Galaxy using the Planetary Nebula Spectrograph

17  Download (0)

Full text


Galaxy using the Planetary Nebula Spectrograph

H. R. Merrett

, M. R. Merrifield , N. G. Douglas

, K. Kuijken

, A. J. Romanowsky

, N. R. Napolitano

, M. Arnaboldi

, M. Capaccioli

, K. C. Freeman

, O. Gerhard


L. Coccato


, D. Carter

, N. W. Evans

, M. I. Wilkinson

, C. Halliday

, T. J. Bridges

The University of Nottingham, NG7 2RD, UK

Kapteyn Institute, The Netherlands

Leiden Observatory, PO Box 9513, NL-2300 RA Leiden, The Netherlands

Departamento de F´ısica, Universidad de Concepci´on, Casilla 160-C, Concepci´on, Chile

Osservatorio di Capodimonte, Via Moiariello 16, Naples 80131, Italy

Research School of Astronomy and Astrophysics, Australian National University, Canberra ACT 2601, Australia

Astronomisches Institut, Universit¨at Basel, Venusstrasse 7, CH 4102 Binningen, Switzerland

Astrophysics Research Institute, Liverpool John Moores University, Twelve Quays House, Egerton Wharf, Birkenhead CH41 1LD, UK

Institute of Astronomy, Madingley Road, Cambridge CB3 OHA, UK

 Osservatorio Astronomico di Padova, Vicolo dell’Osservatorio 5, I-35122 Padova, Italy

 Anglo-Australian Observatory, Epping, NSW 1710, Australia

Accepted 2005 ????? ??. Received 2005 ?????? ??; in original form 2005 July


We present a catalogue of positions, magnitudes and velocities for 3300 emission–line objects found using the Planetary Nebula Spectrograph (PN.S) to survey the Andromeda galaxy with.

Of these 2615 are likely to be planetary nebulae associated with M31. The survey area covers the whole of M31’s disk out to a radius of 1.5 . Beyond this observations have been made along the major and minor axes, the northern spur and southern stream regions. Emission–

line objects in the vicinity of and associated with satellite and background galaxies have been identified.

With the exception of the very central, high surface brightness, region of M31, our survey is complete to a magnitude limit of! " "#%$


, 3.5 magnitudes into the PN luminosity function.

Estimated accuracies of the measured parameters are 0.16''in declination; 0.34''in right ascension; 0.1 for the magnitude of the 5007 ˚A [OIII] line; and 17 kms(*) for our velocities.

The cutoff between HIIregions and PNe is investigated and a limit on the [OIII] to (H+ + [NII]) flux ratio for PNe is proposed such that,.-[OIII]/102,3-H+ + [NII]/5476 &89:

 1 1 ;#=<

 & :

.The Planetary Nebula Luminosity Function has been plotted for our sample and is found to be in excellent agreement with the usual function both for the galaxy as a whole and for radially binned data. No evidence is found of varaition in any kinematic parameters with magnitude. The distribution of PNe as a function of distance along the major and minor axes is also found to be in excellent agreement with the respective surface brightness profiles.

Key words: Local Group – galaxies: individual: M31 – galaxies: kinematics and dynamics – galaxies: structure


The rotation of a spiral galaxy is the principal means for deter- mining its mass distribution, understanding its dynamics, evolution and ultimately formation. Rotation curves are most easily and most commonly measured using neutral hydrogen, integrated starlight

> E-mail:

and emission lines that trace interstellar gases. Fabry-Perot spec- trometers and integral field spectrographs allow measurements to be made in 2 dimensions over the whole of a galaxy’s disk. These methods, however, are based on the bulk motion of stars or gas in the disk. Until recently it has been impractical, even in our clos- est neighbours, to measure velocities of individual stars in external galaxies in large enough quantities to provide a serviceable tracer population.


The Andromeda galaxy, our nearest large spiral neighbour

(@BADC=CFE kpc), provides a unique observational target. Its size

and proximity serve as both a help and hindrance to observation:

structural details can be seen yet it is difficult and time consuming to cover the entire area covered by the galaxy. Previous dynamical studies have either measured gas in the disk (e.g. HIBraun (1991) and HIIRubin & Ford (1970)) or tracer populations in the halo (e.g.

globular clusters Perrett et al. (2002); and more recently, RGB stars Reitzel & Guhathakurta (2002) and Ibata et al. (2004)).

Planetary nebulae have long been recognised as a potential source for a dynamical tracer population, Nolthenius & Ford (1984) report on kinematic measurements of 34 PNe in M31. Not until re- cently, with the development of fibre spectroscopy, has it been pos- sible to take this further (Hurley-Keller et al. 2004), (Halliday et al.

2006). However, this is still a difficult and time consuming process.

The recent commissioning of the Planetary Nebula Spectrograph (PN.S) (Douglas et al. 2002) on the William Herschel Telescope in La Palma has eased this task.

We have performed a survey, using PN.S, covering the bulk of Andromeda’s visible area. Thirteen nights (eight of which were clear) were allocated over two observing runs at WHT to perform this survey. An additional six night run (five clear) was allocated to use the wide field camera on the Isaac Newton Telescope, also in La Palma.

The PN.S survey covers some 5.5 square-degrees on the sky taking in the entire disk out to a radius of 1.5G ; extending along both the major and minor axes, the southern stream and northern spur; as well as a selection of halo fields picked out by the INT survey. We have detected and measured velocities for some 3300 objects, 2730 of which are most likely to be PNe.

This paper takes the following form. InH2 we briefly describe PN.S and the observations made. A detailed description of the data reduction is presented inH3 along with calibration against the fibre spectroscopy data of Halliday et al. (2006), additional data from Massey et al. (2002) and comparisons to other published data sets.

InH 4 PNe in that are associated with other galaxies within our sur- vey are are discussed. H6 presents fits to the planetary nebula lu- minosity function and H7 a comparison to the surface brightness profile.


