Introduction to STIPS for Roman
The STIPS (Space Telescope Image Product Simulator) software produces simulated imaging data for complex wide-area astronomical scenes, based on user inputs, instrument models and library catalogs for a range of stellar and/or galactic populations. It was originally developed for the James Webb Space Telescope (JWST) mission, and has been extended to include Nancy Grace Roman Space Telescope functionality as well. The current Roman Space Telescope version produces images covering a single detector in the Wide Field Instrument (WFI) field of view. STIPS includes current information about the telescope sensitivity, spectral elements, and detector properties. It uses the PSF model generated by WebbPSF1, and it calls the appropriate Pandeia modules to compute instrumental throughput and count rates.
A new version of STIPS (v2.0) was released in September 2022. The release includes the implementation of empirical PSFs to improve the astrometry of the sources, as well as several bug fixes. STIPS is available on GitHub as a Python module, and also via PyPI.
Functionality
At the heart of STIPS is a generator that allows varying degrees of user control. STIPS can generate a randomized set of stars or idealized galaxies based on very concise user specifications, or it can generate point or extended sources based on a detailed input catalog. In addition, an existing image can be used as background.
Accordion
Each of the sources populating a STIPS simulated scene can be a point source or an extended source. Point sources are characterized by their position and spectral energy distribution. Extended sources are in addition characterized by major axis, minor axis, orientation, and light profile, assumed to follow a Sérsic parametrization. For either point or extended sources, the spectral energy distribution is converted into a count rate at the detector depending on the filter(s) chosen and the current instrument model; the user can choose to specify directly the count rate in each filter instead.
Randomized Source Catalog
For a quick implementation that will satisfy many user needs, the distribution of point or extended sources can be specified in terms of a stellar or galaxy population based on a number of global parameters. The parameters for point sources are designed to mimic a stellar population; those for extended sources are more suited to a galaxy population.
While the source distributions are random, they are also repeatable. The user can explicitly set the seed value for the random number generator used to generate the source distributions, in which case two scenes produced with the same input parameters and the same random number seed will be identical.
Stellar Populations
Stellar populations are built on the basis of the Padova isochrones, specifically the version CMD 2.7. The basic parameters required are the number of stars, their metallicity and age range, and the characteristics of the Initial Mass Function (IMF). In addition, the spatial distribution of the stars can be specified via its shape and radius, and a distance range can be specified as well. If desired, the mass distribution can be modified to allow more massive stars to be closer to the center, in order to mimic the effect of mass segregation. Mass segregation is done by sorting the sources by their distance from the center, and then placing high-mass stars (stars with masses greater than 75% of the maximum mass) at shorter distances with a slightly higher probability than would be produced by random distribution. Multiple populations can be included. The module will generate a catalog of stars, which can be retrieved as part of the simulation products. Stellar spectral energy distributions are then generated using the PHOENIX models.
Galaxy Populations
Galaxy populations are built on the basis of a user-specified range of sizes, brightnesses, and redshifts. Spectral energy distributions are drawn at random from a predefined set of Bruzual-Charlot models, and are assigned randomly either an exponential or a de Vaucouleurs profile with randomly generated shape and orientation. The spatial distribution can be specified in the same way as for stellar populations. As for stellar populations, the catalog of galaxies can be retrieved together with the simulation results.
User-provided Input Catalogs
If the populations-based description is not sufficient, the user can have full control of the definition of the astronomical scene by providing one or more input catalogs specifying the properties of each individual source, including position, brightness, and, for extended sources, size, elongation, orientation, and light profile (in the form of a Sérsic index). When an input catalog is provided, the apparent brightness of each source needs to be specified in each filter; a future upgrade will allow spectral energy distributions to be specified instead.
The user can specify a background image to be added to the scene. The WCS of the image may be derived from its FITS header, or the image may be treated as centered at the scene center. The image's flux units and pixel scale must be specified, and the image will be remapped to the Roman detector pixel scale prior to being added to the scene.
The user specifies the desired exposures by filter and exposure time. Exposures can be oversampled if desired. A background level based on a model of zodiacal light is included. Currently, STIPS uses the JWST Background Tool (JBT) to calculate background flux based on source position, and then combines the resulting flux with the instrumental throughput to generate the background value.
STIPS includes a full description of the instrumental PSF as obtained from WebbPSF.
