Robust Data-driven Metallicities for 175 Million Stars from Gaia XP Spectra

Encouraged from the results in Rix et al. (2022), we train XGBoost models using as data features both XP coefficients (Carrasco et al. 2021; De Angeli et al. 2022) and synthesized photometry that was computed with GaiaXPy (e.g., Gaia Collaboration et al. 2022b). For the most part, we do this for stars with literature labels from APOGEE DR17 (Abdurro'uf & Aerts 2022). However, APOGEE DR17 has no metallicity estimates below [M/H] ∼ −2.5, and XGBoost cannot extrapolate beyond the metallicity range of its training sample. ⁴ Therefore, we augment the APOGEE DR17 training sample by the set of very/ultra-metal-poor stars from Li et al. (2022), which provide a consistent and fairly extensive set of [M/H] determinations for a set of (apparently) bright stars. We also replace the AllWISE photometry (Cutri et al. 2021) used in Rix et al. (2022) by CatWISE photometry (Marocco et al. 2021), which is deeper and thus achieves higher completeness.

APOGEE DR17 contains a total of 733,901 stars. Of these, 647,025 actually have the stellar parameters T_eff, $\mathrm{log}g$, and [M/H], and 64,401 have a crossmatch to Gaia. Among these, only 599,662 achieve S/Ns above 50 in the APOGEE spectra. For the purpose of this paper, we also require XP spectra to be available, which is the case in the Gaia DR3 data for 537,412 of these stars. We also require CatWISE photometry in the W₁ and W₂ bands as these bands greatly aid reducing the temperature—extinction degeneracy. This reduces the set of APOGEE DR17 training to 510,413, which only contains 485,850 unique Gaia source IDs, i.e., there are "duplicates" which represent repeated APOGEE observations of the same star. In such cases, we adopt the mean APOGEE parameters averaged over all repeat observations as training labels. The results of Li et al. (2022) encompass 385 stars, of which 291 stars have published XP spectra, as well as W₁ and W₂ photometry in CatWISE.

The resulting [M/H] distribution of this training sample is shown in Figure 1. Already from APOGEE, we have good coverage for [M/H] < −1. But this figure also shows how critical the inclusion of stars from Li et al. (2022) is to cover metallicities below −2.5, eventually down to a minimum value of −4.37. This expanded training sample removes an important limitation at low metallicity of the work by Rix et al. (2022).

Zoom In Zoom Out Reset image size

It is worth taking a closer look at the distribution of all stellar parameters in our training sample in Figure 2, to understand over which range we can expect XGBoost to return robust estimates. While Figure 2(c) suggests that we have a good coverage of main-sequence dwarfs and red-giant stars, the temperature range is limited to 3107 to 6867 K. In particular, we have no OBA stars, no white dwarfs, or ultracool dwarfs in our training sample. Furthermore, Figure 2(a) shows that we have essentially no training examples for [M/H] < −2 and T_eff < 4000 K. Also, Figure 2(b) shows that we have only very few training examples for metal-poor dwarfs with $\mathrm{log}g\gt 3.5$ and [M/H] < −1. This will likely preclude robust and precise parameter estimates in this regime.

Zoom In Zoom Out Reset image size

Rix et al. (2022) noticed that some bands synthesized from the XP coefficients with GaiaXPy had negative fluxes and thus invalid magnitudes, particularly narrow bands in the blue where fluxes are often low. This leads to a rapidly decreasing completeness of XGBoost predictions in Rix et al. (2022) at the faint end. Here, we address this issue more systematically to keep the completeness toward the faint end as high as possible. First, for all stars in our application sample, we synthesize their photometry with GaiaXPy in the following photometric systems that we a priori believe to be useful for estimating metallicities (for details see Gaia Collaboration et al. 2022b):

• 1.  
  Pristine,
• 2.  
  Stromgren_Std,
• 3.  
  JPLUS,
• 4.  
  PanSTARRS1_Std,
• 5.  
  Sky_Mapper,
• 6.  
  Gaia_2.

