13.1.1 gaia_source
This table has an entry for every Gaia observed source as listed in the Main Database accumulating catalogue version from which the catalogue release has been generated. It contains the basic source parameters, that is only final data (no epoch data) and no spectra (neither final nor epoch).
Columns description:
All Gaia data processed by the Data Processing and Analysis Consortium comes tagged with a solution identifier. This is a numeric field attached to each table row that can be used to unequivocally identify the version of all the subsystems that where used in the generation of the data as well as the input data used. It is mainly for internal DPAC use but is included in the published data releases to enable end users to examine the provenance of processed data products. To decode a given solution ID visit https://gaia.esac.esa.int/decoder/solnDecoder.jsp
A source designation, unique across all Gaia Data Releases, that is constructed from the prefix “Gaia DRx ” followed by a string of digits corresponding to source_id (3 space–separated words in total). Note that the integer source identifier source_id is not guaranteed to be unique across Data Releases; moreover it is not guaranteed that the same astronomical source will always have the same source_id in different Data Releases. Hence the only safe way to compare source records between different Data Releases in general is to check the records of proximal source(s) in the same small part of the sky.
A unique numerical identifier of the source, encoding the approximate position of the source (roughly to the nearest arcmin), the provenance (data processing centre where it was created), a running number, and a component number.
The approximate equatorial (ICRS) position is encoded using the nested HEALPix scheme at level 12 (Nside = 4096), which divides the sky into $\simeq 200$ million pixels of about 0.7 arcmin${}^{2}$.
The source ID consists of a 64bit integer, least significant bit = 1 and most significant bit = 64, comprising:

•
a HEALPix index number (sky pixel) in bits 36  63; by definition the smallest HEALPix index number is zero.

•
a 3bit Data Processing Centre code in bits 33  35; for example MOD(source_id / 4294967296, 8) can be used to distinguish between sources initialised via the Initial Gaia Source List by the Torino DPC (code = 0) and sources otherwise detected and assigned by Gaia observations (code $>0$)

•
a 25bit plus 7 bit sequence number within the HEALPix pixel in bits 1  32 split into:

–
a 25 bit running number in bits 8 – 32; the running numbers are defined to be positive, i.e. never zero

–
a 7bit component number in bits 1 – 7

–
This means that the HEALpix index at level 12 of a given source is contained in the most significant bits. HEALpix index of level 12 and lower can thus be retrieved as follows:

•
HEALpix index at level 12 = source_id / 34359738368

•
HEALpix index at level 11 = source_id / 137438953472

•
HEALpix index level 10 = source_id / 549755813888

•
HEALpix index at level $n$ = source_id / $({2}^{35}\times {4}^{(12n)})$ = source_id / ${2}^{(592n)}$
Additional details can be found in the Gaia DPAC public document Source Identifiers — Assignment and Usage throughout DPAC (document code GAIA–C3–TN–ARI–BAS–020) available from https://www.cosmos.esa.int/web/gaia/publicdpacdocuments
Random index which can be used to select smaller subsets of the data that are still representative. The column contains a random permutation of the numbers from 0 to $N1$, where $N$ is the number of sources in the table.
The random index can be useful for validation (testing on 10 different random subsets), visualization (displaying 1% of the data), and statistical exploration of the data, without the need to download all the data.
Reference epoch to which the astrometric source parameters are referred, expressed as a Julian Year in TCB.
Barycentric right ascension $\alpha $ of the source in ICRS at the reference epoch ref_epoch
Standard error ${\sigma}_{\alpha *}\equiv {\sigma}_{\alpha}\mathrm{cos}\delta $ of the right ascension of the source in ICRS at the reference epoch ref_epoch.
Barycentric declination $\delta $ of the source in ICRS at the reference epoch ref_epoch
Standard error ${\sigma}_{\delta}$ of the declination of the source in ICRS at the reference epoch ref_epoch
Absolute stellar parallax $\varpi $ of the source at the reference epoch ref_epoch
Standard error ${\sigma}_{\varpi}$ of the stellar parallax at the reference epoch ref_epoch
Parallax divided by its standard error
The total proper motion calculated as the magnitude of the resultant vector of the proper motion component vectors pmra and pmdec, i.e. ${\text{\U0001d699\U0001d696}}^{2}={\text{\U0001d699\U0001d696\U0001d69b\U0001d68a}}^{2}+{\text{\U0001d699\U0001d696\U0001d68d\U0001d68e\U0001d68c}}^{2}$.