Images produced by STIPS by default include noise and residual errors estimated on the basis of a simple instrument model. This model includes reasonable estimates of possible post-pipeline residuals from flat fielding, dark-current subtraction, and imperfect cosmic ray correction. Error terms are calculated as follows:
- Poisson Noise: Since the counts in the expected Roman/WFI image are quite high, STIPS assumes that the Gaussian and Poisson distributions are approximately equal. Therefore the noise in the simulated image is given as a Gaussian random number with mean zero and variance equal to the signal level
- Read noise: STIPS currently calculates readnoise by dividing the exposure into groups of no more than one thousand seconds, and generating the per pixel read noise, Rp, as \((\sqrt {N_g}) * k * T ^a_g\) where Ng is the number of groups, Tg is the time per group, and k and a are per-instrument constants.
- Flat field: For each instrument, a fixed-pattern flatfield residual file (with a mean of 0) was created, and this file is multiplied by the data.
- Dark: As with flat field, but using a separate file, multiplying the residual file by the exposure time, and adding the resulting dark residual to the image rather than multiplying it.
- Cosmic Rays: The probability of a cosmic ray hit (per pixel) is calculated based on pixel size and exposure time. For each hit, a cosmic ray size and energy are calculated based on the distribution of cosmic rays, and the noise in the pixels associated with the cosmic ray is increased. The pipeline process is assumed to remove 99% of the impact of cosmic rays.
- Source catalogs for each included population
- Image data in FITS format
- Quick-look images in JPEG format
- PSF images
- As-observed catalogs for each observation of the included populations.
STIPS produces quick-look JPEG images, FITS images of each individual detector and offset, mosaic images combining detectors into instruments, and combining detector dither patterns. In addition, it provides copies of generated catalogs, both as-created and as-observed. The as-observed catalogs are generated during the observation, and provide the detector coordinates and count rates of every source appearing in the exposure. Because they are generated before PSF convolution and before residual noise is introduced, the as-observed catalog can be used to test the efficiency and accuracy of a source-detection program (e.g., DAOPHOT) on the output data.
STIPS mosaics are generated by the Montage program, and are built without any coordinate distortion. They are not intended as accurate representations of drizzled images, but as a quick-look to see a rough result of (e.g., a combined dither pattern).
Using STIPS
STIPS can be run by users locally by cloning the GitHub repository or installing the PyPI package. STIPS Documentation is being maintained on Read the Docs.
Accordion
STIPS is still in development. The present implementation is available to the community as a beta version.
The Roman mission is currently in development, and observatory designs continue to evolve. Hence, STIPS-Roman simulations may not accurately reflect the actual future observatory.
STIPS makes several assumptions:
- The current implementation of STIPS uses the Phase C payload design, which incorporates information from the GSFC Instrument Reference Information files.
- STIPS relies in part on calculations with WebbPSF and Pandeia. Therefore, the same assumptions and caveats apply as listed for those tools.
- The Spectral Energy Distribution (SED) for stars and galaxies are obtained from the PHOENIX models and a subset of Bruzual-Charlot models, as described above, and are subject to the same limitations as those models. SEDs for galaxies are assumed constant over their images.
- STIPS post-calibration residuals from flat field, dark-current subtraction, and cosmic-ray correction are based on HST WFC3/IR observations.
- The STIPS background count rate is calculated with the JWST Background Tool (JBT), which determines the observability of a particular source using JWST’s orbital parameters.
- Images are generated in units of electrons.
- The photometric accuracy is undergoing further validation.
- Pixel effects (such as intrapixel sensitivity, interpixel capacitance; pixel response functions, and/or non-linearities) are not addressed.
Enhancements currently planned in the functionality and fidelity of STIPS include:
- Simultaneous images from all 18 detectors with appropriate geometric distortion;
- A more realistic treatment of readout modes and noise estimates;
- Saturation and sub-pixel effects;
- More customizable user input (e.g., per-source SED specification).
If you have suggestions on additional features, please send us suggestions at help@stsci.edu.
Credits, Acknowledgements, and Feedback
STIPS is currently managed and developed by the Roman Telescope Branch in the Instrument Division at STScI. Credit for initial development of the software go to Brian York and Robel Geda, with guidance and contributions from Klaus Pontoppidan, Jason Kalirai, Pey Lian Lim, and Stefano Casertano. STIPS makes use of open source astronomical software developed by Justin Bird, Ahmed Fasih, and Adam Ginsburg.
Users are encouraged to address questions, suggestions, and bug reports to help@stsci.edu with "STIPS-Roman question" in the subject line. The message will then be directed to the appropriate members of the STIPS-Roman team at STScI. In addition, bug reports may be filed against the STIPS GitHub module.
1 The Roman PSF module in WebbPSF is accessed through the following: 'from WebbPSF import roman', as of Webb PSF v1.0.0.
The NASA Nancy Grace Roman Space Telescope is managed by NASA/GSFC with participation of STScI, Caltech/IPAC, and NASA/JPL
Contact the Roman Team