Second, for all photometric bands, derived from the XP spectra, CatWISE, and AllWISE, we investigate the completeness of the magnitudes as a function of G_BP in Figure 3. Evidently, the completeness of the synthesized photometry diminishes much earlier in some bands than in others. Further investigation reveals that bands with pivot wavelengths below ≈420 nm are the first to be affected by incompleteness. This is consistent with our interpretation of the incompleteness arising through noise in the XP coefficients, given that the BP spectra have low transmission and thus low S/Ns for wavelengths below 420 nm. Furthermore, Figure 3 shows that AllWISE would limit the completeness at all magnitudes, and that CatWISE can reach much higher completeness especially at the faint end where there are numerous stars. Still, even CatWISE does not reach full completeness even at the bright end.

Zoom In Zoom Out Reset image size

For the final set of input features for XGBoost, we adopt all bands that achieve a completeness of 95% or higher at G_BP = 18. These are 31 bands and for each band, the XGBoost input feature is the color obtained from the apparent G magnitude minus the magnitude in this band. We choose the G magnitude for all colors for two reasons: first, the G magnitude is measured independently from the XP spectra from which all synthetic photometry are derived; and second, the G magnitudes have very high S/Ns. Additionally, we use the three Gaia colors G − G_BP, G − G_RP, and G_BP − G_RP as input features, as well as several colors including CatWISE photometry (namely W₁ − W₂, G − W₁, G − W₂, and G_BP − W₂). Therefore, our final set of data features comprises 38 colors and all 110 XP coefficients normalized to G = 15. This may appear confusing at first, because the synthesized photometry is fully redundant with the XP coefficients and adding redundant features could even be detrimental to the scientific performance (curse of dimensionality). Ultimately though, the choice to include both XP coefficients and photometry synthesized from XP as input features for XGBoost is a matter of feature selection that we test during cross-validation (see Table 1 in Rix et al. 2022). As it turns out, both are required in order to achieve optimal [M/H] results and the omission of either XP coefficients or synthetic photometry would lead to a noteworthy increase in the [M/H] errors during cross-validation and later application. This implies that XGBoost is unable to extract all information fully from the XP coefficients alone. Instead, our manual help to "rephrase" the information in terms of synthesized photometry is required in order to make the information more easily accessible for XGBoost.

For deriving stellar parameters, in particular $\mathrm{log}g$, the absolute magnitude is highly informative: e.g., it straightforwardly differentiates between giants and dwarfs. While a substantive subset of the stars with XP spectra have good parallax S/Ns, from which we can estimate absolute magnitudes, many sample members have parallaxes that are consistent with zero or even negative. Therefore, we added a data feature that reflects or places limits on the absolute magnitude, but is linear in the parallax in order to remain well behaved in cases of noisy or even negative parallaxes. Specifically, we opted for input features of the form

$$\begin{eqnarray}&&\varpi \cdot {10}^{{m}_{X}/5}={10}^{{\mathrm{log}}_{10}\varpi +{m}_{X}/5}={10}^{({M}_{X}+{A}_{X}-5)/5},\end{eqnarray} \tag{ 1 }$$

where ϖ denotes the parallax, m_X is the apparent magnitude in some band X, while M_X and A_X are, respectively, the absolute magnitude and dust attenuation in the same band X. ⁵ We added five such input features, for the photometric bands X = G, G_BP, G_RP, W₁, and W₂. The parallax in Equation (1) has been corrected for the parallax zero-point according to Lindegren et al. (2021). We find that these additional features do not only help to estimate $\mathrm{log}g$, but they also improve our [M/H] estimates by ∼10% where metal-poor giants benefit in particular. The complete list of all input features and the details of the XGBoost configuration are provided in Appendix A.

We use the exact same set of input features for training the XGBoost models for [M/H], T_eff, and $\mathrm{log}g$, all based on training labels (see Section 2.1). Our objective is to maximize the number of stars for which all required input features are available. In that case, the completeness of our results would be dominated by the completeness of the AllWISE photometry (see Figure 3).

For internal validation, we assess the quality of the XGBoost results on the training sample, using 20-fold cross-validation: 20 times we set aside disjoint sets comprising 5% of the data for subsequent testing of a model trained on the other 95% of the data. In the end, the data features for each object in the training sample have been compared to a statistically independent XGBoost model prediction for them. These cross-validation results are summarized in Figure 4.

