Public Data Release 10

Last updated: 2022-03-08

Contents

  1. Summary of the ZTF Surveys
    1. Acknowledging ZTF and Referencing Data Product usage
  2. Scope of this Release
  3. Summary of Data Products
  4. File-based Product Statistics
  5. Source and Lightcurve Statistics
  6. DR10 Sky Coverage
  7. Lightcurve products: timespans, cadence, and epoch distributions
  8. File-based Products in Archive
    1. Single-Exposure Image Products
    2. Reference Image Products
    3. Other File Products
  9. Epoch-based Data Quality Flags and Filtering
    1. Quality Filtering of Single-Exposure File Products
    2. Quality Filtering of Epochs in Lightcurves
  10. What you should know about the ZTF Release Products
  11. Extended (detailed) Advisories and Cautionary Notes
  12. How to Access ZTF Data Products
    1. Product-specific Access GUIs and Recipes
    2. Example Queries using the APIs
      1. Querying Single-Exposure Science Image Products
      2. Querying Reference Image Products
      3. Querying Lightcurves
      4. Solar System Object Precovery
    3. Bulk Download of all Lightcurves
  13. Help Desk
  14. Further Documentation & Relevant Publications

 

An overview of ZTF and the science objectives is given in Bellm et al. 2019 and Graham et al. 2019. The ZTF Science Data System (ZSDS) is housed at IPAC/Caltech and is described in Masci et al. 2019. This page provides a summary and quick guide to the ZTF Public DR10 products.

Please familiarize yourself with all advisories highlighted in red before accessing and using the products.

Click to jump to the Data Access & Recipes section.


1. Summary of the ZTF Surveys

During Phase-I of the survey (March 2018 - September 2020), ZTF observing time was split three ways:
(i) The Public Survey (40%);
(ii) Private Collaboration and Partnership Surveys (40%);
(iii) Programs granted by the Caltech Time Allocation Committee (20%).
Funding is provided by both the ZTF collaboration and the U.S. National Science Foundation through the Mid-Scale Innovations Program (MSIP).

The Public Survey in particular enabled a wide range of community science and constituted the following:

Phase-II began in December 2020, with 50% of available observing time allocated to a public survey of the Northern sky in g and r bands with a two-night cadence; and automatic classification of supernovae using the SED Machine spectrometer on the Palomar 60-inch telescope.

1.a. Acknowledging ZTF and Referencing Data Product usage

For publications that use ZTF Science Data Products from Phase-I of the survey (taken before December 1, 2020), please include the following text in your acknowledgments:

Based on observations obtained with the Samuel Oschin 48-inch Telescope at the Palomar Observatory as part of the Zwicky Transient Facility project. ZTF is supported by the National Science Foundation under Grant No. AST-1440341 and a collaboration including Caltech, IPAC, the Weizmann Institute for Science, the Oskar Klein Center at Stockholm University, the University of Maryland, the University of Washington, Deutsches Elektronen-Synchrotron and Humboldt University, Los Alamos National Laboratories, the TANGO Consortium of Taiwan, the University of Wisconsin at Milwaukee, and Lawrence Berkeley National Laboratories. Operations are conducted by COO, IPAC, and UW.

For publications using products from Phase-II of the survey (taken on or after December 1, 2020), please include this text:

Based on observations obtained with the Samuel Oschin Telescope 48-inch and the 60-inch Telescope at the Palomar Observatory as part of the Zwicky Transient Facility project. ZTF is supported by the National Science Foundation under Grant No. AST-2034437 and a collaboration including Caltech, IPAC, the Weizmann Institute for Science, the Oskar Klein Center at Stockholm University, the University of Maryland, Deutsches Elektronen-Synchrotron and Humboldt University, the TANGO Consortium of Taiwan, the University of Wisconsin at Milwaukee, Trinity College Dublin, Lawrence Livermore National Laboratories, and IN2P3, France. Operations are conducted by COO, IPAC, and UW.

Furthermore, we appreciate you citing the following publication:

Masci, F. J., Laher, R. R., Rusholme, B., et al. 2018, The Zwicky Transient Facility: Data Processing, Products, and Archive, PASP, 131, 995.


2. Scope of this Release

The Tenth ZTF Public Data Release (DR10) builds upon the Ninth Data Release (DR9) to include products from (i) an additional two months of data acquired under the Public Survey (+ July 2021 to Sept 2021) and (ii) data acquired under Private Survey time (from both partnership and Caltech programs) from approximately the first 30.0 months of science operations (~ March 2018 to September 2020). The latter is included due to the 18 month proprietary period up to this release date.

More specifically, DR10 includes the following:

  1. Single Exposure Products associated with individual science observation epochs (lightcurve epochs) in g and r filters only satisfying:
  2. Single Exposure Products associated with individual science observation epochs (lightcurve epochs) in g, r, and i filters satisfying:
  3. All Reference (co-add) Image Products available in the ZTF Archive. Only good quality science exposures from all survey programs, irrespective of the epoch spans above, are included in the co-adds. Reference images are updated continually as data are taken.

NOTE: The public distribution of transient alerts extracted from difference images constructed from products (1) and (3) above commenced on June 4, 2018. These alerts continue to be generated and distributed using new (unreleased) image data from Phase-II of the Public Survey.

