1. Introduction
The Indian Space Research Organisation (ISRO) launched Astrosat (Agrawal Reference Agrawal2006) on September 28, 2015. It was India’s first dedicated multi-wavelength astronomy satellite and included 5 payloads: the Soft X-Ray imaging Telescope (SXT); the Large Area X-Ray Proportional Counters (LAXPC); the Cadmium Zinc- Telluride Imager (CZTI); the Scanning Sky Monitor (SSM); and the Ultra Violet Imaging Telescope (UVIT), the only ultraviolet (UV) telescope onboard.
There are two telescopes on the UVIT payload which simultaneously image the sky in the far ultraviolet (FUV: 1 300–1 800 Å) in one telescope, and the near ultraviolet (NUV: 2 000–3 000 Å) and visible (VIS: 3 200–5 300 Å) in the second telescope, with the bands separated using a dichroic mirror. The field of view (FOV) is
$\sim$
$28$
′ diameter with a spatial resolution of
$\sim$
$1.4$
′′. Specific bands are selected through a filter wheel on each telescope. The VIS data were intended to be used only for astrometric correction and have no scientific utility. The instrument has been described in Kumar et al. (Reference Kumar2012).
An important product of a wide-field imaging mission is a point source catalogue that may be correlated with catalogues in other wavelengths to yield a multi-wavelength picture of the sky. While numerous catalogues exist in the optical (e.g. Sloan Digital Sky Survey (SDSS) and Panoramic Survey Telescope and Rapid Response System (Pan-STARRS)) and infrared (e.g. 2-Mass Catalogue), fewer options are available in the UV due to the need for space-based observations. An early attempt was the TD-1 catalogue (Boksenberg et al. Reference Boksenberg1973; Thompson et al. Reference Thompson, Nandy, Jamar, Monfils, Houziaux, Carnochan and Wilson1978) from the TD-1 mission, a satellite operated by the European Space Research Organisation (ESRO). The Ultraviolet Sky Survey Telescope onboard TD1 measured the absolute ultraviolet flux distribution of point sources between 1 350–2 550Å, containing about 31 215 stars with a visual magnitude limit of
$\sim$
10 for unreddened early B type stars in its six-month observation period. The deepest UV catalogue to date is from the Galaxy Evolution Explorer (GALEX) (Martin et al. Reference Martin2005; Bianchi, Shiao, & Thilker Reference Bianchi, Shiao and Thilker2017), which was observed in two main bands (1 350–1 750Å and 1 750–2 750Å). The GALEX catalogue includes data from three surveys: All-sky Imaging Survey (AIS), Medium Imaging Survey (MIS), and Deep Imaging Survey (DIS), covering approximately 25 000 sq. degrees of sky in its final data release. The AIS provides a magnitude limit of
$\sim$
20, while the MIS and DIS can detect sources up to magnitude limits of
$\sim$
23 and
$\sim$
25, respectively.
There exist catalogues of small portions of the sky observed by UVIT. Leahy et al. (Reference Leahy, Postma, Chen and Buick2020) observed about 18 regions of the M31 and generated a catalogue of
$\sim$
75 000 sources with a limiting magnitude of
$\sim$
23 in FUV CaF2-1 filter. Devaraj et al. (Reference Devaraj, Joseph, Stalin, Tandon and Ghosh2023) created a point source catalogue from the observation of 3 overlapping fields in the outskirts of the SMC by the UVIT. The total number of sources detected was
$\sim$
11 241. The catalogue provided information about their AB magnitude in 7 UVIT filters, of which 3 are in the FUV and 4 in the NUV. In our work, we cover a larger area of the sky with a higher number of field images, including a few fields of SMC.
