A&A 403, 955-974 (2003)
DOI: 10.1051/0004-6361:20030416
F. Schuller1 - S. Ganesh2,1 - M. Messineo3 - A. Moneti1 - J. A. D. L. Blommaert4 - C. Alard1,5 - B. Aracil1 -
M.-A. Miville-Deschênes6 -
A. Omont1 - M. Schultheis1 - G. Simon5 -
A. Soive1 - L. Testi7
1 - Institut d'Astrophysique de Paris, CNRS, 98 bis Bd Arago,
75014 Paris, France
2 -
Physical Research Laboratory, Navarangpura,
Ahmedabad 380009, India
3 -
Leiden Observatory, University of Leiden, PO Box 9513,
2300 RA Leiden, The Netherlands
4 -
Instituut voor Sterrenkunde, KU Leuven, Celestijnenlaan 200 B,
3001 Leuven, Belgium
5 -
GEPI, Observatoire de Paris, 61 Av. de l'Observatoire,
75014 Paris, France
6 -
Laboratoire de radioastronomie millimétrique, École Normale
Supérieure & Observatoire de Paris, France
7 -
Osservatorio Astrofisico di Arcetri, Largo E. Fermi, 5,
50125 Firenze, Italy
Received 20 August 2002 / Accepted 13 March 2003
Abstract
We present version 1.0 of the ISOGAL-DENIS Point Source Catalogue
(PSC), containing more than 100 000 point sources detected at 7 and/or
15 m in the ISOGAL survey of the inner Galaxy with the ISOCAM instrument
on board the Infrared Space Observatory (ISO). These sources are
cross-identified, wherever possible, with near-infrared (0.8-2.2
m)
data from the DENIS survey. The overall surface covered by the ISOGAL
survey is about 16 square degrees, mostly (95
)
distributed near the
Galactic plane (
), where the source extraction
can become confusion limited and perturbed by the high background emission.
Therefore, special care has been taken aimed at limiting the photometric error
to
0.2 mag down to a sensitivity limit of typically 10 mJy.
The present paper gives a complete description of the entries and the
information which can be found in this catalogue, as well as a detailed
discussion of the data processing and the quality checks which have been
completed. The catalogue is available at the Centre
de Données Astronomiques de Strasbourg
(via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsweb.u-strasbg.fr/cgi-bin/qcat?J/A+A/403/955)
and also via the server at the Institut d'Astrophysique
de Paris (http://www-isogal.iap.fr/).
A more complete version of this paper, including a detailed
description of the data processing, is available in electronic
form through the ADS service and at http://www.edpsciences.org.
Key words: catalogs - stars: circumstellar matter - Galaxy: bulge - Galaxy: disk - Galaxy: stellar content - infrared: stars
The ISOGAL survey is the most sensitive mid-infrared wide-field survey dedicated to the inner Galaxy (see the accompanying paper Omont et al. 2003 and references therein for a review of its scientific goals and results). The large amount of ISO observations collected, in combination with the near-infrared data of the DENIS survey, has resulted in the production of a catalogue of 105 point sources, the PSC. The first scientific results obtained include studies of the Galactic structure, analysis of the stellar populations comprising completely detected AGB stars with their mass-loss in particular fields (Pérault et al. 1996; Omont et al. 1999; Glass et al. 1999; Ojha et al. 2003), characterisation of interstellar extinction (Jiang et al. 2003), of infrared dark clouds (Hennebelle et al. 2001), and of young stellar objects (Felli et al. 2000, 2002; Schuller 2002).
A total of 16 square degrees of the inner Galactic disk
(
)
were observed, with strong emphasis on
the inner Galactic bulge, at wavelengths of 7 and 15
m, with a pixel
scale of usually 6
and sometimes 3
,
down to a sensitivity limit of
typically 10 mJy. A total of
250 hours of ISO
time were used, making ISOGAL one of the largest programs performed by ISO.
For the southern sky the results were combined with the I, J,
(effective wavelengths equal to 0.79, 1.22 and 2.14
m) ground-based
data from the DENIS survey (Epchtein et al. 1994, 1997)
in order to produce
an (up to) 5-wavelength catalogue of point sources. Given the emphasis of
ISOGAL on the inner Galactic regions, the DENIS coverage is available
for 95% of the fields surveyed with ISOCAM.
As a comparison, the IRAS satellite, which made a breakthrough in the
infrared window in 1983,
performed an all sky survey resulting in a
point source
catalogue, with a typical sensitivity (or 90% completeness level) around
0.5 Jy in low source density regions and at the shortest wavelengths.
The four IRAS bands were centred at 12, 25, 60 and 100
m, thus covering
the mid- to far-infrared range, with a spatial resolution ranging from
less than 1
at 12
m to about 4
at 100
m.
The sensitivity of ISOCAM is about two orders of magnitude better than that
provided by the IRAS detectors at 12
m in the high source density
regions (thus in particular in the Galactic plane). Indeed, as explained in
the IRAS Explanatory Supplement (Sect. VIII), the typical 50% completeness
limit flux density was about 1 Jy at 12 and 25
m in the Galactic Plane,
and even brighter at longer wavelengths.
More recently, the MSX (Midcourse Space Experiment, see Mill et al.
1994 for an overview) mission surveyed the complete
Galactic Disk in the range
in the
mid-infrared, using a 33 cm aperture telescope called SPIRIT III
(Price et al. 2001). Six bands between 4 and 25
m were surveyed
simultaneously at a spatial resolution of
18
.
The most sensitive
band was the A band, centred at 8.3
m, for which
the present point source sensitivity limit is about 0.1 Jy. The survey
of the Galactic Plane has presently resulted in a catalogue of
sources (Price et al. 2001), which permits a
complete analysis of the most luminous infrared Galactic populations. The
images of this survey have also led to
the detection of more than 2000 infrared dark clouds (Egan et al.
1998). A very recent analysis (Lumsden et al. 2002)
of the MSX PSC has produced a large sample
of massive young stellar objects in the Galactic disk.
Among the many large observing programs conducted by ISO, including deep
and wide-field extragalactic surveys, worth mentioning are the European
Large-Area ISO Survey, ELAIS (Rowan-Robinson et al. 1999),
ISOCAM deep surveys using guaranteed time observations (Elbaz et al.
1999), and FIRBACK, a deep 170 m imaging survey carried out
with ISOPHOT (Dole et al. 2001). Apart from these there were
also a number of observations of specific targets in the Galaxy. The following
ISOCAM
studies were with sensitivities comparable to or slightly deeper than ISOGAL (in
more limited areas): LW2 and LW3 imaging surveys of nearby star forming regions
(Nordh et al. 1998; Bontemps et al. 2001),
photometric studies of other Galactic HII
regions (Zavagno & Ducci 2001), and
the GPSURVEY (Burgdorf et al. 2000), which provided observations
of about 0.25 deg2 in the central Galaxy at mid-infrared wavelengths.
In this paper, we give a detailed description of the ISOGAL observations in Sect. 2, and of their processing and the related quality checks in Sect. 3. The DENIS data are presented in Sect. 4. The content of the Point Source Catalogue (PSC) is explained in Sect. 5, and the complete descriptions of various support tables are given in the relevant sections. Finally, the main characteristics of the catalogue are briefly summarised in Sect. 8.
The mid-infrared observations were obtained with the ISOCAM instrument
(Cesarsky et al. 1996; Blommaert et al. 2001)
on ISO (Kessler et al. 1996) using
filters centred at
and 15
m and with a pixel scale of 6
,
or 3
in a few cases. Table 1 lists the filters used.
Most observations were performed with the broad filters LW2 and LW3, with a
field selection avoiding bright IRAS sources susceptible to detector array
saturation. However, a few regions with stronger sources
(around the Galactic Centre and in a few star forming regions) were
observed with the narrow filters LW5 or LW6, and LW9, and with smaller
pixel field of view (3
).
