Imaging Calibrations
Scheduling to be Finalized
Overview
After detector effects are corrected, imaging calibration converts the signal at the Wide Field Instrument (WFI) focal plane into a faithful representation of incoming sky flux for direct imaging observations. Imaging calibration activities address both astrometric and photometric effects, as well as other critical components such as the careful determination of the Point Spread Function (PSF) from imaging data. The PSF is needed to determine accurate point-source positions and fluxes, especially in regions with elevated crowding, as well as to determine the accurate shape of images of extended sources.
Roman will observe many astrophysical sources during its 5-year primary mission, in particular through its community-defined surveys, which can also be readily used for calibration purposes. For example, the many observations taken for the Galactic Bulge Time Domain Survey can help improve the characterization of Roman's PSF. Nonetheless, dedicated observations are still required for several imaging calibrations (e.g., relative astrometry). The frequency of these calibration activities is currently somewhat conservative and based on experience obtained from other missions (primarily HST and JWST), and will be adjusted on orbit. The Roman Science Requirement Document provides a detailed flow-down from science requirements to calibration requirements and the required accuracy in the calibration is the strongest driver for the observation sequences and observation cadence.
A summary of the imaging calibrations are given in the Table of Imaging Calibrations and individual sections provide a more detailed description (including links to the WFI Detectors articles), summarize the requirements on the calibration (if they exist), briefly describe the nominal activities planned for the calibrations and their cadence, and connect to the calibration files that are provided for users. Links to other relevant content related to the Roman Community-defined Surveys and Science Data Pipelines are provided when applicable.
Table of Imaging Calibrations
| Calibration Name | Short Description | Observation Cadence | SOC Products |
|---|---|---|---|
| Spatial Uniformity | variations on sub-field-of-view scales | Yearly | large-scale component of the flat reference file |
| variations larger than a field-of-view | self-calibration procedure | ||
| Bandpass Variation | variations in the effective band-pass due to angle-of-incidence variations over the WFI field-of-view | Twice in the Prime Mission | |
| Flux Calibration | calibration of the counts on the detector to an absolute flux system | Varies | photom reference file |
| Temporal Photometric Stability | tracing changes to flux during the mission | Varies | photom reference file |
| Cross-Survey Calibration | understanding of how to use Roman surveys with other relevant survey data | Varies | |
| Relative Astrometry | accuracy of positions within detectors | Yearly | distortion reference file, science instrument aperture file |
| Point Spread Function | description of the distribution of light for point sources on the detector | Monitored in Astrometric Touchstone Field | ePSF reference file |
Spatial Uniformity
Description
Many of the Roman Community-defined Surveys require that the photometric response of the instrument remains uniform both across the WFI field-of-view as well as on much larger scales (up to several degrees), albeit the requirements are at different levels of precision. Over the WFI field-of-view, the variation in photometric response has at least two primary sources: (1) non-uniformity in the overall photometric response as a function of position and (2) field-dependent variations in filter bandpasses. Spatial uniformity over the scale of the WFI is addressed in the large-scale component of the flat field (the l-flat) and applied in the Exposure Pipeline.
Requirements
On a scales smaller than the WFI field-of-view, the photometric response must be uniform to 3 milli-magnitudes RMS to avoid impacting the weak-lensing figure of merit. Systematic spatial non-uniformity affects the photometric redshift determination and source selection for both High Latitude Time Domain Survey and High Latitude Wide Area Survey science goals. The relevant scale for the High Latitude Time Domain Survey is for non-uniformity to be controlled down to about 5 milli-magnitudes on scales of about 30 square degrees. For the High Latitude Wide Area Survey, the scales of interest a few milli-magnitudes over a few degrees.
Calibration Activity
Sub-Field-of-View Scale
To achieve sub-FOV-scale uniformity, a medium density stellar field with at least 100 bright and relatively isolated stars per detector will be observed multiple times in a two-dimensional dither pattern on arcminute scales. The primary target for this activity is the Large Magellanic Cloud Touchstone Field.
