Best Practices for Data Publication in the Astronomical Literature

The most common and basic type of data in any observational article is the naming of astronomical sources. Unambiguous names that follow recommended nomenclature standards are essential prerequisites for clear communication of observational results and scientific conclusions, and for ensuring follow-up observations target the proper source.

2.1.1. IAU Conventions

When assigning object names in an article, we recommend the authors follow the established conventions from the IAU such as "Naming of Astronomical Objects" (International Astronomical Union 2022), "How to refer to a source or designate a new one" (IAU Working Group on Designations 2018a), and "Specifications concerning designations for astronomical radiation sources outside the solar system" (IAU Working Group on Designations 2018b). The IAU Dictionary of Nomenclature of Celestial Objects (Lortet et al. 1994; SIMBAD Team 2022b) provides a list of acronyms that are currently in use, and should be consulted to confirm the correct acronyms and formats for known objects, and to avoid reusing the same ones for newly discovered objects. When publishing simultaneous independent discoveries of the same astrophysical objects, reasonable efforts should be made to coordinate and avoid publishing conflicting designations.

Following the IAU guidelines, we recommend:

• 1.  
  Provide the complete object name. A name that has the coordinate part truncated will likely be ambiguous and can be confused with a nearby object (see, e.g., first example in Table 1).
• 2.  
  Explicitly include the "J" in names based on J2000 coordinates. Without the "J", this could be misinterpreted as B1950, resulting in an incorrect object position and telescope pointing. For example, use "BR J0529-3526" instead of "BR 0529-3526".
• 3.  
  Insert spacers between a catalog name and the identifiers within the catalog. For example, use B3 2327+391, not B32327+391. This will prevent mixing and misinterpreting the catalog name and the identifiers, and allow both the readers and name resolver services at the archives (e.g., SIMBAD, NED, ²¹ NASA Exoplanet Archive ²² ) to recognize object names more efficiently.
• 4.  
  Distinguish between part of an object and the object itself. For example, use "3C 295 cluster" instead of "3C 295" when referring to the cluster. Similarly, the hosts for transients, like supernovae and gamma-ray bursts, should be referred to with the proper names when host properties are discussed, instead of just using the names of the transients.
• 5.  
  Do not use the same name for different objects. Once a name has been assigned to an object in a published catalog or journal article, it should not be reused for a different source in the future, even if an object's existence is refuted. For example, the tau Ceti system now has four planets: e, f, g, and h. Since tau Ceti b, c, and d were refuted, the letter designations b, c, and d were not reused for the newer planets to avoid confusion.

Table 1. Examples of Improper Astronomical Designations in Literature

As Published	Why It Is Improper	Recommended Usage
		(Notes if Available)
SDSS J1441+0948	Insufficient precision in R.A. and	SDSS J144157.24+094859.1, or
	decl. can cause confusion with	SDSS J144156.97+094856.5, or
	nearby sources.	SDSS J144157.26+094853.7
SN 05J	Incomplete name can be interpreted	SN 1905J, or
	into different objects.	SN 2005J
HESS J232+202	Leading zero in R.A. is missing and	HESS J0232+202
	can cause misinterpretation of the
	R.A. at 23 hr instead of 2 hr.
BR 0529-3526	Missing letter J to specify J2000	BR J0529-3526
	equatorial coordinates.
0008+006	Name prefix is needed to distinguish	ZC 0008+006 (Redshift z = 2.3), or
	between different objects.	IVS B0008+006 (Redshift z = 1.5)
DEM45	H ii regions in LMC or SMC should	DEM L 045, or
	be indicated with "L" or "S" to	DEM S 045
	avoid ambiguity.
SDSS 587729386611212320	Database objectID numbers are used	SDSS DR6 587729386611212320
	without specifying release number.
	The same running number may refer to
	a different source in a different release.
Gaia DR 2 2.7904e18	ID is written in scientific notation, making	Gaia DR2 2790494815860044544
	it impossible to retrieve the actual object.
mu cep	Ambiguous name can be interpreted	μ Cep
	into different objects.	(21h43m3046, +58d46m482, ICRS J2000), or
		MU Cep
		(22h23m3863, +57d40m508, ICRS J2000)

Download table as:  ASCIITypeset image

Table 1 illustrates some ambiguous/improper astronomical designations that have appeared in the literature, along with the recommended proper usage.

