/www/6df/ .

References:

Bell, E. F., et al., 2003, ApJS,149, 289Blanton, M. R., et al., 2005, ApJ,631, 208Cole, S., et al., (2dFGRS team),2001, MNRAS, 326, 255Driver, S. P., et al., 2005,MNRAS, 360, 81Eke, V. R., et al., 2005, MNRAS,362, 1233Jones, D. H., et al., 2004,MNRAS, 355, 747Jones, D. H., et al., 2005,PASA, 22, 277Jones, D. H., et al., 2006,MNRAS, submittedKochanek, C. S., et al., 2001,ApJ, 560, 566Loveday, J., et al., 1992, ApJ,400, L43Norberg, P., et al., (2dFGRSteam), 2002, MNRAS, 336, 907Zucca, E., et al., 1997, A&A,327, 477

Figure 5. Integrated luminosity density for 6dFGS acrossbJrFJHK. Comparative values from other recent surveys arealso shown. We have also reproduced the spectral energydistribution for a 12 Gyr-old stellar population from Bell etal. (2003). Details in the text.

(a)

(c) (d)

(b)

b K

b K

b Kb K

89"

60" 63"57"

60"

60"

73"

60"

Figure 4. Example spectra of the most luminous contributors to 6dFGS. Imagesfrom 2MASS (K-band) and SuperCOSMOS (bJ-band) are also shown. In these,north is up, east is left, and the field size in arcsec indicated in each corner.



OB

SE

RV

AT

OR

Y N

EW

S

THE RADIAL VELOCITYEXPERIMENT: A SURVEY TOEXPLORE THE DYNAMICAL ANDCHEMICAL EVOLUTION OF THEMILKY WAYThe RAVE team1

The Radial Velocity Experiment (RAVE) is an ambitiousspectroscopic survey of the southern hemisphere tomeasure radial velocities and stellar atmosphereparameters (temperature, metallicity, surface gravity) ofup to one million stars using the 6dF multi-objectspectrograph on the 1.2 m UK Schmidt Telescope ofthe Anglo-Australian Observatory (AAO). RAVE willprovide a giant leap forward in our understanding of theGalaxy, providing a vast stellar kinematic databaselarger than any other medium or high-resolution surveyproposed for the next ten years. It is a multinationalendeavour involving scientists from Australia, Canada,France, Germany, Italy, the Netherlands, Slovenia,Switzerland, the UK and the USA. The RAVE programstarted in April 2003 and so far has delivered over 80,000spectra in the Ca-triplet region (8410–8790 Å) forsouthern hemisphere stars in the magnitude range9<I<12 at a resolution of R=9000. The radial velocitiesmeasured in this survey are accurate to a few kilometresper second. The pilot phase of the survey has been lastdiscussed in the December 2004 issue of the Newsletter;here we summarize the survey motivation, design andthe verification of data included in the first data releasescheduled for February 2006. At this year’s AmericanAstronomical Society meeting in Washington its 3000attendees were presented with 10 contributionsdiscussing Galactic and stellar applications of collecteddata.

The case for RAVE

In the past ten years it has been increasingly recognizedthat many of the clues to the fundamental problem ofgalaxy formation lie locked up in the motions andchemical composition of stars in our Milky Way. Stellarspectroscopy plays a crucial role in unravelling thisinformation, not only providing radial velocities as a keycomponent of the 6-dimensional phase space of stellarpositions and velocities, but also providing much-neededinformation on the chemical composition of individualstars.

However, to our knowledge no systematic survey isplanned so far that provides continuous coverage of a

substantial fraction of the Galaxy, that includes radialvelocities, and is thus capable of filling this gap in sizeand time between existing astrometric surveys and theGAIA mission, which is expected to provide a completecensus of our Galaxy by the end of the next decade.

With the magnitude limit of RAVE (I<12), the survey willprimarily focus on stars in the extended solarneighbourhood out to distances of a few kpc. The mainscience goals are:

• Increasing kinematical and chemical databaseof Galactic stars by 2 orders of magnitude.

• Searching for unique chemical and kinematicalsignatures of stellar streams in the halo, outer bulgeand thick disk due to satellite accretion.

• Determining the dynamical influence of thelocal spiral arms and inner bar.

• Measuring the degree of ellipticity, warpingand lop-sidedness of the disk.

• Obtaining the first non-local measurement ofthe surface density of the disk, including the localescape velocity and so the overall mass of the MilkyWay.

• Studying detailed structure of the spiral armsand stellar associations.

Survey design and status

RAVE is designed to obtain spectra in the red Calciumtriplet region during the period 2003–2010. So far, RAVEhas gathered over 80,000 spectra for targets down togalactic latitudes of |b|=20° (see Fig. 1). The spatialcoverage of the survey is expected to become morehomogeneous as the observations proceed and as morefields are added to cover the gaps. Due to theunavailability of proper I magnitude, the pilot study(2003–2005) was built using the Tycho-2 and Super-Cosmos catalogs. The input catalog is divided into twosub-samples according to the pseudo-I magnitude. Thebright sample is defined with 9<I<11, the faint part using11<I<12. Both sub-samples are randomly selected in aspecified 5.7° circular region, corresponding to the 6dFfield of view. The a posteriori derived selection functionbased on DENIS I-band magnitudes is shown in thebottom panel of Fig. 2. No color selection was appliedto construct the input catalog. By the end of the survey,RAVE is expected to mimic a magnitude limited surveydown to a limiting magnitude of 12+ in each field.

RAVE started its full scale operation in July 2005. Dataare now obtained at a much increased pace makinguse of any useful night. So we expect that by 2010

1 see: http://www.rave-survey.aip.de/rave/pages/project/Team.jsp


page 15

OB

SE

RV

AT

OR

Y N

EW

S

RAVE may observe up to 1 million stars providing aunique sample to study the formation and evolution ofthe Galaxy.

Data processing and catalog validation

The RAVE survey can reach the science goals statedabove only by a very careful reduction of observedspectra. The issues of optimal spectral extraction,scattered light within the spectrograph, reliablewavelength calibration and background subtraction needto be studied in detail in order to allow a determinationof accurate radial velocity and stellar atmosphereparameters, i.e. its temperature, metallicity, gravity etc.So a dedicated IRAF pipeline for basic data reductionhas been developed and tested. Another dedicated IRAF-based pipeline is then used to calculate the values ofradial velocity and stellar atmosphere parameters. Itinvolves the minimum distance techniques, together withcross-correlation procedures (XCSAO). The appropriatetemplate is chosen from a set of ~20,000 syntheticspectra (~60,000 for the second year data) selectedfrom Munari et al. (2005) and Zwitter et al. (2004). Themain result is determination of the radial velocity to afew km/s. The distribution of the radial velocity internalerrors is given in Fig. 2. More specifically, 50% of the1st data release has radial velocity errors below 2 km/s,87% below 3 km/s. The figure shows the variation of

the internal accuracy with DENIS I magnitude for~18,000 stars in common with the 1st data release.The bottom panel presents the distribution of DENIS Imagnitudes for those stars. The two peaks in thisdistribution correspond to the faint and bright subset,respectively, as defined in the input catalog.

