J-VAR: Javalambre Variability Survey

• J-VAR is a time-domain survey that uses seven strategically chosen optical filters to capture variability tied to specific stellar spectral features.
• The survey employs an observing strategy of 11 epochs (up to ~33 independent samples) with a 40-minute basic block to ensure accurate period recovery and amplitude measurement.
• DR1 provides 1.3 million point-source light curves over 202 deg², supporting studies of variable stars, optical transients, and minor bodies with robust photometric and astrometric calibration.

The Javalambre Variability Survey (J-VAR) is a multi-band time-domain project carried out with the Javalambre Auxiliary Survey Telescope (JAST80) and its wide-field camera T80Cam at the Observatorio Astrofísico de Javalambre. Its defining characteristic is the use of seven optical filters—three broad-bands and four medium/narrow bands strategically placed on stellar spectral features—within a standardized observing sequence designed for variability studies. The first data release (DR1) introduced time-series photometry over approximately 202 deg², with 1.3 million light curves of point sources, detections of optical transients, and a substantial Solar-System component, while subsequent analysis of 315 Gaia DR3 RR Lyrae established the survey’s utility for period recovery and wavelength-dependent pulsation studies (Ederoclite et al., 29 Aug 2025, Kulkarni et al., 3 Sep 2025).

1. Survey configuration and instrumental basis

J-VAR is conducted on the 83 cm Javalambre Auxiliary Survey Telescope, described as an $f/4.5$ modified Ritchey–Chrétien reflector on a German equatorial mount, and uses T80Cam as its focal-plane instrument. T80Cam is described as a 16-amplifier CCD in the DR1 survey paper and as a 9.2 k $\times$ 9.2 k CCD camera in the RR Lyrae analysis; both descriptions refer to the same wide-field imager delivering a 2 deg² field of view at approximately $0.55$ arcsec pixel$^{-1}$ (Ederoclite et al., 29 Aug 2025).

The field of view is given as $1.4^\circ \times 1.4^\circ$ (2 deg²), and the system yields $\sim 10^4$ detectable sources per exposure. The survey footprint is in the northern sky and, for DR1, spans Galactic latitudes $|b|>20^\circ$ to optimize both variable-star and Solar-System science. This operational choice places J-VAR between classical stellar variability programs and broader synoptic surveys: it is wide-field enough for population studies, but its filter system is explicitly tuned to astrophysical diagnostics beyond continuum monitoring.

A common simplification is to treat J-VAR as merely another optical time-domain survey. That is incomplete. Its core distinction is not sky coverage alone, but the combination of wide-field cadence with filters placed on Ca II H+K, Mg $b$, H$\alpha$, and the Ca II triplet, enabling variability measurements tied directly to line-sensitive passbands rather than only broad-band flux changes.

2. Filter system and cadence design

J-VAR employs seven filters drawn from the J-PLUS photometric system. The filter set combines broad-band continuum sampling with targeted coverage of key stellar features.

Filter	Central wavelength / FWHM	Spectral feature or role
J0395	$\lambda_0 = 395$ nm, $\times$0 nm	Ca II H+K
g	$\times$1 nm, $\times$2 nm	Broad-band continuum
J0515	$\times$3 nm, $\times$4 nm	Mg $\times$5 triplet
r	$\times$6 nm, $\times$7 nm	Broad-band continuum
J0660	$\times$8 nm, $\times$9 nm	H$0.55$0
i	$0.55$1 nm, $0.55$2 nm	Broad-band continuum
J0861	$0.55$3 nm, $0.55$4 nm	Ca II triplet

The observing strategy is organized around a Basic Observing Block (BBO) consisting of three full cycles through all seven filters, with orthogonal dithers between cycles. A BBO lasts approximately 40 min, and the median interval between consecutive exposures in the same filter is approximately 12.7 min. Each field is observed in 11 epochs, with a temporal baseline of at least 1 year to maximize transient phase coverage. When enabled, a High-Frequency mode monitors a single field for approximately 3 h while preserving the standard 7-filter sequence (Ederoclite et al., 29 Aug 2025).

The RR Lyrae study describes each field as observed on $0.55$5 independent epochs and DR1 light curves as having $0.55$6 epochs over the first season (Kulkarni et al., 3 Sep 2025). This suggests that the three full cycles within each BBO are treated as independent time samples in that analysis. The distinction is operationally important: “11 epochs” describes the survey visit structure, whereas “$0.55$7 independent epochs” describes the effective number of photometric samples available for period analysis.

The cadence was designed to support robust period recovery. The DR1 survey paper states that 11 epochs ensure approximately 75% success for RR Lyrae with $0.55$8 in one band, while the RR Lyrae analysis finds agreement with Gaia DR3 periods at the level of approximately 80% for RRab and approximately 90% for RRc at $0.55$9 mag under the criterion $^{-1}$0 (Ederoclite et al., 29 Aug 2025, Kulkarni et al., 3 Sep 2025).

