Stars That Approach within One Parsec of the Sun: New and More Accurate Encounters Identified in Gaia Data Release 3

Close encounters of stars to the Sun could affect life on Earth through gravitational perturbations of comets in the Oort cloud or exposure to ionizing radiation. By integrating orbits through the Galactic potential, I identify which of 33 million stars in Gaia DR3 with complete phase space information come close to the Sun. 61 stars formally approach within 1 pc, although there is high confidence in only 42 (two thirds) of these, the rest being spurious measurements or (in) binary systems. Most of the stars will encounter within the past or future 6 Myr; earlier/later encounters are less common due to the magnitude limit of the Gaia radial velocities (RVs). Several close encountering stars are identified for the first time, and the encounter times, distances, and velocities of previously known close encounters are determined more precisely on account of the significantly improved precision of Gaia DR3 over earlier releases. The K7 dwarf Gl 710 remains the closest known encounter, with an estimated (median) encounter distance of 0.0636 pc (90% confidence interval 0.0595–0.0678 pc) to take place in 1.3 Myr. The new second closest encounter took place 2.8 Myr ago: this was the G3 dwarf HD 7977, now 76 pc away, which approached within less than 0.05 pc of the Sun with a probability of one third. The apparent close encounter of the white dwarf UPM J0812-3529 is probably spurious due to an incorrect RV in Gaia DR3.


INTRODUCTION
Close encounters between stars can influence the evolution of stellar systems.Early in stars' lives, interactions can disrupt circumstellar disks, potentially influencing planet formation.Later in stars' lives, loosely bound reservoirs of comets, such as the Oort cloud around the Sun, can be perturbed by a close encounter.This can fling comets inward -where they could impact planets -or eject them out of the system, possibly leading to capture by the other star.The close passage of an active star, in particular one that goes supernova, could even jeopardize -or perhaps assist -the evolution of life on a planet.
Of particular interest is the stellar encounter history of the Sun.To identify past or future encounters we need to know stars' current positions and velocities.Extensive searches for encounters were therefore only possible starting with the publication of a hundred thousand parallaxes by Hipparcos in 1997, which increased by orders of magnitude two decades later with Gaia.These data led to the identification of many close encounters (e.g.García-Sánchez et al. 2001;Bobylev & Bajkova 2017;Bailer-Jones et al. 2018b;Wysoczańska et al. 2020b) which fed into modelling how these encounters affected our Oort cloud (e.g.Feng & Bailer-Jones 2015;Torres et al. 2019;Wysoczańska et al. 2020a;Dybczyński & Królikowska 2022).
The present paper continues a study to discover and characterize close encounters, one that started with Hipparcos (Bailer-Jones 2015a; paper 1), then Gaia DR1 (Bailer-Jones 2018; paper 2) -both complemented by non-Gaia data -and most recently Gaia DR2 (Bailer-Jones et al. 2018b;paper 3).Since the first Gaia data release, astrometry has been in abundance and the comparative lack of relative radial velocities (RVs) has been the limiting factor of these studies, a complete reversal of the pre-Hipparcos situation.Gaia DR3 (Gaia Collaboration et al. 2022) now provides radial velocities for 34 million bright stars (99% with G < 15.7 mag) with median uncertainties of 3.3 km s −1 (central 90% range 0.4-7.8km s −1 ).This is nearly a five-fold increase in the number of sources with radial velocities in Gaia DR2, and constitutes the largest radial velocity survey to date.
Here I use these data to identify stars that approach within 1 pc of the Sun.Previous works by various authors have used a larger limit.But as the average spac-ing 1 between stellar systems in the solar neighbourhood (a few tens of pc) is about 2.2 pc, and the closest stellar neighbour to the Sun -the Alpha Centauri system -is currently just 1.3 pc away, 1 pc seems a more meaningful upper limit.The Oort cloud extends to perhaps 0.25-0.50pc from the Sun (e.g.Brasser et al. 2012), and even massive stars passing much further from the Sun than this are expected to have only a small effect on the Oort cloud (e.g. Brown & Rein 2022).