Zoom In Zoom Out Reset image size

The first row of Figure 4 is most important because it shows the cross-validation of the [M/H] estimates. Panel (a) shows that, for the most part, our results are accurate, i.e., unbiased with respect to the APOGEE reference [M/H]. There are only a few outliers where XGBoost assigns a higher [M/H] value than APOGEE. However, for training labels [M/H] < −3 XGBoost tends to overestimate [M/H]. Most likely, this is a consequence of mixing different definitions of "metallicity" in our training sample: while APOGEE provides [M/H] estimates, Li et al. (2022) provide estimates of [Fe/H]. Astrophysically, very old stars will have low iron content but may already have been enhanced in other elements, such that the stars from Li et al. (2022) may be genuinely low in [Fe/H] but have higher [M/H], which is recognized by XGBoost learning [M/H] from the majority of training examples provided by APOGEE. Panel (b) shows how our [M/H] estimates depend on T_eff values: the agreement is overall very good, and for stars hotter than ∼5500 K our [M/H] estimates are closer to the training labels than for cooler stars. The main reason for this is that for T_eff > 5500 K the APOGEE sample contains virtually no stars with [M/H] below −0.75, limiting the comparison to the "easy" metal-rich regime. Panel (c) shows that our [M/H] residuals do not exhibit any noteworthy trends with the training sample's $\mathrm{log}g$, i.e., our [M/H] estimates work just as good for main-sequence dwarfs as they work for red-giant stars. Panel (d) shows that our [M/H] estimates remain robust (though obviously less precise) to the very faint end of G_BP ∼ 20. The overall RMSD to the training sample's [M/H] estimates is 0.106 and half of the stars differ by less than 0.044 from their reference values. This performance is about 10% better than that found for the sample of only red giants in Rix et al. (2022; see Table 1 therein). This improvement, despite the expanded coverage of the CMD, is mainly due to the inclusion of luminosity estimates (see Equation (1)) as features. Like in Rix et al. (2022), the current [M/H] estimates remain unbiased as the A_K extinction increases as is evident from Figure 5(a). Including CatWISE photometry is the key here. Furthermore, Figure 5(b) establishes that there are also no systematics with parallax (i.e., inverse distance).

Zoom In Zoom Out Reset image size

In particular, Rix et al. (2022) restricted their analysis to bright (G_BP < 16 mag) red giant branch (RGB) stars (teff_xgboost < 5500 K and logg_xgboost < 3.5). Figures 4(b) and (c) suggest that the [M/H] estimates from the current work are also robust outside the RGB and panel (d) suggests that this also holds down to the faintest stars which have their XP spectra published in Gaia DR3. Note that we cannot test with this validation sample whether our [M/H] estimates remain so precise and robust also for very-metal-poor stars.

The other two rows of Figure 4 show the XGBoost residuals for T_eff (middle) and $\mathrm{log}g$ (bottom). The RMSDs are remarkably small: 54 K for T_eff and 0.089 for $\mathrm{log}g$. Furthermore, the residuals do not show obvious systematics and appear to remain robust down to G_BP ∼ 20.

We define the sample to which we apply the XGBoost estimator as all sources in Gaia DR3 that have XP spectra, valid parallaxes, and proper motions, and valid XP-derived and CatWISE photometry; the parallaxes do not have to differ significantly from zero. The following AQDL query

• SELECT
• source_id
• FROM gaiadr3.gaia_source_lite
• WHERE has_xp_continuous='true'
• AND parallax IS NOT NULL

results in 218,132,063 stars. Since we require complete photometry in CatWISE and synthesized passbands (see Section 2.2), not all of them have the complete set of input features to XGBoost. The final number of stars that satisfy these additional conditions is 174,922,161 (∼80.2%). Their apparent magnitude distributions are shown in Figure 6. It is important to note that in Gaia DR3 XP spectra were only published for sources brighter than G = 17.65, yet Figure 6 shows sources fainter than that. The reason is that the XP spectra of presumed QSOs, galaxies, and ultracool dwarfs were exempt from the Gaia DR3 publication limit of G = 17.65. Consequently, these objects may be contaminants in our stellar parameter catalog: they manifest as a small bump at G ∼ 19 in the distributions of G and G_RP.

