3 Analysis

It is not the intent of the paper to attempt to examine in detail every IRAS-CO source in Table 1. Rather in this section we conduct a general analysis of the objects contained in the table in terms of their IRAS colours and their $N_{\rm E}$ values to give the reader some appreciation for the overall content. In Table 3 some basic properties of the IRAS-CO sample are summarized.

An obvious question that can be raised is how do the results of this study differ from the targeted WB89 study within the OGS region? Within the OGS region (completely encompassed by the WB89 survey region) WB89 examined 292 IRAS sources of which 255 had CO detected along the line of sight. Of the 255 WB89 associations, 244 are matched by a rank = 1 BKP association ($\sim$96%). In 6 of the other cases (WB89 200, 287, 321, 376, 392, and 460) the BKP catalogue does include the WB89 associated CO but we assign it a lower ranking based upon the relative $N_{\rm E}$ values of the various CO clouds along the line of sight. For the remaining 5 cases (WB89 298, 365, 378, 398, and 439) there is no BKP cloud associated with the WB89 CO component, because it is either too faint or has too small a spatial extent (see Sect. 2 of BKP for details of the cloud selection criteria). For four of these sources (WB89 365, 378, 398, and 439) we associate, with $N_{\rm E} \geq 0.1$ in all cases, another CO cloud along the line of sight with the IRAS source. This leaves us with 4061 other IRAS-CO objects to investigate - what are these objects? A histogram showing the distribution of $N_{\rm E}$ values for these objects is shown in Fig. 2 in comparison with the histogram for the 254 WB89 sources included in the BKP catalogue. The $N_{\rm E}$bins for all of the histograms shown in this paper are evenly distributed in log space from 10^-3.5 to 10^-0.5. Clearly some of our 4061 IRAS-CO sources are spurious IRAS-CO associations as suggested by the much larger high $N_{\rm E}$ tail seen in our sample. However it is also clear that there is a substantial population of previously unexamined low $N_{\rm E}$ value IRAS-CO sources where a true IRAS-CO association is highly probable.

What sorts of objects would be missed by a targeted survey like WB89? The IRAS sources examined in WB89 were selected to have no upper limits at 25, 60 and 100 $\mu$m and have colours typical of star-forming regions: $\log(25/12) > 0$, $0.38 < \log(60/25) < 1.88$, and $-0.77 < \log(100/60) < 0.39+0.23\log(60/25)$. The two general categories of IRAS sources that avoid these criteria are those IRAS sources with no upper limits that lie outside the WB89 colour cuts, and those IRAS sources with upper limits at 25, 60 and/or 100 $\mu$m. The next two subsections investigate these two subsets in more detail.

 

 

Table 3: IRAS-CO sources.

Number	Notes
6698	IRAS sources in the OGS region
4315	IRAS sources associated with CO emission (IRAS-CO)
2547	1 CO component
1237	2 CO components
411	3 CO components
102	4 CO components
17	5 CO components
1	6 CO components
4025	CO clouds (from BKP) associated with IRAS sources
384	IRAS-CO sources with no IRAS upper limits
541	IRAS-CO sources with one IRAS upper limit

  
3.1 Good IRAS colours

In total there are 384 IRAS-CO sources (including 214 WB89 sources) with no IRAS flux density upper limits in any of the four bands. Figure 3 shows the distribution of $N_{\rm E}$ values for the 170 IRAS-CO objects defined in this study and the 214 WB89 objects. More of the non-targeted sample is found in the high $N_{\rm E}$ bins compared with the targeted WB89 sample; 77% of the WB89 sample is found in the first four bins compared with 50% for the non-targeted sample. The number of IRAS-CO sources in the high $N_{\rm E}$bins ( $N_{\rm E} > 0.01$) is higher for the non-targeted sample (85 objects) compared with the WB89 survey (50 objects). This is due to the targeted nature of the WB89 survey which reduces the number of probably spurious IRAS-CO associations at high $N_{\rm E}$.