2.1 The Planetary Nebula Spectrograph

The Planetary Nebula Spectrograph (PN.S) was designed specifi- cally to measure the wavelength (and hence velocity) of the 5007 ˚A [OIII] emission line that dominates the spectra of planetary nebu- lae. It does this using a technique called “counter–dispersed imag- ing” (CDI). The basic principle of CDI being that light passes through a narrow band filter (AJIK ˚A) and is then dispersed by a pair of oppositely oriented gratings onto separate CCDs, producing a pair of spectrographic images. Stellar continua appear as elon- gated streaks and PNe are visible as dots from the [OIII] emission lines.

Comparing the positions of the emission dots in the two im- ages relative to a calibration mask allows the wavelength, position and magnitude of the PN to be calculated from just one observa- tion. A full description of the instrument can be found in Douglas et al. (2002).

PN.S’s ability to yield so much information from a single observation makes it extremely efficient, considerably more so

than the traditional technique of optical identification and follow–

up spectroscopy. The major drawback being that PN.S has a somewhat lower velocity accuracy than traditional spectroscopy (A 15-20 kmsL  compared toA 2-10 kmsL (Halliday et al. 2006), (Hurley-Keller et al. 2004), (Ciardullo et al. 2004)). However, this is balanced by the increased number of objects detected for the same amount of time at the telescope.

2.2 Observations

We have used PN.S on the William Herschel Telescope in La Palma on two occasions: October 8th to 13th 2002 and September 29th to October 5th 2003. The Wide Field Camera on the Isaac Newton Telescope, also in La Palma, was used August 16th to 21st 2003.

Fig. 1 shows the locations of observed fields.

During the initial PN.S run a straightforward tiling strategy was adopted to cover the disk area (Fig. 1, fields 1–163). The INT run that followed used on–off band pre-imaging to identify candi- dates in the halo where PNe are scarce, thus increasing observa- tional efficiency (fields INT*). These candidates were followed up as well as weather conditions permitted during the second PN.S run. At this time the tiled area was also extended out along the ma- jor axis, the inner stream region and northern spur (fields 200–399).

Only halo fields were pre-imaged.

PN.S was used with the EEV12 and EEV13 CCDs on the left and right arms, respectively. These have 2150M 4200, 13.5N m pix- els, which are windowed down to 2150M 2500 pixels for use with PN.S. Pixel scales are 3.32 pixels/arcsecond in the dispersion di- rection and 3.67 pixels/arcsecond in the spatial direction. Readout noise in EEV12 and EEV13 are 3.5 and 3.6 electrons; gains are 1.16 and 1.20 electrons per ADU, respectively.

Weather conditions varied considerably during the PN.S runs with 2 half nights lost to cloud cover during the first run and 4 nights lost in the second. Seeing varied from 0.8–3.1OO, see Fig. 2.

Our ‘A’ filter was used with 0G tilt. This has a central wave- length of 5002.2 ˚A and a 36.5 ˚A bandwidth (the equivalent velocity range forPQKEEC is -1369–817 kmsL ).

Fifteen minute exposures were used for the majority of fields, multiple observations were made for a few fields and shorter ex- posures were used where bright PN candidates had been identified from the INT runs. INT WFC exposures were twenty minutes long when using the [OIII] filter and two minutes with the gO filter.


The INT-WFC data was debiassed and flat-fielded. The gO band im- ages were remapped to match the [OIII] images.SExtractor was used in double image mode to find sources in the [OIII] image and measure the flux of counterparts in the gO band. Objects with more than five times as many [OIII] counts as gO were selected as can- didates. Thumbnails of these objects were made and visually con- firmed. Rough positions were calculated based on location within the field, more accurate astrometry was not required as PN.S has a large field of view.

The Planetary Nebula Spectrograph is a unique instrument and as such requires a purpose built pipeline to reduce the data. This takes the form of anIRAF pipeline with Fortran and PG PLOT extensions.


Figure 1. Observed fields. The small square fields are PN.S field locations; the larger fields with dotted outlines are the INT WFC fields. PN.S fields located within the INT fields are following up pre-identified candidate objects. Poor weather conditions mean not all these objects have had follow observations. The ellipse marks a 2G disk radius (26.8 kpc), the approximate radial extent of the optical disk (Walterbos & Kennicutt 1988)

3.1 Reduction Pipeline

Images are debiased by subtracting a surface function fit to the under- and over-scan regions (2nd order polynomial inR and 5th order inS ). The function produced is very close to flat.

The bias frames are used to create a map of pixels where charge transfer problems have occurred and skyflats are used to make maps of the bad pixel locations (done by combining flats and comparing to a median smoothed version, pixels with greater than 9T difference between the two are taken to be bad). The two chips differ in quality with EEV12 having relatively few bad pixelsA 300 and EEV13 having significantly moreA 18000. This is clear in the images with the right hand, EEV13, images displaying a number of columns of bad pixels.

These maps are then combined with a list of known bad pixel regions to make a mask which is fed to the standardIRAF routine fixpix to correct all the flats and science images.

Normalised pixel response maps are made by dividing flats by a median-filtered image (to remove spatial structures) then com- bining with weights (to keep the noise Poissonian) and rejection (to

eliminate cosmic ray events), the correction range is then clipped at 10%. The science and arc images are corrected by dividing through by the map.

The next step is to remove the residues of cosmic rays else they will be transformed into objects with a finite PSF at the next stage of reduction and may produce false detections. This has been done using van Dokkum (2001)’s Laplacian cosmic ray identifi- cation routine, using 5 iterations and a contrast limit between the cosmic rays and the underlying object of 1.2.