2.1.2. New Objects

2.1.3. Known Objects

2.1.4. Cross Identifications

One of the primary attributes for an astronomical source is its location in the sky. Accurate celestial coordinates for the objects are especially important for follow-up observations and study. When presenting coordinates in a publication, we recommend to:

• 1.  
  Provide the best available coordinates. Precise positions of the sources are indispensable for the usability of observational data and for planning follow-up observations. All objects studied in an article, especially those from private catalogs, need to be presented with coordinates. Complete celestial coordinates are preferred, e.g., 12^h34^m5678, +12^d34^m567 (equatorial J2000). When positional offsets are published instead, it is essential that authors include the coordinates of the reference position, as well as the angle of rotation (if north is not up or east is not to the left), and the sense of the offset (i.e., reference point minus source, or source minus reference point).
• 2.  
  Specify the celestial reference system and/or frame. Indicate the reference system and/or frame for the coordinates (Urban & Seidelmann 2013, Chapters 4 and 7). The current IAU celestial reference system is the International Celestial Reference System (ICRS; Arias et al. 1995; IERS 2013a), and it is realized through the International Celestial Reference Frame (ICRF; Ma et al. 1998; IERS 2013b).
• 3.  
  Indicate the equinox and epoch of observations when necessary. This is particularly important for the positions of stars and exoplanet systems. Nearby stars can have a quite significant proper motion (>1'' yr⁻¹), which means that the position changes with time and therefore requires the equinox and epoch of observation in order to compute its position at another epoch. The standard equinox and epoch currently in use are J2000.0 (Aoki et al. 1983).
• 4.  
  State the wavelength range from which astrometry is obtained, where appropriate. Astronomical sources can be detected at slightly different positions at various wavelengths due to different emission mechanisms of the components. Therefore, providing the wavelength range of the detection is important for understanding the various components of the object and for cross identifications between sources detected at various wavelengths.

We caution that although the standard equinox and epoch currently in use are J2000.0, when citing coordinates from a catalog, a web page or other sources, one cannot assume the equinox/epoch is always J2000.0. For example, the reference epoch for the Gaia Early Data Release 3 is J2016.0, while it is J2015.5 for Gaia Data Release 2 and J2015.0 for Gaia Data Release 1. ²⁷

The flux or intensity of light radiated by astronomical objects is another key observable in astronomy. To properly represent the photometry data and to enable easy comparisons of the results, we strongly recommend:

• 1.  
  State the facility, telescope, and instrument used. Specify whether the facility is ground based or space based. Authors also need to provide any other relevant instrument configuration information, the specific camera on the instrument, and/or the specific CCD chips of the camera that were used at the time of the observation. This information is crucial to the proper interpretation of the data, e.g., hardware changes such as the replacement of a filter at a particular time will lead to slightly different instrument responses. Good documentation of the metadata will also facilitate the correct and fast ingestion of photometry data in services such as the NED spectral energy distribution (SED) plots and the VizieR Photometry viewer, ²⁸ which in return will better serve the community.
• 2.  
  Describe the method used to estimate photometry. Indicate if an estimate is from point-spread-function fitting, aperture photometry, isophotal measurements, etc. If it is aperture photometry, report the size of the aperture and background annulus, and any corrections made in the calculation.
• 3.  
  Use standard passband/filter identifiers. Do not modify or abbreviate the identifier as it may conflict with a different standard identifier. For example, indicate "Johnson B" or "Cousins B" instead of just "B"; use "2MASS K_s " instead of just "K". A listing of commonly used identifiers is available at the Spanish Virtual Observatory Filter Profile Service (Rodrigo et al. 2012; Rodrigo & Solano 2020; Spanish Virtual Observatory 2022). ²⁹ When possible, authors should provide a link to the instrument documentation with the actual response curve for the filter in question.
• 4.  
  Clarify the magnitude system. Explicitly indicate whether a magnitude is on the AB, Vega, ST, or some other magnitude system. For example, the absolute magnitudes of the Sun in the 2MASS J band, for the AB and ST systems, are 4.54 mag and 6.31 mag, respectively (Willmer 2018).
• 5.  
  Specify spectral transitions completely. Describe the molecular species, transitions, and frequencies/wavelengths. For example, carbon monoxide (CO) has several detectable transitions as do ¹³CO and C¹⁷O. The most commonly observed transition is (J = 1–0) and each is between 110 and 115 GHz. To clearly define a spectral transition, one should use, e.g., "CO (J = 1–0) ν = 115 GHz".