Figure 2: Histogram showing the distribution of $N_{\rm E}$ for IRAS-CO sources. WB89 sources are shown by the dot-dash line and non-WB89 IRAS-CO sources are indicated by the solid line. Note the comparable numbers at low $N_{\rm E}$ and the large increase in non-WB89 sources at higher $N_{\rm E}$ presumably due to false IRAS-CO associations.

Figure 3: Histogram for IRAS-CO sources having no IRAS flux density upper limits. The solid line shows the distribution of $N_{\rm E}$ for the 170 IRAS-CO sources defined in this study, and the dot-dash line shows the 214 WB89 sources.

Figure 4: Distribution of objects in the IRAS 12-25-60 $\mu$m colour plane. The boxes show the main concentrations of a variety of different astronomical objects: a - carbon stars, b - blue planetary nebulae, c - red planetary nebulae, d - T Tauri stars, e - quasars, f - Seyferts, g - blue reflection nebulae, h - red reflection nebulae, i - red galaxies, j - blue galaxies, k - embedded intermediate-mass stars, l - H II regions. a-j from Walker et al. (1989); k from Kerton (2002), l from Hughes & MacLeod (1989).

Since we have four well-defined IRAS fluxes for these objects the IRAS colours can be used to make a rough identification of the types of objects. For reference, Fig. 4 shows the main regions of the IRAS 12-25-60 $\mu$m colour-colour plane occupied by various astronomical objects. As one can see in most cases the identification cannot be exact since there can be substantial overlap of objects at a given position of the plane. In Figs. 5 and 6 we show the positions in the IRAS 12-25-60 $\mu$m colour-colour plane of the non-targeted sample of 170 IRAS-CO sources for the various $N_{\rm E}$ bins, along with the applicable WB89 colour constraints shown as solid lines. There is a population of objects, especially in the low $N_{\rm E}$ bins, that lies in and around the WB89 colour criteria. These low $N_{\rm E}$ objects are almost certainly associated with star-forming regions. As $N_{\rm E}$ increases more of the objects move away from the star-forming region area the colour-colour plane.

Figure 5: IRAS 12-25-60 $\mu$m colour plane. IRAS-CO sources with no IRAS flux density upper limits and not included in the WB89 study are plotted for the four lowest $N_{\rm E}$ bins. Bins are indicated by the number in the upper left of each panel: 1- $N_{\rm E} \leq 10^{-3.5}$, 2- $10^{-3.5} < N_{\rm E} \leq 10^{-3.0}$, 3- $10^{-3.0} < N_{\rm E} \leq 10^{-2.5}$ , 4- $10^{-2.5} < N_{\rm E} \leq 10^{-2.0}$. The solid lines indicate the WB89 colour criteria representative of star forming regions.

Figure 6: IRAS 12-25-60 $\mu$m colour plane. As Fig. 5 for the four highest $N_{\rm E}$ bins. Bins are indicated by the number in the upper left of each panel: 5- $10^{-2.0} < N_{\rm E} \leq 10^{-1.5}$, 6- $10^{-1.5} < N_{\rm E} \leq 10^{-1.0}$, 7- $10^{-1.0} < N_{\rm E} \leq 10^{-0.5}$, 8- $10^{-0.5}< N_{\rm E}$.

Since the non-targeted sample was of a manageable size we searched the SIMBAD database for any known cross-identifications along with any previous observations and/or detections of the IRAS sources with other tracers. The results of this search are summarized in Table 4. In the two lowest $N_{\rm E}$ bins the majority of the objects are either unstudied or poorly studied, however the two known objects are both Galactic - the H II region GLMP 1072 (Garcia-Lario et al. 1997) and the PP1 nebula (Parsamian & Petrosian 1979). As you move to higher $N_{\rm E}$ values one starts to see more stellar sources entering the sample - these are in most cases spurious associations, as will be discussed below, but there are also a few known Galactic H II regions in the sample confirming that some true IRAS-CO associations can be found at the $N_{\rm E} \sim 0.01$ level. It is at this level that the first known extragalactic sources start to appear. If we just take the objects in the two lowest bins ( $N_{\rm E} \leq 10^{-3}$) as being likely star-forming regions (because of their low $N_{\rm E}$ and the fact their IR colours are very close to the WB89 colours) we obtain a sample of 18 new objects, an increase of $\sim$21% over the 84 WB89 sources found in the same two bins. Alternatively, since most of the non-stellar known Galactic objects are associated with star forming regions, for each bin we can use the ratio of the number of known Galactic objects:total number of known objects to select the same fraction of unknown objects as likely star forming regions. Doing this we obtain a sample of 55 objects (assuming the $\rm ratio=1$ for both of the lowest $N_{\rm E}$ bins), an increase of $\sim$22% over the entire WB89 sample. We note that while such an analysis does not identify the star-forming regions in question explicitly it does single out a subsample of IRAS sources where star forming regions are likely to be discovered and as such can provide a good starting point for further studies.