Proper motion in right ascension ${\mu}_{\alpha *}\equiv {\mu}_{\alpha}\mathrm{cos}\delta $ of the source in ICRS at the reference epoch ref_epoch. This is the local tangent plane projection of the proper motion vector in the direction of increasing right ascension.
pmra_error : Standard error of proper motion in right ascension direction (float, Angular Velocity[mas/year] )
Standard error ${\sigma}_{\mu \alpha *}$ of the local tangent plane projection of the proper motion vector in the direction of increasing right ascension at the reference epoch ref_epoch
Proper motion in declination ${\mu}_{\delta}$ of the source at the reference epoch ref_epoch. This is the projection of the proper motion vector in the direction of increasing declination.
pmdec_error : Standard error of proper motion in declination direction (float, Angular Velocity[mas/year] )
Standard error ${\sigma}_{\mu \delta}$ of the proper motion component in declination at the reference epoch ref_epoch
ra_dec_corr : Correlation between right ascension and declination (float, Dimensionless[see description])
Correlation coefficient $\rho (\alpha ,\delta )$ between right ascension and declination, a dimensionless quantity in the range [1,+1]
ra_parallax_corr : Correlation between right ascension and parallax (float, Dimensionless[see description])
Correlation coefficient $\rho (\alpha ,\varpi )$ between right ascension and parallax, a dimensionless quantity in the range [1,+1]
ra_pmra_corr : Correlation between right ascension and proper motion in right ascension (float, Dimensionless[see description])
Correlation coefficient $\rho (\alpha ,{\mu}_{\alpha *})$ between right ascension and proper motion in right ascension, a dimensionless quantity in the range [1,+1]
ra_pmdec_corr : Correlation between right ascension and proper motion in declination (float, Dimensionless[see description])
Correlation coefficient $\rho (\alpha ,{\mu}_{\delta})$ between right ascension and proper motion in declination, a dimensionless quantity in the range [1,+1]
dec_parallax_corr : Correlation between declination and parallax (float, Dimensionless[see description])
Correlation coefficient $\rho (\delta ,\varpi )$ between declination and parallax, a dimensionless quantity in the range [1,+1]
dec_pmra_corr : Correlation between declination and proper motion in right ascension (float, Dimensionless[see description])
Correlation coefficient $\rho (\delta ,{\mu}_{\alpha *})$ between declination and proper motion in right ascension, a dimensionless quantity in the range [1,+1]
dec_pmdec_corr : Correlation between declination and proper motion in declination (float, Dimensionless[see description])
Correlation coefficient $\rho (\delta ,{\mu}_{\delta})$ between declination and proper motion in declination, a dimensionless quantity in the range [1,+1]
parallax_pmra_corr : Correlation between parallax and proper motion in right ascension (float, Dimensionless[see description])
Correlation coefficient $\rho (\varpi ,{\mu}_{\alpha *})$ between parallax and proper motion in right ascension, a dimensionless quantity in the range [1,+1]
parallax_pmdec_corr : Correlation between parallax and proper motion in declination (float, Dimensionless[see description])
Correlation coefficient $\rho (\varpi ,{\mu}_{\delta})$ between parallax and proper motion in declination, a dimensionless quantity in the range [1,+1]
pmra_pmdec_corr : Correlation between proper motion in right ascension and proper motion in declination (float, Dimensionless[see description])
Correlation coefficient $\rho ({\mu}_{\alpha *},{\mu}_{\delta})$ between proper motion in right ascension and proper motion in declination, a dimensionless quantity in the range [1,+1]
Total number of AL observations (= CCD transits) used in the astrometric solution of the source, independent of their weight. Note that some observations may be strongly downweighted (see
astrometric_n_bad_obs_al).
Total number of AC observations (= CCD transits) used in the astrometric solution of the source, independent of their weight (note that some observations may be strongly downweighted). Nearly all sources having G $$ will have AC observations from 2d windows, while fainter than that limit only $\sim 1$% of transit observations (the so–called ‘calibration faint stars’) are assigned 2d windows resulting in AC observations.
Number of AL observations (= CCD transits) that were not strongly downweighted in the astrometric solution of the source. Strongly downweighted observations (with downweighting factor $$) are instead counted in astrometric_n_bad_obs_al. The sum of astrometric_n_good_obs_al and astrometric_n_bad_obs_al equals
astrometric_n_obs_al, the total number of AL observations used in the astrometric solution of the source.
Number of AL observations (= CCD transits) that were strongly downweighted in the astrometric solution of the source, and therefore contributed little to the determination of the astrometric parameters. An observation is considered to be strongly downweighted if its downweighting factor $$, which means that the absolute value of the astrometric residual exceeds 4.83 times the total uncertainty of the observation, calculated as the quadratic sum of the centroiding uncertainty, excess source noise, and excess attitude noise.
Goodnessoffit statistic of the astrometric solution for the source in the alongscan direction. This is the ‘gaussianized chisquare’, which for good fits should approximately follow a normal distribution with zero mean value and unit standard deviation. Values exceeding, say, $+3$ thus indicate a bad fit to the data.
This statistic is computed according to the formula
$\mathtt{\text{astrometric\_gof\_al}}={(9\nu /2)}^{1/2}[{\text{\U0001d69b\U0001d69e\U0001d6a0\U0001d68e}}^{2/3}+2/(9\nu )1]$
where ruwe is the renormalised unit weight error and
$\nu =\mathtt{\text{astrometric\_n\_good\_obs\_al}}N$
is the number of degrees of freedom for a source update. Here $N=5$ for 2parameter and 5parameter solutions (respectively astrometric_params_solved = 3 or 31) and 6 for 6parameter solutions (astrometric_params_solved = 95).
Note that only ‘good’ (i.e. not strongly downweighted) observations are included in $\nu $. For further details please see Lindegren et al. (2020).
