Automated optical identification of a large complete northern hemisphere sample of flat-spectrum radio sources with [formmu1]S6 cm>200 mJy

This paper describes the automated optical APM identification of radio sources from the Jodrell Bank-VLA Astrometric Survey (JVAS), as used for the search for distant radio-loud quasars. Since JVAS was not intended to be complete, a new complete sample, JVAS++, has been constructed with selection criteria similar to those of JVAS (S_5GHz>200mJy, α_1.4-5GHz>-0.5), and with the use of the more accurate GB6 and NVSS surveys. Comparison between this sample and JVAS indicates that the completeness and reliability of the JVAS survey are ~90 and ~70 per cent respectively. The complete sample has been used to investigate possible relations between optical and radio properties of flat-spectrum radio sources. From the 915 sources in the sample, 756 have an optical APM identification on a red (e) and/or blue (o) plate, resulting in an identification fraction of 83 per cent with a completeness and reliability of 98 and 99 per cent respectively. About 20 per cent are optically identified with extended APM objects on the red plates, e.g., galaxies. However, the distinction between galaxies and quasars can not be made properly near the magnitude limit of the POSS-I plates. The identification fraction appears to decrease from >90 per cent for sources with a 5-GHz flux density of >1Jy, to <80 per cent for sources at 0.2Jy. The identification fraction, in particular that for unresolved quasars, is found to be lower for sources with steeper radio spectra. In agreement with previous studies, we find that the quasars at low radio flux density levels also tend to have fainter optical magnitudes, although there is a large spread. In addition, objects with a steep radio-to-optical spectral index are found to be mainly highly polarized quasars, supporting the idea that in these objects the polarized synchrotron component is more prominent. It is shown that the large spread in radio-to-optical spectral index is possibly caused by source-to-source variations in the Doppler boosting of the synchrotron component.