5. The star formation history of SagDIG
The heavy foreground contamination of SagDIG makes the analysis of the SFH difficult. It is based on the distribution of stars in the CMD and uses star counts in low populated areas of the CMD also (see Gallart et al. 1999). Fortunately enough, the bluest part of the SagDIG CMD, populated by youngest stars, is free from contamination and can provide information about the very recent SFH. Besides this, the RGB area can be used, after appropriate correction of foreground contamination to estimate the averaged SFR for ages older than Gyr (Aparicio et al. 2000a).
To start with, a qualitative idea of the stellar ages can be obtained from a glance at Fig. 10, which shows the CMD of SagDIG corrected from the reddening and distance given in Sect. 3. Five isochrones from the Padua library (see Bertelli et al. 1994 for the key reference) have been over-plotted with metallicity and ages 20, 50, 200 Myr and 1 and 10 Gyr. The RGB and AGB are shown for the latter. Only the stars within a radius of from the center of the galaxy have been plotted, which approximately corresponds to the maximum extension of the galaxy (see Sect. 4). It is worth noting that at least some of the stars in the strip extending from to are probably extended intermediate-mass AGBs belonging to SagDIG (see below). Note also the separation between the MS and the blue-loop sequence, and that the brightest blue stars are blue-loopers.
A simplified version of the method proposed by Aparicio et al. (1997b) has been employed to obtain the SFH of SagDIG. In practice, a synthetic CMD with arbitrary, constant SFR of value , the IMF of Kroupa et al. (1993) with lower and upper cut-offs and , respectively, and a metallicity, Z, taking random values from to , independently of age have been used. To avoid small number statistics effects, the synthetic CMD have been computed with 50000 stars with . This guarantees that the relevant regions of the CMD (see below) are well populated and that the synthetic CMD does not introduce further statistical errors to the SFH result. In Sect. 3 the metallicity of SagDIG was estimated as [Fe/H] . This is an extremely low value, probably corresponding to . However, the Padua library is not complete for such low metallicities. For this reason the former metallicity range has been used as representative of a very low metallicity galaxy. Finally, since our results for the SFR can only be an estimate, we have neglected the effects of binary stars.
The resulting synthetic CMD has then been divided into three age intervals: Gyr, Gyr and Gyr. Following the nomenclature introduced in Aparicio et al. (1997b), each of the synthetic diagrams corresponding to the three previously defined age intervals will be called partial model CMDs and any linear combination of them will be denoted as global model . Three regions have been defined in the observed and partial model CMDs as shown in Fig. 10, with the criterion that they sample different age intervals and stellar evolutionary phases, namely the youngest blue-loops [; ], the MS plus young blue-loops [; ], and the RGB+AGB region below the TRGB [; ]. In practice, the two youngest time intervals are sampled by blue stars only, the oldest age included in box 2 of Fig. 10 being in fact about 0.2 Gyr. Although stars of any age above 0.2 Gyr populate box 3 (Fig. 10), this box is in practice dominated by low-mass stars, which are therefore older than about 1 Gyr, so that the sampling of stars in the age interval 0.2-1 Gyr remains poor. This interval should be solved using AGB stars brighter than the TRGB. But the low star counts usually found in the upper AGB together with the high foreground contamination prevent us from making any estimate based on that region. In summary, we will give the average SFR for the 0.2-15 Gyr interval, but the bad sampling of the 0.2-1 Gyr interval must be borne in mind.
We denote by the number of stars of the observed CMD lying in region j and by the number of stars of partial model (age interval) i populating region j. After completeness and foreground correction, take the following values: , , and (inner ). The number of stars populating a given region in a global model is then given by
and the corresponding SFR
where are the linear combination coefficients; k is a scaling constant which transforms from the arbitrary units used in the computation to final, physical units; if t is inside the interval corresponding to partial model i and otherwise. In the simple approach we are using, the coefficients can be analytically solved to produce . This results in the SFRs for the three considered intervals of time given in Table 1. The first three lines give the SFR for the time intervals Gyr; Gyr and Gyr. The three next lines give the same normalized to the area of SagDIG, considered to extend to the Holmberg radius. Quoted errors have been calculated assuming Poisson statistics in the star counts.
Table 1. Star Formation Rates of SagDIG for different age intervals (in Gyr)
Fig. 11 shows the synthetic CMD corresponding to the former solution of the SFH. No simulation of observational effects has been done. As a result, the lowest part of the diagram is much more clearly defined than that of Fig. 4. Note in particular the separation between MS and blue-loop stars, which is only marginally visible in the observational CMD. Interestingly a large amount of bright AGB stars populate the synthetic CMD, indicating, as previously stated, that, at least, some of the stars in the strip extending from to of Fig. 4 are AGBs. Unfortunately, the strong foreground contamination prevents using these stars to improve the SFH result.
The current SFR of SagDIG can also be estimated from the flux given by Strobel et al. (1991). Using our estimate of the distance and following the procedure shown in Aparicio et al. (2000b) with an upper mass for stars , the current SFR results ; i.e., an order of magnitude smaller than the value obtained from the CMD and given in Table 1 for the last 50 Myr (). The disagreement can be solved if an upper cut-off for stellar masses of is imposed.
Summarizing, SagDIG seems to be experiencing a strong burst of star formation which drives it to form stars at a rate 10 times larger than the average of its entire life. This picture is frequently found in galaxies classified as dIrr: (NGC 6822: Gallart et al. 1996b,c; Pegasus: Aparicio et al. 1997a, Gallagher et al. 1998; Sextans A: Dohm-Palmer et al. 1997, Van Dyk et al. 1998; Antlia: Aparicio et al. 1997c; DDO 187: Aparicio et al. 1999). This might imply that an important bias could exist in the classification of dwarfs as bona-fide dIrrs towards objects experiencing strong star formation bursts.
© European Southern Observatory (ESO) 1999
Online publication: December 2, 1999