next previous
Up: A radio catalog of wavelengths


Subsections

3 The Synthetic Catalog at 2.7 GHz

The final task in bringing together the wide-spread radio data on HII regions is to construct a readily accessible catalog summarizing the basic information on each of the 1442 sources covered in the comprehensive data base of the Master Catalog. This Synthetic Catalog gives the best available data on flux density, angular diameter and, where available, the line velocity.

2.7 GHz was chosen as the base frequency for the Synthetic Catalog; it lies in the middle of the frequency range of those catalogs containing most of the data, namely 1.4, 2.7 and 5 GHz. Since there is not complete source coverage at any one frequency, we derive the flux density at 2.7 GHz from the observed frequencies for all the sources. We now describe how best-estimate values of the flux density, diameter and velocity with corresponding error estimates are derived from the Master Catalog.

3.1 Flux density estimates at 2.7 GHz

We estimate the flux density at 2.7 GHz in four different ways:

1.
we use the flux density measures at 2.7 GHz when these are available;
2.
if there are no flux density measures at 2.7 GHz and there are available measures at only one other frequency, we extrapolate these values to 2.7 GHz with the theoretical spectral index $\alpha =-0.1$ ( $S \propto \nu^{\alpha}$) which is typical of thermal bremsstrahlung emission in a thin plasma;
3.
if there are flux density measures only at frequencies $\ge$14.7 GHz, we first select the lowest among the given frequencies - averaging over 14.7 and 15 GHz when both are available - and then extrapolate to 2.7 GHz with $\alpha =-0.1$;
4.
if there are only flux density measures at frequencies other than 2.7 GHz and at least one measure at $\nu \le 5$ GHz, we follow these lines: we exclude the values at 3.9, 14.7, 15 and 86 GHz, we take the weighted average between 4.8 and 5 GHz if both measures are available, we interpolate the remaining values to 2.7 GHz (or extrapolate with $\alpha =-0.1$ if only one frequency measure is left).

We consider the data at 3.9, 14.7, 15 and 86 GHz less reliable for the following reasons: the 3.9 GHz measure may be significantly affected by the ellipticity of the antenna pattern; at 14.7 and 15 GHz the observing resolution is much higher than the typical reference catalogs resolution; at 86 GHz, in addition to the higher resolution, dust emission may contribute largely to the observed flux. Moreover, we prefer to interpolate between 1.4 and 5 GHz rather than extrapolate to 2.7 GHz directly from only one of these two frequencies because we have no information on the frequency location of the free-free knee.

In cases when multiple observations are available at the same frequency, a weighted mean flux density is computed, using the errors discussed below.


  \begin{figure}
\par\includegraphics[width=15cm,clip]{aah3866f1.eps}\end{figure} Figure 1: Results of the flux error estimate for: earlier Parkes survey at 2.7 GHz (top left); Mezger $\&$ Henderson (1967) (top right); Wink et al. (1983) (bottom center). Plotted points in each panel correspond to the $\sigma (S_{*})$ distribution (where $\sigma $ is defined as 100(S* - S)/S). Overlaid - solid line - is the best fit function. See Sect. 3.1 for more details on the quantities.

For the majority of the sources in the reference catalogs, flux density errors are quoted and these are used in deriving the flux density and its error in the Synthetic Catalog. In the catalogs where no errors are given, an estimate of the error is made by comparing the flux S* from a catalog without errors with the flux S from a catalog (or some catalogs) giving errors. In particular, we compute the relative (%) dispersion, $\sigma = 100 (S_{*} - S)/S$, and try to fit the resulting distribution in the $S_{*} - \sigma$ plane with a constant, $\sigma = a$ , a linear, $\sigma = a + b S_{*}$, and a quadratic, $\sigma = a + b S_{*} + c S_{*}^{2}$, dependence of $\sigma $ on S*. In these equations the fluxes are in Jy. Before fitting the $\sigma (S_{*})$ distribution, we remove the relatively small number of points with a dispersion $\simeq$130%. Although we have considered also fits with a constant, linear, and quadratic dependence of $\rm {log} \sigma$ on ${\rm log} S_{*}$, we prefer to use the distributions recovered from the fit carried out with linear variables, since they give, as expected, typical values of $\sigma (S_{*})$ larger than those obtained in the case of logarithmic variables and so giving more conservative error estimates. Figure 1 shows the results of the procedure we have described.

