6 Characteristics of the Optical Counterparts to FIRST Sources

The objects in the FIRST catalog span over four orders of magnitude in radio flux density, and the identification rate as a function of flux density is displayed in Figure 18. Even for sources as bright as 1 Jy, only one-third of the objects have counterparts brighter than the plate limit. The identification rate falls monotonically with flux density to 20 mJy where it reaches a minimum of $\sim 12$% and then climbs again toward the survey limit. The rise at low flux densities reflects the change in the composition of the radio source population which is dominated by distant AGN at high flux densities and by nearby star-forming galaxies below 1 mJy (Condon 1992). The fraction of sources classified as stellar on at least one of the two POSS plates is also reflective of this trend (Figure 19). Above 100 mJy over 80% of all counterparts are stellar (quasars), whereas this fraction decreases monotonically until at 1 mJy, 60% of the counterparts are classified as galaxies; given the strong tendency toward the misclassification of faint galaxies as stellar, the true fraction of galaxy counterparts is considerably higher.

Figure 18: The fraction of FIRST sources with optical counterparts (matching to within $2^{\prime \prime }$) as a function of radio flux density. The shaded histogram includes objects classified as stars on at least one of the POSS-I plates; the remaining objects are classified as galaxies.

Figure 19: The fraction of FIRST sources with stellar optical counterparts (matching to within $2^{\prime \prime }$) as a function of radio flux density. Given the tendency toward misclassification of faint galaxies as stellar, the true fraction of galaxy counterparts at faint flux levels is considerably higher than 60%.

One way to explore the types of objects associated with FIRST sources is to place the optical counterparts onto a color - magnitude diagram. In Figures 20 and 21 we show the color-magnitude diagrams for the `stars' and galaxies (based on a consistent APM classification from both POSS plates) within $1.5\arcsec$ and $4\arcsec$ match radii, respectively. For comparison, these figures also display similar diagrams for an equal number of stellar and nonstellar sources taken from random positions on the same plates.

Figure 20: Color-magnitude diagrams for the 7877 objects classified as stellar on both plates in the APM catalog. (a) Objects lying within $1.5^{\prime \prime }$ of FIRST sources. (b) An equal number of stellar objects randomly selected from the same plates. The FIRST counterparts, fainter and bluer than the typical star, are mostly quasars.

Figure 21: Color-magnitude diagrams for the 16,047 objects classified as nonstellar on both plates in the APM catalog. (a) Objects lying within $4^{\prime \prime }$ of FIRST sources. (b) An equal number of nonstellar objects randomly selected from the same plates. The FIRST counterparts are, on average, brighter.

The radio counterparts classified as stellar are clearly much fainter and much bluer than most `stars'; this is illustrated in Figure 22, where we plot the fraction of the radio-detected `stars' as a function of location in the color-magnitude plane. The detected fraction peaks at $\sim10$% for colors around $O-E=0$ for $18<E<20$. These objects are nearly all quasars. The increasing fraction of detected objects with all colors fainter than $18^{th}$ magnitude is largely an artifact produced by the failure of the classifier to distinguish stars from galaxies near the plate limit; in fact, most of the faint, red stellar objects are actually galaxies, although a handful of very red objects have been shown to be high redshift quasars (Hook et al. 1998).

Figure 22: A contour and greyscale display of the fraction of objects classified as stellar on both plates which have radio counterparts. Darker regions have higher detected fractions. The fraction reaches nearly 10% for colors of $O-E=0$. Contour levels are 0.0001, 0.0002, 0.005, 0.001, 0.002, 0.005, 0.01, 0.02, and 0.05.

For the nonstellar objects, it is clear that the radio detections are biased toward brighter magnitudes, but are found for galaxies of all colors. In Figure 23, we show the detected fraction of galaxies in the color-magnitude plane. The roughly uniform detection rate as a function of color and the bias towards brighter galaxies are apparent; up to 10% of the brightest galaxies are detected and even for E$>$16, $>$2% of all galaxies have FIRST counterparts.

Figure 23: A contour and greyscale plot of the fraction of objects classified as nonstellar on both plates which have radio counterparts. More than 10% of all galaxies are detected at the brighter magnitudes. Contour levels are 0.002, 0.005, 0.01, 0.02, 0.05, 0.1, and 0.2.