next previous
Up: The 1997 reference

11. Diffuse galactic light

11.1. Overview

Historically, the term Diffuse Galactic Light (DGL) denotes the diffuse component of the galactic background radiation which is produced by scattering of stellar photons by dust grains in interstellar space (Elvey & Roach 1937; Roach & Gordon 1973). This scattering process is the dominant contributor to the general interstellar extinction of starlight; thus, the DGL is most intense in directions where the dust column density and the integrated stellar emissivity are both high. This is generally the case at the lowest galactic latitudes, in all spectral regions extending from the far-ultraviolet into the near-infrared. Typically, the DGL contributes between 20% to 30% of the total integrated light from the Milky Way.

However, for the purpose of this reference we are also interested in other sources of diffuse galactic background radiation, and they will be mentioned in the following where appropriate.

11.2 Visual

No comprehensive map of the DGL for the entire sky or even a significant fraction of the sky exists at this time. Groundbased observations in the visual face the difficult requirement that airglow, zodiacal light, and integrated starlight all need to be known to very high precision (tex2html_wrap_inline15403) if the DGL is to be derived by subtraction of the above components from the total sky brightness. In addition, the problems of atmospheric extinction and atmospheric scattering (Staude 1975) need to be solved.

Observations of the DGL at visual wavelengths carried out with rocket- or satellite-borne photometers still have the same major sources of uncertainty, i.e. the integrated starlight and the zodiacal light, remain principal contributors to the measured intensity.

The best prospect for a comprehensive measurement of the DGL in the visual was offered by the Pioneer 10 probe (see the more detailed description in Sect. 10.4 (click here)), which carried out an all-sky photometric mapping in two wavebands centered near 440 nm and 640 nm from beyond the asteroid belt (R > 3 AU), where the contributions from zodiacal light are negligible (Hanner et al.\ 1974). The instantaneous field of view of the Pioneer 10 photometer was tex2html_wrap_inline15407 square, which due to spacecraft spin (12.5 s period) and finite integration time (0.2 s) was drawn into elongated effective fields-of-view of variable size depending on the look angle. Contributions due to unresolved stars begin to enter the Pioneer 10 data at tex2html_wrap_inline15409 for an average galactic latitude; thus, stars dominate the measured fluxes.

Toller (1981) derived DGL intensities from the Pioneer 10 blue data (440 nm) by subtracting integrated starlight intensities of Roach & Megill (1961) and Sharov & Lipaeva (1973) at the positions of 194 Selected Areas (Blaauw & Elvius 1965). The residuals, interpreted as the sum of DGL and extragalactic background light, are most representative in terms of sky coverage.

Figure 76: The average intensity of the DGL as a function of galactic latitude based on the analysis by Toller (1981) of Pioneer 10 photometry of 194 Selected Areas at tex2html_wrap_inline15411 nm. Error bars denoting one standard deviation of the means are a measure of the longitudinal variation of the DGL intensity

For reference purposes, several directions may be taken to estimate the intensity of the DGL at tex2html_wrap_inline15413 nm:

In Fig. 76 (click here) we present the mean galactic latitude dependence of Toller's (1981) values of the sum of DGL and extragalactic background, averaged over all galactic longitudes. The error bars, representing one standard deviation of the mean, reflect in part the real variations of the DGL intensity with galactic longitude, especially at lower latitudes.

A second avenue toward a DGL estimate can be found in ratios of DGL to total line-of-sight starlight (LOS*) intensities. In Table 39 (click here) we list the average ratios of DGL/LOS* for tex2html_wrap_inline15413 nm based on Toller's data. The use of the values in Table 39 (click here) may be advisable, if one wants to estimate the variation of DGL with galactic longitude, where large differences in LOS* may occur. Due to the strongly forward scattering nature of interstellar grains the DGL intensity generally tracks the LOS* intensity at constant latitude.


b [tex2html_wrap_inline15429]

- N [SA] tex2html_wrap_inline15433
0-|5| 19 tex2html_wrap_inline15437
|5|-|10| 11 tex2html_wrap_inline15443
|10|-|15| 28 tex2html_wrap_inline15449
|15|-|20| 15 tex2html_wrap_inline15455
|20|-|30| 29 tex2html_wrap_inline15461
|30|-|40| 22 tex2html_wrap_inline15467
|40|-|60| 41 tex2html_wrap_inline15473
|60|-|90| 29 tex2html_wrap_inline15479