The cleaned science images contain curvatures and rotations which must be eliminated in order to align the left-right image pairs. This is done using IRAF’s standard geometrical mapping routines,geomap and geotran, with parameters calculated from masks which have been taken throughout the observations. These calibration images are short exposure arc frames taken through a mask that is moved into the focal plane by a motor. The mask has 178 regularly spaced holes in it each giving rise to a spectrum (see Fig. 3). The locations of the brightest emission lines (P 5017) are compared to those in an ‘ideal’ mask, thus mapping the distortion.

(The ‘ideal’ mask is defined to be the average of the left and right


Figure 2. Seeing variations across the survey. Light greys representing good seeing; dark, poor.

Figure 3. Detailed arc spectrum seen through filter-A set at 0G arm masks taken on a specific date.) The required image transfor- mation is derived from this and saved as a database file. A separate routine reads the database and transforms the required image, cre- ating a new one.

Two other wavelengths, P 4990 and P 5009, are used in con- junction with the bright line to characterise the dispersion across the image using a 3rd order polynomial solution, seeH3.3.

Following the spatial correction step it is possible to co-ad images in cases where multiple observations of the same field have been taken. Positions of startrails relating to stars in the USNO- B catalogue Monet et al. (2003) are identified. A transformation function is calculated to map subsequent images onto the initial

‘reference’ image. This function is convolved with the undistortion function, so that there is only one interpolation of the data. Weights are calculated for each image in the stack. TheIRAF routine fitprofs

Figure 4.UWV cut to the data set

is used to fit a one-dimensional Gaussian across each star trail. This fit is used to measure the transparency, seeing and background level of each field and the subsequent weight is calculated from

XZY []\^1_:`bac!dFegfhidjclk;egm1nQo




3.2 Identification of Emission–line objects

Emission–line objects are identified by a semi-automated routine.

The initial step usesSExtractor, run on the undistorted, median sub- tracted images and at the same positions on an image that has been convolved with an elliptical Gaussian function elongated in the dispersion direction. A cut is then performed to select only those points where the median subtracted and Gaussian convolved val- ues are different, thereby eliminating detections of stellar continua.

Detections close to the edges are also eliminated.

Sources are initially matched in left–right pairs where they have suitableRQy{z|RQ} values and~S:y|z{S:}~ less than 2 pixels.

These pairs are passed on to aFortran routine which allows the two images to be viewed and examined side–by–side. Any false detec- tions, such as points within extended structures, are eliminated at this point. Also any detections that have been missed, usually due to proximity to the field edge or a star, are added back in. These lists of emission–line objects are passes tophot for centring and photometry (using an aperture radius equal to the mean FWHM seeing for that field), thenimexam to measure the Moffat FWHM using the ‘,’ command.

Once final positions have been found it is possible to improve upon the two pixel€2S limit by performing a statistical 3T clip on the mean value, see Fig. 4, producing upper and lower limits of 1.17 and -1.80 pixels.

3.3 Wavelength Calibration

As was mentioned previously, calibration masks were taken at reg- ular intervals whilst observing. These fields have been used to char- acterise the geometric distortions in the images and are also used to calibrate the wavelength information in the form of a file with 178

M the three wavelengths recognised.

These positions are compared with those in the ‘ideal’ mask and a mulitvariant fit is computed between the measuredR.S posi- tions and the meta-parameters:













This information is saved in the form of coefficients of various combinations of terms. Given the left and right–hand positions of a point–like objects this can be inverted such that`R]‡3S2‡3Rˆ‰‚S2ˆ o

can be used to obtain`R  S  !P o.


Figure 5. Comparison to Halliday et al. (2006) velocities

Figure 6. Velocity Shifts within the PN.S data set

3.3.1 Comparison to Halliday et al. (2006)

A kinematic survey of PNe in the Andromeda galaxy has been per- formed by Halliday et al. (2006) (hereafter H06). Velocities were measured for 723 PNe. This provides an exceptional source for comparison of velocities against the PN.S data set. The original tiling of the PN.S survey was designed to cover the same area as this survey, and we find 715 of the H06 PN within 4OOof a PN.S object. The few that are missing are likely obscured in the PN.S images behind stellar trails.

Plotting the difference in the two velocities against the PN.S velocity, as in Fig. 5, shows a slight trend and a general shift. Fitting a straight line through this gives the trend as

Š‹9Œ.Ž z






Estimating M31’s system velocity from the PN.S data shows a similar shift with respect to the standard value. It therefore seems likely that the PN.S wavelength calibration has a systematic shift, which can be corrected for using this relation. All subsequent PN.S velocities have been corrected for thus. The source of this shift has yet to be identified.

Figure 7. Comparison to Halliday et al. (2006) velocities

Figure 8. Positional shifts within the PN.S data set

Fitting a Gaussian to the corrected velocity difference gives a combined dispersion of 15.4 kmsL  (Fig. 7).

It is also possible to compare velocities within the PN.S data set. There are 732 objects detected in more than one field (643 in two fields, 88 in three fields, 1 in four fields), comparing the ve- locities for these objects allows an approximate dispersion to be calculated for PN.S, see Fig. 6. A Gaussian fit to this data gives



kmsL , a dispersion from PN.S ofA”WC kmsL  . This value is larger than the total dispersion from PN.S and the H06 data set combined, indicating a lower velocity accuracy at the field edges where the majority of the duplicates are located than in the centre of the field. Considering that our wavelength solution varies across the field this is quite reasonable.

The system velocity is estimated from our data by averaging the mean velocities from radial bins on either side of the major axis. A value ofAJIEK kmsL  is obtained, in agreement with the standard value of 300 kmsL .

3.4 Astrometry

Astrometry is performed using the same basic routine as for image stacking. Star positions are read in from theUSNO B catalogue.

Using an approximate position shift and rotation, the stars are found in the PN.S images and centred usingxregister. These are fed to ccmap to calculate the astrometric solution for the field.

In the central fields, the bright background means there are very fewUSNO stars available. In these fields PN positions from neighbouring fields have been added to the star list that is fed to ccmap.