2. DATA Gaia DR3 contains 33 653 049 sources with complete six-dimensional phase space coordinates (three positions and three velocities).Of these, 29 947 046 have a parallax signal-to-noise ratio (SNR), parallax over error, greater than 5 (prior to adjusting the parallax zeropoint).I adopt this as the main sample to search for encounters.Less precise parallaxes correspond to less precise encounter parameters, such that even if the median perihelion distance (d med ph ) were small, its probability distribution would be broad.I nonetheless applied the procedure described in section 3 to the 3 311 812 sources with 0 < parallax over error < 5, but none reached d med ph < 1 pc.I did not consider at all the 394 191 sources with parallax over error ≤ 0. Although it is possible to infer distances for sources with negative parallaxes (e.g.Bailer-Jones 2015b), their distance posteriors are prior-dominated and generally very broad, so the encounter parameters remain very imprecise.
Once selected, I corrected the parallax zeropoint of all the sources in the main sample using the procedure described by Lindegren et al. (2021a).This corresponds to increasing the raw parallaxes by 0.006 to 0.058 mas (central 98% range), the exact value depending on magnitude (as well as colour and sky position).Other analyses have suggested that the parallaxes may have to be increased further for the brighter stars in Gaia, by up to 0.015 mas (e.g.Zinn 2021; Ren et al. 2021).This would have little impact on my encounter results, however: Although the raw parallaxes of the main sample cover a broad range, 0.12 to 7.9 mas (central 98% range), a star will generally have to be closer than about 1 kpc in order to have a high probability of approaching within 1 pc of the Sun, such that any residual error in the parallax zeropoint is small compared to the parallax.
1 By counting the number of stars within some distance we can compute the spacing they would have if they were on a regular cubic lattice.From the Gaia Catalogue of Nearby Stars (Gaia Collaboration et al. 2021) we get values of 2.23 to 2.33 pc depending on the distance used and whether correcting for incompleteness.Using all sources in Gaia DR3 we get 1.95-2.37pc for various distances out to 100 pc.
There is some evidence that some of the parallax uncertainties in Gaia DR3 are underestimated (e.g.Maíz Apellániz 2022), but I have not inflated them in this study.
An observed radial velocity is often not equal to the true radial velocity for a number of physical reasons beyond measurement or calibration errors.The gravitational redshift increases the observed radial velocity by about 0.6 km s −1 for main sequence stars (less for giants), although this is offset by convective blueshifts that reduce the radial velocity by a similar size: As discussed by Gullberg & Lindegren (2002), the net effect is that the observed minus true radial velocity ranges from about −0.4 km s −1 for F stars to +0.4 km s −1 for K dwarfs.These offsets are generally less than the Gaia DR3 radial velocity uncertainties, and as the specific correction depends on properties that are not well determined for most stars, no correction has been made.

METHOD
Once we have the phase space coordinates and specify the Galactic gravitational potential, the orbits of stars relative to the Sun can be computed by numerical integration.From this we can determine the time, distance, and velocity of closest approach.An accurate numerical integration for all 30 million stars is not only time-consuming, it is also unnecessary, because the vast majority of the stars will never come anywhere near the Sun.Moreover, the uncertainties in (and covariances between) the phase space coordinates must be taken into account.I therefore adopt the following procedure for each star, which is an improvement over the procedure used in papers 1-3 (but the method of integration is unchanged).
1. Resample the phase space coordinates 50 times assuming the mean and covariances from Gaia DR3 represent a six-dimensional Gaussian probability distribution over the coordinates, to give 50 "surrogate" stars.
2. Find the closest approach between each surrogate and the Sun under the assumption of nonaccelerated motion, i.e. neglecting gravity.This linear motion approximation (LMA) has a simple analytic solution (section 3.2 of paper 1).
