The Stellar Abundances and Galactic Evolution Survey (SAGES). I. General Description and the First Data Release (DR1)

The SAGES project is an international cooperative survey project that utilized four survey telescope facilities worldwide. We aim to observe a large part of the northern sky (except the sky area of the Galactic plane, ∣b∣ > 10°, >12,000 deg², 58.2% of the northern sky) and succeeding in obtaining a large intermediate-band or narrowband photometric catalog of stars much deeper (∼8–9 mag) than those of GCS and HM of the limiting magnitude for constraining the stellar parameters. Thus, we can derive the stellar parameters from our SAGES catalog, and the magnitude coverage can be well combined for the SAGES faint end and GCS bright end. Nightly survey operations are planned by autonomous scheduler software that can execute the entire survey without human intervention.

In our survey, we will carry out the observations in u_s /v_s /g/r/i/α_n /α_w /DDO51 bands. The filters are chosen for the following reasons. We aim to derive the stellar parameters for a large sample of northern sky with photometry, which is similar to the SC system. As mentioned in Section 1, the SAGES u_s is a similar u passband of the SC system, which covers the Balmer jump, and it is sensitive to stellar photospheric gravity. We designed the v_s-band filter, and it covers the Ca ii H and K absorption lines, which are very sensitive to stellar metallicity. The gri bands are the same as the SDSS passbands, which are used to estimate the effective temperature T_eff. The intermediate band DDO51 measures the MgH feature in KM dwarfs (Bessell 2005), which is sensitive to the gravity of late-type stars. The other two bands, α_w and α_n , are used to estimate the interstellar extinctions, as the value of α = α_w − α_n , which we designed, and they are similar to β = β_w − β_n in the SC system (Crawford 1958). It is only sensitive to the effective temperature T_eff, and it is independent of interstellar extinction. Thus, the photometry of the two passbands can be used to estimate the interstellar extinction. However, for the CCD photometry, the quantum efficiency (QE) is much higher in the wavelength of α_w , and the α_w absorption line is stronger for the FGK stars. We will describe the advantages and sensitivity tests in the sections that follow.

Table 1 shows the central wavelength and bandwidths of the filters of the SAGES photometric system. We used the prime focus system of the 2.3 m (90 inch) Bok telescope for observations in the u_s and v_s passbands. The Bok telescope belongs to the Steward Observatory, University of Arizona, which is located at KPNO.

We ordered the SAGES filters from different manufactures: the u_s -band filter on the 90prime telescope was made in the Omega Optical Inc., USA; the v_s filter on the 90prime telescope was made in Asahi Spectra Co., Ltd, Japan, which has very high efficiency; the α_w and DDO51 on the Xuyi 1 m telescope were made in Beijing Bodian Optical Technology Co. Ltd; the α_w and α_n and DDO51 filters of the Maidanak Astronomical Observatory (MAO) 1 m telescope were made in Asahi Spectra Co., Ltd, Japan, which also have very high efficiency; and the gri-band filters on 1 m telescope of Nanshan station of XAO were made in Custom Scientific, Inc., USA, which are the standard SDSS system.

Since FGK-type stars are good tracers of the nature of the Milky Way, we focus on the stellar parameters of FGK-type stars. Figure 1 shows the spectra of the MILES ¹³ of the FGK stars with the filter transmission of the SAGES passbands for FGK-type stars. We see that it is easy to distinguish different types of stars with the color of the SAGES photometry.

Zoom In Zoom Out Reset image size

The advantage of the SAGES filter system is that it is more sensitive to stellar parameters than the traditional SC filter system. We utilized the Kurucz model (Munari et al. 2005) to analyze the dependence of the colors of SAGES and the stellar parameters. We focus on the analysis of the gravity values log g = 2 and 4.5, and the metallicity range of [Fe/H] = −2.5 to 0.5. Figure 2 shows the effective temperature versus colors for different metallicities with two given gravities log g = 2.0 (a) and 4.5 (b). It shows that the SAGES system is clearly much more sensitive to the effective temperature than that of the SC system. We can see that for the same effective temperature range, no matter which type of stars (FGK), the color range of SAGES is ∼2–3 times of that in the SC system. In this case, the uncertainty of the effective temperature T_eff is one-third to one-half that in the SC system, improving the accuracy by a factor of ∼2–3; i.e., the uncertainty of the effective temperature T_eff of a star of V = 15 mag in the SC system is comparable to an uncertainty V ∼ 16–16.5 mag for the same star in the SAGES system (Fan et al. 2018).