The astrometric solutions produced are somewhat better in declination than right ascension (Fig. 8). Gaussian fits to these give



E…I “ OO andT— \™˜ Y E…rWš OO. RA corresponds to the disper-

sion direction in the spectrographic images, centring up on stars by fitting gaussians is therefore less accurate in this direction and field–to–field shifts of up to 1 arcsecond are observed, thus ex-


Figure 9. Position comparison to H06. The black line represents all the PN in both catalogues; the grey shaded areas those PNe in our central fields (60, 61, 70, 71). Our positions for these central fields are offset by +1OOin declination and show a large spread.

panding the overall spread in RA positional accuracy. This level of accuracy is adequate for our purposes.

3.4.1 Comparison to Halliday et al. (2006)

Astrometry can also be compared against H06’s data set, shown in Fig. 9. The RA distribution is similar to that found within the PN.S data,T Yœ›T ‹9Œ. Žž T  jŸ

™ !



“¡ . In declination a general offset of +0.18OO, with a long tail to the side is seen. The PNe in this tail are all located in the central four PN.S fields (60, 61, 70 and 71), the grey shaded histogram in Fig. 9. As has been mentioned before, it was difficult to obtain astrometry for these central fields, with calculated PN positions from one field being used to map the next. Therefore a shift such as this, with a similar value for all four fields is not unreasonable.

3.5 Flux Calibrations vignetting...

The system efficiency for PN.S with Filter A at 0G tilt as mounted on the William Herscel Telescope is estimated from spec- trophotometric standard stars, see Table 1. This is done by cal- culating the total counts from the star within a column through the brightest region of the stellar trail. The observed flux per unit wavelength is then calculated and compared to the published value.

The average system efficiency is found to be 30.3¢ 0.5 % (15.3 % through left and 15.0 % through the right arm).

Total fluxes are calculated as follows



Y¤`¥ ‡ …¦ ‡  ¥ ˆ …¦ ˆ o‚§



[ \±^1_



½!¾¿ Ÿ

(2) where¥ ‡ and¥ ˆ are the total counts from the left and right arms;

¦ ‡ and¦ ˆ the gains;¯ the collecting area;[ \±^"_ the exposure time,

Table 1. Instrumental efficiencies calculated from spectrophotometric stan- dard stars.

Star F‚ Date Efficiency

(erg sL 


) Left Right Total

LDS 749BÀ 139.339 2002-10-10 0.1582 0.1629 0.3211 BD+33 2642Á 5732.797 2002-10-10 0.1450 0.1469 0.2918 G193-74Â 57.117 2002-10-13 0.1703 0.1547 0.3250 BD+17 4708Â 16479.320 2003-09-29 0.1615 0.1594 0.3210 BD+28 4211Â 7840.906 2002-10-08 0.1517 0.1472 0.2989 BD+28 4211Â 7840.906 2002-10-10 0.1478 0.1433 0.2911 BD+28 4211Â 7840.906 2002-10-11 0.1475 0.1442 0.2917 BD+28 4211Â 7840.906 2002-10-11 0.1468 0.1457 0.2925 BD+28 4211Â 7840.906 2002-10-12 0.1478 0.1442 0.2920 BD+28 4211Â 7840.906 2003-09-29 0.1597 0.1570 0.3167 BD+28 4211Â 7840.906 2003-09-30 0.1522 0.1435 0.2957

References:À Oke (1974),Á Oke & Gunn (1983), Oke (1990)

Table 2. Extinction per unit airmass inà O from the Carlsberg Meridian Tele- scope. The conversion to Ext‚ is calculated from Table 1 in RGO/La Palma Technical Note 31 to be a factor 1.70.

Night ExtÄ!Å Error Ext‚ Error Comments 2002-10-08 0.101 0.004 0.172 0.007 Rain at end 2002-10-09 0.129: 0.029 0.219: 0.049 Cloud first half 2002-10-10 0.106 0.006 0.180 0.010

2002-10-11 0.107 0.005 0.182 0.009 2002-10-12 0.106 0.005 0.180 0.009 2002-10-13 0.112 0.009 0.190 0.015 2003-09-29 0.234: 0.024 0.398: 0.041

2003-09-30 0.213: 0.012 0.362: 0.020 Cloudy at start 2003-10-01 0.317: 0.068 0.539: 0.116 Mostly cloudy

2003-10-02 - - - - Mostly rain

2003-10-03 - - - - Rain and cloud

2003-10-04 - - - - Rain

2003-10-05 - - - - Rain

k1Æa!Ç°È!É the nightly sky extinction taken from the Carlsberg Merid-

ian Telescope (Table 2) multiplied by the airmass of the field; and


­® the Galactic extinction taken from Schlegel et al. (1998) and corrected to our filter central wavelength using the relation from Cardelli et al. (1989) giving a value ofk"Æ9a"Ê




š .

The conversion to magnitudes follows the Jacoby (1989) rela- tionship


‚ Y z ‘ …KÍÌv=s`


‚ o z|WI…C “

(3) The spread in magnitudes from the duplicated PNe within the PN.S catalogue is shown in Fig. 10. A Gaussian fit to this distribu- tion givesTQÎ ¿™Ï¼Ï‚Ð Y E9…EC .

Magnitudes for extended objects will be systematically under- estimated by this technique as no allowance has been made in the choice of aperture size for objects that are larger than the field p.s.f.

3.5.1 Fluxes from The Local Group Survey, Massey et al. (2002) In recent years the Andromeda galaxy has been a popular target for observation, being the subject of a number of different surveys. The Local Group Survey of Massey et al. (2002) (hereafter M02) has been particularly useful for comparing to the PN.S data set. This project aims to image the major constituents of the local group at a


Figure 10. Magnitude variations within the PN.S data set.