3. If any of the 50 surrogates from the previous step approaches within d lim = 7.07 pc of the Sun, numerically integrate the orbits of all 50 surrogates through the Galactic potential.This uses 50 time steps distributed uniformly over the interval [0, 2t LMA ph ], where t LMA ph is the time of closest approach computed from the LMA (for that surrogate).
4. If any of the 50 surrogates from the previous step approaches within d lim = 7.07 pc of the Sun, do a higher resolution numerical integration using 1000 surrogates each integrated for 500 time steps.
5. Determine whether the median (over the 1000 surrogates; d med ph ) encounter distance is below 1 pc.
Only the stars that make it through the final step are reported here.Using the distributions over the perihelion time (t ph , signed), distance (d ph , non-negative), and velocity (v ph , non-negative), I compute their medians as well as their 5% and 95% quantiles, the latter two providing the 90% central confidence interval (CI) for each parameter.d lim was chosen to be large enough to make it likely that the selections in steps 3 and 4 include all encounters that will come within 1 pc when integrated at high resolution in step 5.If d lim is lowered to 5 pc then in fact we miss two encounters.2It is therefore possible that a few more encounters would be found if we used a larger d lim , although this gets increasingly unlikely.
The LMA in step 2 is reasonably accurate in many cases, and good enough for the first liberal selection, because many stars travel only a short distance from their current location to perihelion, so their paths are relatively unaccelerated by the Galactic potential.The gravitational potential used in steps 3 and 4 is a simple, three-component axisymmetric model described in section 3.3 of paper 1.Because the accelerations are not large, the results are not very sensitive to the exact potential or location of the Sun in the Galaxy.As discussed in section 5.2 of paper 1, close encounters of a star on its journey to perihelion with other individual stars will hardly affect its orbit in the vast majority of cases.We may, therefore, reasonably adopt a smooth potential.
Light travel times have been neglected in the computations.This makes the inferred encounter times too late (too positive) by an amount roughly equal to the light travel time, which in the worst case is about 1 kyr.The fractional error this introduces in the encounter time is approximately the ratio of the radial velocity to the speed of light, which is of order 4 × 10 −4 .As we will see, this is much smaller than the uncertainties arising from the Gaia measurements.
For a very close encountering star, or one with large measurement uncertainties, it is possible that the orbits of the surrogates -which represent these uncertaintiespass the Sun on opposite sides.If we then calculated the average of the signed Cartesian coordinates of these surrogates at perihelion, this average could be arbitrarily close to zero, on account of the cancelling of positive and negative coordinates (surrogates passing on either side).But this is clearly not a useful measure of the average perihelion distance.We must instead look at the distribution of the (non-negative) perihelia distances over the surrogates, and then average those.This is what I do in this work (and did in papers 1-3), using the quantiles of the distribution to characterize the distribution.Not all studies do this (e.g.Dybczyński et al. 2022), and the average signed coordinate will always produce a lower "distance" than the average of the non-negative distances.In principle a similar problem can also occur with the encounter time or velocity if the radial velocity is near zero.The encounter times I compute are signed, but we can see from the distribution (see paper 1)summarized by the 90% CI -whether some times have a different sign.This does not occur for any of the encounters in this study.

Overall results
Of the 29 947 046 stars subject to the LMA analysis in step 2 of the procedure in section 3, 11 199 had at least one surrogate with d ph < 7.07 pc and so were processed in step 3. Of those, 6407 had at least one surrogate with d ph < 7.07 pc in the low resolution orbit integration and so were processed in step 4. Of those, 1892 had a median encounter distance (d med ph ) below 5 pc following the high resolution orbital integration.61 have d med ph < 1 pc; this is the sample that I examine in the rest of this paper.
