2000MNRAS.319..517L

In this paper we present semi-analytic models and Monte Carlo simulations of quasi-stellar object (QSO, quasar) Lyα absorption-line systems that originate in gaseous galactic haloes, galaxy discs and dark matter (DM) satellites around big central haloes. The aim is to estimate the number density per unit redshift of Lyα absorption lines related to galaxies, and to investigate the properties of the predicted galaxy-absorber systems, such as equivalent widths W_r, projected distances ρ, galaxy luminosities L_B, as well as absorber redshifts z. It is found that, for strong Lyα absorption lines {formmu1} galactic haloes and satellites can explain ∼20per cent and 40per cent of the line number density of the HST QSO Absorption Line Key Project respectively. The population of DM satellites is adopted from recent numerical simulations by Klypin et al. If big galaxies indeed possess such large numbers of DM satellites and they possess gas, these satellites may play an important role for strong Lyα lines. However, the predicted number density of Lyman-limit systems by satellites is ∼0.1 (per unit redshift), which is four times smaller than that by halo clouds. Including galactic haloes, satellites and Hi discs of spirals, the predicted number density of strong lines can be as much as 60per cent of the HST result. The models can also predict all of the observed Lyman-limit systems. For strong lines the average covering factor within 250h^–1kpc is estimated to be ∼0.36, which is in good agreement with observations. Also, the effective absorption radius of a galaxy (with unit covering factor) is estimated to be ∼150h^–1kpc. There exist correlations of W_r versus ρ, L_B and z. The models predict {formmu2} with {formmu3} {formmu4} and {formmu5}

To compare with the results of imaging and spectroscopic surveys, we study the selection effects of selection criteria similar to the surveys. We simulate mock observations through known QSO lines of sight and find that selection effects can statistically tighten the dependence of linewidth on projected distance. This result confirms previous suggestions in the literature. After applying selection criteria, the models can predict similar distributions of W_r, ρ, L_B, absolute magnitudes and absorber redshifts to those of imaging and spectroscopic surveys. Finally we find that the total redshift interval of the present observations (∼5) is not large enough for the models to reveal the real relationships if adopting the selection criteria. An adequate total redshift interval might be ∼10. This may reconcile contradictory conclusions about the anticorrelation of equivalent widths versus projected distances by different authors.