IRAS Explanatory Supplement
V. Data Reduction
H. The Point Source Catalog
Table of Contents | Index | Previous Section | Next Section
- Processing Overview
- Clean-Up Processing
- Neighbor Tagging
- Cirrus Flagging
- Average Flux Computation and Variability Analysis
- High Source Density Regions
- Catalog Source Selection
- Low-Resolution Spectral Associations
- Associations
H. 1 Processing Overview
After the Working Survey Data Base (WSDB) was completed by the sequential processing of every SOP, several programs processed that data base to create the catalog.
The major steps were as follows:
- Final calibration corrections,
- Clean-up processing. which forced sources closer than 30" (in-scan) and 90" (cross-scan) to weeks-confirm,
- Point source and small extended source neighbor tagging,
- Cirrus (highly structured 100 µm emission) flagging,
- High source density processing,
- Variability tagging and average flux computation,
- Association and classification of low-resolution spectra,
- Associations with previously known astronomical sources,
- Catalog source selections,
- Transformation of coordinates, assignment of position uncertainties and assignment of source names.
Each of these steps is discussed separately below except for step 1. Step 1, incorporating the final calibration, is discussed in Section VI.B.
H.2 Clean-Up Processing
An analysis of sources in the WSDB that were within several arc minutes of each other showed that roughly 5% of all sources did not have every possible hours-confirmed detection combined into one source. There were three reasons for this. First, sources scanned more than five times within one hours-confirming coverage were separated into two distinct hours-confirming detections due to an upper limit of four scans per hours-confirming detection imposed by the processing. Only one of these two detections was allowed to weeks-confirm with other hours-confirmed detections already present in the WSDB. Second sources that suffered band merging difficulties occasionally produced two hours-confirmed detections out of one hours-confirming coverage. Only one piece was allowed to weeks-confirm. Finally, due to the way the minisurvey scans were scheduled the first survey coverage was not allowed to weeks-confirm with some of the minisurvey sources. Each of these cases was sufficiently well defined to allow a clean-up processor to give these pieces a chance to weeks-confirm with each other. Additional problems were caused by radiation hits, source confusion and by multiple scans inadverently spaced too close together in time to allow weeks-confirmation. These sources were cleaned up in a two-step process. First, all weeks-confirmed sources were given the opportunity to confirm using the standard threshold with any hours- or weeks-confirmed neighbors located within 30" in-scan and 90" cross-scan. Most sources that confirmed at this stage were ones that had earlier been denied the opportunity for purely technical reasons. A single source with a refined position resulted.
The second stage of clean-up was to force all weeks-confirmed sources to merge with any neighbors located within 30" in-scan and 90" cross-scan, whether or not the sources were confirmable according to the weeks-confirmation decision process. No position refinement attempt was made. This step proved necessary because analysis of near neighbors showed that most were caused by the incorrect splitting of single sources in confused regions by the various confirmation processors.
A separate processing problem was partially solved by the clean-up processor as well. Two normal survey scans were placed within one layer of the minisurvey (Section III.C.11, Table III.C.1). These scans were inadverently processed twice; once as par of the minisurvey and once as par of the regular survey. Thus some sources (those missed by that layer of the minisurvey) had identical hours-confirming detections present as two separate entries within the source. One of these was deleted by the clean-up processor, although nothing could be done to remove the double weight given to that sighting in the position refinement process.
H.3 Neighbor Tagging
The relative isolation of a source provides the user with an indication of its quality. Associated with each source are the numbers of hours- and weeks-confirmed point sources and small extended sources. The search area used was a box 6' (half-width) in-scan, which was the largest detection shadowing distance (Section V.C.7), and 4.5 (half-width) cross-scan, which was a detector length. Neighbors were tagged before the clean-up processor was applied. However, neighbors that were confirmed with the source during clean-up were not counted, nor, due to an error, were any sources within a box that had been successfully cleaned up.
If no neighbors were found, it is unlikely that processing or confusion problems exist. The presence of neighbors should make the user check those neighbors to investigate the possibility that the source was extended, was in a confused patch of sky, or suffered some other problem.
H.4 Cirrus Flagging
At 100 µm the infrared sky is characterized by emission from interstellar dust on all spatial scales, known affectionately as "infrared cirrus". A significant chance exists that ANY catalog source has been affected by components of this long-wavelength emission on the point source scale. Four separate quantities were derived for each source to assess the importance of contamination by cirrus. While these are imperfect flags (a source may be affected by cirrus without any of these flags being set, or may not be affected at all even at high values of the flags) they work for the majority of sources and provide a simple description of the sky at 100 µm in the vicinity of each catalog object.