Table 39: Ratio of DGL to line-of-sight starlight

A third approach toward a DGL estimate might rely on the mean correlation between DGL intensities found in Selected Areas by Toller (1981) and corresponding column densities of atomic hydrogen. Toller finds:
A good source for tex2html_wrap_inline15483 values is the Bell Lab HI survey by Stark et al. (1992). This approach is based on the fact that the dust and HI column densities are well-correlated (Bohlin et al.\ 1978) and that the DGL intensity is determined in part by the dust column density, as long as the line-of-sight is not optically thick. This third approach is therefore recommended mostly for higher galactic latitudes, or tex2html_wrap_inline15485 atoms cm-2. Estimates based on Eq. (31 (click here)) are at best good to within a factor of two, because Eq. (31 (click here)) reflects only the dependence of the DGL intensity on the dust column density and ignores the dependence on the intensity of the illuminating radiation field.

The red band (tex2html_wrap_inline15489 nm) data from Pioneer 10 have not been subjected to a DGL analysis so far for lack of suitable star count data.

The U-B and B-V colours of the DGL have been measured and have been found to be bluer than the colour of the integrated starlight, as expected from scattering by interstellar grains with scattering cross sections varying as tex2html_wrap_inline15495 in the visible region (Witt 1968; Mattila 1970). Table 40 (click here), to give an example, contains UBV colours of the DGL and of the integrated starlight in Cygnus (upper panel), respectively in Crux (lower panel).



DGL Integrated Starlight Reference
l, |b| U-B B-V U-B B-V
tex2html_wrap_inline15511, tex2html_wrap_inline15513 -0.05 +0.57 +0.07 +0.73 Witt
tex2html_wrap_inline15517, tex2html_wrap_inline15519 -0.10 +0.44 +0.00 +0.68 (1968)

tex2html_wrap_inline15523, tex2html_wrap_inline15513

-0.10 +0.50 -0.01 +0.71Mattila

Table 40: Colour of the DGL

Recently, Gordon (1997) reported the detection of extended red emission (ERE) on a galaxy-wide scale in the diffuse interstellar medium of the Milky Way Galaxy (see also Gordon et al. 1997; Gordon & Witt 1997). The ERE consists of a broad emission band (FWHM 800 Å) with a peak wavelength found in the 6500 Å\ to 8000 Å\ range, depending on environment, with a long-wavelength tail extending well into the I-band. The ERE is believed to result from a photoluminescence process in hydrogenated carbonaceous grain mantles, and it has been previously detected photometrically and spectroscopically in numerous reflection nebulae (Witt & Schild 1988; Witt & Boroson 1990), in carbon-rich planetary nebulae (Furton & Witt 1990, 1992), in HII regions (Perrin & Sivan 1992; Sivan & Perrin 1993), and in the scattered light halo of the starburst galaxy M 82 (Perrin et al. 1995). Gordon (1997) derived the galactic ERE intensity from Pioneer 10 and 11 sky photometry obtained at heliocentric distances greater than 3.3 AU, where the contribution from zodiacal light is no longer detectable (see Sect.nbsp;10.4 (click here)). The integrated star light due to stars of m > 6.5 was determined by integrating recent starcount data from the APS Catalog (Pennington et al. 1993), the HST Guide Star Catalog, and photometric catalogs on brighter stars and was subtracted from the Pioneer 10 and 11 in both the blue and red bands. The diffuse residuals consist of DGL in the blue band, and of a sum of DGL and ERE in the red band. As a result, the B - R colour of the diffuse galactic background radiation is substantially redder than that of the DGL alone. The excess ERE in the R-band can be estimated to be about equal in intensity to the R-band DGL. This ERE intensity is consistent with the measured B-R and B-I colour excesses of individual galactic cirrus filaments (Guhathakurta & Tyson 1989), found to be 0.5 - 1.0 mag and 1.5 - 2.0 mag redder, respectively, than expected for scattered disk starlight.