Figure 1 shows the encounter times, distances, and velocities of the 61 closest encounters.Most of the encounters take place within 6 Myr of the present date.Larger absolute times correspond to stars that are currently further away and/or which have low velocities relative to the Sun.This "time horizon" of the encountering stars arises from them having a characteristic maximum distance due to the magnitude limit of the main sample.This, combined with a characteristic velocity relative to the Sun, means most stars cannot travel for more than a few million years to/from their encounter and be visible now.
Table 1 lists the encounter parameters as well as the parallaxes, proper motions, and radial velocities of the 61 closest encounters.The final column of the table gives the ratio of the current absolute radial velocity (v r ) to  1.The colour of each point indicates the median encounter velocity; those faster than 100 km s −1 are white.
the current transverse velocity (v t ) of each star.The fact that these ratios are generally much larger than 1 is a selection effect: stars that approach close to the Sun are those that are currently moving almost directly toward or away from it, so have |v r | >> v t .It is therefore not surprising that many of the close encounters have large absolute radial velocities.Nonetheless, we see several encounters with radial velocities (and thus space velocities) above several hundred km s −1 .While such velocities do occur, we should be suspicious of these, as they could be a result of an incorrect template or failure of the cross-correlation in the Gaia processing reported in Katz et al. (2022).This will be discussed for some individual cases in section 4.2.
Table 2 provides additional information on the encounters.M G is the absolute magnitude of the source on the assumption of zero extinction and 1/ being a reasonable distance estimate, which it is because the parallax SNR is above 30 for all but three of the sources (and those still have SNRs above 10).ruwe is a calibrated reduced χ 2 of the astrometric solution: values much more than a few could indicate the astrometric solution is poor, although this is not a definitive metric.astrometric params solved is 31 if the astrometric solution was for the standard five parameters (2 positions, parallax, 2 proper motions), and 95 if the colour of the source was solved for in addition, for reasons discussed in Lindegren et al. (2021b).ipd frac multi peak indicates the percentage of image windows used in the astrometry which contain more than one peak.This may indicate it is a double star (physical or otherwise), although only two of the stars in the table have a definite indication of physical close binarity in Gaia DR3 (rows 13 and 58).rv expected sig to noise is the estimated SNR of the spectrum used to determine the RV, and rv nb transits is the number of measurement epochs used in that determination.l and b are the Galactic longitude and latitude of the star.
Figure 2 shows the location of encounters in the colour-absolute magnitude diagram.The median distance to these stars is 100 pc (maximum 380 pc), so in most cases the assumption of negligible extinction is valid.The location of the stars in this diagram, as well as the stellar parameters provided in Gaia DR3 for many of them by Andrae et al. (2022), show that most of the stars are main sequence FGKM stars.In addition there are two white dwarfs, two possible very young stars lying above the main sequence, and two stars lying unexpectedly between the white dwarf and main sequences.These will be discussed in the next section.
Encounters that are dubious are prefixed with a † in Table 1.This assignment occurs because the RV is sus-  1.The 13 encounters with d med ph < 0.5 pc plus several others of possible interest are labelled by their order of increasing distance (row number in Table 1).For orientation, the black lines are unreddened solar metallicity PARSEC isochrones for 1 Gyr (solid) and 10 Gyr (dashed) from Marigo et al. (2017), and the small blue dots are a random subset of all sources in Gaia DR3 with > 50 mas and ruwe < 1.2.
piciously large, the ruwe is large, or for another reason given in the next section.Eight stars that are probably in wide binaries according to the criteria of El-Badry et al. ( 2021) are prefixed with a "b".Unless the orbital period is very long compared to the time to encounter (t med ph ), the orbital motion will be significant, and hence the Galactic trajectory computed in the present work will be incorrect.El-Badry et al. (2021) does not list periods, and only projected separations (which range from 10 2 to 10 4 au for these eight stars), so until a long period can be confirmed for any of these stars, their solar encounters should be regarded as unreliable.Two other stars in Table 1 (6608946489396474752 and 3001468183198140800) are also listed in this wide binary catalogue, but they are probably chance alignments (values of R chance align of 0.44 and 2.3 respectively, whereas the ones denoted "b" have 0.07 or -generally -much less).