Zoom In Zoom Out Reset image size

Further, we have also investigated the relationship between the metallicity and colors of SC system (left panel) and SAGES (right panel). Figure 3 shows the relations between metallicity and colors of the SAGES for different effective temperatures with a given gravity of log g = 2.0 and 4.5. Clearly, the SAGES system has higher sensitivity of the metallicity than that of the SC system. We can see that for the same metallicity range, no matter which type of stars (FGK), the color range of SAGES is ∼2–4 times that of the SC system. In this case, the uncertainty of the metallicity [Fe/H] is 1/4–1/2 that in the SC system, improving the accuracy by a factor of ∼2–4; i.e., the uncertainty of the metallicity [Fe/H] for a star of V = 15 mag in the SC system is comparable to that of v_s ∼ 16–17 mag for the same star in the SAGES system, which can be seen from the spectra of FGK stars shown in Figure 1.

Zoom In Zoom Out Reset image size

Figure 4 presents the relations between the gravity and colors of the SC system and SAGES for different effective temperatures. We consider two cases where the metallicity varies widely, e.g., for the metallicity [Fe/H] = −2.5 and 0. For FG-type stars, the SAGE system is slightly more sensitive than the SC system (as shown in Figure 4). For K-type stars, the SC system has a serious "S-shape," i.e., the nonmonotonic relation. Although the SAGES system also shows a nonmonotonic condition, the relationship between gravity and color is a monotonic function in the interval between log g ≳ 2 and log g ≲ 2 if the distinction is made at log g ≈ 2. We use the DDO51 filter, which can effectively distinguish dwarf stars and giant stars, to provide a feasible solution of gravity log g for K-type stars, which is clearly the advantage of the SAGES photometric system (Fan et al. 2018).

Zoom In Zoom Out Reset image size

Figure 5 shows the relation between the effective temperature T_eff and colors of SAGES for different gravity log g with solar metallicity, where log g = 4.5 for panels (a), (b), (c), and (d), and log g = 2.0 for panels (e), (f), (g), and (h). For comparison, panels (a), (c), (e), and (g) are for the SC system and panels (b), (d), (f), and (e) panels are for SAGES. For G- and K-type stars, the absorption line strength of Hβ is weaker than that of Hα, proving the intensity measurement accuracy of Hα is higher. The relationship between the color index and different extinctions can be obtained by calculation. We take log g = 2 and 4.5 and [Fe/H] = 0. Figure 5 shows the extinction variations in the relations between effective temperature and color: g − I is more sensitive to effective temperature than b − y. For different extinction, β_n − β_w has a certain color change (∼0.01 mag), while for α_n − α_w , the change with extinction is almost invisible (E(B − V) ranges from 0–0.2), suggesting that the latter is more independent of interstellar extinction. Thus, the solution will be more accurate (Fan et al. 2018) for α_n − α_w than that for β_n − β_w .

Zoom In Zoom Out Reset image size

The u_s - and v_s -passband observations were carried out with the 90 inch (2.3 m) Bok telescope at Kitt Peak in Arizona, USA. The prime focus is adopted for the survey, with a corrected focal ratio of f/2.98 and corrected focal length of 6829.2 mm. The altitude of KPNO (111°36'016W, +30°57'465N) is 2071 m. The location offers very stable seeing conditions and a fairly low horizon in all directions, except for the northeast. The sky brightness measurements of Kitt Peak in 2009 were 22.79 mag arcsec⁻² in the B band and 21.95 mag arcsec⁻² in the V band (Neugent & Massey 2010). For the location of the Bok telescope, the typical seeing was 15 during the observations. For the Bok telescope, an 8k × 8k CCD mosaic is composed of four 4k × 4k blue optimized sensitive back luminous CCDs, with gaps along both the R.A. (166'') and decl. directions (54''), by the University of Arizona Imaging Technology Laboratory. In Figure 6, we can see the distribution of the four CCD arrays at the prime focus of the Bok telescope. The QE is ∼80% for the u band (central wavelength of ∼3538 Å with FWHM of 520 Å) and the edge-to-edge FOV is about 108 × 103 (Zou et al. 2015, 2016; Zhou et al. 2016). In this paper, only the observing data of the 2.3 m Bok telescope are released.

