The FIRST Catalogs: Version History

Since the FIRST survey is a continuing project, the current FIRST catalog is not a final data product. We will periodically release new versions of the catalogs as our sky coverage increases and when improvements are made in the catalog construction process. We make new catalogs publicly available as soon as we have completed the necessary quality checks.

For new projects the most recent version of the catalog is always preferred. However, to make it easier to carry out research projects using the FIRST catalogs, we maintain an archive of all publicly released versions of the catalog. When there are questions about sources that have changed (or disappeared) in a new catalog, it is always possible to go back to the older version of the catalog for comparison.

Here we briefly summarize the changes in the various catalog releases. Catalogs are identified by the date of their release.

Current Catalog: 2014 December 17

This catalog from the 1993 through 2011 observations contains 946,432 sources from the north and south Galactic caps. It covers a total of 10,575 square degrees of sky (8,444 square degrees in the north and 2,131 square degrees in the south.)

This catalog is identical to the previous release (14Mar04) except that it includes more accurate information on which FIRST sources are not covered by the SDSS DR10 survey.

The catalog format is the same as the previous two versions (14Mar04 and 13Jun05), including information on the mean and rms epoch of observation for each source. As for versions since 12Feb16, the catalog includes a sidelobe probability estimate, and columns have been added with information on optical and infrared counterparts from the SDSS (DR10) and 2MASS catalogs.

In this version of the catalog, images taken in the the new EVLA configuration have been re-reduced using shallower CLEAN thresholds in order to reduce the "CLEAN bias" in those images. Also, the EVLA images are not co-added with older VLA images to avoid problems resulting from the different frequencies and noise properties of the configurations. That leads to small gaps in the sky coverage at boundaries between the EVLA and VLA regions. As a result, the area covered by this release of the catalog is about 60 square degrees smaller than the 13Jun05 release of the catalog, and the total number of sources is reduced by nearly 25,000. The older version of the catalog does have sources in the overlap regions, but their flux densities are considered unreliable due to calibration errors. The flux densities should be more accurate in this catalog, biases are smaller, and the incidence of spurious sources is also reduced.

2014 March 04

This catalog is exactly identical to the newer 14Dec17 release except that has inaccuracies in the information on which FIRST sources are not covered by the SDSS DR10 survey. See the catalog description for more information on this issue, which stemmed from problems in the SDSS CasJobs interface. The problem is fixed now, and the newer catalog should be used instead of this version.

2013 June 05

This catalog from the 1993 through 2011 observations contains 971,268 sources from the north and south Galactic caps. It covers a total of 10,635 square degrees of sky (8444 square degrees in the north and 2191 square degrees in the south.)

The catalog format differs from the previous version. Information on the mean and rms epoch of observation for each source has been added to the catalog. As for the preceding version (12Feb16), the catalog includes a sidelobe probability estimate, and columns have been added with information on optical and infrared counterparts from the SDSS (DR9) and 2MASS catalogs.

2012 February 16

This catalog from the 1993 through 2011 observations contains 946,464 sources from the north and south Galactic caps. It covers a total of 10,635 square degrees of sky (8444 square degrees in the north and 2191 square degrees in the south.)

The catalog format differs from the previous version. The contents of the sidelobe flag column has changed to a sidelobe probability estimate, and columns have been added with information on optical and infrared counterparts from the SDSS and 2MASS catalogs. There is no GSC2 information in this version.

2008 July 16

This catalog from the 1993 through 2004 observations contains 816,331 sources from the north and south Galactic caps. It covers a total of 9055 square degrees of sky (8444 square degrees in the north and 611 square degrees in the south.)

2003 April 11

This catalog from the 1993 through 2002 observations contains 811,117 sources from the north and south Galactic caps. It covers a total of 9033 square degrees of sky (8422 square degrees in the north and 611 square degrees in the south.)

The catalog format is the same as the previous version.

2001 October 15

This catalog from the 1993 through 2001 observations contains 771,076 sources from the north and south Galactic caps. It covers a total of 8565 square degrees of sky (7954 square degrees in the north and 611 square degrees in the south.)

For this version of the catalog, most of the maps constructed from the previous round of observations (in 1999) were remade to correct a small error in the annual aberration correction (due to a Y2K error in the observation date string.) This led to numerous small changes in the positions and flux densities for sources in the area covered by the 1999 observations. The differences are generally much smaller than the uncertainties: the mean position shift is 0.032 arcsec and the mean flux change is 0.015 mJy, though in rare cases they can be larger.

The catalog format is the same as the previous version.