To gain further insight into the range of the quality of the IRAS-CO associations spanned by this sample we examined the contents of the highest and lowest $N_{\rm E}$ bins in more detail (see Table 5). For the highest bin (number 8, see Fig. 7) there is one unambiguous random IRAS-CO association - IRAS 02381+5923 (topmost point in the Bin 8 panel of Fig. 6) is emission associated with the galaxy Maffei 2. Inspection of the full catalogue entry for this object shows that the IRAS-CO association is indeed very poor - the "associated'' CO is a local cloud ( $V_{\rm lsr} = -2.3$) and the offset between the IRAS coordinates and the CO $T_{\rm p}$ position is $17~\farcm3$.

Three of the IRAS-CO sources in Bin 8 are stars, IRAS 02473+5738, IRAS 03008+5637 and IRAS 22197+6028 (the three leftmost points in the Bin 8 panel of Fig. 6). This fact does not immediately mean that the IRAS-CO association is spurious as CO emission has been observed towards AGB stars where it originates from an expanding circumstellar shell of material (Kwok 2000). However, the CO lines associated with AGB stars are observed to be very broad $\sim$30 km s^-1 due to the expansion motions and, since they originate in a circumstellar shell, very good positional correspondence between the CO peak position and the IRAS source is expected. IRAS 02473+5738 is a M2Iab star with a measured radial velocity of -30.9 km s^-1. In this case the CO association is clearly random as the radial velocity of the associated CO cloud is only -12.2 km s^-1. In addition the CO cloud (BKP 4486) has a linewidth of only 1.54 km s^-1. IRAS 03008+5637 is a known M9 AGB star. It has an "E-type'' IRAS-LRS spectrum where the 9.7 $\mu$m silicate dust feature is in emission thus implying the star is an oxygen-rich AGB with a relatively optically thin circumstellar envelope (Kwok et al. 1997). The cloud associated with IRAS 03008+5637, BKP 12070, has a linewidth of only 1.78 km s^-1 and thus is also a spurious association. Finally for IRAS 22197+6028 the linewidth of the associated CO cloud is again too narrow (3.2 km s^-1) to be a believable CO association. We conclude that all of the stellar CO associations here are spurious.

 

 

Table 4: Good IRAS colours - SIMBAD results.

$N_{\rm E}$ Bina		Known Objects			Unknown Objects
Low	High	Starb	Galacticc	Extragal.	No det.d	Not Studied
	10-3.5	0	0	0	3	3
10-3.5	10-3.0	0	2	0	5	7
10-3.0	10-2.5	2	6	1	11	10
10-2.5	10-2.0	4	5	0	11	15
10-2.0	10-1.5	4	3	1	6	13
10-1.5	10-1.0	13	2	0	4	13
10-1.0	10-0.5	10	1	0	6	3
10-0.5		3	0	1	1	1
All Bins		36	19	3	47	65

^a Bins cover Low $< N_{\rm E} \leq$ High.

^b Star refers primarily to AGB stars.

^c E.g., H II regions, reflection nebulae.

^d Examined for other gas tracers but no detections made.

Figure 7: Maps of CO emission for the IRAS-CO sources having no flux density upper limits and $N_{\rm E} > 10^{-0.5}$ (Bin 8). The greyscale shows CO emission (K) in the peak channel (the $V_{\rm lsr}$ (km s-1) for the channel is indicated in parentheses above each panel). The white crosses indicate the position of the IRAS source identified above each panel.