Knowing the time of observation is essential to understand the observed properties of many astronomical objects, and often for calculating positions, especially for transient or variable sources and moving objects. Calibrations sometimes may also vary depending on when the instrument was installed on a telescope. Authors are advised to:

• 1.  
  Provide the time of observation and exposure time. Any times should be explicitly described in terms of both the frame of reference (e.g., JD, BJD, HJD), and the time system used (e.g., UTC, TDB, TAI). For example, use "BJD-TDB" to indicate Barycentric Julian Date in the Barycentric Dynamical Time standard (preferred). This is particularly important when precise timing is needed, such as the measurement of exoplanet transit timing variations. See Eastman et al. (2010) and Urban & Seidelmann (2013, Chapter 3) for a helpful discussion of precise time standards. It is also important to specify the duration (exposure time) and whether the presented observation time is the beginning, midpoint, or end of the exposure time. When a precise time is not meaningful (such as detections from stacked images), a time range where these observations occurred should be provided.
• 2.  
  Favor full Julian Dates over abbreviated or offset Julian Dates. When reporting Julian Dates, the full unmodified date (e.g., 2,456,789.123) is preferred over any offset variation (e.g., 6789.123), to avoid confusion. This also helps archives avoid having to track down and add often arbitrary offsets to put observations on a uniform timescale, which can add an opportunity for errors to be introduced. Where an offset variation must be used, be sure to clearly indicate the value of the offset (e.g., JD-2,454,833.0) and to refer to the abbreviated date as the Modified Julian Date (MJD) only when the offset meets the IAU-defined formal definition of Julian Date minus 2,400,000.5; all other abbreviated Julian Dates should be referred to as the Reduced Julian Date. The IAU has recommended, "that where there is any possibility of doubt regarding the usage of Modified Julian Date, care be exercised to state its definition specifically" (The 23rd International Astronomical Union General Assembly 1997).
• 3.  
  Include phase-timing measures along with reported periods, where relevant and practical. This allows for future observations to be phased against your data, and combined. For example, for a transiting exoplanet orbit where the period is known, include a time of transit.
• 4.  
  State when observations from multiple missions are executed simultaneously. Coordinated observations in high-energy astronomy are critical for making a simultaneous spectral fit, which can help determine the physical processes responsible for emission in a particular energy band. It is also important to specify which instruments/cameras/chips were taking data and analyzed during these coordinated observations. If possible, include a graphical representation of the times that the missions obtained the data to help visualize where the simultaneity occurs (e.g., Figure 2 of Abbott et al. 2017).

The redshift or the recessional radial velocity of an object is of major importance to many astrophysical fields. The recommendations on publishing redshift/velocity data include:

• 1.  
  Describe the method used to obtain redshift. This includes the particular method (spectroscopic, photometric, friends-of-friends, etc.) and base assumptions used in the models (template fitting, machine learning, etc). When available, include a reference to the model/method used to determine the redshift.
• 2.  
  Specify the reference frame of the redshift measurements. Be sure to include a clear indication of the reference frame, e.g., heliocentric, barycentric, Galactocentric, or local standard of rest. Additional definition on the solar velocities adopted and likely also the knowledge of the Sun's distance to the Galactic center is needed for Galactocentric velocities.
• 3.  
  Provide the frequency/wavelength of the measurement. This is important due to the fact that the measured redshift (thus the derived radial velocity) measured from different spectral features may represent very different physical phenomena. For example, a redshift measured from the H i 21 cm emission line may have a significantly different systematic velocity than a redshift measured from stellar absorption lines in the same galaxy.
• 4.  
  State the velocity definition (radio or optical) when reporting radial velocity. Velocity (V) is conventionally defined from the resultant frequency (ν) shifts in radio astronomy, i.e., V = c(ν_rest − ν_obs)/ν_rest, and defined from the resultant wavelength (λ) shifts in optical astronomy, i.e., V = c(λ_obs − λ_rest)/λ_rest (= c(ν_rest − ν_obs)/ν_obs). Here the subscripts "obs" and "rest" refer to the observed and the rest frame (i.e., emitted), and c is the speed of light. The radio velocity increment depends upon the rest frequency, whereas the optical velocity increment depends on the observing frequency. Therefore, the velocities defined from these two conventions can differ significantly, especially for large values of V. ³⁰
• 5.  
  Indicate the quality of the measurement when possible. A qualitative assessment of the quality of a measurement may add useful information for readers and for databases in guiding researchers to use the measurement appropriately in their analyses. Examples include indications of poor seeing or blended objects, uncertain deblending of spectral lines, redshifts based on a single spectral line assuming identification of the proper feature, etc.

There are many spectral and morphological/phenomenological classifications for astronomical objects, and these classifications are constantly evolving. We suggest:

• 1.  
  Utilize established classifications as available. For basic morphological types, use the well-established schemes (e.g., Sandage 2005, and references therein). Authors are encouraged to refer to NED's extensive suite of searchable galaxy classifications and attributes ³¹ or SIMBAD's Object Classification ³² , which have been standardized to enable unified queries across journal articles and catalogs.
• 2.  
  Define new classifications clearly. If new classifications must be created, define them clearly so they can be easily adopted by other researchers, and quickly integrated into databases.

Reports of measured orbital parameters often suffer from ambiguity in terminology used because the conventions used are not clear. It is advised to avoid ambiguous terms, and to explicitly define all terms and symbols wherever there is any possibility of confusion. For instance:

• 1.  
  Avoid using the term "longitude of periapsis (or periastron)" when "argument of periapsis (or periastron)" is really the term intended. Only use "longitude of periapsis" when referring to the sum of the argument of periapsis and the longitude of the ascending node.
• 2.  
  Be explicit about which body's orbit a measured argument (or longitude) of periapsis refers to. The argument of periapsis for a planet or a secondary star's orbit differs from that of the host or primary star's reflex motion by 180°. Note also that the quantity is affected by the sign convention adopted in radial velocity analyses: as a general rule, positive radial velocity should indicate redshift, and negative should indicate blueshift.
• 3.  
  Include time of periapsis as appropriate when reporting orbital elements. For example, when reporting timing for a nontransiting eccentric orbit for which argument of periapsis is measured, report time of periapsis in preference to (or in addition to) time of inferior conjunction. Both are preferred if possible.

Observational data are often published in the form of tables in an article. For tabular data presentation:

• 1.  
  Provide a clear title and unambiguous labels of columns. Indicate the units for each column when applicable. This is especially important when data are to be compared, or used outside a specialized field.
• 2.  
  Explain the content of each column. Clarify all special symbols or flags in a column (e.g., make a clear distinction between z the redshift and z the filter), and give references to cited values.
• 3.  
  Keep each column homogeneous. A single column should not present measurements with different units, mix errors with limits or comments, or append flags to values (see, e.g., the left-hand side of Table 2).
• 4.  
  Use clearly defined, nonnumeric representations for missing (null) values. It is recommended to use null values that are supported and documented by widely used tool kits, e.g., "NaN" (Not a Number) for floating-point data in Astropy. ³³ Avoid using numerical values, as they could be misinterpreted as actual measurements by people or software. For example, a computer will not hesitate to calculate the average of a list of redshift values that include "−999" or "0.0," deriving a misleading result, but it will properly ignore or issue a warning when "NaN" is encountered. ³⁴ Use the same representation for missing data and have a separate field that explains the reasons for a missing value. Do not use different representations to indicate the different reasons, e.g., blank for "not observed", and "NaN" for "no detection".
• 5.  
  Accompany a machine-readable table (MRT) with a ReadMe file. Authors should include a human-readable description of the data, with at least the column descriptions, units, and references (on the origin of the measurements or instruments for observations when relevant) in a ReadMe file. A comprehensive example of the contents and structure of a ReadMe file can be found in the CDS document "Astronomical Catalogues and Tables Adopted Standards" (Ochsenbein 2000). ³⁵ Authors are encouraged to utilize existing Python packages (e.g., cdspyreadme ³⁶ ; Landais 2016) to generate ReadMes or MRTs from a wide list of standard table formats.