The zero point of our radial velocity solution is obtainedfrom sky emission lines in the Calcium triplet region.With our exposure time of 50 minutes and resolution of9000, the signal to noise ratio in those lines enables usto estimate the zero point with an accuracy of ~1 km/s.

During the course of the project, the radial velocitysolution is checked for stability and zero point accuracy.The stability is estimated with re-observation of a subsetof RAVE targets, while the zero point accuracy ischecked using external radial velocity data taken withother telescopes and derived from other surveys. Fig. 3and Fig. 4 present a summary for the first data release.In both cases, the agreement between RAVE radialvelocities and external or re-observed sources is inexcellent agreement with our expected uncertainties.We thus estimate an average total uncertainty for the1st data release to be ~3 km/s. This level of precisionis more than enough for almost all studies of galactickinematics and dynamics.

Fig. 1: Status of RAVE observations as of December 12th 2005 presented in equatorial coordinates. The color codingcorresponds to the number of times a field has been visited (including re-observations for calibration purposes). Yellowcorresponds to one, green to two, blue to three and red to four observations. So far over 80,000 spectra have beenobtained.



OB

SE

RV

AT

OR

Y N

EW

S

Fig. 2: Distribution of RAVE radial velocity error (internal) as a function of I magnitude for a subset of 18,000 stars of the firstRAVE data release that were observed also by DENIS (top panel). Bottom : distribution of DENIS I magnitudes for thesame sample. The double peak distribution is a property of the input catalog.

Fig. 3: Analysis of RAVE re-observations. This figure presents the variation around the mean RV for 428 RAVE targets (943individual spectra). The mean variation is consistent with zero while the rms is compatible with our internal error estimate,showing a good stability of our reduction.


page 17

OB

SE

RV

AT

OR

Y N

EW

S

Fig. 4: Comparison of RAVE RVs with RVs obtained in other programs. Blue corresponds to calibrationobservations performed with ELODIE, red to the Geneva-Copenhagen data and green to calibrationdata taken with the ANU 2.3 m. The general agreement of the RVs is very good, showing no significantzero point offset.

First data release

The first data release of the RAVE project is scheduledfor February 2006. The catalog will be accessible fromthe RAVE website (www.rave-survey.aip.de and itsmirrors), via the Virtual Observatory and from the VizieRdatabase @ CDS. This release will include radialvelocities for ~25,000 stars in 240 6dF fields. The catalogalso provides cross-identifications with the standardphotometric catalogs USNO-B, DENIS, Tycho-2 and2MASS, giving accurate optical and near-infraredmagnitudes for the sub-sample selection. Proper

motions are taken from Tycho-2 and UCAC. The medianaccuracy for the proper motion is ~5 mas/yr. Stellarparameters and spectra are not included in this firstrelease, but they will be an integral part of future datareleases. The fact that a typical noise per wavelengthbin is around 2% of the signal at I=10, and 5% at I=12indicates that the capabilities of RAVE reach well beyondradial velocities. Actually, AAO is the birthplace of thelargest spectroscopic survey of stellar properties yetundertaken.



HIS

TO

RY

OF

AS

TR

ON

OM

Y

MELBOURNE OBSERVATORY ANDTHE GENESIS OF ASTROPHYSICSIN AUSTRALIAJenny Andropoulos & Wayne Orchiston(James Cook University)

Out With the Old and in With the New

The second half of the nineteenth century witnessedthe emergence of the “new astronomy”, astrophysics,which gradually replaced positional astronomy.Positional astronomy was descriptive and dealt with whatcelestial objects looked like, where they were locatedand how they moved, while astrophysics was dynamicand examined the origins, compositions and evolutionof celestial bodies. Astrophysics flourished because itbrought two invaluable new analytical tools toastronomy: the spectroscope and the photographicplate. In this paper we shall focus solely on theemergence of astronomical spectroscopy in Australia.

At an international level, astronomical spectroscopy wasborn in 1814 when Munich’s Joseph Fraunhofer used atiny 2.5 cm telescope to view the spectra of Sirius. Nineyears later he examined Betelgeuse, Capella, Castor,Pollux, Procyon and Sirius with a 10 cm refractor, andnoted the dark absorption lines that now bear his name(see Hearnshaw, 1986). Stellar spectroscopy then hadto wait forty years before there was further progress,but when this did occur it involved astronomers fromthree different nations and two different continents.

During the early 1860s, Donati and Secchi in Italy, Airyand Huggins in England and Rutherfurd in New York allpublished papers on stellar spectra, marking what isnow recognized as the real start of astronomicalspectroscopy. Hearnshaw (1986: 55) remarks that “It iscertainly remarkable that all these pioneers should havebeen working practically simultaneously andindependently on similar problems.”

Of the five astronomers, Rutherfurd – who was anamateur, just like Huggins – was the first to attempt toclassify the spectra of stars into different groups. Heidentified three groups. Stars in the first group resembledthe Sun, were all reddish or golden coloured, and hadmany lines and bands in their spectra. In the secondgroup were white stars like Sirius, while the third groupalso contained white stars, but they displayed noabsorption lines. Today it is easy to associateRutherfurd’s groups with the MK system.

These initial attempts to classify stellar spectra werefollowed by others, and from the 1870s the ability topermanently preserve spectra on photographic platesmarked a new era in astronomical spectroscopy. In 1873

Vogel and Lohse at Potsdam observed the spectra ofall stars brighter than magnitude 4.5 between –10° and20° and developed their own classification scheme. Thiswas then widely used by astronomers, as was Secchi’searly scheme, but controversy continued to surroundthe interpretation of the different spectra and this wasto blight future research programs, including thosetentative studies carried out in Australia.

Spectroscopic Astronomy at MelbourneObservatory

Melbourne Observatory (Figure 1) was arguablyAustralia’s foremost nineteenth century professionalobservatory (see Haynes et al., 1996), and wasestablished in 1863 following the closure of the Flagstaffand Williamstown Observatories. Founding director atMelbourne Observatory (as at Williamstown) was thecharismatic Robert Ellery.

Holding pride of place at Melbourne Observatory from1869 was the 48 inch Great Melbourne Telescope(Figure 2), and soon after it became operational le Sueurcarried out spectroscopic observations of Eta Carina,publishing the following description in the Proceedingsof the Royal Society:

“The spectrum of this star is crossed with bright lines… The most marked lines I make out to be, if notcoincident with, very near to C, D, b, F and the principalgreen nitrogen line. There are possibly other lines, butthose mentioned are the only ones manageable. Theyellow (or orange?) line in the star has not yet receivedsufficient attention; it is however very near D …; atpresent it cannot be said whether the line may not beslightly more refrangible than D … Owing to the faintnessof the spectrum no dark lines were made out; one inred is strongly suspected. (le Sueur, 1870: 245).”