Zoom In Zoom Out Reset image size

The observing data of the three telescopes that follow are not included in this paper and will be released in subsequent works.

The gri passband, NOWT of Xinjiang Observatory, CAS is from a 1 m wide-field astronomical telescope with an Alt-Az mount, operating at prime focus with a field corrector. The NOWT provides excellent optical quality, pointing accuracy, and tracking accuracy. It is located at Nanshan (87°1067E, 43°2827N) with an altitude of 2080 m, which is ∼75 km away from Urumqi city. A 4k × 4k CCD was mounted on the prime focus of the telescope. The FOV at prime focus is 1.5° × 1.5°, and pointing accuracy is better than 5'' rms for each axis after pointing model correction (Liu et al. 2013).

The α_n passband and part of the α_w passband are from the 1 m Zeiss telescope of MAO (66°53'47''E, 38°40'22''N), which belongs to Ulugh Beg Astronomical Institute (UBAI), Uzbekistan. UBAI is an observational facility of the Uzbekistan Academy of Sciences. The total amount of clear-night time is 2000 hr with median seeing of 070. The altitude is 2593 m. Based on a 2 month observation performed in 1976, it shows that the sky background of Mount Maidanak varies between 22.3 and 22.9 mag arcsec⁻² in the B band and between 21.4 and 22.0 mag arcsec⁻² in the V band (Kardopolov & Filip'ev 1976). Its current FOV is $32\buildrel{\,\prime}\over{.} 9\times 32\buildrel{\,\prime}\over{.} 9$. For narrowband photometry, if the focal ratio is too fast, the central wavelength and the bandwidth will shift to some extent. For its Cassegrain focus, the focal ratio is f/13. However, in order to enlarge the FOV, we installed a focal reducer to make the focal ratio f/6.5, which is still suitable for the narrowband survey. A 2K × 2K Andor DZ936 CCD was mounted on the f/6.5 Cassegrain focus of the telescope, to ensure that there is no wavelength shift or bandwidth changes for the narrowband filters (Ehgamberdiev et al. 2000; Ehgamberdiev 2018).

The α_w and DDO51 bands are from the Xuyi 1 m Schmidt telescope of the Purple Mountain Observatory of CAS. The Xuyi Schmidt Telescope is a traditional ground-based refractive-reflective telescope with a diameter of 1.04/1.20 m. It is equipped with a 10k × 10k thinned CCD camera, yielding a 3.02° × 3.02° effective FOV at a sampling of 103 per pixel projected on the sky. The QE of the CCD, at the cooled working temperature of −103.45° C, has a peak value of 90% in the blue and remains above 70% even to wavelengths as long as 8000 Å . The XSTPS-GAC was carried out with the SDSS g, r, and i filters. The current work presents measurements of the Xuyi atmospheric extinction coefficients and the night-sky brightness in the three SDSS filters based on the images collected by the XSTPS-GAC. The night-sky brightness determined from images with good quality has median values of 21.7, 20.8, and 20.0 mag arcsec⁻² and reaches 22.1, 21.2, and 20.4 mag arcsec⁻² under the best observing conditions for the g, r, and i bands, respectively (Zhang et al.2013). The typical limiting magnitude is m(r) ∼ 20.0 mag.

Figure 7 shows the survey area (in gray) in equatorial coordinates. SAGES covers most of the northern sky area except the Galactic plane, for which the Galactic latitude ∣b∣ > 10°, and the decl. is from −5° to +90°. The total planned survey area >12,000 deg², which is ∼60% of the Northern Hemisphere.

Zoom In Zoom Out Reset image size

The top panel of Figure 8 presents the sky coverage for SAGES in u_s /v_s observations carried out with the Bok 2.3 m telescope. It covers a sky area of ∼9960 deg², which is actually ∼87.8% of the fields in total observed for SAGES in both passbands. We obtained 23,980 frames for the u_s band and 17,565 frames for the v_s band. Among these images, 36,092 frames have better quality, while the rest frames, which have been excluded, are defined as follows:

• 1.  
  The frames with failed astrometry in data processing have been removed;
• 2.  
  Early testing frames of very short exposure (i.e., 3–6 s) were removed;
• 3.  
  Frames with no LAMOST standard stars have been removed, as have the number of common sources in the surrounding celestial region, which is very small.

Zoom In Zoom Out Reset image size

These "better-quality" frames are involved the flux calibration, which have been released in this paper.