Quantitatively, Gordon (1997) finds the galactic ERE and the atomic hydrogen column density at intermediate and high latitudes (tex2html_wrap_inline15549) to be well-correlated, yielding an average ERE intensity of (tex2html_wrap_inline15551 H-atom-1. This correlation may therefore be used to estimate the expected ERE intensity in the R-band in different portions of the sky.

Partial linear polarization of the DGL at a level of tex2html_wrap_inline15557 is expected, and some tentative detections of this polarization have been reported by Schmidt & Leinert (1966), Weinberg (1969), Sparrow & Ney (1972), and Bandermann & Wolstencroft (1976). Both the scattering by grains partially aligned with their short axes parallel to the galactic plane and the scattering of the non-isotropic galactic radiation field by dust in the galactic plane should produce partially polarized scattered light with the electric vector perpendicular to the galactic plane when observed near tex2html_wrap_inline15559. A review of existing polarization measurements is given by Leinert (1990).

11.3. Near-infrared

The diffuse galactic background radiation in the near-infrared (near-IR) is composed of several components, each produced by different constituents of the diffuse interstellar medium by different physical processes. The most important ones are the DGL, caused by scattering of star light on larger interstellar grains; the near-IR continuum emission, caused by a non-equilibrium emission process probably associated with small carbonaceous grains; and the set of so-called unidentified infrared bands which have been attributed to emission from interstellar aromatic hydrocarbon molecules, such as polycyclic aromatic hydrocarbons (PAH). We will refer to them as aromatic hydrocarbon bands.

No separate detection of the DGL at near-IR wavelengths has been accomplished so far, although the galactic component of the near-IR background at 1.25 tex2html_wrap_inline10901m and 2.2 tex2html_wrap_inline10901m observed by the DIRBE experiment (Silverberg et al. 1993; Hauser 1996) undoubtedly contains a scattered light contribution. Recent evidence (Witt et al. 1994; Lehtinen & Mattila 1996) provides a strong indication that the dust albedo remains as high as it is in the visible out through the K-band (2.2 tex2html_wrap_inline10901m). The K-optical depth is about 10% of that at V; hence, only at quite low galactic latitudes (tex2html_wrap_inline15599) can one find the required dust column densities which will give rise to substantial (scattered) DGL. At the galactic equator, however, the ratio of DGL/LOS* should be similar to the values listed in Table 39 (click here). At higher galactic latitudes the ratio DGL/LOS* will be substantially lower than the values listed in Table 39 (click here).

The near-IR continuum emission was first recognized in reflection nebulae whose surface brightnesses in the 1 tex2html_wrap_inline10901m - 10 tex2html_wrap_inline10901m wavelength range exceeded that expected from scattering by factors of several (Sellgren et al. 1983; Sellgren 1984). Absence of polarization provided additional confirmation of the non-scattering origin of this radiation. The non-equilibrium nature of the radiation process was recognized from the fact that the colour temperature of the emerging radiation was independent of distance from the exciting star and thus independent of the density of the exciting radiation. This leaves as the cause of this radiation non-equilibrium processes which depend upon excitation by single photons, e.g. photoluminescence of grain mantles or, alternatively, non-equilibrium heating of tiny grains resulting in large temperature fluctuations. The galactic distribution of this radiation component has yet to be studied; it depends on a very accurate assessment of the near-IR integrated starlight (see Sect.nbsp;10.5 (click here)) and the near-IR zodiacal light (see Sect.nbsp;8.5 (click here)), which need to be subtracted from photometries of the near-IR sky background.

The aromatic hydrocarbon bands centered at wavelengths 3.3 tex2html_wrap_inline10901m, 6.2 tex2html_wrap_inline10901m, 7.7 tex2html_wrap_inline10901m, 8.6 tex2html_wrap_inline10901m, and 11.3 tex2html_wrap_inline10901m, with widths in the range of 0.03 to 0.5 tex2html_wrap_inline10901m, were first observed in bright nebulous regions by Gillett et al. (1973). Thanks to the successful AROME balloon-borne experiment (Giard et al. 1988; Ristorcelli et al. 1994) and the more recent missions of the Infrared Telescope in Space (IRTS, Onaka et al. 1996) and the Infrared Space Observatory (ISO, Mattila et al. 1996; Lemke et al. 1997), they have now been observed in the diffuse interstellar medium at low galactic latitudes. The relative bandstrengths and widths are very similar to those observed in reflection nebulae, planetary nebulae, and HII regions, pointing toward a common emission mechanism. Onaka et al. (1996) show that the band intensities at 3.3 tex2html_wrap_inline10901m and 7.7 tex2html_wrap_inline10901m and the far-IR background intensities at 100 tex2html_wrap_inline10901m along identical lines of sight are correlated very tightly, suggesting that the respective emitters, presumably PAH molecules in the case of the aromatic hydrocarbon bands and classical sub-micron grains for the 100-tex2html_wrap_inline10901m thermal continuum, are well-mixed spatially and are excited by the same interstellar radiation field. The correlation of the band intensities with the atomic hydrogen column density is also excellent, reflected in the dust emission spectrum per hydrogen atom given in Table 43 (click here).



