If the age and IMF can be constrained independently, the observed far-UV spectral energy distribution is mostly a measure of the dust attenuation. The continua of star-forming galaxies are known to obey a well-defined average obscuration curve above 1200 Å (Calzetti 2001). The curve accounts for the total absorption and encompasses the net effects of dust/star geometry, absorption, scattering, and grain-size distribution. Extension of this curve down to the Lyman limit using HUT and FUSE observations of starburst galaxies was done by Leitherer et al. (2002) and Buat et al. (2002), respectively. Their results are compared to stellar data and to theoretical predictions in Fig. 3. The reddening curve of Sasseen et al. (2002) applies to individual stars; it is significantly steeper than the curves derived for galaxies. The physical interpretation of the "grayer" starburst reddening curve is a non-uniform attenuation. Most of the stars are totally hidden from view, and the observed UV light is emitted by those few stars which happen to have low attenuation. This effect becomes progressively more important for shorter and shorter wavelengths. The implication is that far-UV observations sample only the tip of the iceberg and could be severely biased. For instance, if there were an age- or IMF-dependence of the reddening, even the interpretation of spectral lines in the far-UV would be compromised with the assumption of a simple foreground dust screen. Circumstantial evidence for this effect to play a role has been presented by Chandar et al. (2004).
 |
Figure 3. Comparison of attenuation
curves. Solid:
Buat et al. (2002);
dotted:
Leitherer et
al. (2002);
short-dashed:
Calzetti et al. (2000);
dot-dashed:
Sasseen et al. (2002);
long-dashed: model of
Witt & Gordon
(2000)
for a shell distribution, a clumpy dust and an optical depth in the V
band equal to
|
Witt & Gordon
(2000)
found that the empirical starburst attenuation law is most closely
reproduced by a clumpy shell model with SMC-like dust and a dust column
density equivalent to
V = 1.5. This
corresponds to a far-UV
attenuation correction factor of order 10 (Fig. 3).