Large Scale
The primary method to achieve photometric uniformity over large scales is an application of self-calibration akin to the Ubercal procedure (Padmanabhan et al. 2008, Schlafly at al. 2012) used for SDSS Imaging Survey (Padmanabhan et al. 2008), Pan-STARRS (Schlafly at al. 2012), as well as other other ground-based large-area imaging surveys. This method has demonstrated the ability to achieve the necessary ∼ 5 milli-magnitudes spatial uniformity. Roman will have greater stability than ground-based observatories due to space conditions, which eliminate airmass and seeing variations. We expect the statistical error from the Ubercal procedures to be less than 1 milli-magnitude.
Cadence
Sub-field-of-view calibration activities will occur yearly.
The self-calibration activity can be continuously derived from standard survey data. The typical time scale for an independent solution depends on angular scale and precision.
Calibration Files
Information from sub-FOV uniformity provide the large scale flat (l-flat) component that modifies the pixel-level flat fields on the Calibration Reference Data System (CRDS). Large-scale calibration activities are applied to catalog level products.
Bandpass Variation
Description
WFI Ground Testing Campaigns characterized bandpass shifts over the field-of-view of WFI. These shifts are due to the distribution of angles of incidence of the light contributing to each source at the exit pupil where the filters are located. As a result, the single-band photometry and the effective color of a source will depend on where it is located in the field-of-view.
Requirements
Uncertainty requirements on single-band photometry over the field-of-view are between 3 to 5 milli-magnitudes.
Calibration Activity
A rich stellar field will be observed through a variety of large dithers and analyzed to model the color-dependent photometric residual variations after corrections for small-scale uniformity have been taken into account.
Cadence
Two calibration runs are baselined over the nominal 5-year primary mission. The regular observations of the Touchstone Fields provide regular monitoring and, if an issue arises, can trigger more frequent repeats of this activity.
Calibration Files
This type of photometric correction depends on the properties of each source as determined through scientific investigations.
Flux Calibration
Description
A point source mean flux over a passband can be defined as
| <F_{\lambda}> = \frac{\int F_{\lambda }\cdot \lambda \cdot R\cdot d\lambda}{\int \lambda \cdot R\cdot d\lambda} = S_{\lambda}\cdot N_{e} |
where R is the system throughput, S_\lambda is the wavelength-dependent instrument sensitivity, and N_\textrm{e} is the detector count rate in electrons per second. A similar expression can be derived for the frequency-dependent flux <F_\nu>. The detector count rate can be expressed as:
| N_{e} = \frac{A}{hc}\int F_{\lambda }\cdot \lambda \cdot R \cdot d\lambda |
where Ais the telescope collecting area. From these we can derive the ``inverse sensitivity" S_\lambda as:
| S_{\lambda} = \frac{<F_{\lambda}>}{N_e} = \frac{hc}{A\cdot \int \lambda \cdot R\cdot d\lambda } |
since it scales inversely to detector count rate (N_\textrm{e}). (A similar expression can be derived for its frequency-dependent counterpart). The inverse sensitivities are measured through observations of spectrophotometric standard stars, whose spectral energy distributions are well-established. The sensitivity is then inferred via the measured signal (N_\textrm{st}) from spectrophotometric standard stars.
Requirements
Key requirements come from both High Latitude Time Domain Survey and High Latitude Wide Area Survey science, resulting in the following requirements:
- better than 2% per-bandpass absolute photometry;
- better than 0.5% in the flux ratio between any two filters; and
- better than 0.5% in the variation of the flux calibration.
Calibration Activity
Spectrophotometric standard stars over a range of brightness and spectral type will be observed with each WFI detector for all the filters using a 4-point sub-pixel dither pattern and with a minimum signal-to-noise ratio of 100 per exposure. Suitable stars are selected from the CALSPEC database and cover spectral types from white dwarfs to K dwarfs. Stars of different spectral type are used to ensure that there are no systematic color-dependent effects in the resulting calibration.