Flag 1 (CIRR1) is the density of WSDB sources (hours and/or weeks-confirmed) detected only at 100 µm within a ± ½ ° ecliptic coordinate box of the source position. Cirrus typically produces strings of such sources, giving rise to a high density. Weak cirrus, producing only a few 100 µm sources, may still affect the quality of catalog sources.
Flags 2 (CIRR2) and 3 (CIRR3) come from the 100 µm extended source data at a resolution of ½ ° (see Section V.G.5). Although the angular resolution is a serious limitation for these flags, since the angular scale of point sources is 10-30 times smaller, cirrus on the point source scale usually shows structure on the ½ ° scale as well.
CIRR2 was derived from a spatially filtered version of the ½ ° averaged 100 µm emission. The value of the filtered sky brightness at a point, Ji, is given by
(V.H.1)
(V.H.2)
A value of CIRR2=1 corresponds to the cirrus flux being less than or
equal to 0.01 of the source flux while CIRR2=9 corresponds to the cirrus
flux being 10 times the source flux. As described in
Section VII.H, CIRR2 less than about 4 is
indicative of little or no cirrus contamination, while larger values probably
indicate significant contamination. A value of CIRR2=0 indicates that no 100
µm extended emission data were available.
Flag 3 (CIRR3) is the total intensity of the 100 µm extended
emission in a ½ ° beam (MJy sr-1) High values are
indicative of a large column density of interstellar dust.
A fourth cirrus indicator is the presence of nearby small extended sources
at 100 µm. Since the small extended sources have sizes much
closer to point sources than the ½ ° data, this flag (SESl), which
gives the number of hours-confirmed small extended sources, will normally
be an accurate cirrus indicator. However, because of the higher thresholds
used for this processing. weak cirrus may not be detected.
As discussed in more detail in Section VII.H,
if any of these flags takes on a large value, one should be cautious. Cirrus
can affect the point source flux in any band by causing band-merging
difficulties and can certainly affect the quality of a quoted 100 µm
flux.
In the computation of an average flux for each source, three levels
of flux quality were recognized:
High and moderate quality fluxes were obtained by log-flux averaging
of individual acceptable hours-confined fluxes weighted by the inverse
of the log-flux variances. If the flux discrepancy flag was set (see below)
and the source had more than two hours-confirmed fluxes, then one flux
was removed from the averaging process. The rejected flux was the one farhest
from the median flux in units of the error. The median flux was arbitarily
picked to be the larger of the middle two fluxes when there was an even
number of fluxes present. The rational behind this was that discrepant
fluxes occasionally occurred because of radiation hits, a passing asteroid
a failed detector, or processing problems. With more than two measurements
it was possible to weed out the discrepant flux. For quoted fluxes, a
Uncertainties quoted for high and moderate quality fluxes were the maximum
of a) the uncertianty of the mean derived from the accepted fluxes; or
b) the uncertainty of the mean derived by propagating the quoted uncertainties
of the individual, accepted flux measurements.
Fluxes quoted as upper limits were obtained from the the largest flux
(or limit) for sources with only two hours-confirmed sightings and from
the median flux for sources with more than two detections. The median was
taken as the largest of the two middle fluxes for sources with an even
number of observations. The flux discrepancy flag was set if the largest
and smallest flux from all the limits for low quality detections differed
by more than a factor of 3.
Sources with high or moderate quality fluxes at both 12 and 25 µm
were examined for variability. A percentage probability of being variable
was quoted for sources whose fluxes at BOTH 12 and 25 µm either
increased or decreased significantly from one hours-confirmed sighting
to another. In other words, the changes in flux at 12 and 25 µm
had to be correlated. Although variability could have been detected in
other ways, this method led to a simple and reliable determination of
variability of a large number of sources. For sources with more than one pair
of sightings,
the value determined for the pair of sightings with the greatest likelihood
of variability was given.