NOTE: IRSA provides UI and API access to object tables and associated lightcurves for the last 5 releases (due to database size; files remain in the archive). DR10 thus displaces DR5.


3. Summary of Data Products

In accord with the criteria for data-inclusion in Section 2, the following products are available in DR10:

For details on data and file formats, associated ancillary products, product construction and usage, see the ZSDS Explanatory Supplement. See Section 10 below for a list of MUST READ ADVISORIES about these products. See Section 11 below for a more detailed list of advisories and known caveats.


4. File-based Product Statistics

The ZTF camera consists of 16 CCDs, each of which is partitioned into 4 readout quadrants. Therefore, 64 CCD-quadrant images are generated per ZTF exposure. A CCD-quadrant is the basic image-unit for pipeline processing and from which all science data products are derived. A CCD-quadrant covers ≈ 0.854° x 0.854° on the sky. Table 1 summarizes the CCD-quadrant-based (single exposure) image counts. The second column shows the number of full camera exposures from which these were generated. The CCD-quadrant image counts are separated into "bad" (probably unusable due to non-photometric conditions) and "good" images. The criteria used to identify "bad" vs "good" images are described in Section 2.4 of the ZSDS Advisories & Cautionary Notes. The relative fractions of bad and good images in this table provide an indicator of the overall fraction of "bad" (likely suspect) single-epoch photometry measurements within individual lightcurves.

Table 1: Number of exposures and "good / bad" CCD-quadrant-based science images in DR10
Filter(s) #Exposures #Good (usable) #Bad (suspect) Total number Bad / Total [%]
g 231,928 13,063,666 1,170,666 14,234,332 8.22
r 332,506 18,107,826 2,299,794 20,407,620 11.27
i 25,820 890,994 679,645 1,570,639 43.27
g + r + i 590,254 32,062,486 4,150,105 36,212,591 11.46

The number of CCD-quadrant-based reference images (co-adds) per filter in DR10 are shown in Table 2. Note: unless brought to our attention, all of these are expected to have good photometric and astrometric quality in general. This is because they were constructed from good-quality single-exposure images satisfying a range of criteria, which are more stringent than those used to tag the "good" images in Table 1. The criteria used to select inputs for generating reference images are described in Section 6.7 of the ZSDS Explanatory Supplement.

The existence of a reference image for a specific survey field, CCD-quadrant, and filter is important since it means that for this region of sky: (i) image differencing and alerts can be triggered, and (ii) objects can be extracted and then used as seeds to generate lightcurves. The third column in Table 2 shows the percentage of ZTF survey fields/CCD-quadrants with at least one exposure (regardless of quality) that have a reference image, for a snapshot taken on 2021 March 23. Note: at least 15 good-quality images are required to generate a reference image. There are fields/CCD-quadrants not visited as often during the survey and therefore lack reference images. The fourth column in Table 2 shows the approximate percentage of P48-accessible sky covered in reference images, as constrained by the area covered by any exposures on the primary survey grid.

Table 2: Number of CCD-quadrant-based Reference Images and coverage in DR10
Filter(s) #Reference Images Visited-sky coverage [%] Accessible-sky coverage [%]
g 67,102 84.68 97.28
r 69,705 85.89 97.23
i 26,536 41.78 62.23
g + r + i 163,343 71.76 85.58

5. Source and Lightcurve Statistics

Table 3 reports the number of sources in all PSF-fit-based and aperture-based catalog file products (summarized in Section 3), separately for single-exposure (sci) images and reference (ref or co-add) images. The PSF-fit and aperture catalogs are treated independently (with no merging or cross-linking of records) in the Data System. This means a large fraction of the same sources are present in both flavors of catalog. The PSF-fit-based catalogs contain more sources since first, PSF-fitting generally yields more accurate photometry at faint fluxes and hence is more sensitive, and second, it includes deblending of closely-separated sources. Deblending is more important in high source-density regions such as the Galactic plane.

Only sources from the PSF-fit-based sci catalogs are positionally-matched across observation epochs to construct lightcurves (see Table 4 for numbers). As mentioned above, sources in the deeper PSF-fit-based ref catalogs are initially used as seeds to trigger source-matching and generate lightcurves.

Table 3: Number of sources in CCD-quadrant-based catalog files in DR10, according to extraction and image type
Filter(s) #PSFcat-sci sources #Aperturecat-sci sources #PSFcat-ref sources #Aperturecat-ref sources
g 142,078,098,290 90,340,912,752 2,376,300,904 739,186,134
r 419,766,895,288 261,422,093,131 3,244,074,284 1,100,504,871
i 37,769,208,576 20,724,174,282 1,223,561,900 390,639,951
g + r + i 599,614,202,154 372,487,180,165 6,843,937,088 2,230,330,956

Table 4 reports the number of lightcurves from positional-matching of sources in the single-exposure PSF-fit-based science catalogs with length (number of independent epochs) greater than some limit. The distribution in the number of observation epochs and corresponding timespans is shown in Figure 3. All observation epochs, regardless of data quality are included in the counts in Table 4. For an estimate of the average fraction of "bad" epochs per lightcurve (with likely suspect and unusable photometry), see Table 1. The criteria used to identify "bad" vs "good" epochal data are described in Section 2.4 of the ZSDS Advisories & Cautionary Notes.