An example of the estimates of errors is given by the early Parkes 2.7 GHz catalogs. We compare the list of Parkes sources to the reference catalogs at the same frequency - 2.7 GHz - (namely, Altenhoff et al. 1970; F $\ddot{\rm u}$rst et al. 1987; Reich et al. 1986; Wendker 1970), in order to retrieve the subset of Parkes sources whose flux has been quoted with an estimated error. In particular, the comparison retrieves 105 sources in common with the Altenhoff et al. (1970) catalog. The scatter $\sigma $ in the flux density differences, quite well fitted by the law $\sigma \simeq 33.5{-}0.1\times S$, is comparable to the errors of 10-30% of the Altenhoff et al. catalog alone.

Similar comparisons have been carried out for the Mezger $\&$ Henderson (1967) data and for Wink et al. (1983) data. However, as for the Mezger $\&$ Henderson data, the comparison with other single catalogs does not provide a significant statistical subset of sources. Therefore, we estimate the error by considering simultaneously the sources from the Mezger $\&$ Henderson catalog and from all the other catalogs at 5 GHz (which leads to consider altogether 38 common sources). For the Wink et al. (1983) catalog, we derive an error estimate from the comparison with the Wink et al. (1982) catalog for which we have 21 common sources.

After having estimated the errors on the flux density measures for all catalogs, we compute the errors on the flux densities of the Synthetic Catalog at 2.7 GHz. When an extrapolation with a spectral index $\alpha =-0.1$ from a measure at a single frequency different from 2.7 GHz is applied, we use the standard error propagation rules. No error on $\alpha$ is included in the error propagation, for simplicity.

Table 3 summarizes the errors quoted in the original catalogs and those estimated in this work.

3.2 Angular diameter

We discuss here the assignment of an angular diameter with its associated error for every source in the Synthetic Catalog. The published diameter data are not as comprehensive as the flux density data described in Sect. 3.1; nevertheless we will assign a value of the diameter and an error for each source. For 42% of the sources (derived from the Parkes catalogs and those of Mezger $\&$ Henderson 1967; Wendker 1970; Wink et al. 1982; Wink et al. 1983; Berlin et al. 1985) there are no listed diameter errors. For 14% of the sources no diameter is given; for these we derive an indicative diameter.

We assign a diameter for sources in the Synthetic Catalog as follows:

1.
when 2.7 GHz diameters are given, we use these values (possibly their weighted average);
2.
if there is at least one diameter measure at 5 GHz or at 4.8 GHz, we neglect possible measurements at other frequencies and use this value or, alternatively, the weighted average of the diameter measures at 4.8 and 5 GHz when available, using the errors discussed below;
3.
if diameter data are given only at frequencies other than $\sim $4.8-5 GHz, we use the weighted average of the available frequencies, after previous exclusion of the measures at $\nu \sim 14.7$ GHz, using the errors discussed below;
4.
if no diameter data is given at frequencies lower than $\sim $14.7 GHz, we use the available measure at 14.7 GHz or at 15 GHz or, alternatively, their weighted average when both values are provided using the errors discussed below (we remember that no diameter data are available at 86 GHz).


  \begin{figure}
\par\includegraphics[width=8.5cm, clip]{aah3866f2.eps}\end{figure} Figure 2: Correlation between flux densities at 2.7 GHz and angular diameters observed at the same frequency. The solid line represents the best-fit for the plotted distribution.

Values at 14.7 and 15 GHz are preferably excluded because of the much higher resolution at these frequencies than the typical reference catalog resolution. In each case, when multiple observations are available at the same frequency, a weighted mean is computed, using the errors discussed below.