Astrometric goodnessoffit (${\chi}^{2}$) in the AL direction.
${\chi}^{2}$ values were computed for the ‘good’ AL observations of the source, without taking into account the astrometric_excess_noise (if any) of the source. They do however take into account the attitude excess noise (if any) of each observation.
This is the excess noise ${\u03f5}_{i}$ of the source. It measures the disagreement, expressed as an angle, between the observations of a source and the bestfitting standard astrometric model (using five astrometric parameters). The assumed observational noise in each observation is quadratically increased by ${\u03f5}_{i}$ in order to statistically match the residuals in the astrometric solution. A value of 0 signifies that the source is astrometrically wellbehaved, i.e. that the residuals of the fit statistically agree with the assumed observational noise. A positive value signifies that the residuals are statistically larger than expected.
The significance of ${\u03f5}_{i}$ is given by astrometric_excess_noise_sig ($D$). If $D\le 2$ then ${\u03f5}_{i}$ is probably not significant, and the source may be astrometrically wellbehaved even if ${\u03f5}_{i}$ is large.
The excess noise ${\u03f5}_{i}$ may absorb all kinds of modelling errors that are not accounted for by the observational noise (image centroiding error) or the excess attitude noise. Such modelling errors include LSF and PSF calibration errors, geometric instrument calibration errors, and part of the highfrequency attitude noise. These modelling errors are particularly important in the early data releases, but should decrease as the astrometric modelling of the instrument and attitude improves over the years.
Additionally, sources that deviate from the standard fiveparameter astrometric model (e.g. unresolved binaries, exoplanet systems, etc.) may have positive ${\u03f5}_{i}$. Given the many other possible contributions to the excess noise, the user must study the empirical distributions of ${\u03f5}_{i}$ and $D$ to make sensible cutoffs before filtering out sources for their particular application.
The excess source noise is further explained in Sects. 3.6 and 5.1.2 of
Lindegren et al. (2012).
A dimensionless measure ($D$) of the significance of the calculated astrometric_excess_noise (${\u03f5}_{i}$). A value $D>2$ indicates that the given ${\u03f5}_{i}$ is probably significant.
For good fits in the limit of a large number of observations, $D$ should be zero in half of the cases and approximately follow the positive half of a normal distribution with zero mean and unit standard deviation for the other half. Consequently, $D$ is expected to be greater than 2 for only a few percent of the sources with wellbehaved astrometric solutions.
In the early data releases ${\u03f5}_{i}$ will however include instrument and attitude modelling errors that are statistically significant and could result in large values of ${\u03f5}_{i}$ and $D$. The user must study the empirical distributions of these statistics and make sensible cutoffs before filtering out sources for their particular application.
The excess noise significance is further explained in Sect. 5.1.2 of
Lindegren et al. (2012).
The seven bits of astrometric_params_solved indicate which parameters have been estimated in AGIS for this source. A set bit means the parameter was updated, an unset bit means the parameter was not updated. The leastsignificant bit corresponds to ra. The table below shows the values of astrometric_params_solved for relevant combinations of the parameters.
The radial proper motion (${\mu}_{r}$) is formally considered to be one of the astrometric parameters of a source, and the sixth bit is therefore reserved for it. It is also in principle updatable in AGIS, but in practice it will always be computed from a spectroscopic radial velocity and the estimated parallax, in which case the bit is not set.
$C$ is the pseudocolour of the source, i.e. the astrometrically estimated effective wavenumber.
astrometric_params_solved  ra  dec  parallax  pmra  pmdec  ${\mu}_{r}$  $C$ 
${0000011}_{2}=3$  ✓  ✓  
${0000111}_{2}=7$  ✓  ✓  ✓  
${0011011}_{2}=27$  ✓  ✓  ✓  ✓  
${0011111}_{2}=31$  ✓  ✓  ✓  ✓  ✓  
${0111111}_{2}=63$  ✓  ✓  ✓  ✓  ✓  ✓  
${1011111}_{2}=95$  ✓  ✓  ✓  ✓  ✓  ✓ 
In practice all the sources in DR3 have only values of 3, 31 or 95 for the $\mathrm{\U0001d68a\U0001d69c\U0001d69d\U0001d69b\U0001d698\U0001d696\U0001d68e\U0001d69d\U0001d69b\U0001d692\U0001d68c}\mathrm{\_}\mathrm{\U0001d699\U0001d68a\U0001d69b\U0001d68a\U0001d696\U0001d69c}\mathrm{\_}\mathrm{\U0001d69c\U0001d698\U0001d695\U0001d69f\U0001d68e\U0001d68d}$, corresponding to twoparameter (position), fiveparameter (position, parallax, and proper motion) and sixparameter (position, parallax, proper motion and astrometrically estimated effective wavenumber) solutions.
Flag indicating if this source was used as a primary source (true) or secondary source (false). Only primary sources contribute to the estimation of attitude, calibration, and global parameters. The estimation of source parameters is otherwise done in exactly the same way for primary and secondary sources.
nu_eff_used_in_astrometry : Effective wavenumber of the source used in the astrometric solution (float, Misc[$\mu {m}^{1}$])
Effective wavenumber of the source, ${\nu}_{\text{eff}}$, in $\mu $m${}^{1}$.