Filter |
![]() |
![]() |
ZPa | Fmag=0 | Area |
[
![]() |
[![]() |
[mag] | [Jy] | [deg2] | |
LW2 | 6.7 | 3.5 | 12.39 | 90.36 | 9.17 |
LW5 | 6.8 | 0.5 | 12.28 | 81.66 | 0.64 |
LW6 | 7.7 | 1.5 | 12.02 | 64.27 | 2.97 |
LW3 | 14.3 | 6.0 | 10.74 | 19.77 | 9.92 |
LW9 | 14.9 | 2.0 | 10.62 | 17.70 | 3.53 |
For standard ISOGAL observations (broad filters LW2 and LW3), we estimated
that, to avoid saturation of the detector, no IRAS source with
6 Jy should be observed. This limit was further relaxed
up to
Jy with narrow filters; however, even with such
a high limit value, it implied that a few regions, including the Galactic
Centre itself, could not be observed. A quick inspection of the images showed
that only very few observed pixels among all ISOGAL observations
were slightly above the limit of the linear
domain of the detector. The profiles of the associated
point sources do not deviate much from the average point spread function
(PSF, see Sect. 3.2.1), so that no source suffers
strongly from saturation in the published point source catalogue.
The observations were performed as rasters. The basic ISOCAM observation
is a
pixel image of 0.28 s integration time. Due to
limitations in the downlink data rate, these basic images were coadded in
groups of four and downlinked, making the unit frame one of 1.12 s
integration time. At each raster position 19 such frames were obtained,
resulting in an integration time of
21 s per raster position.
The rasters were oriented along galactic latitude and longitude, which
differed from the direction of the sides of the detector array, resulting in
"saw-tooth'' edges of the final mosaics. With 6
pixels, the raster steps
were typically 90
in one direction and 150
in the perpendicular one
(and a factor of two smaller with 3
pixels), in order to observe
each sky position about twice. However, because of the non-alignment of the raster
and detector axes, each sky position was not as regularly observed.
The actual number of observations per sky point varied from four to
exceptionally zero (for the dead ISOCAM column close to a raster
edge), with an average of
1.5.
The total area covered by the ISOGAL survey is 15.6 square degrees,
of which 10.7 were observed at both 7 and 15
m, 2.1 were observed at
7
m only, and 2.8 were observed at 15
m only. This survey is
the result of three successive proposals developed over the lifetime of
ISO.
As a consequence, most fields were observed at
7 and 15
m at different dates, and some fields were
observed at one wavelength only, in particular because the planned
targets were not observable at the very end of the mission.
A total of 696 observations compose the ISOGAL survey. Of all these
observations, 29 could not be used because of instrument failures or other
problems during the data reduction. Another 18 observations are single
ISOCAM frames (
pixels) observed in the spectroscopic
Circular Variable Filter (CVF) mode; they are treated in a
different way (Blommaert et al., in preparation). A further 186 images are
"dummy'' observations, containing only one
pixel image
- acquired after repositioning of the telescope to allow for
reconfiguring the camera from the CAM parallel mode to that of
the observation - and have not been used for the catalogue. As a result, only
463 raster-observations are considered as relevant for the imaging survey.
Col. | Name | Format | Units [range] | Description |
1 | ION | a8 | ISO Observation number | |
2 | name | a13 | ISOGAL observation name | |
3 | date | a6 | YYMMDD | date of observation |
4 | j_day | i4 | Julian day of observation - 2 450 000 | |
5 | qual | i1 | [1,2] | quality of imagea |
6 | l_off | f5.1 | arcsec | applied offset in Galactic longitudeb |
7 | b_off | f5.1 | arcsec | applied offset in Galactic latitude |
8 | G_lon | f8.4 | deg [-180-+180] | Galactic longitude of raster centre |
9 | G_lat | f8.4 | deg [-90-+90] | Galactic latitude of raster centre |
10 | dl | f6.4 | deg | half width of raster in longitude |
11 | db | f6.4 | deg | half width of raster in latitude |
12 | RA | f8.4 | deg | RA (J2000) of raster centre |
13 | DEC | f8.4 | deg | Dec (J2000) of raster centre |
14 | filt | i1 | [2,3,5,6,9] | LW filter number |
15 | pfov | i1 | arcsec [3,6] | pixel field of view |
16 | mag_lim | f5.2 | mag | ISO magnitude cutoffc |
17 | nb_sour | i4 | number of extracted sources brighter than mag_lim | |
18 | rot | i1 | [0,1] | applied transformation (270![]() |
19 | x_inv | i1 | [0,1] | applied transformation (x-inversion) to the raster |
20 | y_inv | i1 | [0,1] | applied transformation (y-inversion) to the raster |
21 | m | i2 | number of raster steps in x in final raster | |
22 | n | i2 | number of raster steps in y in final raster | |
23 | dm | i3 | arcsec | size of step between x (final) raster positions |
24 | dn | i3 | arcsec | size of step between y (final) raster positions |
25 | angle | f6.2 | deg | angle from the upward axis to the north in the final raster |
26 | NX | i3 | pixel | number of pixels in x of final raster |
27 | NY | i3 | pixel | number of pixels in y of final raster |
![]() |
Figure 1: Example of one ISOGAL observation which has been used for one FA and one FC fields. The formal limits of both fields are shown with rectangular frames: FC field (upper frame) and FA field (lower frame). The different symbols correspond to the different catalogues of sources (see Sect. 5): squares (FC, regular), crosses (FC, edge), diamonds (FA, regular) and plus signs (FA, edge). |
To avoid redundancy in the published catalogue (due e.g. to various
observations of a test field with several filters, but also to small
overlapping areas between two observations in many cases), we decided to
use, for the present version of the PSC, only one observation at 7 m
and one at 15
m for a given position
.
Thus, we had to choose the best observation in the case of
overlapping images at the same wavelength. The selection criteria were:
first, if the different observations are obviously of different quality,
the best quality one was selected. Then, if the observations were made
with different filters, we chose to keep the one with a broad filter (if
it exists) because the number of detected sources is larger. In the very
few cases where the filter
is the same but the pixel size is different, we selected the large (6
)
pixel observations in order to have more homogeneous data. If the quality
and the observational setup were approximately the same in different
observations, we then selected the most recent one (the one with higher
ISO observation number), because on average the data quality was better
certified. Finally, 384 raster images have been used to build the PSC.
All the raster images used are published with the PSC (and available through
the CDS and IAP web sites), and the electronic
version of the catalogue of ISOGAL observations of the PSC contains
384 entries, each entry having the format described in Table 2.
Two examples are shown in Table 3, for the 7 and 15
m
observations composing a test field of 0.027 deg2 centred at
(l,b) = (0.0,1.0), hereafter called the "C32'' field.
Col. | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |
Name | ION | Name | date | j_day | qual | l_off | b_off | G_lon | G_lat | dl | db | RA |
Ex. 1 | 83600418 | 2P00P10B | 980228 | 873 | 1 | -4.8 | -5.6 | 0.0001 | 0.9988 | 0.1633 | 0.0758 | 265.4353 |
Ex. 2 | 83600523 | 3P00P10B | 980228 | 873 | 1 | -6.3 | -3.1 | -0.0003 | 0.9995 | 0.1633 | 0.0758 | 265.4353 |
Col. | 13 | 14 | 15 | 16 | 17 | 18 | 19 | 20 | 21 | 22 | 23 | 24 | 25 | 26 | 27 |
name | DEC | filt | pfov | mag_lim | nb_sour | rot | x_inv | y_inv | m | n | dm | dn | angle | NX | NY |
Ex. 1 | -28.4136 | 2 | 6 | 8.89 | 331 | 1 | 0 | 0 | 7 | 4 | 150 | 90 | 58.95 | 196 | 91 |
Ex. 2 | -28.4136 | 3 | 6 | 8.00 | 220 | 1 | 0 | 0 | 7 | 4 | 150 | 90 | 58.97 | 196 | 91 |
We define an ISOGAL "field'' as a rectangular area of the sky whose edges
are aligned with the galactic axes, and which has been completely observed
with ISOCAM. There are three kinds of fields, depending on the available
observations: the "FA'' fields were observed only at 7 m, the
"FB'' fields were observed only at 15
m, and the "FC'' fields
were observed at both 7
m and 15
m.