Specific close encounters
Here I discuss selected sources from Table 1 giving the row number, Gaia source id, and the primary Simbad name for the source, if available.The source id has not changed from EDR3 to DR3 (other than this prefix).

1: Gaia DR3 4270814637616488064 = Gl 710
Since Hipparcos, this K7 dwarf has often held the title of the closest known bona-fide encounter.In paper 1 (Hipparcos) it had a median encounter distance of 0.267 pc with a 90% CI of 0.101-0.444pc.
In paper 3 (Gaia DR2) this was significantly reduced to 0.0676 pc (90% CI 0.0519-0.0842pc) where it was the closest encounter found.This does not change much using Gaia DR3 -the median is 0.0636 pc -but it is more precise with a 90% CI of 0.0595-0.0678pc.This is mostly due to a factor of three improvement in the precision of both its parallax and proper motions.Assuming Gl 710 has a mass of 0.7 M (paper 3) the gravitational attraction by the Sun only lowers the median perihelion by 7 au (35 µpc, 0.06%) demonstrating that gravitational focusing can be neglected in all these encounters.
2: Gaia DR3 510911618569239040 = HD 7977 This G3 dwarf was found in Gaia DR2 (paper 3) as a close encounter with a median distance of 0.429 pc (90% CI 0.368-0.494pc), but now comes much closer with a median of 0.0641 pc (90% CI 0.0191-0.1171pc).This is mostly due to a sig- respectively.The columns labelled 5% and 95% are the bounds of the corresponding confidence intervals.Columns 11-16 list the current parallax ( , including the zeropoint offset), total proper motion (µ), and radial velocity (vr), along with their 1-sigma uncertainties.Column 17 is the ratio of the current absolute radial velocity to the current transverse velocity, the latter computed as 4.74047 µ/ .On the far left, † indicates that the encounter is dubious for a variety of possible reasons, and b indicates it is probably in a wide binary system according to El-Badry et al. ( 2021), so may also not be reliable.A machine readable version of this table reporting values to more significant figures is available from this URL.nificant reduction in its proper motion.There is a large relative uncertainty in the encounter distance because the SNR of the proper motion is low, just 4.2.Yet this is the only known star with a significant probability -32% -of approaching within 0.05 pc of the Sun.The ruwe is slightly inflated (2.01), which could suggest a problem with the astrometric solution, although there is no indication of close binarity.Dybczyński et al. (2022) found a much closer minimum perihelion distance for this star using Gaia EDR3 (which has the same astrometry as DR3, but different RVs), but this is partly because they averaged the signed coordinates of the surrogates (section 3), which underestimates the distance in this case (see their Figure 5).
3: Gaia DR3 5544743925212648320 = UPM J0812-3529 This was discovered to be a nearby white dwarf by Finch et al. (2018), but Gaia DR3 is the first publication of a radial velocity.If a real encounter, then it will pass just 0.11 pc from the Sun in 29 kyr.However, the radial velocity is suspiciously large (−374 km s −1 ), even though the ratio to the transverse velocity is modest (it has a large proper motion and is currently just 11.2 pc away).It is hard to measure radial velocities for white dwarfs accurately on account of their relatively featureless spectra, especially in this case due to its apparently strong magnetic fields (Bagnulo & Landstreet 2020).The Gaia DR3 RV pipeline does not include any white dwarf templates, and although the main sequence dwarf template T eff of 6000 K is close to the value of 6090 K estimated by Bagnulo & Landstreet (2020), the Gaia DR3 RV is likely wrong (Gaia CU6 (RV) team, private communication).If it is correct, the strong gravitational redshift in white dwarfs means that the true RV of the star would be tens of km s −1 more negative.