Figure 11. Positions of other surveys w.r.t the PN.S survey. The PN.S sur- vey area is shown in grey; Massey et al. (2002)’s Local Group Survey fields are the large fields with solid outlines; the Ciardullo et al. (1989) data are the solid black points near the centre; the open circles are the Hurley-Keller et al. (2004) data; and the small crosses are H06’s fibre spectroscopy data.

number of wavelengths. In the case of M31, M02 have imaged 10 fields covering the majority of the disk of M31 and hence a large portion of the PN.S survey (some 2745 objects), see Fig. 11. We have used the M02 [OIII] imaging to check our flux calibration and the HÑ + [NII] imaging to help eliminate HIIregions.

Fluxes were obtained at the positions of our sources using a simplephot examination, with a fixed aperture of 8 pixels – some- what larger then the seeing FWHM ( 3.2-5.7 pixels) to allow for our positional variations – without re-centring. Factors ofI9…Ò ‘ MӐWE L 


 were used to convert the [OIII] and HÑ + [NII] photon counts to fluxes. Galactic extinctions were calculated in from Schlegel et al. (1998) and Cardelli et al. (1989) in the same way as before, yielding extinctions of 0.225 mag and 0.164 mag for [OIII] and HÑ + [NII] respectively.

Even with this simplified method for photometry we find ex- cellent agreement between the PN.S and M02 data after making a few simple cuts (Fig. 12). PN brighter thanA 24.4 mag are excluded as objects fainter than this show large dispersion. Possible HIIre- gions (H3.8) are also excluded as their fluxes are expected to be un-

Figure 12. Magnitude comparison to Massey et al. (2002) images. In the upper panel the black outline represents all the overlapping objects and the grey shaded area PN brighter thanÖØ×ÚÙ;ÛÜÛ . A Gaussian fit to this dis- tribution has a shift in magnitude of 0.02 andÝÞגß=Üà . The lower panels show these cuts, the larger area histogram being the total cut made the lower histograms being for the individual cuts.

derestimated in the PN.S images. Fitting a Gaussian to a histogram of the 1769 PN remaining gives a magnitude shift of 0.02 mag and a combined dispersion of 0.16 mag. There was no evidence of dif- ferent offsets for different fields or data taken on different nights.

3.6 Duplicates from Fields Overlaps

Duplicate detections from field overlap areas have been eliminated by averaging the fluxes, and velocities for object with FWHM dif- ferences less than 1.5 pixels (0.5OO). For those with differences larger than this the information for the better resolved object has been used.

3.7 Geometry

M31 system coordinates`bá ‚â o and`p㠁ä o are calculated following the geometric relations of Huchra et al. (1991).

á Y

fq†e.`påçæ z åèæ

o2m;vfF`péêo (4)

â Y


 o z m;v=fj`påèæ z åçæ


 o (5)

ã Y z áªfqre3`±ëæèo zÞâ m;vfj`±ë]æèo

Ÿ (6)

ä Y z á*m;v=fj`±ëæèo

 â fqre3`±ëæèo (7)

For M31åèæ  Y EE=ì “‘힓=“ DžI ,é  Yî“  G Wš OE=Ò OO (J2000.0);

and the position angle of the major axis, ëæ Y I=C9…GC (de Vau- couleurs 1958). Positiveã is located southwest of the centre of M31 and positiveä , northwest.

3.8 HIIRegions v. PNe

Identification of emission–line objects in PN.S images is fairly un- ambiguous. Yet there being no other wavelengths available is not


Figure 13. FWHM/seeing cut to the data set.

possible to distinguish PNe from HIIregions or supernova rem- nants by any means other than spatial extent.

Bright PNe in the Milky Way are small, less than AB pc across Acker et al. (1992) corresponding toA 0…OO27 at the distance of M31. In the small Magellanic cloud PNe are observed with di- ameters typically around 1pc, though a few are larger than this, up to 3pc in diameter,A 0O…O80 at the distance of M31. At best our res- olution isA 0O…O85, hence all spatially resolved objects must be HII

regions, or supernova remnants.

Attempts were made at the identification stage to exclude obviously extended or structured objects. Distinguishing between marginally extended and point sources was not possible by eye.

Hence, plotting a histogram of the ratio FWHM/seeing (Fig.13) shows a long tail of extended objects. An upper limit of 1.255 has been imposed on this value in order to maintain a normal distribu- tion and eliminate extended objects.

When examining the distribution of the extended objects it was found that they largely follow a ring around M31’s centre, as would be expected of a population of HIIregions. However, a clump of objects were seen originating in one PN.S field. This field was the only one taken near the centre of M31 under exceptionally good seeing conditions (A 0O…O86, see Fig. 2). With the seeing this good the variation in instrumental focus across the field and differ- ences between the two PN.S images becomes significant. Hence, these mainly faint, marginally extended objects may actually be PN but will be excluded for consistency.

3.8.1 Flux Ratios from The Local Group Survey, Massey et al.


The ratio of [OIII] to (HÑ + [NII]) fluxes (} ) can help distinguish between PNe and HIIregions. PNe emit most of their energy via the [OIII] transition, hence have large values of} . Attempts to quantify this for bright PNe have been made (Ciardullo et al. 2004) (Ciardullo et al. 2002), leading to cutoff limits in} ofA 1–2. How- ever, it is clear in these data sets and in ours (see Fig. 14) that a single value cutoff between PN and HIIregions at all magnitudes is not sufficient.

We have attempted to quantify how} varies with magnitude by examining histograms of the flux ratio at regular magnitude bins, see Fig. 14. The distribution of extended objects (shaded his- tograms) is clearly distinct from the rest of the sample for the first three magnitude bins. For objects with‘ Iðï

Ë ï

‘F“ there appears to be two populations of extended objects, this is due in part to the uncertainty in the FWHM cutoff. Fainter still the extended objects largely overlap with the non-extended population. We have chosen to only eliminate objects that are clearly different to the bulk of the population and have estimated this point in the top four magnitude

Figure 14. Flux ratio: [OIII] to Hñ + [NII]. Extended objects are shown in black, the rest in grey. The histograms show the same data in 1 mag bins.