Col. | Name | Format | Units [range] | Description |
1 | Name | a14 | ISOGAL field identifier | |
2 | ION7 | a8 | ION for 7 ![]() |
|
3 | ION15 | a8 | ION for 15 ![]() |
|
4 | filt7 | i1 | [2,5,6] | 7 ![]() |
5 | filt15 | i1 | [3,9] | 15 ![]() |
6 | pfov | i1 | arcsec [3,6] | pixel field of view |
7 | G_lon | f8.4 | deg [-180-+180] | Galactic longitude of field centre |
8 | G_lat | f8.4 | deg [-90-+90] | Galactic latitude of field centre |
9 | dl | f6.4 | deg | half width of field in longitudea |
10 | db | f6.4 | deg | half width of field in latitudea |
11 | area | f6.4 | deg2 | area of field |
12 | dens7 | i5 | deg-2 | density of 7 ![]() |
13 | dens15 | i5 | deg-2 | density of 15 ![]() |
14 | RMS_II | f4.2 | arcsec | RMS separation of 7-15 ![]() |
15 | RMS_ID | f4.2 | arcsec | RMS separation of ISO-DENIS associated sources |
16 | K_max1 | f4.1 | mag | DENIS ![]() |
17 | K_max2 | f4.1 | mag | DENIS ![]() |
18 | dens_K2 | i5 | deg-2 | density of DENIS ![]() |
To build the present version of the PSC, we have defined a total of 43 FA fields, 57 FB fields and 163 FC fields.
In some cases, a fraction of an ISOGAL observation was
used for an FA (or FB) field, and another fraction was used for an FC field
(see e.g. Fig. 1),
so that only 384 different observations were required for these 263 fields.
These peculiar configurations can result in the presence of a few redundant
sources: because of edge effects, two sources at the same position may
appear in two different catalogues; nine such cases can be seen in
Fig. 1 (see also Sect. 5.1).
The complete catalogue of the 263 ISOGAL fields is available
electronically and contains 18 columns, as described in Table 4, and an example is given
in Table 5.
The field names are generated using 14 characters, and the first two indicate the type of the field (FA, FB or FC).
The 12 last characters of the field names are the
galactic coordinates in decimal degrees of the centre of the field.
A graphical view of the observed fields is given in Fig. 2.
Col. | Name | C32 field |
1 | Name | FC+00000+00100 |
2 | ION7 | 83600418 |
3 | ION15 | 83600523 |
4 | filt7 | 2 |
5 | filt15 | 3 |
6 | pfov | 6 |
7 | G_lon | -0.0011 |
8 | G_lat | 0.9990 |
9 | dl | 0.1441 |
10 | db | 0.0471 |
11 | area | 0.0271 |
12 | dens7 | 9225 |
13 | dens15 | 6125 |
14 | RMS_II | 2.24 |
15 | RMS_ID | 1.70 |
16 | K_max1 | 9.6 |
17 | K_max2 | 10.6 |
18 | dens_K2 | 35979 |
A complete description of the data processing and of the procedures that were
run to quantify the quality of the data is given in the electronic version of
this paper, available through the ADS service.
Only the main results,
which may be useful to all users of the catalogue, are summarised in this
section and the following one.
Data reduction was performed with standard procedures of the CAM Interactive Analysis (CIA, Ott et al. 1997) package version 3.0 on data products produced with version 7.0 of the ISO Off-Line Processing (OLP) pipeline (Blommaert et al. 2001). A particular care was taken to correct the effect of the slow detector responsitivity. Two methods of transient correction were used: the IAS model transient correction (Abergel et al. 1998), also called the "inversion'' method, was used to get the best photometry for non-stabilised signals, while the auxiliary "vision'' method (Starck 1998; Starck et al. 1998) was used to remove most of the latent images (or remnants) due to memory effects of the detectors to strong sources (Coulais & Abergel 2000). We thus have two sets of reduced data: a) the main one treated with "inversion'', which performs a correction for the missing signal, (though this correction is not perfect, see Sect. 3.3), but which still contains the remnants; b) an auxiliary one, roughly treated with "vision'', where most remnants have been removed, but with wrong photometry. The two rasters are converted to physical units (mJy), using the standard conversion factors (Blommaert 1998).
A dedicated PSF fitting procedure worked out by C. Alard has been used to extract point sources from all "inversion'' and "vision'' processed images. First, a search for local maxima is performed on the complete image, resulting in a list of pixel positions of point source candidates. Then, an analytical expression of the PSF is fitted at each position to compute the flux density of the point sources, and to discard the local maxima whose shapes do not correspond to the instrumental response to a point source.
The source detection is first performed on an oversampled image, using pixels a factor of two smaller than in the initial image. This image is used only for the detection step of the source extraction. The oversampling is performed by a convolution of the initial pixels with an analytical expression of a theoretical PSF. As a result, the sources can be localised on a thinner grid.
The detection procedure looks for local maxima in the oversampled image.
This step is controlled by a mesh parameter, which can take
values of 1 or 2, and defines the size of the grid on which local maxima are
looked for. With mesh=1, all local maxima are detected, even those
corresponding to bright spots in the background rather than to point sources.
On the other hand, with mesh=2,
oversampled pixel meshes are
used to find local maxima, resulting in a smoothing of the irregularities in
the background, without any significant loss in the detection of relatively
bright (
100 mJy) point sources, but with a more confusion
limited extraction of the faintest sources.
The extraction procedure which has been used to build the ISOGAL PSC performed a complete extraction with each value of mesh. For each observation, the two resulting catalogues have been cross associated to check the quality and the reality of the detected sources (see Sect. 3.2.2). Obviously the extraction performed with mesh=1 is the most efficient to correctly extract blended sources; on the other hand, a non negligible fraction of the sources extracted only with mesh=1(with no association in the extraction performed with mesh=2) seem to be spurious (see the discussion in Sect. 3.5.2).
Another procedure is used to measure the flux density of the sources
on the original image, and to estimate the correlation of their profile
with the PSF. For each observational setup (combination of one filter
and one pixel scale), a single reference PSF has been determined for all
the observations from a sample of relatively bright and isolated sources.
A least square fit between the reference profile and a
(not
oversampled) pixel mesh is computed at the position of each source candidate,
starting with the brightest one. The background
is estimated from the median value of the pixels in an
annulus of inner and outer radii equal to 3 and 5 pixels, respectively.
The results of this operation are the flux density of the source and the
uncertainty on its measurement, computed as the RMS of the residual between the
scaled PSF profile and the actual source profile. This flux density uncertainty
is later converted to a magnitude uncertainty, hereafter called
.
The reality of each
point source is estimated by the ratio of the fitted flux density to the
RMS uncertainty of the fit, and only sources with this ratio greater than 3
are considered valid and stored in the resulting catalogue. Then, the
profile of the source is subtracted from the image, and the procedure
runs iteratively going to fainter and fainter sources. This method is
powerful even in crowded fields, where it is able to estimate correctly
the flux densities of blended sources.
Four catalogues have been built for each observation, combining the
two possible values of mesh (1 or 2) and the "inversion'' and "vision''
processed rasters. Considering the high background level in the Galactic
Disk, we decided to anyhow limit the published catalogue to a flux density
of 5 mJy (
and
)
to reduce the number of
spurious sources (another limit was eventually later applied depending
on the field, see Sect. 3.4).
Then, the sources found in "inversion'' processed images that were associated
with a "vision'' source within a search radius of one observed pixel
were considered valid, while those found
only in the "inversion'' images were considered spurious (these can be
remnants of bright sources, or other non real point-like sources).
The distance between the "inversion'' and the "vision'' sources gives a good
estimate of the quality of the sources: it is generally smaller than 1
for
real sources, while a separation larger than 3
may be due to artifacts
(see also Sect. 3.2.5).
The final data (position and photometry) in the catalogue come only from
the "inversion'' processed rasters, with elimination of the remnant sources
using the "vision'' results.
The majority (70%) of the extracted sources could be associated between the
mesh=1 and the mesh=2 catalogues (with a 6
association radius for all
observations), while the remaining 30% are only found with mesh=1.
Since less than 1% of the
extracted sources were detected with mesh=2 with no counterpart in the
mesh=1 catalogue, while almost 30% of the extracted sources were only
detected with mesh=1, the published data (position and photometry) come
from the mesh=1 results for the sources which were detected with both
values, in order to get a homogeneous set of data. Further quality selection
criteria are applied later in the processing (see Sect. 3.2.5),
so that only
10% of the sources in the published catalogue
have been detected with mesh=1 only.
