6.2. The Narrow Line Region
The observations of narrow lines in high luminosity AGNs are not nearly as good as those of the broad lines. In particular, it is difficult to separate the broad and the narrow ultraviolet lines and there are no reliable measurements of the narrow CIV1549 and CIII]1909 lines in quasars and Seyfert 1 galaxies. The situation is likely to be improved with the HST observations but so far the only narrow ultraviolet lines that have been measured are in Seyfert 2 galaxies.
There are very good observations of optical narrow lines that can be compared with the model predictions. The overall agreement is very good and narrow line models, like the one shown in Fig. 12, reproduce the relative strength of [OIII]5007, [OII]3727, [OI]6300 and H quite well. This is not the case for [FeX]6734, and the line is observed to be much stronger than predicted. A similar, although somewhat smaller discrepancy, occurs for the lines of [NeV] and [FeVII].
Line profile observations (chapter 9) indicate a large density gradient in the NLR. The validity of the model in Fig. 12, where the density gradient is quite small, is thus questionable and the very high ionization lines may come from a much denser part of the NLR. Another, very different suggestion is that the high ionization lines come from the interstellar medium of the host galaxy.
Lower excitation spectra, such as in LINERs, cannot be explained by the relatively high ionization parameter model of Fig. 12. Such spectra are discussed in chapter 11.
The intensity of the narrow Balmer lines are easy to calculate. The H optical depth is not likely to be large, and the H / H ratio is closed to the Case B value. A comparison with the theoretical L intensity is somewhat less reliable. First, the line is likely to be collisionally excited by a density dependent amount. In addition, the typical NLR density is close to the critical density of the 2-photon transition (~ 1.5 × 104 cm-3) and the relative population of the hydrogen 2s and 2p levels may be different in different clouds. Because of this the recombination L / H ratio can vary from about 23 (low density limit, the 2s and 2p levels are not coupled) to 34 (high density limit, the 2s to 2p population ratio is 1:3). Combined with the collisional enhancement of L, the overall expected range in the L / H ratio is about 25-100. Accurate modeling is required for comparing this line ratio with the observations.