The bottom panel shows the gri passbands of SAGES. So far we have finished the observations for the designed area with the NOWT, which can be combined with the SDSS data for the brighter part of the catalog. The gri-band data of NOWT will be released in subsequent papers.

Figure 9 shows the airmass distribution for SAGES observations, with a median value of 1.11, extending to an airmass of 1.5 (as the airmass limit of our observations was set to 1.5 in the observing strategy). Clearly for most of the fields, the airmass is less than 1.2, which can ensure that our images maintain good quality. This aspect has been incorporated in our survey strategy.

Zoom In Zoom Out Reset image size

For the gri passbands, we have completed the observations for the designed area with the NOWT of XAO, which can be combined with the SDSS data. NOWT gri+SDSS survey imaging coverage is the total area for the initial plan of SAGES. The gri-band data of NOWT are not included in this paper and will be released in subsequent papers. However, after the Pan-STARRS DR1 (PS1) data were released, since the sky coverage is large and the observing is homogeneous, it is a better choice for us to use the PS1 data of the gri band instead of the NOWT gri +SDSS survey data for the current work. Another reason is the PS1 can match our SAGES results better in magnitude range. However, in future work, the NOWT gri data will also be used for stellar parameter estimates.

The raw data of SAGES are processed with the pipeline developed by our data reduction team. The basic corrections of overscan, bias, flat, and crosstalk effect are the same as the data reduction pipelines for most similar surveys.

For the bias correction, we took a set of 10 bias frames before and after the sky survey observations, respectively, every observing night. We constructed the combined master bias frame every night by using the median value of the 20 exposures at the same pixel. Thus both the science images and the flat-field images will be corrected for bias by applying this master bias image, which are not only corrected for the mean value of bias but also for the structure of bias (Zheng et al. 2018, 2019).

For the flat-fielding, we took the dome flats with a screen and a UV lamp to correct the pixel-to-pixel variations. We also took the twilight flats to correct the large-scale illumination trend if possible. Otherwise, a super night-sky flat will be used instead, which is the combination of all of the science images taken over the night. For most nights, the uncertainty of the master flat images was less than 1%.

The highly efficient readout mode can significantly reduce the time spent on reading the detector. However, it induces amplifier crosstalk, which may cause contamination across the output amplifiers at the symmetric place, typically at the level of 1:10,000 of the flux. This effect usually is quite significant for bright saturated stars on the CCD chips. A large number of images have been used to estimate the crosstalk coefficients between amplifiers. The overall ratio of crosstalk is at the level of 5:10,000 for 90Prime, while the inter-CCD crosstalk ratios are greater, and intra-CCD ratios are lower.

Figure 6 is one image in the SAGES v_s -band of the survey. The CCD mosaic is composed of four 4k × 4k blue sensitive back-luminated CCDs. For each CCD, the four-amplifier readout mode is applied, which has a faster readout speed.

As mentioned above in the Section 2.4, we obtained 23,980 frames for the u_s band and 17,565 frames for the v_s band. In total we have 41,545 frames for the u_s and v_s bands. Among these images, we have 36,092 frames with better quality and that involve flux calibration.

In our SAGES pipeline for photometry, we applied SExtractor (Bertin & Armouts 1996) for detecting sources and photometry. The detection threshold is set to 4 (σ above the background rms), which can make sure that most sources could be detected and measured precisely. SExtractor could provide the following measurements and errors: the central positions of each source in both CCD physical coordinates and celestial coordinates, the roundness and sharpness, and instrumental magnitudes (which are included in a series of given apertures).

As we know, for the different photometric methods, the routine will produce different results. We use the software SExtractor MAG_AUTO as our primary output for parameters, as this output is in general reasonable for both point sources and extended sources. In our photometric pipeline, the aperture correction needs to be applied to aperture photometry, which is using the aperture growth-curve method.

In the SAGES photometric calibration, we adopted the "AB system" (Oke & Gunn 1983), which is more commonly used for the photometric system, and is well known for the SDSS (York et al. 2000) by Fukugita et al. (1996). For our calibration, the comparisons show that the SAGES implementation of the AB system has an accuracy of ∼0.02 mag (90% confidence). The dominant contribution is the uncertainty in how well spectrophotometry matches the AB system.