Dust emission spectrum
4tex2html_wrap_inline15695 tex2html_wrap_inline10971
(tex2html_wrap_inline10901m) (erg s-1 H-atom-1)
0.65 1.17 10-24
3.5 1.27 10-25
4.9 1.36 10-25
7.7 1.13 10-25
12 1.02 10-24
25 4.11 10-25
60 7.04 10-25
100 2.64 10-24
140 3.64 10-24
240 1.41 10-24
346 5.77 10-25
490 1.77 10-25
535 1.20 10-25
736 3.33 10-26
1100 5.49 10-27

Table 43: Dust emission spectrum per hydrogen atom in the average interstellar medium

11.4. Thermal infrared

The infrared emission from the diffuse galactic ISM is dominated by thermal and other emissions by dust, with some additional contributions from interstellar cooling lines, mainly from CII and NII. At wavelengths < 100 tex2html_wrap_inline10901m the galactic diffuse emission is weaker than the infrared emission from the zodiacal dust cloud (see Fig. 1); at wavelengths > 400 tex2html_wrap_inline10901m the cosmic background radiation dominates over the galactic thermal radiation. Only in the 100 - 400 tex2html_wrap_inline10901m band is the galactic emission the primary background component. However, as the composite spectrum of all night sky components in Fig. 1 schematically indicates, the thermal IR spectrum of galactic dust is complex in structure, suggesting significant contributions from grains covering a wide range of temperatures. In particular, there is substantial excess emission in the 5 to 50 tex2html_wrap_inline10901m spectral range. This excess is generally attributed to stochastically heated very small grains with mean temperatures in the range tex2html_wrap_inline15647 (Draine & Anderson 1985; Weiland et al. 1986), while the main thermal emission peak near 150 tex2html_wrap_inline10901m is attributed to classical-sized dust grains in equilibrium with the galactic interstellar radiation field, resulting in temperatures around 20 K.

The exploration of the infrared background has been greatly advanced by the highly successful missions of the Infrared Astronomical Satellite (IRAS; Neugebauer et al. 1984), the Diffuse Infrared Background Experiment (DIRBE; Boggess et al. 1992) and the Far-Infrared Absolute Spectrophotometer (FIRAS; Fixsen et al. 1994) on board of the COBE satellite, the Infrared Telescope in Space (IRTS; Murakami et al. 1994, 1996), and the AROME balloon-borne experiment (Giard et al. 1988). Before mentioning the relation to interstellar gas, we first comment on the maps of galactic far-infrared emission available from these experiments.

From IRAS, the so-called ISSA maps are available for wavelengths of 12 tex2html_wrap_inline10901m, 25 tex2html_wrap_inline10901m, 60 tex2html_wrap_inline10901m and 100 tex2html_wrap_inline10901m. These present, after subtraction of a zodiacal light model, tex2html_wrap_inline15659 fields with 1.5' resolution, covering the sky on a 10tex2html_wrap_inline11647 grid. For the two longer wavelengths this gives a good picture of the variation of galactic emission (in MJy/sr). The absolute value of these maps is not reliable, which can be seen from the regions with negative intensities. To access the data one can use the world wide web address

With better modeling of the zodiacal light contribution, Rowan-Robinson et al. (1991) produced sky maps of galactic plus extragalactic far-infrared emission with 0.5tex2html_wrap_inline11647 pixel size. These maps give realistic absolute values and still show the spatial variation of the galactic diffuse emission in some detail. Tables 41 and 42 give a version of the two tables for 60 tex2html_wrap_inline10901m and 100 tex2html_wrap_inline10901m in ecliptic coordinates. They are not printed here but available in electronic form at the CDS by anonymous ftp to