This ${\nu}_{\text{eff}}$ is the value used in the image parameter determination and in the astrometric calibration if reliable mean BP and RP photometry were available. It is the photonflux weighted inverse wavelength, as estimated from the BP and RP bands. The field is provided for astrometric solutions with five parameters but is empty for those with two or six parameters.
Due to cyclic processing of the astrometry and the photometry, this effective wavenumber might be different from the one computed using the latest available photometry. Moreover, if no reliable photometry was available at the time of the astrometric processing, this field is empty and an astrometrically estimated value of the effective wavenumber may instead be given in the pseudocolour field.
Effective wavenumber of the source estimated in the final astrometric processing.
The pseudocolour is the astrometrically estimated effective wavenumber of the photon flux distribution in the astrometric ($G$) band, measured in $\mu $m${}^{1}$. The value in this field was estimated from the chromatic displacements of image centroids, calibrated by means of the photometrically determined effective wavenumbers (${\nu}_{\text{eff}}$) of primary sources.
The field is empty when chromaticity was instead taken into account using the photometrically determined ${\nu}_{\text{eff}}$ given in the field nu_eff_used_in_astrometry.
Standard error ${\sigma}_{\mathrm{\U0001d699\U0001d69c\U0001d68e\U0001d69e\U0001d68d\U0001d698\U0001d68c\U0001d698\U0001d695\U0001d698\U0001d69e\U0001d69b}}$ of the astrometrically determined pseudocolour of the source.
ra_pseudocolour_corr : Correlation between right ascension and pseudocolour (float, Dimensionless[see description])
Correlation coefficient $\rho (\alpha ,\mathrm{\U0001d699\U0001d69c\U0001d68e\U0001d69e\U0001d68d\U0001d698\U0001d68c\U0001d698\U0001d695\U0001d698\U0001d69e\U0001d69b})$ between right ascension ra and pseudocolour, a dimensionless quantity in the range [1,+1]
dec_pseudocolour_corr : Correlation between declination and pseudocolour (float, Dimensionless[see description])
Correlation coefficient $\rho (\delta ,\mathrm{\U0001d699\U0001d69c\U0001d68e\U0001d69e\U0001d68d\U0001d698\U0001d68c\U0001d698\U0001d695\U0001d698\U0001d69e\U0001d69b})$ between declination dec and pseudocolour, a dimensionless quantity in the range [1,+1]
parallax_pseudocolour_corr : Correlation between parallax and pseudocolour (float, Dimensionless[see description])
Correlation coefficient $\rho (\varpi ,\mathrm{\U0001d699\U0001d69c\U0001d68e\U0001d69e\U0001d68d\U0001d698\U0001d68c\U0001d698\U0001d695\U0001d698\U0001d69e\U0001d69b})$ between parallax and pseudocolour, a dimensionless quantity in the range [1,+1]
pmra_pseudocolour_corr : Correlation between proper motion in right asension and pseudocolour (float, Dimensionless[see description])
Correlation coefficient $\rho ({\mu}_{\alpha *},\mathrm{\U0001d699\U0001d69c\U0001d68e\U0001d69e\U0001d68d\U0001d698\U0001d68c\U0001d698\U0001d695\U0001d698\U0001d69e\U0001d69b})$ between proper motion in right ascension pmra and pseudocolour, a dimensionless quantity in the range [1,+1]
pmdec_pseudocolour_corr : Correlation between proper motion in declination and pseudocolour (float, Dimensionless[see description])
Correlation coefficient $\rho ({\mu}_{\delta},\mathrm{\U0001d699\U0001d69c\U0001d68e\U0001d69e\U0001d68d\U0001d698\U0001d68c\U0001d698\U0001d695\U0001d698\U0001d69e\U0001d69b})$ between proper motion in declination pmdec and pseudocolour, a dimensionless quantity in the range [1,+1]
The number of field–of–view transits matched to this source, counting only the transits containing CCD observations actually used to compute the astrometric solution.
This number will always be equal to or smaller than matched_transits, the difference being the FOV transits that were not used in the astrometric solution because of bad data or excluded time intervals.
Number of visibility periods used in the astrometric solution.
A visibility period is a group of observations separated from other groups by a gap of at least 4 days. A source may have from one to tens of field–of–view transits in a visibility period, but with a small spread in time, direction of scanning, and parallax factor. From one visibility period to the next these variables have usually changed significantly. A high number of visibility periods is therefore a better indicator of an astrometrically well–observed source than a large number of field–of–view transits (matched_transits or astrometric_matched_transits) or CCD observations (astrometric_n_obs_al). A small value (e.g. less than 10) indicates that the calculated parallax could be more vulnerable to errors, e.g. from the calibration model, not reflected in the formal uncertainties. See Lindegren et al. (2018) for a discussion of this and other astrometric quality indicators.
astrometric_sigma5d_max : The longest semimajor axis of the 5d error ellipsoid (float, Angle[mas])
The longest principal axis in the 5dimensional error ellipsoid.