The remaining two sources, IRAS 00040+6645 and IRAS 03118+6058, are more curious. IRAS 03118+6058 (lowest point in the Bin 8 panel of Fig. 6) was examined by Wouterloot et al. (1993) for H₂O, OH, CH₃OH emission with no detections. There is no 21 cm continuum emission visible in CGPS images of the region. DSS images of the object show that there is nothing exactly at the position of the IRAS source but the carbon star Kiso C5- 65 (Maehara & Soyano 1991) is just within the error ellipse of the position of IRAS 03118+6058 ( $19''\times 3''$ at 68 $\hbox{$^\circ$ }$). Based upon the highly negative $\log(25/12)$ colour and the proximity of a known carbon star, we conclude that IRAS 03118+6058 is the carbon star Kiso C5- 65 and that the IRAS-CO association is spurious. Finally, IRAS 00040+6645 falls within the WB89 12-25-60 $\mu$m colour criteria and has associated extended Midcourse Space Experiment (MSX, Price et al. 2001) Band A (8.3 $\mu$m) emission (see Fig. 8). The associated CO cloud has $V_{\rm lsr} \sim -13$ km s^-1 which is consistent with it being associated with the nearby Sh 2-171 region (Yang & Fukui 1992). We conclude this object is a photodissociation region (PDR) found at the edge of a CO cloud related to Sh 2-171. This object is a good example of the "type 2'' error (a false high $N_{\rm E}$) discussed previously in Sect. 2. As is clear in Fig. 8 the associated CO cloud has peaks. The contrast between the two peaks is very low however and the algorithm did not break the cloud into separate structures. Since the IRAS source position is then compared to the more distant maximum peak of the cloud a large $N_{\rm E}$ value results. We expect this sort of error to occur primarily in some of the local, more extended, clouds. At larger distances a cloud such as this one would be a much more compact object and the $\delta r$ would be significantly smaller.

 

 

Table 5: Good IRAS colours - $N_{\rm E} \leq 10^{-3.5}$ and $N_{\rm E} > 10^{-0.5}$ sources.

	IRAS	Notes
$N_{\rm E} \leq 10^{-3.5}$	03062+5742	0 references
	03083+5618	WBF93 - no detection
	22451+5906	WBF93, WWH88, WW86 - no det.
	22460+6341	0 references
	23089+5914	WWH88, WW86 - no detection
	23369+6142	0 references
$N_{\rm E} > 10^{-0.5}$	00040+6645	0 references , Kiso C5-65
	02381+5923	Maffei 2 galaxy
	02473+5738	HD 237010 M2Iab star
	03118+6058	WBF93 - no detection, PDR
	03008+5637	M9 AGB star
	22197+6028	V662 Cep - S star
WBF93	Wouterloot et al. (1993) - H2O, OH, CH3OH
WWH88	Wouterloot et al. (1988) - NH3
WW86	Wouterloot & Walmsley (1986) - H2O.

In contrast, now consider the six objects found in the low $N_{\rm E}$bin (see Fig. 9). In this case the SIMBAD search reveals no other known cross-identifications. Three of the sources have not been investigated before and the other three have been examined for a variety of high density gas tracers but with no detections. Inspection of the IRAS colour-colour plots show that all of these objects are very close to the original WB89 colour criteria and just missed being included in their sample. Given their position in the colour-colour plane and the very good association with CO they most likely represent a sample of star-forming regions. The lack of detections in the high density gas tracers suggest they may be slightly evolved regions. Four of the objects, IRAS 22451+5906, IRAS 22460+6341, IRAS 23089+5914, and IRAS 03083+5618 have been identified as likely embedded intermediate-mass stars (Kerton 2002).

Figure 8: IRAS 00040+6645 and Sh 2-171. Greyscale shows 21-cm radio continuum emission (K) from the eastern portion of Sh 2-171. MSX Band A (8.3 $\mu$m) emission (black contours at 1.5 and $2.5\times 10^{-6}$ W m-2 sr-1) and integrated CO emission (white contours at 5 and 7 K km s-1; integrated from -9.3 to -15.1 km s-1 ( $V_{\rm lsr}$)) are overlayed. The position of IRAS 00040+6645 is indicated by the cross, and the centroid of the associated CO cloud is indicated by the asterisk. The location of the IRAS source at the edge of the CO cloud and the H II region suggests that it is a PDR.