3. Reduction, calibration, and light-curve construction

All images are processed with the jype pipeline (v3.1.9), which handles the T80Cam 16-amplifier readout and applies standard preprocessing steps for wide-field photometry. The reduction sequence includes overscan and bias removal using prescan/overscan regions per amplifier, master flat-field correction and illumination correction to remove large-scale 2D biases at the level of tens of mmag, fringing correction where required, and masking of cosmic rays and satellite trails (Ederoclite et al., 29 Aug 2025).

Astrometric calibration is solved per image to $^{-1}$1 arcsec accuracy using standard reference catalogs. Instrumental magnitudes are measured in a 6″ diameter aperture, and an aperture-correction vector is computed per image. Zero-points are transferred from contemporaneous J-PLUS DR3 calibration via overlap of reference stars. The final calibrated magnitude is

$^{-1}$2

where $^{-1}$3 is the differential curve from ensemble photometry (Ederoclite et al., 29 Aug 2025).

Light-curve extraction uses ensemble differential photometry. For each target, at least 15 comparison stars are selected within a radius of at most 9′. The signal-to-noise ratio is computed as

$^{-1}$4

where $^{-1}$5 is source flux (e$^{-1}$6), $^{-1}$7 is aperture area, $^{-1}$8 is sky noise (e$^{-1}$9), and $1.4^\circ \times 1.4^\circ$0 is read noise. Photometric precision is reported as about 2% at $1.4^\circ \times 1.4^\circ$1 and about 5% at $1.4^\circ \times 1.4^\circ$2 in $1.4^\circ \times 1.4^\circ$3, $1.4^\circ \times 1.4^\circ$4, and $1.4^\circ \times 1.4^\circ$5, and flags from SExtractor and custom masks are retained for quality filtering (Ederoclite et al., 29 Aug 2025).

The RR Lyrae analysis adds two further points about DR1 photometry. First, observations were taken on non-photometric nights, with photometric zero-points tied to the single-epoch J-PLUS calibration. Second, the public archive contains multi-band light curves for 1,335,279 objects classified as stars or QSOs by BANNJOS, with photometry reported in the AB system and typical per-point uncertainties $1.4^\circ \times 1.4^\circ$6 mag for $1.4^\circ \times 1.4^\circ$7 (Kulkarni et al., 3 Sep 2025).

4. First data release: coverage, depth, and catalog content

DR1 was publicly released on 19 July 2024. It covers 101 pointings, corresponding to approximately 202 deg², with at least 11 epochs in each of the seven filters. The release contains 1.3 million point-source light curves, 131,900 individual detections of Solar-System objects corresponding to 6,570 unique asteroids, and 10 optical transients detected via image subtraction. Among the transient detections are 4 submitted to TNS, including SN 2020admb (Ia, $1.4^\circ \times 1.4^\circ$8) and SN 2024slh (Ia, $1.4^\circ \times 1.4^\circ$9) (Ederoclite et al., 29 Aug 2025).

The median 5$\sim 10^4$0 limiting magnitude in a 6″ aperture for individual exposures is reported as $\sim 10^4$1 mag in J0395 and J0515, $\sim 10^4$2 mag in J0660 and J0861, $\sim 10^4$3 mag in $\sim 10^4$4, $\sim 10^4$5 mag in $\sim 10^4$6, and $\sim 10^4$7 mag in $\sim 10^4$8. The limiting-magnitude relation is

$\sim 10^4$9

Typical image quality is characterized by a median point-source FWHM of approximately 1.5″ in the $|b|>20^\circ$0 band, with a range of about 1.0″–2.5″ across filters (Ederoclite et al., 29 Aug 2025).

The survey paper summarizes the data density as approximately 5000 light curves per square degree of 11 epochs in 7 bands, while the RR Lyrae analysis emphasizes $|b|>20^\circ$1 epoch sampling for that specific use case (Ederoclite et al., 29 Aug 2025, Kulkarni et al., 3 Sep 2025). A plausible implication is that J-VAR’s archive supports multiple granularities of time-domain analysis: visit-level interpretations for survey characterization and exposure-level interpretations for period fitting.

It is also important to distinguish J-VAR from a purely stellar catalog. DR1 explicitly includes variable stars, optical transients, and minor bodies. The release is therefore a mixed time-domain resource rather than a single-purpose variability list.

5. Scientific domains enabled by DR1

For stellar variability, DR1 demonstrates sensitivity to pulsators, eclipsing binaries, rotational variables, and long-period sources. The survey paper lists RR Lyrae, $|b|>20^\circ$2 Scuti, and Cepheids among pulsators; EA, EB, and EW systems among eclipsing binaries; RS CVn, BY Dra, and solar-like rotators among rotational variables; and Mira and SR stars among long-period and semi-regular variables. Long-period and semi-regular sources are monitored over more than 1 yr, and example folded light curves achieve sub-0.02 mag RMS at $|b|>20^\circ$3, enabling precise period and amplitude measurements (Ederoclite et al., 29 Aug 2025).