The probability that a source brightness varied was calculated by comparing
the number of sources with correlated flux excursions exceeding
m (V.H.3)
The density of sources in the WSDB exceeds the resolving capability
of the instrument over some parts of the sky. The nature of both the sky
and the instrument make this a wavelength-dependent effect; at 12 µm
the instrumental resolution was less than 1' and the sky was dominated
by point sources, while at 100 µm the resolution was about 4'
and much of the sky was dominated by highly structured diffuse emission.
The overriding concern in developing a strategy for regions of high
source density was to insure the reliability of the information presented
in the point source catalog. Reliability, in this context, has three meanings:
1) a point source must be an inertially fixed celestial source (the basic
meaning of reliability throughout the catalog); 2) a source must represent
a very compact structure that, to the extent possible, is not merely a
fragment of a complex background; and 3) the intensity measurements reported
in the catalog must repeatably represent the brightness of a source above
the local background.
The first step in processing high source density regions was to determine
whether the number of sources in a given wavelength band in a
1°2
bin exceeded a "confusion limit". Sources in bins with more than
the threshold number of sources were subject to additional criteria for
inclusion in the catalog. The decision to apply the criteria was made independently
on a band-by-band basis. Thus, a multi-band source might have high source
density rules applied to its 100 µm measurements, but not to
its 12 µm measurements. Once the high source density rules were
invoked in a band, each source in that band in that bin was examined for
the quality and repeatability of its individual measurements and for its
isolation from other potentially confusing sources. The high source density
processor determined whether the measurement of the source in that band
was of high, medium or low (upper limit) quality, where these definitions
differ from those in lower density areas, and calculated an average flux
for the source in that band. Many sources were found wanting and were excluded
from the catalog.
The high source density selection criteria used only information contained
within the WSDB to judge the relative merits of a source. The confusion
processing algorithms were developed and tuned in two confused regions,
one in the Large Magellanic Cloud (LMC) and one near the Galactic Center.
Detector strip charts were examined for many of the sources to verify the
validity of the selection criteria.
H.5 Average Flux Computation and Variability
Analysis
values.
2
test was applied to each band. If the value of the reduced
2
was greater than 9, the flux discrepancy flag was set in that band.
at 12 and 25
µm,
Ncorr(>/-m), with those sources showing anti-correlated
flux excursions exceeding m
in
both bands, Nanti(>/-m). The probability that a source is
variable is given by
H.6 High Source Density Regions
H.6.a Location of High Source Density
Regions
 |
|
Figure V.H.1.1 The parts of the 12 µm sky processed according
to high source density rules are shown in an Aitoff projection in
Galactic coordinates. The black regions contain more than 45 sources per
sq. deg, the threshold for high source density processing; the
grey areas contain more than 25 sources per sq. deg, one-half
the confusion limit. larger largest |
The confusion-limited source density is determined by the instrumental
angular resolution and how many detector beam areas are, on average, required
per source for a reliable measurement. A conservative limit of 25 beams
per source leads to maximum source densities of 50, 50, 25 and 12 sources
per sq. deg at 12, 25, 60 and 100 µm (assuming a detector area
of the nominal in-scan detector size times the detector cross-scan width).
Aitoff projection maps in Galactic coordinates showing the density of sources
with moderate or high quality fluxes (according to the basic rules described
in V.H.5) in a given wavelength band are shown in
Figs. V.H.1a-d. At 12, 25 and 60 µm the
region within 3 - 5°
of the Galactic plane and within ± 100° of longitude of the
Galactic
center plus small regions in Orion, Ophiuchus and the LMC present the major
problem areas. The 100 µm sky presents a more complicated picture
because the regions of high source density cover large parts of the sky
due to the infrared cirrus.
 |
|
Figure V.H.1.2 The parts of the 25 µm sky processed according
to high source density rules are shown in an Aitoff projection in
Galactic coordinates. The black regions contain more than 45 sources per
sq. deg, the threshold for high source density processing; the
grey areas contain more than 25 sources per sq. deg, one-half
the confusion limit. larger largest |
To set the thresholds for high source density processing, the numbers
of sources in 1°2 and ¼°2 bins were
examined in the WSDB. It was determined that processing all
1°2 bins containing at least
45, 45, 16, and 6 sources (at 12, 25, 60 and 100 µm) would clean
up 90% of all ¼°2 areas with confusion-limited source
densities. Thus, any 1°2 bin (defined in ecliptic coordinates,
cf. Appendix X.1) containing at least the
threshold number of sources in a particular
band was processed according to high source density rules in that band.