Cadence
The current plan, subject to change, is to take imaging observations of at least four of the primary standards per WFI detector during the first year. In Years 2 to 5, each star in the set is observed once per year. Standard-star observations are distributed evenly throughout the year, with one or more stars per quarter.
To monitor temporal or spatial variations in detector response, observations of the Touchstone Fields will be taken four times per year.
Calibration Files
The photom reference file contains the throughput and zero points for each of the WFI detectors and is hosted on the Calibration Reference Data System (CRDS).
Temporal Photometric Stability
Description
Several Roman Community-defined Surveys acquire data at different times during the 5-year primary mission (see e.g., Roman Observations in the First Two Years of Science Operations) and it is critical that these data can be directly intercompared. This implies that the sensitivity of the overall system remains the same to sufficient accuracy over the time scales of interest. For large-area surveys such as the High Latitude Wide Area Survey, variations over time induce spatial variations that can introduce unwanted systematics in the analysis.
Requirements
Each core survey has a different timescale over which photometric stability must be maintained. These go from 1 milli-magnitude over each 72-day season and across seasons for Galactic Bulge Time Domain Survey to 10 milli-magnitude RMS on spatial scales of about 10% of the WFI FOV over its duration (High Latitude Wide Area Survey).
Calibration Activity
High-precision monitoring of the photometric stability can in most cases be achieved by self-calibration. For instance, the Galactic Bulge Time Domain Survey and the High Latitude Time Domain Survey will observe the same fields multiple times within each season, allowing for very precise characterizations of photometric variations. The analysis for the High Latitude Wide Area Survey is more complex, as observations do not repeat in the same pattern for each field across the survey. However, repeated observations of the same sources can still be used to determine and correct for drifts in the detector photometric response.
Cadence
In addition to self-calibration within each survey, additional observations of the Touchstone Fields can also be used for this activity.
Calibration Files
The photom reference file contains the throughput and zero points for each of the WFI detectors and is hosted on the Calibration Reference Data System (CRDS). The file will be updated with revised throughput measurements.
Cross-Survey Calibration
Description
Roman Community-defined Surveys and general-investigator programs produce data products that complement surveys and programs carried out with other facilities both on the ground and in space (e.g., 2MASS, UKIDSS, Pan-STARRS, DESI, HST, JWST, among others).
Requirements
To address dark-energy science objectives, the photometric agreement between Roman and other data must be 2 milli-magnitudes.
Calibration Activity
Joint analyses of these data to meet specific science objectives confront challenges in calibration to account for instrument specific characteristics (bandpass, detector effects, plate scale, etc.), site specific corrections (seeing, transparency, etc.) and observing strategies (completeness, cadence, etc.). Where the data overlap in wavelength, the photometric transformation (color terms) must be well quantified. When non-overlapping wavelength regions are in play, the flux calibration scale
becomes important.
Cadence
The Roman Community-defined Surveys were designed with sufficient overlap with relevant major astronomical surveys.
Calibration Files
No direct calibration files are made for this activity.
Relative Astrometry
Description
The goal of an astrometric solution for the WFI is to provide an accurate mapping from the position of a source in raw detector coordinates to position in distortion-corrected coordinates. These can be either expressed on a rectified pixel grid or as relative angles on a sphere. Absolute astrometry then takes these relative angles on a sphere and applies Gaia-based zero points to obtain absolute sky coordinates.
The absolute astrometric calibration will be carried out on each individual exposure in the Exposure Pipeline using common Gaia stars as a reference, and is not part of the on-orbit calibration plan.
Requirements
There are several science requirements on relative astrometry, the most stringent of which come from the Galactic Bulge Time Domain Survey: 1 milliarcsecond per measurement for a star of HAB=21.4 in at least two filters, and a systematic precision better than 0.1 milliarcseconds per season for sources measured in at least two filters. Relative astrometry requirements for Roman are 10 times more stringent than HST and JWST and, therefore, require a more sophisticated calibration plan and data analysis.