For tables containing astronomical objects, it is preferred that authors give the complete names of the objects (Section 2.1) in each table, and keep the same names in all the tables and text throughout the article when possible. This will not only enable the readers to quickly access related data about the objects, but also greatly expedite the linking and ingestion of data into data archives. The coordinates of the objects (Section 2.2) should be given in the table where the objects are first presented.

We recommend not to use LaTeX for large tables as the markups degrade the reusability of that data. Authors can improve the transfer of results by avoiding elaborate LaTeX tricks and treating inline tables as regular data arrays. For instance, method or other qualifying note marks could be given numerical values and displayed in an additional column instead of using table note marks which are not easily parsed by a machine (see Table 2). Additional details of the table should be included in an accompanying metadata file (see Section 4). Authors are also advised to report important numerical results for more than a couple of data points in tables, not just in text. When a number is reported in both text and a table, make sure it is consistent in both places.

Figures are commonly used for data visualization in an astronomical article, especially for images, spectra, SED plots, etc. For figures in a publication, we recommend:

• 1.  
  Provide clear caption, legend, and axis labels for each figure. Describe in detail what is presented in the figure, what different colors, symbols, and lines represent. Units of the axis labels should be included when applicable. In practice, figures should be able to stand alone without requiring much reading of the main text.
• 2.  
  Create the graphics with accessibility in mind. Accessibility should be taken into consideration when designing the graphics as it will help a broader audience to properly understand the information conveyed in these plots. For example, color-blind users would benefit from symbols that vary in shape in addition to colors. See the AAS ³⁶ journals' graphics guide ³⁷ for more advice on this.
• 3.  
  Include information for "data behind the plots". Make the original data files used to generate the figures publicly available, as this will greatly enhance the ability to reproduce, validate, or build upon published results.

We include an example of a clearly labeled, accessible plot in Figure 1. This was revised from a previously published article (Cook et al. 2019) with the first author's permission. The published version of this figure, although very accessible with information conveyed using variations in both colors and symbols/line styles, is lacking the units for the axis labels, and missing the explanation for some of the lines in the plot. The revision has addressed these issues.

Zoom In Zoom Out Reset image size

The AAS president in 2009 May/June, Dr. John Huchra, wrote: "Authors have a strong tendency to underattribute, that is not to properly cite both previous ideas and basic data. The phrases "data are available" or "it is known that" should never be used; someone had to work fairly hard to get that data or develop that idea and they deserve credit, even if they are your competitors. Also complicated is the use of master compilations, easy, but again not giving credit where credit is really due." ⁴¹

To properly cite data in an article, one should always:

• 1.  
  Cite the original references. For example, if the redshift presented is obtained from NED, use phrases such as "We adopted a heliocentric redshift of 1.234 (Smith et al. 2012) via NED", where "Smith et al. 2012" is listed correctly in your bibliography.
• 2.  
  Use preferred citations by the authors. Follow the instructions provided by the authors of the original papers, software etc. and use what the authors preferred for citations. For example, if 2MASS data are used in your analysis, the 2MASS web page ⁴² requests that you cite the canonical paper by Skrutskie et al. (2006), instead of the Explanatory Supplement. ⁴³ A separate standard acknowledgment is also listed there to be included in the acknowledgments.
• 3.  
  Provide full provenance of the data. Ensure full reproducibility of the analysis performed in the paper by citing all data/sources/software used: explicitly list the data identifiers provided by the data center in a table or in the text. For example, if 2MASS data is used via TOPCAT accessing the VizieR table, then in addition to any preferred citations of the data itself, cite the software and compilation used, e.g.,"2MASS (Skrutskie et al. 2006) as downloaded with TOPCAT (version 4.8–3, Taylor 2005) via VizieR (II/246, Cutri et al. 2003)". It is also recommended to include the names of principal investigators who acquired the original data sets.
• 4.  
  Include all references in the bibliography. Make sure all appropriate references to papers, software, and data products are included in a paper's bibliography section, not just in footnotes. This will ensure proper attribution of citations to them. More details and examples on this can be found in the AAS Journal Reference Instructions. ⁴⁴
• 5.  
  Distinguish original data in your article and data taken from other work. Use phrases such as "This work" to clearly identify original data in your article. It is important to archives, data services, and other authors that papers distinguish new information and measurements from those made by previous work.

Proper credit needs to be given to the facility or service used to obtain the data in an article. This may be hardware (telescopes and instruments) or online services (databases). Not only are the specifics essential to properly understand the data, the acknowledgment metric is also used as a basis for productivity by organizations maintaining telescopes, archives serving the data, and funding agencies for those facilities. We therefore recommend:

• 1.  
  Explicitly indicate the facility involved. Always describe the facilities or services used, and make sure the name is unique. Examples of some of the ambiguous names seen in the literature are given in Table 3.
• 2.  
  Use standardized keywords when possible. The AAS has created keyword tags ⁴⁵ to be used with AASTeX \facility and \facilities. Many major astronomical archives, databases, and computational centers have been recently added to this list (e.g., CDS, Exoplanet Archive, IRSA ⁴⁶ , NED).
• 3.  
  Include the facility's own statement if available. Refer to the phrase that was in place at the time the facility was used. For example, the NASA Exoplanet Archive asks authors to include the following standard acknowledgment in any published material that makes use of its services: "This research has made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program." ⁴⁷

The increasing volume and complexity of the astronomical data require a lot of software processing in obtaining, calibrating, and analyzing the data and building the final product. Similar to acknowledging the facilities and services, we recommend:

• 1.  
  List the software and version used in the production of the article. Use the preferred citation if available, e.g., the paper describing the software. If not, include the name of the author(s), title of the program/source code, the code version and a URL link to the code publisher. For example, "Stanford Classifier v3.9.2, The Stanford Natural Language Processing Group, https://nlp.stanford.edu/software/classifier.shtml" is in the Acknowledgment section in Chen et al. (2022). This will ensure proper credit is given, and at the same time help with reproducibility.

The Astrophysics Source Code Library (ASCL) ⁴⁸ is a free registry for authors to submit the software used in their research that has appeared in or been submitted to peer-reviewed journals. A unique ASCL ID is assigned to each code which enables accurate citation.

A digital object identifier (DOI) ⁴⁹ provides a persistent and unique identification of online resources, and is widely used to identify content related to published articles. Authors are encouraged to:

• 1.  
  Use DOIs to cite related content if available. This includes specific data sets, software, and services used to produce results in the published articles. Archive these in persistent repositories and link them to the article through DOIs minted by the repositories. The DOI links should be included in the bibliography to ensure proper citation (see the bibliography section of this article for examples), and also be put where the data are discussed to make it easier for readers to locate and access the data.

Some resources for obtaining DOIs for specific data sets at the archives, as well as guidelines for using them in the journal articles, are provided in Appendix C. When DOIs are not available, authors can use the exact data set identifiers from the archives where the data were obtained. For example, the Spitzer Survey of Stellar Structure in Galaxies (S4G) data set in IRSA was cited (by footnote) as http://irsa.ipac.caltech.edu/data/SPITZER/S4G/ in Ciambur (2015) before the DOI for this data set was generated. Better reference methods currently include citing this data set URL in the bibliography, e.g., S4G Team (2022), or citing the specific IRSA S4G DOI, S4G Team (2020).