At that time Eta Carinae was – and indeed still is – oneof the most enigmatic stars in the entire sky, and wenow know that it would hardly be the object of choice if

Figure 1: Melbourne Observatory soon after itsfounding (Orchiston Collection)


page 19

HIS

TO

RY

OF

AS

TR

ON

OM

Yone were launching a new research program in stellarspectroscopy! It is little wonder that le Sueur made noattempt to interpret the spectrum that he described.

Yet this impasse did not dissuade Ellery from furtherspectroscopic forays and both he and his ultimatesuccessor, Pietro Baracchi, were to embark on these inthe late 1880s. But in each instance they made use of theObservatory’s 8 inch refractor (Figure 3) and a Macleandirect-vision spectroscope, rather than the Great MelbourneTelescope. Both surveys were completed by 1889 andpublished in two papers in Monthly Notices of the RoyalAstronomical Society (Ellery, 1889; Baracchi, 1889).

In all, the two surveys involved two hundred stars, but onlynine of these were common to both surveys. Seventy-fiveof the stars showed continuous spectra, and thepredominant colours were blue and green; violet was rarelyseen. About forty of the remaining stars revealed spectrawith some combination of very faint C, D, F and G Fraunhoferlines; many of these stars appeared to be white or yellow.The remaining eighty-five stars had Fraunhofer lines andbands in their spectra, and Ellery and Baracchi made mostfrequent mention of the F, G and C lines, but the E, b, Hand D lines were sometimes referred to. It was also notedthat Gamma Crucis, 20 Librae, Eta Sagittarii and Delta 2Gruis had ‘flutings’ towards the red ends of their spectra.

Nine stars appeared in both surveys, but only three of these(Beta Lupi, Pi Scorpii and Delta 2 Gruis) yielded consistentresults. Of the other six, Ellery neglected to include anydata for Beta Centauri in his paper, while he and Baracchidescribed differing spectra in the case of the other stars.For instance, Baracchi assigned a continuous spectrumto Beta Scorpii, while Ellery identified faint Fraunhofer linesin the blue and violet. In the case of Alpha Pavonis, Baracchidescribed a spectrum with dark bands and lines, includinga very thick dark band or groups of dark lines far in theviolet, and with similar bands near the Fraunhofer line Gand a dark band or group of dark lines near the F line. In

stark contrast, Ellery reported aspectrum with very faint Fraunhoferlines in the blue, violet, orange andred.

Despite the size of their collectivesample, all Ellery and Baracchi coulddo was describe the spectra theyobserved. What they could not do wasinterpret the spectra in any meaningfulway, and this merely reflected theconfusing state of spectralclassification at that time. In a sense,these two Melbourne astronomerslaunched their stellar spectroscopiccareers at just the wrong moment, for

had they waited a few short years they would havehad access to Wien’s Law. In 1893 Wilhelm Wiensuggested that the wavelength at which the radiatedenergy reached a maximum is inversely proportionalto the absolute temperature of the radiator: “Thecolour of a radiating body is thus a function of itssurface temperature … [and] At a stroke, stellarspectral classification became physically meaningful.Stars of spectral class M appear to be ‘red’, K orange,G yellow, F creamy, A white and B and O blue-white,with surface temperatures increasing from 3,000 Kto 35,000 K.” (Hughes, 2005: 108; our italics).

The spectral studies undertaken by Ellery andBaracchi were initially planned as forerunners to amore extensive survey to be carried out with the GreatMelbourne Telescope, but the afore-mentionedinterpretive problems, staff shortages, other observing

Figure 3: The 8 inch refractor at MelbourneObservatory (Orchiston Collection).

Figure 2: The Great Melbourne Telescope (courtesy: RAS Library)



HIS

TO

RY

OF

AS

TR

ON

OM

Y

commitments with the large reflector, and the newly-adopted Astrographic Program all conspired to preventthese laudable plans from coming to fruition.

Other Australian Spectroscopic Studies

Between 1869 and 1871, le Sueur – and perhaps otherMelbourne Observatory staff members – also used the48 in Great Melbourne Telescope to examine thespectrum of 30 Doradus in the LMC. The GreatMelbourne Telescope has a fascinating history (e.g. seePerdrix, 1992), but Malin’s claim that “In research terms,its results were clearly disappointing.” (Haynes et al.,1996: 107) is typical of the prevailing view. The first authorof this paper is currently carrying out doctoral researchon the observational work accomplished with thisinstrument, and one of her areas of special interest willbe to document and evaluate the entire suite ofspectroscopic observations made with the GreatMelbourne Telescope.

Melbourne Observatory was not the only Australianprofessional observatory to delve into astronomicalspectroscopy during the second half of the nineteenthcentury. In 1881 Sydney Observatory director, HenryChamberlain Russell, subjected the Great Comet of thatyear (C/1881 K1 Tebbutt) to spectroscopic scrutiny andpresented his findings in a paper published in the Journaland Proceedings of the Royal Society of New SouthWales (Russell, 1881). This comet (Figure 4) made afortuitous appearance as both astronomicalspectroscopy and astronomical photography werebeginning to play a crucial rôle in internationalastronomy and it was possible to apply both of thesetechniques to this comet. Consequently, it came tooccupy an important place in the history of cometary

astronomy (Orchiston, 1999), but because Russell’sspectroscopic paper was ‘buried’ in a local non-astronomical journal it failed to reach an internationalaudience and its true importance was all but lost.

Finally, it should be mentioned that as elsewhere in theworld, amateur astronomers in Australia were quick tofamiliarize themselves with new developments inastrophysics. In Hobart, Tasmania’s foremostastronomer, Francis Abbott, sought to popularizeastronomy by publishing three booklets in 1878, 1878and 1880, and all of these contained readable up-to-date information about the emerging field of astrophysics(see Orchiston, 1992). Meanwhile, Tasmania’s second-ranked astronomer at this time, Alfred Barrett Biggs(Figure 5), was fascinated by the brilliant red sunsets of1884 and subjected them to protracted spectroscopicanalysis (Orchiston, 1985). He even went and publisheda paper on these so-called “sky glows” in the Papersand Proceedings of the Royal Society of Tasmania(Biggs, 1884). With the benefit of hindsight, we nowknow this impressive atmospheric phenomenon wasassociated with the spectacular eruption of Krakatoa inwhat was then the Dutch East Indies.

References

Baracchi, P., 1889. MNRAS, 49, 439.Biggs, A.B., 1884. Papers Proc. Roy. Soc. Tasm., 202.Ellery, R.L.J., 1889.. MNRAS, 50, 66.Haynes, R., Haynes, R., Malin, D., and McGee, D., 1996.