As the RV of this star is negative, the true absolute value of its RV would be even larger and the encounter even closer.If the gravitational redshift were 30 km s −1 , for example, the star would come about 0.008 pc (7%) closer.
4: Gaia DR3 213090546082530816=UCAC4 689-035468 This was found in paper 3 but with a much larger median encounter distance of 1.44 pc and large uncertainty (90% CI) of 0.703-2.175pc.These have now decreased to 0.248 pc and 0.072-0.489pc respectively thanks to more precise astrometry.
6: Gaia DR3 5469802896279029504 = CD-25 8217 At 11±1 Myr, this is the latest encounter.It is also the slowest encounter, moving at just 3.2 km s −1 relative to the Sun.Both of these are a consequence of its small radial velocity of −2.66 ± 0.17 km s −1 .RAVE DR6 (Steinmetz et al. 2020) reports an RV of −4.02±1.06km s −1 .If we use this in the orbital integration we get a larger median perihelion distance of 0.92 pc with a wider 90% CI of 0.32-1.10pc, and an encounter that probably takes places sooner, at 7.7 Myr, but with a larger spread due to the larger RV uncertainty (90% CI 5.6-12.9Myr).
There is a second source in Gaia DR3 1.0" away that is 1.1 mag fainter; this could have interfered with the radial velocity determination.The RV spectrum also has a very slow SNR.The star also lies well above the main sequence (Figure 2).This can indicate binarity, although 3 magnitudes offset is far too large.It could instead be a pre-main sequence star (see object 39 below), in which case the RV is likely wrong due to a lack of appropriate templates in the data processing.
13: Gaia DR3 3118526069444386944 This was identified as a young stellar object candidate by Zari et al. (2018).It was found in paper 3 to encounter at 1.03 pc; now apparently reduced to 0.49 pc.However, it is just one of two stars in Table 1 that is identified as a close physical binary in Gaia DR3 via the non single star flag (the other is object 58).Specifically, it is an astrometric binary with a significant proper motion acceleration (a so-called 7-parameter solution).The table nss acceleration astro lists improved parallaxes and proper motions that take into account this acceleration, but as they still may not be representative of the centre-of-mass of the binary system, I have not recomputed the encounter.This encounter should be disregarded until the system motion has been adequately characterized.

20: Gaia DR3 4116451378388951424
This star is strange because it lies in the otherwise relatively empty region between the white dwarf and main sequences.After Gaia DR2 it was found that many stars in this region were very faint (G 19.5 mag) and had spuriously large parallaxes.This situation has been greatly improved in Gaia DR3, but as this star has a large ruwe, its astrometric solution may well be inappropriate.It lies very close to the Galactic centre in a crowded field, so it is possible that a nearby star

CONCLUSIONS
This study has revealed 61 stars that, based on Gaia DR3 data, have passed -or will pass -within 1 pc of the Sun, most within ±6 Myr.Closer inspection reveals that 12 of these are probably not real encounters due to spurious measurements or being in close binaries.A further seven are in wide binaries, so their encounters are probably spurious unless their orbital periods are very long.As Gaia DR3 has much better precision and accuracy than previous Gaia releases, encounters within 1 pc identified using earlier Gaia releases that do not appear in Table 1 should generally be considered as obsolete.Potential exceptions are multiple systems in which the orbit of the centre-of-mass has been traced, and stars with reliable non-Gaia astrometry.
In principle one could use these results to infer an overall stellar encounter rate, as was done in Bailer-Jones (2018) and Bailer-Jones et al. (2018b).Yet correcting for the significant and poorly characterized incompleteness of the sample makes this difficult, and the result uncertain.This will be tackled using the larger sample expected in Gaia DR4.As close encounters generally have large ratios of radial velocities to transverse motions, contamination from spuriously large RVs is a problem.Resolved binaries not identified as such are also a source of contamination.But the individual encounters identified in this paper can be used to reassess Oort cloud comet perturbation or effects on specific comets (e.g.Dybczyński & Królikowska 2022).