2000 July 5

This catalog from the 1993 through 2000 observations contains 722,354 sources from the north and south Galactic caps. It covers a total of 7988 square degrees of sky (7377 square degrees in the north and 611 square degrees in the south.)

The catalog format is the same as the previous version.

1999 July 21

This catalog from the 1993 through 1998 observations contains 549,707 sources from the north and south Galactic caps. It covers a total of 6060 square degrees of sky (5450 square degrees in the north and 610 square degrees in the south.)

The catalog format is the same as the previous version. Note that the northern and southern catalogs are now combined into one file.

1998 February 4

This pair of catalogs from the 1993 through 1997 observations contains 382,892 sources from the north Galactic cap and 54,537 sources from the south Galactic cap. In the north it covers about 4150 square degrees of sky, including most of the area 7^h20^m < RA(2000) < 17^h20^m, 22.2° < Dec < 57.6°. In the south it covers about 610 square degrees of sky, including two narrow strips in the area 21^h20^m < RA(2000) < 3^h20^m, -11.5° < Dec < 1.6°.

The catalog format is the same as the previous version.

1997 April 24

This pair of catalogs from the 1993 through 1996 observations contains 236,177 sources from the north Galactic cap and 31,870 sources from the south Galactic cap. In the north it covers about 2575 square degrees of sky, including most of the area 7^h20^m < RA(2000) < 17^h20^m, 22.2° < Dec < 42.5°. In the south it covers about 350 square degrees of sky, including most of the area 21^h20^m < RA(2000) < 3^h20^m, -2.5° < Dec < 1.6°.

This is the first catalog of the southern survey region. It has a different format than previous versions, with additional columns giving the fitted major axis, minor axis, and position angle (before deconvolution). The old format was judged too difficult to use with the elliptical point-spread function of the southern images. The new format also gives the major and minor axes to higher precision than before.

1997 February 27

This catalog contains 236,040 sources from the 1993 through 1996 observations. It covers about 2575 square degrees of sky, including most of the area 7^h20^m < RA(2000) < 17^h20^m, 22.2° < Dec < 42.5°. The format of this catalog is identical to that used for the previous version.

In addition to having 1.7 times as many sources as the previous catalog, both from adding new survey area and from filling in holes in the coverage, this catalog has these changes and improvements:

The images used for the catalog have had a small emprical rotation and scale change applied, which can shift sources near the grid image corners by up to 0.2 arcsec. Also, higher-order interpolation has been used in summing the overlapping grid images. The result is that the catalog sources have peak flux densities that are systematically slightly higher (by about 0.7%) and sizes that are slightly smaller (typically by 0.15 arcsec in the deconvolved major axis for nearly unresolved sources.) All the data taken to date were reprocessed using the new imaging pipeline. The new coadded images have `E' as the final character of the field name.
Source positions are also shifted slightly due to the improved maps. Bright sources shift less than 0.2 arcsec (95% shift less than 0.06 arcsec). Fainter sources can shift more due to the changing effects of noise. For the entire catalog, 95% of the sources detected in the previous catalog have moved less than 0.22 arcsec and 99% less than 0.82 arcsec.
A small fraction (2.5%) of the sources from the previous catalog have disappeared (or moved by more than the 5.4 arcsec beam size). Almost all of these are near the flux detection threshold and have dropped below the level of detectability. Some are in regions where the maps have been improved through the addition of new data. Some are in complex regions where a different set of overlapping Gaussian components was used to model the radio emission. Many were actually sidelobes and were flagged as such in the previous catalog. Note that these same effects can also cause new sources near the detection thresholds to appear in the catalog.

If you are doing research on sources that have disappeared in this catalog, check their images using the FIRST Cutout Server to see whether they appear to be real sources or not. We would be interested to hear of cases where real radio sources have been omitted from this catalog; the number of such sources is known to be small, however.
We corrected a mistake in the assignment of field names to sources, so the field name has changed for many sources. We now choose the coadded field in which the source is farthest from any edge.

1996 May 28

This catalog is identical to the 95Oct16 version except for the position angles of sources with fitted minor axes smaller than the 5.4 arcsec beam (i.e. sources with negative values for the deconvolved minor axis.) Those sources all have zero position angles in the 95Oct16 catalog. The majority of such sources are unresolved so that the position angle is indeterminate. However, a minority have unresolved minor axes and resolved major axes; in those cases, the position angles in the 95Oct16 catalog are incorrect and should not be used.

Since this catalog is essentially identical to the previous version (it has the same number of sources, identical positions, and the same sort order), it should be straightforward to switch ongoing research projects based on the 95Oct16 version to the 96May28 catalog.