A special MESH flag is included in the catalogue to indicate for which
value(s) of mesh a source has been extracted, and the global QUALITY
flag is decreased for sources without association between the
mesh=1 and the mesh=2 results (see next section).
The quality of the derived photometry as well as the reliability of the extracted sources can be affected by several factors, and different quality flags have been computed to warn the user when effects degrading the photometric quality are present, and to finally estimate the global quality of the point sources.
A global quality flag Q was computed by combining the flags defined above
and the photometric uncertainty .
Its value ranges from 1 to 3 for
sources with MESH = 1 or 2, and from 2 to 4 for sources with
MESH=3, the higher the better quality.
The distribution of this flag for all the sources in the
catalogue is shown for the different filters in Fig. 3. As can be
seen, more than one half of the sources have a very good photometric quality
(Q = 4). A value of 3 for this flag can also be considered as reasonably good
quality. Finally, only
15% of the sources in the catalogue have a
moderate photometric quality (
). They should be used with much
caution since their reliability is not warranted.
![]() |
Figure 3: Distribution of the quality flag Q for the different filters. The gray scale corresponds to the different values of this flag, from 4 (lightest grey) to 1 (darkest grey). |
Additional estimates of the reliability of the sources are provided by
the analysis of repeated or overlapping observations (see
Sect. 3.5.1), but also by the combination of several wavelengths,
including DENIS ones:
a source with a moderate quality flag at, for example, 7 m, but with
a good quality association at 15
m (see Sect. 3.6)
finally has a very large probability to be a real source.
The first version of the ISOGAL PSC only contains point sources, and sources of very small extension. The extraction of extended objects will be performed with a dedicated procedure for the second version of the catalogue.
The present version of the PSC contains a small proportion of
sources of small extension, with typical sizes
around 10-20
(FWHM). These slightly-extended sources are characterised
by relatively high values of the photometric uncertainty, with typical
0.15 mag for bright (
1 Jy) sources,
while bright point sources generally have
mag.
Aperture photometry performed on a small sample of such bright slightly
extended sources has shown that their magnitudes can be underestimated
by about 1 mag (Schuller 2002). It is planned to perform
accurate photometry and to
include a relevant extension flag in the second version of the PSC.
The flux densities of the point sources, as obtained by the PSF fitting procedure, lead to a good relative photometry, but have to be calibrated in an absolute way. Two factors introduce biases in the photometry. First, the integration time for the standard ISOGAL observations was too short to allow the signal to stabilise. A correction to this transient problem is applied with the "inversion'' method (see Sect. 3.1). However, this method only allows proper correction for extended emission, but is insufficient for point sources (Coulais & Abergel 2000; Blommaert 1998). A few ISOGAL fields were observed with longer integration times. A comparison between regular and long measurements showed that the photometry from the regular raster is about 0.2 mag too high (too faint).
Our PSF photometry introduces a second bias, because a fraction of the signal from a point source falls outside the mesh used to model the PSF. The general flux calibration of ISOCAM was established from measurements on standard stars (Blommaert 1998; Blommaert et al. 2000). The observed signal was measured using aperture photometry, which was corrected for the part of the PSF falling outside the aperture. To convert our PSF-fitting photometry to absolute photometry, a comparison was made with photometry obtained using the same techniques as in the ISOCAM general flux calibration. The aperture magnitudes were found to be lower (brighter) than the PSF magnitudes by 0.2-0.4 mag, revealing a bias in the PSF normalisation.
The total correction that has to be applied is between -0.37 and -0.59 mag for the different setups. As the uncertainty on each determined correction is at least 0.1 mag we decided to apply the same constant offset of -0.45 mag to all the sources and for all observational setups. This correction leads to photometry in good agreement with external comparison data, as is explained below.
The first publications based on ISOGAL data made use of a non-corrected
photometry. The mid-infrared magnitudes presented there should thus
be corrected by a -0.45 mag offset (with a possible 0.1
mag additional discrepancy from field to field). This concerns in particular
the results published in Pérault et al. (1996), Testi et al.
(1997), Omont et al. (1999), Glass et al. (1999),
Schultheis et al. (2000) and Felli et al. (2000).
Appropriate errata will be published for these papers.
The comparison of the observed with the predicted photometry for stars
with known spectral types and distances provides an absolute calibration.
Comparing the predicted and the corrected PSF magnitudes for three stars
from the Hipparcos Input Catalogue we obtain:
A second check on the photometry is provided by the cross calibration with
the published catalogue of bright sources detected by the MSX
survey of the Galactic Plane (Price et al. 2001).
A comparison with the band D photometry of MSX, which
used a filter similar to the ISOCAM 15 m filters, showed good agreement
between the corrected ISO magnitudes and the MSX ones. For 650 stars
(424 observed with LW3 and 226 with LW9) we find:
Artificial star experiments (see Bellazzini et al. 2002 and references therein for a general description) were conducted on the ISOGAL images in order to study the effects of a crowded field on the photometric quality and the completeness of the extracted point source catalogue. A procedure was created for adding artificial stars to the ISOGAL images, for extracting the sources with the same pipeline as the one used to generate the ISOGAL catalogue, and for checking how well the input sources are extracted.
Artificial star experiments enabled us to evaluate both random and systematic photometric errors due to crowding, as well as the completeness level of the extraction. The output magnitudes were found brighter than the input ones. This bias is very small for bright stars, but can reach 0.3 mag for the faintest ones in the densest fields, where the probability of blending with real stars is higher (see e.g. Fig. E-10 in the electronic version of this paper).
The completeness of the extraction can be quantified as follows.
For each observation, we can plot the
fraction of simulated sources which were retrieved as a function of
input magnitude. We observe a smooth curve which drops for the
faintest magnitudes. The magnitude where this fraction becomes
less than 50% strongly depends on the density of the field.
We used this trend to define the limiting magnitudes for each observation,
corresponding to the faintest sources that were included in the published
catalogue.
These magnitudes are generally consistent with the magnitudes above
which the bias reaches 0.1 mag and its standard deviation reaches 0.3 mag. We derived relations between source density and limiting magnitudes
for the different observational setups (see also Figs. E-13 and E-14 in the
electronic version). We make a distinction between the
core of the ISOGAL survey observed with broad filters and 6
pixels
and the peculiar observations of difficult fields observed with
narrow filters and 6
or 3
pixels.
For the 6
pixel observations with LW2 and LW3 filters,
we computed the following linear relations:
The results of our artificial source simulations show that the completeness
level is generally worse in LW5, LW6 and LW9 observations, which can
be interpreted as an effect of the much brighter diffuse background in the
peculiar regions which needed the use of such narrow filters. Therefore,
we applied 0.5 mag brighter cutting criteria for the 6
observations
with these filters:
The situation is more complicated for the 3
pixel observations, because
they are too few and peculiar to allow a global statistical treatment.
Artificial source simulations have been run on all the 3''
pixel observations used in the PSC,
and the
results show good agreement between the different observations with a given
filter. Therefore we used a single limiting magnitude for each filter, and the
different values are given in Table 6.
These limits give reasonably good results in terms of bias and completeness.
Filter | LW2 | LW5 | LW3 | LW9 |
![]() |
10.0 | 8.4 | 8.5 | 7.0 |
The distribution of the limiting magnitudes, as defined in the previous section
(Eqs. (1)-(4) for 6
pixel observations, Table 6
for 3
pixel observations) for all ISOGAL observations is shown in
Fig. 4. Since most observations were done with the
broad LW2 and LW3 filters, these histograms show that the typical reached
sensitivity is around 20 mJy at 7
m and 12 mJy at 15
m.
![]() |
Figure 4:
Distribution of the magnitudes mag
![]() ![]() |
When we apply these relations to all the ISOGAL catalogues, we eliminate
25% of the sources. This photometric cut is far more severe for
moderate quality sources
than for good quality ones: if we consider the QUALITY flag as defined
in Sect. 3.2.3, it appears that about one half of the sources
with
QUALITY =1 or 2 are discarded, while
30% of the sources
with QUALITY=3 and
12% of the sources with
QUALITY = 4 are
removed by this cut.