Table 4: Number of lightcurves with number of DR10 observation epochs (Nobs) greater than some limit
Filter(s) #lightcurves with Nobs ≥ 1 #lightcurves with Nobs ≥ 2 #lightcurves with Nobs ≥ 5 #lightcurves with Nobs ≥ 10 #lightcurves with Nobs ≥ 20
g

1,356,061,500

1,117,111,394

933,630,593

816,818,038

694,110,875

r

2,158,697,597

1,848,325,736

1,572,495,172

1,405,725,599

1,234,469,607

i

559,365,948

452,347,284

367,360,293

323,757,711

280,740,497

g + r + i 4,074,125,045 3,417,784,414 2,873,486,058 2,546,301,348 2,209,320,979

6. DR10 Sky Coverage

Figure 1 shows the approximate spatial distribution in the number of single-exposure epochs in each filter included in DR10, as Aitoff-Hammer projections in equatorial coordinates. These are approximate since the spatial resolution (bin size) of each map is 1° while a CCD-quadrant subtends ≈ 0.854° x 0.854° at the center of a map. The small-scale dark regions are not holes in coverage, but due to aliasing from the warped reprojection and resampling of CCD-quadrant centers onto larger scales. The ~ 3 arcmin gaps between the camera's CCDs also contribute to these holes.

FITS-formatted image representations of Figure 1 with Word Coordinate System metadata in their headers can be downloaded for each filter:

These maps can be visualized using a FITS-image viewer to determine if a specific sky-position is covered by a survey-field/CCD-quadrant. If so, the approximate number of overlapping epochs on/near this position (to within ≈ 1°) can be determined from examining the pixel values.

All-sky coverage map for g-band images All-sky coverage map for r-band images All-sky coverage map for i-band images
Figure 1 - Sky coverage and number of observation epochs per approximate CCD-quadrant footprint represented in DR10 in equatorial coordinates centered at ra, dec = 0, 0. Left: g-filter; Center: r-filter; Right: i-filter. Color bar represents the number of observation epochs. Dark regions on small scales are not holes in coverage, but due to the coarse resampling of CCD-quadrant centers on 1° scales. This resampling also distorts the true number of epochs per pointing (see Figure 3a for the true epoch distribution). Click on a panel to enlarge.

 

Figure 2 shows the approximate spatial distribution in CCD-quadrants per filter from which lightcurves were generated for inclusion in DR10. These maps closely track the Reference Image coverage since (PSF-catalog-based) sources extracted therefrom were used as seeds to trigger the lightcurve generation. As in Figure 1, these maps are approximate due to their finite spatial resolution (bin size) of 1° compared to the remapped CCD-quadrants that subtend ≈ 0.854° x 0.854° at the center of a map. The small-scale dark regions are not holes in coverage, but due to aliasing from the warped reprojection and resampling of CCD-quadrant centers onto larger scales. The ~ 3 arcmin gaps between the camera's CCDs also contribute to these holes.

FITS-formatted image representations of Figure 2 with World Coordinate System metadata in their headers can be downloaded for each filter:

These maps can be visualized using a FITS-image viewer to determine if a specific position is covered by a survey field/CCD-quadrant. If so, one can determine if lightcurves (and/or Reference Images) exist on/near this position, to within ≈ 1°.

All-sky coverage map for reference images in g-band All-sky coverage map for reference images in r-band All-sky coverage map for reference images in i-band
Figure 2 - Spatial distribution representing lightcurve (and ≈ Reference Image) coverage in DR10, binned on approximate CCD-quadrant scales, in equatorial coordinates centered at ra, dec = 0, 0. Left: g-filter; Center: r-filter; Right: i-filter. The color bar does not represent the number of observation epochs, but the number of neighboring CCD-quadrants that happen to overlap due to the fixed sky-tiling and coarse resampling of their centers on 1° scales. Click on a panel to enlarge.

7. Lightcurve products: timespans, cadence, and epoch distributions

To complement the epoch-coverage maps in Figure 1 and lightcurve numbers in Table 4, Figure 3 shows the distribution in overall timespan per survey-field (latest minus earliest exposure in DR10) versus the number of observation epochs in each.

Figure 3 - Timespan of observations per survey-field as a function of the number of epochs in each. These timespans are proxies for the overall spans of lightcurves in DR10.

 

Figure 4 encapsulates the available timespans and cadences (separation between consecutive epochs) in DR10, for each filter. These plots can be used as follows: for a given observation or effective lightcurve timespan per survey field (or pointing), one would take a horizontal slice to infer the distribution of visit separations (cadences) therein. The number of times a lightcurve is sampled at that cadence is indicated by the color bar.

Figure 4 - Effective timespan per pointing (per ≈ 7° x 7° survey field) versus visit cadence. See text above for details. Left: g-filter; Center: r-filter; Right: i-filter. Click on a panel to enlarge.