Approximate limits between the two populations are indicated by black lines on the histograms and solid triangles in the upper panel, where a least–

squares–fit has been applied producing the cutoff shown by the dashed line.

bins (indicated by lines and points in Fig. 14). This gives a cutoff value of}”ò’zóE…‘ IE



“ …šI . For objects with


‚ ò



the PN.S and LGS magnitudes show significant scatter, however at these fainter magnitudes the cutoff requires bright HÑ fluxes which should still stand out, hence the cutoff has been applied to all objects.

Figure 15 shows the locations in the (ã ,ä ) and (ã ,Š ) planes of these extended and flux ratio selected objects. Nearly all of the objects with particularly non-M31 like velocities are flagged in this sample and are likely to be emission from background galaxies.

The remaining objects are mainly located in the disk of the galaxy in the star forming ring (radius = 0.8G ) discussed by Devereux et al.

(1994) and the spiral arms, as would be expected of HIIregions.

3.9 Completeness of the Sample

Comparing the distribution of PNe with radius at different magni- tudes allows us to estimate the limit of the complete sample. This


1 0:41:01.7 42:07:58.6 25.48 -283.3 - - - - - -

2 0:40:35.4 42:10:32.0 22.16 -133.9 - - - - - -

3 0:39:35.9 42:09:39.5 25.51 268.9 - 2MASXi - - - -

4 0:39:37.3 42:09:59.6 23.52 354.5 E 2MASXi - - - -

5 0:39:31.6 42:11:56.6 23.28 -343.3 - - - - - -

6 0:38:54.0 42:09:39.3 25.36 -266.7 E - - - - -

7 0:37:54.3 42:14:49.3 23.37 -328.1 - - - - - -

8 0:43:11.6 42:07:13.0 22.31 -194.0 - - PN 10 3 1 - - -

9 0:43:10.5 42:11:33.3 23.58 -286.1 - - PN 10 3 2 - - -

10 0:43:09.5 42:14:24.2 24.87 -314.2 - - - - - -

11 0:42:49.6 42:15:03.7 24.38 107.3 E - - - - -

12 0:44:08.6 42:06:00.0 25.62 -202.5 - - - - - -

13 0:44:09.8 42:06:40.2 22.99 -108.0 - - PN 10 3 7 - - -

14 0:43:31.0 42:09:25.7 24.20 -136.6 - - - - - -

15 0:43:39.8 42:13:46.3 25.07 -210.8 - - - - - -

16 0:43:37.6 42:06:58.6 24.65 -554.4 E Stream? - - - -

17 0:44:59.8 42:07:44.8 21.76 -193.6 - - PN 9 3 4 - - 1091

18 0:45:00.9 42:08:44.8 23.53 -72.2 - - - - - -

19 0:44:30.5 42:08:55.2 22.36 -81.9 - - PN 10 3 8 - - 1005

20 0:44:51.4 42:09:34.2 23.81 -148.4 E,R - - - - -

... ... ... ... ... ... ... ... ... ... ...

À Probable HIIregions (or background galaxies): E – extended in the PN.S data, R – Flux ratio÷ below cutoff value (see Fig. 14), other – excluded as a PN due to flux .

Á Sources of non-M31 objects: M32 / M32? – Objects associated with and in the vicinity of M32. NGC205 / NGC205? – Objects associated with and in the vicinity of NGC205. AndIV – Objects associated with the galaxy Andromeda IV.

2MASXi – Non-M31 source in the 2MASS catalogue. MLA93 – Non-M31 source listed in Meyssonnier et al. (1993) NS – Objects in the region of the Northern Spur. Stream? – Objects referred to in Merrett et al. (2003) as being a possible extension of the Southern Stream.

IDs from other catalogues: H06 - Halliday et al. (2006), C89 - Ciardullo et al. (1989), HK04 - Hurley-Keller et al. (2004), MLA93 - Meyssonnier et al. (1993).

can be seen in Fig. 16, where the bright and intermediate PN show distributions that are statistically indistinguishable, but the faint PN display a dearth of objects at small radii. This is expected given a combination of the high surface brightness of the bulge region and the poor seeing conditions during observations of the central fields.

Overall the sample is complete to




I over the entire survey;


‚ Yø‘

I…C=K at radii larger than 0.2G ; and


‚ A ‘ K

beyond 1G .

In terms of position, M31 has been completely covered out to a disk radius of 1.5G . Beyond that radius the major and minor axes are fairly well sampled though unevenly from one side to the other.

3.10 Comparison to Ciardullo et al. (1989)

Comparison to the Ciardullo et al. (1989) data set shows a good match in magnitude (Fig. 17). A Gaussian fit centres at zóE9…EE=I=K with a combined dispersion of 0.13 mag. This is similar to the dis- persion found when comparing to M02 fluxes. Again there was no evidence of different offsets for different fields or data taken on different nights.

3.11 Comparison to Hurley-Keller et al. (2004)

Hurley-Keller et al. (2004) have surveyed one quadrant of M31’s halo finding 135 PNe, see Fig. 11. We have covered nearly their

whole survey area finding all but 3 of their PNe. We also extended to somewhat fainter magnitudes.

Comparing our positions to those in Hurley-Keller et al.

(2004)’s produces very similar error estimates as those estimated from the duplicates in our sample.

Velocities have also been compared (Fig. 18). A Gaussian fit to this distribution gives a combined dispersion of 12 kmsL , lower that the calculated PN.S velocty error. This Gaussian does not fit to the bump atŠ ‹9Œ.Ž z Š Qù Y  IE , hence it is likely this underesti- mates the dispersion. The PNe in this bump are scattered through- out the area, with no indication of positional inaccuracy.