Explorers of the Southern Sky. A History of AustralianAstronomy. Cambridge University Press.

Hearnshaw, J.B., 1986. The Analysis of Starlight. OneHundred and Fifty Years of Astronomical Spectroscopy.Cambridge University Press.

Hughes, D.W., 2005. J. Astr. Hist. & Heritage, 8, 107.Le Sueur, A., 1870. Proc. Roy. Soc., 18, 245.Orchiston, W., 1985. Rec. Queen Victoria Museum, 89, 1.Orchiston, W., 1992. Vistas in Astron., 35, 315.Orchiston, W., 1999. Irish Astron. J., 26, 33.Perdrix, J.L., 1992. Austr. J. Astron., 4, 149.Russell, H.C., 1881. J. Proc. Roy. Soc. NSW, 15, 81.

Figure 5: Alfred BarrettBiggs (OrchistonCollection).

Figure 4: The Great Comet of 1881(courtesy Chapin Library, WilliamsCollege).


page 21

OB

SE

RV

AT

OR

Y N

EW

SAAOMEGA IS COMMISSIONEDRob Sharp and Will Saunders for theAAOmega team

AAOmega was commissioned in three runs inNovember, December and January. November markedfirst light for the new fibre feed, 39 m in total from therefurbished 2dF top end down into Coude West on thefourth floor. Simply routing the snake-like fibre-opticcable, encased in its armoured conduit, safely downthe Coude mirror train, proved quite a feat for theindomitable day crew, who worked without (much)complaint in the full and terrible knowledge of the wrathof Kristin, John and the fibres team should anythinghappen to the fibres en route.

Once safely out of the telescope, the fibres enteredCoude West and were removed from their protectivetransit torpedo to be mounted in the AAOmega slitexchanger, and system tests began in earnest. For theNovember run, the 2dF field plates were not fullypopulated, with only enough fibres and guide bundlesfor testing purposes being mounted. 2dF had at thistime been through a major 10th birthday refit and upgrade,and there was much about the system that we neededto shake down. Additionally, a manufacturing defect inan LN2 dewar meant that only the blue arm was inoperation for this first run. The run was a great success– integrating a system as complex as AAOmega withthe telescope control and support astronomer is a majortask and is one which requires AAOmega to be actuallyon the telescope and observing in order to complete.Test driving the control software, and confirming theastrometric accuracy of a number of 2dF top-end relatedupgrades progressed well and the blue camera achievedfirst light on a field of astrometric Tycho-2 stars. Theseobservations may represent some of the least interestingspectra to be taken with the AAOmega system duringwhat is hoped to be a long and illustrious career.

For the second commissioning run in December, botharms of the spectrograph were operational and the fieldplates fully populated with their 800 fibres (all 32 km-worth). While the first run focused on engineering andinstrument control, the second run focused on multi-wavelength and multi-resolution observations of realscience targets. High resolution radial velocity data wastaken, to investigate what will be achievable in the future.Data was also taken for a number of fields from the2SLAQ survey, to provide comparison data on faint LRGsand quasars.

As we go to press, the third and final commissioningrun and the subsequent Science Verificationobservations have just been completed. Data has been

taken for 12 programs in a wide variety of setups anddata-taking modes. Nod&shuffle and cross-beamswitching have been successfully commissioned, andsome test mini-shuffled data (nod&shuffle using all fibres)looks very promising. Although many loose ends remainto be tidied up, AAOmega is now a working instrument.

Instrument performance and sensitivity

There were several areas where we were particularlyconcerned that the instrument live up to predictions.These were:

1) Would the fibres, with their input buttons andoutput slitlets, have the uniformity and quality required?

2) Would the light propagate through 39 m of fibreoptics without significant focal ratio degradation?

3) Would the spectrograph optics perform asdesigned, giving the required imaging quality anduniformity (crucial for accurate sky subtraction and radialvelocities)?

4) Would the overall throughput yield the expectedgains of a factor of 2–3 with respect to 2dF?

The answers to all these questions look to be ‘yes’ (orat least ‘probably’). The uniformity of the data is superb,with fibre throughputs varying by about 10% both inabsolute terms and spectrophotometric variations. Thefigure on the back cover illustrates this – it is 2 hours ofco-added raw, low-resolution red data on faint targets –350 simultaneous, long-slit-quality spectra.

The point spread function is remarkably Gaussian. It isextremely stable across the CCDs – varying in width byonly a few percent for a given wavelength (as requiredfor 1% sky subtraction). The blue camera gives imagesthat meet or exceed specification (3–3.25 pixel FWHM).The red camera focus is less sharp (3.2–3.5 pixel FWHM,though still constant); there are known imperfections inthe red camera alignment which will be rectified beforeroutine science observations begin in late February.

The red arm throughput looks to be 75–80% of thepredicted value. This is despite a telescope primarymirror that has not been cleaned for a year or aluminisedfor 2 years (re-aluminising will occur mid-February), andsome pupil misalignment in the spectrograph. So weare confident that the red arm will live up to throughputexpectations.

For the blue arm, the throughput is still unknown. Thefield lens in the camera remains frosted (from assemblyin a hot and humid Sydney), and the moisture has provedstubborn to remove. A baffle accidentally both anodisedand painted (hence trapping large amounts of moisture)



OB

SE

RV

AT

OR

Y N

EW

S

is strongly suspected to bethe cause of the problem.The amount of light lost tothis frost – judging from itsreflectivity and the scatteredlight on the detector – isclearly large, and we cannotestimate a reliablethroughput at this stage. Astrip-down of the bluecamera is programmedbefore the first routinescience observations in lateFebruary. The blue CCD isalso less good cosmeticallythan the red one, withseveral bad columns nearthe middle of the detectorwhich must be interpolatedover.

Sky subtraction accuracy, using the traditional dedicatedsky fibre technique will be crucial to many AAOmegaprojects, particularly at longer wavelengths. Preliminaryextractions show sky subtraction often (but not always)better than a few percent and sometimes 1%. Withimprovements to the calibration system and datareduction software in the coming months, we hope toreach 1% consistently. This would mean that manyprograms will not have to resort to nod&shuffletechniques, which have significant efficiency overheadscompared to dedicated sky-fibres.

Data reduction

An essential part of the success of 2dF was theavailability of a fast, automated reduction packageallowing observers to leave the telescope with fullyreduced data under their belt. For AAOmega, we requirereductions of equal robustness, but with a much higherlevel of precision to do justice to the data. As we go topress, work on upgrading 2dfdr to AAOmegadr isunderway. Basic reductions are now straightforward,though some workarounds remain necessary at thisstage. Further improvements and tuning will beimplemented as we gather more experience with theinstrument.