This is a 5dimensional equivalent to the semimajor axis of the position error ellipse and is therefore useful for filtering out cases where one of the five parameters, or some linear combination of several parameters, is particularly illdetermined. It is measured in mas and computed as the square root of the largest singular value of the scaled $5\times 5$ covariance matrix of the astrometric parameters. The matrix is scaled so as to put the five parameters on a comparable scale, taking into account the maximum alongscan parallax factor for the parallax and the time coverage of the observations for the proper motion components. If $C$ is the unscaled covariance matrix, the scaled matrix is $SCS$, where $S=\text{diag}(1,1,\mathrm{sin}\xi ,T/2,T/2)$, $\xi ={45}^{\circ}$ is the solar aspect angle in the nominal scanning law, and $T$ the time coverage of the data used in the solution.
astrometric_sigma5d_max is given for all the solutions, as its size is one of the criteria for accepting or rejecting the 5 or 6parameter solution. In case of a 2 parameter solution (astrometric_params_solved = 3) it gives the value for the rejected 5 or 6parameter solution, and can then be arbitrarily large.
The total number of field–of–view transits matched to this source.
new_matched_transits : The number of transits newly incorporated into an existing source in the current cycle (short)
Individual field–of–view transits are crossmatched into unique sources at the start of each reprocessing cycle taking the source list from the previous cycle as a starting point. During that process a combination of appending, merging and splitting operations is performed to create a more complete and reliable map of unique sources given the available information. Existing individual sources may accrete further transits, may be merged into fewer unique sources, or may split into two or more new, unique sources as more measurements are accumulated. Field new_matched_transits logs the number of transits newly appended to an existing source during the most recent cyclic reprocessing crossmatch. It refers exclusively to the source_id.
matched_transits_removed : The number of transits removed from an existing source in the current cycle (short)
Individual field–of–view transits are crossmatched into unique sources at the start of each reprocessing cycle taking the source list from the previous cycle as a starting point. During that process a combination of appending, merging and splitting operations is performed to create a more complete and reliable map of unique sources given the available information. Existing individual sources may accrete further transits, may be merged into fewer unique sources, or may split into two or more new, unique sources as more measurements are accumulated. Field matched_transits_removed logs the number of transits removed during the most recent cyclic reprocessing crossmatch from those allocated to an existing source during all previous cycles. It refers exclusively to the source_id.
ipd_gof_harmonic_amplitude : Amplitude of the IPD GoF versus position angle of scan (float, Dimensionless[see description])
This statistic measures the amplitude of the variation of the IPD GoF (reduced chisquare) as function of the position angle of the scan direction. A large amplitude indicates that the source is double, in which case the phase indicates the position angle of the pair modulo 180 degrees. The quantity was computed using only transits used in the astrometric solution, for example those without the EPSL and without outliers.
Let $\psi $ be the position angle of the scan direction. The following expression is fitted to the IPD GoF for all the AF observations of the source:
$$\mathrm{ln}(\mathrm{GoF})={c}_{0}+{c}_{2}\mathrm{cos}(2\psi )+{s}_{2}\mathrm{sin}(2\psi )$$ 
The amplitude and phase of the variation are calculated as
$$\mathtt{\text{ipd\_gof\_harmonic\_amplitude}}=\sqrt{{c}_{2}^{2}+{s}_{2}^{2}}$$ 
$$\mathtt{\text{ipd\_gof\_harmonic\_phase}}=\frac{1}{2}\mathrm{atan2}({s}_{2},{c}_{2})\mathit{\hspace{1em}}(+{180}^{\circ})$$ 
where atan2 returns the angle in degrees, and 180 is added for negative values.
This statistic measures the phase of the variation of the IPD GoF (reduced chisquare) as function of the position angle of the scan direction. The quantity was computed using only transits used in the astrometric solution, for example those without the EPSL and without outliers. See the description of parameter ipd_gof_harmonic_amplitude for further details.
This field provides information on the raw windows used for the astrometric processing of this source coming from the Image Parameters Determination (IPD) module in the core processing. It provides the fraction of windows (having a successful IPD result), as percentage (from 0 to 100), for which the IPD algorithm has identified a double peak, meaning that the detection may be a visually resolved double star (either just visual double or real binary). The quantity was computed using all transits where the IPD was successful.
This field is calculated during AGIS and provides information on the raw windows used for the astrometric processing of this source. It provides the fraction (as a percentage, from 0 to 100) of transits having either truncation or multiple gates flagged in one or more windows. Such a situation invariably means that the onboard VPU detected some nearby source (which may be just a spurious detection, but typically could be some real nearby source — having another distinct transit and most probably assigned to a different source). So in general a non–zero fraction indicates that this source may be contaminated by another nearby source. The quantity was computed using all transits where the IPD was successful.
The Renormalised Unit Weight Error is computed as
$$\text{\U0001d69b\U0001d69e\U0001d6a0\U0001d68e}=\frac{\sqrt{\mathtt{\text{astrometric\_chi2\_al}}/(\mathtt{\text{astrometric\_n\_good\_obs\_al}}m)}}{f(G,{G}_{\text{BP}}{G}_{\text{RP}})}$$ 
where $m$ is the number of parameters solved (the number of set bits in paramsSolved) and $f$ is a renormalising function.