  
3.2 Poor IRAS colours

The remainder of the sample (3931 IRAS-CO sources) have a flux density upper limit in at least one of the IRAS bands, this includes 40 sources with upper limits in the IRAS 12 $\mu$m band included in the WB89 sample. Table 6 summarizes the $N_{\rm E}$ distribution of this sample.

Figure 9: IRAS-CO sources with no IRAS flux density upper limits and $N_{\rm E} \leq 10^{-3.5}$. The white crosses indicate the position of each IRAS source. The greyscale shows the CO emission (K) in the peak channel, and the velocity of the channel ( $V_{\rm lsr}$ (km s-1)) is shown in parentheses above each panel.

It is clear from looking at the number of objects in the lower $N_{\rm E}$bins that there is potentially a very large number of interesting objects in this sample. For this paper we will limit our analysis to the 68 sources in the $N_{\rm E} \leq 10^{-3.5}$ bin. We first queried the SIMBAD database for these 68 sources. Seven of the IRAS sources were previously known objects - three H II regions, one reflection nebula, and three star-forming cores. Twenty-one of the objects had been included in other surveys for dense gas tracers but with no detections, and 40 of the IRAS sources had not been studied at all beyond their catalogue identification.

 

 

Table 6: Poor IRAS colours - $N_{\rm E}$ distribution.

$N_{\rm E}$ Bina
Low	High	Non-Targeted	WB89	Total
	10-3.5	68	0	68
10-3.5	10-3.0	110	3	113
10-3.0	10-2.5	272	5	277
10-2.5	10-2.0	468	5	473
10-2.0	10-1.5	739	6	745
10-1.5	10-1.0	988	11	999
10-1.0	10-0.5	992	10	1002
10-0.5		254	0	254
All Bins		3891	40	3931

^a Bins cover Low $< N_{\rm E} \leq$ High.

Because of the upper limits on the IRAS flux densities one cannot use the IRAS colour-colour plane to effectively look at the sources. However if we limit ourselves only to those objects with a single flux density upper limit then some progress can be made. Objects with upper limits at either 12 or 100 $\mu$m can be unambiguously placed on the 25-60-100 or 12-25-60 colour plane respectively. On the other colour plane one colour can be fixed while the other will be a limit. The situation is somewhat more poor for sources with only upper limits at either 25 or 60 $\mu$m. In this case the source will have a limit in both colors and the allowed position of the point will be along a diagonal line in the colour-colour plane - not ideal but at least certain regions of the plane can be eliminated.

For the 68 very low $N_{\rm E}$ sources, 19 have a single upper limit: 3 at 12 $\mu$m, 4 at 60 $\mu$m, and 12 at 100 $\mu$m. Table 7 shows details of the objects based upon the results of a SIMBAD query and Fig. 10 shows this subsample of IRAS-CO sources in both IRAS colour planes. Examining the 12-25-60 $\mu$m plane in more detail we see that 11 of the 12 sources with good colours in this plane (i.e., with 100 $\mu$m upper limits) lie within the WB89 region and the other source is just outside of the WB89 region. Eight of the objects are known to be associated with star forming regions and we think it is highly likely that all twelve of these objects are associated with star-forming regions. One of the 3 12 $\mu$m upper limit sources lies within the WB89 region and all of these sources have $\log~(60/25)$ colours consistent with the WB89 criteria. Given their close association with CO it is likely these are also all associated with star-forming regions. The sources with 60 $\mu$m upper limits are slightly harder to interpret. Unfortunately there are no known cross-identifications for these objects, but all of them could be associated with star formation if the upper limits are not too far off the true flux densities.

 

 

Table 7: $N_{\rm E} \leq 10^{-3.5}$, single upper limit sources.