A few ISOGAL fields have been observed twice or more with exactly the same
observational setup (filter and pixel size), and a large
number of fields have overlapping regions. The total surface of such
repeatedly observed areas is 0.7 deg2.
A comparison of the photometry extracted from such independent observations
of the same regions of the sky was performed, and the main results for each
observational setup are given in Table 7. Note that,
because of the variability of some sources,
the quoted standard deviations in Table 7
are slightly above the true photometric uncertainty of the final
catalogue multiplied by
.
It is also possible to get information about the completeness from the
fraction of sources detected in both overlapping observations.
It is however difficult to accurately estimate the completeness
level by this method, as neither of the two catalogues is complete.
It is nevertheless possible to have a rough estimate by comparing the catalogue
extracted from a 6
pixel observation with the more complete one, extracted
from a 3
pixel observation of the same region. Then we can compute the
magnitude above which the fraction of 3
sources found in the 6
catalogue
is below 50%. Taking into account all the limitations inherent to this
method, the final results are essentially consistent with those derived from
the artificial sources simulations, and also confirm that more care
has to be taken for the observations performed with narrow filters.
Filter | Overlap | Nb. of |
![]() ![]() |
RMS |
surface (deg2) | sources | |||
LW2 | 0.166 | 2793 | 0.008 | 0.21 |
LW6 | 0.098 | 1974 | 0.005 | 0.22 |
LW3 | 0.275 | 2244 | 0.009 | 0.23 |
LW9 | 0.111 | 1250 | 0.007 | 0.28 |
Total | 0.650 | 8261 | 0.003 | 0.23 |
An additional check of the reality of the sources can be performed as
follows. The sources extracted from 6
pixel observations should also
be found in a 3
pixel observation of the same region, because the
sensitivity is generally greater in the latter, since the source extraction
is much less limited by confusion. Also sources detected
at one wavelength and with a good quality association at another
ISO or DENIS wavelength have a very large probability to be real.
But sources found only in a 6
pixel observation, with counterparts
neither in the overlapping 3
pixel observation nor at other wavelengths
(or with a bad quality association) may be spurious.
From the available set of overlapping 3
and 6
observations,
we have determined that the overall
fraction of such doubtful sources is very small (
7%),
with a large difference between the 7
m (
4%) and
the 15
m (
11%) sources. This fraction also strongly
depends on the quality of the sources, and ranges from less than 1% (at both wavelengths) for sources with quality flags Q=4,
to
15% (resp.
30%) for sources with Q=1 or 2 or with MESH=1 or 2 at 7
m (resp. at 15
m).
Therefore sources with quality flags less than 3 should be
considered with extreme caution, especially at 15
m.
The initial astrometric accuracy of the ISOCAM data is limited by the errors
in the pointing of the telescope and in the positioning of the lens wheels.
The global astrometric uncertainty can reach 10
(Blommaert et al.
2001, see also Ott 2002), and the
offset between two independent observations can reach twice this value.
Therefore an offset correction between the 7
m and the 15
m
observations was needed before the two catalogues could be cross identified.
The found offsets are typically of order a few arcseconds, but can
reach 15
,
in agreement with expectations.
In addition, there can be a small error in the positioning of the individual
images within the final raster, due to a combination of possible long term
drifts and the lens wheel jitter. Only very small amplitude "distortion''
effects have been observed, but a low order polynomial correction was
systematically applied to the 15 m coordinates to best match the
7
m ones.
After the 15 m coordinates were corrected to match those at
7
m, an association between 7
m and 15
m sources
was performed with a search radius equal to two pixels. This rather large
radius was chosen in order not to miss 7-15
m associations for slightly
extended sources, and because the density of 15
m sources is low enough
to limit the probability of chance associations to a few percent in most
cases. Only associations with the smallest separation are retained. The
mean values of the 7-15
m separations are typically in the range 1-3
in all ISOGAL FC fields, with standard deviations in the same range, as shown
in Fig. 5. At the end of this step, the catalogued source
coordinates are the most accurate available, namely the 7
m coordinates
for the sources detected at 7
m, or the 15
m coordinates translated
to the 7
m referential for the 15
m sources with no 7
m
association in the FC fields. We kept the initial 15
m coordinates
only for the sources in FB fields without 7
m observations.
Finally, a 7-15 m association quality flag is computed for each associated
source. The value of this flag is defined as follows:
In addition to these mid-infrared wavelengths, all the observations in
the southern hemisphere (almost 95% of the total area) have been
systematically cross-identified with the DENIS (Epchtein et al. 1994,
1997)
data, which provide measurements in the three near infrared bands I, Jand .
In coordination with the ISOGAL project, dedicated observations with the DENIS
instrument on the ESO 1 meter telescope at La Silla have been performed, along
the inner Galactic plane, between -30 and +10 degrees in galactic longitude, -2and +2 degrees in latitude, (4 degrees in the inner Bulge) using a
specific technique (Simon et al. in preparation). The individual images
(
)
were taken in a raster mode, covering typically 3 square degrees.
Between +10 and +30 degrees in longitude, regular
30
DENIS strips (see Epchtein et al. 1994) were used, with a
special reduction procedure.
All the DENIS images which have been used to build the ISOGAL PSC are
described in the Table of DENIS Observations, whose format is given in
Table 8.
Col. | Name | Format | Units [range] | Description |
1 | Name | a7 | image number | |
2 | date | a6 | YYMMDD | date of observation |
3 | j_day | i4 | Julian day of observation - 2 450 000 | |
4 | RA | f8.4 | deg | RA (J2000) of image centre |
5 | Dec | f8.4 | deg | Dec (J2000) of image centre |
6 | G_lon | f7.3 | deg [-180-+180] | Galactic longitude of image centre |
7 | G_lat | f7.3 | deg [-90-+90] | Galactic latitude of image centre |
8 | ![]() |
i1 | quality flag of I image | |
9 | ![]() |
i1 | quality flag of J image | |
10 | ![]() |
i1 | quality flag of ![]() |
The source extraction has been made through PSF fitting, using the same
extraction code as for ISOCAM images. The PSF is modelled in 9 squares
on each
individual image and adjusted with
respect to the source position. The derived correlation factor gives an
evaluation of the photometric uncertainty of the source extraction. For each
band, we preserve only the sources with a correlation factor greater than 0.6.
The correlation factors are given for each DENIS source in the ISOGAL PSC
(Sect. 5).
The saturation of DENIS detectors occurs around magnitude 10 in I, 7.5 in J and 6 in ,
and results in severely underestimated flux densities.
Therefore, the brightest DENIS sources have been removed from the catalogue.
The absolute photometry results from the zero point derived from standard stars
observed
through the night. A mean value is applied. These magnitudes can be converted
to flux densities using the zero points given in Table 9
(from Fouqué et al. 2000).
The limiting sensitivity is about 0.05 mJy (mag. 19) in I, 0.5 mJy (mag. 16)
in J and 2.5 mJy (mag. 13.5) in
but the extraction can become
confusion limited in the dense Galactic environment. The relative accuracy of
the photometry is checked through the comparison of the measurements in the
overlaps (2
between adjacent images).
The average differences are better than 0.03 mag down to magnitudes 17 in I(standard deviation <0.1 mag), 14 in J and 12 in
(standard deviation
<0.2 mag), which remains very good given the difficulty inherent to such
dense regions.
Finally, an image quality flag has been evaluated from the overlapping regions of each DENIS frame covering the ISOGAL rasters. In each band the standard deviation of magnitude differences over a defined magnitude range is calculated (see Table 10) and we assigned a quality flag ranging from 0 (very bad) to 3 (very good). More than one half of the used images have this flag equal to 3, and less than 20% have a flag equal to 1 or 0.
The astrometry is calculated for each image from the present
association between I and
the USNO_A2 catalogue. Then, the cross associations of J data over I, and
of
data over J are relatively straightforward since all three images
have been observed simultaneously. The resulting relative accuracy is better
than 0.2
(RMS) in I and 0.4
in J and
.