1995 October 16

This catalog contains 138,665 sources from the 1993 and 1994 observations. It covers about 1550 square degrees of sky, including most of the area 6^h35^m < RA(2000) < 17^h30^m, 28° < Dec < 42°.

In addition to having 5 times as many sources as the previous catalog, this catalog has these changes and improvements:

The rms noise level is more accurate in fields with bright sources.
Sources that are probably sidelobes of bright objects are flagged.
The CLEAN bias correction has been added to flux densities.
The interpretation of the major and minor axis sizes has changed for unresolved sources.
The format has changed slightly.

1995 January 6

The initial FIRST catalog was a preliminary release known to have some shortcomings, including regions of bad data. It covered only about 300 square degrees and included 27,000 sources. All this area is included in subsequent higher-quality catalogs. The current catalog is of sufficiently improved quality that it should be used in place of the 95Jan06 version for new or existing research projects.

The FIRST image archive at the Space Telescope Science Institute is now fully up-to-date with the current catalog and the cutout server. Access to the full-sized FIRST images is now quite easy and no longer requires a Hubble archive account. The data can be retrieved either through a search interface or through anonymous ftp.

XXX

Over most of the survey area, the detection limit is 1 mJy. A region along the equatorial strip (RA = 21.3 to 3.3 hrs, Dec = -1 to 1 deg) has a deeper detection threshold because two epochs of observation were combined. The typical detection threshold in this region is 0.75 mJy. There are approximately 4,500 sources below the 1 mJy threshold used for most previous versions of the catalog.

The catalog is on-line and can be searched using the FIRST Search Engine. However, for large-scale investigations it will be necessary to obtain the complete catalog. This document describes the catalog format.

The catalog is sorted by decreasing declination and has the following format:

#                                                                                                               ------SDSS------ -----2MASS----  Epoch     Epoch     Epoch  
#  RA           Dec        P(S)    Fpeak      Fint    RMS     Maj    Min   PA    fMaj   fMin   fPA Field         #  Sep    i  Cl  #  Sep    K   Mean-yr   Mean-MJD   rms-MJD
07 27 34.289 +64 40 59.80 0.197     1.00      1.12   0.139   2.13   1.58   0.2   5.80   5.63   0.2 07300+64243J -1 99.00 99.00 -  1  6.76 14.62 2002.687 2452526.1     0.002
07 38 39.304 +64 40 16.28 0.014     2.39      5.73   0.139   9.31   3.58 136.6  10.76   6.48 136.6 07360+64243J -1 99.00 99.00 -  0 99.00 99.00 2002.687 2452526.1     0.001
07 50 24.019 +64 40 01.21 0.014    22.09     24.00   0.140   1.96   1.13   6.8   5.74   5.52   6.8 07480+64243J -1 99.00 99.00 -  0 99.00 99.00 2002.687 2452526.1     0.002
07 38 45.622 +64 39 50.12 0.014     2.39      4.33   0.140   6.43   3.27  13.5   8.40   6.31  13.5 07360+64243J -1 99.00 99.00 -  0 99.00 99.00 2002.687 2452526.1     0.000
07 39 32.799 +64 39 18.03 0.082     1.41      1.25   0.139   2.63   0.00  37.5   6.01   4.30  37.5 07420+64243J -1 99.00 99.00 -  0 99.00 99.00 2002.687 2452526.1     0.002

The format of this catalog is the same as the previous release (13Jun05) but differs from earlier versions of the catalog. The last 3 columns give information on the epoch of observation for each source and are described below in the observation epoch section. The P(S) column, which indicates the probability that the source is a sidelobe, replaces the previous binary sidelobe flag column. The columns following the field name give information on counterparts to the FIRST source in the SDSS DR10 catalog and the 2MASS catalog. These parameters are described in detail below in the description of the P(S) column and the counterparts columns. Other catalog columns are common with FIRST catalog releases extending back over the past decade.

Names for FIRST sources are derived from the positions according to IAU recommendations.
RA, Dec are the position (J2000) of the source. The accuracy of the position depends on the brightness and size of the source and the noise in the map. Point sources at the detection limit of the catalog have positions accurate to better than 1 arcsec at 90% confidence; 2 mJy point sources in regions with typical noise levels have positions good to 0.5 arcsec. An empirical expression for the positional accuracy is
```
   unc(90% confidence) = Size  (1/SNR + 1/20)  arcsec
```
where Size is either the major or minor axis fitted FWHM (fMaj or fMin) as given in the catalog and SNR is the peak flux density signal-to-noise ratio:
```
   SNR = (Fpeak-0.25) / Rms
```
(The positional uncertainty is of course elliptical for elliptical sources.) The best possible positional uncertainty is limited to about 0.1 arcsec by our ability to fit source positions in maps with 1.8 arcsec pixels and by various calibration uncertainties. Systematic errors in the positions are smaller than 0.05 arcsec.