In practice $f$ is determined in an offline statistical analysis of the secondary solutions — see for example ‘Renormalising the astrometric chisquare in Gaia DR2’ (Lindegren 2018). Also note that this value is set to null for sources with only a twoparameter solution, since this value would be difficult to interpret in such cases.
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 quantify the distribution of AL scan directions across the source. scan_direction_strength_k1 (and similarly 2,3,4) are the absolute value of the trigonometric moments ${m}_{k}=\u27e8\mathrm{exp}(ik\theta )\u27e9$ for $k=1,2,3,4$ where $\theta $ is the position angle of the scan and the mean value is taken over the astrometric_n_good_obs_al observations contributing to the astrometric parameters of the source. $\theta $ is defined in the usual astronomical sense: $\theta =0$ when the FoV is moving towards local North, and $\theta ={90}^{\circ}$ towards local East.
N.B. When astrometric_n_obs_ac $>0$ the scan direction attributes are not provided at Gaia EDR3. Hence for all sources brighter than G $\approx 13$, and for a tiny fraction of fainter sources ($\approx 1$%), these 8 scan direction fields will be NULL.
The scan_direction_strength_k1…4 are numbers between 0 and 1, where 0 means that the scan directions are well spread out in different directions, while 1 means that they are concentrated in a single direction (given by the corresponding scan_direction_mean_k1…4).
The different orders $k$ are statistics of the scan directions modulo ${360}^{\circ}/k$. For example, at first order ($k=1$), $\theta ={10}^{\circ}$ and $\theta ={190}^{\circ}$ count as different directions, but at second order ($k=2$) they are the same. Thus, scan_direction_strength_k1 is the degree of concentration when the sense of direction is taken into account, while scan_direction_strength_k2 is the degree of concentration without regard to the sense of direction. A large value of scan_direction_strength_k4 indicates that the scans are concentrated in two nearly orthogonal directions.
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 attributes quantify the distribution of AL scan directions across the source.
See the description for attribute scan_direction_strength_k1 for further details.
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 attributes quantify the distribution of AL scan directions across the source.
See the description for attribute scan_direction_strength_k1 for further details.
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 attributes quantify the distribution of AL scan directions across the source.
See the description for attribute scan_direction_strength_k1 for further details.
scan_direction_mean_k1 : Mean position angle of scan directions across the source (float, Angle[deg])
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 attributes quantify the distribution of AL scan directions across the source. scan_direction_mean_k1 (and similarly for $k=2,3,4$) is $1/k$ times the argument of the trigonometric moments ${m}_{k}=\u27e8\mathrm{exp}(ik\theta )\u27e9$, where $\theta $ is the position angle of the scan and the mean value is taken over the astrometric_n_good_obs_al observations contributing to the astrometric parameters of the source. $\theta $ is defined in the usual astronomical sense: $\theta =0$ when the FoV is moving towards local North, and $\theta ={90}^{\circ}$ towards local East.
N.B. When astrometric_n_obs_ac $>0$ the scan direction attributes are not provided at Gaia EDR3. Hence for all sources brighter than G $\approx 13$, and for a tiny fraction of fainter sources ($\approx 1$%), these 8 scan direction fields will be NULL.
scan_direction_mean_k1 (and similarly for $k=2,3,4$) is an angle between ${180}^{\circ}/k$ and $+{180}^{\circ}/k$, giving the mean position angle of the scans at order $k$.
The different orders $k$ are statistics of the scan directions modulo ${360}^{\circ}/k$. For example, at first order ($k=1$), $\theta ={10}^{\circ}$ and $\theta ={190}^{\circ}$ count as different directions, but at second order ($k=2$) they are the same. Thus, scan_direction_mean_k1 is the mean direction when the sense of direction is taken into account, while scan_direction_mean_k2 is the mean direction without regard to the sense of the direction. For example, scan_direction_mean_k1 = 0 means that the scans preferentially go towards North, while scan_direction_mean_k2 = 0 means that they preferentially go in the NorthSouth direction, and scan_direction_mean_k4 = 0 that they preferentially go either in the NorthSouth or in the EastWest direction.
scan_direction_mean_k2 : Mean position angle of scan directions across the source (float, Angle[deg])
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 attributes quantify the distribution of AL scan directions across the source.
See the description for attribute scan_direction_mean_k1 for further details.
scan_direction_mean_k3 : Mean position angle of scan directions across the source (float, Angle[deg])
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 attributes quantify the distribution of AL scan directions across the source.
See the description for attribute scan_direction_mean_k1 for further details.
scan_direction_mean_k4 : Mean position angle of scan directions across the source (float, Angle[deg])
The scan_direction_strength_k1…4 and scan_direction_mean_k1…4 attributes quantify the distribution of AL scan directions across the source.
See the description for attribute scan_direction_mean_k1 for further details.
During data processing, this source happened to be duplicated and only one source identifier has been kept. Observations assigned to the discarded source identifier(s) were not used. This may indicate observational, crossmatching or processing problems, or stellar multiplicity, and probable astrometric or photometric problems in all cases. The duplicity criterion used for Gaia DR3 is an angular distance of
0.18 arcsec, while a limit of 0.4 arcsec was used for Gaia DR2.