The derived position for I is kept
for I/J/
associations, and the J position is given for the
J/
associated sources. From a comparison made with the TYCHO catalogue in the SgrI
field in the Baade's Window, no systematic offset was found. The mean value of
the distances was 0.36
,
with a 0.19
standard deviation (Simon et al.,
in preparation).
Altogether the present accuracy of the DENIS coordinates used is thus
better than 0.5
.
It will be improved in the future since it is greatly
limited by the accuracy of the astrometry of the USNO_A2 catalogue.
The general method that we used to associate DENIS sources with ISOGAL
sources is similar to the procedure we used to associate 7 m and 15
m
data. The only difference arises from the very high density of DENIS sources,
so that we used a much smaller association radius, and we cut out the faintest
DENIS sources when the source density was too high, in order to reduce the
probability of chance associations.
As explained in Sect. 4.2, the absolute accuracy of the
DENIS coordinates is better than 0.5
,
thus much better than the
ISO astrometry. Therefore we took the DENIS coordinates as the
reference system, and computed the global translation offset between the
ISOGAL and the DENIS catalogues with the same procedure as for
the 7-15
m associations. The resulting offsets are typically in the
range 3-9
,
and can be explained
by the lens wheel jitter of ISOCAM (Sect. 3.6.1). This
also implies that the coordinates of ISOGAL sources outside the region with
DENIS observations can be wrong by this range of distances.
An approximate polynomial distortion correction was computed with the same
procedure as for the 7-15
m associations, in order to match as best as
possible the previous ISO reference coordinates with the DENIS ones. Again,
the observed effects were of very small amplitude, but this correction was
required to correct for small rotations in the ISOCAM rasters.
Band |
![]() ![]() |
![]() |
I | 0.791 | 2499 |
J | 1.228 | 1595 |
![]() |
2.145 | 665 |
Mag. range Sigma range | |||||
Flag | 0 | 1 | 2 | 3 | |
I | 11-16 | >0.15 | 0.1-0.15 | 0.07-0.1 | <0.07 |
J | 9-14 | >0.20 | 0.16-0.20 | 0.13-0.16 | <0.13 |
![]() |
7-12 | >0.20 | 0.16-0.20 | 0.13-0.16 | <0.13 |
The catalogue of DENIS sources covering each ISOGAL field was first
limited to sources
with a
detection, since a J-7
m association without
counterpart has a large probability of being a misidentification.
The density remains very high at this stage,
exceeding 105 sources/deg2in the Galactic Centre region. Therefore we further cut the DENIS catalogue
to a
magnitude that gave a source density of 72 000 sources/deg2for the ISO 3
pixel observations. For the observations with 6
pixels,
we proceeded in two steps, first limiting the DENIS source density to
18 000 sources/deg2 and then to 36 000 sources/deg2 (see below).
This confusion cut, with the procedure described
below, enabled us to limit the probability of
chance associations to a few percent even in the most crowded fields.
The search for DENIS associations was done with the same procedure as for the
7-15 m associations, with a smaller search radius.
The mean values of ISO-DENIS separations that we found are
typically in the 1-2
range for all ISOGAL fields, with a
few larger values for the FB fields, in which the association is done between
DENIS and 15
m coordinates (see e.g. Fig. E-31 in the electronic version
of this paper). The corresponding standard deviations are
mainly in the 1-1.5
range. An association radius of
3-4
is thus
appropriate to find most good associations with a low probability of
spurious results. However, a close inspection of the distribution
of association radii shows that, in a few fields with poor data quality, a few
real associations may have a larger association radius, in particular for
blended or extended sources with 6
pixels. Therefore, for the ISO 3
/pixel
observations, we used a 3.6
search radius. But for the ISO 6
/pixel
observations, we pushed the search up to a radius of 7
;
however, we carefully
distinguished by quality flags the associations with separations smaller or
larger than 3.5
.
With such values, the probabilities of random associations may appear
high. However, as discussed below, because of the large fraction of real
associations with smaller separation, the actual fraction of spurious
associations with reasonably good quality flags remains lower than a few
percent. The chance of spurious association is larger for weaker sources allowed with the higher density limit. The final ISO-DENIS quality
flag (Sect. 4.3.4) takes this point into account.
The ISO-DENIS association is characterised by a specific quality flag,
,
which ranges in values from 5 (highest quality) to 0
(no association). The computation of
this flag takes into account:
Finally when the derivation leads to
0, the association is
considered as invalid and no DENIS association is
given in the catalogue. With this
definition, associations with a quality flag equal to 4 or 5 can be considered
as secure, while a value of 3 is more uncertain but remains a high probability
association, and values of 1 or 2 are more doubtful but still include an
appreciable fraction of real associations. The distribution of
the computed ISO-DENIS association flags is shown in Fig. 7,
where it can be seen that
87% of the associations found have a good
quality (
), while fewer than 8% of the 7
m sources (LW2,
LW5 and LW6 filters) within the area observed by DENIS have no association.
The Point Source Catalogue contains a total of 106 150 sources, and is
composed of two sections. For each field, the "regular'' catalogue contains
all the sources inside the formal limits of the rectangular field, as defined
in Table 4 (see example in Fig. 1).
These limits have been
computed to avoid any border effects: all the sources inside this area
are located at more than two pixels from the saw-tooth edges of the observed
raster, both at 7 and 15 m for FC fields. This differs from the EDGE flag
computed for each wavelength (see Sect. 3.2.3) since the "regular''
region is limited to a rectangular area (whose axis are aligned along the
galactic ones) which has been fully observed at both wavelengths.
Then, the "edge'' catalogue contains the sources outside the limits of the
rectangular field, but excluding the measurements at less than two pixels
from the saw-tooth edges. This means that in the "edge'' region
of an FC field, it is possible to find a source with for example a 7 m
detection and no 15
m counterpart, simply because the edges of the
15
m raster do not exactly match the ones of the 7
m raster, so
that the source can be outside the region observed at 15
m or within 2 pixels of one saw-tooth edge.
As a result,
53% of the 7
m sources and
81% of the
15
m sources in the "regular'' regions of all FC fields have
an association at the other ISO wavelength, while these fractions become
47% for 7
m sources and
70% for 15
m sources
in the "edge'' regions.
Both the "regular'' and the "edge'' catalogues have the format described in Table 11, and a few examples of entries are given in Table 12. The final Catalogue contains 93 385 sources in the "regular'' regions, and 12 765 sources in the "edge'' regions.
