This page has the current version of Roman WFI Performance. This is v1.1, January 2025.
Roman Performance Information
WFI Overview
The Wide Field Instrument (WFI) is Roman’s primary science instrument and enables large scale surveys. The WFI is a 300-megapixel visible-to-NIR imaging camera and slitless spectrometer. In the full Roman Observatory, the WFI is composed of two major components: the Cold Sensing Module (CSM) and the Warm Electronics Module (WEM).
The CSM is situated between the telescope and spacecraft bus and latched to a structure called the Instrument Carrier (see figure below). At the core of the CSM is a detector array of 18 Teledyne H4RG-10 Sensor Chip Assemblies (SCAs) and their associated readout and control electronics.
The SCAs in the focal plane array produce an 0.4 x 0.8 deg (0.281 deg2 excluding detector gaps) field-of-view, ≈200× larger than Hubble’s WFC3-IR camera but with better sensitivity and comparable spatial resolution.
Before reaching the detector array, light from Roman’s optical system encounters the Element Wheel Assembly (EWA). The EWA is a rotating wheel that contains multiple filters for imaging and grism and prism dispersers for slitless spectroscopy across the full field.
The focal plane array is mounted into a mechanism called the Alignment Compensation Mechanism (ACM) that allows for fine focus and alignment.
The WFI also carries an internal relative calibration system to accurately calibrate and trend detector response, particularly linearity, over the full mission lifetime.
The WEM provides instrument commanding and mechanism control. It is located in one of the spacecraft bus bays.
The figure below shows the CSM in context with the observatory and provides details on the major subsystems (adapted from Schlieder et al. 2024). The table below provides a top-level overview of WFI parameters that are driven by mission science requirements.
Parameter
Value
Focal Plane Array
Detectors
18 Teledyne H4RG-10 detectors with 4096 x 4096 pixels
Field of View
0.8 deg x 0.4 deg (0.281 deg2, excluding gaps)
Spatial Sampling
0.11 arcsec/pixel
Pixel size
10 μm
Image Stability
1.0 nm RMS wave front error (WFE) in 180 sec
Guiding
Guide star sensing interleaved with science data collection
Element Wheel
Modes
Imaging, spectroscopy, and calibrations
Imaging
8 imaging filters spanning 0.48 to 2.3 μm
Spectroscopy
Prism and grism for full-field, slitless spectroscopy spanning 0.75 to 1.93 μm
Calibrations
Dark element and filter mask diffusers allow darks, flat fields, and other calibrations with WFI's internal relative calibration system
WFI carries 8 science filters with overlapping band passes spanning 0.48 – 2.3 microns. The filters are located at the telescope exit pupil. They are housed in the element wheel and are rotated into the instrument light path for multiwavelength imaging.
Filter Point Spread Functions
This update is based on the post-CDR optical design. The wavefront error model includes design residuals and Monte Carlo estimates of surface figure errors and alignment tolerances. We plan to issue a new release in 2025 once we have wavefront measurements of the integrated telescope and Wide-Field Instrument.
Point spread functions (PSFs) for the Nancy Grace Roman Space Telescope have been created using WebbPSF version 1.0, a Python-based package. This tool takes into account properties of the telescope and the instruments, including detector pixel scale, rotations, filter profiles, and point source spectra. These are not full optical models, simply a tool that transforms the optical path difference maps, into the resulting Roman PSFs.
Imaging PSFs were calculated at the center of each SCA and also around the periphery of the focal plane; located at the center of a pixel and at the corner of a pixel.
*Note: PSF FWHM in arcseconds simulated for a detector near the center of the WFI FOV using an input spectrum for a K0V type star. Please click on the FWHM value for each filter to view the simulated PSF.
The above table provides representative PSF FWHM values for a detector (SCA 1) near the center of the WFI FOV. Wavelength distribution is that of a K0V star, sampled over each filter bandpass. N Eff Pix is number of effective pixels (aka noise pixels) under the PSF (1/sum of squares of pixel values). Fluxes normalized at telescope exit pupil. Images are ~7 arcsec square and typically contain 97%-99% of the incident flux. Two cases are provided: at the corner of 4 pixels, and at the center of a pixel.
FWHM (arcsec) of PSF is computed from 8-times oversampled PSF. Gaussian pointing jitter with FWHM = 8 mas is included. The number of noise pixels and maximum flux per pixel are computed on native detector pixels.
The Roman effective area has been updated to reflect recalibration of the sensor ship assembly (SCA) quantum efficiency, and preliminary updates to the filter, prism, and grism bandpasses. The tables have also been broken out by SCA to illustrate the small differences in QE and the shift in filter bandpasses with field angle. Download the Filter Effective Area tables [.ZIP]
The tables are in ECSV format, which is can be read via astropy.io.ascii.read() and any text editor.
The plot below illustrates both these effects by comparing the effective area for SCAs 1 and 9 for filter F158.
The bandpass shifts given here are representative models; these will be updated once the full set of data from the wide-field instrument thermal vacuum test has been obtained and analyzed.