Number of observations (CCD transits) that contributed to the G mean flux and mean flux error.
Mean flux in the Gband.
Standard deviation of the Gband fluxes divided by sqrt(phot_g_n_obs)
Mean flux in the Gband divided by its error.
Mean magnitude in the G band. This is computed from the Gband mean flux applying the magnitude zeropoint in the Vega scale.
No error is provided for this quantity as the error distribution is only symmetric in flux space. This converts to an asymmetric error distribution in magnitude space which cannot be represented by a single error value.
Number of observations (CCD transits) that contributed to the integrated BP mean flux and mean flux error.
Mean flux in the integrated BP band.
Error on the mean flux in the integrated BP band (errors are computed from the dispersion about the weighted mean of input calibrated photometry).
A handful of sources have error equal to zero. More details and the list of the sources affected are in Section 5.4.2.
Integrated BP mean flux divided by its error.
A handful of sources have error equal to zero, meaning that the ratio is NULL. More details and the list of the sources affected are in Section 5.4.2.
Mean magnitude in the integrated BP band. This is computed from the BPband mean flux applying the magnitude zeropoint in the Vega scale.
No error is provided for this quantity as the error distribution is only symmetric in flux space. This converts to an asymmetric error distribution in magnitude space which cannot be represented by a single error value.
Number of observations (CCD transits) that contributed to the integrated RP mean flux and mean flux error.
Mean flux in the integrated RP band.
Error on the mean flux in the integrated RP band (errors are computed from the dispersion about the weighted mean of input calibrated photometry).
A handful of sources have error equal to zero. More details and the list of the sources affected are in Section 5.4.2.
Integrated RP mean flux divided by its error. A handful of sources have error equal to zero, meaning that the ratio is NULL. More details and the list of the sources affected are in Section 5.4.2.
Mean magnitude in the integrated RP band. This is computed from the RPband mean flux applying the magnitude zeropoint in the Vega scale.
No error is provided for this quantity as the error distribution is only symmetric in flux space. This converts to an asymmetric error distribution in magnitude space which cannot be represented by a single error value.
Number of BP transits that contributed to the mean photometry and were considered to be contaminated by one or more nearby sources. The contaminating sources may come from the other field of view.
Number of BP transits that contributed to the mean photometry and were flagged to be blends of more than one source (i.e. more than one source is present in the observing window). The blended sources may come from different fields of view.
Number of RP transits that contributed to the mean photometry and were considered to be contaminated by one or more nearby sources. The contaminating sources may come from the other field of view.
Number of RP transits that contributed to the mean photometry and were flagged to be blends of more than one source (i.e. more than one source is present in the observing window). The blended sources may come from different fields of view.
This flag indicates the photometric calibration process used for the source. The process is determined by the availability of colour information derived from the internally calibrated mean BP and RP source spectra. The following values are defined for Gaia EDR3:

•
0: this corresponds to the “gold” photometric dataset. Sources in this dataset have complete colour information.

•
1: this corresponds to the “silver” photometric dataset. Sources in this dataset have incomplete colour information and therefore were calibrated using an iterative process that estimated the missing colour information from the source mean G and either BP or RP photometry (depending on which band had full colour information available) using empirical relationships derived from the gold dataset.

•
2: this corresponds to the “bronze” photometric dataset. Sources in this dataset had insufficient colour information and therefore were calibrated using default colour information derived from the gold dataset.
Because the process of generating the mean BP and RP spectra and the process of producing mean BP and RP integrated photometry are very different and have different requirements it is possible for gold sources to be missing any of the bands, i.e. “gold” does not imply anything about the availability of mean G, BP and RP photometry. Similarly for silver and bronze sources it is possible to have photometry available in any bands (and possible combinations).
More details about the different calibration procedures are available in Chapter 5 of the Gaia EDR3 online documentation and in
Riello et al. (2020)
BP/RP excess factor estimated from the comparison of the sum of integrated BP and RP fluxes with respect to the flux in the G band. This measures the excess of flux in the BP and RP integrated photometry with respect to the G band. A deviation from the norm means that there is a consistency issue between the fluxes. This could generally imply a problem with the G, BP or RP measurements. More details on how to best interpret this metric can be found in Riello et al. (2020).
BP$$RP colour (phot_bp_mean_mag – phot_rp_mean_mag).
BP$$G colour (phot_bp_mean_mag – phot_g_mean_mag).
G$$RP colour (phot_g_mean_mag – phot_rp_mean_mag).
Spectroscopic radial velocity in the solar barycentric reference frame. The radial velocity provided is the median value of the radial velocity measurements at all epochs. At Gaia EDR3 this value is simply that copied in from Gaia DR2.
The Gaia DR2 values are assigned to the Gaia EDR3 sources through an internal crossmatch operation and about 11000 sources could not be matched. In addition about 4000 identified spurious radial velocities are not copied. For further details see Section 6.1.1 in the online documentation for the release.