Figure 1: An r=20.23 galaxy spectrum at low resolution

Figure 2: An r=22.03 galaxy observed with nod and shuffle in two hours


page 23

OB

SE

RV

AT

OR

Y N

EW

S

Figure 3: High resolution R=10 000 spectrum of a K=10.77 star in the calciumtriplet region with S/N>100, revealing a wealth of absorption lines as well as theobvious calcium triplet features.

Figure 4: An r=19.86 galaxy with [OII] emission at z=0.37 as observed by the blue AAOmegacamera with the 580V grating.



OB

SE

RV

AT

OR

Y N

EW

S

Sample data

Here we show some commissioning data showing whatAAOmega is capable of in its various setups and modes.

Figure 1 shows an r=20.23 spectrum from the raw datashown above. The sky subtraction accuracy is about2%; we are still working on improving this accuracy downto the target of 1%. Such an accuracy would rendernod&shuffle unnecessary for most programs (it is 2–8times less efficient than observations with dedicatedsky fibres).

For the deepest data, nod&shuffle will still be needed,to achieve Poisson-limited sky subtraction. Figure 2shows an r=22.03 nod&shuffle spectrum representingjust 2 hours ON + 2 hours OFF data.

Figure 3 shows high resolution Calcium triplet data fora reddened K=10.77 star, with S/N>100 and resolutionR=10,000, revealing a wealth of lines for abundancestudies and promising sub-km/s radial velocityaccuracy.

Figure 4 shows an r=19.86 galaxy with [OII] emissionat z=0.37 as observed by the blue AAOmega camerawith the 580V grating (2.5 hrs exposure). In this plotwe have zoomed in on the interesting region of thespectrum (the actual spectrum extends from 370nm to580nm) and show the plotting interface within 2dfdr.

SPIRAL IFU

As well as feeds from the two 2dF field plates, AAOmegawill also support integral field observations with arefurbished SPIRAL IFU mounted at Cass. The SPIRALsystem will be available with any AAOmega gratingconfiguration, and will give 50% higher spectral resolutionbecause of its smaller fibres. SPIRAL has a field of viewof 22x11 arcsec. SPIRAL IFU mode is due forcommissioning in June 2006.

This article is dedicated to Terry Bridges, for hisdedicated work as project scientist in the design stages,and also for the wonderful and sadly missed bottle ofwhisky he left with us to celebrate first light.

A NEW FIELD CONFIGURATIONALGORITHM FOR AAOMEGABrent Miszalski (Macquarie), KeithShortridge (AAO), Will Saunders (AAO),Quentin Parker (AAO/Macquarie) andScott Croom (AAO)

Introduction

The complexity of multi-object spectroscopy (MOS)instruments such as 2dF has tended to divert attentionfrom some aspects of the observation process. Perhapsthe most important area that has not been fully exploredis the field configuration algorithm (FCA) used to selecttargets for observation. The FCA is primarily responsiblefor allocating as many targets as possible, therebymaintaining the multiplex advantage of the instrument.However, there are other more subtle FCA issues, suchas target sampling, which must be as uniform aspossible. In some cases an FCA, by its inherent design,can imprint artificial power or bias onto observed targetdistributions selected by the FCA. Such subtle biasescan be detrimental to statistical analyses applied tolarge-scale astronomical surveys that attempt toproduce a fair representation of large-scale structure inthe Universe.

The advent of the AAOmega spectrograph for 2dF hasmotivated an investigation into current FCA strategiesand alternative methods within the stable and maturedevelopment environment of 2dF CONFIGURE. Thedefault ‘Oxford’ algorithm for CONFIGURE was designedspecifically for the 2dF Galaxy Redshift Survey(2dFGRS; see §5.1 of Colless et al. 2001) to achievehigh target yield for fields with a similar number of targetsand fibres. Indeed, the ‘Oxford’ algorithm achieved itstask with acclaim, contributing to the success of the2dFGRS.

Despite its heavy usage during the lifetime of 2dF, it isonly recently that sufficiently sensitive measurementsof the sampling behaviour of the ‘Oxford’ algorithm havebeen made by Outram (2004) and Miszalski (2005).These tests have revealed that the fundamental designof the ‘Oxford’ algorithm does not satisfactorily deal withthe higher density, multiple-priority fields anticipated tobe the staple diet of AAOmega.

Here we present an overview of a newly developed FCAbased on simulated annealing (SA) that is currently beingintegrated into 2dF CONFIGURE. It addresses all themajor shortcomings of previous FCAs and providessupport for new observational modes available withAAOmega. It is planned to become the default FCA forAAOmega observations. The generic nature of the


page 25

OB

SE

RV

AT

OR

Y N

EW

Salgorithm means that it can be used for 6dF fields, andis suitable for VLT FLAMES, Subaru FMOS and futureMOS instruments. Further details are presented inMiszalski (2005) and Miszalski et al. (2006, in prep).

Algorithm Overview

SA is a generic method of solving constrainedoptimisation problems, developed by Kirkpatrick et al.(1983). SA employs the probabilistic Metropolisalgorithm (Metropolis et al. 1953) to simulate the thermalmotion of a system (in this case a field) whilst slowlycooling the system to simulate the annealing processof physical systems. Many small, random perturbations(fibre swaps) are made at each stage in the coolingprocess (known as the annealing schedule). Theperturbations explore the set of possible solutions (theparameter space) under the guidance of the Metropolisalgorithm that is designed to efficiently find near-optimalsolutions. If slow cooling takes place then the solutionobtained may also be uniform, with progressively finerstructure being ‘ironed-out’ with decreasing temperature.SA has been successfully applied to problems similarto field configuration, most notably the tiling of the 6dFGalaxy Survey (Campbell et al. 2004).

We designed the new algorithm to satisfy the followingoptimality criteria:

• High overall yield independent of target priority.

• Highest priority targets have highest possibleyields.

• Uniform sampling overall and for each priority.

• Observational flexibility.

The optimality criteria ensure that the multiplexadvantage of the instrument is maintained, that targetpriorities are weighted correctly, that minimal artificialpower is imprinted on target distributions, and thatdifferent sky subtraction techniques and a large varietyof fields are supported. These criteria are not mutuallyexclusive. For instance, it would be undesirable toallocate sky targets at the expense of high priority targetswhen low priority targets remain allocated.

The randomisation process of the SA algorithm entails~105 fibre swaps. Traditionally CONFIGURE spends amajority of its time re-calculating collisions betweenfibres and buttons each time a new allocation is made.We have developed a novel allocation sub-system thatpre-calculates all possible collisions within a field andstores them within a sparse, indexed collision matrix.This has made the previously prohibitive amount of swapseasily achievable – within minutes rather than hours!The removal of this bottleneck has enabled the SAalgorithm to explore much more of the parameter spaceof a field than previous FCAs to produce a more optimalsolution.