8. File-based Products in Archive

8.a. Single-Exposure Image Products

The calibrated CCD-quadrant-based science images have the following generic root paths and filenames in the archive:

      YYYY/MMDD/fracday/ztf_filefracday_field_filtercode_cccdid_imgtypecode_qqid_sciimg.fits
  where:
      YYYY = year
      MMDD = two-digit month and two-digit day
      fracday = fractional day since UT 0hrs
      filefracday = YYYYMMDDfracday
      field = six-digit survey field ID 
      filtercode = zg, zr, or zi for g, r, i filters respectively
      ccdid = two-digit CCD identifier: 01..16
      imgtypecode = "o" for on-sky data; "f" for dome flat; "b" for bias
      qid = one-digit quadrant ID within specific ccdid: 1..4

An archived "sci" image can be prepended with the following root URL to access and download the specific sciimg.fits file:
https://irsa.ipac.caltech.edu/ibe/data/ztf/products/sci/

For example:
https://irsa.ipac.caltech.edu/ibe/data/ztf/products/sci/2018/0325/251921/ztf_20180325251921_000314_zg_c08_o_q2_sciimg.fits

The path and filename identifiers for science images covering a sky position and/or time range can be obtained using an API query (see Section 12b.i).

Alternatively, the sciimg.fits suffix can be replaced with any of the following ancillary file suffixes, with products summarized in Section 3:

  mskimg.fits - bit-mask image corresponding to sciimg.fits
  psfcat.fits - PSF-fit source catalog for sciimg.fits
  sexcat.fits - aperture-based (SExtractor) photometry catalog for sciimg.fits
  sciimgdao.psf - PSF for sciimg.fits in DAOPhot look-up-table format
  sciimgdaopsfcent.fits - PSF stamp at center of sciimg.fits
  scimrefdiffimg.fits.fz - difference image in fpack'd FITS format
  diffimgpsf.fits - effective PSF stamp for difference-image
  alerts.tar.gz - gzipped tar-directory of all alert packets
  sciimlog.txt - processing log for sciimg.fits (instrumental calibration)
  diffimlog.txt - processing log for difference image and event extraction
  log.txt - overall processing log for realtime pipeline run
8.b. Reference Image Products

Reference images (co-added science images) have the following generic root paths and filenames in the archive

      prefield/fieldfield/filtercode/ccdccdid/qqid/ztf_field_filtercode_cccdid_qqid_refimg.fits
  where:
      prefield = first three digits of survey field ID  
      field = six-digit survey field ID
      filtercode = zg, zr, or zi for g, r, i filters respectively
      ccdid = two-digit CCD identifier: 01..16:
      qid = one-digit quadrant ID within specific ccdid: 1..4

An archived "ref" image can be prepended with the following root URL to access and download the specific refimg.fits file:
https://irsa.ipac.caltech.edu/ibe/data/ztf/products/ref/

For example:
https://irsa.ipac.caltech.edu/ibe/data/ztf/products/ref/000/field000245/zg/ccd02/q1/ztf_000245_zg_c02_q1_refimg.fits

The path and filename identifiers for reference images covering a sky position can be obtained using an API query (see Section 12b.ii).

Alternatively, the refimg.fits suffix can be replaced with any of the following ancillary file suffixes, with products summarized in Section 3:

  refcov.fits - pixel depth-of-coverage image
  refunc.fits - pixel (1σ) uncertainty image
  refpsfcat.fits - PSF-fit source catalog for refimg.fits
  refsexcat.fits - aperture-based (SExtractor) photometry catalog for refimg.fits
  refimlog.txt - processing log for sciimg.fits (instrumental calibration)
  log.txt - overall processing log for realtime pipeline run
8.c. Other File Products

Other relevant root URLs to access file-based products in DR10 are as follows.
The root URL to access raw CCD-based image files is: https://irsa.ipac.caltech.edu/ibe/data/ztf/products/raw/
The root URL to access calibration (cal) related files is: https://irsa.ipac.caltech.edu/ibe/data/ztf/products/cal/

For a description of all path and filename identifiers, see Section 7 of the ZSDS Explanatory Supplement.


9. Epoch-based Data Quality Flags and Filtering

The ZTF Observing System is fully robotic. One limitation is that it cannot determine in advance when conditions are non-photometric, for example, when intermittent clouds partially (or fully) cover the field-of-view during an exposure. Scattered moonlight can also wreak havoc. This severely affects the accuracy of the derived photometric calibration solutions. There is one estimate of the photometric zeropoint (magzp) with accompanying color term (clrcoeff) per CCD-quadrant, where each of the 64 quadrants is calibrated independently in processing. Spatial variations in transparency at the intra-quadrant level will have led to unusable science products. The lightcurve photometry extracted from the single-exposure image products will also be suspect.

There are two flavors of science products that can be queried from the archive for which data quality flags are available for filtering: (a) single-exposure CCD-quadrant-based files; and (b) Lightcurves. The quality filters for each are as follows:

9.a. Quality Filtering of Single-Exposure File Products

Bad-quality or generally unusable CCD-quadrant-based images from individual epochs (including accompanying source catalog files) can be omitted when querying the archive by thresholding the INFOBITS value in the archive metadata. If INFOBITS for an image has value < 33554432 (i.e., does not contain bit 25), the image and catalog data are probably usable.