Figure 10 shows the flow chart of the photometry pipeline for SAGES, which uses SExtractor as the main routine for photometry. The bias and overscan are combined for the correction. We use the super-flats as the final flat for the correction, which is more reasonable for our data reduction.

Zoom In Zoom Out Reset image size

Figure 11 shows the photometry uncertainties versus the magnitude of SAGES in the u_s and v_s bands. Here the photometry uncertainties are just from the Poisson statistics, which are not the rms of repeat observations, since for most sources we just observed for only one time. It can be seen that the limiting magnitude is ∼17 mag for both the u_s band and the v_s band of SAGES, for the uncertainties of magnitude 0.01 mag, which correspond to S/N ∼ 100 and are different from the complete magnitude.

Zoom In Zoom Out Reset image size

The astrometric calibration is realized in two steps. First, the pipeline will work out a linear solution assuming the image has no distortion, based on the information stored in image headers, including the telescope pointing coordinates, the rotation angle, and the pixel scale. For further corrections, the pipeline employs the Software for Calibrating AstroMetry and Photometry (SCAMP) package, ¹⁴ which is a computer program that computes astrometric projection parameters. Otherwise, SCAMP is mature and robust software that is widely used in astrometric calibration by matching SExtractor's output catalog with an online or local reference catalog (Bertin & Armouts 1996).

Our pipeline runs SCAMP twice: for the first run, we use a loose criterion for crossmatching detected sources with reference sources, which is in order to maintain enough stars in the image for calibration; for the second run, we applied a more strict criterion to obtain a more precise solution based on the previous results. Table 2 shows the configuration parameters for SCAMP in our pipeline for the two runs. It is found that DISTORT DEGREES = 3 is the most proper configuration after a series of tests (Zheng et al. 2018, 2019).

In our astrometric pipeline, we use the position and proper motion extended (PPMX; Roser et al. 2008) as the astrometric reference, which contains ∼18 million stars. The reference stars are distributed across the whole sky with the astrometric accuracy of ∼002 in both R.A. and decl. (Zheng et al. 2018, 2019).

Although Pan-STARRS DR1 (PS1; Chambers et al. 2016) has an accurate position estimate, i.e., both uncertainties are lower than 0005 in both R.A. and decl. directions, PS1 does not have information on proper motions. Thus, we do not use the PS1 catalog as the astrometric reference, and we use it as the flux reference (see Section 3.4). We use PPMX as the astrometric reference. In fact, in the matching with our SAGES, >85% of PPMX stars are in the magnitude range of 10.0 < V < 15.0 mag. As Gaia DR3 has now been released, we may improve the precision of the astrometric reference in future work. Due to the lack of providing the astrometric residuals, we cannot check if the solution is correct or not. In addition, other than SCAMP, we developed another astrometric calibration method, simple imaging polynomial (SIP; Shupe et al. 2005) for providing an astrometric solution, since SCAMP does not provide fitting errors, which can be used to judge whether or not the results are good enough. Therefore, in the astrometric pipeline, we applied both methods to make sure the solution is correct and accurate.

In our astrometric pipeline, we first convert the original pixel coordinates (x, y) with the origin at the bottom-left corner of one single frame (one chip), to intermediate pixel coordinates (u_s , v_s ) with the origin at the center of the whole FOV. The second step is to convert the (u_s , v_s ) to the intermediate world coordinates (, η) with parameters CD_ij recorded in the fits header. The parameters CD present a matrix which transfers from (u, v) to (, η). Initially, they can be computed by the rotate angle and the pixel scale of the detector. They are provided in headers after the header fixing process. Finally, the intermediate world coordinates (, η) are projected to the world coordinates (α, δ), with a projection type of "TAN." In order to resolve the nonlinear correlation transformation between (, η) and (α, δ), we applied the SIP convention to represent image distortion as it introduces high-order correction polynomials f and g to u_s , v_s to express the distortion, as in Equation (1) (Zheng et al. 2018, 2019) as follows:

$$\begin{eqnarray}&&\left(\begin{array}{c}\epsilon \\ \eta \end{array}\right)=\mathrm{CD}\times \left(\begin{array}{c}u+f(u,v)\\ v+g(u,v)\end{array}\right).\end{eqnarray} \tag{ 1 }$$

In our transform of the astrometry, A_pq and B_pq are used as the coefficients of u^p v^q , as shown in Equation (2) to determine polynomials f and g, in which N_A and N_B are the highest order to correct u_s and v_s . We adopt appropriate parameters NA = NB = 3 for the SAGES.