Algorithm Advantages

The optimality criteria are largely satisfied by theobjective function of the SA algorithm. The objectivefunction is a measure of the quality of the fieldconfiguration, i.e. how optimal the configuration is. It isthis quantity that the algorithm endeavours to maximiseto obtain the most optimal solution possible. It alsoprovides a simple yet powerful and flexible means toinfluence the final solution. Additional terms can beintroduced to maximise such quantities as fibre

Fig. 1: Gain in overall target yield of the SA algorithm over the ‘Oxford’ algorithm. The left panel shows the gainfor fields extracted from mock 2dF catalogues (Cole et al. 1998) with moderate target clustering. The rightpanel shows the yields and gains (inset) obtained from generated Gaussian fields parameterised by sigma(lower sigma corresponds to higher central clustering).



OB

SE

RV

AT

OR

Y N

EW

S

straightness or close pair yield. For simplicity we coverhere the advantages of the SA algorithm rather thandelve into the objective function and its implications.

The new algorithm provides the following advantagesover the ‘Oxford’ algorithm:

• Increase in overall target yield by up to 11%(~5% for typical fields) (Fig. 1).

• Increase in target yield of highest prioritytargets by up to 30% (Fig. 2).

• Much more uniform sampling of targets andremoval of the panoply of artificial structuresimprinted by the ‘Oxford’ algorithm at high targetdensities (e.g. Fig. 3).

• Removal of previously unforeseen artificialstructure in different priority distributions (Fig. 4).

• Effective allocation of sky targets byreassigning fibres from lowest priority targets (Fig. 2).

• Support for nod&shuffle with Cross-BeamSwitching (Saunders 2005) with at least 98 pairsallocated under difficult circumstances (Fig. 5).

• Ability to weight close target pairs of givenseparation.

• Well documented implementation withcharacterised sampling statistics in terms of two-

Fig. 2: Average priority distributions for fields with 100 targets of each priority configuredby the ‘Oxford’ (left column) and SA (right column) algorithms. The handling of targetpriorities by the SA algorithm produces a very optimal priority distribution with and withoutsky targets allocated. The ‘infiltration’ of low priority targets in the ‘Oxford’ algorithm (top)is further compounded by the poor redistribution of fibres to sky targets (bottom).

point correlation functions, completeness and targetyields.

• Simplified algorithm maintenance via objectivefunction to allow future support of specialisedobservational requirements.

Algorithm Limitations

The advantages leave few limitations that can beassociated with the new algorithm. Such performancedoes entail a time penalty, but this is not necessarilyproblematic.

The most significant of these penalties is the calculationof the collision matrix that allows for orders of magnitudegreater fibre swaps than previous algorithms. This inturn allows more optimal solutions to be found than werehitherto possible. Fig. 6 shows actual calculation timesfor fields with uniformly distributed targets, in additionto fields with moderately clustered targets from mock2dF catalogues. The data show that on average it takes~5 minutes to set up the collision matrix on a reasonablypowerful desktop computer.

Further limitations to the collision matrix setup includean intolerance of fields with very high target numbers orheavily clustered targets. Such fields may be handledprovided that sufficient physical memory is available(~2GB recommended). However, such fields may involvelengthy collision matrix setup times of ~30–60 minutes.


page 27

OB

SE

RV

AT

OR

Y N

EW

S

For these fields a reduction in field complexity is requiredto reduce collision matrix setup times to acceptablevalues. This is not to say that this will be the norm formost users; indeed the vast majority of fields configuredduring algorithm development were accommodated bymodest computer specifications.

The actual algorithm runtime is not so easilybenchmarked as the collision matrix setup time,primarily because of the large number of variablesinvolved. The default runtime of the annealing schedule(the heart of the algorithm) is typically ~5 minutes,roughly the same time as the collision matrix setuptime. This is highly dependent on the amount ofclustering in a field and specific algorithm parametersthat determine the number of fibre swaps made at eachtemperature decrement, the latter of which can also be

controlled. Fortunately, we plan to provide the user withat least three presets of thorough, intermediate and quickconfiguration speeds. More detail on these settings willbe provided in revised CONFIGURE documentation.

Fortunately, the collision matrix is reusable and oncecalculated it will be stored as a binary ‘.matrix’ file (typicalsize ~50–100MB). This enables the user to reconfigurea field with different algorithm parameters as many timesas desired until either an optimal solution is found orthe field changes (either field centre changes or targetsare changed). The latter requires that the matrix berecalculated before configuration can recommence. Theinclusion of sky targets is therefore recommendedbefore the collision matrix is established. All theaforementioned limitations classify the algorithm assemi-interactive. Indeed, the algorithm generates

Fig. 3: Completeness C(x,y) (fraction ofallocated targets) as a function of field plateposition for fields with uniformly distributedtargets (number as indicated) configured bythe ‘Oxford’ (top) and SA (bottom) algorithms.The SA algorithm eliminates all of theartificial structure imprinted by the ‘Oxford’algorithm at high target densities save foronly a slight gradient present in the radialdirection.

Fig. 4: Radial completeness C(r) (C(x,y) averaged over theta) for fields with 200 P9, 200 P8, 100 P7 and 100 P6 uniformlydistributed targets configured by the ‘Oxford’ (left) and SA (right) algorithms. The SA algorithm completely removes thepanoply of previously unknown structure imprinted on lower priority targets.



OB

SE

RV

AT

OR

Y N

EW

S

sufficiently optimal solutions that batch modefield configuration is a viable alternative thatmay be utilised by tiling algorithms ofupcoming AAOmega surveys.

Algorithm Status

The algorithm is currently implemented withina new batch version of CONFIGUREdeveloped alongside the algorithm. It iscurrently being reworked back into thestandard CONFIGURE. This work alsoinvolves the:

• Delegation of the old FCAs to‘expert’ mode.

• Simplification of the new algorithmparameters.

• Collision matrix integration to betransparent to the user.

• Implementation of a histogram ofpriority yields and a plot of objective functionwith temperature to show optimality ofconfigured field.

There will be an updated CONFIGURE manualwith sample fields and example runs of thealgorithm explained to help ease thetransition. It is anticipated that the new versionof CONFIGURE will be released before mid

2006.

Acknowledgements

BM acknowledges an AAO/MacquarieHonours scholarship that enabled thiswork. The authors wish to thank P.J.Outram for his foresight and assistance.

References

Campbell, L., Saunders, W., Colless, M.2004, MNRAS, 350, 1467

Cole, S., Hatton, S., Weinberg, D. H.,Frenk, C. S. 1998, MNRAS, 300, 945

Donnelly, R. H., Brodie, J. P., Allen, S. L.1992, PASP, 104, 752

Kirkpatrick, S., Gelatt, C. D., Vecchi, M. P.1983, Science, 220, 671

Metropolis, N., Rosenbluth, A.,Rosenbluth, M., Teller, A., Teller, E.1953, Journal of Chemical Physics,21, 1087

Miszalski, B. 2005, Honours Thesis,Macquarie University

Miszalski, B., et al. (2006, in prep)Outram, P.J. 2004, http://www.aao.gov.au/

local/www/brent/configureSaunders, W. 2005, AAO Newsletter,

Number 108, Page 8

Fig. 6: Time needed to setup the collision matrix before the actual FCAmay configure a field. The main series with error bars are fields withNTargets uniformly distributed targets. Other data represent moderatelyclustered fields from mock 2dF catalogues. Statistics obtained with a2.4 GHz P4 Xeon CPU.

