6.1.1 Ground-Based Selection
Ground-based surveys have used morphological criteria, colour selection and emission line selection. Lists of blue compact galaxies were pioneered by Zwicky, followed by Fairall and others who isolated objects from their anomalous high surface brightness as seen on the Palomar Sky Survey. Spectroscopic follow ups have revealed a large proportion of H II galaxies and AGNs (Kunth et al. 1981)
The colour selection proceeds by searching for blue or ultraviolet excess objects involving various techniques such as the use of very low dispersed images or multiple colour direct images. Dispersed images have been used by the First and Second Byurakan Survey (FBS, SBS) by Markarian (1967) on IIaF emulsion and later the University Michigan survey (UM, MacAlpine et al. 1977) and Case survey (Pesch and Sanduleak 1983) with IIIaJ emulsion. The second method has been pioneered by Haro (1956) and extensively developed by the Kiso Observatory Survey (Takase and Miyauchi-Isobe 1984). Low resolution slitless spectroscopy enables to detect [O II] 3727, H, [O III] 4959, 5007 and H lines depending on the chosen emulsion or filter. Good seeing and excellent guiding are a requisite to avoid trailing and loss in detectivity. These techniques face a trade off between the dispersion and the spectral range covered. The higher the dispersion, the easier it becomes to detect weak emission lines against the continuum, while a narrow spectral range cuts significantly the sky background at the expense of the redshift range. The recent surveys conducted by Gallego et al. (1997) and Salzer (1999) use the H line which can be bright even in low-excitation or very metal-poor objects. Because each technique involves specific observational biases, modern surveys tend to combine various approaches. The use of large CCD arrays equipped with scanning Fabry Perot interferometry or slitless spectroscopy offer deeper limits at the expense of the reduced field of view. In the future, these combinations will probe distant H II galaxies populations. The most difficult problem that these surveys have to face is that of the follow-up observations (Terlevich et al. 1991). Getting even a rough oxygen abundance for an object fainter than the 17th magnitude requires long telescope time and suggests the use of multi-object-spectroscopy.