The metrics and criteria used to set this "bad" data quality flag are described in Section 2.4 of the ZSDS Advisories & Cautionary Notes.

Examples of image-based queries that include filtering on INFOBITS are given in Section 12b.i below.

9.b. Quality Filtering of Epochs in Lightcurves

Analogous to the flagging of “bad-quality” images (Section 9a), bad or generally unusable observation epochs in lightcurves can be omitted by thresholding the catflags column in the lightcurve metadata. If catflags for an image has value < 32768 (i.e., does not contain bit 15), the photometry at that epoch is probably usable.

This flagging removes epochs based on their overall image/calibration quality. catflags also encodes possible issues at the source level, for example, contamination by bad pixels. These bits are defined in Section 10.6 of the ZSDS Explanatory Supplement. If you demand perfectly clean extractions at every epoch, we advise specifying catflags = 0 when querying lightcurve epochs.

Examples of lightcurve queries that include filtering on catflags are given in Section 12b.iii below.


10. What you should know about the ZTF Release Products

The following are high-level advisories. Visit the link in Section 11 for a more detailed list.

  1. DR10 contains products from both the Public and Private Surveys, derived from individual exposures covering the observation spans summarized in Section 2. All i-filter data was acquired under private survey time. All exposures in the public survey have an integration time of 30 seconds. Exposures in the private surveys have a mixture of integration times: 30, 60, 90, 120, 240, and 300 seconds, with the most common being 30 seconds.
  2. The archive contains products with basic automated data-quality assurance applied during processing. The primary criterion for archiving a science image (along with its derived products) is that it must have passed astrometric and photometric calibration using thresholds on metrics tuned to yield potentially usable science products. "Bad" (unusable) data in the archive is still possible. This is data affected by clouds, low atmospheric transparency and spatial variations therein, moon contamination, bright source scattering artifacts, unmasked aircraft and artificial satellites, transient detector behavior including cross-talk, bad seeing or focus, and/or inaccurate telescope tracking. The archive metadata contains flags at both the image-level and epoch-level within individual lightcurves to identify and omit likely unusable data from queries (see Section 9).
  3. Calibrated ZTF photometry can show biases of up to 0.025 mag for predominately bright sources with g, r, i magnitudes <~ 15.5 relative to source photometry in the Pan-STARRS1 (PS1) MeanObject DR1 Catalog. However, see note 17 below. These biases are present in both the PSF-fit-derived and aperture-derived catalogs. These biases are field, CCD-quadrant, and magnitude dependent. The lightcurve photometry (when retrieved through either the GUI, API, or tar balls) includes corrections for various biases, resulting in final photometric accuracies (with respect to PS1) of typically <~ 0.01 mag.
  4. There is no separate searchable database storing source measurements from the single-exposure catalog files. These measurements can only be accessed by downloading the catalog files, which are in FITS-binary table format. Each table file corresponds to a single CCD-quadrant.
  5. Single-exposure source photometry can be accessed at the source-level indirectly by retrieving entire lightcurves through either the GUI, API, or tar balls. These products also have various photometric corrections applied (note 3 above), while the single-exposure catalog files do not include these refinements.
  6. Large spatial queries using the GUI-driven lightcurve service may take longer than expected depending on user-load. An alternative for retrieving lightcurves is to submit source positions to the lightcurve API. We recommend first querying the Objects Table to ensure your positions return data (however see notes 9, 10, and 11 below).
  7. Lightcurves were constructed using exclusively measurements from the calibrated single-exposure PSF-fit-derived catalogs, not the aperture-based photometry catalogs.
  8. The photometry catalogs, including lightcurve products, contain measurements of both point-like and extended objects. The PSF-fit photometry measurements are optimal for point-sources only. Concentric aperture photometry measurements are more optimal for extended objects.
  9. Lightcurves were constructed using seed detections from the deeper (co-add) Reference Image PSF-fit catalogs. These "seed sources" were loaded into the searchable Objects Table where their photometry is ~ 2 to 2.5 magnitudes deeper than the single-exposure catalogs. Therefore, a majority of faint sources in the Objects Table (fainter than the single-exposure sensitivity limits) will not have lightcurves. Futhermore, objects with photometry hovering near the single-exposure detection limits will have incomplete temporal-coverage and therefore more sparsely sampled lightcurves.
  10. Reference Images and their catalogs (loaded into the Objects Table) were generated by stacking single-exposures acquired from all science programs (public, private, and Caltech-time observations). These programs covered two overlapping tiling grids on the sky, a primary grid and a secondary grid, to mitigate gaps between CCDs on the focal plane. The single-exposures included in DR10 however (from which lightcurves were constructed), do not cover all survey fields from both grids. Therefore, expect to see missing lightcurves for many sources in the Object Table.
  11. When querying the Objects Table to retrieve lightcurves, the nobsrel (and ngoodobsrel) metrics report the total (and total unflagged) number of lightcurve epochs included in DR10. The total number of epochs, including unreleased epochs from all science programs (public, private, and Caltech-time observations) are indicated by the nobs and ngoodobs metrics. The latter will usually exceed nobsrel and ngoodobsrel respectively. All lightcurve-collapsed metrics in the Objects Table are computed using all (including unreleased) epochal data from all science programs.
  12. The lightcurve photometry measurements (mag values) already have their epoch-dependent photometric zeropoints applied (magzp column in returned metadata). All photometry is in the native ZTF photometric system. No color corrections are applied (or rather, zero color in the AB system is assumed). To transform photometry onto the AB (PS1) system, you will require knowledge of the source color in that system. The exact color used during initial photometric calibration against PS1 is defined by column pcolor and the color coefficient to transform to the PS1 system is given by column clrcoeff. The procedure for applying color corrections is described in Section 10.1.1 of the ZSDS Explanatory Supplement.
  13. In DR10, a difference image (file suffix *scimrefdiffimg.fits.fz; see Section 8a) will only accompany a single-exposure science image in the archive if a Reference Image corresponding to the same field/CCD-quadrant/filter existed on/before the single-exposure image was processed. Reference Image generation continued throughout the survey and image differencing could only be triggered if a Reference Image was present. Therefore, missing difference image products are more prevalent during the early phase of the survey.
  14. The units of the pixel values in all image products are in Digital Numbers (DN). Methods on how to apply the photometric calibration solutions to your own measurements using these products are outlined in Section 10 of the ZSDS Explanatory Supplement.
  15. When querying lightcurves using the GUI, the mjd timestamps attached to the lightcurve measurements pertain to the start of each exposure while hjd and hmjd pertain to the middle, i.e., with 0.5*EXPTIME added to the start observation times, where EXPTIME represents the integration time (FITS header keyword) of the corresponding exposure. The date/timestamps attached to all file-based products however (i.e., values for the FITS header keywords: OBSJD, OBSMJD, SHUTOPEN, DATE-OBS) pertain to the start of the exposure, not the middle.
  16. Varying atmospheric transparency on the scale of an individual CCD-quadrant can have a significant impact on the quality of the i-filter fringe corrections. Fringe corrections were calibrated from high throughput images acquired under photometric conditions. Therefore, the presence of intermittent cloud cover, depending on its spatial variation, can lead to erroneous corrections and unusable images in the i-filter.
  17. Be warned that calibrated photometry extracted from image data acquired with CCD #06 after October 23, 2019 may show systematic residuals with respect to PS1 of up to 10%. This was traced to a camera tuning update. Future reprocessing will attempt to correct the biased photometry.
  18. The lightcurve photometry measurements (mag values) are sometimes (but seldomly) accompanined by negative error estimates (magerr). These negative errors are meaningless and imply that PSF-fit solutions did not converge. We advise treating all measurements with negative errors as bogus in analyses.