$$\begin{eqnarray}&&f(u,v)=\displaystyle \sum _{p,q}{A}_{{pq}}\cdot {u}^{p}{v}^{q},2\leqslant p+q\leqslant {N}_{A}\end{eqnarray} \tag{ 2 }$$

$$\begin{eqnarray*}&&g(u,v)=\displaystyle \sum _{p,q}{B}_{{pq}}\cdot {u}^{p}{v}^{q},2\leqslant p+q\leqslant {N}_{B}.\end{eqnarray*}$$

We use multiple visits, no matter if they are in the same band or not, to estimate the differences in coordinates between different visits. A typical internal astrometric error of one image shows that ΔR.A. = 0014 ± 0145 and Δdecl. = −0002 ± 0166. The external astrometric errors are estimated by comparing the difference between the coordinates from our catalog and those from reference catalog PPMX. Figure 12 shows a typical distribution of external astrometric calibration errors in one observing field. It can be seen that the standard deviations are quite small, i.e., ∼01 in both the R.A. and decl. directions, as marked in the lower-left panel, which includes both internal and external astrometric uncertainties in both directions (Zheng et al. 2018, 2019). In future work, this will be improved when Gaia DR3 is employed as a reference catalog.

Zoom In Zoom Out Reset image size

Figure 13 is the flow chart of the SIP for the SAGES astrometry, which adopts the SExtractor results as the input catalog and applies the fits header as the WCS initial value. The task SCAMP is applied twice, and we use the PPMX as the reference catalog.

Zoom In Zoom Out Reset image size

Table 2 shows the configuration parameters of the astrometry table for SCAMP, which has been applied to our data reduction pipeline.

In this work, using the spectroscopic data from LAMOST (Cui et al. 2012; Zhao et al. 2012) DR5, photometric data from the Gaia DR2 (Gaia Collaboration et al. 2016, 2018), and overlapping observations of SAGES, we first perform a relative flux calibration of SAGES DR1 u_s /v_s bands by combining the stellar color regression (SCR; Yuan et al. 2015) method and the ubercalibration method (Padmanabhan et al. 2008). The absolute calibration is then carried out by comparing with synthetic colors from the MILES library (http://research.iac.es/proyecto/miles/; Sanchez-Blazquez et al. 2006).

In fact, Yuan et al. (2015) proposed the spectroscopy-based SCR method to perform precise color calibrations by using millions of spectroscopically observed stars as color standards, with the star-pair technique (Yuan et al. 2013), which considers that stellar colors can be accurately predicted by large-scale spectroscopic surveys, e.g., SDSS and LAMOST.

Combining with the accurate and homogeneous photometric data from Gaia DR2 and Early Data Release 3 (EDR3; Gaia Collaboration et al. 2021a, 2021b), the method can further accurately predict the magnitudes of stars in various passbands and perform high-precision photometric calibration. Such a method has been applied to a number of surveys, including the SDSS Stripe 82 (Yuan et al. 2015; Y. Huang et al. 2022), the SkyMapper Southern Survey DR2 (Huang et al. 2021), Gaia DR2 and EDR3 (Niu et al. 2021a, 2021b), and PS1 DR1 (Xiao & Yuan 2022). A precision of 1 mmag to a few millimagnitudes is usually achieved. A recent review of the method and its implementations can be found in Huang et al. (2022).

The ubercalibration method was originally developed for SDSS, and it achieved a precision of 1% in the griz bands, and 2% in the u band (Padmanabhan et al. 2008). It requires a significant amount of overlapping/repeating observations and assumes that the physical magnitude of the same object under different observational conditions should be the same. Schlafly et al. (2012) applied the method to PS1 catalogs, achieving a precision of better than 1%. The method has also been applied to the Beijing-Arizona Sky Survey (Zou et al. 2017; Zhou et al. 2018), achieving a precision of better than 1%. A detailed summary and discussion of limitations of the method can be found in Huang et al. (2022).

We combine the two aforementioned methods to perform the relative flux calibration of SAGES DR1 u_s /v_s bands. A detailed description of the calibration process will be presented in a separate paper (H. Yuan et al. 2023, in preparation). Here we briefly outline the calibration strategy.