P(S) indicates the probability that the source is spurious (most commonly because it is a sidelobe of a nearby bright source.) Low values mean the source is unlikely to be spurious. Here is the distribution of P(S) values in the catalog:

Probability Range	Fraction of sources
0.00 < P_S < 0.05	76.3%
0.05 < P_S < 0.15	7.0%
0.15 < P_S < 0.25	4.4%
0.25 < P_S < 0.35	3.5%
0.35 < P_S < 0.45	2.1%
0.45 < P_S < 0.55	1.7%
0.55 < P_S < 0.65	1.6%
0.65 < P_S < 0.75	1.5%
0.75 < P_S < 0.85	1.1%
0.85 < P_S < 0.90	0.9%

Sidelobe probabilities for this version of the catalog have been computed using an improved algorithm based on multiple voting oblique decision tree classifiers. The classifiers were trained using deep VLA fields that give reliable assessments of the reality of FIRST sources. The algorithm will be described in detail in a future paper on the final FIRST catalog; we believe it is a substantial improvement over the previous sidelobe flagging approach. Nevertheless, we still recommend checking the images using the FIRST Cutout Server if there is doubt about the reality of particular sources. This is easily done when using the FIRST Search Engine to search the catalog, since each source selected in the search has a link to the Cutout Server.

Fpeak and Fint are the peak and integrated flux densities measured in mJy. They are derived by fitting an elliptical Gaussian model to the source. To correct for the "CLEAN bias" effect, 0.25 mJy has been added to the peak flux density and the integrated flux density has been multiplied by (1+0.25/Fpeak) (see our Astrophysical Journal paper and our catalog paper for more details.)
The uncertainty in Fpeak is given by the rms noise at the source position, while the uncertainty in Fint can be considerably greater depending on the source size and morphology. For bright sources the accuracies of Fpeak and Fint are limited to about 5% by systematic effects. Note that for sources that are not well-described by an elliptical Gaussian model, Fint is not an accurate measure of the integrated flux density.
Rms is a local noise estimate at the source position measured in mJy. Rms is computed by combining the measured noise from all grid pointing images contributing to this coadded map position. Note that the significance of detection for a source is (Fpeak-0.25)/Rms, not Fpeak/Rms, because of the CLEAN bias correction to the peak flux density. The catalog includes only sources brighter than 5 Rms.
FITS images giving the rms noise as a function of position on the sky are available for the northern and the southern areas. These images give the rms in mJy/beam tabulated on a ~3 arcmin grid in RA and Declination. If there is no source in the catalog at a given position, the source peak flux density (before CLEAN bias correction) is less than 5 times the coverage map rms value at that position. The sky area covered is displayed in the images below (which are linked to postscript versions of the images):

The coverage in the southern Galactic cap is ragged (and discontiguous) at the northern edge because poor weather and equipment failures prevented the planned survey area from being filled in there. Small gaps can be seen around the edges of the area observed using the EVLA configuration in 2011 (see above for more details).

Note that the FIRST Search Engine gives the rms noise estimate at the specified sky position, as derived from these coverage maps. That is the most convenient way to determine the noise at a particular position in the survey (or to determine whether a position has been observed.)
Maj, Min, and PA give the major and minor axes (FWHM in arcsec) and position angle (degrees east of north) derived from the elliptical Gaussian model for the source. Maj and Min have been deconvolved to remove blurring by the elliptical Gaussian point-spread function. (The fitted parameters before deconvolution are given in the fMaj, fMin, and fPA columns.) In the north the beam is circular 5.4 arcsec FWHM; south of declination +4°33'21", the beam is elliptical, 6.4x5.4 arcsec FWHM, with the major axis running north-south. In the southern Galactic cap (RA = 21hrs to 3hrs), the elliptical beam size increases further to 6.8x5.4 arcsec south of declination -2°30'25".
Noise can cause the fitted values of the major and minor axes (before deconvolution) to be smaller than the beam. The corresponding deconvolved size is given as zero in that case.