Zoom In Zoom Out Reset image size

The overall result of this analysis is given in Table 1: the three stellar parameters, [M/H], T_eff, and $\mathrm{log}g$ for 175 million sources, specified by their Gaia DR3 source ID and with a label whether these sources were included in the XGBoost training.

We can validate these XGBoost results by a comparison to other surveys not used in the training. Specifically, we compare to results from GSP-Spec after calibration ⁶ in Gaia DR3 (Recio-Blanco et al. 2022), GALAH DR3 (Buder et al. 2021), and SkyMapper DR2 (Chiti et al. 2021).

To start, we compare the [M/H] estimates from XGBoost with other metallicity estimates. For GSP-Spec (Figure 7(a)), the agreement is excellent: there are no discernible systematics and very few outliers. Importantly, the GSP-Spec comparison is mostly limited to [M/H] > −1, where we expect the [M/H] estimates to be robust. For GALAH (Figure 7(b)), we still see good overall agreement across the full metallicity range. However, there are some outliers, where GALAH estimates [Fe/H] below −1, while XGBoost estimates [M/H] above −0.5. We also note a small systematic offset below an [Fe/H] of −1 where XGBoost's [M/H] is ∼0.2 lower than GALAH's [Fe/H]. These outliers and the slight offset have also been observed in the results of Rix et al. (2022). Yet, unlike in Rix et al. (2022), we no longer see a saturation of XGBoost metallicities below −2, where we now see a continuation of the one-to-one relation with GALAH. This is the result of including the very-metal-poor stars of Li et al. (2022) in our training sample, thus extending the APOGEE metallicity range. For the SkyMapper photometric metallicities (Figure 7(c)), substantial scatter is evident both visually and quantitatively. Successful comparisons to the external spectroscopic surveys suggest that this scatter is probably inherent to SkyMapper.

Zoom In Zoom Out Reset image size

We further investigate the origin of the outliers and systematics of our [M/H] estimates in Figure 8 where we directly compare the metallicity estimates from APOGEE (i.e., the training sample underlying XGBoost) to those from GALAH. First, we observe the same small systematic offset below an [Fe/H] of −1 between GALAH and APOGEE, i.e., the XGBoost model has correctly learned from APOGEE and simply reflects this difference. Second, we can also see the outliers where GALAH votes for [Fe/H] < −1 whereas APOGEE votes for [M/H] > −0.5, so again XGBoost has faithfully learned from its APOGEE training sample. Consequently, both effects are traced back to genuine differences between GALAH and APOGEE and are thus not introduced by XGBoost. In fact, these outliers in Figure 7(b) mostly have high temperatures in GALAH and potentially correspond to outliers in GALAH DR3 itself.

Zoom In Zoom Out Reset image size

Quantitatively, Figure 7 shows that our XGBoost [M/H] estimates compare very well with those from GSP-Spec and GALAH, with half of the stars differing by no more than 0.092 and 0.068, respectively. This external validation error is somewhat larger than the cross-validation error of 0.042 found for APOGEE in Figure 4. This most likely reflects subtle differences between APOGEE's and GSPSpec's [M/H] estimates (our XGBoost estimates are tied to the APOGEE scale), as similar scatter is found in the direct comparison of these surveys.

For the scatter of temperatures and surface gravities when comparing with GSP-Spec we find RMSDs of 112 K for T_eff and 0.223 for $\mathrm{log}g$; and when comparing with GALAH, we find 166 K for T_eff and 0.119 for $\mathrm{log}g$. These are again slightly higher than the RMSDs from the 20-fold cross-validation on APOGEE (54 K and 0.089, respectively) quoted in Figure 4. For $\mathrm{log}g$, the difference to GSP-Spec is larger than for APOGEE or GALAH, but we also did not apply the empirical corrections for GSP-Spec's $\mathrm{log}g$ recommended in Recio-Blanco et al. (2022).