UVIT has completed about
$\sim$
1 700 observations (including multiple observations of a single field) after almost 9 yr of observations, and we believe that it is time to construct a point source catalogue (‘UVIT-cat’). We begin with a sample of 428 fields in the FUV and 54 fields in the NUV covering about
$\sim$
63 square degrees of the sky and plan to expand the catalogue to the entire UVIT data set. In Section 2, we explain the details of the UVIT instrumentation and data collection; in Section 3, we discuss the analysis and results; and in Section 4, we give a summary of our catalogue and future work.
2. UVIT instrument and data
The Ultra Violet Imaging Telescope (UVIT) observed the sky in two UV and one visible band using two 38 cm telescopes. One telescope observed the far ultraviolet (FUV: 1 300–1 800 Å) while the other split the light into the near ultraviolet (NUV: 2 000–3 000 Å) and visible (VIS: 3 200–5 300 Å) bands. There are three identical intensified microchannel plates with an aperture of 40mm, and only the photocathode differs as given in Kumar et al. (Reference Kumar2012). A filter wheel on each telescope divided the spectral region into several bands (Table 1). The UVIT image pixel size is
$\simeq$
0.4168
$^{\prime\prime}$
by 0.4168
$^{\prime\prime}$
. The two UV detectors were operated in photon counting mode while the VIS channel was used only for attitude correction and was operated in integration mode.
Table 1. Filter details used for observations.
UVIT data are processed and archived at the Indian Space Science Data Centre (Singh et al. Reference Singh2014) and may be downloaded from the Astrobrowse Archive Footnote a in the following formats:
-
1. Level 0: The Level 0 is the raw binary data as observed by the UVIT payload onboard the AstroSat, along with its auxiliary data. This is sent to the AstroSat Data Centre at Indian Space Science Data Centre (ISSDC) for further processing into Level 1 data.
-
2. Level 1: The Level 1 data from the ISSDC is sent to the Payload Operation Centre (POC) at the Indian Institute of Astrophysics (IIA), Bengaluru. The files in the Level 1 data are FITS binary table files organised according to the orbit number under a single top-level directory as described in Rahna, Murthy, & Safonova (Reference Rahna, Murthy and Safonova2021).
-
3. Level 2: These are the final products after processing the Level 1 data through the UVIT pipeline software. The final product is a FITS image file, containing the coordinate information and can be read by any common astronomical data processing program.
As of 1st June 2024, there were a total of 1 713 observations in the UVIT archive in both channels. The NUV channel ceased operation in March 2018, and only FUV data were taken since then. We have only used observations with an exposure time greater than 200 s. These are distributed in the sky as shown in Fig. 1 and exclude the Galactic Plane, where there were no observations due to concerns about the sky brightness. The blue dots represent the images observed only through the FUV filter, while the green dots represent sources observed only through the NUV filter. We represent the pointings as red dots for field images observed through both FUV and NUV.
Figure 1. Area of sky covered by UVIT-cat. Blue: FUV fields only, Green: NUV fields only, and Red: Field images containing both FUV and NUV.
2.1 Astrometric correction and source extraction
The astrometry in the Level 2 archival data file is off by several arcminutes, and each field has to be corrected individually. We first ran each image through Astrometry.net by Lang et al. (Reference Lang, Hogg, Mierle, Blanton and Roweis2009), which used a set of stars (or galaxies) from a reference catalogue (typically optical catalogues such as Tycho-2 and 2MASS) to solve for the astrometry in that image. This worked well for the NUV, where the sky looks similar to the optical, but the FUV sky is quite different from the visible sky, and we had to create a custom catalogue based on GALEX FUV observations. There were still a few problematic fields, generally because there were too few stars in the field, which we corrected by matching stars by eye. The FITS files of the image headers having the updated World Coordinate system are in the GitHub repository.Footnote b
We used SExtractor v2.25.0 (Bertin & Arnouts Reference Bertin and Arnouts1996) to extract point sources from the astrometrically-corrected FITS files. We used the default parameters for extraction except as listed below:
-
1. We changed DETECT_MINAREA (the minimum number of adjacent pixels that have to be above a certain value for a detection) to 5 from 3 as per the PSF of the instrument.