11. Extended (detailed) Advisories and Cautionary Notes

A more detailed compilation of advisories and known caveats specific to each data product are given in:
ZSDS Advisories & Cautionary Notes

These notes are progressively updated as issues are resolved, new ones become known, or as we learn of new tips or software that could be of use to the community.


12. How to Access ZTF Data Products

Access to all ZTF data products is through the online (GUI-based) web-tools and API services of the NASA/IPAC Infrared Science Archive (IRSA) linked from: https://irsa.ipac.caltech.edu/Missions/ztf.html.

Below we provide links to specific GUIs and documentation describing how to access the archive products. Examples using the API services are also given.

12.a. Product-specific Access GUIs and Recipes

 

Workflow for retrieving and analyzing lightcurves using GUI services

 

Figure 5 - Schematic of the workflow for retrieving and analyzing lightcurves using the GUI services (see Section 12a). Click to enlarge.

 

12.b. Example Queries using the APIs

Application Programming Interfaces (APIs) accompany most of the GUI data-retrieval services. These APIs can be executed from within your own software to enable repetitive and/or bulk data downloads.

An overview for retrieving file-based products using APIs is given on the Image/Catalog API Page. API queries can include thresholding on any of the available archive image metadata parameters.

Examples of using APIs to retrieve file-based products and lightcurves are given below.

i. Querying Single-Exposure Science Image Products using the API

To retrieve file-based single-exposure science-image products that touch a fixed R.A.,Dec position, fall within an observation JD range, and are likely to be of "good" (usable) quality with INFOBITS < 33554432 (see Section 9a), first query their metadata using the wget utility, e.g.,

wget "https://irsa.ipac.caltech.edu/ibe/search/ztf/products/sci?
      POS=255.9302,11.8654&WHERE=obsjd>2458219.9678+AND+obsjd<2458228.8155+
      AND+infobits<33554432" -O out.tbl

where all inputs reside on one line. Then use the output metadata table (out.tbl) contents to construct the image URL-paths/filenames. Using the metadata column names, a science-image will have the generic URL-path/filename:

https://irsa.ipac.caltech.edu/ibe/data/ztf/products/sci/
      YYYY/MMDD/fracday/ztf_filefracday_000field_filtercode_
      cccdid_imgtypecode_qqid_sciimg.fits

where identifiers in bold green are the values of actual columns in the out.tbl table, and YYYY (year), MM (month), DD (day), and fracday (fractional time of day) can be extracted from filefracday. These images can then be retrieved using the wget utility. For the example above, the first metadata record in out.tbl would be retrieved using:

wget https://irsa.ipac.caltech.edu/ibe/data/ztf/products/sci/
     2018/0411/467847/ztf_20180411467847_000535_zr_c11_o_q3_sciimg.fits