Of particular interest is the publication limit of XP spectra in Gaia DR3, which was set at G = 17.65. Figure 4(d) suggests that our XGBoost results may remain robust as we approach the publication limit, but we would like to confirm this with an independent validation sample. Unfortunately, both GSP-Spec and GALAH DR3 are of no use in exploring this regime, as both samples are limited to bright stars. Therefore, we make use of the LAMOST DR6 ⁷ data (Wu et al. 2011, 2014). As is evident from Figure 9, the [M/H] differences between XGBoost and LAMOST degrade "gracefully" toward the faint end, which means that the random scatter increases smoothly and no systematics appear. At G = 16 the central 68% interval ranges from −0.2 to +0.2 and even at G = 17.65 it ranges from −0.3 to +0.4. In fact, these variances include the [Fe/H] uncertainties from LAMOST, which typically are of the order of 0.25 around G = 17.65. Assuming that these uncertainties add in quadrature, the −0.3 to +0.4 interval at G = 17.65 implies an uncertainty of 0.17—0.32 attributable to XGBoost.

Zoom In Zoom Out Reset image size

While a random error of ∼0.33 in [M/H] at G ∼ 17 is acceptable, it still represents a substantial increase from the error of 0.1 at the bright end. What are the possible origins of this increased noise? First, an earlier version of our catalog was based on AllWISE photometry instead of CatWISE, but apart from having significantly lower completeness (see Figure 3) it produced the same results at the faint end. This rules out the CatWISE photometry as origin for the increased noise. Second, Gaia DR3 parallaxes can become very noisy toward the faint end, such that the features defined in Equation (1) could begin to confuse XGBoost at the faint end. However, if we remove these features from the XGBoost input and thus become entirely independent from the parallax, we find no improvement either. This only leaves the XP spectra as the source of the increased noise toward the faint end. More precisely, we suspect that it is not the XP coefficients themselves but rather the synthesized narrowband photometry which is becoming increasingly susceptible to noise toward the faint end. This interpretation is also supported by Figure 3, which reminds us that the synthetic photometry becomes incomplete due to noise, leading to negative flux values even when XP spectra are available.

Gaia Collaboration et al. (2022a) compiled a list of 5863 solar-analog candidates, whereof 5759 are in our sample. According to XGBoost, their mean [M/H] is 0.012 ± 0.105 and the central 90% interval ranges from −0.167 to 0.178. Figure 10 shows their distribution, which is consistent with a Gaussian of standard deviation 0.1. This is in excellent agreement with the solar value and demonstrates that our [M/H] estimates are also reliable at least for solar-like main-sequence dwarfs, whereas the estimates from Rix et al. (2022) were applicable only to giant stars.

Zoom In Zoom Out Reset image size

Among our XGBoost results, we find 22,477 member stars in 36 open clusters from Gaia Collaboration et al. (2018). As a first instructive example, Figure 11(b) shows how XGBoost's metallicity estimates vary with G_BP − G_RP color in the Praesepe cluster. The expected literature value is recovered only within a certain color range but otherwise XGBoost systematically underestimates the metallicity. This underestimate is related to the limited temperature range of the training sample (3107–6867 K, see Section 2.1). In the absence of interstellar extinction (such as for Praesepe), this temperature range roughly corresponds to a G_BP − G_RP color range from 0.5 to 2.3. Indeed, Figure 11(b) shows that the underestimated [M/H] values occur mainly for colors bluer than 0.5 or redder than 2.5, with a less pronounced underestimation of about 0.15 also in the range from 1.5 to 2.5. As is also evident from Figure 11(a), the Praesepe member stars are dominated by main-sequence dwarfs. We also note that the solar analogs showing excellent agreement in Figure 10 have intrinsic colors of G_BP − G_RP = 0.818 ± 0.029 (Gaia Collaboration et al. 2022a), and would thus fall well within the regime of good agreement between 0.5 and 1.5 in Figure 11(b).

