Performance
What one 7DT exposure delivers: which bands it can be taken in, how sharp it is, how well it is calibrated, and how deep it goes — the five numbers an observing plan starts from.
Filter transmission
Every unit carries a nine-slot filter wheel. Three slots in each wheel hold Sloan g, r and i; one unit adds u and three units add z. The remaining slots hold medium-band filters, distributed across the array so that the full set is covered in a small number of exposures. The original twenty medium bands are spaced regularly at 25 nm from 400 to 875 nm with 25 nm FWHM. Fifteen more, procured from Edmund Optics and installed in late 2025, fill the gaps between them with central wavelengths from 412 to 832 nm and bandwidths of 14 to 41 nm. The current suite of 35 filters covers 375 to 875 nm, advancing toward the designed complement of 40 medium bands at 12.5 nm spacing.
The additional fifteen filters depart from the regularity of the original set: their central wavelengths are not precisely aligned to the 12.5 nm grid, their bandwidths vary, and no filter between 700 and 800 nm is included in the second batch. These departures reflect availability and will be addressed as the remaining five filters become available. Their spectrophotometric calibration is in preparation; the original twenty remain the calibrated set in operational use.
The same response is shipped as reference data with supy and is what its simulator module computes with, so a response derived there matches this figure. Which bands exist on a particular tile is reported on the data access page, and how much of the sky each band has reached is on the status page, which draws the same curves.
Point-spread function
Across the sixteen operational units the point-spread function measured at field center on good nights ranges from 1.4 to 2.2 arcseconds FWHM, with an array median of 2.0 arcseconds closely tracking the median site seeing. Unit-to-unit scatter in delivered FWHM is 0.2 arcseconds, and the PSF grows by 0.3 arcseconds from field center to corner while ellipticity stays below 0.1 over the central 80 percent of the field. Delivered image quality is therefore consistent across the array. Median delivered FWHM has held stable to within 0.3 arcseconds since routine survey operations began in July 2024.
Measured from the point-spread function across the field of view by the Py7DT pipeline as part of routine astrometric and photometric processing, so the figures are delivered image quality in survey operation rather than a specification. Image quality is monitored continuously; units are re-aligned individually as needed.
| Pixel scale | 0.505″ per pixel |
|---|---|
| Field of view per unit | 1.34° × 0.90° (1.25 deg²) |
| PSF FWHM at field center | 1.4–2.2″ |
| Array median FWHM | 2.0″ |
| Unit-to-unit scatter | 0.2″ |
| Center-to-corner growth | +0.3″ |
| Ellipticity, central 80% of field | < 0.1 |
| Median site seeing | 1.5″ |
Photometric zero point
Photometric calibration runs against synthetic photometry derived from Gaia DR3 BP/RP spectra, homogenized to correct the color- and magnitude-dependent residuals reported by the Gaia collaboration. The procedure was established during commissioning on 68 spectrophotometric standard stars, including CALSPEC sources, with non-variable point sources selected following criteria adapted from SkyMapper DR4. Zero-point uncertainty across the twenty medium bands in operational use is 15 to 25 mmag, with the larger values redward of 775 nm where detector quantum efficiency falls and signal-to-noise drops accordingly.
| Reference | Gaia DR3 BP/RP synthetic photometry |
|---|---|
| Homogenization | Color- and magnitude-dependent residuals corrected |
| Established on | 68 spectrophotometric standards, incl. CALSPEC |
| Zero-point uncertainty | 15–25mmag |
| Redward of 775 nm | Toward the upper end of that range |
| Calibrated set | The original 20 medium bands |
Zero points are determined per image by Py7DT with 3σ clipping across multiple aperture sizes, against corrected synthetic photometry of matched Gaia sources. The fifteen filters installed in late 2025 are not yet spectrophotometrically calibrated and are excluded from the figures above; a calibration campaign following the same procedure is planned. Full methodology is in preparation (Paek et al.).
A 100-second exposure
For the canonical 100-second exposure the 5-sigma point-source depth reaches 19.06 mag in the bluest medium band (m400) and 16.60 mag at the longest wavelength (m875), peaking at 19.61 mag in m475 near maximum system throughput. The Sloan broad bands reach 20.59, 20.25 and 19.17 mag in g, r and i. These are nominal-condition figures: seeing better than 2.0 arcseconds, airmass below 1.5, and non-bright nights.
| m400 (bluest medium band) | 19.06 mag |
|---|---|
| m475 (peak throughput) | 19.61 mag |
| m875 (reddest medium band) | 16.60 mag |
| Sloan g | 20.59 mag |
| Sloan r | 20.25 mag |
| Sloan i | 19.17 mag |
One hundred seconds is the fiducial 7DS exposure: every survey visit is built from it, so a depth quoted for any program is this number scaled by the number of frames coadded. Background-limited, so four frames buy 0.75 mag.
Nominal conditions: seeing better than 2.0 arcseconds, airmass below 1.5, and a non-bright night. The spread around each figure on a real night is the width of the violins below.
What each survey reaches
The three surveys spend that fiducial exposure differently — over the whole southern sky once, over a smaller area every ten to fourteen days, or on one field every night — so they arrive at very different depths from the same instrument.
| Survey | Area | Cadence | Depth |
|---|---|---|---|
| RIS | 23,000 deg² | Single visit | 19.1 mag |
| WTS | 800–1,200 deg² | 10–14 days | 22.2 mag* |
| IMS | 8.5 deg² | 1 day | 23.6 mag* |
The RIS figure is one visit of 3 × 100 s. Figures marked with an asterisk are cumulative over the planned operation rather than the depth of any single visit, and are targets: status and overview reports what has actually been observed, and the coverage map on the data access page gives the integration time and estimated depth reached on any individual tile.
Planning an observation
To turn these figures into an expected signal-to-noise, combine the depths with the response curves: the supy package has a simulator module that generates filter and detector response for the 7DT bands, and an observer module for target visibility from El Sauce.