To extend the identification of close encounters significantly beyond ±6 Myr we need to obtain RVs for fainter stars.There will be perhaps 100 million RVs in Gaia DR4 (Katz et al. 2022), and many will be near the faint limit of the RV survey, which is expected to extend two magnitudes deeper from the current limit of G RVS = 14.0 mag to G RVS = 16.0 mag (the bright limit in Gaia DR3 is G RVS = 2.8 mag).Although one can estimate an RV (or rather a probability distribution over the RV) for stars using their distances, transverse velocities, and a Galaxy model, and this will help to infer the statistics of encounters for fainter stars, this cannot be used to identify individual encounters reliably.
Whilst we need more RVs, more precise proper motions will help to reduce the uncertainties on the encounter distances.This is because the closest encountering stars tend to have very low proper motions, and the lower the proper motion, the lower its uncertainty must be to retain a given SNR.Yet as proper motion precision improves as the 3/2 power of the observation baseline duration, and Gaia DR3 is based on just under 3 years of observations, we can expect significant improvements in Gaia DR4 (5 years) and Gaia DR5 (expected to be 10.5-11 years).The detection of astrometric orbits with Gaia will also help include or remove multiple stellar systems from the encounter lists.

Figure 1 .
Figure 1.Perihelion (encounter) times and distances for the 61 stars that have a median perihelion distance (d med ph ) below 1 pc.Negative times indicate past encounters, positive times future ones.The circles/squares show the median of the perihelion time and distance distributions computed from the 1000 data resamples (surrogates); the error bars show the 5th and 95th percentiles.Circles denote good encounters; squares denote the questionable encounters, which are marked with a † or b in Table1.The colour of each point indicates the median encounter velocity; those faster than 100 km s −1 are white.

Figure 2 .
Figure 2. Colour-absolute magnitude diagram, with MG computed assuming zero extinction.The red/yellow coloured points are the 61 encounters with d med ph < 1 pc identified in this work, colour-coded according to d med ph .Circles denote good encounters; squares denote the questionable encounters, which are marked with a † or b in Table1.The 13 encounters with d med ph < 0.5 pc plus several others of possible interest are labelled by their order of increasing distance (row number in Table1).For orientation, the black lines are unreddened solar metallicity PARSEC isochrones for 1 Gyr (solid) and 10 Gyr (dashed) fromMarigo et al. (2017), and the small blue dots are a random subset of all sources in Gaia DR3 with > 50 mas and ruwe < 1.2.

Table 1 .
Perihelion (encounter) parameters for all stars with a median perihelion distance (d med ph ) below 1 pc, sorted by increasing d med ph .Columns 2, 5, and 8 are t med ph , d med ph , and v med ph

Table 2 .
Additional data on the close encounters listed in Table 1.All fields are taken directly from the Gaia DR3 gaia source table, except for MG which is G + 5 log 10 ( /100), where is the zeropoint-corrected parallax in mas.Descriptions of the other fields can be found in the Gaia DR3 online documentation.Gaia DR3 1926461164913660160 = Ross 248 This is currently one of the closest stars to the Sun at 3.2 pc.It will come closer in 36 kyr, to 0.934 pc.7 kyr before this it will pass within 0.46 pc of the interstellar object 'Oumuamua (Bailer-Jones et al. 2018a).56: Gaia DR3 5853498713190525696 = Proxima Cen.It appears that our closest neighbour will get slightly closer in 27 kyr, although its orbit around Alpha Centauri AB needs to be taken into account.This was done by Wysoczańska et al. (2020b).They infer similar encounter parameters to the medians reported in Table 1, which is not surprising because Proxima Centauri's orbital period of 547 +66 −40 kyr (Kervella et al. 2017) is much longer than the time to perihelion of 26.622 kyr (90% CI 26.580-26.653kyr).