Zoom In Zoom Out Reset image size

Obviously, many Praesepe member stars fall into a temperature range that is not covered sufficiently by our training sample. Unfortunately, we cannot select based on our catalog's XGBoost temperature estimates because these values are also strictly confined to the range 3107–6867 K of the training examples. Instead, we select on input features as shown in Figure 12: for every individual cluster member star, we ask where its features fall into the two diagrams and we reject it from further consideration if and only if it has a sufficiently high number of training examples nearby in these diagrams. ⁸ After this filtering procedure, Figure 13 shows that the XGBoost [M/H] estimates agree reasonably well with the adopted mean metallicities of the 36 open clusters from Gaia Collaboration et al. (2018). We do see a slight positive offset below an [Fe/H] of −1 and a slight negative offset around solar [Fe/H]. The latter is probably similar to the offset seen in Figure 11 for Praesepe and color G_BP − G_RP > 1.4. The slight positive offset below an [Fe/H] of −1 may be due to XGBoost occasionally overestimating [M/H] in that regime (see Figure 4(a)).

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Given the catalog of El-Badry et al. (2021), we find 55,033 pairs of wide binaries where each component star is observed as an individual source by Gaia. Since both stars from each binary pair have formed from the same gas cloud, XGBoost should estimate the same metallicity. In fact, Figure 14 shows that their [M/H] estimates are consistent with each other. The RMSD divided by $\sqrt{2}$ is 0.105.

Zoom In Zoom Out Reset image size

An inspection of the XGBoost surface gravities reveals that essentially all wide binary members are main-sequence dwarf stars, which most likely is a selection effect from El-Badry et al. (2021) focusing on high-quality astrometry in Gaia DR2. Being main-sequence stars, the limitations from Figure 11(b) apply: if we restrict the comparison to wide binaries in which both stars fall within the color range 0.5 < G_BP − G_RP < 1.5, the RMSD divided by $\sqrt{2}$ drops from 0.105 to 0.079 and the mean difference is 0.004. In contrast, if we only consider pairs where one component is in the good color range, whereas the second component is within 1.5 < G_BP − G_RP < 2.5, the redder stars have on average a 0.05 lower [M/H] than the (unbiased) bluer stars. This offset is slightly less than the systematic underestimation of –0.15 seen in Praesepe in Figure 11(b).

We also note that while most of the wide binaries in Figure 14 are at [M/H] > −0.5, there are a handful of systems with [M/H] < −1 and that in those cases the XGBoost estimates still hold.

In this section, we investigate the results for OBA stars. Gaia Collaboration et al. (2022a) compiled a list of 3,023,388 OBA stars, whereof 2,371 ,118 are in our sample. These are beyond the temperature range of the training sample, i.e., we intentionally break the model assumptions of our XGBoost model. Given that absorption lines are often washed out in very hot stars, we expect XGBoost to misinterpret such stars as metal poor. Since OBA stars are mostly young, they should have solar-like metallicities or higher. However, XGBoost assigns a median [M/H] of −0.443 to the stars in this sample and about 11% of the OBA stars are assigned an [M/H] lower than −1 by XGBoost. Consequently, the XGBoost estimates are clearly not viable for OBA stars. We note, however, that Gaia Collaboration et al. (2022a) report contamination from metal-poor stars, i.e., not all of those may actually be hot OBA stars.

Perhaps the most compact way to present the sample is to show its [M/H] distribution. We show this distribution in Figure 15 for all 174,922,161 stars, for 43,520,755 likely RGB stars, and for 18,858,968 RGB stars with high-quality parallaxes (see below). The [M/H] distribution for this last subset in Figure 15, restricted to the Milky Way within about 10 kpc, is the one that can be taken most at face value as an observational approximation of the "total" metallicity distribution of the Galaxy. Of course this is a flux-limited sample, whose extent is limited by distance, dust extinction, and (in part of the sky) crowding. Proper volume corrections of this [M/H] distribution would be a complex exercise (see, e.g., Rix et al. 2021) that is beyond the scope of this work. Such an analysis must include the G ≤ 17.65 publication limit of XP spectra in Gaia DR3 (e.g., introducing foreground dust extinction), the completeness of photometry synthesized from XP spectra (i.e., loss of stars due to negative synthetic fluxes caused by noisy XP spectra, see Section 2.2), and the crowding-afflicted completeness of AllWISE photometry.