Any of the accompanying products can be retrieved by replacing sciimg.fits with another filename suffix (see Section 8a). Instead of retrieving the entire image, you can download a square cutout centered at a specific R.A.,Dec and specific size, and save the output to a file (optionally gzipped). For the example above:

wget "https://irsa.ipac.caltech.edu/ibe/data/ztf/products/sci/
      2018/0411/467847/ztf_20180411467847_000535_zr_c11_o_q3_sciimg.fits?
      center=255.8535,12.0503&size=60arcsec&gzip=false" -O cutout.fits

If the API fails to return a file (or its metadata), you can try navigating the archive tree directly to search for the specific product: https://irsa.ipac.caltech.edu/ibe/data/ztf/products/sci/. For guidance, listings of the DR10 single-exposure Science products are given in Section 8a. A description of the path and filename identifiers is given in Section 7 of the ZSDS Explanatory Supplement.

ii. Querying Reference Image Products using the API

To retrieve file-based Reference Image products that touch a fixed R.A.,Dec position in only the r-filter (fid=2), first query their metadata using the wget utility, e.g.,

wget "https://irsa.ipac.caltech.edu/ibe/search/ztf/products/ref?
      POS=358,25.6&WHERE=fid=2" -O out.tbl

where all inputs reside on one line. Then use the output metadata table (out.tbl) contents to construct the image URL-paths/filenames. Using the metadata column names, a Reference Image will have the generic URL-path/filename:

https://irsa.ipac.caltech.edu/ibe/data/ztf/products/ref/
      prefield/fieldfield/filtercode/ccdccdid/qqid/
      ztf_field_filtercode_cccdid_qqid_refimg.fits

where identifiers in bold green are the values of actual columns in the out.tbl table, and prefield is the first three (left-zero-padded) digits of the six-digit field value. These images can then be retrieved using the wget utility. For the example above, the first metadata record in out.tbl would be retrieved using:

wget https://irsa.ipac.caltech.edu/ibe/data/ztf/products/ref/
     001/field001596/zr/ccd16/q2/ztf_001596_zr_c16_q2_refimg.fits

Any of the accompanying products can be retrieved by replacing refimg.fits with another filename suffix (see Section 8b). Instead of retrieving the entire image, you can download a square cutout centered at a specific R.A.,Dec and specific size, and save the output to a file (optionally gzipped). For the example above:

wget "https://irsa.ipac.caltech.edu/ibe/data/ztf/products/ref/
      001/field001596/zr/ccd16/q2/ztf_001596_zr_c16_q2_refimg.fits?
      center=357.46871,26.00549&size=100arcsec&gzip=false" -O cutout.fits

If the API fails to return a file (or its metadata), you can try navigating the archive tree directly to search for the specific product: https://irsa.ipac.caltech.edu/ibe/data/ztf/products/ref/. For guidance, listings of the DR10 Reference Image products are given in Section 8b. A description of the path and filename identifiers is given in Section 7 of the ZSDS Explanatory Supplement.

iii. Querying Lightcurves using the API

The API functionality corresponding to the GUI-driven recipe for retrieving lightcurves is described on the Lightcurve Programming Interface Page. Numerous examples are given on this page. Below we provide a more generic example that includes most parameters of interest. You will always want to exclude observation epochs associated with bad/unusable data by masking catflags values equal to 32768, -- the "cloud-affected and/or moon-contamination" flag (decimal bit 15); see Section 9b. Alternatively, you can exclude epochs with any non-zero bit by using the catflags mask value 65535 in the query below.

The following query returns a concatenated list of lightcurves for objects falling within a 10 arcsec (~0.0028°) radius from position R.A.,Dec = 255.9302°, 11.8654° in the r-filter, where each has ≥ 3 epochs that fall within the MJD range 58194.0...58483.0 and all consist of only "good" (likely usable) measurements. Lightcurves are stored in the output ASCII table: out.tbl. The concatenated lightcurves are distinguished by their ObjectIDs therein (column oid).

wget "https://irsa.ipac.caltech.edu/cgi-bin/ZTF/nph_light_curves?
      POS=CIRCLE+255.9302+11.8654+0.0028&BANDNAME=r&NOBS_MIN=3&
      TIME=58194.0+58483.0&BAD_CATFLAGS_MASK=32768&FORMAT=ipac_table" -O out.tbl

All inputs reside on one line. A lightcurve (with common oid values) is represented by columns mag ± magerr versus mjd or hjd. The out.tbl also contains metadata to reconstruct the single-exposure science image path/file names that you can later download using the wget utility, or alternatively, cutouts on your position of interest (see Section 12b.i).