Col. | Name | Format | Units [range] | Description |
1 | Number | a5 | source identification number in the field | |
2 | Name | a25 | ISOGAL-PJ
![]() |
source identifier (J2000)a |
3 | RAJ2000 | f8.4 | deg [0-360] | Right Ascension (J2000)b |
4 | DEJ2000 | f8.4 | deg [-90-+90] | Declination (J2000) |
5 | RAISOGAL | f8.4 | deg [0-360] | ISOGAL RA (J2000) |
6 | DEISOGAL | f8.4 | deg [-90-+90] | ISOGAL Dec (J2000) |
7 | G_lon | f8.4 | deg [-180-+180] | Galactic longitude |
8 | G_lat | f8.4 | deg [-90-+90] | Galactic latitude |
9 | I_field | a14 | Fxslllllsbbbbb | ISOGAL field name |
10 | D_field | a7 | DENIS image namec | |
11 | Imag | f5.2 | mag | DENIS I-band magnituded |
12 | Icorr | f4.2 | [0-1] | DENIS I-band correlation factor |
13 | x_I | f5.1 | pixel | x-position in DENIS I-band image |
14 | y_I | f5.1 | pixel | y-position in DENIS I-band image |
15 | Jmag | f5.2 | mag | DENIS J-band magnituded |
16 | Jcorr | f4.2 | [0-1] | DENIS J-band correlation factor |
17 | x_J | f5.1 | pixel | x-position in DENIS J-band image |
18 | y_J | f5.1 | pixel | y-position in DENIS J-band image |
19 | Kmag | f5.2 | mag | DENIS ![]() |
20 | Kcorr | f4.2 | [0-1] | DENIS ![]() |
21 | x_K | f5.1 | pixel | x-position in DENIS ![]() |
22 | y_K | f5.1 | pixel | y-position in DENIS ![]() |
23 | mag7 | f5.2 | mag | ISOGAL 7 ![]() |
24 | e_mag7 | f4.2 | mag | uncertainty in 7 ![]() |
25 | filt_7 | i1 | [2,5,6] | LW number of filter used |
26 | pfov_7 | i1 | arcsec [3,6] | pixel field of view |
27 | x_7 | f6.2 | pixel | x-position on ISOGAL final 7 ![]() |
28 | y_7 | f6.2 | pixel | y-position on ISOGAL final 7 ![]() |
29 | npix_7 | i1 | [0-7] | npix flag at 7 ![]() |
30 | mesh_7 | i1 | [1,2,3] | mesh flag at 7 ![]() |
31 | edge_7 | i1 | [0,1] | edge flag at 7 ![]() |
32 | qual_7 | i1 | [0-4] | global quality flag at 7 ![]() |
33 | mag15 | f5.2 | mag | ISOGAL 15 ![]() |
34 | e_mag15 | f4.2 | mag | uncertainty in 15 ![]() |
35 | filt_15 | i1 | [3,9] | LW number of filter used |
36 | pfov_15 | i1 | arcsec [3,6] | pixel field of view |
37 | x_15 | f6.2 | pixel | x-position on ISOGAL final 15 ![]() |
38 | y_15 | f6.2 | pixel | y-position on ISOGAL final 15 ![]() |
39 | npix_15 | i1 | [0-7] | npix flag at 15 ![]() |
40 | mesh_15 | i1 | [1,2,3] | mesh flag at 15 ![]() |
41 | edge_15 | i1 | [0,1] | edge flag at 15 ![]() |
42 | qual_15 | i1 | [0-4] | global quality flag at 15 ![]() |
43 | dis_II | f5.2 | arcsec | separation 7 to 15 ![]() |
44 | ass_II | i1 | [0-4] | 7-15 ![]() |
45 | dis_ID | f5.2 | arcsec | separation ISOGAL to DENIS associated sources |
46 | ass_ID | i1 | [0-5] | ISOGAL-DENIS association quality flag |
For the ISOGAL sources within the area observed by DENIS but with no
DENIS association, the I, J and magnitudes are set to 99.99,
while they are set to 88.88 for all the sources located outside
the region surveyed by DENIS. In these two cases,
the PSF correlation factors and pixel coordinates are set to 0.
The correlation factors with the PSF give an indication of the
photometric quality (see Simon et al., in preparation):
the uncertainty on the measured magnitude
is small when this factor is 0.95. On the other hand,
a value
0.85 means that the photometry is more uncertain
(typically by 0.1 to 0.2 mag). For bright sources, this
may come from moderate saturation effects, while for faint sources,
a value
0.80 is more typical. Nevertheless, a factor
0.70 indicates a poor photometric quality, which
may be caused by blending effects or confusion with the
background.
Col. | Name | Example 1 | Example 2 | Example 3 |
1 | Number | 0008 | 0017 | 0007 |
2 | Name | ISOGAL-PJ174118.0-282916 | ISOGAL-PJ174122.7-283146 | ISOGAL-PJ174117.6-282901 |
3 | RAJ2000 | 265.3250 | 265.3446 | 265.3234 |
4 | DEJ2000 | -28.4880 | -28.5296 | -28.4838 |
5 | RAISOGAL | 265.3251 | 265.3446 | 265.3230 |
6 | DEISOGAL | -28.4880 | -28.5296 | -28.4837 |
7 | G_lon | -0.1158 | -0.1419 | -0.1129 |
8 | G_lat | 1.0415 | 1.0048 | 1.0449 |
9 | I_field | FC+00000+00100 | FC+00000+00100 | FC+00000+00100 |
10 | D_field | 0955338 | 0000000 | 0955339 |
11 | Imag | 16.49 | 99.99 | 16.36 |
12 | Icorr | 0.96 | 0.0 | 0.91 |
13 | x_I | 367.7 | 0.0 | 376.6 |
14 | y_I | 153.8 | 0.0 | 735.6 |
15 | Jmag | 10.87 | 99.99 | 10.63 |
16 | Jcorr | 0.99 | 0.0 | 0.98 |
17 | x_J | 369.9 | 0.0 | 371.1 |
18 | y_J | 154.9 | 0.0 | 745.1 |
19 | Kmag | 8.32 | 99.99 | 8.02 |
20 | Kcorr | 0.98 | 0.0 | 0.99 |
21 | x_K | 368.1 | 0.0 | 370.5 |
22 | y_K | 151.8 | 0.0 | 750.9 |
23 | mag7 | 7.60 | 3.47 | 7.36 |
24 | e_mag7 | 0.03 | 0.01 | 0.03 |
25 | filt_7 | 2 | 2 | 2 |
26 | pfov_7 | 6 | 6 | 6 |
27 | x_7 | 165.76 | 181.81 | 164.10 |
28 | y_7 | 71.76 | 50.07 | 74.02 |
29 | npix_7 | 2 | 0 | 2 |
30 | mesh_7 | 3 | 3 | 3 |
31 | edge_7 | 0 | 0 | 0 |
32 | qual_7 | 4 | 4 | 4 |
33 | mag15 | 99.99 | 1.54 | 5.84 |
34 | e_mag15 | 0.00 | 0.03 | 0.06 |
35 | filt_15 | 0 | 3 | 3 |
36 | pfov_15 | 0 | 6 | 6 |
37 | x_15 | 0.00 | 181.64 | 163.77 |
38 | y_15 | 0.00 | 49.86 | 73.49 |
39 | npix_15 | 0 | 1 | 2 |
40 | mesh_15 | 0 | 3 | 3 |
41 | edge_15 | 0 | 0 | 0 |
42 | qual_15 | 0 | 4 | 4 |
43 | dis_II | 0.00 | 1.06 | 1.16 |
44 | ass_II | 0 | 4 | 4 |
45 | dis_ID | 0.32 | 0.00 | 1.25 |
46 | ass_ID | 5 | 0 | 5 |
The value of the ISOGAL 7-15 m association flag (see definition in
Sect. 3.6.3) is given in Col. 44, and the separation (in arcseconds)
between the 7
m and the 15
m positions (after correction of the field
offset) is given in Col. 43. This flag and the corresponding separation are set to
zero for sources with no 7-15
m association.
For the ISO-DENIS association, the quality flag (see definition in Sect. 4.3.4) is given in Col. 46, and the separation (in arcseconds) between the ISO and the DENIS positions (after correction of the field offset) is given in Col. 45. Again, these two entries are set to zero when there is no ISO-DENIS association.
As explained in Sect. 3.2.5, three kinds of extracted sources brighter than the limiting magnitude of each field are considered spurious: (1) the sources found only in the "inversion'' processed raster, with no counterpart in a 1 pixel search radius in the "vision'' raster, (2) the sources with simultaneously a doubtful inversion-vision association (with a separation between 0.5 and 1 pixel) and with a poor detection confirmation (i.e. with no association between the mesh=1 and the mesh=2 results), and (3) the possible remnants of bright sources, found by a procedure that looked at the same pixel location in the five successive images of the implied raster.
These sources are published in three distinct tables. Their format is defined in Table 13. The numbers, as they appear in Col. 1, are preceded by an "I'' for the "inversion-only'' sources, by an "M'' for the sources of the second class and by an "R'' for the probable remnants.
Note that most spurious sources of the first two kinds are probably artifacts, but can also be related to faint extended structures, for which different parameters in the extraction process result in slightly different coordinates. The third class of spurious sources is essentially composed of spurious remnants, but may contain a few real sources, which have been accidentally discarded by the procedure because of spatial coincidence with a putative remnant.
The ISOCAM images have been initially processed using version 7.0 of
the off-line processing pipeline (Sect. 3.1).
Similar images processed with the latest version of OLP
are now publically available through the Data Archive on the
ISO web site.
However, we make available here the OLP7 images together with version 1
of the PSC for consistency, because they have been used for the
extraction of the sources of this catalogue. Improved ISOGAL images
(Miville-Deschênes et al. 2000 and in preparation) will
be published with version 2 of the catalogue.