No attempt was made to join together high source density regions into a
few simply connected areas. A list of the high source density bins in each
band is available with the machine readable version of the catalog.
 |
|
Figure V.H.1.3 The parts of the 60 µm sky processed according
to high source density rules are shown in an Aitoff projection in
Galactic coordinates. The black regions contain more than 16 sources per
sq. deg, the threshold for high source density processing; the
grey areas contain more than 12 sources per sq. deg, one-half
the confusion limit. larger largest |
High source density rules were occasionally invoked for wavelength bands
with fewer than the threshold number of sources in a bin if failure to
include that band would have resulted in a non-adjacent set of bands being
processed, i.e., a "spectral hole". Thus, for example, if the
number of sources at 25 and 100 µm, but not at 60 µm,
exceeded the thresholds, the decision was made to process 60 µm
sources as well.
 |
|
Figure V.H.1.4 The parts of the 100 µm sky processed according
to high source density rules are shown in an Aitoff projection in
Galactic coordinates. The black regions contain more than 12 sources per
sq. deg, the threshold for high source density processing; the
grey areas contain more than 6 sources per sq. deg, one-half
the confusion limit. larger largest |
H.6.b Catalog Selection Criteria in High
Source Density Regions
In unconfused regions where the reliability of a single hours-confirmed source is high, the criteria for including sources in the catalog emphasized completeness. In high source density regions, however, the criteria emphasized reliability. Thus, the first step toward weeding out unreliable point sources was to stiffen the requirements for accepted detections and valid weeks-confirmed sources in a given band.
The concept of high, medium and low quality fluxes is crucial to understanding how sources were included in the catalog. High quality measurements were those which passed through the entire data processing chain without blemishes such as more than one missing detection (per hours-confirmed sightings), even those missing due to failed detectors. Medium quality fluxes could suffer from a variety of problems due either to the faintness of the source, the presence of failed detectors, the complexity of the background or peculiarities of the data processing. Low quality fluxes were upper limits based on an estimate of the local noise. The definition of the relative quality of fluxes and the way that sources were selected for inclusion in the catalog on the basis of their quality differs depending on whether a source was found in a region of high or low source density.
In regions of high source density, tests were applied successively to the measurements of a source to determine its quality. If all of the criteria described below were satisfied, the source was deemed to have a high quality measurement in that band. If only a subset of the criteria were satisfied, then the source was considered to have a medium quality measurement in that band. If none, or only a few, of the criteria were satisfied then an upper limit was given for the flux of that source. To be included in the catalog a source had to have a high quality measurement in at least one band.
For high source density regions the requirements for a high quality flux were stiffened by demanding that at least two hours-confirmed sightings in one band each meet all of the following criteria: 1) seconds-confirmed detections on at least two orbits; 2) a minimum correlation coefficient of 0.97, corresponding to a local signal-to-noise ratio of about 12; and 3) fewer than four detections within the seconds-confirmation window on a given scan. Simple edge detections (3 detections) were permitted.
A high quality measurement had to be repeatable. The ratio of brightness to faintest high quality flux measurements in a band had to differ by less than a factor of 3.0. If this repeatability requirement was not satisfied, then the brightest measurement was taken as a medium quality
measurement. Although this requirement in principle biases the catalog against variable sources in high source density regions, in practice, relatively few sources were rejected for this reason.
H.6.c Weaker Neighbors
The above rules all pertain to a single source, independent of its environment. The weaker neighbor algorithm was designed to remove sources adversely affected by brighter, neighboring sources. The "neighbor" of a source with a high quality flux in some band must itself have at least two hours-confirmed measurements in that same band and lie within a wavelength dependent position window. If the brightest flux of one source was a factor of 1.2 weaker than the faintest flux of the other object in the same
band then the fainter object was marked as a "weaker neighbor" in that band. The flux status of the weaker neighbor was demoted to low quality and the upper limit assigned was that of the flux of the brighter object. The half-widths of the weaker neighbor window at (12, 25, 60 and 100 µm) were (90", 90", 180" and 360") and the half-lengths were (270", 272",286", 300"-a detector length).
H.6.d Confused Neighbors
A good indicator that a source was part of a more extended structure was to have two or more sources located very close together. A source with a high quality measurement in one band was examined for neighbors in the same window as described for weaker neighbors. If another source were located sufficiently close to the first and if any confusion status flags other than edge detection flags (Section V. D.2) were set in either source, then the fluxes of both were marked low quality and the brightest of the two fluxes was assigned as an upper limit to each source. If no confusion status flags were set, neither source was changed. The definition of a neighbor and the size of the confused neighbor window were as described above for the weaker neighbor test.