The following figure shows the effective area for the full set of Roman filters and dispersers, for SCA #1.
Imaging Sensitivity
Imaging Sensitivity Overview
(Updated June 3, 2024)
The table below gives the 5-sigma AB magnitude limiting sensitivity, for twice the minimum zodiacal light background (roughly equivalent to that obtained at an ecliptic latitude of 25 degrees at a Solar elongation of 90 degrees), for 57 second and one-hour integrations, for point sources and a compact galaxy with half-light radius of 0.3 arcseconds.
S/N was computed for photometric apertures of radius 2 pixels for point sources and 6 pixels for the galaxies.
Scalings for other deep cases: The flux density limit for other deep integrations, between about ten minutes and tens of hours, can be estimated from the 1 hour depths using a flux ∝ t-1/2 scaling. For integrations below about 10 minutes, this scaling becomes over-optimistic by > 30% due to read noise and overheads. Limits for a half-light radius of 0.2” are slightly closer to the 0.3” case than the point source case. For more extended sources, the limiting flux density fν,lim for fixed SNR and integration time should be scaled from the 0.3” case approximately as fν,lim∝r50 or ΔABlim=2.5 log(0.3"/r50). Reducing the assumed zodiacal background from 2x to 1.44x minimum improves deep imaging flux limits by about 0.15 mag for the F062 through F158 filters, 0.08 mag for F184 and F146, and negligibly for F213.
Fast/Wide Limit: The final two lines in the table give magnitude limits achieved at 5σ in 55 seconds (with a single exposure). At this integration time, Roman can cover approximately 8 contiguous square degrees per hour in one spectral element, and slew-and-settle overheads slightly exceed integration time. There is little point considering faster survey speeds, because sensitivity drops rapidly for modest increases in survey speed at yet shorter integrations.
Zodiacal Light and Thermal Background
(Updated June 3, 2024)
The tables below provides the count rate per pixel at minimum Zodiacal light in each filter and the estimated thermal background. For observations at high galactic latitudes, the Zodi intensity is typically ~1.5x the minimum. For observation into the galactic bulge, the Zodi intensity is typically 2.5-7x the minimum.
On this page you will find tools designed to help you, the user, calculate the exposure time required for a given source at a given signal to noise, or vice versa.
We have provided a Jupyter notebook which walks through each step of the calculation, and a python script that will open a GUI in which you can input your objects information.
To run both scripts you will need to download the accompanying data files.
Both the notebook and the GUI require the user to specify the filter, zodiacal light contribution, type of source, fitting method, and signal to noise. In this notebook we will be doing exposure time calculations for point sources, and extended sources with half-light radii of 0.2 arcsec or 0.3 arcsec.
WFI carries 2 dispersive elements for slitless, multi-object spectroscopy. The grism band pass spans 1.0 – 1.93 microns and has a resolution of ~600. The prism band pass spans 0.75 – 1.80 microns and has a resolution of ~100. The dispersing elements are housed in the element wheel and are rotated into the instrument light path for slitless spectroscopy across the WFI FOV.
The effective area as a function of wavelength for the Grism and Prism.
Grism and Prism zodiacal light
The table below provides the count rate per pixel at minimum zodiacal light for the grism and prism. For observations at high galactic latitudes, the Zodi intensity is typically ~1.5x the minimum. For observation into the galactic bulge the Zodi intensity is typically 2.5-7x the minimum.
The Roman WFI slitless grism has a spectral range of 1.00-1.93 microns and a dispersion of about 1.1 nm/pixel, essentially independent of wavelength, yielding a 2-pixel resolving power of R = λ / δλ = 460 λ / μm for a point source. Table gives 5-sigma detection limits for a one-hour exposure time with zodiacal light background at twice the minimum intensity. This is representative of an ecliptic latitude of 25 degrees and 90 degrees ecliptic longitude relative to the Sun (the middle of the object visibility window). Typical HLWAS backgrounds are ~30% lower.
For emission lines, the values are integrated line fluxes in units of 10-17 ergs/cm2/sec.
For continua, the values are the AB magnitude at which S/N=5 per pixel.
The values are averages over all 18 detectors; typical variations from one detector to another are 0.05 to 0.1 mag for the continuum cases and ~10% for the emission line limits.
5σ limits for Roman WFI grism in 1 hour on source, 2x minimum zodiacal background.
Emission line limits are in units of 10-17 erg cm-2 s-1, and continuum limits are in AB mags for 1 pixel
Sensitivities for other integration times (between a few minutes and tens of hours) and zodiacal backgrounds can be scaled from the above using f lim∝t-1/2 b1/2, where t is integration time and b the zodiacal background level. Because this is slitless spectroscopy, the scaling of sensitivity with size for r∝r50 > 0.3" differs between line and continuum sensitivity, with limiting behaviors of flne ∝r50, and fcont ∝r501/2.