Because of the difference in orientation between the individual images (aligned along the satellite axis, thus with the equatorial coordinates) and the mosaiced rasters (aligned along the galactic axes), and of different times of observations, the orientation obtained after the OLP7 processing was different from one raster to another. We therefore decided to change this orientation if necessary, in order to use the same convention for all rasters, and set the orientation to l along decreasing Xand b along increasing Y.
A more important improvement provided by the construction of the ISOGAL PSC
deals with the astrometry, which has been tied to DENIS whenever possible.
The offsets
that we applied to the source coordinates in order to associate the
ISO sources with DENIS have also been applied to the rasters, as
indicated in Table 2.
For the FC fields with no DENIS observation, the astrometry of
the 15 m rasters has been tied to that of the 7
m ones.
The corrected images are available through the CDS and the
IAP
server.
Col. | Name | Format | Units [range] | Description |
1 | Number | a4 | identification number in the ION | |
2 | RAJ2000 | f8.4 | deg [0-360] | ISOGAL RA (J2000) |
3 | DEJ2000 | f8.4 | deg [-90-+90] | ISOGAL Dec (J2000) |
4 | Mag | f5.2 | mag | ISOGAL magnitude |
5 | ION | a8 | ISO observation number | |
6 | x | f6.2 | pixel | x-position on ISOGAL final image |
7 | y | f6.2 | pixel | y-position on ISOGAL final image |
The first version of the ISOGAL-DENIS Point Source Catalogue contains
a total of 106 000 sources, with one or two magnitude measurements
in the mid-infrared (7 and 15 m), and up to three magnitude
measurements in the near-infrared (I, J and
bands of the
DENIS survey, see Tables 11 and 12).
The data are presented in two similar tables, corresponding
to the "regular'' and the "edge'' regions of the observed fields.
The latter contains the sources from the edges of the ISOCAM rasters, where
border effects can occur, which can lead to non-association between the
two ISO bands.
The typical RMS photometric uncertainty is at most 0.1 mag for the
DENIS bands, and better than 0.15 mag for the ISO bands in most cases,
but it can reach 0.3 mag for the faintest sources in the densest fields.
For the most numerous fields observed with broad filters, the limiting
magnitudes of the published catalogues range between 8.8
and 10.1 at 7
(with a median value equal to 9.46 mag, or
15 mJy), and between 7.7 and 8.8 at 15
(median 8.16 mag,
11 mJy), depending on the source density.
For the most difficult fields observed with narrow filters, these
limits range between 8.2 and 9.6 mag at 7
and between
7.0 and 8.2 mag at 15
.
These limits are conservative and the fainter sources have been rejected
in the present version of the PSC
.
The current astrometric accuracy of the DENIS data used is better than 0.5
(RMS).
The final coordinates (as they appear in Cols. 3 and 4 of the catalogue - see
Table 11 - in equatorial J2000 system, in Cols. 7 and 8 in
the galactic system, and in the name of the source, Col. 2)
of all ISOGAL sources with a DENIS counterpart are the DENIS ones, and should
also be accurate to 0.5
.
The astrometry of the ISOGAL sources with no DENIS
association, but within the fields observed by DENIS, is also tied to the
DENIS coordinates, and should therefore be accurate to
2
(RMS).
Finally, ISOGAL sources located
outside the area surveyed by DENIS may suffer from the lens wheel jitter
of ISOCAM, resulting in a maximum
10
systematic offset in the
extracted coordinates.
Several flags have been implemented to characterise the reliability of the
sources, the quality of their photometry and of the associations between
the different bands. An indication of the reliability of the mid-infrared
detection is also given by the mesh flag (Col. 30 for 7 m and Col. 40
for 15
m, see Table 11). A value of 3 indicates a good reliability
level, while a value of 1 or 2 shows that the extraction was not perfectly
confirmed, making the real point-like nature of the source doubtful.
The global quality of the ISO photometry and reliability of each source
is quantified by one
quality flag for each band. These two flags are given in Col. 32 for 7 m
and in Col. 42 for 15
m, and range from 1 to 4, the highest value
corresponding to the best quality. Thus sources with quality flags equal to 1
or 2 should be considered with caution.
The quality of the association between the two ISO bands is also characterised
by a specific flag, which appears in Col. 44, together with the separation
of the association in Col. 43.
When this flag is equal to 3 or 4, which means that the separation between
the 7 m coordinates and the 15
m ones is smaller than one pixel,
the validity of the association is almost certain,
while a value of 1 or 2 means that the association has to be carefully
checked, but it may be a real association for slightly extended sources.
Finally, the quality of the ISO-DENIS association is quantified by a flag given in Col. 46 (and the ISO-DENIS separation appears in Col. 45). Here, values of 4 or 5 correspond to secure associations, while a value of 3 means that the association was not straightforward, but it still has a good probability to be real. When this flag is equal to 1 or 2, the reality of the association has to be checked carefully, using for instance colour compatibility criteria.
With the first public version of the ISOGAL-DENIS Point Source Catalogue,
we provide the astronomical community with a catalogue containing about
105 mid-infrared sources, detected at 7 and/or 15 m in the
obscured centre of the Galaxy. The bulk
of them are associated with near-infrared data from the DENIS survey. We
also provide nearly 400 mid-infrared images, with an astrometric
accuracy of
1
for most of them.
All the data were reduced using data products of version 7 of the ISO off-line processing pipeline. Additional specific procedures enabled us to greatly reduce the number of artifacts and to reduce the photometric uncertainty to typically 0.15 mag, at the cost of limiting the published catalogue in the densest observed fields to levels well above the sensitivity limit of a few mJy.
A second version of the catalogue is already under development, based on a systematic reprocessing of the raw data using the most up-to-date specialised procedures (Miville-Deschênes 2000 and in preparation). This second version will also contain systematic cross-associations with the near-infrared data of the 2MASS survey, and with the mid-infrared data of the MSX survey.
Acknowledgements
We thank the whole ISOGAL Team for its contribution to the project and to the production of the present catalogue.
The ISOCAM data presented in this paper were analysed using "CIA'', a joint development by the ESA Astrophysics Division and the ISOCAM Consortium. The ISOCAM Consortium is led by the ISOCAM PI, C. Cesarsky. We thank A. Abergel, H. Aussel, A. Coulais, R. Gastaud, M. Pérault, J. L. Starck and many other members of the ISOCAM team, of the ISO/ESA team at Villafranca and especially of the CIA team for their help in the ISOGAL data reduction. We are very grateful to all people who contributed to the ISOGAL data reduction, including T. August, X. Bertou, E. Copet and M. Unavane.
We thank the whole DENIS Team, and especially its PI, N. Epchtein, and S. Bégon, J. Borsenberger, B. de Batz, P. Fouqué, S. Kineswenger & D. Thiphène for making available the DENIS data. The DENIS project is supported, in France by the Institut National des Sciences de l'Univers, the Education Ministry and the Centre National de la Recherche Scientifique, in Germany by the State of Baden-Würtemberg, in Spain by the DGICYT, in Italy by the Consiglio Nazionale delle Ricerche, in Austria by the Fonds zur Förderung der wissenschaflichen Forschung and the Bundesministerium für Wissenschaft und Forschung.
This publication made use of data products from the Midcourse Space Experiment. Processing of the data was funded by the Ballistic Missile Defense Organization with additional support from NASA Office of Space Science.
This work was carried out in the context of EARA, the European Association for Research in Astronomy.
S. Ganesh was supported by a fellowship from the Ministère des Affaires Étrangères, France, and this research was supported by the Project 1910-1 of Indo-French Center for the Promotion of Advanced Research (CEFIPRA). SG also acknowledges the support he received from the French CNRS for participating in the astronomical school in Les Houches in 1998. M. Schultheis acknowledges the receipt of an ESA fellowship. B. Aracil and A. Soive were posted to the ISOGAL Project by the Délegation Générale de l'Armement, France.
We are grateful to Dr. M. Cohen for his help in the calibration of ISOCAM data, and to Dr. S. Ott and Prof. I. S. Glass for their useful comments and inputs.
A complete description of the data processing and of the procedures that were run to quantify the quality of the data is given in this complete version as Online Material.
Explanatory Supplement of the ISOGAL-DENIS Point Source Catalogue PDF File (1.29 Mb)