H.6.e Very Near Neighbors
One of the original selection rules was to exclude from the catalog all sources, in ail bands, with a neighbor closer than 30" in-scan and 90" cross-scan. Because the clean-up processor described in Section V.H.2 forced the weeks-confirmation of all sources within-n such a box, there should have been no such sources remaining to delete. Due, however, to slight differences in the way in which the neighbor boxes were calculated by the two programs, about twenty sources at the periphery of the neighbor box were deleted by the high source density processor.
H.6.f Moderate Quality Fluxes
Measurements that failed one or another of the above rules could still carry useful information about the strength of a source in a given band. Moderate quality fluxes had to have at least two hours-confirmed measurements with flux status, FSTAT = 3, 4, 5 or 7, and correlation coefficient > 0.95. Fluxes of otherwise moderate quality could be demoted to low quality by either the weaker or confused neighbor rules.
H.6.g Low Quality Fluxes (Upper Limits)
Measurements in a given band that failed to meet either the high or
moderate quality criteria become upper limits for the source. If multiple
detections were available, the brightest one was given. As mentioned above,
confused or weaker neighbors were given the brightest flux of either of
the two sources as an upper limit. Sources of low quality were examined
for brighter neighbors and were assigned the brightest neighboring flux
(see above for neighbor rules). Sources with no detections on any sightings
(i.e., with only noise fills present) and no neighbors were given an upper
limit based on 12 times the
1- value
reported by the noise estimator. This value was chosen since the correlation
coefficient requirement of >/- 0.97 corresponds to roughly a local
signal-to-noise ratio of 12.
H.6.h Flux Averaging and Uncertainties
In all cases, averages and uncertainties were computed using the logarithms of the fluxes, not the fluxes themselves. However, the technique used to obtain the average values for both high and moderate quality fluxes in high source density regions was different from the one used outside of dense regions. Rather than computing inverse-variance weighted averages in regions where the noise estimator was suspect, simple averages of the logarithms of the valid high (or moderate) quality fluxes were computed. The uncertainty associated with the measurement was the greater of: 1) the average of the N uncertainties quoted for the N valid hours-confirmed measurements, divided by N ½; or 2) the standard deviation of the mean of the N values averaged together to obtain the flux average.
H.7 Catalog Source Selection
Outside of regions of high source density, two rules were used to select weeks-confirmed WSDB sources for inclusion in the catalog: sources had to have either a high quality flux in at least one band or adjacent bands, both of which have an HCON with FSTAT = 7 and another with FSTAT = 3 or 5 (Section V.H.5). The minimum time separation of 123,000 sec. required for two hours-confirmed sightings to weeks-confirm was found to be long enough to reject even slow moving asteroids.
Within high source density regions, as described in V.H.6, sources had to have at least one high quality flux in one band and be relatively isolated from other sources to be included in the catalog.
H.8 Low-Resolution Spectral Associations
For each source in the WSDB, a check was made to see whether a low-resolution spectrum was available. See Chapter IX for more details.
H.9 Associations
Positional associations of IRAS sources are made with objects in other astronomical catalogs. The associations in the IRAS catalog were based purely on positional agreement, with no attempt made to distinguish between multiple sources associated with a particular IRAS source. Any number of sources from a variety of catalogs could be associated with a specific IRAS object as long as the position test was met. No attempt was made to "identify" an IRAS source with a source from another catalog, by requiring the IRAS source to have a "reasonable" energy distribution for the identification. The only attempt to preclude spurious associations was to forbid a source with only a 100 µm flux (i.e., a source having a significant likehood of being Galactic "cirrus") from being associated with stars.
Because the catalogs used for associations had a wide range of positional accuracies, and the IRAS positions were intermediate between the very high positional accuracy catalogs and the lower accuracy catalogs, the following procedures were adopted to make the associations. For the high-positional-accuracy catalogs, a window of half width 8" in-scan and 45" cross-scan was used. If the position of the IRAS source lay within this distance of the cataloged source, then an association was made. For those catalogs where the positional uncertainty was greater than that of the IRAS source, an association was made if the source position and the IRAS position agreed to within a fixed radius or to within the radius (see below) of the cataloged source, if available.