The calibration process of SAGES DR1 is carried out for each gate separately in the u_s and v_s bands. We assume that the relative calibrated magnitude of an object m_rel can be derived from its instrumental magnitude m by

$$\begin{eqnarray}&&\begin{array}{l}{m}_{\mathrm{rel}}\,=\\ m+{zp}\_{frame}(i)+{zp}\_{gate}\_{corr}(i,j)+{flat}\_{corr}({day},j,X,Y),\end{array}\end{eqnarray} \tag{ 3 }$$

where zp_frame(i) is the zero-point of the ith frame, zp_gate_corr(i, j) is the zero-point correction of the jth gate of the ith frame, and flat_corr(day, j, X, Y) is the daily star flat correction of the jth gate and depends on CCD position (X, Y). Taking the u_s band as an example, the detailed calibration strategy is listed as follows:

• (a)  
  Combine the SAGES DR1 photometric data with LAMOST DR5 and Gaia DR2 to create a sample for the flux calibration, which includes data from all three databases above. Select dwarfs as the calibration stars with the following constraints: (1) 5300 < T_eff < 6800 K and −0.8 < [Fe/H] < 0.2, a narrow temperature and metallicity range for robust fitting of stellar colors as a function of stellar atmospheric parameters; and (2) S/N_g of the LAMOST spectra >20. About 1.5 and 1.1 million calibration stars are selected for the u_s and v_s bands, respectively.
• (b)  
  Fit stellar intrinsic color—stellar atmospheric parameters relation ${({G}_{\mathrm{BP}}-{u}_{s})}_{0}$ = f(T_eff, [Fe/H]) using stars from two well-selected neighboring fields, which have more LAMOST targets for calibration. Here f(T_eff, [Fe/H]) is a second-order 2D polynomial with six free parameters. The neighboring fields means that they have similar position and observing time; thus, they should have similar zero-points so that they can be regarded as one field and have more sources.

  $$\begin{eqnarray}&&\begin{array}{l}f({T}_{\mathrm{eff}},{[\mathrm{Fe}/{\rm{H}}])={a}_{0}\cdot {T}_{\mathrm{eff}}^{2}+{a}_{1}\cdot [\mathrm{Fe}/{\rm{H}}]}^{2}\\ \quad +\,{a}_{2}\cdot {T}_{\mathrm{eff}}\cdot [\mathrm{Fe}/{\rm{H}}]+{a}_{3}\cdot {T}_{\mathrm{eff}}+{a}_{4}\cdot [\mathrm{Fe}/{\rm{H}}]+{a}_{5}.\end{array}\end{eqnarray} \tag{ 4 }$$
• (c)  
  Estimate the predicted magnitudes for all of the calibration stars with a typical precision of 2%–3%, and obtain zero-points of each gate of each image file.
• (d)  
  Construct the daily star flat for each gate flat_corr(day, j, X, Y) using second-order 2D polynomials as a function of X and Y.
• (e)  
  After star flat correction, obtain the zero-point of each image file zp_frame(i) and zero-point correction zp_gate_corr(i, j) of each gate and each image file.
• (f)  
  Go to step (b), reconstruct the relation, and iterate.
• (g)  
  Combine the ubercalibration method and the SCR method to further improve the calibration, particularly for the image files with nil or a small number of LAMOST standard stars.
• (h)  
  Finally, perform absolute calibration by comparing the color–color relationship (G_BP − u_s versus G_BP − G_RP) between the observed and synthesized cases from the empirical MILES spectral library.

Reddening correction in steps (b) and (c) is performed using E(BP − RP) and empirical temperature- and reddening-dependent reddening coefficients, both of which are obtained using the star-pair technique similar to that in Sun et al. (2022). Applying the above procedure, we have achieved an internal calibration precision of around 5 mmag for the SAGES DR1 data by comparing repeated observations.

Figure 14 shows the color–magnitude and color–color diagrams of M67 with the SAGES and Gaia DR2 photometry but the DR3 will not change much. It can be seen that our SAGES photometric precision is comparable to that of the Gaia BP/RP. The red dots are the giants and black dots are the dwarfs. We can see that the giants and dwarfs can be separated very well.

Zoom In Zoom Out Reset image size

As in SAGES DR1, we mainly crossmatch the master table with PS1 catalogs, as its observing depth is comparable to our u_s and v_s passband. Although we have the complete gri-passband observations of the 1 m telescope at the Nanshan station of XAO, it is not deep enough. We previously planned to combine our observations of XAO to complement the SDSS imaging survey. However, it is well known that the combination of the two kinds of survey data may not be homogeneous for the observational depth. We also need the transformations of the passband between the Nanshan data and the SDSS data.