Finally, we consider the assignment of a diameter to the 14% of HII regions in the Synthetic Catalog which have no quoted diameter in the reference catalogs. It is possible to derive a first-order indicative diameter by noting that the flux density and the diameter of HII regions are weakly correlated, as shown in Fig. 2 which includes all the sources at 2.7 GHz with measured diameters. The best fit to the data gives as indicative diameter $\theta = 2.25 \times S$ with a typical error of a factor of 2-3. Each source with an indicative diameter is annotated in the Synthetic Catalog; such diameter data clearly should not be used in astrophysical analysis of the catalog.

Following the same strategy as in Sect. 3.1, where a catalog has no quoted diameter errors we estimate an overall error for that catalog by comparing the observed diameters of that catalog with another catalog having an adequate number of sources in common. However, with respect to the flux density case, we can now relax the strict requirement of comparing only different datasets at the same frequency. As shown in Fig. 3, there is no significant frequency dependence in the measured diameters at 2.7 GHz (the Parkes survey) and at 5 GHz (Altenhoff et al. 1970; Mezger $\&$ Henderson 1967; Reifenstein et al. 1970). On the other hand, provided that an adequate number of sources in common is retrieved, we prefer to consider in the comparison surveys at the same frequency and with similar angular resolution. Therefore, let $\Theta_{*}$ ($\Theta$) be the source diameter measure in the catalog without quoted errors (with quoted errors). We compute the relative (%) dispersion $\sigma $ of the diameter measures, $\sigma = 100 (\Theta_{*} - \Theta)/\Theta$, and try to fit the resulting distribution in the $\Theta_{*} - \sigma$ plane again with a constant, linear, and quadratic, dependence of $\sigma $ on $\Theta_{*}$ (or of $\rm {log} \sigma$ on ${\rm log} \Theta_{*}$; in this case the fit error estimates are again less conservative) and after removing the points with a dispersion $\simeq$130%. In these equations the angular diameters are in arcmin.


  \begin{figure}
\par\includegraphics[width=9cm,clip]{aah3866f3.eps}\end{figure} Figure 3: Diameter measures at 2.7 GHz (y-axis) versus diameter measures and at 5 GHz (x-axis). It is evident the good agreement between the two datasets. The best fit function is: $\Theta_{\rm 2.7~GHz} = 1.17+1.17 \times \Theta_{\rm 5~GHz}$. Data are from: earlier Parkes survey (2.7 GHz data); Altenhoff et al. (1970)a, Mezger $\&$ Henderson (1967), Reifenstein et al. (1970) (5 GHz data).


  \begin{figure}
\par\includegraphics[width=12.5cm,clip]{aah3866f4.eps} %
\end{figure} Figure 4: Results of the diameter error estimate for: Altenhoff et al. (1970)a (top left); Altenhoff et al. (1970)b (top right); earlier Parkes survey at 2.7 GHz (left, second row); Caswell $\&$ Haynes (1987) (right, second row); Felli $\&$ Churchwell (1972) (left, third row); Wendker (1970) (right, third row); Wink et al. (1982) (bottom left); Wink et al. (1983) (bottom left). Plotted points in each panel correspond to the $\sigma (\Theta _{*})$ distribution (where $\sigma $ is defined as $100(\Theta _{*} - \Theta )/\Theta $). Overlaid - solid line - is the best fit function. See Sect. 3.2 for more details on the quantities.

Figure 4 shows the results of the error estimates. In particular, we consider the comparison between the following datasets (the first one is that for which we are estimating the errors): Altenhoff et al. (1970)a vs. Reifenstein et al. (1970) (48 common sources); Altenhoff et al. (1970)b vs. Kuchar $\&$ Clark (1997) (63 common sources); Caswell $\&$ Haynes (1987) vs. Wilson et al. (1970) (79 common sources); early Parkes 2.7 GHz Survey vs. Reifenstein et al. (1970) (31 common sources); Felli $\&$ Churchwell (1972) vs. Kuchar $\&$ Clark (1997) (21 common sources); Wendker (1970) vs. Reifenstein et al. (1970) (8 common sources); Wink et al. (1983) vs. Downes (1980) (59 common sources); Wink et al. (1982) vs. Altenhoff et al. (1979) (53 common sources).