-
2. The value for DETECT_THRESH was set to be 3 for the FUV and 5 for NUV images to minimise the number of false detections. Setting a smaller threshold value (e.g. 1.5 times background) increased the false detection counts of higher exposure images. Hence, we chose an optimal value for the detection threshold such that there is minimum detection of false positives across the entire dataset.
-
3. We used the pixel scale of UVIT (0.4168
$^{\prime\prime}$
) for PIXEL_SCALE -
4. The zero point magnitude (MAG_ZEROPOINT) of the instrument was varied for each filter, as per Tandon et al. (Reference Tandon2020).
The detailed parameters used for source extraction are listed in Table 2.
Table 2. Source-Extractor default parameters.
Table 3. Total number of sources through FUV filter.
Table 4. Total number of sources through NUV filter.
Table 5. UVIT-cat catalogue summary.
The source-extractor calculates the error in magnitudes by considering the error in the flux, which is given as:
\begin{align} \sigma_{F} = \sqrt{\sigma^2_{photon}+ \sigma^2_{sky} + \sigma^2_{read}} \end{align}
where,
$\sigma_{photon}$
is the photon noise,
$\sigma_{sky}$
is the background noise and
$\sigma_{read}$
is the readout noise associated with the detector.
The photon noise can be represented as the square root of flux in photon counts (F) from the source as:
\begin{align} \sigma_{F} = \sqrt{F} \end{align}
The magnitude error,
$\Delta m$
, is calculated using the flux error value from equation (1) as:
\begin{align} \Delta m &= \frac{2.5}{\ln10}\left(\frac{\sigma_{F}}{F}\right) \end{align}
The detection threshold parameters and the error calculation algorithm can be found in the manual at Source-extractor official website.Footnote c
Figure 2. Exposure time histogram of 3 bins in FUV band, separated by (a) green: 2 000 s and (b) red: 10 000 s.
Figure 3. AB magnitude vs Source count distribution for 3 bins in FUV band.
Figure 4. (a) Histogram of exposure times in NUV band (b) AB magnitude vs Source count distribution of NUV filters.
2.2 Merging of duplicate sources
We combined the point source list from each observation into a single catalogue. Any detection within 1.5
$^{\prime\prime}$
of another was merged into a single source. A single binary FITS binary table is created from all observations of a single field, regardless of detector or filter. If multiple detections of the same source were present in the same filter, we took the mean of the magnitudes and flux. If multiple detections of a source were present in different filters, all the information is condensed into a single row in the FITS table. The number of sources having entries in FUV and NUV are given in Tables 3 and 4.
3. Result and analysis
3.1 Image to catalogue script
We have modified jude_call_astrometry procedure of JUDE (Jayant’s UVIT Data Explorer) (Murthy et al. Reference Murthy, Rahna, Sutaria, Safonova, Gudennavar and Bubbly2017; Rahna, Murthy, & Safonova Reference Rahna, Murthy and Safonova2021) to process the Level 2 UVIT images, correct the astrometry, and output a FITS binary table catalogue file for each image. The columns are listed in Table 5. We then merged all the field catalogues into a single file (UVIT-cat). We have included all instances of different observations of the same field.
The UVIT catalogue without any merging had about
$\simeq117\,598$
sources, of which
$14\,825$
were duplicates. We merged the duplicate rows to come up with the UVIT-cat catalogue that contains 102 773 NUV and FUV sources, observed through different filters of UVIT.
The matching algorithm merges all sources within a 1.5
$^{\prime\prime}$
radius to give us a final merged catalogue. The algorithm finds all sources within the radius by flagging them as duplicates, extracts the magnitudes, fluxes, and their errors from each row in a single filter, and computes the average, which is taken as the final photometry, and merged into a single row. If duplicate sources are found in separate filters, all the rows are merged into a single row with photometric entries in separate columns.