Thus, we combined our u_s - and v_s -band data with PS1, which is homogeneous and has large enough sky coverage, i.e., ∼3/4 of the total sky. For large photometric catalogs, we determine the matching external object and record its ID and projected distance within the master table. In our work, we match in the reverse direction and distance in the external catalog as DR1-ID. However, the proper-motion information is not included in the catalog. The maximum distance for all crossmatches is 2'', motivated by the region in which our 1D point-spread function (PSF) magnitudes may be affected.

Such a matching process has resulted in a catalog with half of the original sources. There are 39,857,439 sources and 42,457,643 sources left in the u_s and v_s passbands, respectively, after combining with the PS1 catalog.

The SAGES DR1 contains total photometry from a total of 36,092 image exposures, including 23,980 images in the u_s passband and 17,565 images in the v_s passband. In total, we have 41,545 frames for the u_s and v_s bands. Among these images, 36,092 frames have high quality and involve flux calibration. After combining with the PS1 catalog, the SAGES master catalog thus contains a total number of 48,553,987 sources, which are released in this paper.

4.1.1. The SAGES Master Table

4.1.2. Limitation of Auto- and Aperture Magnitudes

Figure 15 shows the magnitude distribution of the SAGES u_s and v_s bands for all of the sources of the SAGES DR1 catalog in all of our observing fields, including sources fainter than the "limiting magnitude." We can see that for the u_s band the limiting magnitude is around 20 mag, while for the v_s band, the limiting magnitude is around 21 mag.

Zoom In Zoom Out Reset image size

Figures 16–19 are the distributions of observing depth (i.e., limiting magnitude of different S/N and complete magnitude, which is defined as the turnover in the magnitude histogram) of the SAGES survey for the best magnitude for the u_s /v_s passbands.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Figures 20–21 show the distributions of the complete magnitude (which is defined as the turnover in the magnitude histogram) in the SAGES u_s and v_s passbands. The median values are u_s ∼ 20.4 mag and v_s ∼ 20.3 mag.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

It is also possible that the photometry table does not reveal entries for an object in an image, which is clearly visible in the image in question. Unless SExtractor overlooked the object, given the parameters we chose, this should only happen when SExtractor extracted two objects in this image while they are considered children of a single parent object for this filter. If all images within one filter show multiple objects for what is taken to be a single merged object, all measurements for this object will be excluded from the photometry table, and hence no distilled summary photometry for this filter shall be included in the master table. Table 4 shows the meanings of the Merge Bits for each source in Table 3.

Table 4. Contents and Descriptions of the SAGES Merge Bits Table

Bits	Represents Meaning
0	Detected only once
1	Has other close objects but rejected while merging
2	Flux not good enough, at least one source is beyond 3σ
3	Position not good enough, at least one source beyond 3σ
4–27	Reserved
28	No matched object
29–31	Reserved

Download table as:  ASCIITypeset image

Finally, we have obtained a master catalog (please see Table 5) for the SAGES DR1, which includes 93,293,120 items in the u_s band and 91,185,024 items in the v_s band. After combing with PS1, the numbers are reduced to 48,553,987 in our detection u_s or v_s band. The SAGES DR1 data set will be available to the community through the China-VO platform and the National Astronomical Data Center. ¹⁵ The release can be browsed at the website, either in color or in each SAGES per passband.

The catalog structure includes the following:

• 1.  
  The crossmatching to external catalogs now includes SAGES DR1 and PS1 DR1. Further, in the DR2 version, it will include the information of LAMOST and Gaia DR3, as well as the Two Micron Ally Sky Survey (2MASS), AllWISE, GALEX, and UCAC4. The LAMOST catalog contains the stellar parameters of gravity, metallicity, and temperature, while the Gaia DR3 contains photometry of the G, GB, and GR bands as well as parallax and other information.
• 2.  
  For the SAGES u_s and v_s passbands, the photometry table also includes a column that shows if the sources have been measured one, two, or four times for the observations, as overlapped with the adjacent fields. We only provide the combined magnitude in DR1. Thus for the exposure of two or four times, the photometry uncertainties are calculated as the variations of the magnitudes for different times. However, in the next version, the individual measurements of photometry may also be released.