Astron. Astrophys. 364, L93-L96 (2000)

3. Gamma-ray sources in the Gould Belt

The distribution on the sky of a subset of the unidentified EGRET sources is strongly reminiscent of the tilted Gould Belt. The subset includes all persistent sources at from the 3rd EGRET catalogue, i.e. those detected with a significance ( at ) using the cumulative data from April 1991 to October 1995. Sources listed as likely artifacts were discarded. The 67 sources thus selected are displayed in Fig. 1. 80% of them show no time variability, the others are only moderately variable (Tompkins 1999). Strong biases due to the instrument exposure and the bright interstellar background does influence their apparent spatial distribution. To assess their correlation with various structures, a maximum-likelihood test was applied that takes these biases into account (Grenier 1997). An important degree of freedom is the unknown luminosity function of the parent sources. The flatter the function, the sharper the longitude and latitude profiles of any Galactic population because more sources remain visible to large distances. This effect being quite strong, was kept as a free parameter over a wide range of values: . EGRET exposure maps above 100 MeV for the 4.5-year survey (Hartman et al. 1999), the observed isotropic background of ( cm^-2 s^-1 sr^-1, and the interstellar background model (Hunter et al. 1997) of the EGRET survey were used to model the source visibility in by bins across the sky.

Fig. 1. visibility map, in Galactic coordinates, from the best fit "Iso+Belt" model, the Belt being traced by its young massive stars. The colour codes the number of sources detectable per bin in the EGRET survey. The 67 persistent, unidentified EGRET sources at are marked as white dots which closely follow the curved lane of Belt stars.

To allow for possible extragalactic sources among the unidentified ones, linear combinations of an isotropic and a Galactic component were tested (and nicknamed as in Table 1). The choice of Galactic distributions cover the various sites likely to display sources at mid latitudes: 1) sources uniformly distributed in a spherical Galactic halo, 20 kpc in radius; 2) sources uniformly distributed in longitude in the "local Galactic disc" with any exponential or gaussian scale height; 3) sources spread in a "thick Galaxy" with a gaussian radial scale length of 9.3 kpc and an exponential scale height of 0.4 kpc, typical of radio pulsars (Lyne et al. 1985); 4) sources distributed in the interstellar medium as mapped by the NHN(HI+2H₂) column-densities (Grenier 1997); 5) sources in the Gould Belt as traced by the column-densities of stars with spectral type B3, fitted from the Yale Bright Star catalogue data according to the function ), with the resulting width and median latitude of the star distribution displayed in Fig. 1. Distributions 4) and 5) provide independent ways to trace the Belt through its young stars as well as its clouds since the latter dominate the interstellar maps at medium latitudes.

Table 1. max-likelihood results and their 1 errors for selected models against the 67 persistent unidentified sources

The reliability of the analysis was checked by applying it to the 67 active galactic nuclei discovered by EGRET and spread over the sky. For them, an isotropic distribution indeed yields the best fit and no Galactic component, however large in scale height, is detected: the maximum log-likelihood value remains constant within 0.1 for all the models tested in Table 1. 67 8 sources are attributed to the isotropic component, 0 6 to the local Gal., thick Gal., NH, or Belt components. The 9 16 sources attributed to a halo component are not significant. The luminosity index of the parent population is .

Results obtained for the 67 unidentified sources are displayed in Table 1. Modelled source counts in the isotropic and anisotropic components, and respectively, were summed over the given latitude intervals. The probability is determined from the log-likelihood increase between two models, with left free. It is the chance probability that a random fluctuation from model 1 yields as good a fit as model 2. The dramatic improvement in the fit for any Galactic model over a pure isotropic population shows that over 50 sources have a Galactic origin. The fit improves very significantly from a spherical halo to a flatter Galactic distribution, , and even more so with the tilted geometry of the Belt clouds or stars, . The good quality of the BELT fit is illustrated in Fig. 1 & Fig. 2. The fact that the LOC and GAL models yield equally good fits implies a large fraction of nearby sources: no contrast in longitude is detected that would require distances of a few kpc to the sources. Observed source counts are indeed equivalent in the centre and anticentre quadrants in Fig. 1. Not only are the sources local, the significant likelihood increase between the LOC and BELT fits, with a chance probability , gives strong support to their origin in the inclined Belt system. On the other hand, the sources at were shown to be distinctly fainter and softer than those at lower latitudes and a subset of 20 were pointed out along the Belt (Gehrels et al. 2000). Together these findings provide compelling evidence that a distinct population of 20 to 40 EGRET sources belong to the Gould Belt. Based on their spatial distribution only, they could be as numerous as 40 5 at . The luminosity index of their parent population is .

Fig. 2. latitude profiles of the persistent unidentified EGRET sources (crosses), of the best fit "Iso+Belt" model (thick line), and of the isotropic contribution to this model (dotted line). The "Iso+Local Gal." model (dashed line) yields a significantly poorer fit () with systematic deviations from the data over wide latitude bands.

© European Southern Observatory (ESO) 2000

Online publication: January 29, 2001
helpdesk.link@springer.de