Prism Spectroscopy Sensitivity
(updated June 11, 2024)
The Roman WFI slitless prism has a spectral range of 0.75-1.80 microns and a resolution that is strongly wavelength dependent, with 80 < λ / δλ < 180. The highest resolution is at the blue end of the prism wavelength coverage. In addition to its lower dispersion, the prism has higher throughput than the grism, making it more sensitive to continuum. The table below has AB magnitude at which S/N=5 per pixel (not per 2 pixels as earlier), at zodiacal light at twice minimum. This is representative of an ecliptic latitude of 25 degrees and 90 degrees ecliptic latitude relative to the Sun (middle of the object visibility window). Typical HLWAS backgrounds are ~30% lower.
These are averages over all 18 detectors; typical variations from one detector to another are 0.05 to 0.1 mag.
5σ limits for Roman WFI prism, 2x minimum zodiacal background
The same scalings that apply to grism spectroscopy can be used to scale other deep prism sensitivities from the 1-hour case.
Grism and Prism dispersion
The grism has constant dispersion and linearly increasing resolving power. The prism provides higher throughput and lower dispersion than the grism. The prism dispersion varies with wavelength and varies slightly with field angle.
These measurements were performed on either individual flight SCAs in the NASA Goddard Detector Characterization Lab (DCL) at a temperature of 95 K or on the full flight focal plane during WFI Thermal Vaccuum Test #2 (TVAC2) in the nominal operation (NomOp) test plateau at a temperature of 89.5 K. The test parameters for each performance measurement are noted in the descriptions. In these data tables individual detectors are described by either their Sensor Control Unit (SCU) number, which defines their position in the focal plane array as an integer 1 through 18, or their SCA serial number, which is a 5 digit integer that defines each individual detector. Please refer to the SCU to SCA mapping table and labeled diagram of the flight focal plane array for reference.
Roman/WFI SCU to SCA mapping
Sensor control unit (SCU) number to sensor ship assembly (SCA) serial number mapping.
Roman/WFI TVAC2 Nominal Operation (89.5 K) total noise measurements (means, medians, and percentage of pixels passing the total noise requirement) for each SCA.
Roman/WFI total noise measured for each Sensor Control Unit (SCU). Total noise data were acquired from WFI TVAC2 measurements of 100 total exposures, each with a ~170-second integration time, consisting of 55 non-destructive reads and 0 skip frames. Specifically, 50 exposures were taken with the Guide Window (GW) at the bottom left of each sensor chip assembly, and 50 with the GW at the top right. Data analysis included top and column reference pixel correction (using a Savitzky-Golay filter), slope calculation of digital number per unit time per pixel using all frames, and total noise calculation per pixel as the standard deviation of the slopes multipliedby the total exposure time (~170-seconds). A global conversion gain was then applied to convert the results to electrons.
Roman/WFI TVAC2 Nominal Operation (89.5 K) dark current measurements (means, medians, and percentage of pixels passing the dark current requirement) for each SCA.
NOTE: These measurements are representative of the instrument internal thermal background rather than the true dark current floor of the detectors.
Roman/WFI TVAC2 Nominal Operation (89.5 K) persistence measurements as a function of time was measured using 10 regularly spaced darks spanning a total time of 1770 secs (~30 min) following exposures of 56 frames each for each SCA. The persistence decay where the detectors were exposed to ~900 e-/s flat field illumination, for a total of ~159 ke- of charge accumulation.
Roman/WFI TVAC2 Nominal Operation (89.5 K) exponential decay fits to persistence measurements as a function of time for each SCA. The median persistence (e-/s over the dark current) was fit to an exponential function with this form: a*e-bx+d*e-ex+c, where x is the time in the decay curve and a, b, c, d, and e are fit coefficients provided in this file.
WFI focal plane persistence decay curves. Curves are shown for each SCA in the focal plane during a series of dark measurements that occurred after 56 frames of ~900 e-/s flat field illumination. These measurements were performed during WFI TVAC2 with the detectors at 89.5 K.
WFI focal plane persistence response in the first dark after flat field illumination (56 frames at ~900 e-/s). The color scale is designed to emphasize the small differences in persistence response at the pixel level both within a given SCA and across the full focal plane. These measurements were performed during WFI TVAC2 with the detectors at 89.5 K.
WFI Quantum Efficiency
Roman/WFI Detector Characterization Lab measurements of the median quantum efficiency (QE) versus wavelength (linearly interpolated) for each SCA.
The Roman/WFI quantum efficiency for each of its 18 SCAs.
Roman/WFI TVAC2 Nominal Operation (89.5 K) pixel operability statistics. For each SCA this includes the number of pixels deemed inoperable, which do not meet requirements ("bad pixels"), and the percentage of pixels deemed operable that do meet requirements ("good pixels") for each SCA. The performance metrics that can contribute to a pixel being inoperable are dark current, total noise, linearity, persistence, QE, flat field response, gain, and electrical connectivity.
NOTE: SCU/SCA 4/21115 has a significantly lower operable pixel fraction due to an extended region of the detector with persistence that is somewhat larger than the requirement.