Fig. 5: Average number of allocated pairs from fields with 400 uniformlydistributed target pairs with a variety of separations and orientations. The SAalgorithm (filled circles) consistently outperforms the ‘Oxford’ algorithm (opencircles) by 33%. At least 98 pairs are allocated in all cases, meeting the limitof observable target pairs – a constraint imposed by the AAOmega CCDarea.


page 29

OB

SE

RV

AT

OR

Y N

EW

S

NOTES FROM AATACMartin Asplund, AATAC chair (RSAA, ANU)

The inaugural meeting of the Anglo-Australian TimeAllocation Committee (AATAC) was held at the AAOheadquarters in Epping on November 2–3, 2005. Gonenow are the days when AU and UK astronomers wouldsubmit proposals to separate TACs (ATAC and PATT;ATAC will however continue to handle AU applicationsfor Gemini time). The institution of a new joint committeeshould improve the efficiency of the AAT as well asstimulate further scientific collaborations between thetwo countries. The new AATAC is very much lookingforward to helping ensure that the AAT remains productiveand will continue to generate high-impact science.

The format of AATAC is largely moulded on ATAC. It willhave seven members, currently four from AU and threefrom UK; as the UK financial contribution to AAO willramp down in the coming years, the membership ofAATAC will be reviewed annually by the AAT Board inorder to reflect the partner shares. The board has decidedthat the AATAC chair will always be Australian while thedeputy chair will be British. The current membership is:

• Martin Asplund (chair, ANU),

• Michael Drinkwater (Univ. of Queensland),

• Seb Oliver (Univ. of Sussex),

• Sean Ryan (deputy chair, Univ. of Hertfordshire),

• Peter Tuthill (Univ. of Sydney),

• Jacco van Loon (Keele Univ.) and

• Rachel Webster (Univ. of Melbourne).

The members of the service time sub-committee arepresently Peter Tuthill (AU representative), Sean Ryan(UK) and Antony Horton (AAO). In general, AATAC willnot rely on external referees to assess proposals buthas the discretion to do so when the need arises, suchas for large programs (more below) and for applicationsoutside the scientific expertise of the various AATACmembers. The application deadline will be March 15(“B” semesters) and September 15 (“A” semesters).

The allocation of observing time will roughly reflect thepartner shares. Naturally, this can not and should notbe done exactly. Instead, the AAO Director hassuggested an allocation formula, which hassubsequently been endorsed by the AAT Board andadopted by AATAC. The formula is designed to stimulatecontinued collaboration between the two countries whilebeing cognizant of the varying partner contributions; a

detailed description of the accounting of the observingtime is available in the August 2005 AAONewsletter (p. 17). For the recent 06A round, this workedextremely well, with both AU and UK receiving almostexactly their actual share (imbalance relative to partnershare <0.5 night).

The oversubscription rate for 06A was a healthy 3.6(bright time), 2.5 (grey) and 2.1 (dark). This semesterwas the first with a dedicated call for large programs(>50 nights) with AATAC aiming to set aside >20% ofthe available time for large programs in future years.Although this new feature was instigated by theupcoming availability of the new AAOmega spectrograph,large programs can be proposed for any of the AAT facilityinstruments (AAOmega, UCLES, UHRF, WFI andIRIS2). For 06A there were 11 proposals for largeprograms. AATAC was very pleased with the largenumber of such proposals presenting compelling sciencecases of the highest quality. Besides being extensivelydiscussed within AATAC, these proposals were alsoexternally reviewed. Each proposal was sent to threereferees with expertise in the relevant field forassessment; the return rate for reports was quite high(after some prodding and begging...) with all proposalsreceiving at least two referee reports.

For 06A, AATAC decided to take a somewhatconservative approach, only approving one such largeprogram, the Anglo-Australian Planet Search program(PI: Chris Tinney). The main reason for this is thatAAOmega – the instrument most large programsproposed to use – had not yet been commissioned andits on-telescope efficiency therefore not accuratelyknown. Furthermore, many of the programs requiredpilot studies to demonstrate convincingly that thetechniques and methods are appropriate before justifyingallocating a large number of AAT nights. Individualfeedback on each proposal has been distributed to thePIs. Another call for large programs will be made in 06Band thereafter in each B semester. AATAC remainscommitted to the goal of awarding >20% (with nopredetermined upper limit) of the time on AAT to suchlarge projects.

Please do not hesitate to contact the chair([email protected]) or any of the other membersif you have questions about AATAC and its work. Wealso appreciate feedback from the AU and UKcommunities, in particular now in the beginning as weare still trying to decide on appropriate ways to operateand the policies for the new committee. For the interestedreader, the current version of the AATAC guidelines isavailable at: http://www.aao.gov.au/AAO/astro/apply/aatacguidelines.v3.pdf



LO

CA

L N

EW

S

AUSTRALIAN SUMMER STUDENTSFOR 2005/06Scott Croom

The Anglo-Australian Observatory runs a summerstudent program for undergraduates from both the UK(during the northern summer) and Australia (during thesouthern summer). The current Australian students areShane Hengst, from Macquarie University, Ryan Cookefrom Queensland University of Technology and ChrisBanks from the University of Queensland. They areworking on a number of observational projects usingdata from the AAT and also the VLT.

Shane Hengst is working with Heath Jones on deep on/off-band narrowband imaging of 20 low-redshift (z ~ 0.3and 0.45) QSOs. The 20 QSOs are drawn from theoptically-selected Calan-Tololo QSO Survey of J. Mazaand collaborators. The imaging was obtained with FORS2on the ESO VLT and is being put to two uses. The firstis to reveal the possible existence of extended nebularline emission on 10 to 100 kpc scales in the vicinity ofthe QSO. The second purpose is to locate emission-line galaxies at the same redshift as the QSO and whosepresence would indicate clustering in the QSOenvironment. One would expect to see a connectionbetween these two phenomena if cluster cooling flowsare indeed the mechanism responsible for extendedQSO emission. Supporting broad band imaging providesa ready discriminant between foreground and backgroundemission lines.

This observational project marks a change from thetheoretical work that Shane has done previously. His2005 Honours thesis at Macquarie University modelleddust grains in different turbulent structures,encapsulating the dynamics of the dust grains in proto-planetary discs.