But in particular the DIRBE Zodiacal Light-Subtracted Mission Average ("ZSMA") maps at 60 tex2html_wrap_inline10901m, 100 tex2html_wrap_inline10901m, 140 tex2html_wrap_inline10901m and 240 tex2html_wrap_inline10901m give a good estimate of the observed intensity resulting from the sum of galactic diffuse infrared emission and the extragalactic background at each of these wavelengths, apart from errors in the zodiacal light model model. Figure 77 (click here) shows the DIRBE 240 micron ZSMA map, along with representative intensity profiles at fixed galactic latitude. Since the extragalactic background light, which is spatially uniform, is not necessarily negligible at these wavelengths (see Table 47 (click here) in Sect. 12 (click here)), caution should be exercised when making quantitative statements about the absolute level of the diffuse Galactic infrared emission as derived from these maps. The Galactic contribution is certainly dominant at latitudes tex2html_wrap_inline15681, but probably also all over the sky. Aside from this issue of absolute levels (how much extragalactic background light radiation exists and has to be subtracted?), and aside from some visible zodiacal light model artifacts in the ecliptic plane at 60 microns, the ZSMA maps give a good picture of the spatial variation of the diffuse Galactic emission at all latitudes. The intensities recorded in the ZSMA maps are reported in MJy/sr, assuming a nominal wavelength and a spectral shape of tex2html_wrap_inline13261 = constant. The ZSMA maps are available from the NSSDC through the COBE homepage website at

Figure 77: Map of the sky brightness at 240 tex2html_wrap_inline10901m after removal of zodiacal light, obtained from the COBE/DIRBE experiment. The map is a Mollweide projection in Galactic coordinates, with intensities on a logarithmic stretch from 1 to 1000 MJy/sr. Smoothed intensity profiles at fixed Galactic latitudes of 0tex2html_wrap_inline11647, tex2html_wrap_inline15765, tex2html_wrap_inline15767 and tex2html_wrap_inline15769 are also plotted; positive latitudes are represented by the thin black, negative latitudes by the thick grey line

Interstellar dust appears to be well-mixed with all phases of the interstellar gas (Boulanger & Perault 1988; Sodroski et al. 1997); however, to obtain a first-order representation of the emissions from galactic dust, the well-established correlations with N(HI) provide the best guide. The average dust emission spectrum per H-atom is given in Table 43 (click here), as derived from the following original sources: ERE at 0.65 tex2html_wrap_inline10901m, Gordon (1997); galactic emission in the 3.3 tex2html_wrap_inline10901m aromatic feature, Giard et al. (1989), Bernard et al. (1994); dust emission in the 6.2, 7.7, 8.6, and 11.3 tex2html_wrap_inline10901m mid-infrared unidentified bands, Onaka et al. (1996); and the broad-band thermal dust emissions, Boulanger et al. (1996), Reach et al. (1995a), and Dwek et al. (1997).

Compared with emission from dust, the radiation from infrared cooling lines of the gas is comparatively weak, reflecting the fact that dust in interstellar space absorbs approximately one third of all energy emitted by stellar sources. The three strongest lines are the [CII] transition at 158 tex2html_wrap_inline10901m and the [NII] lines at 122 and 205 tex2html_wrap_inline10901m (Wright et al. 1991). At low galactic latitudes, the [CII] line emission is well correlated with N(HI), yielding (2.65 tex2html_wrap_inline11177 0.15) 10-26 ergs s-1 H-atom-1 (Bennett et al. 1994). With latitudes increasing beyond tex2html_wrap_inline15749, the ratio of I([158])/I(FIRtex2html_wrap_inline15755) decreases rapidly, leaving the 158 tex2html_wrap_inline10901m line unmeasurably weak at tex2html_wrap_inline15759.

11.5. Ultraviolet

11.5.1. FUV diffuse galactic light (91.2 nm to 180 nm)

Intense Lyman tex2html_wrap_inline10825 flux is present in this bandpass, and this, combined with various instrumental limitations, meant that most of the effort to measure a diffuse flux in this band was carried out at wavelengths longer than 121.6 nm. In retrospect, the measurement of a diffuse flux in this bandpass is far more difficult than was originally imagined, and the potential for obtaining erroneous results is substantial. The literature is filled with controversial and erroneous results.