Zoom In Zoom Out Reset image size

Nonetheless, this distribution shows a number of remarkable features, extending over a factor of 10,000 in metallicity within a single galaxy. It starts at [M/H] ≈ −3.5, rises steeply to [M/H] ≈ −2.5, and then follows $d\mathrm{log}N/d[{\rm{M}}/{\rm{H}}]\sim 1$ to [M/H] ≈ −1.0. At this point, the onset of the old disk in metallicity, the [M/H] distribution rises quickly to a maximum near [M/H] ≈ −0.4, stays flat to [M/H] ≈ +0.2, dropping steeply beyond. The implications of this distribution in terms of chemical enrichment warrant to be studied in a framework of chemical evolution models, such as that of Weinberg et al. (2017). The interpretation in terms of halo, old disk, thin disk, etc., will be most powerful when combining this information with orbital information. We do not pursue these avenues in this paper, but stress only two points: first, our training set, extended in [M/H] compared to Rix et al. (2022), shows that there are likely hundreds of mostly bright (G < 16) extremely metal-poor giants ([M/H] ≥ −3) and over 10,000 very-metal-poor giants ([M/H] ≥ −2) observable in the Milky Way. Second, the steep slope of the [M/H] distribution argues that spurious [M/H] determinations remain rare also at the lowest accessible metallicities.

Zoom In Zoom Out Reset image size

Since we have an all-sky sample with precise and robust metallicities, it behooves us to make all-sky maps as a function of metallicity to illustrate it. Figures 17 and 18 provide all-sky maps of the sample's number density in various metallicity ranges for two samples: first, the complete (unfiltered) sample of all 174,922,161 stars, and second a vetted sample of 17,558,141 RGB stars. The vetted RGB sample was designed to eliminate spurious [M/H] estimates at the expense of sample size, in particular to eliminate sample contamination among the metal-poor stars that result from unrecognized instances of hotter but reddened stars. After some experimentation, we adopted the following selection criteria illustrated in Figure 16:

• 1.  
  phot_g_mean_mag < 16;
• 2.  
  ϖ/σ_ϖ > 4;
• 3.  
  logg_xgboost < 3.5;
• 4.  
  teff_xgboost < 5200 K;
• 5.  
  M_W1 > −0.3−0.006 × (5500 − teff_xgboost);
• 6.  
  M_W1 > −0.01 × (5300 − teff_xgboost);
• 7.  
  (G − W₂) < 0.2 + 0.77 × (G_BP − W₁);

where ${M}_{W1}={W}_{1}+5\cdot {\mathrm{log}}_{10}(\varpi /100)$.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Only 11,853 of the 2,371,118 of the OBA stars identified by Gaia Collaboration et al. (2022a) that are in our sample pass these quality cuts, i.e., these cuts succeed to eliminate 99.61% of OBA stars from this sample. Notably, Gaia Collaboration et al. (2022a) find that their OBA star sample has some contamination from "halo" stars (i.e., metal-poor stars), which they eliminate kinematically, but cannot eliminate for metal-poor stars with disk-like orbits. Hence, our metal-poor sample may be even purer than the above comparison implies.

For the convenience of the user, this vetted RGB subset is provided as a separate table, as described in Table 2. In this table, we also provide auxiliary information from Gaia about each source's astrometry, photometry, and RVS radial velocity that are available for a substantive fraction of the sample. This provides all the information necessary for the user to compute stellar orbits.

Table 2. Table Description of the 17,558,141 Vetted RGB Results Provided Online (Andrae et al. 2023)

Column name	Description
source_id	GDR3 identifier in ascending order
l	Galactic longitude [deg]
b	Galactic latitude [deg]
ra	R.A. [deg]
dec	decl. [deg]
parallax_corrected	parallax with zero-point correction [mas]
parallax_error	parallax error [mas]
pmra	proper motion R.A. [mas/yr]
pmra_error	error of proper motion R.A. [mas/yr]
pmdec	proper motion decl. [mas/yr]
pmdec_error	error of proper motion decl. [mas/yr]
ruwe	astrometric quality flag
radial_velocity	radial velocity [km/s]
radial_velocity_error	radial velocity error [km/s]
phot_g_mean_mag	apparent G magnitude [mag]
phot_bp_mean_mag	apparent GBP magnitude [mag]
phot_rp_mean_mag	apparent GRP magnitude [mag]
catwise_w1	apparent W1 magnitude [mag]
catwise_w2	apparent W2 magnitude [mag]
mh_xgboost	XGBoost estimate of [M/H]
teff_xgboost	XGBoost estimate of Teff [K]
logg_xgboost	XGBoost estimate of $\mathrm{log}g$
in_training_sample	membership in training sample

Note. We adopt the column names from the Gaia DR3 archive where appropriate. We emphasize that the zero-point correction of Lindegren et al. (2021) has been applied to the parallaxes in this table.