In addition to the other catalogs, associations with IRAS small extended sources have been included so that the user will know if a small extended source is close to the point source.
Table V.H.1 identifies the catalogs used and gives the search radius for the association and an indication of whether a source slae was used for the association radius. The information carried with the association is described in detail in Section X.B and includes such information as visible magnitude, size, spectral type, and morphological type. In addition, the distance and position angle from the IRAS position to the location of the cataloged source are given.
| Catalog No. |
Catalog Type |
Catalog Name | Search Radius |
Source Size Used in Association |
|---|---|---|---|---|
| 01 | 2 | General Catalogue of Variable Stars Kukarkin, et al. | 90" | |
| 02 | 2 | Dearborn Observatory Catalogue of Faint Red Stars, Lee, et al. | 90" | |
| 03 | 3 | Air Force Geophysical Laboratory Four-Color Survey, Price & Murdock | 90" | |
| 04 | 2 | Two Micron Sky Survey Neugebauer and Leighton | 90" | |
| 05 | 3 | Globule List Wesselius | 90" | X |
| 06 | 1 | Second Reference Catalogue of Bright Galaxies, de Vaucouleurs, et al. | 90" | |
| 07 | 2 | Early Type Stars with Emission Lines Wackerling | 90" | |
| 08 | 3 | Equatorial Infrared Catalogue (Sweeney et al.) | 90" | |
| 09 | 1 | Uppsala General Catalogue of Galaxies Nilson | 90" | |
| 10 | 1 | Morphological Catalogof Galaxies Vorontsov-Velyaminov, et al. | 180" | |
| 11 | 3 | Strasbourg Planetary Nebulae | 90" | |
| 12 | 1 | Catalogue of Galaxies and Clusters of Galaxies, Zwicky, et al. | 90" | |
| 13 | 2 | Smithsonian Astronomical Observatory Star Catalog | 45"(x-scan) x8"(in-scan) |
|
| 14 | 3 | ESO/Uppsala Survey of the ESO (B) Atlas, Lauberts | 90" | |
| 15 | 2 | Bright Star Catalogue - 4th Edition Hoffieit | 45"x8" | |
| 16 | 2 | New Catalog of Suspected Variable Stars Kukarkin, et al. | 90" | |
| 17 | 2 | General Catalogue of Cool Carbon Stars | 90" | |
| 18 | 2 | Catalog of Nearby Stars Gliese | 45"x30" | |
| 19 | 2 | General Catalog of S Stars Stephenson | 90" | |
| 20 | 3 | Parkes HII Region Survey Haynes, et al. | 120" | X |
| 21 | 3 | Bonn HII Region Survey Altenhoff et al. | 80" | |
| 22 | 3 | Catalog of CO Radial Velocities Toward Galactic HII Regions, Blitz et al. | 80" | X |
| 23 | 3 |
|
X | |
| 24 | 2 | Two Micron Sky Survey with Improved Positions, Kleinmann and Joyce | 45"x8" | |
| 25 | 1 | Catalog of Dwarf Galaxies van den Bergh | 90" | |
| 26 | 1 | Atlas of Peculiar Galaxies Arp | 120" | |
| 27 | 1 | Galaxies withanUltraviolet Continuum, Markarian, et al. | 90" | |
| 28 | 1 | Catalog of Extragalactic Radio Sources Having Flux Densities Greater than 1 Jy at 5 GHz, Kuhr, et al. | 60" | |
| 29 | 1 | Catalogue of Quasars and Active Nuclei, Veron-Cetty and Veron | 90" | |
| 30 | 1 | Lists of Galaxies Zwicky | 90" | |
| 31 | 1 | Atlas and Catalog of Interacting Galaxies, Vorontsov-Velyaminov | 120" | |
| 32 | 3 | IRAS Small Scale Structure Catalog | min(120", diameter/2) |
X |
| 39 | 3 | Ohio State University Radio Catalog | 120" | |
| 40 | 2 | University of Michigan Spectral Atlas | 60" 45" x 8" |
(Vol. 1) (Vols. 2,3) |
| 41 | 3 | IRAS Serendipitous Survey Catalog | 60" | |
| Catalog types include (1) extragalactic, (2) stellar or (3) other, e.g. dark clouds, HII regions, etc. | ||||
| Catalog numbers 33-38 reserved for internal use. | ||||