4 Colour-colour diagrams

In Fig. 6 we show different IR diagrams derived from the data of the present observational programme (given in Table 2). The internal consistency of the derived colours given the small scatter around the average loci in all diagrams (discarding obviously a few individual measurements) is noteworthy. The dispersion in the (H-K) colour may account for an effect of metallicity similar to that found for main sequence stars (Paper II). The similarity between the (J-L) : (H-L) and (V-K) : (U-V) diagrams supports this possibility. The absence of stars at the turn-over of the (H-K):(J-H) diagram shows that the biasing of the sample with late K and M dwarfs is practically negligible. Discrepancies in the mean trends of solar metallicity lines published by other authors are seen. These could be ascribed partly to the uncertainty of the transformations applied in this work; however, zero-point and/or calibration problems in the definition of those photometric systems should not be completely discarded.

 

Figure 6: Infrared colour diagrams for the stars of the sample measured at the Observatorio del Teide. Solid circles : stars measured more than twice; open circles: stars measured once; triangles: TCS standards taken from Kidger (1992). The superimposed lines correspond to the intrinsic relation extracted from literature and transformed to the TCS system. Solid line: Bessel & Brett (1988); dashed line: Bouchet et al. (1991)

In order to ascertain the influence of metallicity on near-IR colours we have plotted the (H-K):(J-H) diagram for the whole sample, separating stars in three groups of decreasing metallicity which roughly correspond to the disk, halo/disk transition and halo populations (Fig. 7). The data of each of the groups in the range F5-K2 have been fitted, by using the least-squares method, to straight lines of the form (H-K)=a+b(J-H). The group with [Fe/H] $\ge\, -0.5$ consists of 140 stars, in this case the fit yields a=0.01 and b=0.19 with a rms error of 0.013 mag. The group with $-0.5\, \ge$ [Fe/H] $\ge\, -1.5$ consists of 84 stars, in this case the fit yields a=-0.01 and b=0.22 with a rms error of 0.017 mag. The group with $-1.5\, \ge$ [Fe/H] consists of 88 stars, in this case the fit yields a=-0.02 and b=0.25 with a rms error of 0.015 mag. It can be appreciated that the slope of the mean traces increases slightly with decreasing metal content. This effect is similar to that found for dwarf stars in Paper II. It is probably related to the ion H^-, which is the dominant source of continuum opacity in the near infrared wavelength range for the spectral types under consideration.

 

Figure 7: (H-K):(J-H) diagram for the stars of the whole sample separated according to metallicity. Open circles: [Fe/H] $\ge -0.5$ (Disk); squares: $-0.5\ge$[Fe/H] $\ge -1.5$ (Disk/Halo transition); triangles $-1.5\ge$ [Fe/H] (Halo). The data in the range F5-K2 have been fitted to straight lines of the form (J-H)=a+b(H-K). The mean lines obtained are overimposed on the graph to show the sensitivity of IR colours to metallicity. Solid line [Fe/H] $\ge -0.5$; dashed line $-0.5\ge$[Fe/H] $\ge -1.5$; dotted line $-1.5\ge$ [Fe/H]

Notice that the possible bias of insterstellar reddening can be safely neglected in this analysis, since on the one hand the size of the expected dereddening vectors is small, and on the other, their direction is approximately parallel to the mean traces of the giant stars on the (H-K) : (J-H) diagram.
In Table 4, we provide the computed IR colours in the TCS system according to spectral type and luminosity class. For the computation of the mean colours we have considered the photometric data of stars contained in Paper II and in the present sample with reliable spectral classification.

  

Table 4: Computed mean colours in the system of the TCS (Carlos Sánchez Telescope) for giant and main sequence stars, separated according to spectral types

The interstellar extinction has not been taken into account, however most of the stars considered are population I giants with well determined spectral types which are close to the solar neighbourhood. We have adopted a rejection criteria in order to calculate the mean values, for this reason the number of stars considered to obtain each colour varies slightly, this is the cause of the small discrepancies observed when adding (J-H) and (H-K) to obtain (J-K). The small differences between the mean lines of giants and dwarfs are significant, indeed both lines should only strongly diverge for types later than K3III ((V-K)>2.5).

Diagrams which combine optical and IR colours are suitable for analysing separately the effects of effective temperature and metallicity. In the (V-K):(B-V) diagram (Fig. 8) the spread under the intrinsic line for population I stars is due mainly to the blanketing effect on the (B-V) colour, since (V-K) is hardly sensitive to metallicity. We show the mean lines for the giant branch of 47 Tuc ([Fe/H] $\sim-1$) and M 15 ([Fe/H] $\sim-2$). The different variation of the (V-K) colour with metallicity when compared with that found for dwarf stars, especially in the redder part of the diagrams (see Paper II), is remarkable. In the (V-K):(U-V) diagram (Fig. 9) this point is even more conspicuous than in the (V-K):(B-V) diagram, although a considerable number of stars in our sample lack U measurements. In this case, the stretching of the metallicity axis is larger, since the blanketing effects in the U band are stronger. Superimposed on the sample stars we show, for the sake of comparison, the intrinsic lines for 47 Tuc (Frogel et al. 1981), typical for disc globular clusters, and the average for M 3, M 13, and M 92 (Cohen et al. 1978) as representative of halo globular clusters. It is clear from this point that the different Galactic populations of giant stars are well represented.

 

Figure 8: (V-K):(B-V) diagram for the stars of the sample. The superimposed lines correspond to the intrinsic relation extracted from Johnson (1966) for Population I giants. Black circles correspond to stars with IR colours measured in the present work; the remaining symbols correspond to stars taken from the sources described in Sect. 2. Dotted line: intrinsic line for 47 Tuc (Martínez-Roger 1985). Dashed line: intrinsic line for M 15 (Martínez-Roger 1985)

 

Figure 9: (V-K):(U-V) diagram for the stars of the sample. The superimposed lines correspond to the intrinsic relation extracted from Johnson (1966) for Population I giants. Black circles correspond to stars with IR colours measured in the present work; the remining symbols correspond to stars taken from the sources described in Sect. 2. This diagram provides an instructive characterization of the sample insofar as the (V-K) axis reflects the extension in temperature, and the (U-V) axis the blanketing effect associated with the metal abundance of stars