3.12 Meyssonnier et al. (1993)

Meyssonnier et al. (1993) (hereafter MLA93) produced a cata- logue of 1312 emission–line objects with identifiable natures in the Andromeda galaxy, based on slitless spectroscopy with a spectral range of 4350–5300 ˚A. We find 856 of these objects to be within 4OO of an object in our survey, the majority of which are classified as PN by both MLA93 and ourselves.

94 objects are listed by MLA93 as non-PNe, two-thirds of these we have flagged as HIIregions. Of the remaining 31 objects MLA93 have determined 25 to be possible Wolf-Rayet stars. Iden- tification as a Wolf-Rayet star is based on emission in the region of 4686 ˚A. It is believed that 6–10% of PNe have Wolf-Rayet central


Figure 15. Locations of extended objects and those below the ÷ cutoff.

The majority appear to be HIIregions in the star formation ring atÃûú

ß=Üü G and in the spiral arms. Those with non-M31 like velocities are likely to be emission from background galaxies.

Figure 16. Cumulative distribution of radius in the disk plane for bright (solid line) and faint (dotted line) PNe. A Kolmogorov-Smirnov test shows these to be statistically the same at the 95% confidence level. with

Figure 17. Comparison to Ciardullo et al. (1989) magnitudes

Figure 18. Comparison to Hurley-Keller et al. (2004) velocities stars (G´orny & Stasi´nska 1995), hence these can be safely left in the PN catalogue.

The last five objects are of uncertain classification in the MLA93 catalogue and will be left in out PN catalogue.

Only one object has a velocity dissimilar to that of M31, this object has been identified and flagged as a non-M31 source (MLA93 0953), see Fig. 20. MLA93 list this object as a possible Wolf-Rayet star or QSO. Given the peculiar velocity a background QSO seems likely.


In addition to the Andromeda galaxy a number of other galaxies lie within the PN.S survey area. These include M31’s two bright satellite galaxies, M32 and NGC 205; Andromeda VIII, a satellite galaxy proposed to explain a velocity substructure; and Andromeda IV, a faint background galaxy. There are also a number of objects with velocities unlike that of the M31 system. Mainly these are extended or selected as having a low [OIII]/HÑ ratio and are likely to be background galaxies. A couple of these have been identified in other catalogues.

The locations of known galaxies are shown in Figure 19.

Emission–line objects associated with these objects have ini- tially been selected based on position, then cuts made in velocity where appropriate.

4.1 M32

M32 is a dwarf elliptical galaxy (cE2) projected in line with the disk of M31. It has major and minor diameters ofA 8.7O and š…KjO (de Vaucouleurs et al. 1991); a system velocity of -200 kmsL  (Huchra et al. 1999);and a velocity dispersion ofKFEó¢ý;E kmsL , rising up toA ¡ E kmsL  at the centre (Simien & Prugniel 2002).

We have selected all objects within a 5.7Oradius of the centre of M32 as potentially belonging to the satellite (Table 4 and Fig.

20). A velocity cut has been made at -314 kmsL  to distinguish between objects belonging to Andromeda’s disk and M32. A fairly clean cut is inferred when the velocities of PN at slightly larger radii from M32 are considered, though the boundary between the two object is not certain.

4.2 NGC 205

NGC 205 (also knows as M 110) is a dwarf elliptical galaxy (E5) projectedA 35O from the centre of M31 along the minor axis. Its major and minor diameters areA 21.9O andA 11.0O respectively at a


(c) (d) (e)

Figure 20. (a) M32, (b) NGC 205, (c) And IV, (d) 2MASXi J0039374+420956, (e) MLA93 0953. The area around each minor galaxy is enlarged in the upper panels. Emission–line objects are shown as crosses and stars, the size of which represents an object’s velocity with negative and positive values respectively, w.r.t M31’s system velocity. The large circles indicate the approximate spatial extent of each galaxy (1.5 scale lengths) and PNe with circles around them have both positions and velocity consistent with that galaxy. The lower panels shows velocity histograms. The dotted line being the whole PN.S sample; the solid line an area including the minor galaxy, but extending well beyond its influence; the hatched histograms are the PNe within the area of galaxy, and the cross-hatched agree in velocity as well as position.