Many field images with extended sources like galaxies, nebulas, and densely crowded fields, are not yet included, and we intend to put forward updated versions of the catalogue by including more UVIT field images, both in FUV and NUV.
3.2 Completeness and analysis of the catalogue
The number of fields along with their respective number of sources observed through each FUV and NUV filter of UVIT is given in Tables 3 and 4, respectively. We find the highest number of observations and detections in the first filter of both NUV and FUV channels. 44 289 sources were found in the F148W filter (CaF2-1; F1) of FUV channel out of the 190 observations. A similar (157) number of observations were made through the filter F154W (BaF2; F2), but significantly less number of detections were made (16 161) as shown in Table 3. The number of sources detected is about
$\simeq$
$24\,562$
as detected in the 26 fields of the NUV N242W filter. The next set of filters (N219M, N245M, N263M, and N279N) detected about 546, 9 965, 1 942, and 3 238 sources, respectively. We found that the highest exposure time images were mostly observed through the F1 filter of FUV and NUV compared to the other filters, and this accounts for the highest number of detections of the first filter in each channel.
The images that were analysed had a large variation of exposure time (from
$\sim$
400 s to more than 45 000 s), influencing the catalogue’s completeness. To facilitate statistical studies, we divided the images into three bins based on exposure times. The first bin contained images with 2 000 s or less exposure timing. The second bin ranged from 2 000 to 10 000 s, and any image with exposure time greater than 10 000 s was put into the third bin. The bins contained 246, 142, and 40 FUV images, respectively. As shown in Fig. 2, the magnitude limit will correspond to the median value of each of the three bins. For the first bin, we find the typical exposure of FUV images to be
$\sim$
$1\,798$
s. The second and third bins had the median exposure time at
$\sim$
2 500 and
$\sim$
11 000 s, respectively. The bins are separated with two vertical lines, green and red at the 2 000 and 10 000 s, respectively. The histogram of the exposure time of NUV images is shown in Fig. 4a, and we find out the median value to be
$\sim$
400 s.
Table 6. Peak AB magnitude of source-count distribution of each filter of FUV.
Table 7. Magnitude limit of all the filters of FUV in three bins.
The distribution of the AB magnitude of the detected sources in each of the 3 bins of the FUV channel and NUV channel is shown in Figs. 3 and 4b. The plots are shown for images observed via different filters of the FUV and NUV channels of UVIT.
We find the peak of the source count distribution as 20.527, 20.81, and 21.98 for the F1 filter in the three respective bins of exposure times as shown in Fig. 3a, b, and c. The peak magnitude distribution for the remaining filters of UVIT in FUV is tabulated in Table 6.
To compute the completeness of the catalogue from the source count distribution in Figs. 3 and 4b, we find the turnover point where the number count of the sources falls drastically. The results are given in Table 7. We see that the completeness increases in Bin3, that is, with higher exposure time.
In the NUV band, the N242W band with the highest number of detections has the turnover point around the magnitude
$\sim$
22.602.
To determine the completeness from the magnitude vs magnitude error plot, we find the 5-
$\sigma$
detection limit of the catalogue’s data. This corresponds to an error cut of 0.198; that is, the faintest source which has a magnitude error of 0.198 or lower. The variation of magnitude error as a function of brightness (in AB magnitude) in the three bins is shown in Fig. 5. For the F148W filter, the plot generated from each of the three bins shows the completeness at
$\sim$
23.98, 24.30, and 24.91, respectively, as shown in Fig. 5a, c, and e. For the second filter, we find the 5-
$\sigma$
point at
$\sim$
23.35, 23.80, and 25.12 in all the three bins, as given in Fig. 5b, d, and f. We find the limiting magnitude for the NUV N242W filter at about
$\sim$
23.027 (Fig. 6a).
Figure 5. AB Magnitude vs AB Magnitude error for filters: FUV CaF2-1 (148W) and BaF2 (F154W) of UVIT. (a),(b): Bin 1, (c),(d): Bin 2 and (e),(f): Bin 3.