IRAS	Limita	Notesb
00206+6555	100	UCHII region - BNM96 CS(2-1) det.
02157+6053	100	K01 - submm sources
02227+6127	100	K02 - embedded B star
02570+6028	100	CHS00 - embedded cluster
03054+6407	100
22111+5845	100	S00 - no det.
22163+5555	100	K02 - embedded B star
22333+5744	100	K02 - embedded B star
23033+5951	100	UCHII region - BNM96 CS(2-1) det.
23140+6042	100	WWH88, WW86 - no det
23377+6059	100
23483+6325	100	K02 - embedded B star
22510+6153	60	WBF93, WWH88, WW86 - no det.
22521+6205	60	WBF93, WWH88, WW86 - no det.
00153+6532	60
03116+5951	60
00510+6550	12
02366+5845	12	WBF93 no det.
02499+5752	12

^a IRAS band ($\mu$m) with flux density upper limit.

^b Blank indicates unstudied object;

BNM96 - Bronfman et al. (1996),

K01 - Kerton et al. (2001),

K02 - Kerton (2002),

S00 - Szymczak et al. (2000),

other abbreviations as in Table 5.

The 25-60-100 $\mu$m colour plane confirms these observations. There is not much of a surprise with the 100 $\mu$m upper limit sources since we already knew that they had $\log~(60/25)$ colours consistent with WB89. We see that all of these sources could lie within or close to the WB89 region, again depending upon the true value of the 100 $\mu$m flux density. One of the 3 12 $\mu$m upper limit sources lies inside of the WB89 region and the other two lie just outside of the region and could again easily be star forming regions. As with the other colour plane the identity of the 60 $\mu$m upper limit sources depends critically upon how far off the upper limit is from the true 60 $\mu$m flux density.

Figure 10: IRAS colour-colour planes showing IRAS-CO sources with $N_{\rm E} \leq 10^{-3.5}$ and one IRAS flux density upper limit. The solid lines show the WB89 colour criteria representative of star forming regions. Arrows indicate the effect of the single flux density upper limit on the IRAS colour.

  
3.3 IRAS sources with no CO associations

As mentioned in Sect. 2 there are 2383 IRAS sources in the OGS that are not associated with any CO emission as accounted for in the BKP catalogue. Because of the lack of associated CO we expect that most of these sources will either be stars or extragalactic sources. This idea is supported when one examines the average flux densities of the associated and non-associated IRAS sample (see Table 8). The average flux densities for the non-associated IRAS sample are significantly lower than the IRAS-CO sample.

Of the non-associated IRAS sources, only 41 have good IRAS colours in all four IRAS bands and thus can be placed on the IRAS colour plane unambiguously. Figure 11 shows these 41 objects on the 12-25-60 $\mu$m IRAS colour plane. A search of SIMBAD reveals that 15 of the objects are stars, and 2 are known planetary nebulae. Also, even though a number of these objects occupy regions of the colour-colour plane consistent with star forming regions, the SIMBAD search reveals no known star forming regions in this sample. These objects are most likely galaxies or red reflection nebulae (see Fig. 4). For the former we do not expect to observe associated CO in the $V_{\rm lsr}$ range of the survey while the latter objects are fairly evolved objects where a CO association is not highly probable. In this region of the colour-colour plane the lack of associated CO is thus a very useful means to remove the degeneracy present in the IRAS colour identification. Section 4.1 provides an example of using the presence or lack of a CO association to clarify the true nature of an object.

 

 

Table 8: Average IRAS flux density comparison.

	Average Flux Density
IRAS Band	IRAS-CO Samplea	No CO Associationb
<F12>	$2.42\pm0.28^{c}$	$2.14\pm0.04$
<F25>	$5.52\pm1.4$	$1.59\pm0.03$
<F60>	$40.79\pm8.05$	$2.39\pm0.05$
<F100>	$114.90\pm16.59$	$21.14\pm0.4$

^a 4315 IRAS sources.

^b 2383 IRAS sources.

^c standard error of the mean.

Figure 11: IRAS sources with no flux density upper limits and no CO associations. Many of the objects plotted in this 12-25-60 $\mu$m colour plane are obviously stellar objects where one does not expect associated CO. For objects near and within the WB89 colour criteria (shown as the solid lines) the lack of CO emission indicates that the objects are most likely not star forming regions in our Galaxy.