This bandpass has been studied extensively since the beginning of the Space Age, because the zodiacal light component is not present and contributions from stellar sources were expected to be sufficiently low that emission from a hot (105 K) or very hot (106 K) intergalactic medium might be detected.

A review of this literature is provided by Bowyer (1991), but for an alternative view and a detailed examination of an particular data set, see Henry (1991).

Substantial progress has been made in identifying the components that contribute to the diffuse flux in this band. The vast majority of the diffuse flux is starlight scattered by interstellar dust. In particular, Haikala et al. (1995) found FUV emission at a high galactic latitude from a cirrus cloud detected at 100 tex2html_wrap_inline10901m with IRAS. Emission from hot (tex2html_wrap_inline10939105 K) gas has been detected. An analysis of this radiation establishes that the emitting gas is well above the Galactic plane. Two-photon emission from recombining ionized hydrogen has been recognized as a component of this background. Molecular hydrogen fluorescence has been found in low density clouds. Any extragalactic flux is quite small; this component is discussed in Sect. 12.1 (click here).

In Fig. 78 (click here) we provide examples of the best available data on the diffuse far ultraviolet background. The data from 912 to tex2html_wrap_inline109391200 Angstrom are from Holberg (1986) and are upper limits to the background from a high galactic latitude view direction. Two data sets are shown for the 1400 to 1840 Angstrom band. The upper line is from Hurwitz et al. (1991) and shows data obtained at a low galactic latitude. These data are typical of what is observed in viewing directions with an optical depth of tex2html_wrap_inline15771tex2html_wrap_inline13353 1. Molecular hydrogen fluorescence is evident as an additional component at wavelengths from 1550 to 1650 Angstrom. The lower line is from Martin & Bowyer (1990) and shows data obtained at a high galactic latitude and a low total galactic neutral hydrogen column. The CIV 1550 Angstrom line is clearly evident in emission, and the 1663 Angstrom line of forbidden O III is also apparent though at a lower signal-to-noise.

As already mentioned, the major components of the cosmic far ultraviolet background are summarized in Table 44 (click here) above.


Total intensity 300-1500
Scattering by dust 200-1500
H II two-photon emission 50
H2 fluorescence 100
(in molecular clouds)
Hot gas line emission 10
from hot Galactic gas
Extragalactic 50 to 200
Unexplained none to 200
Table 44: Components of the diffuse cosmic far ultraviolet background with approximate intensitiesa (photons cm-2 s-1 sr-1 Å-1)

a Intensities dependent upon view direction.
Intensities of processes producing discrete features are averaged over the 1400-1859 Å band.

Figure 78: Summary data on the diffuse cosmic far ultraviolet background. The data from 902 to   1200 Å are from Holberg (1986) and are upper limits to the flux from a high Galactic latitude view direction. Two data sets are shown for the 1400 to 1850 Å band. The upper line is from Hurwitz et al. (1991) and shows typical data obtained in view directions with tex2html_wrap_inline15771tex2html_wrap_inline13353 1. Note the H2 fluorescence emission around 160 nm. The lower line is from Martin & Bowyer (1990) and shows data obtained at a high Galactic latitude; the CIV 1550 Å line is clearly evident in emission and the 1663 Å line of OIII] is also apparent, though at lower signal-to-noise ratio. The extragalactic contribution to these data probably is small (see Table 44 (click here))

11.5.2. Near-ultraviolet (180 nm - 300 nm)

The diffuse radiation in this band is the sum of zodiacal light and starlight scattered by interstellar dust. A few first studies of the zodiacal light in this band have been carried out, which suggest this component exhibits characteristics similar to that observed in the visible (see Sect. 8.6 (click here)). A few studies of scattering by dust by early type stars have been carried out. The results obtained differ, and independent of these differences, the scattering varies tremendously from place to place in the galaxy. We refer the reader to Dring et al. (1996) and references therein for a discussion of these results.

next previous
Up: The 1997 reference

Copyright by the European Southern Observatory (ESO)