The uncertainties in the deconvolved sizes depend on both the brightness and the sizes. Objects at the catalog flux density limit have uncertainties of about 2 arcsec in their sizes (so faint objects with Maj < 2 arcsec are consistent with point sources.) An simple empirical estimate of the uncertainty is
```
   Sigma(Size) = 10 arcsec  (1/SNR + 1/75)
```
where SNR is the signal-to-noise ratio defined above.
fMaj, fMin, and fPA give the major and minor axes (FWHM in arcsec) and position angle (degrees east of north) derived from the elliptical Gaussian model for the source. These are the fitted sizes measured directly from the image; the elliptical point-spread function has not been deconvolved.
The Field Name is the name of the coadded image containing the source. Note that the field name encodes the position of the field center: field hhmmm+ddmmm is centered at RA=hh mm.m, Dec=+dd mm.m. The images are available from several archives and through the FIRST Cutout Server.
All field names in the current catalog end with a letter E through X, depending on the date of the last catalog release in which the image was modified. The W and X fields are new in this catalog, while sources extracted from the E-T fields are essentially identical to those in the previous version of the catalog. The W fields include contributions from EVLA data taken in Spring 2011 and have both a slightly different central frequency (1.335 GHz instead of 1.400 GHz) and typically higher noise levels than the older images. The X fields are images that neighbor the EVLA fields but differ from the previous release in that they omit nearby EVLA observations from the co-adding of overlapping grid images.

The next 7 columns give information on optical counterparts from the DR10 release of the Sloan Digital Sky Survey and infrared counterparts from the 2MASS catalog. The SDSS and 2MASS matches were performed using the SDSS-III CasJobs web interface.

For each catalog there is a column giving the number of matches within a fiducial radius of 8 arcsec and, for the nearest match, the separation from the FIRST position in arcsec and the magnitude. For SDSS the morphological classification is also given (s=stellar, g=nonstellar/galaxy). A count of zero indicates there are no sources within this radius (which is also indicated by a separation of 99.00 and a classification of '-'). For SDSS, a count of -1 indicates that the FIRST source falls outside the SDSS DR10 survey area so that no SDSS data are available. All of these catalogs have multiple wavebands available; the given magnitudes are for i (SDSS) and K (2MASS). For SDSS, a magnitude of -1 indicates that the magnitude in the DR10 catalog was given as -9999. Note that these parameters can be used for catalog searches via the "Additional Constraints" field in the FIRST Search Engine web form. See the Additional Constraints Help Page for more information.

Warning: The sky coverage for SDSS DR10 relies on the fInFootprintEq SQL function, which currently gives the wrong answer in some sky areas because it includes observations that were not actually included in the DR10 catalog. We have attempted to adjust the coverage empirically, but unfortunately the information in this catalog on the SDSS sky coverage is somewhat unreliable in this version of our catalog. The errors affect a few percent of the FIRST sources. We will release an updated catalog when this problem with the DR10 database has been corrected.

Some information summarizing the frequency of counterparts in the various catalogs is given in the table below:

Survey	Characteristic	Percentage
SDSS	1 or more counterparts < 8"	54.2%
SDSS	Counterpart < 1.5"	27.9%
SDSS	Covered by DR9	93.1%
2MASS	1 or more counterparts < 8"	12.5%
2MASS	Counterpart < 1.5"	7.4%

The final 3 columns give information on the mean epoch of the FIRST flux density measurement. The FIRST survey consists of 3-minute integrations at a grid of points on the sky. But the catalogs are derived from coadded images that use a weighted sum of the individual pointing maps to achieve nearly uniform sensitivity. (See the FIRST survey paper for more details.) That means that each measurement is a weighted sum of multiple observations.
The Epoch Mean columns give the weighted mean of all the contributing pointing epochs at the position of the source. It is given both in decimal years and in MJD for convenience. The Epoch rms column (MJD) gives the weighted rms of the pointing epochs at the source position. It is a measure of the spread in epochs that contribute to the measurement. Many sources have small rms values of only a few minutes (dominated by a single 3-minute pointings or by 2 adjacent pointings), but values of days to weeks are also common (for sources observed in the overlap between declination strips), and some objects have rms values of years (for sources observed at the edges of regions in different observing seasons or that were observed multiple times due to data problems). The largest epoch rms in the survey is 7.7 years.

The epoch rms should be used as a guide to identify objects that do not have well-defined epochs. The table below gives an indication of the frequency of different epoch rms values in the catalog.

Epoch rms range Fraction

< 5 minutes 6%

5 min — 1 day 28%

1 — 10 days 44%

10 — 100 days 11%

100 days — 1 yr 7%

1 — 2 yrs 1%

2 — 5 yrs 3%

> 5 yrs 1%

Epoch rms range	Fraction
< 5 minutes	6%
5 min — 1 day	28%
1 — 10 days	44%
10 — 100 days	11%
100 days — 1 yr	7%
1 — 2 yrs	1%
2 — 5 yrs	3%
> 5 yrs	1%

The catalog history page describes previously released versions of the FIRST catalogs, which are still available for historical purposes. We recommend that the new catalog be used where possible for all projects.