Download table as:  ASCIITypeset image

The three panels of the all-sky maps for the unfiltered sample in Figure 17 clearly show the imprint of the Gaia scanning law: in particular two "crescents" of lower sample density at high latitudes are attributable to too few transits that prevented the publication of XP spectra in Gaia DR3 (see De Angeli et al. 2022, Figure 29 therein). Moreover, dust extinction in the Galactic plane causes many stars to be dimmed below G < 17.65; they may be too faint to have XP spectra published or even too faint to be in the Gaia catalog at all. Note that the extinction in the Galactic plane appears most dramatic among the two metal-poor bins (top and middle panels), as these have far fewer foreground stars. The top map in Figure 17 also shows an implausible set of seemingly metal-poor stars near the disk, preferentially in star-forming regions. Most likely, these objects have spuriously low [M/H] estimates, and actually are reddened OBA stars of presumably higher metallicity that are misinterpreted as metal poor (see Section 3.7).

It is these stars that most immediately show that the full unfiltered sample must contain some spurious [M/H] estimates, which motivated our vetted sample of bright giant stars. Comparison of the two top panels of Figures 17 and 18 shows that these spurious sources are absent in the vetted giant sample.

Apart from these issues, the top maps in Figures 17 and 18 both clearly show the central concentration of metal-poor stars toward the Galactic center, extensively discussed as the "Poor Old Heart of the Milky Way" in Rix et al. (2022). Figure 17 also shows the two Magellanic Clouds, which are missing from Figure 18 due to the cut on parallax quality of $\tfrac{\varpi }{{\sigma }_{\varpi }}\gt 4$.

While we have illustrated only two examples of how to vet or filter the overall table of [M/H] estimates, it is clear that there are further limitations of our stellar parameter estimates that may compromise the use of our catalog. Here, we provide some guidance on potential filtering by the user:

• 1.  
  In Rix et al. (2022), we used a bright RGB sample defined by teff_xgboost < 5300 K, logg_xgboost < 3.5, and G_BP < 16. While this is still possible, we point out that our results in this work also hold for main-sequence dwarfs (see Figure 4(c)). Nevertheless, if the focus is on RGB stars, we recommend to drop or relax the selection on G_BP, given that our results are robust toward the faint end (see Figure 9).
• 2.  
  Given that OBA stars are problematic (see Section 3.7), one could take the golden sample of OBA stars from Gaia Collaboration et al. (2022a) and remove all known OBA stars from the sample.
• 3.  
  The Gaia DR3 publication limit of G = 17.65 for XP spectra was not strict. Rather, XP spectra for 162,686 QSOs and 26,500 galaxies were also published down to the survey detection limit. Furthermore, XP spectra of ultracool dwarfs were published beyond G = 17.65. Given our training sample's temperature range of 3107 K—6867 K, ultracool dwarfs are not covered. Therefore, the user may consider removing all 129,997 results for G > 17.65.
• 4.  
  The user may want to give special consideration to globular clusters, as those represent regions of high source density where the color–color diagram windows assigned to the XP spectra may begin to overlap, thus compromising the XP spectra and our derived [M/H] estimates. This effect was illustrated for Omega Centauri in Figure 27 of Creevey et al. (2022).
• 5.  
  When working with [M/H] for stars of the main sequence, we recommend limiting the color range to 0.5 < G_BP − G_RP < 1.5 for unbiased results (see Figures 10 and 11). The T_eff estimates of stars on the main sequence may be precise over a wider range.
• 6.  
  In order to prevent invalid extrapolations beyond the training sample, the user can check if a source's colors fall within the training sample, as we did for clusters in Section 3.5 and Figure 12. To this end, our catalog contains a column named in_training_sample, which is a Boolean flag that indicates if a source was part of the training sample.