The dr2_radial_velocity_error is the error on the median to which a constant noise floor of 0.11 km/s has been added in quadrature to take into account the calibration contribution. At Gaia EDR3 this column is simply that copied in from Gaia DR2. The Gaia DR2 radial velocities and associated quantities are assigned to the Gaia EDR3 sources through an internal crossmatch operation and about 11000 sources could not be matched. In addition about 4000 identified spurious radial velocities, along with their associated values, are not copied. For further details see Section 6.1.1 in the online documentation for the release.
In detail, dr2_radial_velocity_error = $\sqrt{{\sigma}_{{V}_{\mathrm{rad}}}^{2}+{0.11}^{2}}$ where ${\sigma}_{{V}_{\mathrm{rad}}}$ is the error on the median:
$${\sigma}_{{V}_{\mathrm{rad}}}=\sqrt{\frac{\pi}{2}}.\frac{\sigma ({V}_{\mathrm{rad}}^{\mathrm{t}})}{\sqrt{\mathtt{\text{dr2\_rv\_nb\_transits}}}}$$ 
where $\sigma ({V}_{\mathrm{rad}}^{\mathrm{t}})$ is the standard deviation of the epoch radial velocities and dr2_rv_nb_transits the number of transits for which a ${V}_{\mathrm{rad}}^{\mathrm{t}}$ has been obtained.
The number of transits (epochs) used to compute dr2_radial_velocity. At Gaia EDR3 this value is simply that copied in from Gaia DR2.
The Gaia DR2 radial velocities and associated quantities are assigned to the Gaia EDR3 sources through an internal crossmatch operation and about 11000 sources could not be matched. In addition about 4000 identified spurious radial velocities, along with their associated values, are not copied. For further details see Section 6.1.1 in the online documentation for the release.
dr2_rv_template_teff : Teff of the template used to compute radial velocity in Gaia DR2 (float, Temperature[K])
Effective temperature of the synthetic spectrum template used to determine dr2_radial_velocity. N.B. the purpose of this parameter is to provide information on the synthetic template spectrum used to determine dr2_radial_velocity, and not to provide an estimate of the stellar effective temperature of this source.
At Gaia EDR3 this value is simply that copied in from Gaia DR2. The Gaia DR2 radial velocities and associated quantities are assigned to the Gaia EDR3 sources through an internal crossmatch operation and about 11000 sources could not be matched. In addition about 4000 identified spurious radial velocities, along with their associated values, are not copied. For further details see Section 6.1.1 in the online documentation for the release.
dr2_rv_template_logg : logg of the template used to compute radial velocity in Gaia DR2 (float, GravitySurface[log cgs])
$\mathrm{log}g$ of the synthetic spectrum template used to determine dr2_radial_velocity. N.B. the purpose of this parameter is to provide information on the synthetic template spectrum used to determine dr2_radial_velocity, and not to provide an estimate of the $\mathrm{log}g$ of this source.
At Gaia EDR3 this value is simply that copied in from Gaia DR2. The Gaia DR2 radial velocities and associated quantities are assigned to the Gaia EDR3 sources through an internal crossmatch operation and about 11000 sources could not be matched. In addition about 4000 identified spurious radial velocities, along with their associated values, are not copied. For further details see Section 6.1.1 in the online documentation for the release.
dr2_rv_template_fe_h : Fe/H of the template used to compute radial velocity in Gaia DR2 (float, Abundances[dex])
Fe/H of the synthetic spectrum template used to determine dr2_radial_velocity. N.B. the purpose of this parameter is to provide information on the synthetic template spectrum used to determine dr2_radial_velocity, and not to provide an estimate of the stellar atmospheric Fe/H of this source.
At Gaia EDR3 this value is simply that copied in from Gaia DR2. The Gaia DR2 radial velocities and associated quantities are assigned to the Gaia EDR3 sources through an internal crossmatch operation and about 11000 sources could not be matched. In addition about 4000 identified spurious radial velocities, along with their associated values, are not copied. For further details see Section 6.1.1 in the online documentation for the release.
Galactic Longitude of the object at reference epoch ref_epoch, see Section 4.1.7 of the release documentation for conversion details.
Galactic Latitude of the object at reference epoch ref_epoch, see Section 4.1.7 of the release documentation for conversion details.
Ecliptic Longitude of the object at reference epoch ref_epoch, obtained from the equatorial coordinates using the transformation defined in Volume 1, Section 1.5.3 of ESA (1997).
Note that in the transformation applied here the ICRS origin is shifted in the equatorial plane from $\mathrm{\Gamma}$ by $\varphi =0.05542$ arcsec, positive from $\mathrm{\Gamma}$ to the ICRS origin (Chapront et al. 2002). The ICRS has an unambiguous definition with an origin in the ICRF equator defined by the realisation of the ICRF. The ecliptic system is less welldefined, potentially depending on additional conventions in dynamical theories. The transformation employed here corresponds to the inertial mean ecliptic with obliquity and $\mathrm{\Gamma}$ defined by reference to the ICRS equator. Both the obliquity and the position of $\mathrm{\Gamma}$ on the ICRS equator with respect to the ICRS origin have been obtained from Lunar Laser Ranging measurements. This has no time dependence – there is no secular variation of the obliquity and no precession – and it simply defines the relative situation of the various planes at J2000.
Ecliptic Latitude of the object at reference epoch ref_epoch. For further details see the description for attribute ecl_lon.