Ryan Cooke is working with Joss Bland-Hawthorn onnew data taken with the VLT VIMOS IFU. Theobservations are of an emission line nebula thatsurrounds the x-ray binary source LMC X-1. Theseobservations were inspired by a TTF H-alpha image ofthe source that appears to show streamers possiblyarising from a jet flow. A new Spitzer mid-infrared imagereveals that the emission line nebula sits within a “hole”in the dust distribution which may support the idea of awind-blown cavity. The new spectroscopic data cubesshould tell us whether the streamers arise from a jetimpacting on the inside of a dusty cloud.

The VIMOS IFU obtained a total of 4 hours of data onthe source in four contiguous fields. These data are beingreduced with the recently released VIMOS reductionpipeline software. The final data cube will have dimensions

150 x 150 x 2048, and will be visualized using Karmasoftware developed for radio astronomy. Emission linewidths, line kinematics and line ratios will be examinedat each spectrum within the cube. If the jet picture iscorrect, the properties of the emission will be highlycomplex over the full field, and this will be the firstevidence of a jet in this extraordinary source.

Chris Banks is working with Rob Sharp and Tim Schmidt(from the Physical Chemistry group at SydneyUniversity) on the reduction and analysis of UCLESspectroscopy of the Red Rectangle nebula.

The Red Rectangle, one of the brightest sources in themid-IR as seen by IRAS, is a proto-planetary nebulawith some unusual spectral properties. The spectrumexhibits unidentified molecular band emission at a widerange of wavelengths, much of which is commonlyclaimed to be due to emission from Polycyclic AromaticHydrocarbon molecules (PAHs). Identifying the carriersof these bands will require detailed laboratoryspectroscopy, which is in progress in the Sydney Laserlabs. However, accurate astronomical spectra arerequired to compare to the laboratory data.

The unusual environment of the nebula complicates thestructure of the molecular emission bands. Highresolution, spatial resolved spectroscopy of the nebulais required to disentangle the emission structure. OurUCLES data was taken in support of Integral Fieldobservations obtained with the ARGUS IFU on the VLT(through the allocation of two nights of guaranteed timefrom delivery of the Oz-Poz instrument by the AAO).While UCLES provides only the one spatial dimensionwith its 8 arcsecond long slit, the extended wavelengthcoverage provided is the complementary information werequire to interpret the ARGUS spectra.

Weather was not kind during the observing run.However, in a single night UCLES was able to recordalmost complete wavelength coverage from 4000–8000Åand at a resolution of R~40,000, covering most of themolecular emission features we wish to cross-referencewith the limited wavelength coverage IFU data. The taskof reducing the long slit echelle spectra is one whichChris is currently taking in his stride. Once the data isreduced Chris will determine the wavelength and band-widths for a range of molecular emission features seenin the data.


page 31

LO

CA

L N

EW

SLETTER FROM COONABARABRANRhonda Martin

Wishing our lives away, we look pantingly forward tothe autumn when it will be, hopefully, cooler and we cangain relief from the breathtaking, mind-numbing heat thathas plagued us of late. Alternatively, we can eagerlycheck the schedule for Paul Butler’s next visit – itALWAYS rains when Paul makes a visit. Usually atCoona we can cope with the heat because the nightsare cool, but lately ………… My goodness, it’s likeliving on the coast. Without any rain since earlyDecember lawns are crunchy and leaves hang limply,so – we await Paul.

It doesn’t help, of course, to have Frank Freemanextolling the wonders of retirement and sending picturesof white beaches and big fish.

But then – the project that has driven us for what seemsforever, has made staff bleary-eyed and short of temper,has taxed ingenuity to the edge of brilliance and been abit of a windfall for fibres manufacturers – AAOmega,arrived at Site along with a horde of Epping staff. Iremember Greg Smith complaining about the cold.Cold? What cold? The old Mappitt room is now theAAOmega room and very smart it looks too. But thegreat thing is, it works! There are glitches, of course,but these are being ironed out at a rate of knots.

It’s all gone rather quiet now – the long tables havebeen removed from the walkway, the fibres lab is nolonger frenetically busy or full of people helping get thefibres all together, the 4th Floor is tidy again (more orless) and we have toasted a job well done with whisky,wine and water (not all together). Congratulations aredue to a lot of tired people taking a well-earned rest.

Site is also host to John Danson, a local lad studyingelectrical engineering at ADFA. He will be with theelectronics people for the next six weeks or so.

Frank Freeman working hard in retirement!

EPPING NEWSSandra Ricketts

There have been quite a number of comings and goingssince the last newsletter, spread over a couple ofgenerations.

The previous writer of this column, Greta Simms, left ustowards the end of last year, after 10 years at the AAO.She was farewelled at a barbecue at Epping in November.We are pleased to have Carolyn Hampele back againto make sure we are all paid! Pat Roche returned to theUK in December after his sabbatical at the AAO, andJohn Storey also returned to UNSW after his half-timesabbatical

We welcomed Simon O’Toole who will be working onAAO Planet Search data, as well as Antony Horton andMike Birchall who will both be working with the softwaregroup. Antony will be working on TCS and Mike on theFMOS DR project. We hope they enjoy their time atthe AAO.

Congratulations are due to both Chris Tinney and ScottCroom on the births of Nicholas William Tinney andSamuel William Scott Croom in November. The AAOseems to be having another baby boom!

LIBRARY NEWSSandra Ricketts

Life in the AAO library continues much as usual. Towardsthe end of last year library users were asked to respondto a survey on their use of the library. Various usefulsuggestions were made, and by and large people sawthe future of the library as being a mixture of physicaland virtual resources. It is hoped that the varioussuggestions can be acted upon this year.

Some changes have been made in the library at thetelescope – there is a new computer (newer and fastereven than the one this newsletter is being producedon!), as well as a new scanner and printer. Somerearrangement of furniture has also taken place. It ishoped that this equipment will be used more by libraryusers at the AAT – it is not for the exclusive use of thelibrarian by any means!

Don't forget that the major astronomy journals are nolonger received in hard copy at the telescope, but ofcourse are available on-line – use the new PC (and printerif necessary). Older issues are still on the shelves invarious places – the library itself, the corridor outside,and study 4.

TH

E B

AC

K P

AG

E

editor SCOTT CROOM editorial assistant SANDRA RICKETTS

ISSN 0728-5833Published by ANGLO-AUSTRALIAN OBSERVATORYPO Box 296 Epping, NSW 1710 Australia

Epping LabTelephone +61 2 9372 4800 Fax +61 2 9372 4880 email < [email protected] >

AAT/SchmidtTelephone +61 2 6842 6291 Fax +61 2 6884 2298 email < [email protected] >

URL < http://www.aao.gov.au >

Raw low-resolution data from the red arm of AAOmega (2 hours coadded)

Raw AAOmega data at low resolution

Successful commissioning of AAOmega - Australian · PDF file · 2014-04-22fact if one overlooks some frantic activity associated ... and WHT is coalescing into a coherent picture

Documents