Table 4. Emission–line objects in the vicinity of M32

IDÀ RA Dec r Ö ‚ õ

ì \®ö¬

NoteÁ 1843 0:42:40.3 40:52:57.8 67 25.18 -235.3 2210 0:42:40.9 40:50:27.4 88 25.48 -311.5 2212 0:42:30.2 40:50:43.8 188 26.04 -225.6 2213 0:42:44.3 40:50:53.2 72 25.11 -141.6 E 2214 0:42:47.2 40:51:07.1 94 25.63 -142.5 2215 0:42:44.6 40:51:12.3 59 25.23 -163.0 E 2216 0:42:51.2 40:51:24.0 144 25.32 -133.8 2217 0:42:38.7 40:51:19.9 58 25.07 -172.4 2218 0:42:29.8 40:51:20.4 183 25.40 -448.3 ? 2219 0:42:40.2 40:51:25.8 37 24.94 -206.5 2220 0:42:36.5 40:51:28.6 84 24.74 -266.8 E 2221 0:42:43.8 40:51:41.6 33 24.97 -155.4 2222 0:42:39.2 40:52:01.0 39 26.51 -830.0 ? 2223 0:42:39.8 40:52:09.2 33 25.70 -166.1 2225 0:42:40.4 40:52:33.3 44 24.70 -164.8 2226 0:42:29.5 40:53:09.0 199 25.82 -559.4 ?, E 2228 0:42:47.2 40:54:36.7 181 25.65 -186.9 2229 0:42:26.9 40:54:55.0 288 25.11 -535.1 ? 2230 0:42:58.5 40:55:40.2 337 25.67 -196.0 2232 0:42:35.6 40:56:21.2 283 24.99 -443.0 ?, E 2234 0:42:42.3 40:51:49.5 9 20.13 -147.3 2235 0:42:44.6 40:51:44.1 43 24.81 -229.4 2891 0:42:44.7 40:50:33.0 92 25.63 -169.9 2894 0:42:41.7 40:51:31.6 23 22.30 -210.3 2895 0:42:42.2 40:51:39.8 16 20.78 -193.3 2896 0:42:44.2 40:51:44.2 37 24.68 -211.4 2897 0:42:45.7 40:51:49.7 59 23.22 -203.4 2898 0:42:44.4 40:52:33.4 54 23.78 -235.8 2900 0:42:44.6 40:54:04.0 136 24.66 -431.8 ? 2901 0:42:48.2 40:54:13.6 169 21.36 -199.2 2902 0:42:42.9 40:54:33.5 160 24.52 -433.2 ? 2903 0:43:00.1 40:55:00.4 331 24.17 -194.6 2904 0:42:46.7 40:55:04.8 204 23.47 -380.3 ? 2907 0:42:37.7 40:55:25.6 220 25.36 -285.8 2908 0:42:52.7 40:55:53.2 289 23.51 -150.8 E 2911 0:42:59.7 40:53:46.8 291 25.50 -365.4 ? 2929 0:42:40.6 40:47:33.8 261 22.89 -137.7 2930 0:42:36.8 40:48:04.7 242 24.38 -177.1 2931 0:42:53.0 40:49:00.0 242 21.50 -226.3 2932 0:42:46.3 40:49:02.2 185 24.42 -633.7 ?, E 2933 0:42:40.7 40:49:12.8 163 24.52 -155.8 2934 0:42:27.1 40:49:09.8 275 25.10 -399.6 ? 2935 0:42:41.0 40:49:40.2 135 25.09 -389.3 ?, E 2936 0:42:27.0 40:49:41.2 259 22.02 -458.8 ? 2937 0:42:23.5 40:50:49.2 282 25.41 -537.8 ? 2938 0:42:26.0 40:50:57.9 245 24.99 -469.5 ? 3172 0:42:32.1 40:52:26.6 150 23.64 -377.7 ? 3174 0:42:35.9 40:53:00.7 111 20.91 -207.7 3175 0:42:31.1 40:53:04.7 176 21.27 -414.5 ? 3181 0:42:38.7 40:51:35.5 51 24.32 -167.3 3182 0:42:41.0 40:51:34.6 23 24.01 -131.3 3237 0:42:39.3 40:52:10.7 40 23.41 -149.1 3238 0:42:54.0 40:49:54.2 219 24.01 -167.0 E 3239 0:42:57.0 40:51:01.8 234 23.34 -160.1 3241 0:42:56.0 40:51:12.6 217 22.21 -652.7 ?, E 3242 0:42:49.9 40:51:10.6 129 22.61 -505.3 ? 3243 0:42:40.0 40:49:41.3 136 22.03 -156.6 3245 0:43:02.0 40:49:30.5 336 20.89 -160.0 3267 0:42:48.4 40:50:09.4 144 25.04 -175.6 3273 0:42:23.1 40:50:23.4 295 22.74 -381.9 ?, E 3274 0:42:39.9 40:51:19.0 46 24.23 -180.0 3275 0:42:39.4 40:51:23.9 47 23.80 -132.3 3276 0:42:39.6 40:52:40.4 57 20.94 -250.8 3277 0:42:40.4 40:52:44.0 54 21.26 -196.9 E 3278 0:42:41.4 40:52:32.3 38 23.86 -265.1 E 3300 0:42:40.2 40:51:02.7 57 21.06 -158.9 E

À ID, RA, Dec,Ö ‚ andõ

ì \®ö¬

from PN.S data, see Table 3

Á ? – velocity implies object belongs to M31; E – extended; R – below flux ratio cutoff.

Figure 19. Positions of other galaxies and objects. An objects size and colour reflects its velocity w.r.t. M31’s system velocity; grey objects are moving away, black towards.

position angle of 170G (de Vaucouleurs et al. 1991). It has a system velocity ofz ‘F“ ô¢þI kmsL  (Bender et al. 1991); and a velocity dispersion of“ šç¢ ¡ kmsL , dipping toA ‘ E kmsL  at the very centre of the galaxy (Mateo 1998), (Carter & Sadler 1990), (Simien

& Prugniel 2002); and a rotation velocity ofWIç¢ ‘ kmsL  along the major axis (Geha et al. 2005).

The initial area selection is all objects within 12.3Oof NGC 205’s centre. Velocity cuts at -346 kmsL  and -146 kmsL eliminate objects that are clearly not members of NGC 205.

A number of the remaining objects may actually belong to M31 as the velocity distribution close to NGC 205 and at larger distances overlaps considerably.

4.3 Andromeda IV

Andromeda IV is a dwarf irregular galaxy, lying 40O from the centre of M31. Ferguson et al. (2000) have previously identified a number of emission–line objects associated with this galaxy, the PN.S sur- vey equivalents to these are listed in Table 6 along with the original IDs. Object F1 has not been found in our survey - it was originally detected in HÑ and has no [OIII] counterpart within our magni- tude range. The other seven objects are detected in our survey and velocities have been calculated. Objects F7 and F8 are located a little way from And IV and have velocities similar to the disk of M31 in that region, implying they are associated with M31, rather than And IV, and are excluded from further analysis of this object.

Two additional objects not listed in Ferguson et al. (2000) have been found in the vicinity of And IV with velocities similar to that object.

The seven And IV emission–line objects have a mean velocity

of‘F“ I5¢„WE kmsL  , in reasonable agreement with Ferguson et al.

(2000)’s value of‘ Kšó¢{Ò kmsL  measured from just three of these sources.




Related subjects :