As for the Altenhoff et al. (1970) catalog, the data have been split because the diameters have been measured in two different ways: they have been either taken from a survey of Galactic sources made at 5 GHz with the 140-ft antenna - beamwidth 6' (W. J. Altenhoff, P. G. Mezger and J. Schraml, private communication) - or they have been measured directly from contour maps. We will annotate the set of sources for which the size was measured in the 5 GHz survey Altenhoff et al. (1970)a and the remaining sources Altenhoff et al. (1970)b. For the Berlin et al. (1985) catalog, since none of the comparisons with other catalogs retrieves a significant number of common sources, we derive the error from the mean of the dispersions $\sigma $ given by each comparison. Table 3 summarizes the estimated and quoted errors used in calculating the diameters in the Synthetic Catalog.

3.3 Content of the Synthetic Catalog

The Synthetic Catalog summarizes the known radio frequency information on 1442 Galactic HII regions. It contains the position, flux density, diameter data for each HII region, supplemented by velocity data where available. To those HII region with no published diameter data, an indicative diameter is given (marked by **) on the basis of the flux-size correlation in Fig. 2. For sake of clarity, the first ten lines of the Synthetic Catalog are reported in Table 2. The line velocity value that we quote in Col. 9 is the weighted average of the available data (see Sect. 2.4 for details). Although the original measures can be, for the same source, at different frequencies, the weighted average of these data is a meaningful quantity and provides a useful first-sight kinematic indication. Since the line velocity is an effect of the Galactic rotation motion, it does not strongly depend on the frequency of observation.


 
Table 2: A selection of sources from the Synthetic Catalog is shown. The arrangement of column is as follows:
- Column 1: source-numbering (records from 1 to 1442)
- Columns 2-3: galactic coordinates, l and b
- Columns 4-5: celestial coordinates (epoch 2000)
- Columns 6-7: derived 2.7 GHz flux density and 1-$\sigma $ error (Jy)
- Columns 8-9: derived angular diameter and 1-$\sigma $ error (arcmin)
- Columns 10-11: velocity relative to the LSR and 1-$\sigma $ error (km s-1)
- Column 12: notes on individual sources (see Sect. 2.4 and Appendix 1 for details).
N l b RA DEC S $\sigma_{S}$ $\theta$ $\sigma_{\theta}$ $V_{\rm LSR}$ $\sigma_{V_{\rm LSR}}$ Notes
1 0.1 0.0 17 45 51.3 -28 51 08 230.0 24.1 5.9 0.5 -27.4 4.0 W24
2 0.2 -0.1 17 46 29.0 -28 49 07 209.4 10.5 10.7 0.5 24.5 3.5  
3 0.2 -0.0 17 46 05.6 -28 46 00 177.6 38.1 9.2 0.5 -12.7 3.5 W24
4 0.3 -0.5 17 48 17.0 -28 56 25 2.5 0.7 2.7 1.7 20.0 4.9 C, S
5 0.4 -0.8 17 49 41.7 -29 00 33 8.0 2.6 7.0 2.3 20.0 4.9 C, S
6 0.4 -0.5 17 48 31.1 -28 51 17 4.1 1.0 3.9 1.8 24.0 4.9 C, S
7 0.5 -0.7 17 49 32.2 -28 52 19 2.9 0.9 2.3 1.4 17.5 4.9 C
8 0.5 -0.1 17 47 11.6 -28 33 44 28.3 4.0 3.3 1.4 45.8 5.0 S, X
9 0.5 0.0 17 46 48.2 -28 30 37 40.3 8.6 4.8 0.3 47.1 2.0  
10 0.6 -0.9 17 50 33.3 -28 53 19 2.5 0.8 2.4 1.3 15.0 4.9 S, S21, RCW142



next previous
Up: A radio catalog of wavelengths

Copyright ESO 2003