Calibration Activity
The geometric-distortion solution of each WFI detector will be modeled using several dithered exposures of the Astrometric Touchstone Field. There will be three different sets of observations:
- A "small", custom-made dither pattern made up of 7 x 7 pointings that cover a WFI detector from corner to corner. This set will be used to model the geometric distortion of each of the 18 WFI detectors.
- A "large", custom-made dither pattern made up of 9 x 5 pointings that cover the entire WFI FOV from corner to corner. This set will be used to map the distortion-corrected pixels of each WFI detector into the distortion-corrected pixels of any other WFI detector.
- A "roll" set made up of a single pointing but observed at different position angles (V3PA), spaced by 60 degrees from each other. This set will be used to characterize the skew terms of the distortion.
The calibration itself will be initially carried out referencing the positions from Gaia, but later improved through a self-calibration approach using Roman measurements simultaneously to solve for the distortion and to define a common distortion-free reference-frame system.
Source positions from each exposure, based on precise effective PSF fitting, are transformed on to a reference frame system and compared to distortion-free positions from either Gaia or the self-calibration step. Residuals will be modeled with two-dimensional polynomials and the solution will be augmented by look-up tables of residuals and pixel-based fine corrections to account for detector-related effects.
Cadence
Relative astrometry will be calibrated once a year through all 8 imaging optical elements.
Calibration Files
The full pixel-based distortion model for each of the 8 imaging optical elements will be generated and hosted on the Roman Calibration Reference Data System (CRDS).
The coefficients of the polynomial component of the full solution will also be made available through the science instrument aperture file (SIAF), also in CRDS.
Point Spread Function
Description
The effective PSF (ePSF) includes all linear contributions that lead to smearing out of the light from a point source – this includes optics, but also image motion, lateral charge diffusion, the pixel top-hat function, IPC, etc. Here, the ePSF is not to include non-linear effects (classical or count-rate non-linearity, the brighter-fatter effect, etc.).
Requirements
The ePSF is required to be known with ellipticity errors less than 5.7 × 10−4 RMS per component and relative size errors less than 7.2 × 10−4 RMS over the WFI field-of-view. Here the ellipticity components are e_1=(I_{xx}-I_{yy})/(I_{xx}+I_{yy}) and e_2 = 2 I_{xy}/(I_{xx}+I_{yy}), and the size is defined by (I_{xx}+I_{yy}), where I_{ij} is the adaptive second moment.
Calibration Activity
The ePSF is determined from stars observed in the field. In sparse stellar fields like those in the High Latitude Wide Area Survey, principal-component analysis (PCA) provides a more practical approach. As part of the relative astrometry calibration, the SOC releases spatially-dependent, ePSF library models for each of the 8 optical elements, computed using the formalism first introduced in Anderson & King (2000). These library models might be expanded to account for telescope thermal variations and stellar spectral energy distributions (SEDs).
Cadence
The PCA-based models will be based on the science exposures themselves. The library of PSF principal components will be built up over long timescales, probably from the beginning of science operations to the cutoff date for each data release. Preliminary library models based on the Anderson & King (2000) formalism will be made available early during the 5-year nominal mission and improved as more data is collected.
Calibration Files
ePSF library models based on Anderson & King (2000) formalism are hosted on the Roman Calibration Reference Data System (CRDS). The release of PCA-based ePSF models is to be determined.
For additional questions not answered in this article, please contact the Roman Help Desk.
References
- "An Improved Photometric Calibration of the Sloan Digital Sky Survey Imaging Data", Padmanabhan et al. 2008
- "Photometric Calibration of the First 1.5 Years of the Pan-STARRS1 Survey " Schlafly et al. 2012
- CALSPEC database: https://www.stsci.edu/hst/instrumentation/reference-data-for-calibration-and-tools/astronomicalcatalogs/calspec
- "Toward High-Precision Astrometry with WFPC2. I. Deriving an Accurate Point-Spread Function" Anderson & King 2000
- UKIDSS: Lawrence et al. (2007), MNRAS, 379, 1599