3. Star formation history in the galactic disk
3.1. Previous chromospheric SFH determinations
In Fig. 6, we show a comparison between two SFHs, derived from chromospheric age distributions available in the literature: Barry (1988, SFH given by Noh & Scalo 1990) and Soderblom et al. (1991, SFH given by Rana & Basu 1992). In this plot, as well as in subsequent figures, the SFH will be expressed always as a relative birthrate, which is defined as the star formation rate in units of average past star formation rate (see Miller & Scalo 1979, for rigorous definition).
Note that the SFHs in Fig. 6 are very similar to each other, a result not really surprising since Soderblom et al. have used the same sample used by Barry. On the other hand, the corresponding events in Barry's SFH appears 1 Gyr earlier in Soderblom et al.'s SFH. The different age calibrations used in these works are the sole cause of this discrepancy. Barry makes use of Barry et al. (1987)'s calibration which used a low-resolution index analogous to Mount Wilson , while Soderblom et al. use a calibration derived by themselves. In Fig. 7, we show a comparison of the ages for Barry (1988)'s stars using both age calibrations. The difference in the ages are clearly caused by the slopes of the calibrations. Barry et al. (1987)'s calibration gives higher ages compared to the other calibration, which explains the differences in the corresponding SFHs published.
3.2. Determination of the SFH
The three corrections described in Sect. 2 are applied to our data in the following order: the age distribution is first weighted according to the volume corrections, then each age bin is multiplied by the factor and we iterate the result according to Eqs. (9) and (10). The final result is the best estimate of the star formation history. It is shown in Fig. 8a, for an age bin of 0.4 Gyr and Scalo's scale height. There can be seen three regions where the stars are more concentrated: at 0-1 Gyr, 2-5 Gyr and 7-9 Gyr ago. Beyond 10 Gyr of age, the SFH is very irregular, probably reflecting more the sample incompleteness in this age range, and age errors, than real features. These patterns are still present even considering a smaller age bin of 0.2 Gyr. Fig. 8b shows the same for Holmberg & Flynn (2000) scale heights. The only difference comes from the amplitude of the events. In this plot, the importance of the older bursts is increased, since in Holmberg & Flynn (2000) the difference in the scale heights of the oldest to the youngest stars is greater than the corresponding value in Scalo's scale heights.
We have used an extended nomenclature to that of Majewsky (1993) to refer to the features found. At the age range where bursts B and C were thought to occur double-peaked structures are now seen. Thus, we have used the terms B1 and B2, and C1 and C2, to these substructures. Also shown is the supposed burst D, as Majewski (1993) had suggested. Their meaning will be discussed later. The lulls between the bursts were named AB gap, BC gap and so on. Some of us have previously referred to the most recent lull as `Vaughan-Preston gap'. We now avoid the use of this term because:
Comparing with other studies in the literature, the SFH seems particularly different. There are still three major star formation episodes but their amplitude, extension and time of occurrence are not identical to those that were previously found by other authors. Table 1 summarizes the main characteristics of our SFH comparing to that of Barry (1988, as derived in Noh & Scalo 1990). In the table, the entries with two values stand for the SFH derived with different scale heights. The first number refers to the SFH with Scalo's scale height, and the other refers to that with Holmberg & Flynn's.
Table 1. Main features of the SFH compared with Barry (1988).
As we can see, the main events of our SFH seem to occur earlier than the corresponding events in Barry's SFH, by approximately 1 Gyr. This can also be seen in Fig. 6: the SFR from Soderblom et al. (1991)'s data have features earlier than Barry by about 1 Gyr. This comes mainly from the use of Soderblom et al. (1991)'s age calibration on which we have based our ages. This hypothesis is reinforced by the fact that the fraction of the stars formed in each burst is in reasonable agreement with the corresponding events in Barry's SFH (see Table 1). The events we have found are most likely to be the same that have appeared in previous works, and the difference in the time of occurrence comes from the shrinking of the chronologic scale of the age calibration.
The narrowing of the AB gap is one of the main differences of our SFH and that found by Barry. This can be expected since our sample does not show a well-marked Vaughan-Preston gap, contrary to what is found in the survey of Soderblom (1985), from which Barry (1988) selected his sample.
Some other differences in the amplitude and duration of the bursts can be understood as resulting from the differences in the samples used by us and by Barry. Nearly 70% of our stars come from HSDB survey. We have already shown in Paper I that HSDB and Soderblom (1985) surveys have different chromospheric activity distributions. These are directly reflected in the SFH.
We have found double peaks at bursts B and C. At the present moment we cannot distinguish these features from a real double-peaked burst (that is, two unresolved bursts) or a single smeared peak. However, it is interesting to see that the previous chromospheric SFHs give some evidence for a double burst C. In Fig. 6 burst C also seems to be formed by two peaks. On the other hand, the same does not occur for burst B. The feature called B2 corresponds more closely to burst B in the previous studies, but at the age where we have found B1, the other SFHs show a gap.
The resulting SFH comes directly from the age distribution, in an approach which assumes that the most frequent ages of the stars indicate the epochs when the star formation was more intense. Both the evolutionary and the scale height corrections do not change the clumps of stars already present in the age distribution. The only correction which could introduce spurious patterns in it is the volume correction, which must be applied before the other two. Fig. 1 shows how it affects the age distribution. It is elucidating that the major patterns of the age distributions are not much changed after this correction. We refer basically to the clumps of stars younger than 1 Gyr and stars with ages between 2 and 4 Gyr. These clumps will be identified with burst A and B, respectively, after the application of the other corrections. Note also, that the AB gap is clearly seen in the age distribution before the volume correction. In spite of it, it is necessary to know if the presence of stars with very high weights (due to their proximity and low temperature) could affect the results. Therefore, we have recalculated the SFH now disregarding the stars that have very high weights after the volume correction. We have cut the sample to those stars with weights not exceeding 2 and 3. The resulting SFHs is compared to the SFH of the whole sample in Fig. 9. It is possible to see that the presence of outliers does not affect the global result. The uncertainty introduced affects mainly the amplitude of the events, at a level similar to that introduced by the uncertainty in the scale heights. We believe that the volume correction has not impinged artificial patterns on the data, and that the star formation just derived reflects directly the observed distribution of stellar ages in the solar vicinity.
© European Southern Observatory (ESO) 2000
Online publication: June 20, 2000