For optical transients, differential imaging in seven bands yields early SED evolution. The survey paper cites SN 2020amv as an example showing H$|b|>20^\circ$4 emission in J0660, and states that the multi-filter light-curve sampling enables classification of type II versus Ia without spectroscopy while constraining explosion epochs (Ederoclite et al., 29 Aug 2025). This is a distinctive capability of the filter set: line-sensitive temporal photometry can provide information that broad-band cadence alone would not isolate as directly.

For Solar-System science, the survey recovers approximately 400 detections per epoch near the ecliptic, corresponding to approximately 30 objects, decreasing to fewer than 20 at high Galactic latitudes. Multi-filter colors such as $|b|>20^\circ$5 versus $|b|>20^\circ$6 and $|b|>20^\circ$7 versus $|b|>20^\circ$8 separate asteroid taxonomic classes, and a value-added catalog of magnitudes, observation times, and orbits supports rotational-light-curve and surface-composition studies (Ederoclite et al., 29 Aug 2025).

The survey’s legacy role is framed as complementary to J-PLUS static-sky products, including photometric redshifts and SEDs, and to other time-domain surveys such as ZTF and LSST by adding narrow-band cadence. Future releases are stated to expand the area at approximately 58 deg² yr$|b|>20^\circ$9 toward approximately 1000 deg² by 2035 (Ederoclite et al., 29 Aug 2025). This suggests an evolving infrastructure in which J-VAR functions not only as a source of variability detections but also as a narrow-band temporal layer for multi-survey synthesis.

6. RR Lyrae analysis and the amplitude–wavelength relation

A dedicated analysis of J-VAR DR1 cross-matched the catalog to the Gaia DR3 RR Lyrae table via the shared gaia_sid identifier and identified 315 RR Lyrae among the 1.3 million J-VAR sources. The subtype breakdown is 224 RRab, 82 RRc, and 2 RRd; 7 stars lacked reliable Gaia periods and were left unclassified. The light curves were modeled with the Sesar et al. (2010) SDSS $b$0 multiband RR Lyrae templates using gatspy’s RRLyraeTemplateModeler, with band mapping J0395$b$1, $b$2, J0515$b$3, $b$4, J0660$b$5, $b$6, and J0861$b$7. Fits required at least 6 phase points per band with uniform phase coverage $b$8 (Kulkarni et al., 3 Sep 2025).

The model

$b$9

is fit to observed magnitudes $\alpha$0 by minimizing

$\alpha$1

Periods $\alpha$2 are taken from the best-fitting template family and compared to Gaia DR3 periods $\alpha$3 through

$\alpha$4

With $\alpha$5 adopted as agreement, the analysis finds success rates of approximately 80% for RRab and approximately 90% for RRc at $\alpha$6 mag, and reports strong agreement between J-VAR and Gaia DR3 periods (Kulkarni et al., 3 Sep 2025).

The study then defines the pulsation amplitude in filter $\alpha$7 as half the peak-to-peak magnitude difference of the best-fitting template and introduces the normalized amplitude

$\alpha$8

so that $\alpha$9. The Bailey diagrams in all seven filters show the expected separation of RRab and RRc sequences, and the amplitudes decrease systematically with increasing wavelength. Median normalized amplitudes are reported as $\lambda_0 = 395$0 for J0395, $\lambda_0 = 395$1 for $\lambda_0 = 395$2, $\lambda_0 = 395$3 for J0515, $\lambda_0 = 395$4 for $\lambda_0 = 395$5, $\lambda_0 = 395$6 for J0660, $\lambda_0 = 395$7 for $\lambda_0 = 395$8, and $\lambda_0 = 395$9 for J0861 (Kulkarni et al., 3 Sep 2025).

An exponential fit is given as

$\times$00

with $\times$01 in nm and $\times$02 nm, equivalently

$\times$03

The RR Lyrae analysis states that the bluer J-VAR filter centered at 395 nm has twice the amplitude of the reddest J-VAR passband at 861 nm, and that Gaia and ZTF normalized amplitudes lie within $\times$04 of the J-VAR exponential curve. The paper further interprets the monotonic amplitude decline with wavelength as following from reduced temperature sensitivity of the Planck function at longer wavelengths and identifies the seven-band trend from 395 to 861 nm as an empirical constraint on atmospheric temperature variations, shock-wave strengths, and line-formation physics in RR Lyrae envelopes (Kulkarni et al., 3 Sep 2025).

In methodological terms, one notable result is that SDSS templates derived from broad bands also provide a proper description for the medium and narrow band light curves. This does not remove the need for dedicated template development; rather, the paper argues that next-generation template libraries should incorporate these filters explicitly to refine period–amplitude–phase relations across detailed spectral features (Kulkarni et al., 3 Sep 2025).