Figure 6. AB Magnitude vs AB Magnitude error for NUV filters: (a) NUV Silica-1 (N242W) and (b) NUVB15 (N219M) of UVIT.
Figure 7. Angular separation distribution of UVIT with (a) GAIA EDR3 (b) GALEX.
As we have combined all the observations into a single catalogue, we found that around
$\sim$
1 800 s exposure time, the highest number of observations has been made by UVIT. Hence, we can consider the magnitude limit of Bin 1 as the completeness of the entire ‘UVIT-cat’ catalogue, that is, for FUV, we get the magnitude limit to be around
$\sim$
21, while for NUV, we find the completeness at around
$\sim$
23.
3.3 Positional offset analysis
We cross-matched the positions of our own UVIT catalogue using the GALEX GUVcat and Gaia EDR3 catalogue. For the following analysis, we plot and visualise the histogram distribution of angular separation of matched sources between the UVIT-cat catalogue (
$\alpha$
,
$\delta$
) and the reference catalogue position (
$\alpha$
,
$\delta$
) values, respectively. We keep the separation threshold as 5
$^{\prime\prime}$
and find all matches, that is, we consider the source as a match if it falls within this radius. This is shown in Fig. 7.
We found the angular separation cross-matched with Gaia EDR3 to be
$\simeq$
$0.240$
′′, as given in Fig. 7a. Similarly, we found the average angular separation when compared with GALEX’s GUVcat to be
$\simeq$
$0.602$
′′, as given in Fig. 7b. This shows that our positions of the detected sources are in good agreement astrometrically with the already existing catalogues.
4. Conclusion and future work
In our work, we have analysed 428 FUV and 54 NUV UVIT field pointings and present UVIT-cat catalogue. The UVIT-cat covers a total of
$\sim$
63 square degrees. We have come up with a catalogue of
$\simeq$
102 773 sources in the 5 far ultraviolet and the near ultraviolet bands. The number of entries in each filter is tabulated in Tables 3 and 4. The magnitude limit after statistical analysis is found to be,
$\sim$
21.28, 21.03, 20.59, 19.523, and 21.474 for all FUV band filters, respectively. The catalogue contains 85 columns of data, and the description of the column keys is given in Table 5. The catalogue will help the scientific community study objects such as QSOs, white dwarfs, young galaxies, and various astrophysical sources and provide a clear picture of the sky in the ultraviolet band.
Acknowledgement
This publication uses the data from Indian Space Research Orgranisation’s (ISRO) AstroSat mission. The data was available at the organisation’s archive Indian Space Science Data Centre(ISSDC), where the data was processed at the Payload Operation Center (POC) at the Indian Institute of Astrophysics (IIA). The UVIT mission is accomplished in collaboration between IIA, Inter-University Center for Astronomy and Astrophysics (IUCAA), Tata Institute of Fundamental Research (TIFR), ISRO and Canadian Space Agency (CSA). PS acknowledges Manipal Centre for Natural Sciences, Centre of Excellence, Manipal Academy of Higher Education (MAHE) for facilities and support. The authors would like to thank the anonymous reviewers for their insightful reviews and comments.
Softwares: IRAF (Tody Reference Tody1986), Astropy (Price-Whelan et al. Reference Price-Whelan2022), Numpy (Harris et al. Reference Harris2020), Pandas (McKinney et al. Reference McKinney2011), Matplotlib (Hunter Reference Hunter2007), DS9, JUDE (Murthy et al. Reference Murthy, Rahna, Sutaria, Safonova, Gudennavar and Bubbly2017; Rahna, Murthy, & Safonova Reference Rahna, Murthy and Safonova2021), Astrometry.net (Lang et al. Reference Lang, Hogg, Mierle, Blanton and Roweis2009).
Data Availability: The complete electronic version of the UVIT-cat catalogue in FITS format and the header files containing the WCS information can be found in the GitHub repository.Footnote d