You can also upload the out.tbl file from above to the Time Series Tool to enable visualization, analysis, and period finding, analogous to the GUI-driven method described in Section 12a. Ensure you select "ztf" from the menu before uploading. Following your upload, ensure only one ObjectID (oid) is selected for visualizing, using the filter ( ) icon in the Time Series Tool.

iv. Solar System Object Precovery using the API

The API equivalent of the Moving Object Search Tool (GUI) is described on the MOST Programming Interface Page. For example, to retrieve all single-exposure science images containing Asteroid 438973 Masci, first query their metadata using the wget utility:

wget "https://irsa.ipac.caltech.edu/cgi-bin/MOST/nph-most?
      catalog=ztf&input_type=name_input&obj_name=Masci&
      obs_begin=2018+03+17&obs_end=2018+12+31&output_mode=Brief" -O out.tbl

where all inputs reside on one line. Then use the output metadata table (out.tbl) contents to construct the science image URL-paths/filenames (see Section 12b.i above). These images can then be downloaded using the wget utility. Alternatively, cutouts centered on the object's R.A.,Dec in each image, as listed in out.tbl, can be downloaded.

12.c. Bulk Download of all Lightcurves

The lightcurves included in this release are available in Apache Parquet format. A selection of basic (most necessary) metadata accompany each lightcurve, primarily IDs to enable the retrieval of associated metadata from the archive, for example, images or Object Table metrics using the access methods above.

The following URL contains these bulk-downloadable products, totaling 4.8TB:
https://irsa.ipac.caltech.edu/data/ZTF/lc_dr10/
You can click on individual files to download them, or use the wget utility to automate the downloads. For example, wget -r -np -nH -R "index.html*" https://irsa.ipac.caltech.edu/data/ZTF/lc_dr10/0/field0697 will download all lightcurves for Field 697. A list of corresponding MD5 checksums is also provided for those who want to verify file contents following a download. On a Unix/Linux OS, the files can be verified using md5sum -c checksum.md5.

The files are organized by survey-Field ID subdirectories where each Field spans ≈ 7° x 7°. The DR10 lightcurves are spread across 1,134 fields, according to this table of central coordinates.

Each file corresponds to one field/chip/quadrant/filter combination for a total of 162,036 Parquet files. The files are readable with libraries including Pandas, pyarrow, or Dask. For example, in a Python session on a Unix/Linux system, a Dask dataframe can be constructed from all the files with

import dask.dataframe as dd
ddf = dd.read_parquet('[01]/field*/*parquet', engine='pyarrow')

With a recent version of pyarrow, a table can be constructed for one field and converted to a Pandas dataframe with

import pyarrow.parquet as pq
df = pq.read_table('0/field0697').to_pandas()

The columns in each Parquet file are as follows:

  objectid: Unique Object Identifier. This is the oid that is returned by
            queries on the Objects Database Table. This ID can be used to
            retrieve additional metadata for an Object or lightcurve.

  filterid: Filter for the lightcurve photometry: 1 => g; 2 => r; 3 => i.
  fieldid: The survey Field Identifier. Lightcurves in DR10 are spread
           across 1126 Fields. This identifier can be used to retrieve additional
           archive metadata.
  rcid: The readout-channel Identifier (or CCD-Quadrant index) that contains the
        object. The rcid runs from 0 to 63 over the ZTF focal plane.
        In combination with fieldid, this can be used to retrieve additional
        archive metadata.
  objra: Object Right Ascension from the input Reference Image (seed) catalog
         corresponding to the fieldID, rcid, and filterid [decimal degrees].
  objdec: Object Declination from the input Reference Image (seed) catalog
          corresponding to the fieldid, rcid, and filterid [decimal degrees].
  nepochs: The number of observation epochs in the lightcurve for this object, included in this data release only.
  ----------- multidimensional columns 
  hmjd: Heliocentric-based Modified Julian Date [days] corresponding to the
        middle of each exposure and approximate position of the object.
        The HMJD is derived from "HJD - 2400000.5" where HJD is the Heliocentric
        Julian Date. HMJD is double precision truncated to 5 decimal places in order to resolve high-cadence observations. 
  mag: Calibrated magnitude for a source with color g - r = 0 in the AB photometric
       system. See Cautionary Note #12 in Section 10. 
  magerr: Corresponding 1-σ uncertainty in mag estimate; excludes the (usually small)
          uncertainty associated with the calibration zeropoint and possible systematics
          relative to external photometry catalogs (for example PS1).
  clrcoeff: Linear color coefficient term from photometric calibration
            that can be applied to the mag estimate if you know the
            AB g - r color of your source. See Cautionary Note #12 in Section 10.
  catflags: Photometric/image quality flags encoded as bits (Section 9b).
            In particular, you will always want to exclude observation epochs
            affected by clouds and/or the moon. These epochs have catflags = 32768
            (decimal bit 15).

Users of the zort package with the ascii-format lightcurves released as part of DR1 to DR4 can use this Python script to convert a directory of Parquet files to that format.


13. Help Desk

For any issues related to data-access, GUI tools, APIs, data-quality, formats, processing algorithms or product usage, please email:
irsasupport@ipac.caltech.edu


14. Further Documentation & Relevant Publications

A detailed description of the processing pipelines, data products (including alert packets), instructions for product retrieval, filename identifiers, data formats, and product usage can be found in:

The ZSDS Explanatory Supplement: Pipelines, Definitions, Data Products & Access

Two relevant publications from the special ZTF Focus Issue of PASP: